Plurk paste

Introduction
The essential business of language is to assert or deny facts.
—BERTRAND RUSSELL1 The point of a discourse—at least one central kind of discourse—is the exchange of information.
—ROBERT STALNAKER2 Words can be like tiny doses of arsenic: they are swallowed unnoticed, appear to have no effect, and then after a little time the toxic reaction sets in after all.
—VICTOR KLEMPERER3 IN BOOK 3 of Thucydides’s The Peloponnesian War, the Athenian Cleon represents Mytilene’s revolt against Athens in the most extreme possible terms, claiming, “Mytilene has done us a greater wrong than any other single city.”4 Cleon claims that were the tables turned, Mytilene would slaughter every Athenian citizen—in other words, that the Mytileneans would carry out the very same action toward the Athenians that Cleon urges Athens carry out against Mytilene. Cleon’s speech mobilizes the Athenian citizens to genocidal action against Mytilene by employing the accusation that Mytilene is a genocidal threat to Athens. Cleon’s speech is political propaganda. It stokes irrational fear and desire for revenge, while simultaneously presenting itself as a reasonable contribution to discourse.
It justifies murdering the entire adult population of Mytilene not because of what they did, but because of an imaginary situation that Cleon gives no reason to think would be realized.
Cleon uses the savagery he suggests the Mytileneans would do if the tables were turned to justify the exact same course of action against the Mytileneans. Cleon’s speech is one of antiquity’s classical examples of demagoguery. In Cleon’s speech, he does not represent his own city, Athens, as greater or more exceptional in its value system and history than its enemies.
However, his speech is decidedly also not neutral, as he represents Athens’s interests as vastly more important—with a hypothetical future threat to its citizens judged far more serious than the actual threat to the lives of innocent Mytileneans. The speech completely takes the side of Athens while masquerading as some kind of aperspectival reason. Cleon’s speech centers the interests of Athens completely. The Mytileneans are visible only as genocidal threats.
Cleon’s speech is layered with emotion, values, perspective, and interests. It seeks to mobilize its audience toward action. One way to mobilize an audience is by providing information about the world. This book centers other ways in which language impacts audiences: by emotion, values, perspectives, interests, identity, and shared practices. We build a model of speech that incorporates these aspects as central from the very beginning.
Harmful Speech One way in which speech impacts a group of people is by harming them.
One kind of harmful speech, omnipresent in popular and academic discussions, is slurs—terms that target a group with an ideology that derogates its members. But the category of harmful speech is vastly broader than slurs. For example, Victor Klemperer describes a form of the linguistic process he calls “objectification” as follows: Why does a palpable and undeniable brutality come to light when a female warder in Belsen concentration camp explains to the war crimes trial that on such and such a day she dealt with sixteen “Stück” Gefangenen [prisoner pieces]? … Stück … involves objectification. It is the same objectification expressed by the official term “the utilization of carcasses [Kadaververwertung],” especially when widened to refer to human corpses: fertilizer is made out of the dead of the concentration camps.5 Linguistic objectification is a characteristic feature of various kinds of harmful speech. In chapter 10, we will return in detail to the topic of harmful speech and give our accounts of slurs, genocidal speech, and bureaucratic speech. To do that, we’ll first need to give an account of presupposition, in part II, for we will need to be able to explain, for example, how speaking of prisoners as “pieces” presupposes that they are less than fully human.
To understand Klemperer’s second example, we must also understand the connections between practices and “official terms.” These are connections that must be understood in terms of how speech attunes people to practices, an analysis of which is a central aim of part I of this book.
Here is another illustration, this time from the United States, of how speech attunes people to practices. John DiIulio Jr.’s 1996 magazine article “My Black Crime Problem, and Ours” begins by acknowledging that “violent crime is down in New York and many other cities.”6 DiIulio proceeds to predict “270,000 more young predators on the streets … [in] the next two decades.” He adds, “As many as half of these juvenile super-predators could be young black males.” DiIulio’s prediction was far off; violent crime continued to plummet.7 But the introduction of the term “super-predator” into criminal-justice discourse led (in difficult to quantify yet hard to dispute ways) to the adoption of ever-harsher laws concerning juvenile offenders.
Describing juvenile offenders as “super-predators” suggests that the proper practices toward juvenile offenders are the ones that are reasonable to take against enormous threats to humankind: death, or complete permanent isolation. Use of the term “super-predator” to describe juvenile offenders rationalizes treating them with practices that would only be reasonable to use against deadly enemies.
In the 1990s in the United States, criminal-justice policy had become a proving ground for politicians to demonstrate their putative toughness.
Debate was dominated by an ethos that frowned on expressions of empathy for perpetrators. Dehumanizing vocabulary targeting those caught up in the criminal-justice system was commonplace, and many of the words were racially coded.8 Rehabilitation is hard to envisage for those described as “thugs,” “super-predators,” or “gangsters.” During this period where these terms were part of the political discourse, criminal-justice practices became considerably harsher, and sentences longer.
9 Although the precise mechanisms continue to be a matter of debate, it is widely agreed that the culture surrounding crime policy had an extreme and rapid effect on criminal-justice practices. The incarceration rate in the United States hovered around the norm for liberal democracies of 100 per 100,000 for many decades until the late 1970s.10 Then it started to rise. The Bureau of Justice Statistics’ current rate of 810 for every 100,000 adults (18 years and older) in prison is by far the highest in the world.11 The United States has also developed a culture of policing marked by a level of fear and lack of empathy that is without parallel in liberal democracies (a 2015 headline of an article in the Guardian states, “By the Numbers: U.S. Police Kill More in Days Than Other Countries Do in Years”12). However, the unprecedented two-decade decrease in crime from 1991 until the early 2010s was not strictly due to the intensely punitive criminal-justice path that the United States chose to take in the 1990s. Canada experienced a similarly unprecedented drop in crime during this same time period, without following the United States’ path into mass incarceration.13 How does one investigate the way in which violent language about a targeted group affects attitudes?
As we will argue in part III, focusing on a case like this brings out the limitations of a model of conventional meaning that just theorizes in terms of a connection between words and things. To explain harmful speech, one must recognize conventional connections between words and practices, as well as words and emotions.
Hustle The examples of harmful speech we discussed in the last section involved expressions that attune their audiences to harmful practices in ways that are overt. Calling young Black American men “super-predators,” or, to use an example we will discuss later, calling Rwandan Tutsi “cockroaches” or “snakes,” directly attunes audiences to violent practices toward these populations. These examples highlight the need for a theory of meaning that connects speech not just with information, but with practices. But speech does not just impact an audience directly. It can and often does impact audiences indirectly.
Why would someone choose to impact an audience indirectly with their words, rather than overtly attempting to attune them in the desired manner?
The reason is because the speaker might not wish to be held responsible for their words.
The speaker may want to convey something in a way that allows for plausible deniability that they intended to convey it. Plausible deniability is a symptom of what we call hustle—speech that functions nontransparently. When speech is not transparent, a speaker has latitude to deny that they intended the nontransparent features.
Hustle is a large and diverse category, including insinuation (itself a broad category). One of the goals of the book is to show just how large it is.
While chapter 8 will describe hustle in more detail, this type of speech is our focus throughout the book. To illustrate it with an example, we’re going to focus in this section on one quite specific mechanism of hustle, the mechanism of the dog whistle.
Dog whistling involves employing speech that appears on the surface to be transparent, but, when married to a hearer’s background frame and value systems, communicates a message not obvious to those without that background (i.e., it functions nontransparently). Dog whistling is a mechanism specifically designed to allow plausible deniability. Though it is far from the only such method, dog whistling is useful to focus on in this introduction as it is most obviously a kind of hustle with a linguistic trigger.
In 1981, Lee Atwater, later to lead George H. W. Bush’s 1988 presidential campaign (featuring the notorious Willie Horton ad, funded allegedly by an independent PAC), had an anonymous interview with a journalist that remains one of the clearest expressions of the strategic value of code words to signal allegiance to ideologies that have been explicitly repudiated. In it, he famously said (although we’ve censored the original for obvious reasons), You start out in 1954 by saying, [N-word, N-word, N-word]. By 1968 you can’t say [N-word]—that hurts you, backfires. So, you say stuff like, uh, forced busing, states’ rights, and all that stuff, and you’re getting so abstract. Now, you’re talking about cutting taxes, and all these things you’re talking about are totally economic things and a byproduct of them is, blacks get hurt worse than whites.… “We want to cut this,” is much more abstract than even the busing thing, uh, and a hell of a lot more abstract than [N-word, N-word].14 Subsequent research by the Princeton political science professors Martin Gilens and Tali Mendelberg has confirmed the success of the strategy of linking certain discourse to negative racial stereotypes. Their research shows that expressions like “welfare,” “the poor,” “food stamps,” and “homeless” all contribute to priming the thought that Black Americans are lazy.
15 Gilens finds that “the belief that blacks are lazy is the strongest predictor of the perception that welfare recipients are undeserving.”16 There is a large amount of additional evidence that the word “welfare” has been connected with a flawed ideology of race, in addition to the studies Gilens himself has carried out. Gilens reports similar results from the “welfare mother” experiment from the National Race and Politics Study of 1991: Respondents are asked their impressions of a welfare recipient described as either a black or white woman in her early thirties, who has a ten-year-old child and has been on welfare for the past year.
Respondents are first asked how likely it is that the woman described will try hard to find a job, and second, how likely it is that she will have more children in order to get a bigger welfare check.17 The largest predictor of opposition to programs described as “welfare” was one’s bias against black American mothers receiving various state benefits, where the study found that “nonblack respondents with the most negative views of black welfare recipients are 30 points higher in opposition to welfare than are those with the most positive views of black welfare mothers.”18 But why, one might ask, are these facts linguistic? Perhaps we can explain the political effects of describing a term as “welfare” merely by talking about the social programs that are so described, together with false beliefs, including the ones associated with racist ideology. Why are properties of language at issue here?
What fuels Americans’ obsession with programs called “welfare”? Is it background commitments to individual responsibility? Is it Americans’ supposedly fierce opposition to “big government,” in the form of government programs? Is it background racist beliefs and false empirical beliefs about poverty in the United States? Can we explain the political force of describing a program as “welfare” just by discussing the social programs themselves, without discussing the meaning and use of words? Or do we need some explanation that invokes properties of the word “welfare” itself? Americans are fond of, and committed to, what are by far the United States’ largest social welfare programs: Medicare and Social Security.
19 But perhaps the powerful and widespread support for these programs is due to the facts that they “benefit large numbers of Americans of all social classes”20 and that American opposition to programs described as “welfare” has something to do with attitudes toward poverty, specifically? Here, too, the explanation would be nonlinguistic.
In surveys from the 1990s that measure public support for government responsibilities, those that do not use the term “welfare,” or other terms that evoke paradigmatic programs that Americans think of as instances of welfare, we do not find sentiment against a large government role in providing jobs, housing, and other forms of assistance to needy Americans; in fact, as Martin Gilens writes, quite the opposite is true: When asked about spending for the poor, the public again expresses a desire for more, not less, government activity. Over 70 percent of Americans say we are spending too little on “fighting poverty,” while a similar number think spending for the homeless needs to be increased. Smaller numbers—but still majorities—think we are spending too little on “poor people,” on “assistance to the poor,” and on “child care for poor children.” And as was true for education, health care, child care, and the elderly, very few Americans believe spending for the poor should be reduced from current levels.21 In stark contrast, Gilens observes that in those surveys, between 60 and 70 percent of Americans thought that the government was spending too much on programs described as “welfare,” or on programs described as benefiting “people on welfare.” It is impossible to describe political communication in the United States—dating back to the 1970s, when Ronald Reagan’s campaign introduced the expression “welfare queen” into political discourse22—without talking about the connection between such value systems and the linguistic properties of words like “welfare.” In a 2018 article, Rachel Wetts and Robb Willer integrate multiple studies providing strong evidence that the connection between White racial resentment toward Black Americans and negative reactions to programs described as “welfare” continues unabated.23 If it were a matter simply of Americans rejecting “big government programs,” we would find them rejecting large government programs such as Medicare and Social Security, which are designed to help working-class Americans by providing health insurance and support during retirement.
Indeed, when programs described as “welfare” are described in other terms, not involving this vocabulary, they receive far more support than when they are described as “welfare,” even when they are the same programs.
A long-term goal of many in the US Republican Party is to cut funding to even very popular government programs that provide support to needy populations, including the elderly. In pursuit of this political goal, the fact that “welfare” and similar expressions such as “public assistance” give rise to negative reactions among certain audiences has proven too tempting to ignore. On March 13, 2017, then president Donald Trump issued an executive order authorizing Mick Mulvaney, the director of the Office of Management and Budget, to oversee a complete reorganization of federal agencies.24 A draft of Mulvaney’s proposals was floated, “Delivering Government Solutions in the 21st Century: Reform Plan and Reorganization Recommendations.”25 The second proposal listed is “Consolidate Non-Commodity Nutrition Assistance Programs into HHS [Health and Human Services], Rename HHS the Department of Health and Public Welfare, and Establish the Council on Public Assistance.”26 The proposal “moves a number of nutrition assistance programs
… —most notably SNAP and the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC)—to HHS and, acknowledging the addition of these programs to the Agency, renames HHS the Department of Health and Public Welfare.”27 The focus on renaming programs, and bringing more programs that Republicans hope to dismantle under the description “welfare,” suggests a clear recognition that it is the label that does damage.
This explains why the proposal recommends grouping Health and Human Services and food programs that many Americans use under the heading of “welfare,” in an attempt to tie its racial stigma to these programs. “Public assistance” also carries with it racial stigma; appointing a Council on Public Assistance to monitor a vast sweep of government programs connects government spending to the negative racial sentiments that many Americans associate with the words “public assistance.”28 This makes sense as part of a larger mission to dismantle such programs. The Republican Southern Strategy provides a model for political propaganda, to which we shall return, using the campaign against critical race theory that dominates US politics as of the writing of this book as a contemporary example.
Jennifer Saul’s paper “Dogwhistles, Political Manipulation, and Philosophy of Language” is an investigation of the speech act of dog whistling.29 On Saul’s analysis, a dog whistle’s message is a function of the ideology of the audience. The function of using a term like “welfare” to describe a program is to make that program less popular in the minds of those with a racist ideology (such a description will be less pejorative to those who lack a racist ideology). Descriptions of programs as “welfare” or of persons as “on welfare” are paradigm examples of dog whistling in this sense. Describing a program as a “welfare program” gives rise to a strongly negative reaction to that program among one audience (those with at least some racial bias), and considerably less negative reactions among a different audience (composed of members with few indicators of racial bias). Racial bias is a value system; it is a way of valuing things—or, in this case, persons—on a metric of value at least partly determined by race.
Describing something as “welfare” signals one very negative message about it to an audience who endorses a racist value system and lacks this negative force with audiences who do not share that value system.
Saul makes an important distinction between different categories of dog whistles. The category of overt intentional dog whistles is the most straightforward to define, but perhaps least politically central. Kimberly Witten defines an overt intentional dog whistle as a speech act designed, with intent, to allow two plausible interpretations, with one interpretation being a private, coded message targeted for a subset of the general audience, and concealed in such a way that this general audience is unaware of the existence of the second, coded interpretation.30 An overt intentional dog whistle is the clearest example—it is one that works, as the label suggests, overtly. Overt dog whistles are meant to be understood as such by their target audiences.
Saul introduces another category of dog whistles, covert intentional dog whistles.31 Overt dog whistles are meant to be understood as such by their target audiences. Covert intentional dog whistles are not meant to be recognized as delivering hidden messages. An example Saul provides is “inner city”: this expression is meant to be seen as a race-neutral expression, but hearing it triggers negative responses in those disposed to racial bias; something in the vocabulary triggers value systems that involve degrees of racism.32 A covert intentional dog whistle triggers a response, perhaps a negative affective one, in those who share the relevant value system. But it does so surreptitiously. Many or most uses of “welfare” in the context of the United States are covert intentional dog whistles, in Saul’s sense—those on whom they work most effectively do not realize that the dog whistle is having this effect.
In the 1990s, Bill Clinton appropriated the Republican racial rhetoric with his call to “end welfare as we know it,”33 thereby attracting White voters who otherwise would have been loath to vote for a party connected to the attempt to lift Black American citizens to equality, which might be seen as helping “the undeserving.” Demonizing poor Black Americans has been a successful electoral strategy for both the Democrats and the Republicans in the decades following the Civil Rights Movement, and covert racist dog whistles have been central to this practice. Currently, the Republican campaign against critical race theory continues these strategies.
Covert and overt dog whistles function communicatively by drawing on an ideological background. To understand dog whistles, we must incorporate into our theory of speech the ways in which different ideological backgrounds affect what is communicated by a speech act. The concept we will use to explicate dog whistles, as well as some other kinds of hustle, is presupposition. On our analysis, dog whistling functions by presupposing certain ideologies. In part II, we will be developing a detailed theory of presupposition and ideology.
The example of dog whistles brings out this more general feature of hustle—hustling is characteristically dependent on presupposed narratives, ideology, prejudice, values, and frames. A theory of meaning adequate to explaining hustle must develop and elucidate a novel notion of presupposition that could explain how such notions could be presupposed in a way that enables speakers to hustle their audiences.
The task of explaining dog whistling with presupposition faces an immediate objection, one that will help us elucidate early on some of the ways our project rethinks the terrain. Dog whistling is a paradigm of a speech act that allows for plausible deniability. As Justin Khoo has pointed out, this contrasts starkly with standard examples of presupposition, which cannot be plausibly denied.34 For example, “I am picking up my sister” presupposes that the speaker has a sister, and so it would be odd for a speaker to say: (1) I am picking up my sister from the airport, but I do not have a sister.
In contrast, one can say: (2) That program is nothing other than a welfare program, but I don’t mean to suggest anything negative about Black Americans.
The worry is this: if the negative racist message associated with “welfare” is presupposed, then one cannot explain plausible deniability, the very property that a theory of hustle must explicate.35 Responding to this objection helps us, from the beginning, elucidate the centrality of speech practices to our model.
It is familiar from the work of Saul Kripke, among many others, that words are embedded in speech practices, which give those words meaning; according to Kripke, speech practices explain why proper names have the references they do.36 We agree with Kripke on this point, but we think of speech practices as imbuing significance to words that goes well beyond their referential properties. Every time one uses a word, one presupposes (and manifests) a speech practice, one that is connected to a variety of resonances, emotional and otherwise. The word “welfare” belongs to a racist speech practice that casts a negative shadow on anything sodescribed. Using the word in this way presupposes this speech practice. But most words belong to multiple speech practices—and to understand what speech practice its use presupposes, one must often know the social location, point, and purpose of the speaker. In a paper that has deeply affected us, Anne Quaranto argues that dog whistles function by exploiting the presence of multiple speech practices governing a single word.37 In using a dog whistle, one presupposes one speech practice, while taking advantage of the fact that the word can also be used in other ways. If one is challenged, one claims that one was using it in this other way. What’s needed to complete this analysis is an account of presupposition that can make sense of the claim that using a word can presuppose something like a practice. And we need an account of speech practices that explains the resonances of language and the impact language has on us.
The Path Forward There are clear difficulties in making sense of the multifarious ways in which speech impacts audiences in the terms of the philosophical tradition of semantic analysis that dominate analytic philosophy and linguistic semantics. Let us briefly sketch the problem and where it led us. We start with the tradition that forms the background. It runs through Gottlob Frege at the end of the nineteenth century, the early Ludwig Wittgenstein in the first part of the twentieth century and Richard Montague in the 1960s, and onward into what is now a rich, well-articulated, and diverse academic enterprise, that of compositional formal semantics. In this enterprise, meanings of words are understood in terms of the bits of the world they refer to and in terms of functions on those bits, and the bits are composed to calculate what the sentence says about the world. Adherents of this approach, ourselves included, see an austere beauty in the smooth way these meanings can be composed, as if they were physical building blocks engineered to slide into place.
We place early Wittgenstein at the heart of the tradition in which we were trained because the approach we are describing can be seen as a realization of what he termed in the Tractatus38 the picture theory of meaning. On this view, a sentence functions like a panel in the pictorial instructions accompanying a prefabricated furniture kit: an elongated Tshape with a series of slightly diagonal parallel lines at one end depicts a particular type of bolt, a long rectangle depicts a table leg, and the spatial relationship of these elements together with an arrow depicts an action that the assembler of the furniture must perform. The idea is that the conventions of language determine how arbitrary symbols can be mapped onto real-world objects in the way that pictorial elements are mapped onto real-world objects via iconic similarity. The Frege-Montague line of work makes precise how language can represent in this way, but it creates a quandary (a quandary perhaps not unrelated to the evolution seen in Wittgenstein’s own later work): how can a picture theory of meaning like that we have just caricatured possibly help us understand phenomena like harmful speech?
While we will not directly use Wittgenstein’s picture metaphor in presenting the account that these worries eventually led us to, it might be said that we still presuppose a depiction theory of meaning. But don’t think of a construction manual; think of a picture (from the front page of the October 1936 edition of the Nazi propaganda newspaper Der Stürmer) depicting a rich Jew with vampire teeth eating tiny “ordinary” people whole. He has a Star of David on his forehead, in case other aspects of the caricature were insufficient to indicate his identity, and a masonic symbol on his lapel for good measure.39 Or think of Picasso’s Guernica, also expressly created and exhibited to support a political cause. There are certainly pictorial elements in the Guernica that can be mapped onto things and events in the real world: a bull, a horse, faces and grimaces, a broken sword. Yet what makes the painting so rich is not simply the existence of symbols that stand for things. It is the extraordinary way the elements are chosen, portrayed, and composed so as to immediately evoke powerful emotional reactions, and the way they collectively and holistically bring to salience a peculiarly rich web of social and historical associations, of interwoven half-told narratives, and of practices of war and killing.
Although we neither offer nor presume an analysis of artistic representation, what we seek in this book is a theory of how language can evoke similar emotional reactions, social and historical associations, narratives, and practices. Once one begins to look at language in this way, one begins to see even the simpler cases that have been the mainstay of semantic theory in a very different light, such as the relation between “dog” and “cur,” which the logician and philosopher Gottlob Frege used to motivate the notion of meaning at the heart of the formal semantic tradition.
The view we develop in this book will bring out how even the Ikea instruction manual was never a simple static mapping from 2-D representations to the 3-D furniture of the world, but embodied a complex set of consumer-societal, industrial, and constructional practices. So it is, we will argue, with every piece of language that was ever reduced in a class on semantics or philosophy of language to a sequence of logical symbols.
We are not against the practice of performing such formalizations. But we will argue that what must be made precise is not a simple mapping from expressions to things. A conclusion we draw from Wittgenstein’s later work is that what must ultimately be made precise, if we are to understand how meaning functions, is rather a set of language practices and the social conditions accompanying their use.40 We believe that this is as true for the simplest sentence in a learn-to-read storybook as it is for the more complex and subtle ways in which speech mobilizes audiences toward explicitly political action.
The leading ideas of the new framework we develop in this book are as follows. Linguistic actions, such as speaking a word, exemplify social practices, and have resonances by virtue of the practices they exemplify.
The resonances include things, properties, emotions, practices, and social identities—anything that tends to be around when words are used. The resonances always have ideological significance, and sometimes this is obvious, as when a word like “freedom” is used. The function of speech is to attune audiences to each other and to facts of the world, and this attunement occurs via the resonances of what is said. Some resonances concern effects of the linguistic action on the interlocutors, like the gaining of new attunements to the way the world is, or the experience of pain when a slur is hurled at someone. Other resonances are presupposed. Presupposed resonances are an especially important way that hustle creeps into communication; entire ideologies are presupposed and may be slowly accommodated, and yet the presence of ideological presuppositions can escape attention. Harmonization, which is a generalization of accommodation, is an adaptive process by which attunements change in the face of mismatches, for example, ideological mismatches between the attunements of different interlocutors. Harmonization does not always repair mismatches, but it can do so, allowing individuals and groups to coordinate; they may coordinate, for example, on ways of speaking, on ways of treating others, or ways of voting. Thus we study the influence of speech on political action, but with a particular interest in covert aspects of this influence.
Here is the plan of our book: In part I, we introduce the foundational notions of our model. Words are employed in communicative practices, which lend these words resonances. Groups of people form communities of practice, which shape these resonances. This is the topic of chapter 1, which is motivated in terms of political language, but in which the major new development is a general model of meaning as resonance, a model that is not specific as regards its application area. The use of words by a community of practice attunes its members to these resonances. The work of chapter 2 is to motivate and explain how attunement functions within such a community. This is where we start to get more explicit about the machinery required for questions of social and political significance, laying the groundwork for a model in which we can make sense of issues like ideological change and transmission. In chapter 3, we analyze the process by which attunement changes at both an individual and group level, or, equivalently, the way people and groups adapt to each other through communicative interactions.
We refer to this process as harmonization. What we seek is a model of how speech can affect people in the short term, but a model that allows us to make sense of the process by which ideas and ideologies spread and transform over the larger time scales at which political change occurs.
In part II, we use the notions we develop in part I to redefine the central concepts of formal pragmatics, presupposition and accommodation. Presuppositions reflect the background of communicative practices, the things that are normally so evident to interlocutors that their significance need not be made explicit. In justification of a tradition of philosophers pioneered chiefly by Rae Langton, we argue that presupposition plays a special role in ideological transfer. In our terms, this is because people tend to harmonize with presuppositions nondeliberatively. This both reflects the positive role of presupposition in helping people coordinate and build common ground, and introduces a danger, since a propagandist can take advantage of presupposition in order to persuade covertly. In chapter 5, we generalize standard models of presupposition using the notions introduced in part I.
We use this to make sense of the idea that a communicative action can presuppose a practice, so that, for example, telling sexist jokes can presuppose sexist ideologies.
Accommodation refers to the way people adapt to the communicative situation. We suggest in chapter 6 that accommodation be modeled as a special case of harmonization, as introduced in part I.
Accommodation is harmonization to a group, especially to a group with which people identify. This move helps us to understand a range of complex phenomena, such as the processes that undergird political polarization and the formation of echo chambers.
Our model of speech is more realistic than many more standard views in the sense that we aim to avoid certain common idealizations, because we think these idealizations obstruct the analysis of social and political aspects of language. In part III, we step back to look at theoretical issues involving idealization, in particular the issue of how idealizations about speech can serve as ideological distortions. For the sake of perspicuity, we focus on two idealizations standardly made in linguistic and philosophical work on meaning, which we call neutrality and straight talk. We use these to exhibit two different ways in which idealizations characteristically distort. First, they can distort by being incoherent, as we argue in chapter 7 to be the case with the idealization of neutrality. Words are embedded in practices, and as such are vehicles for ideology. There is no such thing, then, as a neutral word in a human language. The pretense of neutrality functions to mask the way speech transmits ideology.
Secondly, idealizations can distort by limiting attention to an unrepresentative subset of language types, as we argue in chapter 8 to be the case with straight talk. In chapter 9, we situate our project within the broader ambit of attempts across philosophy to critique idealizations.
Finally, in part IV, we turn to the question of the power of speech to harm and liberate. How do we theorize these together? Chapter 10 concerns harmful speech, focusing on several different categories, such as slurs, and bureaucratic speech, which harms by objectifying and masking. In our final chapter, we turn to the question of the liberatory potential of speech. How do we best think of free speech in a democracy, given speech’s power to harm? We conclude that arguments against speech restrictions that are based on the democratic ideal of liberty fail. But this does not mean that no at least partial defenses of a free-speech principle are possible—an approach we suggest, cast in terms of maximizing participation in a process of collective harmonization, is to reconfigure the defense of free speech around the other central democratic ideal, that of equality.
1. The quote is from Russell’s introduction to Wittgenstein’s Tractatus Logico-Philosophicus, x.
2. Stalnaker, “On the Representation of Context,” 5.
3. Klemperer, The Language of the Third Reich, 15–16.
4. Thucydides, The Peloponnesian War, 147.
5. Klemperer, The Language of the Third Reich, 154.
6. DiIulio, “My Black Crime Problem, and Ours,” 14.
7. “Reported Violent Crime Rate in the United States from 1990 to 2017,” Statista: The Statistics Portal, October 10, 2022, accessed March 1, 2023, https://www.statista.com/statistics/191219/reported-violent-crime-rate-in-the-usa-since-1990/ .
8. For a contemporary report on this phenomenon, see Templeton, “Superscapegoating,” 13–14.
9. In 1994, Bill Clinton signed the Violent Crime Control and Law Enforcement Act. This included the “Federal Death Penalty Act of 1994,” which created sixty new death-penalty offenses under forty-one federal statutes (Title VI, §§60001–26); the elimination of higher education for inmates (§20411); registration of sex offenders (Title XVII, Subtitle A, §170101); and making gang membership a crime (Title XV, §§150001–9). See: U.S.
Congress, Violent Crime Control and Law Enforcement Act.
10. Cahalan and Parsons, Historical Corrections Statistics, 30.
11. John Gramlich, “America’s Incarceration Rate Falls to Lowest Level since 1995,” Pew Research Center, August 16, 2021, accessed March 1, 2023, https://www.pewresearch.org/facttank/2021/08/16/americas-incarceration-rate-lowest-since1995/. Although the title of the article appears to contradict this claim, the article confirms the United States’ high incarceration rates.
12. Jamiles Lartey, “By the Numbers: U.S. Police Kill More in Days Than Other Countries Do in Years,” Guardian, June 9, 2015, accessed March 1, 2023, https://www.theguardian.com/usnews/2015/jun/09/the-counted-police-killings-us-vs-other-co untries. 13. Laura Glowacki, “9 Reasons Canada’s Crime Rate Is Falling,” Canadian Broadcasting Corporation, July 23, 2016, accessed March 1, 2023, https://www.cbc.ca/news/canada/manitoba/9- reasons-crime-rate-1.3692193.
14. Rick Perlstein, “Exclusive: Lee Atwater’s Infamous 1981 Interview on the Southern Strategy,” The Nation, November 13, 2012, accessed March 1, 2023, https://www.thenation.com/article/archive/exclusive-lee-atwaters-infamous-1981-interview-so uthernstrategy/.
15. See Gilens, Why Americans Hate Welfare; and Mendelberg, The Race Card, 191–208.
16. Gilens, Why Americans Hate Welfare, 95.
17. Gilens, Why Americans Hate Welfare, 97–98.
18. Gilens, Why Americans Hate Welfare, 99.
19. Gilens, Why Americans Hate Welfare, 30.
20. Gilens, Why Americans Hate Welfare, 27.
21. Gilens, Why Americans Hate Welfare, 29.
22. Gene Demby, “The Truth behind the Lies of the Original ‘Welfare Queen,’ ” NPR, December 20, 2013, accessed March 1, 2023, https://www.npr.org/sections/codeswitch/2013/12/20/255819681/the-truth-behind-the-lies-of-t heoriginal-welfare-queen.
23. Wetts and Willer, “Privilege on the Precipice,” 1–30. 24. Executive Office, “Comprehensive Plan for Reorganizing the Executive Branch.” 25. Executive Office, “Delivering Government Solutions in the 21st Century.” 26. Executive Office, “Delivering Government Solutions in the 21st Century,” 27–29.
27. Executive Office, “Delivering Government Solutions in the 21st Century,” 27.
28. Executive Office, “Delivering Government Solutions in the 21st Century,” 27–29.
29. Saul, “Dogwhistles,” 360–83.
30. Witten, “Dogwhistle Politics,” 2, cited in Saul, “Dogwhistles,” 362.
31. Saul, “Dogwhistles,” 364–67.
32. Saul, “Dogwhistles,” 367.
33. Carcasson, “Ending Welfare as We Know It,” 655.
34. Khoo, “Code Words in Political Discourse.” 35. A technical solution to Khoo’s problem is available within the presupposition literature. One could say that while the presupposition in (1) is both presupposed and entailed, the presupposition in (2) is only presupposed, and not entailed. Putative cases of non-entailed presuppositions (which can thus be canceled even when not embedded under logical operators like negation) have been discussed at least as far back as the Gazdar’s work on presupposition (Pragmatics, Implicature, Presupposition, and Logical Form). The analysis of presupposition developed in part II of the book allows this type of analysis, and also allows that presuppositions are probabilistic, so that there are tendencies for them to hold in contexts of utterance rather than absolute requirements.
However, these facets of our account are not what we take to explain the contrast between (1) and (2). We would submit that while simply analyzing “welfare” as having an unentailed presupposition is possible, this would still leave an explanatory gap, since it is not at all clear why these constructions would be associated with unentailed presuppositions while the bulk of what are standardly taken to be presuppositions are entailed.
36. Kripke, Naming and Necessity.
37. Quaranto, “Dog Whistles, Covertly Coded Speech, and the Practices That Enable Them.” 38. Wittgenstein, Tractatus Logico-Philosophicus, 5–12.
39. Images from Der Stürmer, including the one described in the main text, have been collated by Randall Bytwerk. At time of writing, they can be seen at his Calvin University website, https://research.calvin.edu/german-propaganda-archive/sturmer.htm (site verified March 2022).
40. There is throughout Wittgenstein’s Philosophical Investigations a continuous push away from the inner, mental significance of language, and toward the societal practices within which language is used, or, as he would have it, within which language games are played.
The view is crystalized in an extraordinary remark with respect to which perhaps we err by merely mentioning it in a footnote; it might be said that the current volume, like much other philosophical work of the last seventy years, is really the footnote: “For a large class of cases—though not for all—in which we employ the word ‘meaning’ it can be defined thus: the meaning of a word is its use in the language” (20, proposition 43).
PART I How Words Connect People
CHAPTER ONE Resonance You rely on a sentence to say more than the denotation and the connotation; you revel in the smoke that the words send up.
—TONI MORRISON1 Gedanken ohne Inhalt sind leer, Anschauungen ohne Begriffe sind blind. (Thoughts without content are empty, intuitions without concepts are blind.) —IMMANUEL KANT2 Although we speak of utterances carrying meaning just as of a pipeline carrying oil, we know perfectly well that the parallel is faulty. The meaning of an utterance is not in the utterance as the oil is in the pipeline; rather, awareness of the meaning is triggered or induced in the hearer BY the utterance.
—CHARLES HOCKETT3 1.1. Let Freedom Ring We will presumably not be thought to be making a controversial claim in saying that the words “equality” and “freedom” have strong resonances. As Martin Luther King Jr. reprised in his most famous speech, freedom can ring. The resonances are highly emotional, and intertwined with group identity, with what it is to be an American, or, for that matter, French.
The resonances include ideas, ideals, and broader ideologies. And the resonances also include practices, practices of treating people equally and giving them freedom to live life as they see fit, but also speech practices.
These speech practices themselves have many different levels. There are practices of speaking freely and of permitting free speech, and there are practices of treating competing voices equally. More narrowly, there are the very practices of invoking “equality” and “freedom,” which have become mantras. Through repetition, the words have developed a totemic ability to symbolize ideals without any felt need to state exactly what the ideal is.
Liberté! Egalité! Fraternité! Such words, in whatever language, are repeated almost ritualistically by sloganeering politicians.
It is a remarkable fact that when the slave states of the American South seceded in 1860–1861, both sides went to war with the same avowed goals: freedom and equality. The position of the North requires no explanation to a modern audience, but the position of the slave states has come to be recognized as anathema. The four published state declarations of causes, official state documents drawn up to make the case for secession, are jarring.
The Mississippi declaration, for example, says that the hostility to slavery “seeks not to elevate or to support the slave, but to destroy his present condition without providing a better.”4 Looking at this quote now, it makes no more sense than would an attack on a medical charity on the basis that it “seeks not to elevate or to support” cancer patients, “but to destroy [their] present condition without providing a better.” Yet we can only assume that in 1861, it was, for many people, an effective framing.
The Georgia declaration, which also includes an explicitly economic argument, invokes both liberty and equality in its conclusion. It ends by saying this of the Republican Party under Lincoln: Their avowed purpose is to subvert our society and subject us not only to the loss of our property but the destruction of ourselves, our wives, and our children, and the desolation of our homes, our altars, and our firesides. To avoid these evils we resume the powers which our fathers delegated to the Government of the United States, and henceforth will seek new safeguards for our liberty, equality, security, and tranquility.
5 This excerpt from the declaration by state of Texas, a call for preservation of inequality if ever there was one, literally emphasizes equality, in the sense that the emphasis below is in the original: We hold as undeniable truths that the governments of the various States, and of the confederacy itself, were established exclusively by the white race, for themselves and their posterity; that the African race had no agency in their establishment; that they were rightfully held and regarded as an inferior and dependent race, and in that condition only could their existence in this country be rendered beneficial or tolerable.
That in this free government all white men are and of right ought to be entitled to equal civil and political rights; that the servitude of the African race, as existing in these States, is mutually beneficial to both bond and free, and is abundantly authorized and justified by the experience of mankind, and the revealed will of the Almighty Creator, as recognized by all Christian nations; while the destruction of the existing relations between the two races, as advocated by our sectional enemies, would bring inevitable calamities upon both and desolation upon the fifteen slave-holding states.6 Alexander Stephens was the vice president of the Confederacy. His most famous speech, the “Cornerstone” Speech, is equally jarring. Toward the beginning, Stephens declares that “all, under our system, stand upon the same broad principles of perfect equality.” And yet its most famous (or rather infamous) line, only a few paragraphs further, announces that the new government’s “cornerstone rests, upon the great truth that the negro is not equal to the white man.”7 What can explain how such language resonated effectively for many in the antebellum South, persuading them to accept the need for war?
Bolstered by the authority of an explicitly racist constitution and by an implicit version of trickle-down economics in which the mass of poor White farmers hoped that the opportunity to enslave others would eventually trickle down to them, the secessionists undermined the ideals of freedom and equality, in the very act of appealing to them.
Secessionist rhetoric relied on a narrow framing, a reinterpretation of the domain of application of undefined terms involved in widely held but imprecisely stated ideals. First, the ideal of universal freedom was undermined by reinterpreting freedom as freedom of “free” states and, almost vacuously, freedom of free people. Given that this notion of “freedom” included the freedom to own another human being, the freedom of some required the enslavement of others. Second is the ideal of equality.
The founding of the nation eighty-five years earlier had hinged on a “self-evident” truth expressed with powerful simplicity in the Declaration of Independence: “all men are created equal.” But the Constitution pragmatically undermined the ideal of equality, among other ways through its infamously contorted picture of the value of an enslaved person—threefifths of a free man, for the purposes of congressional representation and taxation. In the Texas declaration of causes, we see a completely unequivocal retrenchment of the ideal that had been expressed in such redolent terms in the Declaration of Independence: “all men are created equal” becomes “all white men are and of right ought to be entitled to equal civil and political rights.” The ritualistic use and totemic quality of “freedom” and “equality” are so strong that the underlying concepts and ideals can often be pushed into the background.
Confederate messaging could be seen as persuading people by subverting ideals through the use of conveniently narrow framings. But we should be careful here. While there was clearly an intent on the part of the authors of the declarations to persuade, it is not at all clear whether these authors would have understood their own framings of the notions of freedom and equality as narrow, and it is unclear whether the authors of the four state declarations viewed what they were doing as subversion, or even manipulation.
Some, such as George Fitzhugh, were clear-eyed about endorsing a conception of human freedom that required slavery.
8 For Fitzhugh, freedom for “us” is enslavement for “them.” And it is likely that this route to the appeal of “freedom” still resonates with many, in various forms. Toni Morrison writes, “There is quite a lot of juice to be extracted from plummy reminiscences of ‘individualism’ and ‘freedom’ if the tree upon which such fruit hangs is a black population forced to serve as freedom’s polar opposite.”9 Morrison’s suggestion is that talk of freedom sometimes resonates with appreciation of a life that is in contrast to that of Black Americans, one that requires their general subordination in order to be fully appreciated. Talk of freedom might be intended to do one thing, but unintentionally it resonates with a degraded unfreedom, with which it is unconsciously contrasted.
This distinction, between what someone intends a communicative act to achieve and what it actually does, relates to a central theme in this book: hidden meaning. We introduced the term hustle to cover the hidden meanings of communicative acts, or, to be a little more precise, the effects of a communicative act that are not mutually identifiable by interlocutors as being intended. Hustle covers many subtle aspects of language that, despite that subtlety, are highly efficacious. It includes choices of framing and optics that politicians and spinmeisters trade in, but it includes equally the way in which a mundane, everyday communicative act drags with it an invisible net of ways of being and doing.
On the political side, the word “freedom” carries much hustle, as does the evocative template “Let freedom ring from,” as Martin Luther King rang it out repeatedly in his famous 1963 speech. People are moved by King’s words, and yet most would not be able to say exactly why they are so stirring. Melvin Rogers singles out the power of aspirational democratic rhetoric as a central theme in African-American political thought in his scholarly work on the topic.10 Rogers draws particular attention to examples in which African-American thinkers not only draw on the language of freedom, but also “appeal, for example, to the Declaration of Independence or the American Founders or they figuratively place themselves into the story of Exodus (not uniquely American).”11 We can see the terrain of US politics partly in terms of a struggle over the meanings of the word “freedom,” that is, over its resonances.
A less obviously stirring word, simply because it is so commonplace, is “mom.” Yet what word could be more emotionally and socially significant?
Just as the word “freedom” carries much hustle, so does “mom,” presupposing certain gender roles and family structures, some general to a culture, and some specific to any situation in which it is used.12 More generally, the resonances of words are often hidden in plain sight: our point is not that the emotional and social significance of the word “mom” is some big secret, but that it is often not brought to awareness in ordinary uses of the word.
Our explanation of how a resonance can be hidden (or hidden in plain sight) will be developed gradually. Before we can even begin to explain what it means for a resonance to be hidden, we must motivate and develop our account of what it is for a word to carry resonance at all. This is the job to which we now turn. 1.2. What’s in a Word?
Kant famously described thoughts as having “contents,” as if thoughts are containers and sitting inside the container there is some sort of objectively real thing waiting to be unpacked.13 This has been a fruitful metaphor. It is the basis for Frege’s analysis of meaning, which is now commonly seen as pivotal in the Western analytic tradition of philosophy of language, as well as laying the foundations for the most formally well-developed accounts of meaning in contemporary theoretical linguistics. By now, the idea of words containing meaning or information is standard in philosophy of language, linguistics, and beyond. Let’s call the general idea that communication consists of conveying meaning inside container-like vessels consisting of symbols, such that the speaker’s job is to wrap the meaning up and the hearer’s job is to unwrap it, the content-delivery model of meaning. This is a special case of a more general (and much discussed) metaphor for communication, the conduit metaphor, to be considered below. In this chapter, we suggest a way to change the focus in theory of communicative meaning, away from the idea of meaning as an object, a lump of special abstract stuff that might be contained inside a word or phrase, and toward the idea of meaning as a process that connects people to the world and to each other.
The idea that the function of words might be to connect people is not new. It was made explicit a century ago by the anthropologist Bronislaw Malinowski in describing a type of discourse he terms “Phatic Communion.” Malinowski is completely explicit that the function of this discourse is not to transmit information about the world but (like the gift exchanges he famously described elsewhere) to increase social cohesion: Are words in Phatic Communion used primarily to convey meaning, the meaning which is symbolically theirs? Certainly not!
They fulfil a social function and that is their principal aim, but they are neither the result of intellectual reflection.… We may say that language does not function here as a means of transmission of thought.… Each utterance is an act serving the direct aim of binding hearer to speaker by a tie of some social sentiment or other.
14 Despite this long history, it remains far from obvious how best to construct a theory in which linguistic meanings can serve the function of connecting people, at least in the strong sense we aim for whereby the connection is sufficiently strong that it can support both collective action and mutual understanding. Further complicating matters, the political interest in language derives in part from the fact that language not only connects, but also divides, setting people apart into groups that lack common understanding. So the task we have set ourselves is a large one.
Indeed, to the extent that the essentially Fregean content-delivery model still marks the center of gravity for current work on linguistic meaning, our project is not only large, but also radical. However, it is not without significant antecedents. The development of our positive view will rest heavily on prior work drawn from a range of disciplines, including sociology and psychology, in addition to linguistics and philosophy.
15 We began the chapter with an example, the resonances of a particular set of words in discourse related to freedom and equality. We now present what we hope is an intuitive picture of why it might be useful to reframe the theory of meaning so that it is not seen as content-delivery, and instead use a metaphor of resonance. To develop the account of resonance, we adapt an idea of David Kaplan’s that provides a way of assigning meanings to expressive language in terms of usage conditions. We argue that at base, and leaving aside sociocultural effects to which we turn later, we can gain purchase on expressive meaning by considering the simplest type of conventionalized signaling possible.
The latter part of this chapter analyzes resonance in simple artificial practices, involving a single signal that is associated with a single canonical response. These cases serve as a microcosm of the broader theory of communication we will develop, a seedling of resonance from which to grow a broader model of ideological attunement. The examples are paradigmatic because they are interactional and associative. Meaning is interactional when functional success of a signaling action is measured in terms of patterns of reaction to it, and when the appropriateness of those actions depends on conformity to practice, rather than centering on questions of truth or descriptive adequacy. Meaning is associative in the examples we will develop in the sense that it involves tendencies for certain actions to co-occur with other actions or with states of the world, rather than involving an absolute one-to-one correspondence between signals and things that the signals represent. In the last two sections of the chapter, we develop this idea of simple associative and interactional meaning using standard probabilistic and game-theoretic tools, which allow us to characterize the meaning of expressive acts in terms of their role within a simple practice.
Our broad project requires a move away from the standard contentdelivery model of language that focuses on exchange of a single proposition between two individuals, and toward an account that can explain effects seen in mass communication, effects that include mass emotion and masscoordinated action. The way that the complex resonances of terms like “freedom” develop through the ages must ultimately be set within such an account, but it is far from obvious how this might be achieved. So before launching into the choppy waters of freedom and equality, let us briefly situate the account of resonance we develop in this chapter within the broader framework developed in this first part of the book.
First, anything can have resonances of any sort. A bowl of soup can carry a resonance of comfort, and a boot can have resonances of socks and feet, of a quaint little cobbler’s shop, and simultaneously of oppression and authoritarianism; Orwell’s “boot stamping on a human face—forever”16 both adapted and modified the resonances of boots—forever. Of most interest to us in this book are the resonances of words, not boots, but to the extent that a boot is symbolic, we hope that the theory covers boots too.
Second, the resonances of things depend on the attunements of people, which include associations and understandings, the emotions that things conjure up, and the ways people tend to interact with or use things. We are particularly interested in the role that the resonances of communicative devices play in shaping the attunements of individuals and groups. We suggest that to understand this process we must look not at particular attunements people have to this or that discursive practice or feature of the world, but at systems of attunements. Our thesis is that the way people’s attunements change, and the way that propagandists change them, depends on people’s need for a type of coherence.
This takes us to a third notion, harmony. Here we follow in the footsteps of the social psychologist Leon Festinger, who prefigured our approach sixty-five years ago in his account of cognitive dissonance.17 We adopt Festinger’s hypothesis that change in attunements (or what he termed “cognitions”) is often driven by a need to dispel dissonance. We refer to processes of re-attunement that act at the level of complete systems of attunement as harmonization. Our thesis is that harmonization is central to the function of political propaganda. But it is not the only type of reattunement in our model. We also assume a more basic type of reattunement, a direct and immediate nondeliberative response to a stimulus. For the purposes of this book, the type of stimulus of greatest interest is communicative acts: someone shouts at you, and you get upset, and perhaps you also come to change your beliefs, now being attuned to the world in a way that either is or is not more consonant with the person doing the shouting. It is characteristic of nondeliberative emotional response to speech that it happens without necessarily resulting in increased harmony.
For example, we might come to believe something that someone tells us, and only slowly resolve tensions that result because the new belief has left us in a state that is in some way incoherent. Perhaps the update has left us with beliefs that are inconsistent with each other, or with beliefs that are inconsistent with our dispositions. Nondeliberative uptake has produced an incoherent state, and harmonization is then a slower process that, with luck, will pick up the pieces and bring us into a state that is more coherent.
We suggest that thinking in terms of the harmonization of attunements has consequences both at the level of individuals and at every larger level of human organization and activity, from cafe conversation to genocide.
Effectiveness of political messaging depends not on logical consistency but on emotional, cognitive, and dispositional resonances, drawing especially on people’s need to align their attunements with those of their in-group.
Indeed, although the terminology may be unfamiliar, we do not think that the claim should be deeply controversial. What is needed is a proper understanding of the notions of resonance and attunement, so as to found claims like this theoretically. We begin, in this chapter, with resonance, and now turn to some paradigmatic examples of the role that resonance plays in political language and thought.
As Michael Reddy observed,18 people pervasively frame their talk about language using a metaphor of transfer with a physical container or carrier, the contents of the container corresponding to the informational content of an utterance. Speakers put their thoughts into words. While insincerity may mean the words are hollow or empty, speakers may pack or cram many ideas into few words, even filling them with more than they can carry. They may give the hearer an idea of what they mean, or push a challenging idea, hoping they will get their thoughts across to a speaker who will unpack them, extract meaning from them, unless the words are hard to unpack or impenetrable, in which case the hearer might just get the general drift, and might not see where it’s coming from. There is always a danger of hearers reading something into our words that wasn’t there, but then again, perhaps the words were loaded intentionally, freighted with unseemly baggage.
Using this “conduit” metaphor, as Reddy named it, to talk about figurative language itself, we might say that the physical-transfer framing is not merely a way to package discussion of communication, but is baked into the meaning of the vocabulary itself. Going still further, metaphor theorists, including not only Reddy but also George Lakoff and Mark Johnson and their many followers,19 suggest that people don’t merely talk about words as containers of meaning, but have come to conceive of communication by analogy to the physical transfer of goods.
Theoretical linguists and philosophers of language perhaps use the conduit metaphor in a more limited way than other people, at least when they’re theorizing about language rather than talking informally about what someone has said. Probably one reason for that is that prevailing semantic theories suggest that meaning is context dependent in a way that doesn’t naturally cohere with that metaphor. An object in a package is shielded from the context around the package: the whole point of packaging is to allow the object to get from one place to another without change. By contrast, contemporary theories suggest that the meaning of particular words is mutable and evanescent. The meaning of a given expression depends on who utters it, and the time and place of the utterance. In cases of so-called presupposition failure, the meaning may not be recoverable by someone who does not share the assumptions of the speaker. Yet even semantic theorists cannot easily resist the containment metaphor. Philosophers talk, after all, of words as meaning-bearing, and of meaning as content. We also use such Kantian phrases as mental content, a phrase reminiscent of the more colloquial notion of having an idea in mind, as if meanings and ideas could be contained not only inside words, but inside heads.
The word content as used in technical talk about meaning has become somewhat divorced from the notion of containment, but it is well to remember that talk of words as something like boxes with content or trays that bear meaning might easily lead any of us quite unconsciously into a simplistic way of thinking, whereby meanings are treated as fixed entities that can be carried in or by an expression. The simplifications invited by the metaphor can be thought of in terms of the idealizations about meaning. On a natural understanding of the metaphor, the speaker’s intention is to give the hearer what is in the box, and the hearer recognizes that intention as soon as they open it, so one is inclined to view the exchange as following an idealization of transparency.
20 Similarly, while we might imagine that context provides tools that help us in packing or unpacking the container, it does not change what’s in there, and if communication is conveying an object in a package, then anyone who has the tools to open the package will find the same thing inside. That suggests an idealization of shared context, there being no relevant variation of context across interactants, and an idealization of language homogeneity, marginalizing cases where people have different understandings of meaning because of linguistic variation within a speech community. Relatedly, the metaphor draws attention away from who is doing the giving and receiving, thus suggesting an idealization of social homogeneity of the linguistic community, whereby we don’t have to consider issues like social roles in the study of meaning. (All of these idealizations will be discussed in chapter 9.) The content-delivery model suggests that conversation can be analyzed in terms of neat exchange of discrete units of information, inviting us to think of individual acts that are complete in themselves, like the act of producing a sentential utterance. It thus defocuses from discourse effects (the idealization we will term extent), and suggests that we convey one content at a time with each sentential utterance, and the content is in some sense a complete unit of information within itself (the idealization we call propositionality). To be clear, we do think that the notion of content has played a useful role in helping people think about what language does, but we also think that it has helped create an unresolved tension in work on meaning. The very theorists who have recognized the incompleteness of meanings divorced from context have nonetheless strived to isolate those meanings like laboratory samples on a microscope slide. We propose that, instead of treating meanings as objects, albeit with little holes in them where context should plug in, we should focus on meaning as a connection between agents and the processes that lead to it. There is no such thing as an objectively true or correct metaphor (which is not to say that metaphors cannot be used to make true statements).
Rather, some metaphors are helpful, some less so, and some lead to productive new ways of thinking. In this latter case, they are, as Donald Schön put it, generative, 21 in the sense that they spur creativity and generate new insights. The concept of transfer of packaged goods is so basic to our cultural understanding that it makes for an effective metaphor, redolent with associations. But an alternative, physical analogy is more apt for the view we develop. We will use another classic concept, one that entered the scientific vernacular with Galileo’s study of pendulums, and say that meaning is resonance. So meaning is not an object but a process, a process of energetic alignment. The resonance metaphor is clearly related to the conduit metaphor in that it involves a type of exchange or transmission. But to the extent that anything is physically transferred, it is, on this view, intrinsically less tangible and discrete than the standard conduit metaphor might suggest. Don’t think of information in packets delivered to your door, but as flowing into you like the energy of heavy bass music vibrating your core. And don’t think of meaning as an object being transferred, but as a state being shared, so that although resonance requires a physical connection, it may continue even in the absence of energy transfer.
22 Words resonate, interlocutors becoming attuned to the world and each other. What you say may ring true, or at least ring a bell, but it might also strike a chord, tug on heartstrings, and produce a ripple of laughter or applause, or a groundswell of enthusiasm. We get with the vibe. Stirred by tender words, two hearts may, for a while, beat as one. Amid the fluxes and flows of conversation, your words may be pitch perfect one moment, striking the right note, or even striking a chord, but hitting a discordant note the next, so that disharmony reigns. In the fierce counterpoint of argument, a crescendo of matched voices reaches a fever pitch. Then one sharp word may create a shock wave that reverberates through the community, or through the ages, yielding a tsunami of anger. Modern marketing is all about buzz, about having your finger on the pulse, while in the political theater, a campaign and its leaders act in concert to produce a movement of followers marching to the same beat. The political piper plays a tune and we dance like puppets; the drummers take us off to war, and in the discordant cacophony of contemporary culture, where it’s hard to distinguish signal from noise, the propagandist can drown out voices of reason, and drive us into an echo chamber divorced of resonance with reality.
All of this is a long way from the tidy metaphor that popped meaning into a box. There is, to answer the question that heads this section, nothing in a word. Words are not like boxes, or pipes, or train carriages. Meanings are neither freight nor passenger. The resonance metaphor suggests that rather than being a “thing” inside or carried by another “thing,” meaning is a process, a process that sustains an informational connection.23 A word only carries meaning in something like the sense that Galileo’s pendulums carried a beat.
On the conduit view, a semanticist’s job would be no harder than delicately cutting off packaging to shine light on the content inside, but meanings, if we are right, are not the sorts of things that sit still while one tries to cast even metaphorical light on them. That is why the resonance metaphor is helpful to our project at this point. As we will discuss in subsequent chapters, idealizations that tend to come along with the metaphor that meanings are object-like are subtle and pervasive. It is far from clear what would constitute firm evidence for or against the contentdelivery model of communication that accompanies the conduit metaphor, but it is problematically suggestive. We believe that the content-delivery aspect of the conduit metaphor sets traps.24 It suggests, for example, that conventions relate terms only to contents that are in some sense neutrally shared. That conventions link words only to neutrally shared contents is not something to take for granted.
1.3. The Scholarship of Resonance The idea of switching to a resonance metaphor is not new. For example, Keith Stenning, Alex Lascarides, and Jo Calder write that the “analogy of resonance is an important antidote to thinking that communication is merely about the transport of ideas.”25 We now discuss the work of some of the many prior scholars who advocate this antidote, inspiring our own use and development of the resonance metaphor.
There is much we have learned from this prior work, but let us observe right away that our project is unlike any we know of in sociology, political science, linguistics, or philosophy (although we know of little philosophical work that uses a resonance metaphor at all). For in prior work, it is at least tacitly assumed that there is, in addition to resonance, some separate account of semantics or symbolic representation. By contrast, we seek to inject the “antidote” directly into the veins of the theory of linguistic meaning. Unlike any of the prior scholars whose work we will describe in this section, we propose analyzing all linguistic meaning as resonance, everything from the meaning of a morpheme to the meaning of a monograph, and we then use our account of meaning as resonance as the core element of our model of how language mediates the connections between people.
It is perhaps ironic that scholarship on resonance appears to have bloomed independently in a number of academic fields, with little apparent connection between these developments—a resonance zeitgeist spiriting itself into the fragmented landscape of contemporary humanities and social science. One large body of prior work is found in sociology, and in the borderlands between sociology and political science. This work is of direct relevance to the current project not merely because it uses the same physical metaphor, but because it concerns the same social and political issues. Prior authors in these areas have usually sought to account for what sorts of messaging resonates or creates resonance in social and political contexts. Resonance is explained at a pragmatic level, as perhaps involving holistic properties of a text. Sociological and political work on resonance barely touches on what linguistic meaning would have to be like to produce such effects. The same can be said of prior work on resonance in the field of linguistics: there is much discussion of how people use language to create emotional and social effects, and provide reinforcement or emphasis of the underlying linguistic meaning. The existence of an underlying linguistic meaning is simply assumed.
Let us turn first to work on the sociology of message framing, where resonance is a completely standard notion, especially in work on protest movements. In the protest movement literature, springing from work of Robert Benford and David Snow, the question is, What makes a frame resonate?26 For example, why would framing a message one way rather than another be more likely to induce someone to become active in a protest movement or attend a particular rally? Although highly relevant to our project, the questions that Benford and Snow and followers ask is a slightly different one from the one we will ask in this chapter. For the moment, we will not ask why a communicative action resonates with someone, but rather what the resonances of that action are. Put differently, the framing literature tends to focus on why something is meaningful to someone, whereas we are considering the related question: what is the meaning? Here we must beg the indulgence of readers interested in questions raised by prior framing literature, for we will only return to the question of why a communicative action resonates with someone in chapter 3, after extensive further theoretical development.
Of particular importance to our project is Deva Woodly’s analysis of the resonance of frames, which she applies to the messaging of political movements seeking to gain political acceptance.27 Although it is set broadly in a tradition of political-frame analysis within which Benford and Snow’s work is central, Woodly’s focus is distinctive. Much work in that tradition of political-movement analysis centers on an empirical methodology that involves taxonomizing so-called collective action frames as regards what effects they seek to accomplish. Woodly’s focus is more on the discursive features of arguments that make them resonate, which she achieves by leaning into a much older tradition, that of Aristotelian rhetoric, and by borrowing from twentieth-century scholarship in sociology and philosophy; in so doing, she argues for an analysis of what makes an argument resonate in terms of the way it is historically embedded in the culture of those it seeks to persuade. Woodly’s arguments are grounded in extensive analysis of the trials and tribulations of two protest movements in the United States: the movements for equality of marriage rights and for minimum-wage laws. She demonstrates that the relative success of the marriage-rights movement in the United States correlates with the use of frames that in her sense are highly resonant. This means that arguments within these frames have just the right combination of new and old: new ways of looking at protest issues that work because they fit in well with preexisting ways of looking at core cultural concepts and practices. Woodly’s work provides strong motivation for developing a better understanding of resonance, because she argues that proper deployment of resonant arguments can allow nonmainstream (“challenger”) movements to create the conditions for political change. Woodly explains her project as follows: “I am not making the argument that successful movements are those that take advantage of serendipitous cultural resonances between their issue(s) and the prevailing ethos of a particular time. Instead, my claim is that movements can actually change the politics surrounding their issue through the disciplined use of resonant arguments over time.”28 Let us give an example, adapting somewhat from her work: two highly resonant arguments used for same-sex marriage in the United States, a romantic argument and a stable monogamy argument. She has found that these arguments frequently co-occur in her sample of newspaper articles.
The first uses the premise that the basis of marriage is love, and requires the additional, possibly unmentioned, premise that love is not specifically heterosexual. The second uses the premise that marriage promotes stability of relationships and monogamy, and requires the possibly unmentioned premise that stability of relationships and monogamy are desirable objectives independently of whether the relationship is heterosexual.
For each of the two arguments, the first premise is widely accepted within the background culture of US politics, and so doesn’t need to be argued for. The second premises would seem to be more disagreeable to opponents of same-sex marriage. Yet it’s also true that in both cases the second premise is a straightforward generalization from themes that are easily comprehensible even to opposing groups. It is in the nature of romantic love as it has been understood for millennia that it sometimes transgresses normative boundaries, and while opponents of gay marriage might choose to push the argument that homosexual love is itself in some sense unnatural, it is hard to even push the argument without at the same time recognizing and understanding the possibility of such love. For the stable-monogamy argument, again even opponents of same-sex marriage cannot help but understand the values of stability and monogamy, since those are two central values attached to their own conservative view of marriage. It would be rhetorically awkward for conservative advocates of the restriction of marriage to heterosexual couples to offer an argument that stability and monogamy are inappropriate for same-sex relationships. More generally, what makes both arguments resonate is that they dig heavily into existing cultural schemas, giving them just enough twist to offer people a new way of looking at a problem, but not so much of a tug that they become unmoored.
Woodly’s project, then, is to identify what it is about arguments that make them resonate and to show why this matters.29 Our project overlaps with hers, but there are major methodological differences. The most glaring difference is that while we start from the way language functions and apply our model of language functioning to political examples, her work starts from deep analysis of political movements and tries to answer the question of how language functions in the cases she observes. Although coming from opposite directions, we see confluence in our projects. Woodly talks about some arguments having resonances and others lacking them, but she doesn’t extend this terminology to talk of argument X having resonances Y. For example, it’s at least consistent with her account that the resonances of the romantic argument for marriage equality might be said to include nineteenth- and twentieth-century depictions in popular culture of romance culminating in marriage, as well as the strengthening practice through this period of people marrying for love. The framework we develop does allow us to talk of X having resonances Y, although we generalize such that X could be any communicative action and not just (the action of putting forward) an argument, and such that Y could be any feature of context, whether a style of depiction, a practice, a narrative, or a particular type of object (say gold rings, wedding dresses, and oversized cakes). When she talks of an argument being resonant, that translates for us, in the terms we define in the coming two chapters, as an argument harmonizing with the collective attunements of a certain community, or, more narrowly, as an argument harmonizing in a consonant way with that community.
Much work using resonance relates more closely to the development in the next chapter of the notion of attunement. In our terminology, an action has resonances, and people can be attuned to those resonances, as well as to all sorts of other things. Robert St. Clair’s use of resonance perhaps falls into this category, as seen when he says, “Chants that accompany rituals … are sources of harmony … and are similar to vibrating tuning forks that beg others to join them in resonance.”30 As we understand his terminology, it is people who can be “in resonance.” In an influential sociological account of human relationships, what it is to lead a good life in the face of widespread alienation, and what this means for society, Hartmut Rosa also takes resonance as the central concept.31 Here there is a significant difference in terms of goals, since in the current work we do not aim for a normative picture of how life should be lived, but for an account of how communication works. While there is much overlap, and while we draw from many of the same scholarly wellsprings as Rosa, the metaphor of resonance is also applied differently than in our work. For Rosa, resonant experiences connect different parts of a person’s world, in a way that may be transformative. Rosa’s (several) resonance relations are closely related to what we will term emotional attunements, involving affective relationships between people, and affective relationships between people and actions or things.
Given the difference in goals, we do not attempt here to do justice either to Rosa’s work or to the growing literature it has seeded. That literature develops, critiques, and applies his account, especially in making sense of strongly resonant experiences that may center on a person’s encounters with individuals who have had strikingly different life experiences that yet interlock with one’s own lived history, or on encounters with a wide range of cultural artefacts whether movies, musical works, or museums. In a recent book-length treatment of Rosa’s notion of resonance, Mathijs Peters and Bareez Majid observe that although resonance can be understood at large scales of time and culture, “it is not coincidence that … [they focus] on particular experiences that are vulnerable and brief.”32 They do not see Rosa’s notion of resonance as being able in and of itself to sustain analysis of broad societal structures (or to provide a normative basis for such structures), but rather see larger-scale political and cultural currents as being reflected in the momentary experiences of resonance that are their central concern. Here there is a clear contrast with what we attempt in this volume.
While we will, in section 3.9, consider what can make cultural objects resonate for a person, and while we will also, in section 6.4, proffer a related definition of what makes something meaningful to someone, our own focus is on developing a notion of resonance that can underpin communicative practice more generally, and at a societal level. For us, resonance is the basis of all communicative meaning, and does not come into play only during striking emotional or social experiences. Within linguistics, John Du Bois has developed an important account of “dialogical resonance” centering on the way that speakers create connections within and across discourse, especially in spontaneous speech.
The ensuing secondary literature builds not only on Du Bois’s theoretical account, but also on the empirical methodology he developed for identifying resonance within spoken text, a methodology broadly set in the tradition of conversation analysis. Du Bois gives the following definition: Resonance is defined as the catalytic activation of affinities across utterances.
Resonance is a property of relations between elements in discourse; as such it cannot be attributed to any element in isolation.
It represents a developing process of activation and elaboration of certain aspects of the perceived relationship between comparable linguistic elements.33 Although Du Bois takes resonance to operate also at the level of systematic affinities people have for using the linguistic structures found in their social environment, his primary focus, and that of many following scholars, is on relatively short-term processes, whereby adjacent or nearby segments of text resonances depend on activation of working memory.
34 While Du Bois is careful to distinguish his notion of resonance from simple textual parallelism, he takes the notions to be intertwined, and paradigmatic examples of dialogical resonance involve clear repetition of discourse segments, often just a few seconds apart.
While we will consider short-term attentional processes in this volume, there is a significant, and, we hope complementary, difference of focus in this regard. As we have emphasized repeatedly in this section, the ambit of our model is set at a broader level of communicative practice and long-term processes of ideological change.
A number of authors explicitly use the terminology of attunement, a philosophical example being the Wittgenstein scholar Stanley Cavell.35 Cavell takes attunement to be a relationship only between people, whereas we will develop a broader notion that allows people to be jointly attuned to practices and all sorts of other things. Approaches to attunement like Cavell’s, whereby attunement is a relationship between people, are found in a rich line of work developed especially by the sociologists Thomas Scheff and Randall Collins.36 Again, we can think of the contrast between prior work and ours in terms of different questions we will seek to answer. The question posed in this line of sociological work is, simplifying massively, Who is someone attuned to (and why)? We, instead, will ask, What are people attuned to (and why)? And most importantly, our answer will be that one of the most important attunements people have is to the resonances of language.
To round out our discussion of the prior work on resonance on which we build, let us observe that one can think of what we are trying to achieve in this volume in terms of the program of the early behaviorist psychologist Ivan Pavlov. In his most famous experiments, he used a bell to activate his dog’s appetite. As it happens, Pavlov experimented primarily not with a bell but with a metronome, which he found to be a more easily manipulable stimulus. Thus, the first behaviorist study of conditioning was a demonstration that autonomic salivary response can resonate with a pendulum, harkening back to Galileo’s demonstration of how pendulums resonate with each other. More generally, the conditioning experiments of the behaviorist psychologists amount to a program of study of the development of what might be termed behavioral attunements animals have for the resonances of experimental and natural stimuli. We will not restrict ourselves, as the later behaviorists did, to the study of behavior independent of postulation of mental state, but we do take inspiration from the now highly unfashionable behaviorist paradigm. Like the behaviorists (and many others, especially in anthropology and sociology), we think that there is value in considering the actions of people independently of their conscious intentions.37 1.4. Associative Resonance How could it be that something is meaningful but lacks content, or at least lacks anything for which the content metaphor is appropriate? Both as an intuition pump as a deeply problematic phenomenon in its own right, expressive language is a natural place to start investigating this question. It is somewhat obvious that ribald fuck and scatological shit, in their expletive uses, function partly by virtue of incongruous resonances, and it is hard to say what “ordinary content” could be attributed to either. If such epithets are content-laden boxes at all, then they are booby-trapped to explode on opening. In the early and mid-twentieth century, philosophers studying expressives, indexicals, and performative utterances concluded that the study of natural language required different tools and resources than were available in “the formalist tradition,” as David Kaplan calls it in his unpublished talk, “Ouch and Oops”: Within philosophy and especially 20th-century philosophy, there are two great traditions of semantic theory, one a formalist tradition in which the great figures are all logicians: Frege, Russell, Tarski, Carnap, Church, and Kripke, and the other an anti-formalist tradition, in which the great figures are Wittgenstein, Strawson, Austin, and Grice.… The formalists for the most part study the idealized languages of science; the anti-formalists studied natural language, especially its context-sensitivity. It’s from Wittgenstein that the slogan “Meaning is use” is derived.38 The idealized artificial languages of science have limited indexicality, typically indexing the subject matter rather than the speaker; for example, they pertain to the structure of an argument, or the values of unknown objects through the use of explicit variables. They do not have slurs or expressives, albeit that one might detect a note of triumph in a logician’s “QED.” We do not find correlates of “I,” “limey,” or “wow” in, say, the language of calculus. Both formalists and informalists, historically, agreed that natural languages demanded a different approach to meaning than idealized languages of science. In his classic paper, “Demonstratives,” Kaplan showed that one could add indexical and demonstrative expressions to a formal language (for example, to the predicate calculus) and provide a characterization of logical consequence for such languages. Thus, Kaplan showed that indexicals were not, after all, a barrier to the formalist tradition.
In “Ouch and Oops,” Kaplan extends this project to expressives, including slurs. Kaplan suggests that expressive interjections like those of his talk’s title do not have ordinary truth conditions, but rather usage conditions.
39 Thus, “damn” is a word a speaker uses as an adjectival modifier in a nominal when the following condition is met: the speaker is negatively disposed to the referent of the nominal.
“Oops” is something you say when you’ve observed, in Kaplan’s words, “a minor mishap.” These usage conditions could be modeled as sets of contexts—the usage conditions for “ouch” are the set of contexts in which the speaker is in pain.
Kaplan uses this to sketch a (nonstandard) account of logical consequence for expressives.
In his talk, Kaplan draws a distinction between descriptives and expressives—descriptives describe, whereas expressives express their contents. The content of an expressive is generally about the state of a speaker; for example, “ouch” expresses that the speaker is in pain. Kaplan relates his discussion of expressions of pain to Wittgenstein’s, and this striking passage from the Philosophical Investigations makes clear the import of Wittgenstein’s work both to Kaplan’s project and to ours: Here is one possibility: words are connected with the primitive, the natural, expressions of the sensation and used in their place. A child has hurt himself and he cries; and then adults talk to him and teach him exclamations and, later, sentences. They teach the child new pain-behaviour. “So you are saying that the word ‘pain’ really means crying?”— On the contrary: the verbal expression of pain replaces crying and does not describe it.40 Wittgenstein’s suggestion that an expression of pain replaces crying is consonant with our theme that whatever is needed to analyze such language, it is not a standard notion of content, for it would seem at the very least somewhat contrived to describe crying as having content.
Just as Kaplan’s “Demonstratives” can be seen as an attempt to expand the formalist tradition to indexicals, “Ouch and Oops” can be seen as an attempt to expand the formalist tradition to account for expressives. Here is Kaplan’s analysis of “goodbye”: I take it that the expressive (in this case, the word goodbye) has as its expressive content something like “You and I are now parting from one another for a significant period of time.” Probably it doesn’t any longer contain the elements of well-wishing, you know, the etymology of goodbye is from “God be with you,” and probably well-wishing isn’t any part of the expressive content any longer because I think nowadays if someone said goodbye and plainly did not wish the person well, you wouldn’t think that that was an insincere use of goodbye, right? Goodbye seems perfectly acceptable to be used in that situation. So that’s the expressive content.41 Kaplan says that the contemporary expressive content of “goodbye” corresponds to a proposition about the interlocutors’ imminent separation.
In contrasting, as Kaplan does, the descriptive meaning of an expression with its use, he says relatively little about what those uses might be. As against a “normal declarative sentence,” which “describes something,” he says, “Let us call an expression an expressive if it expresses or displays something which either is or is not the case.”42 This suggests that an expressive may do one of two things: expressing or displaying, although it’s not clear whether he intends any opposition between these two. However, any further hint as to what expressives might do is lost in Kaplan’s presentation, since he goes on to analyze the semantic contribution of an expressive only in terms of the contexts in which it can be used, and not in terms of what its use is: Now I claim that “ouch” is an expressive that is used to express the content that the agent is in pain, so what is the semantic information on this kind of model-theoretic analysis (if it is that)? The semantic information in the word “ouch” is represented by the set of those contexts at which the word “ouch” is expressively correct, namely, the set of those contexts at which the agent is in pain. That set of contexts represents the semantic information contained in the word “ouch.”43 So, the meaning of an expressive, according to Kaplan, is a set of contexts.
Although Kaplan himself does not describe it in such terms, one might reasonably think that his usage conditions are analogous to Austin’s preconditions on a speech act, the conditions that must prevail in order for someone to perform a licit act, such as sentencing someone in a court of law. So what Kaplan is discussing under the name of use conditions is what must prevail for someone to say “Ouch!,” or at least what does prevail when people thus vocalize and they’re doing so for appropriate reasons.
It is immediately clear that there is much more to say. Establishing that preconditions hold is not the most important part of what the various speech acts discussed by Austin do, and it doesn’t seem to exhaust what some of the expressions Kaplan takes to be expressives do.
Maybe saying “Ouch!” establishes that you’re in pain, but analyzing greetings and farewells solely in terms of preconditions would leave too much unexplained. Why do we typically exchange greetings and farewells rather than simply having one person say “Hello!” or “Goodbye!” and why does a meeting bereft of these niceties seem so lacking, even when it is quite clear to everyone exactly when the meeting started and when it ended? Establishing the preconditions of individual actions of saying “Hello!” and “Goodbye!” might elucidate the social practices that these words play roles in, and elucidate the role that rituals of greeting and farewell play in establishing social connection, but it tells us very little about why those practices exist. It’s as if we were to say that the significance of putting on your seat belt is that it is done after getting into a car, which would give it essentially the same significance as starting the engine. This would fail to clarify that belting up and starting the engine have totally different functions, and would fail to identify the significance of either within the broader practice of driving.
The view we develop in this book suggests that people taking part in a greeting ritual are not merely signaling the state they are in, but are cocreating a state, a state of affiliation as an interactional group, a state of joint attunement to common protocols, and a state of collective harmony.
We can put it this way: greeting is not merely something an individual does when in a state of readiness to enter into an interactional relationship in order to signal that state. Rather, a relationship consists of connections between people, connections consist in joint behaviors and attunements, and greeting rituals are joint behaviors that constitute part of such a relationship. We should not look at the resonances of “hello” as merely reflections of a state that someone is in. Rather, we should look more holistically at the resonances of entire greeting rituals as part of an emergent cocreated state of interactional group membership.
Kaplan’s theory of expressives does, however, give us a starting point.
The starting point is the idea that expressive communicative actions reflect a preexisting state, that reflection being a first example of what we will term the resonances of an action.
We will, for the purposes of this chapter, make a simplification and focus on what could be termed correlational or associative resonance, whereby things have resonance with each other if they tend to co-occur.
Resonance can be dependent on considerably more complex relations of similarity or narrative fit, as we shall later see. But we will simplify and idealize dramatically here and define what it means for an action of a given type to have associative resonances with a feature of context in terms of the propensity of the feature to co-occur with a token of the action. The types of actions that are relevant are communicative practices, and what we are doing here can be seen as characterizing the abstract meaning of a communicative practice in terms of what the context tends to be like when the practice is performed.
We take the context to be the way things are, but relativized to parameters that are significant for a conversational interaction. Standard parameters include the time and location of the event, and/or the identity of an agent whose perspective is relevant, for example, the speaker or hearer.
In the standard terminology of work on meaning in analytic philosophy, our contexts are centered possible worlds. Features of context include anything and everything. They include physical things like tables and physical events like rain, psychosocial features like someone’s feeling of pain or of affiliation, and also cultural features, like the everyday practices that people tend to take part in, including communicative practices. However, our notion of context allows us to distinguish features around which the context is centered, like the current speech event (if any), the view to the left, what the individual around which the context is thinking at the time at which it is centered (if anything), and the weather in the same place but twelve hours earlier.
The following definition, which takes a practice to have an extension consisting of all the instances of the practice, provides the first step of the technical development of our model: Associative resonances of an action: A feature of context is a (positive) resonance of an action in the extension of a practice to the extent that an occurrence of the action changes the probability of that feature (positively).44 So, if people are 90 percent more likely to have just experienced pain after saying “ouch” than if they hadn’t said “ouch,” then the strength of the resonance of the pain feature for this action is +90 percent.
Much of this book is concerned with societal practices, especially communicative practices, and the above definition of resonance provides a crucial component of our analysis of practice, and of the meaning of a practice. This is developed further in the next chapter. In talking of the meaning or resonances of a practice, one might be taken to refer only at a high level to the broad significance the practice has, whereas we are referring at a low level to the resonances associated with individual instantiations of the practice by virtue of instantiating it. For example, in talking of the meaning of the practice of exchanging rings in a wedding ceremony, one might refer to the cultural significance of the fact that a society has such a practice, a sort of sociological meta-analysis of the practice of exchanging rings. By the associative resonances of the action of exchanging rings, we refer to the significance that exchanging rings tends to have each time they are exchanged, not the significance for the culture of the fact that it has a practice of exchanging rings. However, and as will become important in part II of the book, it will turn out that resonances involve much high-level information, information about the ideology of the communities that perform them that goes well beyond the immediate significance that participants in a particular action, whether an action of ring exchange or an action of saying “ouch,” are typically aware of.
In Bayesian terms, the strength of the resonance that an action has for a feature is the probability of the feature given the action minus the prior probability of the feature, what would be written in standard mathematical notation as p(feature| action) - p(feature).45 When considering narrowly how utterances are interpreted by a particular hearer (or, more generally, a particular observer of a communicative act), it would suffice to consider the probabilities as subjective probabilities based on the hearer’s prior information about actions of that type, and the probability boost as being a measure of the evidence that the action provided about the features of context. However, we understand resonances as being independent of particular observers, as inherent properties of actions qua instances of a given type of action, and we must be a little careful here.
There is a well-established (but controversial) tradition of analyzing probabilities as objective, so that, for example, the probability of decay of an atom of a particular isotope given the isotope’s half-life might be regarded as independent of the observer, as an inherent propensity of the atom. However, we are concerned with intrinsically social actions, for which the word “objective” is not entirely apt. Although we can conceive of cases where this might be inadequate,46 we will adopt a frequentist interpretation of probabilities. That is, it suffices for current purposes that the resonance of actions instantiating practices is in principle measurable in terms of frequency of co-occurrence, and frequency of occurrence of one without the other. What is crucial to such an interpretation is that it is clear what set of situations is to be considered in calculating frequencies, and hence probability boosts. We discuss this issue in section 1.8, below, and return to it in chapter 2, especially section 2.5.
Let us examine the resonances of an utterance of “ouch.” To do this, we need to consider a set of contexts centered on someone saying “ouch” and compare it to a set of what we might call index situations. Resonances are probability boosts, and the probability boost is the difference between the probability of some feature given an utterance of “ouch” and the prior probability of that feature. The index situations are those that determine the prior probability.
The choice of index situations is important, and we will have more to say about it, but for the moment let us consider a set of contexts centered on a randomly chosen living human. The features we will consider are (i) recent speaker pain, (ii) speaker being somewhat fluent in English, (iii) speaker being a living human, and (iv) the increasing cost of fish. Then: i. the probability of recent speaker pain given that the speaker said “ouch” is much higher than the prior probability of pain, so this is a strong resonance; ii. the probability of English fluency given that the speaker said “ouch” is somewhat higher than the prior probability, so this is a medium- to high-strength resonance; iii. the prior probability of the centered individual being human and alive is already high in the contexts being considered, so the utterance produces no probability boost, and this is not a resonance; iv. there is presumably little correlation between people saying “ouch” and the increasing cost of fish, so there is only very low resonance, if any. To the extent that someone exposed to a sudden increase in fish prices might signal their feelings using such an exclamation, it is because that cost increase has caused them pain, or because they are jokingly exploiting the resonances of “ouch” to make an analogy with the feeling of pain salient.
Our analysis of resonances is intrinsically scalar rather than categorical. A scalar relation is one that, like the probability of A given B, admits of degrees. The standard entailment relationship of logic from Aristotle on is in this sense categorical. Kaplanian meaning is categorical insofar as it involves the set of contexts in which an expressive is used, and a given context either is or is not a member of that set. Resonance is thus obviously unlike Kaplanian meaning.
For purposes of comparison, a categorical notion of resonance can be derived by assuming some specific threshold degree of probability boost, so that we can talk of a feature of the context being a resonance of an action just in case the feature has that action as a resonance to that threshold degree. For example, that threshold could be the degree that could be a probability boost of .5, or a psychologically determined threshold, say the threshold that would tend to make a connection between the action and a feature noticeable to a random person chosen off the street who’d seen the action three or more times. Leaving aside the details of how the threshold should be set, we can easily define a Kaplanian meaning of an action as the set of contexts that have features that are resonances of the action.
Extending our metaphorical terminology, we can also say that these contexts are resonances of the action.
Kaplanian meaning and resonance can both be related to what is referred to in biology, economics, and game theory as costly signaling. Costly signaling is the sending of a message in a way that requires significant effort, as opposed to cheap signaling, which is usually taken to include much vocal communication. Words are cheap. In many cases, what makes a costly signal effective is that it involves manifesting a property by putting effort into some action that could not have been performed were the property lacking. We could say, for example, that the Kaplanian meaning of driving to work is a complex proposition, something like the conjunction of the propositions that one has access to a vehicle in a drivable state, that one has the physical capacity and legal license to drive, that one has a workplace physically located at some drivable distance from one’s current location, and so on. Therefore at least one way to signal this complex proposition or any of its entailments, a very straightforward and not easily refutable way, would be simply to drive to work. Putting in that effort obviously shows that you have the capacity to do so. In our terms, having access to a car is one of the resonances of the action of driving to work.
One can understand much real-world signaling in terms of resonance.
Absent a medical condition, a resonance of burping after eating is that one has eaten a significant amount of food in a rapid or delightfully haphazard manner. This natural resonance then explains the fact that in some parts of the Middle East and Asia a mild burp after a meal has come to have a conventional nonnatural significance, a polite expression of the quality and quantity of the repast. It carries a convincing message to the extent that it is hard to fake. Thus does a burp become resonant.
Your capabilities are resonances of various actions you perform, specifically, and rather obviously, all the actions that evidence your capability. So, just as with the driving to work case, one can signal one’s ability to invade a country by massing multiple divisions of heavily armed troops on the border, the resonance of this action providing compelling proof that one has the relevant military capacity. Other, more standard examples of costly signals are the provision of a lifetime warranty, indicating a seller’s faith in the reliability of a product, and the ostentatious display of finery, whether a monarch’s clothes and castles or a peacock’s feathers. What such signals lack in subtlety and ease of performance, they make up for in clarity and probative value. Why? Because insofar as a particular costly signal depends on preconditions that are themselves incontrovertible, the signal is completely reliable. Costly signals have reliable resonances.
1.5. Revelations What communicative act does someone perform when they utter an expressive? It is tempting to say that in uttering an expressive, speakers inform hearers of their state, so an expressive act would be an act of informing. This is unsatisfying, since it does not distinguish the act performed by saying “Ouch!” from the act performed by saying “I just felt a sudden sharp pain.” Furthermore, if saying “Ouch!” is an act of informing, why is it that we would not normally describe an “Ouch!” that way? Suppose that Jason says “Ouch!” and David reports the vocalization by saying “Jason just informed me that he’s in pain.” The implication would be that David didn’t think Jason had reflexively responded to pain, but rather had, perhaps humorously, exploited the convention that “Ouch!” expresses pain, and said “Ouch!” in order that David would recognize that Jason was sending a standard pain signal. Such labored ouches are not direct expressions of pain, in much the same way that saying “Ha, ha!” is not a direct way of expressing that you find something funny.
In fact, saying “Ha, Ha!”—like “Ha, bloody ha!”—commonly expresses not the amusement of true laughter, but, sarcastically, the opposite. Such cases of exploitation of a communicative convention are naturally analyzed as involving pragmatic reasoning to figure out the intention behind someone’s communicative act, typically in the style developed by the philosopher Paul Grice.47 More generally, Kaplan’s account of expressive meaning is compatible with a Gricean explanation of how all uses of expressives convey information. The idea would be that when a speaker says “Ouch!” genuinely, a hearer who knows that “Ouch!” is something people say when they just experienced pain can reason that the speaker said “Ouch!” in order that the hearer would recognize their intention to convey that they just experienced pain. On this view, understanding someone who says “Ouch!” involves substantive theory of mind. We will not attempt to provide detailed arguments against such an account, but rather try to describe how a simpler, more direct account of expressive acts that doesn’t revolve around communicative intentions could work. But what could it mean for something to be a direct expression of a feeling, and what would it mean for the effect on its hearer to be more direct than recognition of an intention to communicate that feeling?
The later Wittgenstein begins his quest for an account of meaning in the Philosophical Investigations with a nod sixteen centuries earlier, to St.
Augustine. Here is a quote from the Confessions, going slightly beyond what Wittgenstein cited: My infancy did not go away (for where would it go?). It was simply no longer present; and I was no longer an infant who could not speak, but now a chattering boy. I remember this, and I have since observed how I learned to speak. My elders did not teach me words by rote, as they taught me my letters afterward. But I myself, when I was unable to communicate all I wished to say to whomever I wished by means of whimperings and grunts and various gestures of my limbs (which I used to reinforce my demands), I myself repeated the sounds already stored in my memory by the mind which thou, O my God, hadst given me. When they called some thing by name and pointed it out while they spoke, I saw it and realized that the thing they wished to indicate was called by the name they then uttered.
And what they meant was made plain by the gestures of their bodies, by a kind of natural language, common to all nations, which expresses itself through changes of countenance, glances of the eye, gestures and intonations which indicate a disposition and attitude— either to seek or to possess, to reject or to avoid. So it was that by frequently hearing words, in different phrases, I gradually identified the objects which the words stood for and, having formed my mouth to repeat these signs, I was thereby able to express my will.48 Perhaps the simplest communicative acts are the preverbal whimperings and grunts of which Augustine speaks, and this may be the earliest learned discussion of expressive meaning, notably presented with a Judeo-Christian revision of a Platonic nativist attitude toward the inbuilt nature of “sounds already stored in my memory” and toward “the gestures of their bodies … a kind of natural language, common to all nations.” What caught Wittgenstein’s attention, however, was Augustine’s description of naming and pointing to objects, which Wittgenstein takes as inspiration for his celebrated builders’ game: Let us imagine a language for which the description given by Augustine is right. The language is meant to serve for communication between a builder A and an assistant B. A is building with building-stones: there are blocks, pillars, slabs and beams. B has to pass the stones, and that in the order in which A needs them.
For this purpose they use a language consisting of the words “block,” “pillar,” “slab,” “beam.” A calls them out;—B brings the stone which he has learnt to bring at such-and-such a call.— Conceive this as a complete primitive language.49 Wittgenstein’s builders’ game provides something like the level of simplicity and directness of communication that we are interested in here, a good starting point for the more complex cases we will turn to in later chapters.
And “ouch” and “oops,” like “block” and “pillar,” are not costly signals. Yet even Wittgenstein’s builders’ game could be said to have an extra level of complexity to it beyond what we need for “ouch” and “oops,” insofar as the builders are engaged in an interaction in which both players must act, whereas it is not obvious that “ouch” or “oops” require any action on the part of the hearer. Indeed, like many expressives, it is striking how natural it is to utter them when no one else is present. So let us first consider how these relatively noninteractional expressives function, and then turn to the interactional case.
For “ouch” and “oops,” we agree with Kaplan that the primary function is to “display,” that is, simply to manifest or reveal a state. We will term them revelatory expressives, as opposed to “hello,” “goodbye,” “please,” and “thank you,” which we will call interactional expressives. To utter a revelatory expressive is, as it were, to wear your heart on your sleeve. It makes sense that as social animals humans would tend to reveal various aspects of their states, including their feelings and their status, to those around them, because knowing something about each other enables us to coordinate more effectively. We find the metaphor of “wearing” apt because much of what people actually wear can also be seen as revelatory.
The crown, the bowler hat, and the peaked cap all reveal your place in society, and both the style of your jeans and the ring on your finger can reveal who you roll with. And although the use of “wear” involves a metaphorical extension, something else you can be said to wear is the expression on your face, which can be revelatory, expressing your feelings.50 To say “Ouch!” is like briefly grimacing. The idea we want to pursue, then, is that when someone wears an expression on their face, we can see what they are feeling, and that other revelatory acts function similarly. The standard Gricean view of the communicative function of declarative assertions51 involves the following steps: (i) the speaker has an intention to communicate (their belief that) the world is a certain way, and wishes for the hearer to recognize that intention; (ii) the speaker produces an utterance that portrays the world as being that way; (iii) the hearer grasps that portrayal; (iv) the hearer recognizes the speaker’s intention (or the speaker’s underlying belief).
The Gricean view is idealized and covers only part of what is communicated, but it seems a reasonable description of the way in which much content is conveyed in a simple cooperative two-person act of assertion. Compare this to what happens in a simplified, passive model of perception: (i) entities have externally detectable qualities that are correlated with their state; (ii) an able observer perceiving those qualities recognizes that the entity has the corresponding underlying states.
When you perceive someone to be happy because they are smiling, the basic process, whether the smiler is a baby or an adult, is the second, perceptual process. To put recognition of emotion in terms we will introduce in chapter 2 and discuss at greater length in chapter 3 (especially section 3.4), the observer becomes attuned to the smiler’s emotional state because the observer is attuned to the practice of smiling. Attunement to the resonance of smiling enables a nonreflective and direct perception of that state, direct in the sense that the psychologist Gibson used that term in his development of a theory of visual perception.52 There is no essential need for intention recognition.
Of course, an adult may feign a smile, and they may smile because they have the intention to convey happiness or some other message, and you may or may not perceive this intention, so the communicative process involved in understanding a smile may sometimes be more complicated.
Crucially, it need not be. We may say in such a case that a communicative action drawn from the class of expressives is being used to perform a nonexpressive act, or at least is being use to perform something other than the expressive act with which it is most commonly associated.53 Likewise, when we recognize the pain of someone grimacing, or the pain of someone who utters “ouch,” our recognition of the state of pain does not imply the recognition of an intention. That, we suggest, is the sense in which the communicative process associated with a revelatory act like saying “ouch” is simpler and more direct than that for an utterance of “I just felt a sudden, sharp pain.” If anything, recognition of the significance of a revelatory act implies the absence of recognition of an intention, that is, the absence of recognition of an intention to act in a pained way in order to convey something that might or might not be reflective of the level of pain that the actor is feeling.
The directness of recognition of revelatory acts again suggests a respect in which the resonance metaphor is more appropriate for expressives than is the content-delivery metaphor. An expressive is an outer counterpart of what is always, at least in part, an inner state. Its presence in the air between speaker and hearer enables the hearer to become attuned to that inner state in much the same way as registering someone’s red face allows you to become attuned to someone’s state of excitement, and feeling their pulse allows you to become attuned more specifically to their heartbeat. A red face is not a box containing excitement; a pulse is not a box containing a heartbeat; and neither is “ouch” a container for painful content.
The idea we are proposing is that expressives should be thought of not as containing meanings, but as sustaining social connections. Use of revelatory expressives helps individuals become attuned to each other, in the sense that they allow the feelings of a person to become correlated with the model an observer has of those feelings. Our earlier technical definition of what it means for a feature of the context to be a resonance of an act is that the act is correlated with that feature. From a hearer’s perspective, the act provides probabilistic evidence for the state, although the hearer need not be understood as reflectively weighing up this evidence. Such resonance is what allows a hearer to recognize a smile as the appearance of a happy person and to recognize an “ouch” as the sound of a person who experienced pain. Further, an observer perceiving the state of another may recognize what it would be like to be in that state themselves. So, resonance allows us to empathize with others whose states we can perceive. As discussed in section 3.9, if the connection between people is strong, one person may even feel another’s pain, at least in attenuated form. Resonance can hurt. This is a theme that will become still clearer in section 10.2, where we apply our resonance-based approach to yet another type of expressive word, namely slurs. Slurs, we will argue, can not only reveal emotion, but also reveal ideology. Furthermore, slurs do not merely reveal, they also inflict, and sometimes they may inflict pain.
1.6. Interactional Expressives as Moves in a Game Moving from the revelatory to the interactional case, we now discuss models of interactions as games, so that a standard game-theoretic approach can be used to model expressive acts, and we can make more concrete the technical underpinning for the resonance approach to interaction. Broadly, resonance between communicating agents is analogous to coordination in a game.
However, the real world does not sharply delineate a definition of the game, the strategies adopted by players, and the individual token actions played. This takes us to a consideration of the difference between what is expressed and what the expression presupposes. This distinction, we argue, is crucial in the case of expressives, since the role they play in the social world hinges on presupposed resonances.
An example of a simple but devastatingly effective signaling system is given by insects in a colony under attack. They may release a specific pheromone that spreads appropriate defensive behavior, a concerted counterattack coordinated by a chemical cocktail. In game theory, a signaling system becomes part of a strategy, an idea famously developed in David Lewis’s PhD thesis, which became his 1969 book Convention, where the concept of a signaling game is introduced.54 Let us say that in what we’ll call the coffee game, there are a certain range of states {hunger, thirst, sleepiness} of a player named David (the sender), a certain range of actions {moan, nod, blink, …} that David can perform, and a further set of actions that consist of providing goods {food, coffee, nothing, …} that a second player, Jason (the receiver), can perform in response. Let us assume that Jason cannot directly detect David’s state.
We might suppose that there is a particular pattern of utilities associated with Jason’s actions, so that there is, for example, higher utility to both players for coffee when David is in a state of sleepiness than for any other action, and suboptimal utility for coffee in any other state. Suppose that signaling is cheap, that is, that there is no significant inherent difference in utility between each of David’s actions. In that case, it doesn’t matter exactly what action David produces in any particular state. A communicative strategy for David would consist in a particular pattern of actions by David, for example, always nodding when sleepy. If Jason now has a strategy of performing action coffee when and only when David has performed action nod, the net result will obviously be that Jason provides the coffee exactly when it’s needed, despite not being able to detect that state directly. If both players reliably stick to these strategies, then the combination of strategies will constitute a convention between them. The sleepiness-nod-coffee convention allows the two players to succeed in coordinating, and their joint work can proceed apace, with joint utility for both parties.
Given some facts about how often different combinations of states and actions happen, we can calculate the resonances of the action of nodding.
Suppose further that David is sleepy 25 percent of the time, thirsty 25 percent of the time, and hungry 50 percent of the time. Suppose further that after some large number of iterations of the game, David nods 100 percent of the time when he is sleepy, 40 percent of the time when he is thirsty, and 0 percent of the time when he is hungry. The probability of David’s states corresponds directly to their frequencies, so, for example, p(sleepiness) = 0.25. The boost that nodding gives to David having been sleepy is then p(sleepiness| nod) - p(sleepiness) = 1 - 0.25 = 0.75. As far as David’s initial states are concerned, then, it can be seen that the resonances of nodding are hunger: -0.5, thirst: 0.15, and sleepiness: 0.75. It can be seen that once a convention like sleepiness-nod-coffee is somewhat in place, as in the above scenario, the strongest resonance of David’s nodding, from among his prior states, is sleepiness. Likewise, if Jason always brings coffee when David nods, and rarely brings it otherwise, then both David’s state of sleepiness and David’s action of nodding will be resonances of coffee. An observation of the action coffee would provide evidence for both the prior state and the prior action.
Resonance in the coffee game is asymmetric in the sense that the actions of the sender and receiver are not similar. Symmetric resonance is exemplified by the spread of laughter or barking, whereas the resonances of David’s nodding are asymmetric, more like the Pavlovian metronome than Galileo’s pendulum. The pattern sleepiness-nod-coffee is our first example of an interactional practice. It is much simpler than the more realistic practices we will be considering in future chapters. Crucially, it is the practice that gives actions their significance, or, as we will say, their resonances. That is, we will not be primarily concerned with what, if anything, David thinks he is communicating on any particular occasion when he nods, or with what Jason thinks David meant by nodding when Jason brings David coffee immediately afterward. Our focus will rather be on the significance of nodding within the context of a practice of doing so at certain times and with prototypical outcomes. This provides the blueprint for all the later analyses in the book: the resonances of a communicative action will be given by whatever practices the action can reasonably be taken to instantiate. Note that this does not mean that the intentions or other mental states of interlocutors are irrelevant to meaning as defined in terms of meaning, for it is quite possible for a practice to depend on any arbitrary aspect of an interlocutor’s mental state, whether that be a state of surprise, a state of pain, or a state of intending to make a joke. But we are getting ahead of ourselves: the practice we’re now considering doesn’t involve mental states of surprise, pain, or intention, merely states of sleepiness, hunger, and thirst.
What does it mean to say that two people’s actions are coordinated? It is easy to understand in symmetric cases where coordination simply means that all players do the same, but game theory is helpful in understanding what coordination means in asymmetric cases. The relevant technical notion is that of a coordination game. This is a game in which there are multiple solutions (technically, equilibria, e.g., Nash equilibria), and where players perform actions that accord with the same solution. In the above example, David and Jason coordinate on the strategy pair sleepiness→nod and nod→coffee, but an equally effective solution would have been the pair of strategies sleepiness→blink and blink→coffee. Thus, game theory provides one way of understanding what it means for people to be coordinated, or, as we will put it in the next chapter, to be attuned to each other dispositionally: they are attuned in the sense that they have matched strategies in a coordination game.55 The sense of connection that people feel with each other might perhaps be related not only to their capacity to empathize, as discussed in the previous section, but also to how attuned they are in this technical sense. That is, the depth of connection we feel with someone might depend on how effortlessly we can coordinate shared dispositions that result in mutually advantageous outcomes to our interactions.
The fact that there is no a priori reason for players to prefer one solution to another in a signaling game is analogous to Ferdinand de Saussure’s notion of arbitrariness of the sign in natural languages.56 That is, the coffee game illustrates not merely a convention, but an arbitrary convention, since other conventions would have done just as well. The development of the convention over multiple plays (it being an iterated game) illustrates a difference between costly signaling and cheap talk. Although conventional use of costly signals does emerge over time (the peacock’s biologically costly feathers, for example, have a certain arbitrariness to them), costly signaling can operate without preexisting conventions: doing something that requires property X provides evidence that you have property X to a suitably perceptive observer. Thus, costly signaling builds on an existing resonance. On the other hand, cheap talk in the coffee game displays the emergence of an entirely new resonance. David and Jason initially dance to their own beats, but come to be attuned without negotiation or prior alignment of dispositions.
In the coffee game, nodding is an interactional signal: it functions as part of a strategic pairing of actions by distinct agents. But according to the definitions we’ve given, it is also revelatory: it is correlated with David’s tiredness, and hence could potentially reveal that state to a suitably able observer. So what, then, is the meaning of a nod in the game?
On our proposal, the meaning of a nod is its resonances, a distribution of probability boosts of different features in the context. But in this section, we want both to contrast our proposal with a more Kaplanian approach, and to make a further argument that does not depend crucially on the fact that resonances are scalar. So let us also consider two variants of Kaplan’s notion of meaning, just similar enough to resonance to make comparison easy, but categorical rather than gradient in the way that resonance is.
Consideration of these variants will reveal stark differences between the Kaplanian approach and ours, while also highlighting a crucial issue that will be clarified in the coming chapters, the issue of which contexts are considered in defining resonance. 1.7. Differentiating Kaplanian Expressive Meaning from Resonance Let us say that the Kaplanian meaning of an action is a function mapping features of context to 1 if the conditional probability of the feature given the action is 1, and to 0 otherwise.
Before proceeding, let us touch on two somewhat obvious but inessential differences between how Kaplan actually defined expressive meaning and our definition of Kaplanian meaning. First, Kaplan talked about the meaning of an expression, whereas we talk of the Kaplanian meaning of an action, by which, properly speaking, we mean the Kaplanian meaning of a type of action or practice. At least when the actions in question are utterances, this difference is just terminological. Second, for uniformity with our notion of resonance, we have defined Kaplanian meanings as functions from features to probabilities, whereas Kaplan defined expressive meanings as sets of contexts. However, the Kaplanian meaning of an action corresponds to a set of contexts in an obvious way, that is, to the set of contexts that bear all and only the features the action maps to 1.
There is a third possible, or even probable difference with Kaplan insofar as it is not entirely clear what notion of context he was assuming in the “Ouch and Oops” talk, although one might potentially interpolate what he intended from his well-known writings on the topic elsewhere. A context, as we’ve introduced it, is just reality seen from a certain perspective, and we assume that this is a good enough match for what Kaplan had in mind for current purposes.
Above, we discussed contexts as relevant for an artificial game, the coffee game. We could stipulate that what we took to be reality was much reduced, and contained only the states and actions mentioned in the game, but this would to some extent belie the reason for using the artificial model in the first place. The intuition behind the game is that it is a simplified model of a real-world language community. So, again for the sake of argument, let us assume that in discussing the coffee game, reality has been fictitiously augmented with the two players performing their role in the coffee game, and centered on communicative episodes in the game.
As defined, the basic Kaplanian meaning of an action is a set that includes every feature that is found in all contexts of use. Since in all contexts the players are named David and Jason, part of the basic Kaplanian meaning of a nod is that the players are named David and Jason.
Also, part of the basic Kaplanian meaning is that brown butterfly wings contain melanin, and that every mouse is self-identical. We will tackle these in reverse order, to bring out differences between Kaplanian meaning and the resonance model.
The last of these features of context, that every mouse is self-identical, is a tautology, as is every theorem of mathematics. In classical logics, tautologies have the very special property that they follow from every other proposition. As a result, in many standard approaches to meaning, tautologies are in some sense included in (or at least follow from) every declarative sentence. Relatedly, in many theories of belief based on classical logics, when one believes something, one also believes everything that follows from it, and hence all tautologies—what is known as the problem of logical omniscience. This is generally regarded by philosophers of language and linguistic semanticists as an acceptable price to pay given the benefits of the formal modeling systems they use, and many do not regard it as a negative at all. People do not exhibit logical omniscience, in the sense that they cannot recognize all the consequences of their beliefs, and would not recognize, say, Fermat’s last theorem as following from, say, the statement “It’s raining.” Nonetheless, logical omniscience is commonly viewed as an acceptable idealization, justified by the clarity that logical models bring to models of both belief and meaning.
Given this background, it is not terribly surprising that the Kaplanian meaning of an action should fold in tautologies like the fact that every mouse is self-identical. Presumably, many theorists would not find it to be an objectionable property. However, let us point out that there is here a sharp difference between resonance and Kaplanian meaning. Whereas every tautology is part of the Kaplanian meaning of every expression, resonance has the reverse property: tautologies are never resonances of any expression. This can be seen by reasoning as follows: (i) the resonances of an action are features that are more likely in the presence of the action; (ii) tautologies are features of every context, which have a uniform probability of 1 in every context; and so (iii) tautologies never get a probability boost, and cannot be resonances.57 Similar issues will arise in our discussion, in chapter 3, of how people harmonize their beliefs and other attunements. Quite generally, in this volume we try to avoid strong idealizations about people’s reasoning capacities, including the idealization of logical omniscience. We leave a broader discussion of idealization until chapter 9.
We take it that the property that brown butterfly wings contain melanin is similar to a tautology insofar as it is a uniform physical feature of the real world, and thus a feature that holds in every context, independent of how that context is centered. Thus, by similar reasoning as for tautologies, this property is part of the Kaplanian meaning of every expression, but a resonance of no expression, at least if we anchor our contexts to the real world, which we do in this volume. Note here that this does not force resonances to always be veridical: even though situations in the real world in which real speakers have actual pain are resonances of “ouch,” it does not follow that the speaker is in pain every time it is used. It merely follows that if people expressing themselves in this way were rarely in pain in the set of contexts used to define the resonance, it could not be a strong resonance.
To further elucidate the difference between resonance and Kaplan’s conception of expressive meaning, let us move from one to the other via two intermediary steps. The first intermediary step is a version of Kaplanian meaning where probability plays a nontrivial role: what we can call for the purposes of this section the probabilistic Kaplanian meaning of an action is a function mapping features of context to 1 if the conditional probability of the feature given the action is above 0.5, and to 0 otherwise. The second intermediary step we can call the categorical resonance of an action: a function mapping features of context to 1 if the probability boost given by the action is above .5, and 0 otherwise. By the probability boost we mean, as before, the difference between the conditionalized probability of the feature and the prior probability of the feature, and we call it categorical because it categorizes features as to whether they are resonances (value 1) or not (value 0). So, the categorical resonance of an action is a halfway house between the probabilistic Kaplanian meaning and regular resonance, like the former except based on the boosted probability of a feature of context, rather than absolute probability, and like the latter, except with all values above .5 pushed to 1, and all other values pushed to 0. In the regular resonance definition, the one we will be using in the rest of the book outside of the discussion in this section, features to be mapped into real values between -1 and +1, that is, the actual probability boost or drop of a feature given a certain action.
Let’s consider the differences between Kaplanian meaning and probabilistic Kaplanian meaning. In the above coffee-game scenario where the sleepiness-nod-coffee conventions have emerged, both would assign values to David’s states following his nodding of hunger: 0, thirst: 0, and sleepiness: 1. Similarly, the two notions agree as regards the players in the coffee game being named David and Jason, brown butterfly wings containing melanin, and mice being self-identical. Where the two notions come apart, obviously, is as regards features of context that are merely probable rather than certain; this has both positives and negatives.
Let’s get what we take to be a negative for probabilistic Kaplanian meaning out of the way.
The probabilistic Kaplanian meaning of David nodding includes the proposition that the next dice thrown anywhere in the world will not come up on six, and indeed that the last dice thrown did not come up on six. Indeed, this will be true of the probabilistic Kaplanian meaning of any type of communicative action that is not directly correlated with the value of the next dice thrown (and similarly, mutatis mutandis, for the last dice thrown). This is because if we consider a somewhat arbitrary non-dice-related set of the contexts centered on individuals at a certain time and location, say all those where David nods, then in a majority of them the next dice thrown will not land on six. This is obviously not a property of nonprobabilistic Kaplanian meaning. Neither does it hold for the resonance model: knowing that David had nodded would not boost your estimate of the next dice thrown coming up on six.
It might reasonably be argued that it is a mere artifact of our definition of probabilistic Kaplanian meanings that merely probable events are among the features connoted by all sorts of utterances. However, it does bring with it a striking advantage over the Kaplanian meaning. Suppose that just once somebody has said “ouch” without having first experienced pain, perhaps as a joke, perhaps to mislead, perhaps through misunderstanding of the convention, perhaps through some quirk of speaker performance, and so on.
Whatever the reason, by the definition of Kaplanian meaning, speaker pain is then not a feature associated with, or expressed by, the utterance of “ouch.” Indeed, nothing but tautologies and other eternal physical properties of the world are part of the Kaplanian meaning of anything. Absent some further work, Kaplanian meaning, and, by extension, Kaplan’s own definition of the meaning of expressives, is so idealized as to be inapplicable to meaning in the real world, where no convention is followed 100 percent of the time.
Switching from Kaplanian meaning to probabilistic Kaplanian meaning provides at least one way of ameliorating the problem, although, as we’ve seen, that brings its own problems. One can also imagine variants of the latter with higher probability thresholds for associating features with actions, say 95 percent instead of 50 percent, but simply changing the threshold would not alter a fundamental weakness of the probabilistic Kaplanian meaning, namely, that highly probable features of context are signaled by every action, even if there is no correlation or causal connection at all between the action and the feature. Since, or so we imagine, conventions of language are broken more often than lightning strikes a speaker just after speaking, it follows that there is no way of setting the threshold such that (a) the probabilistic meaning will include what we would think of as conventionalized aspects of meaning, and (b) it is not the case that every action has as part of its probabilistic Kaplanian meaning that the speaker will not be struck by lightning just after speaking.
Here, categorical resonance helps. In the coffee-game scenario, categorial resonance gives the same pattern of values as for Kaplanian meaning and probabilistic Kaplanian meaning (again, hunger: 0, thirst: 0, and sleepiness: 1). Where categorial resonance differs is that it only associates features with an action if their presence in the context is correlated with that action. Therefore, the categorical resonance of an expression includes neither tautologies, nor laws of nature, nor uncorrelated high probability events like those involving dice throws and lightning strikes. And here we see a general advantage of resonance over Kaplanian meaning, since the regular resonance of an action is similar to categorical resonance in this respect: it does not assign any probability boost (or probability drop) to arbitrary features of context that are not correlated (or anticorrelated, respectively) with the action. 1.8. Which Contexts Count?
The question of what features of context are part of the Kaplanian meaning of an expression depends on the set of contexts that are considered. For example, if one were somehow able to restrict the set of contexts over which the Kaplanian meaning of “ouch” was calculated to just those in which people used it to express pain, then the Kaplanian meaning of “ouch” would express speaker pain, since the contexts where people were joking and so on would be excluded. It’s doubtful that this would be a viable way of dealing with the failure of Kaplanian meaning to account for variability in language, such as uses of “ouch” in painless situations, because it is inherently circular, defining the meaning of expressions to be just what happens in the situations chosen because they have the features desired for the meaning. Nonetheless, the dependency of Kaplanian meaning on the particular set of contexts considered highlights an issue that for us is pivotal and affects all four models we considered in the last section, namely, Kaplanian meaning, probabilistic Kaplanian meaning, categorical resonance, and regular associative resonance. Indeed, the issue underlies the way that conventional meaning is transmitted and transformed over time, and also underlies the way that the same word can have different meanings to different groups. The issue is this: since we take the meaning (i.e., resonance) of an action to be a function of features of contexts in which people can make a choice as to whether to perform that action, meaning is not absolute, objective, and fixed, but rather is dependent on the choice of the set of contexts over which it is calculated. The question, then, is what set of contexts to use?
Even with respect to our simple coffee game, the answer to the question of which contexts to consider is not obvious, for it depends on what features we allow to vary in our contexts. If we restrict ourselves to contexts that are complete plays of the game and in which the nod→coffee convention has emerged and is followed all the time, such that nodding is performed always and only when David is sleepy, and coffee is delivered on exactly those occasions and on no others, then the expressive meaning of a nod might be described as a conjunctive proposition: David is tired and Jason is going to get coffee. But why take this notion of context? An alternative context for an action might be incipient, containing only information up to the time that the action took place. This seems like a good candidate for what Kaplan had in mind, but from a purely technical point of view we might equally well use posterior contexts, containing only information about what takes place after an action. We could say that the incipient expressive meaning of nodding in the coffee game is that David is tired, the posterior expressive meaning is that Jason will get coffee, and the full expressive meaning is the conjunction.
Yet why limit ourselves to contexts in which just one particular set of communicative conventions has emerged? In considering the expressive meaning of the move nod, we could also calculate the expressive meaning relative to much bigger sets of contexts, such as a set of all contexts including those in which David and Jason have started with no conventions and iterate the coffee game until one arises, whether that is the nodding convention or another, say that he blinks when tired. In that case, we would end up with disjunctive expressive meanings. For example, the disjunctive incipient meaning would be this: either there is a convention that David nods when tired and David is tired, or there is no convention that David nods when tired and David is not tired.
But why stop there? Why not consider the set of all contexts in which there is an act designated nod? These contexts would include games just like chess except that one nods when one resigns, and games that involve living an otherwise normal life, except that one nods when and only when one recognizes the emptiness of one’s existence. Given a poor choice of a set of contexts, the current notion of expressive meaning of nodding would become distinctly unenlightening. For the notion of resonance to be at all helpful, we somehow need to find a set of contexts like Goldilocks’s ideal chair: neither too big, nor too small.
Intuitively, the set of contexts that are “just right,” in that they capture what resonance is supposed to capture, is what is likely to be informatively signaled by a given action. One thing we don’t normally take an utterance of “oops” to signal is the fact that one tends to say “oops” after a minor mishap. That is, saying “oops” doesn’t normally signal the presence of the very conventions that we follow when we use it. While we can, in providing language instruction, convey a convention by using it, signals are not typically used to communicate their own communicative conventions.58 Conventions are features of context that are resonances of a signal, since every use of a signal provides evidence that the conventions surrounding that signal are in place. So, conventions provide an example of a feature of context that is a resonance of a signal, but which intuitively would not normally be considered part of the expressive meaning. There are many such features. When somebody says “oops,” that provides evidence that they are not dead, and that they are neither a watermelon nor a prime number. And indeed, these features would be part of what Grice called the “natural meaning” of saying “oops.” Crucially, however, while saying “oops” means that you are not a watermelon, “oops” is not generally used to communicate that the speaker is not a watermelon.
Our consideration of signals in the coffee game has brought forth a number of problems with Kaplanian expressive meanings. First, it’s not clear what aspects of an interaction should be considered in a Kaplanian context, and second, it’s not clear whether the conventions of communication can themselves be allowed to vary across Kaplanian contexts. However, we are not concerned with defending or extending Kaplan’s notion of expressive meaning per se, since our notion of resonance is quite different. Even so, our discussion of a Kaplanian approach, and of how to deal with at least some of the problems in that approach, casts light on our proposal for analyzing meaning in terms of resonance.
The resonances of an expressive include both incipient and posterior features, and among the incipient features are the prior conventions of meaning themselves, the very fact that the expressive has the resonances it does. In general, we do not regard the fact that resonances include both incipient and posterior features to be a shortcoming of our model. There is, on our view, no fact of the matter as to whether David’s nod is more like an expression of tiredness or more like a request for future coffee, and no reason why resonances should be restricted to only one of these. However, we do think that it is intuitively clear that David’s nod is not naturally thought of as signaling that there is a convention of nodding when tired.
If the presence of this convention is a resonance of nodding, and if the meaning of nodding just is its resonances, then we are led inevitably to the following conclusion about our model: Meaning/signal inequality: What an action means is more than what is signaled. This result will in fact be central to the account we develop in part II of the book. Much of meaning, we will argue, consists not in what is signaled, but in what is presupposed, and it is the presupposed part that is perhaps of greatest interest when considering the often-invisible transmission of ideology, for example, the very potent hustle of propaganda.
How, then, can we get the right set of features, a set that includes just what an expressive signals, and that leaves out all the features of context that are resonances that are not part of what is signaled? Over the coming chapters, we develop a two-pronged approach to this problem.
One prong of our approach to getting the right set of features involves using a standard method for restricting context: presupposition.
Presupposition will be used to differentiate between what an action means and what it signals, or, more generally, between what an action means and what it does. We postpone discussion of presupposition until the second part of the book where, based on the notions of resonance developed in this chapter, we develop a radically different account of presupposition than is found in prior literature. However, the basic idea of how presupposition is relevant to the Goldilocks problem of finding a set of contexts that is neither too big nor too small, will already be clear to those readers familiar with prior presupposition theory. The presuppositions associated with an expression determine a broad set of contexts in which communicative interactions of a given type take place. For example, “oops,” “ouch,” and “pass the salt, would you darling” all share a common presupposition that some variety of English is being spoken. It is only relative to this subset of contexts that any one of the three is used, and none of them are standardly used to signal that we are in such contexts. Within that subset of contexts, each of them has its standard functions, for example, expressing that what the speaker takes to be a mishap has occurred.
Presupposition will be used to distinguish between the complete set of resonances of an expression, which encode all the information that an observer could possibly derive from an utterance of the expression, and the smaller set of resonances that determine what an expression usually does, including what it is usually used to signal.
The question of which contexts should be considered in evaluating the resonances of an expression, and hence what can possibly be communicated by use of the expression, depends on psychological and social factors. The second prong of our approach involves extending our model in such a way that the resonances of communicative practices become reflections of the way these practices manifest in society. In the next chapter, we begin our exploration of how social and psychological factors constrain context choice, developing an account of what we term the collective attunements of groups of people. The set of contexts to be considered will be constrained by the existence of recognizable groupings of people, for which we adopt from sociological work the term communities of practice, 59 who are collectively attuned to a use of an expression, as well as to many other practices. The set of uses of an expression that are relevant to determining the resonances of that expression will be restricted to a set of interactions taking place within such a community of practice.
While there is clearly much work to do, we hope that in this chapter we have made the case that a theory of meaning centered on resonance rather than content holds promise. We are not claiming that theories of meaning that are framed in terms of “content” are inherently wrong just by virtue of using the term, or that they are wrong by virtue of implicitly invoking other aspects of the conduit metaphor or similar metaphors of transfer. Such metaphors are natural and easy given the culture in which both we authors and readers of this volume are steeped, and we accept that such metaphors often illustrate an idea helpfully. But we also think that framing the theory of meaning in terms of these metaphors presents dangers: the danger of structuring the theory around the metaphor without proper reflection, and the danger of focusing the theory only on phenomena for which the metaphor is apt, and to the exclusion of many communicative phenomena that are of societal and cultural significance.
We are thus suggesting that it is a worthwhile intellectual exercise to attempt to approach the theory of meaning without reference to “content,” and without some of the assumptions commonly associated with its use. We hope that the reader will indulge us in this, and will agree with us also that it is worthwhile to think through an alternative approach to communication that is as yet underexplored (although not unexplored), thence to see what fruit it brings.
It should be clear that resonance has some advantages over Kaplanian meaning. It might be argued that some of these advantages are not particularly surprising. After all, the relevant properties of resonance emerge directly from a straightforward use of Bayesian probability theory, itself a completely standard tool in many areas of research, in common use for analysis of communication and cognition, and the tool that provides the foundation of the bulk of contemporary work in computational linguistics.
What is original in our project is not the use of probabilistic methods per se, but rather the way we leverage these methods to support the development of an account of meaning within society, an account in which meaning is essentially cultural and political.60 1. As quoted by Rosie Blau, Financial Times interview, November 7, 2008, accessed March 1, 2023, https://www.ft.com/content/b1c8c954-ac59-11dd-bf71-000077b07658.
2. Kant, Critique of Pure Reason, 193 (A51/B75).
3. Hockett, Refurbishing Our Foundations, 90.
4. “Confederate States of America—Mississippi Secession,” Yale Law School, 2008, accessed March 1, 2023, https://avalon.law.yale.edu/19th_century/csa_missec.asp.
5. “Declaration of the Causes of Secession, Georgia,” Digital History, 2021, accessed March 1, 2023, https://www.digitalhistory.uh.edu/active_learning/explorations/south_secede/south_secede_g eorgia.c fm.
6. “Confederate States of America - A Declaration of the Causes Which Impel the State of Texas to Secede from the Federal Union,” Yale Law School, 2008, accessed March 1, 2023, https://avalon.law.yale.edu/19th_century/csa_texsec.asp.
7. Alexander Stephens, “Cornerstone Speech,” American Battlefield Trust, 2023, accessed March 1, 2023, https://www.battlefields.org/learn/primary-sources/cornerstone-speech. This sentence continues painfully: “that slavery subordination to the superior race is his natural and normal condition. This, our new government, is the first, in the history of the world, based upon this great physical, philosophical, and moral truth.” 8. Fitzhugh, Cannibals All!
Or, Slaves without Masters.
9. Morrison, Playing in the Dark, 64.
10. Rogers, The Darkened Light of Faith, citation is from unpublished draft.
11. Rogers, The Darkened Light of Faith, introduction.
12. Here, as at many points to come in this volume, we are deeply indebted to Sally Haslanger’s work on social meaning and practice. For discussion of the ideal of motherhood and the importance of the social meaning of the word “mother” (although not “mom”), see Haslanger, “Social Meaning and Philosophical Method.” 13. Although we have not attempted a systematic historical study of the notion of “content” in philosophy of mind and language, it seems likely that Kant’s talk of “content” was taken over from Christian Wolff. See Anderson, “The Wolffian Paradigm and Its Discontent,” 30–32. The idea of one property or term containing another in the sense that if something is “in” the extension of one then it is also “in” the extension of the other seems to date back at least to Aristotle’s Prior Analytics (see Ross, Aristotle’s Prior and Posterior Analytics, 292, notes on 26–28). The idea that a thought might be structured and contain parts is found in Hume, but the structures that Hume is considering are mostly the syntactic Aristotelian subject-predicate structures of sentences used to express thoughts, so that the syntactic expression of a proposition can contain a subject and predicate. He does at one point in the Treatise of Human Nature comment (in discussing existential propositions, for which he denies that existence is a predicate), “We can thus form a proposition, which contains only one idea” (Hume, A Treatise of Human Nature, 67, fn. 20). However, he does not in that work consistently talk of propositions as containing ideas, and he never refers to the content of a proposition or thought.
14. Malinowski, “The Problem of Meaning in Primitive Languages,” 315. If we may be pardoned a historical aside, note that Malinowski is not only explicit about the social function of language, but is also explicit in repeatedly expressing his account in terms of language as action. Thus he writes, some ten years before Wittgenstein dictated the notes that eventually became the Philosophical Investigations, and over thirty years before Austin gave the William James Lectures that became How to Do Things with Words, “In its primitive uses, language functions as a link in concerted human activity, as a piece of human behaviour. It is a mode of action and not an instrument of reflection” (312). Malinowski is explicit in stating that phatic communion also “function[s] not as an instrument of reflection but as a mode of action” (315).
15. There is, to be sure, much strong work by scholars who share some but not all of our goals, including work that we have, as yet, not attempted to reconcile with our own approach despite the presence of a clearly overlapping agenda. A prime example here is recent work of Jennifer Nagel and Evan Westra on the epistemology of dialogue (Nagel, “Epistemic Territory”; Westra and Nagel, “Mindreading in Conversation”). Like us, these authors seek to analyze not merely individual conversational turns, but also complex interactions taking place over many turns. Indeed, they have gone rather further than us in directly connecting their account to detailed theories of dialogue structure developed in the Conversation Analysis literature. However, there are several respects in which our goals crucially differ.
One is that they focus on cooperative conversation, whereas we consider ways in which language can be divisive and drive interlocutors apart. Another is that they focus on recognition of intention, whereas we take a central goal of a theory of political language to account for ways in which what is communicated comes apart from what is intended, or what is intended to be recognized. A third is that they focus on epistemology, how language changes people’s knowledge states, whereas we attempt to centralize emotion and social role. Although the focus of the Nagel and Westra work is quite different from that of our work, we should make it clear that we do not deny the importance either of the type of cooperative conversational situation they study or of epistemological aspects of conversation. Indeed, we see no particular reason why the approach Nagel and Westra develop could not, in principle, be reframed in terms of the resonance framework we develop here, a thought we offer not to suggest any shortcoming in their work, but rather to suggest a direction for future development of our own framework.
16. Orwell, Nineteen Eighty-Four, 274.
17. Festinger, A Theory of Cognitive Dissonance.
18. Reddy, “The Conduit Metaphor.” 19. Lakoff and Johnson, Metaphors We Live By.
20. As we will define it, transparency means transparency of intentions. One could extend the content-delivery metaphor to allow that a speaker accidentally packaged something they didn’t intend to, or that they intended to cover up the true reason they sent the package. We do not claim that the content-delivery metaphor is inconsistent with breakdowns in transparency, i.e., with hustle. Rather we claim that the content-delivery metaphor invites focus on something exchanged that is objective and accessible for all parties. Seen in this light, Gricean pragmatics, with its attendant reasoning about speaker intention, is itself a substantial and non-obvious break from a straightforward use of the content-delivery method. However, while Grice recognized that what is intended differs in interesting ways from what might be thought of as literal content, Gricean pragmatics does not offer a general account of unintended meaning, which is central for us.
21. Schön, “Generative Metaphor.” 22. To say that information is analogous to energy is perhaps underselling the connection between the two concepts, as physicists from James Clerk Maxwell and Ludwig Bolzmann to Albert Einstein and Stephen Hawking have shown that the relationship between energy and information is far reaching and subtle. So, e.g., Hawking’s most famous result involved a relationship between energy and information that led him to predict on information-theoretic grounds that black holes must radiate energy. (For a readable summary, see Stuart Clark, “A Brief History of Stephen Hawking: A Legacy of Paradox,” New Scientist, March 14, 2018, accessed March 1, 2023, https://www.newscientist.com/article/2053929-a-brief-history-of-stephen-hawking-a-legacy-of paradox/.) 23. Here we use “information” in the very broad sense it has in Shannon Information Theory, and in discussions of information in physical systems. In this sense, a wave crashing on the shore carries information about the position of the moon and about far-off storms long since passed. The lay use of “information” is narrower, referring to what is newly and perhaps even explicitly conveyed about the world. When an utterance is described as “uninformative,” what is usually meant is not that the utterance carried no physical information at all, but rather that it failed to describe anything unknown to the hearer. 24. Note that it is possible to use a conduit metaphor for information without assuming content delivery. Shannon Information Theory and the related development of Information Channel Theory illustrate exactly this. These models of information provide mathematical limits on the degree to which a stream of signals across a (possibly noisy) channel of limited bandwidth can result in information being shared between a source and a target, yet many of the results involve no assumptions at all about the particular process used to encode that information. Typically, no assumption is made that particular chunks of signal correspond directly to particular chunks of information. For a standard reference, see Cover and Thomas, Elements of Information Theory, and Hamming, Coding and Information Theory.
25. Stenning et al., Introduction to Cognition and Communication, 8. Robert St. Clair (“Cultural Wisdom, Communication Theory, and the Metaphor of Resonance”) is also explicit in contrasting resonance and transport metaphors for communication, and further contrasts the metaphor of resonance with a broader framing of social relationships and interactions in rhetorical terms.
26. The framing literature on resonance begins with Snow and Benford, “Clarifying the Relationship between Framing and Ideology.” A central text of this work is now Benford and Snow, “Framing Processes and Social Movements,” 611–39. Also highly relevant here is Gamson, Talking Politics, and Rodger Payne’s “Persuasion, Frames and Norm Construction,” which discusses coconstruction of resonant frames, paralleling moves we make when we introduce accommodation in chapter 6. Some recent work on resonance of frames (e.g., McDonnell, Bail, and Tavory, “A Theory of Resonance”) takes a dynamic and interactional view on when resonance occurs that is very much in line with our own development. However, it remains the case that what is studied in this newer work is when resonance occurs and why, whereas we attempt to model what the resonances of an action are. Extending slightly from work on framing, Ottati, Rhoads, and Graesser (“The Effect of Metaphor on Processing Style in a Persuasion Task,” 688) ask not when a frame resonates, but when a metaphor resonates, although there is a similar interest in the value of metaphor for persuasion.
27. Woodly, The Politics of Common Sense.
28. Woodly, The Politics of Common Sense, 96.
29. In The Politics of Common Sense, Woodly gives two different analyses of resonance.
The first, which we focus on in our discussion above, relates to the way that arguments use a shared cultural background for rhetorical advantage. She characterizes this background in terms of the Aristotelian notion of endoxa, i.e., understandings of how the world is shared by the majority, something close to what we term collective attunements in chapter 2. Woodly’s second analysis of resonance is a more practical operationalization of the notion of resonance deployed in her quantitative empirical study of news stories. In this study, she creates a taxonomy of the main arguments appearing in the complete text of many hundreds of news stories on a topic (e.g., marriage equality), identifies each instance of each argument, and then uses statistical methods to make sense of the distribution of arguments.
Departing from analyses of frames in the tradition that includes Benford and Snow’s work, she defines a frame as a collection of commonly co-occurring arguments.
She then characterizes a frame as resonant if it contains five or more commonly co-occurring arguments. Thus resonance on this second definition is a measure of the frame’s ability to connect a significant number of distinct arguments. In our terms, we should say that the arguments that make up a Woodlyan frame resonate highly with each other, because each of them is found in the contexts in which the others are present, and we would further hypothesize that the reason for this is that the arguments cohere with each other to produce consonance of attunements when people accept the arguments together. While it is plausible that Woodly’s quantitative measure relates to her discursive presentation of resonance in terms of shared cultural background, the nature of the connection is not obvious.
30. St. Clair, “Cultural Wisdom,” 80.
31. Rosa, Resonance.
32. Peters and Majid, Exploring Hartmut Rosa’s Concept of Resonance, 150.
33. Du Bois, “Towards a Dialogic Syntax,” 372.
34. An example of a well-developed project applying Du Bois’s notion of resonance (although a wider range of relevant antecedent work is cited) is Mark Sicoli’s analysis of Zapotec speech. He gives the following definition: “To resonate is to build signs that connect to and evoke the other through a parallelism of form that displays the common experience under development in the saying and doing of conversation as joint activity” (Sicoli, Saying and Doing in Zapotec, 17). As the definition suggests, Sicoli, in line with most literature adapting Du Bois’s work, takes resonance to be a local property of discourse, albeit that he views the use of resonance as a somewhat conventionalized aspect of Zapotec speech practice.
35. Cavell, The Claim of Reason. Although we find much to agree with in Cavell’s Wittgensteinian, practice-based outlook, he differs sharply from us not only because he takes attunement to be only a relation between individuals (rather than between an individual and anything else), but also in that he seems to take the foundational nature of human attunement to suggest that much work on understanding language conventions is superfluous, whereas for us it is precisely human (collective) attunement that must be studied in order to understand what language conventions are and how they develop.
36. Scheff, Microsociology; Collins, Interaction Ritual Chains.
37. Let us give credit to two further prior (and well-developed) literatures that invoke concepts of resonance, both concerning psychology of language. Our model shares features with both of these antecedents, although our goals are somewhat different.
First, the term resonance is found in the resonance model of reading comprehension: see, e.g., O’Brien and Myers, “Text Comprehension: A View from the Bottom Up,” and, for contextualization relative to other reading models, McNamara and Magliano, “Toward a Comprehensive Model of Comprehension.” The resonance model of reading comprehension is a processing model concerned with the way earlier parts of a text affect the interpretation and integration of later parts of a text. The idea is that the effects are mediated by a passive process that does not require active inference, but is rather seen as a type of effortless psychological resonance, representation of the earlier text automatically affecting the building of new representations depending on degree of accessibility. Like our own model, resonance is seen as gradient, although the specific factors seen to affect the degree of resonance (e.g., textual similarity and textual distance) are specific to the domain being modeled. Second is the notion of motor resonance: see Zwaan and Taylor, “Seeing, Acting, Understanding: Motor Resonance in Language Comprehension.” Motor resonance concerns the activation of parts of the brain implicated in activation of motor signaling occurring as human movement-related tasks are perceived or comprehended, processes that have been argued to be mediated by “mirror neurons.” The work relates to our discussion of mimicry, especially in section 6.4. The phenomena considered by Zwaan and Taylor are suggestive of the way in which language is “embodied,” but we do not directly discuss embodied cognition or embodiment of language in this volume.
38. Kaplan, “The Meaning of Ouch and Oops,” 2. Cf. fn. 40 in the Introduction. 39. Kaplan, “The Meaning of Ouch and Oops.” 40. Wittgenstein, Philosophical Investigations, 88, para. 241. Elsewhere, in material Wittgenstein was preparing for the Investigations, he makes a related statement, which is perhaps a passage Kaplan is referring to in his talk: “To say, ‘I have pain’ is no more a statement about a particular person than moaning is” (The Blue and Brown Books, 67). The logic of the tight connection that Kaplan draws between expressivity and indexicality mirrors the tight connection that Wittgenstein drew between the two, in remarks central to what has become known in philosophy as “the private language argument.” This is brought out in the following characteristically aphoristic comment of Wittgenstein’s, slightly later in the same notes: “The man who cries out with pain, or says that he has pain, doesn’t choose the mouth which says it” (The Blue and Brown Books, 68).
41. Kaplan, “Ouch and Oops,” 16.
42. Kaplan, “Ouch and Oops,” 5.
43. Kaplan, “Ouch and Oops,” 10.
44. A related definition has been given as a general account of the meaning of animal alarm calls by Dezecache and Berthet, “Working Hypotheses on the Meaning of General Alarm Calls.” They suggest the following definition of the meaning of an alarm call: “The particular set of circumstances in which calls occur more than expected by chance will be referred to as their ‘semantics’ or their ‘semantic domain,’ ” (114). From a philosophical point of view, their definition is clearly Kaplanian, but with the extension that they consider probability boosts rather than uniform presence across uses of a call.
45. A number of standard Bayesian and information-theoretic notions could potentially be used in a definition of resonance. For example, we could have defined the strength of resonance between an action and a feature of context in terms of the mutual information between them. The definition we give is chosen both because it is the simplest that we could conceive of, and because it sets the stage for a strikingly straightforward definition of the presuppositions of an action, although we don’t define this concept until chapter 5.
Technically, associative resonances as we define them can be negative as well as positive.
For example, a negative resonance of someone saying “ouch” would be the speaker feeling completely calm and relaxed, and a negative resonance of people greeting each other would be that they are in the process of parting. The existence of negative resonances plays no special role in this volume, and the reader can safely focus on positive resonance when we use the word resonance. However, we should note that when in chapter 6 we discuss the possibility of divergent accommodation, this could be thought of as accommodating (i.e., adapting) to selected negative resonances of the actions of out-group members.
46. A case where the frequentist interpretation would not be easily applied would be a communicative practice that is rarely carried out but that is established by some sort of fiat, e.g., a religious ritual to be carried out in a certain rare circumstance, like the Second Coming of the Messiah. In principle, this ritual practice has a meaning, but it is not straightforward to express its resonances in terms of frequency-based probability, without considering frequency counterfactually, across alternative worlds in which the Second Coming occurred.
47. Grice, Studies in the Way of Words. 48. Augustine of Hippo, Confessions and Enchiridion, ch. 8, para. 13.
49. Wittgenstein, Philosophical Investigations, 3.
50. The presumed strong link between facial expressions and underlying emotions is such that many analyses of emotions take facial expression as the primary organizing principle behind taxonomies of emotional states, a classic example of such a facial-expression-based taxonomic organization being the Plutchik wheel (Plutchik, The Emotions). Further, even the adage “The face cannot lie,” though clearly not literally true, has some psychological backing. Paul Ekman, famous for a theory of deception identification based on facial cues, comments (in a paper highly relevant to the current discussion), “When an emotion occurs, impulses are always sent to the facial muscles.
There is no choice about that. We can choose to try to interfere with the appearance of that expression, we may be able to interrupt the action of the facial muscles, or dampen them so that nothing is visible, but we cannot choose to prevent the impulses from being sent to the facial nerve.
We can also choose to make a set of facial movements which resemble a facial expression of emotion, but it will differ detectably from an emotional expression” (“Should We Call It Expression or Communication?,” 336).
51. Grice, “Meaning,” 383–85.
52. Gibson, “A Theory of Direct Visual Perception.” 53. It might be said that all communicative acts are expressive, for even when one performs the act of asserting, one must express the thought that one then asserts. We suggest avoiding this terminology and talking of “depicting a thought” rather than “expressing a thought.” A resonance of an utterance of a declarative sentence would then be the act of depicting a thought, and if this declarative is not part of a more complex sentence, then a further resonance would be the assertion of that thought. An utterance of a declarative can then be said to be primarily depictive rather than expressive (although it might additionally have some expressive resonances).
54. Lewis, Convention.
55. We return to the nature of coordination in section 8.2, and provide a fuller definition.
56. Saussure, Course in General Linguistics, 67–69.
57. Note that although tautologies cannot be resonances, both attitudes toward tautologies and actions of representing tautologies can be resonances. Thus, although it cannot be a resonance of the declarative “Mary is self-identical” that the context is one in which Mary is self-identical, it can be a resonance that the context is one in which this tautology is being depicted and that it is a focus of joint attention.
58. Clearly most signals do not serve primarily to establish their own conventions. A large class of signals that challenge this generalization is found within handshaking protocols.
These include saying “hello” to establish both English as lingua franca and availability for further communicative exchange, and computer-networking signals establishing data transmission rates or encryption standards. Yet even here, what is signaled is not the generalization that a given convention exists, but rather a more parochial fact, local to the communicative situation. Handshaking protocols usually establish which conventions are in use, not what the conventions are.
59. The notion of communities of practice was introduced in Lave and Wenger, Situated Learning. This notion was adapted and extended to the linguistic domain in Eckert and McConnellGinet, “Think Practically and Look Locally,” 461–90. We talk of communities of practice in a way that significantly generalizes, and perhaps thereby distorts, the original use. The communities that Lave and Wenger discuss are primarily local, for example, comprised of those working together in a single building or company, and their interest was in the process of learning how people become enculturated to the norms and practices in such settings. We allow that communities of practice exist not only in this local sense, but also in arbitrarily larger groups across which cultural practices are shared, for example, all those affiliated with a particular political party. 60. To be sure, our goal of developing a theory of meaning that is essentially cultural and political, a theory that centers social and emotional aspects of meaning rather than letting them play second string to an account of logical inference or truth conditions, is also not in itself original, as will be obvious to those readers familiar with the works of, say, Habermas or Foucault. Equally obvious to such readers will be that we take a quite different path than either of these figures, using quite different tools in quite different ways, and building on a range of prior scholarship that in large part was unavailable to them.
CHAPTER TWO Attunement Eye contact, although it occurs over a gap of yards, is not a metaphor. When we meet the gaze of another, two nervous systems achieve a palpable and intimate apposition.
So familiar and expected is the neural attunement of limbic resonance that people find its absence disturbing. Scrutinize the eyes of a shark or a sunbathing salamander and you get back no answering echo, no flicker of recognition, nothing. The vacuity behind those glances sends a chill down the mammalian spine. The prelimbic status of mythological creatures that kill with their gaze— the serpent-crowned Medusa, the lizardlike basilisk, hatched from a cock’s egg by toads or snakes—is no accident. These stories create monsters from ordinary reptiles by crediting them with the power to project out of their eyes what any mammal can see already dwells within: cold, inert matter, immune to the stirrings of limbic life.
To the animals capable of bridging the gap between minds, limbic resonance is the door to communal connection. Limbic resonance supplies the wordless harmony we see everywhere but take for granted—between mother and infant, between a boy and his dog, between lovers holding hands across a restaurant table.
—THOMAS LEWIS, FARI AMINI, AND RICHARD LANNON1 2.1. You Must Remember This Think of a favorite movie. Why does it resonate with you? Do you have a favorite line? Why does that line work? What ties it to its context?
Conversely, what makes it … timeless?
If you’re into old classics, you might think of the line “Play it, Sam!” from Casablanca, which was filmed eighty years ago, in the midst of the Second World War.
2 The movie is seen by some as the greatest of all time.
A version of the line (“Play it once, Sam, for old time’s sake”) is uttered first by Ilsa (Ingrid Bergmann). It carries mystery. The audience has seen the conversation from its start, and no referent has been provided for the pronoun “it.” Despite Sam’s protestations (“don’t know what you mean, Miss Ilsa”), it quickly becomes evident that Ilsa and Sam are very much on the same wavelength. To put this in the standard language of semantic theory, interpretation of the pronoun depends on common ground, and we infer that there is much common ground between Ilsa and Sam. The line is repeated by different characters, echoing in our minds, just as the referent of it, namely the song “As Time Goes By,” provides a musical leitmotif.
Together, they strengthen resonances across scenes, picking out dots between which we draw lines.
The second time we hear “Play it,” the line is delivered by Ilsa’s slick saloon-owning ex-lover, Rick (Humphrey Bogart). Rick’s rendition is the best known, though typically misquoted as “Play it again, Sam.” Although there is again no explicit antecedent for the pronoun, by this point we are well enough attuned to the characters that we know the referent immediately.
But constant it is not: the referent changes: when Ilsa’s husband, anti-Nazi resistance fighter Victor (Paul Henried) says, “Play it!” he is referring not to “As Time Goes By,” but to “La Marseillaise,” presented as a stirring symbol of free French resistance. Still, the resonances with the earlier occurrences of the line are strong. “As Time Goes By” has already been tied sentimentally to France with extensive flashback montages, and if Victor is not singing from quite the same score as Ilsa and Rick, he is nonetheless hitting shared themes. Despite their different nationalities (Norwegian, Czech, and “Drunkard,” as Rick famously muses), we see the common ground of the movie’s love triangle stars as much in their common use of language as in their common ideals, and in the united front they present against the evil of their day.
To watch the movie is, in part, to understand the attunements between the characters. But perhaps more important to our current project is the question of how viewers’ attunements change as they watch the movie.
How can we understand the way that the viewer is drawn in, and why?
Casablanca is not a pill that people swallow, or a box containing a set of propositions. That’s not how it works. It works by engaging existing attunements and providing a path for people to focus their energies and emotions, through a process of what we will term harmonization.
Casablanca has many resonances, and those resonances helped shape collective attunements both during the war and afterward, attunements to an ideology in which Americans are plucky yet indefatigable capitalists representing freedom in the face of authoritarianism, all in this together, everybody doing their bit, sometimes it’s dirty work but the ends justify the means, and in which even one of the most intensely romantic loves must be sacrificed for the greater cause of country and freedom from oppression.
It is relevant here that the effects the movie has on the viewer are not mere random happenstance, for the movie was made with a quite explicit goal of drawing in a broad range of people and helping them see the world in a certain light. It is a movie that was made in the wake of the formation of the US Office of War Information under the directorship of Elmer Davis, which had issued a call for patriotic support of the war effort from the movie industry, and which was charged with overseeing a wide range of media production, including the output of Hollywood studios. Here’s why Davis saw movies as an important part of the war effort: “The easiest way to inject a propaganda idea into most people’s minds is to let it go in through the medium of an entertainment picture when they do not realize they are being propagandized.”3 An overarching concept of interest throughout this book is attunement to ideology. We seek to describe a model of attunement that is relevant both for the strong resonances between the characters in a story, and for the ways in which something like a movie, which does not tell us what to believe or why, can be a powerful vehicle for ideological transmission. It is crucial here that attunements can change gradually. This may result from repeated exposure to propaganda or from exposure to a changed world; it is a wellworked theme in literature, history, and psychology that people may gradually become inured to events that were previously unimaginable but have become commonplace, and can adaptively develop ways of living in circumstances that they would have thought unsustainable. The gradualness of these processes implies, we think, that attunement is not an all-or-nothing thing, but that people can be attuned to something by degrees. We will be interested in this chapter and the next in the mechanisms by which degrees of attunement change. There are various sources of change in attunement, including reflective reasoning and random drift. Our interest is specifically in communication as a factor in creating, strengthening, maintaining, and destabilizing such attunements.
An ideology could creep upon an entire society gradually, without anyone in particular, even the elites, fully understanding the system they build and the ideology they propagate. At some point a country might find itself in peril, with no explicit theory of how it got there. In our view, what happens in such cases is this: the strong resonances of messaging it was exposed to (or exposed itself to) lead to it slowly becoming attuned to an ideology it cannot survive. But this is not to say that the effects of strongly resonant messaging are always so dire. Casablanca, in combination with a much larger collection of wartime messaging and educational policies, had positive effects for the war effort and for the country, effects of bringing people together, of developing common attunements.
Ideologies are cultural artifacts consisting of practices, attitudes, affect, and norms (we do not take a stand on the relative interdefinability of these notions). Assuming that ideologies are neither sourced from heaven nor innate, it must be the case that ideologies develop and are maintained through practices, including communicative practices.4 People are attuned to ideologies, and they are attuned to the people who hold to those ideologies. An ideology, in the way we use it in this book, is at its most comprehensive comparable to what Rahel Jaeggi calls a form of life: What are we talking about when we speak of forms of life? In everyday usage, the notion of a form of life refers to a whole series of extremely diverse and more or less comprehensive phenomena.
The nuclear family is a form of life from which one may try to escape with the help of alternative forms of life; the urban form of life is opposed to the provincial form of life; the forms of life in South Texas can be compared to those in Northern California.
Studies are devoted to the fate of nomadic or the decline of bourgeois forms of life. Scholars analyze the forms of life of the Middle Ages, changes in forms of life in the modern era, but sometimes we speak in the singular of the modern or medieval form of life.5 Like Jaeggi’s forms of life, ideologies, in the sense in which we employ it in this book, are “ensembles of social practices,” which are both “products and presuppositions” of our activities, both verbal and nonverbal.6 The remainder of this chapter introduces our model of attunement.
Whereas resonance plays a similar role in our account to that which content plays in other theories of meaning, attunement plays the role that cognitive attitudes like belief play in other models of conversation and cultural transmission. What we will term collective attunement will then equate to what others have termed common ground, which is usually analyzed in terms of mutual belief or knowledge. Of particular interest will be collective attunement to practices, including speech practices. However, before we can analyze collective attunement, we need to introduce the concept of attunement as it relates to individuals, which is the function of the next section. In section 2.3, we discuss attunement to practice in the context of previous scholarship on practice, building especially on recent work of Rahel Jaeggi and Sally Haslanger. In section 2.4, we introduce the notion of collective attunement, contrasting it with standard models of common ground. With this notion of collective attunement in hand, we can then come full circle, in section 2.5, resolving a puzzle introduced in chapter 1. The puzzle centered on the fact that resonances depend on the set of contexts against which they are defined; absent a way of regulating the set of contexts used, there is no fact of the matter as to whether a given feature of context is or is not a resonance. Our goal in section 2.5, then, will be to clarify the way that resonances can exist pseudo-objectively, as if floating free of any individual’s associations or dispositions. Our solution depends on recognizing that the resonances of any particular communicative practice mirror the collective attunements that a certain group of individuals has to that practice, so that we must define resonances relative to interactive contexts that occur within a community of practice.
We end the chapter, in section 2.6, by discussing some of the ways in which attunements change within a community, in the process relating attunement to a trio of central concepts in political theorizing: ideology, persuasion, and power. This sets the scene for the more detailed discussion of attunement change in chapter 3 and in the second part of the book.
2.2. Individual Attunement Attunement has the following properties: (i) it is scalar; (ii) it is externalist, allowing high levels of attunement to something for which the agent has only the thinnest of mental representations; (iii) it does not require the agent’s conscious awareness; (iv) it is not inherently propositional; and (v) it allows for groups to have attunements that individual members of the group lack. All of these properties will play a role in our account of the evolution and transmission of ideology.
In principle, a change in attunement may result from acceptance or rejection of an overt statement of some component of an ideology, but it is doubtful that exchange of explicitly stated ideological precepts is the primary communicative mechanism by which ideology is transmitted, transcended, or transmuted. Features of ideology can be communicated even when they are not what someone intends to communicate, and indeed they may be communicated by an act that was not intended to be communicative at all. More commonly, change results from observation of behavior that exemplifies or reveals practices or other aspects of an ideology; the behavior in question presupposes elements of the ideology.
But this transmission process does not always increase attunement to existing ideologies.
The process we are describing does not simply reproduce ideology like a cultural Xerox machine. There are three reasons for this. First, the production of behaviors never perfectly exemplifies a broader practice, whether by intention or by accident. Second, a finite number of examples can never provide complete information as to the nature of the practice that they (imperfectly) exemplify. And third, when observing someone perform an action, the resulting change in an observer’s dispositions never consists in suddenly gaining a tendency to behave exactly as the performer did, even if that should happen to be the observer’s goal.
In the case of ideological conflict, people may even seek to behave in a way that is as distinct as possible from the performance they observed, a process that, as we will see, has been much studied in communication studies and sociolinguistics.
People can be attuned to many different things, including power tools (if they know how to use them), the weather (e.g., by adjusting their wardrobe), and Catholicism (an ideology).
Here’s a working definition of attunement: Attunement: An agent is attuned to something to the extent that their state and behavior predictably evolve in accordance with its presence in the agent’s context.
So, this notion of attunement is gradient, relating an agent to something else, which we may refer to as the object of attunement. Attunement is defined in terms of three further notions: predictability, accordance of state and behavior, and presence in an agent’s context.
Predictability in the definition of dispositional attunement is inherently correlational, the degree of dispositional attunement being a measure of the degree to which the agent’s behavior is correlated with the presence of something being in the context. However, the notion of predictability here is inherently more complex than that assumed in our definition of resonance in the last chapter, because we are now talking not simply about whether two things co-occur, but about whether a complex behavior is responsive to something’s presence.7 We must also say more about the notions of accordance of state behavior and presence in the context. As regards presence in the context, let’s consider the examples mentioned above. There is an obvious enough interpretation of presence for power tools: presence in a context could mean presence of an instantiation of the type in the physical environment of the individual on whom the context is centered. But we might also interpret presence in the context to mean, less clearly, presence of the concept of power tools in the cultural context of the agent—that the agent’s assumptions and practices contain or reflect that concept, for example. Both interpretations are relevant for us, and we will return to a similar ambiguity (regarding the question of when a practice is present in the context) in the next section. As regards dispositional attunement, it is not of import whether the concept of power tools is present in the cultural context; it will suffice to consider whether practices are present in an agent’s context, practices that may involve particular physical artifacts.
For weather, we can take a similar tack as for the first interpretation of the presence of a power tool: weather is present in the context when an instance of it (say, light rain) is present in the immediate physical environment of the agent. If we want to speak instead of an ideology being present in a context, for example, Catholicism, we will need to think of contexts as cultural.8 We have relatively commonsensical intuitions about when Catholicism is present in a context: it was not present in the contexts of agents in pre-Columbian South America, and some version of it is present in the context of many of the people of this area now. We will take presence of an ideology, such as Catholicism, to be best analyzed at the level of communities of practice, so the question of whether Catholicism is present in a context becomes a question of social affiliation: Catholicism is present for an agent if the agent is a member of a community of practice within which the specific practices of Catholicism are common (which does not necessarily mean that the agent themselves is a practitioner).
What is it for state and behavior to accord with the weather? Our characterization should be general enough to be neutral between different ways of attuning one’s behavior to the weather. For example, someone might subscribe to a religion according to which one should expose oneself to the elements the worse it gets—in the cold, one should remove one’s jacket, for example. This is a way of being attuned to the weather—as is the more standard way of adding clothing as the temperature plunges. Both of these are different ways of acting in reaction to changes in the weather.
How about power tools? Although someone who uses an electric sander to prop a door open acts in a way that is dependent on the object, the action of propping a door open is not one that is distinctively accomplished using an electric sander as opposed to some other similarly weighted object. A power tool has a range of potentialities, of which those lent by its weight alone are a small subset. Let us say that one behavior accords better with a power tool than another behavior does when the first depends more highly upon the potentialities of the object that distinguish it from other objects.
However, since power tools are cultural artifacts, there is more to say.
Using an electric sander as a doorstop is acting in accord with (the presence of) the sander, but not to the same extent as using it to smooth wood, which relies on potentialities unique to sanding equipment, and, again, on standard practices.9 The potentialities of such an artifact, and indeed the potentialities of any natural object that takes on an ideological role, depend upon the culture. Acting in accord with the potentialities of the object implies being disposed to utilize the object as an affordance in the sense of Gibson, meaning something that serves a function supporting an activity, so that a seat affords sitting and a pen affords writing.10 Thus acting in accord with an object involves performing behaviors (the behaviors afforded by the object) that themselves may be culturally transmitted. So, attunement to the instrument will imply attunement to practices involving that instrument. Those practices may themselves have psychological and social dimensions that are connected to the physical practices. Power tools might have the potential to make some feel manly, and demonstrate to others that they are not confined to their office jobs. Yet the complications with power tools go deeper than the fact that the potentialities of a power tool are in part culturally defined. The deep problem with talk of acting in accord with power tools is that absent culture, there is no such thing as a power tool. The very concept of a power tool is culturally bound, and the determination of whether something is or is not a power tool is not culturally neutral. So, it’s not the case that we can talk of someone’s attunement to power tools independently of their attunement to other aspects of culture. Attunements to a cultural artifact bleed into attunements to other aspects of the prevailing culture in which attunement to the artifact was attained. Learn to do something geeky, and you are likely to become better attuned to geek culture; learn to do something manly, and you are likely to become better attuned to male culture. As Jaeggi observes, “Practices are always ‘practices in a nexus.’ ” 11 In our terms: it is both the case that practices are resonances of other practices, and that they are objects of collective attunement in combination, so that attunement to one practice tends to imply attunement to many others. The resonances between various objects of attunement are central to an account of ideological maintenance and change.
Central to this chapter is the development of an account of dispositional attunement to practice. But as we’ve said (uncontroversially, we take it), there’s more to an ideology than a set of practices. To understand attunement to ideology, we also need to understand affective attunement, cognitive (or attitudinal) attunement, and perceptual attunement. All are evident in the case of Catholicism.
Attunement to the ideology of Catholicism clearly involves affective attunement—feeling the same emotions as others in the community would in the relevant places and times. Without speculating on the nature of spiritual states, or their role in different varieties of Catholicism in particular, it is hard to imagine spiritual states not involving emotions of some sort, whether feelings of calm and well-being, or feelings of love or ecstasy. The central practice of prayer in Catholicism is associated with feelings such as adoration and humbleness, and affective attunement is also expected for the artifacts and other physical entities that support the practice and ideologies, for example, the Bible, the cross, and inspirational depictions.
Affective attunement is, furthermore, central in the formation of in-group affinity within and across Catholic communities (i.e., positive feeling in the presence of in-group members, in-group practices, in-group ideology). What we are calling affective attunement, that is, certain patterns of emotional response, is not generally explicitly discussed in the theory of meaning—but it is starting more recently to emerge in theories of presupposition that seek to explain phenomena also of interest to us, such as slurs, as in the work of Teresa Marques and Manuel García-Carpintero.12 That Catholic ideology involves cognitive attunement, in particular the sharing of certain characteristic beliefs (the central case in the theory of meaning), is obvious. Catholicism is, after all, a faith. But the central tenets of the religion find their place in a broader web of attitudes, for example, beliefs about the way one should lead one’s life, the respect different people in and out of the religious community are due, and beliefs that involve Catholic perspectives on everything from history and art to food and wine.
It’s not that there’s one set of attitudes on these issues that all and only Catholics share, but rather that there are tendencies for Catholics to think in certain ways that are not so commonly shared among non-Catholics, and that are distinctive enough to be described as Catholic attitudes. Note here that the phrase “think in certain ways” bespeaks the fact that different types of attunements are not cleanly separable. We do not have one set of cognitive attunements and a separate, orthogonal set of dispositional attunements. The terms cognitive and dispositional characterize aspects of attunements, rather than strictly different types of attunement. Thus, to “think in certain way” reflects an attunement that is both dispositional and cognitive. Dispositions to reason and to judge necessarily have this character.
Last in our list of Catholic attunements, to be attuned to Catholicism is also to be perceptually attuned to it, to experience the world in a Catholic way. Catholics might perceive the world through different eyes than nonCatholics because they have different perceptual categories for events and objects they witness, and possibly also because at mass or elsewhere they might have (to borrow from William James) “varieties of religious experience” to which non-Catholics lack access.13 2.3. Attunement to Practice The question of what acting in accord with a practice means varies enormously, from cases like driving where there are institutionalized methods for determining whether your behavior is legal, to cases like fashion where the fashion police aren’t wearing badges, at least not this year.
14 What is relevant here is that Catholicism, driving, and fashion all involve, among other things, a host of practices. We must say more about attunement to practices, how that relates to a broader ideology, and about other types of attunement, going beyond attuned behavior. But before going on we must note that, like many before us, we were initially inspired by the work of Pierre Bourdieu.15 A theme of Bourdieu’s anthropological and sociological work is that practices as cultural objects can be separated from the intentions of practitioners. This means that reproduction and innovation of practice do not need to be the result of conscious intention, and, from the perspective of an outside observer, that practices can be studied at least somewhat independently of what practitioners report as their intentions, if such reports are even available.
Bourdieu’s development of a practice-based model that is not dependent on intention meshes naturally with the well-established idea that much of what is communicated is to a great extent left tacit or implicit, a web of presuppositions that are reflected in what is said, but is not completely explicit, and with the further idea that there is often no intention to communicate these presuppositions. As we will discuss in chapter 9, whether practices are taken to be performed intentionally or not, an approach that centers on intention and the recognition of intention is not likely to be well suited to the study of what we have called hustle, namely, that which is either communicated without intention, or is communicated without the recognition of communicative intention. Our model of resonance allows that the resonances of words exist at a community level, rather than at the level of individuals, and that these resonances can be significant in communicative interactions even when none of the conversational participants are aware that those resonances are in play. That is, the resonances of a communicative practice can affect change in attunement absent any intention for that to occur.
Comprehensive coverage of the voluminous literature on practice would be impossible here, even if we could claim to be fully versed in sociological and anthropological work on the topic, and we will not even attempt to connect to most of that literature.16 We do not discuss the well-developed methodologies by which actual practices are studied. We will focus on Jaeggi’s theoretical account, noting connections to parallel developments in the work of Sally Haslanger.
17 Being attuned to a social practice means having dispositions that result in behaviors that match the practice. But what is a social practice? Jaeggi introduces her discussion of practices as follows: Practices in the most general sense are complex activities in which we engage alone or with others. Examples of practices are lining up at the checkout when shopping, making a bank transfer, inviting friends over for dinner, throwing a party, playing basketball, playing hide-and-seek with children, conducting a seminar, and taking an exam.18 Jaeggi describes the essential features of her account of practice as follows: “Practices are habitual, rule-governed, socially significant complexes of interlinked actions that have an enabling character and through which purposes are pursued.”19 We illustrate Jaeggi’s multipart characterization with a particular kind of practice, in fact a particular speech practice, the practice of racist dog whistling, as developed through the Republican “Southern Strategy” discussed in the introduction. This is the practice of using terms like “welfare” or “inner city” for a political goal, say, to increase support for cutting taxes on the wealthy (say, by describing taxes as going to support “welfare programs”).
Habituality: The practice of racist dog whistling in this way had to be established, by repeated and habitual performance (think of Reagan’s campaign against the “welfare queen”). Rule-governed: To engage in the practice of dog whistling involves describing programs to be attacked using certain labels; the practice has rules, or at least has sufficient consistency in its form that observers can identify a categorical similarity across different instantiations of the practice. Philosophers and linguists tend to use the word “rule” in a nuanced way that does not necessarily require the presence of an explicit statement or requirement. This is certainly the case for Jaeggi, who indeed goes further, and allows that some rules governing a practice might not even permit of explicitation in principle.20 Sally Haslanger cautions against practices necessarily being frameable in terms of rules,21 arguing rather that descriptive normativity is required for a regularity to be a social practice: Members of a group take the culture’s concepts, scripts, and meanings to be normative for members of the group in the following sense: when encountering others who are similarly socialized, we implicitly begin with the assumption that they will do things in a particular way, taken to be the “right way.” We may be surprised or feel entitled to criticize them if they don’t.22 Despite this disagreement on whether rules are central to practices, we find that on this issue as on most we consider here, Jaeggi’s analysis and Haslanger’s are broadly consonant.23 The disagreement seems to relate more to the nature of rules than to the nature of practice; given that Jaeggi suggests that rules can exist without practitioners being conscious of them, and in some cases without it even being possible in principle to make the rules explicit, it is not clear to us how much light could be shone in the gap between Jaeggi and Haslanger’s accounts of practice on this issue. Jaeggi does not present an analysis of what rules are per se (beyond taxonomizing into different sorts and functions of rules), but what is clear from her discussion is that the crucial distinction between following a rule and performing a non-rule-based action is normative. Jaeggi explains that mere habits are not in this sense normative, because breaking a habit is not in and of itself something that opens one to criticism.
For example, if one is in the habit of putting pepper on one’s food only after adding salt, then switching would be a change in habitual behavior, but not a normatively problematic change in dining behavior that would open one to criticism. On the other hand, dining practices are more broadly normatively constrained. There is extreme cultural variation as regards, for example, the appropriateness of belching at the table, or the constraints, if any, on what counts as a finger food, and dining behaviors are often actively regulated, with overt acculturation of children and foreigners who lack the local mores. Someone’s dining habits may be in conflict with community norms, lacking the proper etiquette, but any habits that are not perceived as themselves conforming to a societal regularity or maintained through societal regulation are not practices in the technical sense that we will assume.
As regards normativity, the practice of dog whistling is an interesting case, for it makes clear that there is a complex relationship between the norms governing practices and what is ethically desirable. There are communities on the political right wing that presumably think that acts that the current authors would describe as dog whistling are morally justified, whereas liberal academics, politicians, and journalists would not. The same communities who are among those most likely to label an act as dog whistling, and most likely to try to pinpoint the essential features of dog whistling justifying such a label, are those who take the activity to be problematic and who are also likely to try to restrict it. Similar considerations apply to all practices that we might term “hate speech” and to other communicative practices that simultaneously embody some community norms and controvert others. Sometimes controversial types of speech morph precisely because of these joint pressures, as when a slur or taboo word occurs in a slightly disguised form (“egad” and “zounds” being historical examples of the morphing of practices). Quite generally, norms that circumscribe the judgment of what counts as an instance of a practice are logically distinct from norms about whether performances of the practice are just or desirable, but not entirely separable.
As should be clear from the above discussion of rules, or maintenance of practice even if not explicitly regulated as such, the norms that constrain and define practices are taken by both Jaeggi and Haslanger to be understood at the level of communities, and not at the level of individuals.
Haslanger’s goal is explicitly to analyze “social practice,” and it is a phrase that Jaeggi also uses on occasion. What Jaeggi describes as the “rule-governed” nature of practices is interwoven with what she describes as their social significance.
Social significance: The norms that constrain and define practices are intrinsically social, regulated at the level of communities or societies, not by individuals.
Practices of dog whistling are socially significant both as regards the cultural background they rely on and as regards their function. Clearly, dog whistling can only be understood with the background of the social construct of race and its particularity in the American context.
Within this practice, to describe programs (say) as “welfare” is to code them as strongly negative, associating them with increased laziness, for example, and drug use, all elements that tie in with a manifestly biased, socially constructed racial stereotype. The function of the practice is manifestly to drive political alignment on the basis of acceptance and use of such constructs.
Haslanger’s and Jaeggi’s characterizations link social practices together with group identity, in a way that our future discussion will exploit. To have a group identity is, minimally, to share in a set of social practices, and it may also require other elements, such as shared persona (in the sense of public manifestation, or comportment) and common ancestry or heritage.
Our discussion of collective attunement in the next section, and the associated notion of a community of practice, both relate to group identity.
We will briefly consider in chapter 3 the separate and in some ways stronger notion of identity fusion, which implies that people not only share various traits or practices, but see themselves as part of a common whole, to some extent merging personalities and developing a rich sense of collective agency.
We take it as definitional that the communicative practices we study in this volume are social practices, and that the politics of language is also definitionally social, but it should be noted that this is not quite the same as claiming that language is itself intrinsically social, a claim that some have denied.24 Enabling character: The enabling character of practices refers both to the idea that practices often depend on other practices in order to even be possible (so, e.g., no amount of celebration would count as a wedding anniversary celebration absent the broader institutional practice of marriage), and to the idea that performing a practice causes changes that allow further practices to take place. The practice of dog whistling has an enabling character in both senses.
First, it is obviously enabled by a substantial linguistic background (as well as a social and political background), without which dog whistling would just be making sounds or squiggles on a page. Second, it reinforces unconscious racial bias, thereby enabling other practices of treating Black Americans as subordinate, for example, by discriminating in social activities such as job application or voter-rights support.
Purposes: Practices serve a function which can be understood as the reason to perform them. The practice of dog whistling has clear purposes: to marshal unconscious racist bias toward political goals, such as cutting taxes for the wealthy, and destroying institutions or individuals to which one is opposed (e.g., opposition candidates or parties, the federal bureaucracy, the Walt Disney Company). Dog whistling has a purpose, in that it has predictable effects, would not exist if it did not have those effects, and certainly can be and often is carried out with the intention to produce those effects. However, Jaeggi is circumspect in her discussion of purposes, allowing that practitioners may not be aware of the purpose of a practice they engage in.25 So, it is consistent with her view, and ours, that someone might dog whistle without intending to, and without intending to have the dog-whistle effects, such as reinforcing unconscious racial bias.
With these defining characteristics of (social) practices as background, we can now say more about what it means for some individual or group to be attuned to a practice. Simply plugging practice into our definition of dispositional attunement, we arrive at the following (which is simply a special case of the existing definition, not a new definition as such): Attunement to practice: An agent is attuned to a practice to the extent that their state and behavior predictably evolve in accordance with its presence in the agent’s context.
Instantiating the definition of attunement in this way leads to a type-token ambiguity briefly foreshadowed in our discussion of attunement to power tools above. At the type level, a practice can be present in a cultural context in which people tend to perform the practice, but at a token level, a practice can be present in a more temporally and physically localized part of the context in the sense that it is being actively instantiated, that is, performed, at a given time and place. That is, people can both be attuned to practices, and be attuned to particular instantiations of the practice. The latter notion, of being attuned to instantiations of a practice, will largely suffice for our purposes, but there is value to considering attunement at the type level too.
Let us first consider an attunement to particular instantiations of a practice, and then return to consider the type level.
What are the tokens of a practice? This is a difficult question, relating closely to the subtle tension between the aforementioned views of Jaeggi and Haslanger regarding whether there are rules governing practices.
Sidestepping the philosophically vexed question of what rules are, and hence whether practices can be defined by them, let us consider the tokens of a practice from a purely extensional point of view, as a certain type of regularity of behaviors. Rather than trying to specify in general terms when a regularity of behaviors constitutes a practice, let us simply assume that there are such regularities, and that the occurrences of individual tokens would be recognized by members of a community as belonging to the category. Note here that recognizing neither implies conscious recognition, nor does it imply a tendency to publicly label the actions as belonging to the practice (a metapractice), although it is common for practices both to be consciously recognized and publicly labeled or at least labelable. What we will call the extension of a practice is just such a set of recognizably like behaviors.
Sometimes, but not always, these regularities are societally regulated, and occasionally they are even regulated through a separate formalized practice, perhaps bureaucratic or legal, but in focusing on the history of a practice, we generalize to a level that includes both regulated and unregulated practices.
The notion of an extension of a practice is metaphysically minimalistic relative to the denser and more theoretically substantive notions of practice itself offered by Jaeggi and Haslanger, and yet the extension of a practice is already a rich notion to work with. In recognizing a practice, community members must have access both to the context in which the behavior occurred, and to sufficient information regarding prior instantiations of the practice and the contexts in which they occur. Contexts are rich and include such features as the attunements of any participants in the practice. So, although the extension of a practice is minimalistic, it manages to be simultaneously just a form of behavior, and much more than that. For example, the extension of the British practice of drinking tea is simultaneously a bland pattern of physical actions involving a kettle, water, and other mundane trappings of British life, and also a set of rich contexts in which these actions take place, cold winter afternoons, moments after bad news, a favorite mug, the anticipation by the drinker of the first hot sip, a sense of well-being, and all the attitudes that make mundane trappings mundane to the participants in an event of drinking a simple cuppa, unremarkable to those whose lives they permeate, and yet paradigmatically defining of an entire society.
Drinking a cup of tea is a fleeting episode in a day, innocuous and commonplace. Yet it is woven into the fabric of society—into a nexus of other practices, as Jaeggi would have it—in such a way that it cannot be cleanly separated from other activities that co-occur, like the exchange of social niceties while enjoying the tea, or the practices that support tea drinking, such as dairy and tea production, water and power supply, and practices of mercantile exchange. Activities of all these sorts are present in the contexts of individual episodes of drinking tea. The extension of the practice of drinking tea is at once almost nothing and nearly everything.
For many practices, the extension is precisely what community members have access to as they acquire the practice, and indeed as they further shape the practice. Or rather, community members typically have access to a thin slice of the extension, a small subset of occurrences seen from a particular angle, with a patina of context. A toddler who witnesses adults drinking tea a few hundred times at a safe distance above where they play has access to almost nothing of the adults’ experiences and attitudes as they sip and chatter.
And yet almost nothing makes for a good start in life. A theme we shall return to is that people shape or extend practices over time by performing them idiosyncratically, and yet with sufficient similarity to prior occurrences that the new performance is seen as a novel instantiation of the same practice rather than as a completely separate practice. Whether a British three-year-old’s imitation of tea drinking is actually an instantiation of the practice may be a little unclear to an observer, but soon enough most children will probably be so well attuned to tea drinking and its nuances that they perform the action without thinking twice, and with very similar attitudes and experiences to the adults who once surrounded them as they played. Yet other children will go on to drink tea in a slightly different way, with less sugar, say, or perhaps by staring at a screen instead of at another human while they drink. A practice steeps and develops new flavors.
Given that a practice has an extension that consists of individual instantiations of the practice, and given that a context is just reality centered on some individual at a particular time and location, we can easily state what it is for a token of the practice to be present in a context. It is simply for the relevant individual at a given time and location to experience the practice over an interval including that time and location, whether as an active participant or observer. The practice consists in a regular pattern of changes of contextual parameters over time, including parameters that concern the activity of participants and the attitudes of observers. To predictably act in accord with tokens of the practice is to be an individual whose behavior has a tendency to evolve in the way that is recognizable as a role of the practice, exemplifying the category given by the history of practice (and in contradistinction to categories of behavior identifiable as distinct and incompatible practices). It is, then, fairly obvious what it means for your behavior to predictably accord with the presence of the British teadrinking practice in your context. Sipping a mug of hot tea made with a teabag of dusty Assam and Darjeeling varieties and with a spot of cow’s milk is recognizably acting in accord with the standard British practice, but slurping sweet green watermelon-flavored tea with coconut milk and tapioca bubbles from a plastic cup through a giant straw is not.
Dunking a biscuit (i.e., a cookie) into the tea fits with the practice, but dunking a croissant, celery, or your finger would not.
Put in terms of the last chapter, the resonances of the British practice of tea drinking include a positive probability boost for cow milk and for biscuit dunking, but not for tapioca bubbles, big fat straws, or the dunking of anything other than a biscuit. Acting in accord with the practice is having an increased tendency to do things for which the resonances provide a positive probability boost when participating in the practice. Not acting in accord with the practice, contrarily, would involve having an increased tendency to do things for which the resonances include a probability decrease.
Let us briefly return to the type-token ambiguity mentioned above. We have focused on attunement to a practice at the level of tokens in the extension of the practice. However, metapractices by which people regulate practices, communicate about the significance of practices, and sometimes consciously innovate practices, are set at the type level.
Attunement to metapractices, say metapractices of maintaining the practice, can involve communicative actions about the practice being maintained at a type level.
For example, a metapractice of maintaining tea-drinking practices, say by explaining that the milk should be put in the cup before the steeped tea is poured from the pot, involves communication about tea drinking at the type level. Metapractices may themselves occur in the immediate contexts of tokens of the practice, as when someone comments on proper tea-drinking practice during an episode of drinking tea, which means that the metapractices themselves can have a boosted probability in the resonances of the practice. Furthermore, since participants and observers of tokens of a practice will tend to be among those who engage in metapractices, even if they do not do so at a predictable time relative to their participation in or observation of the practice, attunements to metapractices will themselves tend to be resonances of a practice. Nonetheless, we suppose that many facets of practices discussed by Jaeggi and Haslanger are best understood at a type level, not only as regards regulation of a practice when rules and regulations are identifiable, but also as regards social significance and purposes, in Jaeggi’s sense.
Attunement to a practice at a type level also buys us something that is not available at the token level in any obvious way. Attunement to a practice at a token level ensures that an agent will tend to act in accord with a practice when an instantiation of it is present in the context, but this does not in itself establish that an agent will initiate the practice when appropriate. To be dispositionally attuned to the practice of tea drinking is not merely to act appropriately when drinking tea, but also to have a tendency to drink tea when that would be appropriate, that is, to actively introduce tea drinking into the context on certain occasions, such as teatime. At the type level, a practice is present for an individual when they are in a community among which participating in instantiations of the practice is common enough to be recognizable as a practice of that community, where by “in” we do not mean necessarily being a member of the community, but also include the broader sense of being among members of the community.
This is roughly the sense of the proverb “When in Rome …,” which concerns the idea that when one is in a location one should perform local practices, even if one is not actually a member of the local community (i.e., one is in Rome, but not romano/romana, or, for that matter, Roman).
Although we omit the math, it is a simple matter to use Bayes’s Theorem and the resonances of a practice to calculate the set of occasions on which the practice itself has a boosted probability, and thus, when practitioners should be expected to initiate it. An agent who is attuned to the practice at the type level will be one who, when in a context for which the practice is present in the sense of being regularly practiced, not only has a tendency to act in accord with instantiations of the practice, but also has a tendency to initiate the practice in accord with its resonances.
Our focus on the extension of practices as histories of behaviors borrows from work in the psychology of categories and concepts over more than forty years, work heavily inspired by Wittgenstein’s critique of rule-based categorization in his Philosophical Investigations. One prominent line of psychological work on categorization is exemplar theory, although, fittingly, it’s perhaps more of a cluster of approaches than a line of work.
Exemplar theory emphasizes the process of structuring data into groups, that is, exemplars, and views categorization as being heavily influenced by comparison with some or all members of those groups. So, these models can be seen as being relatively extensional models of conceptual categories, more focused on category building as structuring of data than as development of rules, or, for that matter, development of abstract prototypes against which new cases might be compared. While we take such work on categorization to be relevant to the categorization of practice, which is itself essential to our core notion of attunement to practice, we make no claims about psychological representation of practices, or of events more generally.
Ours is not a model of categories as psychological exemplars, but as cultural exemplars.
What we claim is that for the practices of most interest for the present volume, there need be no community-accessible representation of the practice that has priority over what people actually do, beyond collective memory of what has occurred (or, to put it more precisely in the terms of the next section, collective attunement to what has occurred).
While we make no commitments about psychological representations, we do take the nature of psychological representation to be important for the development of practice. For our immediate purposes, what is important is not exactly how the human mind works, but rather the fact, or so we assume, that there are strong commonalities in the way that human minds operate. It follows from this assumption that humans will share common ways of developing attunements in response to similar exposures to behavior, and this in turn must contribute to the development of stable practices in human communities.
For better or worse, modern “big data”-based natural-language processing is focused on models that start with an enormous repository of practices, and then try to reproduce those practices.26 The field of machine learning is dominated by a highly empirical methodology in which systems are evaluated primarily on the basis of their ability to predict new instances of phenomena based on exposure to large numbers of prior instances. In this methodology, relatively little weight is given to whether the system’s internal representation of a phenomenon is congruent with what we might think of as an intuitively or scientifically correct model, except insofar as our ability to understand those internal machine representations itself supports scientific and engineering progress in the development of systems that achieve even greater levels of empirical performance. That is, a majority of contemporary computer scientists, unlike more cognitively oriented AI specialists of bygone years, don’t model the cognitive or affective state of humans performing a task; they just try to produce a system that has the right dispositions. In case they are trying to make a system appear human, they model human dispositions. When modern computer scientists build machines to perform the practice of answering questions, what they aim for is the system that most effectively replicates features of a data set of examples of humans answering questions, a data set that is taken to be a large enough sample of the practice that it can be assumed to be representative of it. A deep problem that we won’t discuss in detail here is that the assumption of representativeness of a big data sample is often problematic, an unwarranted assumption leading to troublesome bias. Nonetheless, what we are saying can be put this way: maybe such a data set, if it really were representative, and if it really did encode context in a rich enough way, would be all that was needed to pin down what matters about the practice. That is, with a large enough set of exemplars of a practice, and enough context around them, one could, in principle, identify all the resonances of the practice, and hence all the information that would in principle be needed to model attunement to that practice.
The notion of attunement to a practice that we have so far described has involved only behavioral dispositions. If someone watches professional wrestling all day, it seems natural to say that they will become highly attuned to it, even if they do not thereby become disposed to body slam others over the ropes of a wrestling ring. They need not be behaviorally attuned to the practices performed by wrestlers. Someone could regularly fail to behave in accord with practice, even in the appropriate circumstances (dropped into a professional-wrestling ring, and still be hopeless at the practice), and yet be able to recognize the practice, understand it, and get excited by it. In such cases, there are perceptual, cognitive, and emotional attunements without behavioral attunement. These are all valid notions of attunement, but they are not the notion of behavioral attunement that is a central focus of this chapter.
Dispositional attunement can also be more complex than in the cases we have considered.
An agent’s behavior may be adapted to the dispositions of others who follow a practice, even though the agent does not follow the practice. Examples would be somebody who avoids church on Sundays because it’s too crowded, or a hunter who locates themselves (whether consciously or instinctively) within range of a watering hole at dusk, being thus adapted to the prey’s practice of drinking then without necessarily following the practice themselves. We may speak in such cases of a secondorder attunement to a practice, that is, an attunement to others’ attunement to a practice. In that case, all the other cases of dispositional attunement to a practice we have considered, cases that involve a tendency to perform the actions that we might say constitute the practice, are first-order attunements. This distinction will be useful when we introduce the notion of collective attunement in the next section. 2.4.
Collective Attunement and Common Ground A standard story about communicative interactions, seen most obviously in David Lewis’s and Robert Stalnaker’s work,27 focuses on the function of conversation to increase the store of shared information of the interlocutors.
According to the story, information increase is a very public process. The shared information, what is known as the common ground, is by definition available to everyone involved. We are invited to think of the common ground in any conversation as like a big old scoreboard that hangs over the far end of a baseball field, except that instead of telling everyone the state of a baseball game, it tells us the score in a conversational game. The score determines what information is mutually believed about the world, as well as facts internal to the conversation, like whose turn is next.
There are some laudable aspects of this story, well known to philosophers, and some less laudable ones. The (Wittgensteinian) game metaphor is laudable. But the scoreboard metaphor is problematic. At the very least, communicative interactions that transmit ideology do not typically do so by displaying the ideology they transmit in ten-foot-high letters with 10,000-watt illumination on a stadium jumbotron. This aspect of the metaphor interests us mostly as a symbol of the way that the superficial topic of a conversation can distract from what is going on underneath.
We now develop a way of thinking about common ground in terms of attunement. This will provide an important building block for part II of the book, where we distinguish between the overt move made by a conversational act, and the presuppositions of the act. Presupposition will be central to our account of ideological transmission and conflict, as well as to analyses of linguistic phenomena involving social meaning and context dependence.
An attunement-based notion of common ground will have three properties that are either not available or not usually utilized in standard accounts of common ground. First, the standard notion of common ground is as something like a set of propositions, whereas, as discussed in the previous section, we allow attunement to various things that on most views are not propositions, for example, power tools, the weather, and Catholicism. Second, standardly the common ground, like the scoreboards of Lewis’s youth, is black and white: a given proposition is either in the common ground or not; in contrast, attunement is a gradient notion, so that the degree of attunement can vary continuously. Third, whereas the common ground is usually conceived of as a complex attitude, we conceive of it in terms of the dispositions of agents, specifically, their dispositions to interact with each other.
28 How can an account of individual attunement to context be extended to an account of how groups of individuals are attuned? Just treat groups as individuals! It turns out that this seemingly fatuous answer has significant benefits. Let us first state the definition and then explain how the approach differs from standard approaches and why that is a good thing.
Collective behavioral attunement: A group of agents is behaviorally attuned to something to the extent that their collective behaviors predictably accord with its presence in the group’s context.
Interest in the definition depends on there being nontrivial collective behavior, behavior that goes beyond the sum of individual behaviors. We take it that some inherently joint behaviors are not naturally analyzed at an individual level, for instance, having a conversation, moving a piano, playing a team sport, performing mass protest, and oppressing minorities.
Given the existence of such collective behaviors, our definition simply mirrors the earlier definition of attunement for individuals, and directly leads to a notion of common ground.
This notion will be useful in framing our account of how communicative actions undergird ideological change, but it will also help make sense of our account by allowing a straightforward juxtaposition to a more standard view.
Common ground: The common ground of a group is the collective attunements of that group.
Each collective attunement has a certain strength, so that the common ground of the authors might involve a strong collective attunement to certain practices of arguing with each other, and a weak collective attunement to the price of fish. We do not here attempt to provide a quantitative measure of the different levels that collective attunement might have, but the fact that we assume that there is some measure guarantees various intuitive properties, like transitivity: if a group is attuned to X more than Y, and Y more than Z, then it will be attuned to X more than Z. For current purposes, it will suffice to know that we take collective attunement to be a partial order, whereby in some cases it is possible to say that a change resulted in an attunement becoming stronger or weaker than some other attunement, without specifying any quantitative level.
As will become clear, our use of a scalar notion of common ground is essential to our model, but to make it easier to think about the common ground, and to enable more direct comparison to standard approaches, we can approximate the common ground as a set: Common ground (discrete approximation): The collective common ground is (approximated by) the set of things to which the level of collective attunement of the group is high.
Here “high” might be defined, for example, as high enough that the attunement would be manifest to an observer, such as someone randomly plucked from the streets of New Haven who entered into a lengthy conversation with us.
Our view builds on the standard notion of common ground, and we now explain why the differences matter at an intuitive level. We will then outline reasons to think that our reconceptualization retains the efficacy of earlier accounts in dealing with a cluster of much-studied linguistic phenomena involving presupposition. This will set the stage for introducing our approach to social meaning and ideological change.
In a series of classic papers, Robert Stalnaker defined the common ground of a group to be equivalent to the set of propositions that members of the group mutually believe (or accept).29 He models this set of propositions in terms of the set of possible worlds in which the propositions hold, although nothing in our discussion hinges on the use (or non-use) of possible worlds. This notion of common ground was also assumed by David Lewis and has become the standard for subsequent scholars. Stalnaker used common ground as the basis for an analysis of presupposition and assertion.
In Stalnaker’s view, the (pragmatic) presuppositions of a speaker making an assertion are all the propositions in the common ground between speaker and hearer as the act commences.
The effect of the assertion, once accepted, is then to change this set. In the cases that Stalnaker considered, the change consists of a strict increase of the common ground to include the proposition asserted by the speaker. Equivalently, an accepted assertion yields a strict decrease in the set of worlds in the common ground, by removing from the prior common ground any possible worlds incompatible with the proposition newly asserted.
On an idealized version of the standard Stalnaker-Lewis view, the common ground of a group has the following features:30 1. Propositional: the common ground can be thought of as a set of propositions; 2. Discrete: a proposition is either in the common ground or not; 3.
Highly accessible: it is scoreboard-like, equally apparent to all members of the group; 4.
Intersective: a strict subset of beliefs held by individuals; 5. Incontrovertible: the group cannot disagree about its own common ground.
We are confident that collective attunement can do everything that common ground does in standard semantic and pragmatic theory, in part because standard common ground is a special case of collective attunement. Yet collective attunement generalizes enormously. In a nutshell, the difference between the standard notion of common ground and the variant we are proposing is that our attunement-based notion has none of the above five properties.
First, a system of collective attunements is not propositional. It is not propositional because we can be attuned to things and people, as well as practices and power tools. We can be attuned to them dispositionally, perceptually, emotionally, and cognitively. For example, we can be disposed to perceive something in a certain way. Seeing a certain person as your partner rather than as a stranger, involves attunement to that person. That attunement is not naturally thought of as attunement to a proposition.
Likewise, the emotional attunement of loving someone is not an attunement to a proposition, but to the person. A crowd’s adoration of a preening demagogue involves a collective emotional attunement to that individual, but again what they are attuned to is not propositional. There may be reasons for their adoration, and it may be that the adoration can be explained by the fact that the demagogue is, or appears to be, in possession of various traits, but that does not make the fact of possession of those traits into what is adored. We can describe attunements and their objects using propositions, and then reason about them, but the fact that we can describe them using propositions no more makes the attunements propositional than saying “it was a hard edible fruit of the species Malus domestica that fell onto Newton’s head” makes the apple into a proposition. Propositions can’t fall onto people’s heads.
Second on our list is the discreteness of the fact as to whether a particular proposition is in the common ground or not, as against the way that attunements can take values on a continuous scale of strength. An individual or group can be more or less attuned to a charismatic leader, to the fact that he is lying to them, to the thugs in the street, or to the practices of the leader’s followers. The assumption that things are categorically either in or out of the common ground is an idealizing approximation. It is available to us, not a core part of the model (and neither do we take it to be desireable).31 As regards the high accessibility of common ground in the LewisStalnaker model, it is unclear how seriously to take Lewis’s scoreboard metaphor. Certainly, many philosophers, most obviously Stalnaker, would not assume that agents will always be able to make their beliefs explicit.32 However, at the very least, there is a difference in emphasis between us and many of our predecessors working on common ground, since we are most interested in attunements that people are unaware of. Indeed, some of the most interesting collective attunements, for example, attunements to racist or sexist practices, are attunements of which many practitioners might sincerely deny awareness. By contrast, standard work does not distinguish between elements of the common ground that are manifest to all, and elements that are hidden.
The last two of the above five properties, intersectivity and incontrovertibility, also relate to hiddenness. As described above, on a standard view, the common ground of a group is a set of beliefs, a strict subset of the beliefs of the members of the group (that subset that they not only believe, but believe each other believe, and believe that each other believe they believe, and so on, recursively). However, our model has a weird property: something can be in the common ground of the authors of this book while neither Jason nor David is attuned to it.
But are such failures of intersectivity really so odd? It is certainly not a novel thought that collectives of people have properties that members of the collective lack. Indeed, this can be said to be, from the point of view set out by Émile Durkheim, a fundamental and defining assumption of the field of sociology. Setting out what he takes to be a social fact, he states, There are ways of acting, thinking and feeling which possess the remarkable property of existing outside the consciousness of the individual.
Not only are these types of behaviour and thinking external to the individual, but they are endued with a compelling and coercive power by virtue of which, whether he wishes it or not, they impose themselves upon him. Undoubtedly when I conform to them of my own free will, this coercion is not felt or felt hardly at all, since it is unnecessary.
33 Although our primary motivation for avoiding intersectivity does not involve attunement to propositions, it may be instructive to consider the propositional case in order to facilitate comparison with the most standard view in analytic philosophy on its own terms, terms that mark it out as quite a different notion to that assumed by Durkheim. So let’s briefly restrict attention to a narrow set of dispositions that concern propositions, namely the propensity to make assertions.
A group may jointly avow propositions, or an entire system of thought, that none of its members would individually defend, feeling collectively but not individually prepared to defend it. It should at least be clear that, whether this is normatively desirable or not, it is entirely possible for a group to jointly avow a proposition that some or even all members of the group would individually not avow. More generally, a group may have attunements that its members lack, so that an electorate may collectively support a candidate or proposition that many individuals object to, and a population may have collective dispositions to marginalize and oppress, when it would be a category error to even talk about the individuals having such dispositions. This shows a divergence from the standard notion of common ground, although perhaps not yet a difference to make the case that a switch from the more standard approach is merited. So what are the cases that we feel more clearly merit a departure from the standard approach? The cases we find most interesting as a motivation for reconsidering the notion of common ground involve practices and ideologies. The thought that contexts, and perhaps common grounds, contain something like practices is familiar enough in semantics and formal pragmatics, in particular in semantic accounts of imperatives. In his analysis of different kinds of imperative-containing constructions, Paul Portner has added to-do lists to contexts, which formally represent plans, and some others have followed suit.34 It’s natural to think of an ideology as providing plans; certainly these notions are connected. Ideology, after all, is a source of descriptive normativity in Haslanger’s sense—and so provides prescriptions for actions, often in the form of practices.
Discussions in metaethics that treat prescriptions for actions as nonpropositional plans underlying normativity appeal to something like ideologies in our sense. Semantic discussions of expressivism that try to formalize these notions explore some of the same structural territory as us—adding something like normatively laden practices into the common ground. But less clear in such presentations is the way in which attunement to a single practice can yield radically different plans.
Our methodological outlook has been to look to social and political examples of speech to test the concepts and tools of standard approaches in the theory of meaning. Investigation of imperatives shows that consideration of ordinary speech acts even in nonpolitical usages reveals similar shortcomings—this is why semanticists who have been exploring semantic theories for imperative constructions have had to expand the toolbox, for example, by adding to-do lists or plans to contexts. Could one expand the toolbox enough without considering social and political examples? Our guess is one could not. Without considering social and political examples, certain essentially communicative phenomena will fall outside the purview of a theory meant to capture them. But whether we establish this in the current work is something only future research can decide.
Practices and ideologies operate not only at the level of the conversational interaction, the primary level at which the notion of common ground is usually applied in work on meaning, but also at a societal level.
Consider any practice that involves specialization, such as American football, performing an orchestral symphony, or manufacturing various high-tech widgets. It could be the practice of forming a country’s government. Or it could be a practice of overthrowing that government.
In each of these cases, a collective is attuned to what we might call a macropractice (i.e., an essentially interactive, multiperson practice), and individuals, say individual musicians in the orchestra, are highly attuned to various micropractices (i.e., patterns of individual action), but it may sometimes be of little value to talk about their individual attunement to the larger macropractice to which they contribute their own specialized activity.
In much the way that a dog may be better attuned to the practice of sniffing out prey than any human would be, so it is that collectives of individuals may be better attuned to some collective practices than any individual could be. The practices of swarming, rioting, or oppressing minority groups are just not the sort of thing that an individual can become good at, and although sometimes there are individuals who are perceptually and cognitively attuned to the nuances of the collective process, or who intentionally instigate it, this is not necessary. Similarly, marriage is not the sort of thing that a single person could master, and it takes two to tango.
Attunement to a practice—the general case—is not sharing the same plan, in the sense of the same recipe for behavior. A quarterback in American football and their receiver are attuned to the same practice. Yet this joint attunement is poorly modeled by treating them as presupposing the same recipe of behavior. In virtue of being attuned to the same practice, the quarterback and their receiver enact very different plans of action. The quarterback and the receiver are enacting very different skills via attunement to the very same practice.
Attunement in this more general sense is essential to understanding oppressive or potentially oppressive relationships. When everyone has equal attunement to a joint practice, management of the process can itself be collective, but in a complex system with many specialized roles, it is often an emergent property that the management of the process be delegated to a minority. Thus, there is asymmetric control over resources and knowledge, which in turn presents those in a management role with the opportunity to maintain power over others.
For example, Russian landowners and serfs prior to nineteenth-century emancipation were collectively attuned to a style of agricultural production that no individual had complete competence in, but, equally, individuals were collectively attuned to a system of oppression.
The system of oppression involved a range of behavioral dispositions (as well as attitudes) that could not be mastered by a single person, as they impose conflicting demands. The serfs, who were not allowed the luxury of education, usually learned little beyond the practices and culture of their forefathers, including not only practices essential to agricultural production but also practices of serving and showing deference to their masters.
Recall the above distinction between first-order and second-orderattunements. As regards first-order attunements, the serfs were no more attuned to the refined practices and etiquette of the nobility than were the nobility attuned to manual labor, and the serfs were no more attuned to practices of domination than were the nobility attuned to practices of deference.
Nonetheless, they each had second-order attunements to the practices of others in society with whom they interacted, and were collectively attuned to the oppressive system within which they existed, including its attendant ideology, and to a vast set of subpractices that no individual could have been fully attuned to. And here it might be added that being emotionally attuned to a system does not mean liking it. Prisoners in a retribution-based penal system are supposed to suffer. If as a prisoner you enjoy the cut of the whip, you are not well-attuned to such a penal system.
Lucky you.
Our model explains that even an oppressor may completely fail to understand the role they play. Just as someone may not appreciate that they are driving a getaway car from a bank heist, or that they inadvertently act as a drug courier every time they drive their truck across the border, likewise, people may not appreciate that they are contributing to a system of oppression. The way you carry yourself, the way you keep the floor in public conversation, the confidence you show at a job interview, which neighborhood you live in, and which schools your children attend, these are not faults per se, but they may reflect a system that fails many.
Relatedly, our view also explains what is sometimes called “racism without racists,” the title of an influential book by Eduardo Bonilla-Silva.35 It is clearly possible, on the model we have described, for a group to be collectively attuned to a racist practice, without any member having racist attitudes. We can explain how US policing could be a racist practice—one that continues the oppression of poor marginalized American communities, by appeal to racist attitudes in the past that led people to develop those practices, together with the furthering of practices devised by people with these attitudes, due to the continued attunement of groups to these practices.
We might also speak of the dispositional attunement of institutions to practices, institutions like courts, police forces, companies, and political parties, to the extent that we can identify institutional behaviors, and say of these behaviors whether they are in accord with a given practice or ideology.
2.5. Attunement to Resonance When one is “attuned to a practice,” what one is attuned to is the pattern of activity circumscribed by the resonances of the practice. So, we can equally talk of being attuned to the resonances of the practice. We briefly touched on the relationship between resonance and attunement in our discussion of practice, in section 2.3, above, but other than that we introduced the two notions independently, as if they had no more in common than the metaphorical framing they share. As we will make clear in this section, the connection between them is much tighter than a mere framing. It is no exaggeration to say that as regards the role we give them in sustaining practice, the two concepts are mirror images of each other. This is what will enable us later in the book to move seamlessly between resonance and attunement, using whichever is most convenient. Specifically, when we want to talk of properties of a speech practice, we will focus on the resonances of that practice. In particular, our analysis of the presuppositions of communicative acts will be cast primarily in terms of resonance. When we want to talk about individual and collective psychological states, we will refer to attunement, or collective attunement (i.e., loosely, common ground).
This will be crucial for our discussion of harmony, and of the related process of accommodation.
36 In chapter 1, we introduced the idea of resonance as playing something like the role that content often plays in accounts of meaning, although we intended it to capture conventionalized communicative significance more broadly. We emphasized that resonance operates at an interactional level, illustrated in terms of a simple thought experiment, the coffee game. In the coffee game, a practice emerges whereby when one person is sleepy, they nod, and when they nod, the other brings them coffee. The game showed how conventional resonances can be an emergent property of iterated interactional situations. By presenting our thought experiment in terms of a well-studied paradigm, the two-player game, we adopted what we hope was a natural simplification. But it was a simplification that obscured the way resonance functions within society. A book on political language necessarily concerns speech practices that exist at the level of the polis, that is, a large speech community. We will now consider how resonance functions within these larger groups, within what we have been calling communities of practice. To do this, we must first say more about what a practice is and what a community of practice is. As discussed above, we avoid commitment as to how practices are represented psychologically, beyond assuming some sort of memory (which may be “muscle memory” rather than explicit or clearly representational episodic memory) of what has occurred. To achieve this, we need a general method that, based on a history of actions, can compute for any two actions whether they belong to the same practice. In this vein, we mentioned exemplar theory, as developed in psychological work on concepts and categorization. But let us here adopt a counterpart of exemplar theory found in exploratory data science and machine learning, namely clustering.
Clustering is an approach whereby an algorithm automatically classifies data into groups of similar points. For example, faced with data about the physical properties of animals, a clustering algorithm might be used to guess how the animals are separated into different species.
The practices within a community can be seen as a way of clustering actions together, of viewing subsets of them as belonging to the same category. Often, people introduce labels for practices, for example, labeling certain complex acts as acts of assertion, and certain other complex acts as acts of going on a date. So, each practice is one cluster, a label on a subset of actions (or complexes of actions). It is not, however, a simple clustering, in the way that is standardly assumed in data analysis. Commonly, cluster labels are taken to cleanly separate different groups of data points into nonoverlapping sets. It may help to start by imagining a space of actions that is simple in this way, with like actions clustered together, and each action belonging to only one cluster, but it is as well to note right away that the real situation is more complex. Actions can belong to multiple practices, practices can be special cases of other practices, and patterns of actions can be broken down into smaller actions or patterns of actions that belong to further practices.
We are assuming, with Jaeggi, that practices are massively interconnected and interdependent, so the space of practices has a massively complex structure. Still, the basic idea of what a typical clustering algorithm in data science does is a good first approximation for what an individual must do when faced with the problem of making sense of the actions they experience. The individual assigns labels to actions (or at least groups them together, associating them with each other) with some level of confidence, collecting actions that are similar on dimensions that the individual determines to be useful for classification.
Let us move from an individual level to a group level. For a given community, let us define a set of recognizable practices as those clusterings of actions such that (i) there is a high tendency for individuals in the community to have similar clusterings, and also (ii) there is a high tendency for individuals in the community to perform the actions comprising the practices. These are the community practices.
To say that there is a high tendency for individuals in the community to have similar ways of clustering some set of practices, that is, that they classify actions similarly, is to say that they have strong collective cognitive (and perceptual) attunements to those practices. Similarly, to say that there is a high tendency for individuals in the community to perform the actions comprising the practices is to say that the community has strong collective dispositional attunements to the practices. So let us define a community of practice as follows: Community of practice: A community of practice is a set of individuals with strong collective attunements to a set of practices, such that no larger set of individuals has similarly strong or stronger collective attunements to that set of practices.37 Within such a community, there will obviously be a tight relationship between cognitive and dispositional attunements. We should expect that, ceteris paribus, the practice will be ongoing and that new performances of practices from that set will typically be taken to instantiate those same practices, and with a high level of agreement among community members. We take the extension of a practice to be a strict subset of events that include members of a relevant community of practice as active participants.
In section 1.4, we defined the associative resonances of an action in terms of a probability distribution over features of context, the idea being that some feature is a (positive) resonance of an action to the extent that the occurrence of the action implies a boosted probability for that feature. In principle, the probability boost for a feature can be calculated by taking the set of contexts in which the action has taken place, and comparing how much more often the feature is present in those contexts than it is in other contexts. What we are suggesting now is that the set of contexts relevant to such a calculation is determined by a community of practice. Communities of practice determine the extension of a practice, which in turn determines what contexts are relevant to calculating resonance, and the resonances of practices thus defined depend crucially on the identification of an appropriate community of practice, the community that contains the practitioners.
As Penny Eckert and Étienne Wenger explain, One doesn’t “do” a practice excised from the community. A practice is a way of doing things, as grounded in and shared by a community.
The wearing of the right color of lipstick for the sake of being accepted is a practice by virtue of its role in getting accepted, not by virtue of its disembodied appropriateness. Practice always involves the maintenance of the community and therefore its power structure.
Legitimacy in any community of practice involves not just having access to knowledge necessary for “getting it right,” but being at the table at which “what is right” is continually negotiated.38 Exactly what counts as the relevant community for a particular practice is typically far from obvious. Thus it also follows that the extension of a practice is not fully determinate, and thence that the resonances are not fully determinate. We take this not as an inherent problem with the framework, but as reflective of real-world vagueness of social constructs, and of the fact that reasonable people may sometimes disagree about shades of meaning.
For relatively stable practices, like the practice of saying “ouch,” it will make little difference exactly what the community of practice is taken to be, within reasonable limits of space and time, but for other practices, like a relatively new meme generated on a particular message board, spread across social media, and just starting to lead to idiomatic usages in ordinary conversational interactions, it will matter greatly what is taken to be the relevant community of practice. Furthermore, this is not just an empirical matter, but may become normative.
People may actively police both the communities and the practices, attempting to constrain which actions count as exemplars of the practice.
Many countries legislate the proper use of their national languages through special-purpose academies. An example of policing that gained national attention was the reactions of a broad slice of the US public to the “Ebonics” controversy in 1997, which occurred after the Oakland school board approved a policy allowing teaching of English to flexibly use the variety of English spoken by many of the students, that is, African American Vernacular English, a.k.a. Ebonics, a.k.a. Black English. A political and media firestorm followed, a firestorm in which the facts of what the Oakland school board was attempting to do were largely buried.39 Although it provided a learning moment or two, the Ebonics firestorm was revealing of deep ignorance of the nature of language. Many believe that the language that much of the US population speaks is in some sense degraded both grammatically and morally, and there is deep collective angst at the possibility of what are regarded as standard varieties of US English being in some sense sullied by a vastly less prestigious variety playing any role in education. Implicitly, Black Americans were not being accepted as part of the community of practice for American English except when behaving in accord with the White majority. The urgent desire to push back against use of a particular language variety as a teaching tool is reflective not only of broader negative attitudes to the culture of an oppressed group, but also of a widespread perception that it is the proper role of educational establishments to defend the dominant culture. In other areas of education, say as regards religion, or as regards political attitudes in authoritarian states, common ways in which educational establishments standardize the language of students would be described by nonbelieving outsiders as a combination of policing and indoctrination—a point central to Althusser’s discussion of schools as “ideological state apparatuses.”40 Let us leave aside the admittedly important questions of language attitudes that come up at the level of language policy at a state level and within educational systems. We assume that identifying the community of practice relevant for a particular type of action must then involve a combination of facts about location, time, joint interests, and social identity.
A community of practice may be as large as a nation, or even transnational, or as small as a family or work unit. Further, there may be no fact of the matter as to exactly who is in the community and who is not. One might consider a nation as a single community of practice relevant for various social customs, food preparation and consumption practices, child-rearing practices, and so on, but for any of these practices there will be smaller subcommunities for which the practices differ idiosyncratically but somewhat consistently, as well as larger or overlapping communities (say, English speakers around the world) among whom a subset of the practices is partially shared. Given these considerations, we cannot claim that relativizing practices to communities of practice immediately brings clarity to the definition of particular practices. Rather, we claim that the complexity of identifying the relevant community or communities is inherent to the process of identification and understanding of any social practice.
As discussed both in chapter 1 and in the current chapter, we do not understand resonances primarily in terms of the epistemic state of observers, although we take the resonances of communicative actions to be precisely what allows communicative acts to provide information. Rather, we understand the probability changes that constitute resonance in terms of the collective behavior and states of all those involved in the practice, as well as in the state of the world more generally as it changes when the practice is performed. Let us return to the practice of saying “ouch” as an example. When a member of an English-speaking community of practice performs an act of expressing mild pain, one of the things briefly but dramatically boosted is the probability that they are engaged in making an “ow” sound, and that probability more or less quickly recedes, to be replaced by a high probability that they are making a “ch” sound. More generally, the resonances of a communicative action include boosts to all the typical features of situations in which that practice tends to occur, including features associated with the speaker (in this case, primarily emotional and behavioral) and features associated with the hearer (in this case, primarily cognitive and emotional, to the extent that a hearer might recognize pain and feel empathy).
Our relativization of resonance to community practices allows us to state two important postulates in cases involving the resonances of conventionalized, communicative actions, that is, communicative practices.
We might term these postulates non-idiosyncrasy and causal efficacy. Both concern the standardization of language, the sorts of things that can become resonances.41 Non-idiosyncrasy postulate: For a member of a speech community, the resonances of a speech act that follows the practices of that community are not dependent on idiosyncratic features of that individual, be they the speaker or audience member, but only on properties of the context in accord with which the practice is predictably used.
For example, the exclamation “ouch” is predictably used in situations when the speaker has experienced sudden pain, and its resonances include that property of the speaker, whether the speaker actually experienced such pain or not. Suppose that a particular speaker happens to idiosyncratically say “ouch” on a somewhat regular basis not when they are in pain, but rather when they are about to eat bacon for breakfast. If they are not part of a larger community of practice where their speech behavior is normal, then for any other English-speaking observer, the exclamation will be associated solely with resonances of prior pain, and not imminent bacon. Repeated actions that are recognized as the idiosyncratic behavior of only one individual and not the regular behavior of a meaningful class of individuals are not social practices, and for our purposes should be regarded as habits, not practices at all.
One of the most cited dialogues in philosophy of language, Lewis Carroll’s Humpty Dumpty dialogue, famously touches on this point: “When I use a word,” Humpty Dumpty said in rather a scornful tone, “it means just what I choose it to mean—neither more nor less.” “The question is,” said Alice, “whether you can make words mean different things—that’s all.” “The question is,” said Humpty Dumpty, “which is to be master —that’s all.”42 In our terms, the master is not Humpty Dumpty, but the community of practice to which he belongs, for it is community practice that determines the meaning of a word, that is, its resonances. A community of one is no community at all.
Causal-efficacy postulate: Within a speech community, speech practices will not emerge that have as resonances properties that can neither causally affect whether community members perform the practice, nor are causally related to effects of the practice that community members can recognize.
This second postulate also concerns the limits of conventionalization, but it concerns the types of features that can become conventionalized resonances for a community rather than the question of whether an individual can idiosyncratically determine the resonances for themselves. Imagine a situation in which “ouch” was used to express pain, but in which only a subset of people ever used the term, that subset being determined by an unobservable property. For example, it could be that the only people who ever said “ouch” were those whose maternal grandfather had once dropped a glass of milk. It would then follow in this hypothetical situation that the saying of “ouch” would boost the probability of someone having such a hapless maternal grandfather, at least for some suitably well-informed observer.
Care must be taken here as regards what we take community members to be able to recognize. When we experience pain, certain neurochemical pathways are active (or so we presume, though we lack relevant neuroscientific expertise). So, saying “ouch” boosts the probability that these pathways are active. Does it then follow that the resonances of “ouch” include certain facts about neurochemical pathways? We do not take it to be urgent for us to decide this matter here, but we are prepared to allow that the resonances might include properties of the context that, although not directly recognized by community members, happen to correlate with properties that can be recognized. In that case it might be that when ancient Greeks used the word ὕδωρ (hudōr, “water”), the presence of the chemical H2O in a salient situation was a resonance of the utterance, even though the ancient Greeks lacked knowledge of modern chemistry, a point familiar from the philosophical literature on externalism. One thing that would not follow from allowing for the existence of such resonances is that by virtue of Plato pointing at a glass and saying ὕδωρ, his student Aristotle could have become aware of the chemical structure of the liquid in the glass. More generally, the resonances of speech practices might include all sorts of things that neither the speaker nor the hearer is aware of, but which are nonetheless distinctive features of the contexts in which the practice is instantiated.
Let us now tie down a bit further the relationship between resonance and attunement. Earlier in the chapter, we defined attunement to practice as follows: Attunement to practice: An agent is attuned to a practice to the extent that their state and behavior predictably evolve in accordance with its presence in the agent’s context.
The resonances consist in the set of features that are present in contexts of use, and as noted these include all the relevant facts about speakers, addressees, and other audience members or observers. So, for the state and behavior of agents to predictably evolve in a way that accords with a practice just means that the state and behavior evolve in line with the resonances of that practice. We arrive at the following, which describes the relationship between individual dispositional attunement and resonance, and which we take to have the status not of a new definition, but of something like a lemma, following from the definitions we have already given: Attunement to practice (in terms of resonance): An agent is attuned to a community practice to the extent that their state and behavior tend to evolve in accord with the resonances of actions belonging to the extension of that practice in the community.
So, slightly more narrowly, a dispositional attunement to a (part of a) practice is a tendency to perform the practice in contexts bearing features that are resonances of the practice.
Further, membership of a community implies a tendency to be attuned to the resonances of practices that are prominent within that community. A member of a speech community will tend to have strong attunements to the resonances of the speech practices found in that community, and a member of a sports community will tend to have strong attunements to the resonances of whatever sports practices are found in that community. But since such communities often overlap, a member of a speech community will tend to have strong attunements to the resonances of the sports practices found in that community, and, vice versa, a member of a sports community will tend to have strong attunements to the resonances of the speech practices found in that community. The nexus of practices described by Jaeggi is mediated by overlapping communities.
Let’s consider a simpler case, the coffee game. The community of practice consists only of the two players. The resonances of the only action that David performs include prior sleepiness, the physical act of nodding itself (the locutionary act), and Jason later bringing coffee. David has a dispositional attunement to behave in accord with these resonances, specifically by performing the action when he is sleepy, and also before Jason brings coffee, though of course he only has correlational evidence that Jason will react by bringing coffee.
Similarly, Jason is dispositionally attuned to the practice. When he observes David nodding, he performs an action of bringing coffee. In this case, the two actions (David nodding and Jason bringing coffee), if they are perfectly correlated, completely share the resonances of a larger complex practice that involves both actions being instantiated.
The coffee game illustrates the sense in which resonance and attunement mirror each other.
There is simply no difference between stating the resonances of the actions of nodding and bringing coffee, and stating David and Jason’s individual dispositional attunements. To state the resonances is to state the dispositional attunements of the two members of this tiny community of practice, and vice versa. And this is entirely general. To state the resonances of a speech practice found in a speech community (or any practice in any community of practice) is just to state the attunements that members of the community tend to have to that practice, and vice versa. 2.6. Changing Ideology No two instantiations of a practice are identical. Every time a practice is newly instantiated, differences between that instantiation and prior instantiations can have a small impact on what members of the community who were exposed to that instantiation judge to be correlated with instantiations of the practice, and this can then shift their attunements in a way that brings them either temporarily or permanently out of step with other community members. Practices change. They drift, may be affected by other cultural forces, and sometimes are shifted intentionally. A practice may vary considerably over time, or over space, as when there is a dialect continuum, such that small changes in a speech practice correlate with geographical location. A practice may vary relative to other variables, like social status, education, or genre. Consequently, it may sometimes be hard to identify a unique and stable community of practice within which there is a high level of collective attunement to the practice.
A stable practice is a fixed point of behaviors, emerging when there is a high degree of constancy of collective attunements to the practice over time and space. This will normally imply ongoing shared dispositional and cognitive attunement, that is, dispositions to produce new actions within the practice and high levels of agreement as to whether new events count as instances of the practice.
In case there is substantial stability of many practices, we can think of the community of practice in two ways. We can think of the community as its current set of members, or we can think of it as a collection of individuals over time, such that different individuals at different times would have similar attunements, and so, for example, would have a high degree of agreement as regards whether events instantiate the practice.
Contrarily, sometimes a set of practices, or an entire community of practice, can be unstable and even die out. Since a community of practice can be understood as involving individuals spaced over time, it is possible to be the last speaker in a community of practice, albeit at this point the practice itself is inevitably highly attenuated. There may not be many speaking opportunities for that one speaker.
Here’s a straw-man, assertion-based model of ideological transmission: people assert things, and other people accept them. Clearly transmission can take this form, but there are obvious reasons to doubt its generality. At the very least, there are other mechanisms of ideological transmission and information exchange. Actions can reveal a great deal to an observer, whether those actions are assertions or not, and even an assertion can reveal a great deal more than the proposition that is asserted. How so? There is nothing mysterious about the mechanism we have in mind.
Actions do not take place randomly, but rather are conditioned upon the state of the world. A daffodil blooms, and, unrelatedly, a coauthor sends a text. From the first, a honey bee, in natural harmony with the plant, might glean information to direct its nectar-foraging activities, and from the second, the recipient of the text, who enjoys a somewhat more complex harmony with the sender, might learn of some politician’s latest outrage.
The daffodil does not bloom in order to tell us that spring is in the air, and yet its blooming reflects that very fact. And a coauthor’s text, whatever its content, may reveal that the recipient and their joint project with the sender haven’t been forgotten. It may or may not have been intended to convey this. When we witness an event, we may infer that we are in the sort of situation in which such events occur. The more regularly an event takes place, the more clearly it might reflect its context. An event that involves humans interacting in such a systematic way that we can identify their interaction as an exemplar of a general practice can provide a great deal of information, if we know about the contexts in which such interactions tend to take place.
Every practice is potentially communicative, whether we would normally think of it as a communicative practice or not, because any instantiation of that practice reflects the fact that the context is one in which the practice tends to occur, and thus any instantiation of a practice provides an observer with information about that context. It follows that as regards specifically communicative practices, we get an informational double whammy: first, qua practices, they reflect context, and second, qua communicative practices, they have some functional signaling effect that both provides information about the context and changes it.
The first, the way that a practice reflects its context, is what we will call, adapting standard terminology, presupposition. We reserve until part II of the book detailed discussion of presupposition, of the mechanism by which it supports the spread of attunement, and of its role in group identity and ideological conflict. For now, we merely seek to illustrate the way that ideology can be spread through changes in attunement that result from broad and perhaps repeated exposure to actions with strong ideological resonances, without the necessity of assertion. But first, let us state more explicitly what ideology amounts to within the resonance and attunement model.
So far, we said of ideologies (in section 2.1, above) that they are “cultural artifacts consisting of practices, attitudes, affect, and norms.” On the view we have developed, cultural artifacts, as opposed to material artifacts, are no more and no less than collective attunements of groups of people. For an ideology, the relevant grouping is a community of practice, whether that community is a group of practicing Catholics, Democrats, or carpenters. Ideologies are then collective dispositional, cognitive, and emotional attunements among such groups, as well as attunements to metapractices that regulate practice. Ideology perhaps might be taken to include a separate dimension of moral attunement, but we will not discuss the question of whether moral attunement should or could be separated in any way from other attunements, or whether it would be useful to discuss ethical issues in terms of the attunement model. Leaving morality aside, we arrive at the following definition: Ideology: An ideology is the system of collective attunements among members of a community of practice.43 The ideological resonances of a practice then consist in the increased tendency for practitioners to have attunements belonging to the broader ideology of the community, above and beyond attunement to the practice itself. The resonances can be innocent. An ideological resonance of the practice of saying “oy vey” is a tendency to eat bagels with a certain distinctive texture and density. The resonances can also be pernicious. The practice of labeling angry women as hysterical may have as a resonance an increased tendency to treat women as unworthy of hiring or promotion.
Such practices take place within what we will call discriminatory ideologies: Discriminatory ideology: An ideology that includes attunements to in-group/out-group distinctions (a.k.a.
us-them distinctions), and in which members of out-groups are valued less than members of ingroups, and hence as inherently deserving of less than equal treatment or resources. To say that the ideological resonances of a practice are potent would then be to make the claim that instances of the practice have a powerful effect in shifting collective attunement within a community of practice, perhaps helping alter the makeup of subcommunities. We doubt the potency of the practice of saying “oy vey,” but we suppose that various sexist speech practices, such as the labeling of women as hysterical, may be highly potent. Educational and religious practices often have potent ideological resonances, but so do practices of protest and practices of oppression. Neither of the two definitions we have given, of ideology and discriminatory ideology, are inherently pejorative. According to the notion of ideology as we have characterized it, for example, everyone has an ideology. Even calling something a discriminatory ideology is not, in our sense, to evaluate the ideology negatively—for example, it may be acceptable to discriminate against Nazis as a group. In the literature on ideology, however, the dominant notion of ideology from Marx on has typically been pejorative in nature, the idea being that an ideology (such as the ideology of the ruling class that masks reality to subserve their interests) clouds people’s perceptions and preferences.
We can distinguish two different senses in which the notion of ideology can be pejorative—a moral sense and an epistemic sense. A system of attunements is ideologically problematic in the moral sense if it contributes to the maintenance of unjust social hierarchies. A system of attunements is ideologically problematic in the epistemological sense if it masks socially important truths about the world. When scholars talk of ideologies pejoratively, they typically describe systems of attunement that are simultaneously epistemologically and morally problematic.
At least since the work of the sociologist Talcott Parsons, a nonpejorative view of ideology has also emerged. Parsons defines an ideology as “a system of beliefs, held in common by the members of a collectivity,” adding also that the ideology should include some a norm of affinity to the collective and to shared goals of the collective used to guide action.44 Our own nonpejorative notion of ideology is naturally seen as an extension of Parsons’s notion, generalizing from beliefs to attunements, which then folds in the idea that the ideology should guide action since attunements can be dispositional.45 If the pejorative view prominent in Marxist thought and much critical theory is that ideology distorts reality, then a typical nonpejorative view, like that of Parsons, would be that ideology of one sort or another is both ubiquitous and necessary, that it is impossible to comprehend reality without the conceptual scaffolding of an ideology. We can then distinguish a third, radical view of ideology, turning the standard notion on its head. Quill Kukla expresses it as follows: “Ideologies need not be false; they are not sets of misrepresentations. While they in some sense represent social relations, these are not representations that cover over reality.
Indeed, they play a role in constituting reality.”46 Taking this view to its extreme, ideology would neither be a mask that hides social reality, nor a tool for making sense of social reality, but rather ideology would be the primary construct from which social reality is derived. For our purposes, the weaker view expressed by Kukla suffices, that ideology is not merely some sort of observational or conceptual tool, but at least plays a role in constituting social reality.
We postpone to section 7.5 further consideration of the scholarly background against which this definition of ideology is given, although we must forewarn the reader that even there we make no attempt at a comprehensive discussion of prior literature on ideology, a topic that we could not hope to do justice to in this volume. We also do not attempt a detailed comparison with Jaeggi’s notion of a “form of life,” although that notion is closely related to our notion of ideology, especially in respect of being centered around practice. For the purpose of setting out our framework in these first two parts of the book, what we need of our notion of ideology is that (i) practices and not just propositions are central, (ii) both practices and propositional attitudes are tied to ideology via a notion of attunement shared across a community, and (iii) the community itself is bounded by the shared practices that define it. Thus our notion of ideology inherits the central role given to practice by Jaeggi, but also Haslanger, Kukla, and many others, notably Bourdieu. In the light of our attunement-based approach to ideology, let us now briefly consider an oppressive practice, as an example application. Our interest here is in showing how the model we have developed can be used to explicate the efficacy of dehumanizing speech acts not merely as regards the immediate effect of denigrating somebody, but as a way of driving ideological change. That efficacy consists in the ability of those acts to use an innocuous way of talking to smuggle across (if we may borrow from the conduit metaphor) a vast repertoire of attitudes and practices.
Dehumanizing speech acts can achieve this by implicitly pivoting around a context that supports those attitudes and practices. As a hearer, you see the pivot and infer the context.
That is the process we want to capture.
As the work of Lynne Tirrell has emphasized, when Rwandan Hutus repeatedly referred to Tutsis in the early 1990s as cockroaches and snakes, their actions presupposed, and brought together, many features of context.47 Among those features are practices of treating Tutsis, and enemies more generally, in certain ways, practices of characterizing groups in terms of paradigmatic features, practices of using vivid analogies to achieve those characterizations, practices of speaking Kinyarwanda, and practices of talking specifically associated with cockroaches and snakes. This latter set of practices includes (i) bodily ways of manifesting fear and loathing, (ii) ways of speaking about them that characterize them as individually low and worthless and as collectively presenting a plague-like existential threat, and (iii) ways of eradicating them, violently and without remorse or mercy: one does not grant a cockroach a last-minute reprieve and offer it sustenance and a safe place to live with its growing family in exchange for a solemn promise to be just a little less roachy in the future. Without the existence of these three types of practice, reference to Tutsis as cockroaches and snakes would not be effective.48 But in what sense is reference to Tutsis as cockroaches and snakes appropriate? On the contrary, such talk is not appropriate.
Rather, it is indefensibly immoral. We can say that it is inappropriate and immoral because we take as a fundamental precept that all human life has significant value. Yet to say this is beside the point, because the question we have posed is not quite the right one. The question we need to ask is not what makes reference to Tutsis as cockroaches and snakes appropriate, but what, counterfactually, would make it appropriate? To spell it out: what features would a context have to be like in order for it to be appropriate to refer to Tutsis as cockroaches and snakes? It would have to be a context in which Tutsis posed an existential threat, were of no human value, and where there would be no reason to show them mercy. If you were a Hutu who took yourself to be in such a context, you might be unable to empathize with Tutsis. A context in which one cannot empathize with other human actors is a fertile context for injustice or, in this case, genocide.
What the action of referring to Tutsis as vermin presupposed, then, was certain collective attunements, attunements relative to which the practice of genocide appeared justified, because it is only in a context involving such collective attunements that talking that way would be appropriate. The Rwandan Hutu populace was repeatedly exposed to such presuppositions in many different forms, with broadcasts from the RTLM television station from mid-1993 onward being a primary mechanism for spreading such propaganda. The presuppositions were carried not only through analogy to vermin, but also through other speech acts that carried similar presuppositions, and through increasingly violent nonlinguistic acts.
A dramatic example of such a nonlinguistic act would be the killing of political opponents and ordinary Tutsis from the spring of 1994 on. All actions presuppose the appropriacy, or even reasonableness, of the practices they exemplify. So, a killing can presuppose the appropriacy of a practice of homicide by those who identify with the killers of those with whom the victim is identified. Assassinating a prominent political figure is ideal for such signaling, not only because the prominence guarantees publicity, but also because it is evident who the figure is identified with, it is evident who the enemies are, and it is evident that if the life of a high-status individual can be taken, then so can anyone else’s. At the start of the Rwandan conflict, President Juvénal Habyarimana was shot out of the sky.
Habyarimana was himself a Hutu, and the initial round of killings was presumably intended not merely to send signals, but also, and perhaps more importantly, to directly destabilize control structures; however, later mass killings focused on Tutsis.49 This ethnically based targeting, combined with the increasing brutality of the killings, served to reinforce both the normalcy of a practice of killing Tutsis and the lack of mercy they were due. The machetes and clubs used in these killings mirror the methods used to kill snakes and cockroaches.
Suppose one were faced with repeated exposure to acts with extreme presuppositions. It’s easy to suppose that one can simply ignore them.
Suppose one regularly hears some group being talked of as vermin, and later learns that they are being exterminated in much the way that vermin are exterminated. It might be that with increasing exposure to such presuppositions, one comes to a slightly greater acceptance that the context is just as it has been presupposed to be. And so with each exposure, nonperpetrators become a little more like perpetrators, or at least share a way of talking about the actions that are perpetrated.
Similarly, a Nazi prison guard’s action in referring to genocidal victims as if their bodies were no more than industrial materials presupposes a context that has features in which no respect is due to those victims and no remorse or guilt need be felt over their murder. In the context that the guard presumably both accepted and perhaps had helped promulgate, the victims lacked value as human beings by virtue of the groups they were identified with, whether defined by religion, ethnicity, or sexual preference. It was a context that supported a range of individual and state-sanctioned practices that were not merely unfair, but inhumane.
The prison guard’s act of referring not only presupposed their own speech practice of describing the victims’ corpses as they did, but presupposed a much broader range of practices in which, as a Nazi, and as a state worker, they were complicit.
Likewise telling a sexist or racist joke presupposes a context in which certain negative stereotypes are valid, and in which social practices depending on the negative generalization are warranted.50 In particular, the context will support practices that involve awarding differential reward, status, and opportunity to the “target”—or, perhaps better, the “victim”—of the joke, including the opportunity to have their voice heard. A laugh betrays consciousness of tension. But it also betrays a degree of acceptance of the perspective portrayed, and a degree of like-mindedness.
Presuppositions are commonly discussed at the level of individual sentences or utterances of those sentences. One respect in which we generalize the notion of presupposition is to move beyond the presuppositions associated with individual utterances, to consider instead the presuppositions associated with systematic patterns of language use. We have already seen that people can be attuned to practices, and, likewise, in the model we will develop, both the practices, and their attendant attunements, can be presupposed. So we may talk of utterances presupposing practices, or, indeed, of practices presupposing practices.
Let us turn to another example of a practice and its effect in changing collective attunement.
The systematic exposure to sexist jokes may, over time, cause people to behave differently toward women, to engage in practices that not only denigrate women verbally, but cause them physical harm or curtail their freedom of action. When the authors were young, there was a ubiquitous and public practice of exchanging “my wife” jokes, often in the form “she’s so fat/ugly/easy that …” or “I’m not saying she’s fat/ugly/easy, but.…” This practice persists, although at least in public forums it’s less prevalent. To be sure, there was always a certain uncomfortable edginess about such jokes, even though they were clichéd.
The edginess was a necessary ingredient of the comic tension, but can also be seen in terms of Brown and Levinson’s positive politeness. Positive politeness involves presenting goals as shared, and minimizing distance between the speaker and hearer. Positive politeness commonly involves evidencing common group identity and solidarity.
51 Both speaking of something that is mildly taboo and admitting problems with someone or something close to the speaker are ways of showing solidarity with the audience. You wouldn’t say such a terrible thing about your spouse unless you trusted the person you were speaking to, would you?
In the United Kingdom, “my wife” jokes were a staple of public humor, and were common not only on the men’s working clubs comedy circuit, but in variety shows both on stage and on television for mixed audiences of men, women, and children. They were in place at the workplace, and at home in the living room. There are many presuppositions associated with jokes of this kind. Most obviously, they reflect sexist values and their attendant dispositional attunements to focus on physical attributes of women and their sexuality, and to judge them harshly if they fail to adhere to the ideals of the attendant sexist ideology. Concomitantly, such humor reflects practices in which women’s roles depend on their physical attributes and sexuality, rather than on their skills and intelligence. Comics reinforced these practices by telling the jokes, the media reinforced them by giving such comics a platform, and we all reinforced them whenever we joined the comic in laughing at women who fail to meet physical and sexual ideals. Suppose someone is regularly exposed to sexist jokes that either center on women’s putative inability to comprehend technical matters, or implicitly suggest such inability by valuing other attributes. When such a person is confronted with a need to choose between asking a man or a woman for help with a technical issue, it would be surprising if that exposure had not contributed to a bias toward asking the man. Exposure to a practice of telling sexist jokes, a speech practice, leads to an increased dispositional attunement to other sexist practices, practices that are not limited to speech. In chapter 6, we will explicate such changes in attunement in terms of accommodation.
We reject any stark dichotomy between the speech practice of telling sexist jokes and attendant nonlinguistic sexist practices. Indeed, we regard the drawing of such dichotomies as insidious. There are two issues: first, speech bleeds into other types of action: if someone regularly describes another person as a pig, that person will probably come to think of them in a porcine light, and also naturally come to treat them with more disrespect.52 One does not need to be a psychoanalyst to recognize that thinking of someone in a deeply disdainful way and treating them with disdain tend to go together, even if it would be theoretically possible for some superhuman to keep them apart. Second, even nonspeech practices are communicative: one portrays one’s attitude toward other groups as much by the way one treats them as the way one talks about them. Both issues relate to discussion in the coming chapters. As regards the bleeding of speech into other behavioral dispositions, we suggest that we tend to harmonize across different attunements. As regards the communicative potential of nonspeech actions, the model we have introduced in the last chapter allows arbitrary actions to be associated with resonances, and one large class of those resonances, the presuppositional resonances, will be discussed in the chapters to follow. On the view we develop, all actions presuppose practices, and these are likely to be accommodated by others with whom we are socially affiliated. Both sexist jokes and acts of treating people of different genders differentially presuppose practices, and both seed future differential treatment. A further point is that dichotomies between disrespectful speech practices and other forms of harassment in any case tend to be drawn strategically, to defend disrespectful speech practices by artificially disassociating them from the nonspeech practices that are accommodated along with exposure to the speech practices.
But what is exposure to a practice? Exposure to a practice is not, or at least need not be, passive. By taking part in a practice, becoming a participant, we deepen its psychological hold on us. By telling sexist jokes, jokers thereby deepen their own attachment to the practice they are participating in. Jokers’ actions have effects both on their audiences and on themselves. In this sense, sexist comics expose themselves. The social and psychological processes involved are not simple. For example, to the extent that some would view the act as shameful, performing the act may lead either to being shamed, or imagining that one is being shamed, and thence to defensiveness. Such defensiveness, and consequent discounting of alternate views, is one path that might, or might not, lead to an entrenchment of views and tendencies. Defending oneself against accusations of improper conduct may help inure the joker to the denigration of women. In this case, as for so many of the social phenomena that preoccupy us in this book, practice makes imperfect.
If participating in a practice is a form of exposure to that practice, the reverse is also true: exposure to a practice becomes a form of participation in the practice if one does not resist.
And yet one often fails to resist. It is not obvious that merely hearing a sexist joke is always a form of participation in the speech practices surrounding sexist jokes. Yet, at least where some avenue for protest or rejection is available, it is hard to deny that exposure can be seen as tacit participation, whether one cringes “inside” or not.
One could say the same of viewers of pornography or portrayals of violence, of audience members in talks by extremist politicians and preachers, and of people whose idea of a nice afternoon is to watch circus animals jump through burning hoops or bulls being methodically tortured.
It’s the audience that has a choice of whether to participate, not the animal.
If you have ever been a member of any such audience, it would seem pertinent to wonder how it has affected the way you are attuned to the social world, or how your reaction, or lack of reaction to the spectacle has changed the way others view you, or, more subtly the way you view yourself. Perhaps the true worry is this: how can we even know how the many sexist jokes we’ve heard have affected us?
Sexist jokes do not involve assertion of sexist stereotypes. They are a case of showing, not telling. The speaker shows us what the context they are attuned to is like, and what is normal in their society. The speaker shows us the stereotypical properties that women or classes of women are taken to have in this context, and shows us which ways of treating and talking about women are acceptable in it. But why should we accept what the speaker shows us, ways of thinking and being that draw lines that marginalize people, that denigrate ourselves if we are female, and that denigrate our own mothers and daughters whatever our gender? What pressure are we under? The answer is easy: it is social pressure. The positive politeness of a joke is an offering of the speaker’s hand for a pact, an invitation to scrutinize the speaker’s face up close, an implicit invitation to draw socially closer. Here is Sigmund Freud on the function of jokes: Since we have been obliged to renounce the expression of hostility by deeds … we have, just as in the case of sexual aggressiveness, developed a new technique of invective.… We are now prepared to realize the part played by jokes in hostile aggressiveness. A joke will allow us to exploit something ridiculous in our enemy which we could not … bring forward openly or consciously; … the joke will evade restrictions and open sources of pleasure that have become inaccessible. It will further bribe the hearer with its yield of pleasure into taking sides with us without any very close investigation, just as on other occasions we ourselves have often been bribed by an innocent joke into over estimating the substance of a statement expressed jokingly. This is brought out with perfect aptitude in the common phrase die Lacher auf seine Seite ziehen [to bring the laughers over to our side].53 At least in the moment of laughter or eye-rolling, you take on the perspective needed to understand a joke. You shift your perspective—that is, you accommodate, as we shall say in future chapters—at least for long enough to get the joke, but possibly for much longer, becoming more accepting of the perspective the joker has espoused, more likely to behave like the joke-teller does. By laughing when they laugh and thinking like they think, we become more like them. So there is a danger that when we understand a Hutu rebel, or a Nazi prison guard, or a sexist comedian, we will empathize, understand where they are coming from, and take on board their way of thinking to some extent. Empathy itself has dangers. Might we even say that to accommodate the prejudices of malefactors is to become complicit in their crimes? If so, we pay a high price for laughter.
A central issue for a study of the politics of language is persuasion: how people persuade, what people find persuasive, and how people become persuaded. In our terms: Persuasion: What happens when communicative actions cause someone’s attunements to shift to conform to some preexisting pattern with which their original attunements would have been in tension, typically through intent to produce that change.
As the earlier quote from Freud suggests, persuasion is often best achieved indirectly, for example, by humor rather than by explicit presentation of facts or opinions. Recent psychological literature on persuasion has generalized this idea, and typically adopts a two-process model separating the effects of explicit statements of reasons why someone should change their mind from other channels. The major lines of work descend either from Shelly Chaiken’s opposition between systematic and heuristic persuasion, or from Richard Petty and John Cacioppo, who distinguish between central and peripheral routes that persuasion can follow.
54 Neither of these dichotomies match precisely onto distinctions between speech-act types, but, broadly, we can see both Chaiken’s systematic persuasion and Petty/Cacioppo’s central route as corresponding to the straw-man assertion model of communication from earlier in this section (“people assert things, and other people accept them”).
The systematic and central approaches depend on the power of language to depict ideas that can be used as evidence in a deliberative process of drawing successive conclusions.
By contrast, the heuristic and peripheral approaches to persuasion depend essentially on the power of resonance to drive re-attunement without evidentiary deliberation. This nondeliberative approach is well illustrated by the strategies of Hutu propagandists, discussed above. The propagandists persuaded people to develop dispositions to harm Tutsis, in part, by first encouraging a practice of labeling Tutsis as vermin, a process that involved exploitation of emotional and dispositional attunements toward vermin. Change in epistemic attunements was presumably a by-product of this process, but the new epistemic attunements of those influenced by the propaganda campaign were not formed primarily by presentation of propositional evidence that would justify the new beliefs that were formed.
The powers of speech are varied, although certainly, on our view, the power to drive attunement is foundational. Yet here it is unclear what it means to say of speech that it has power. According to Steven Lukes, “A exercises power over B when A affects B in a manner contrary to B’s interests.”55 But as Lukes discusses at length, building on the typology of power introduced by Peter Bachrach and Morton Baratz, an interesting case occurs when someone changes another’s understanding of their own interests, and hence perhaps changes their interests, by persuasion.56 When persuasion takes place, someone may be induced to act without any demand being made explicit, and without the target of persuasion necessarily recognizing that any power has been wielded.
Power: Let us say that an entity exerts power to the extent that the entity changes someone’s state, shapes their interests, or causes them to act. An entity has power to the extent that it has the ability to exert power.
Note that this definition allows not only individuals to have power, but arbitrary groups or institutions too. Of relevance to our later discussion of oppressive language, in chapter 10, is that a community of practice can exert power.
We allow that some might wish to refine this definition in various ways, perhaps by insisting that power is only exercised when the effects are independent of or contrary to the affected person’s prior interests. While this would be close to Lukes’s specific analysis, several variants involving opposition to goals or preferences are found in the literature on power. As we read Lukes, part of the reason why political theorists have preferred to restrict the exercise of power to antagonistic cases that run against the interests of the affected party is that it is only in such cases that the presence of a power relationship might be empirically tested, by determining which party’s interests were best served by an observable outcome.
We are sympathetic to this desideratum, although it is not a driver in our work.
From our point of view, the addition of a restriction to expressly antagonistic situations is problematic (and elsewhere, for quite different reasons, we oppose limitation of the field of study to expressly cooperative situations—see chapters 8 and 9).
We are interested in this volume in the power of words. We might define the instrumental power of words as the ability they give certain people to exercise power over others. But an interesting special case is the power that words give groups of people to coordinate and effect joint action. Whether in a protest movement or at a family dinner, the power of words to facilitate coordination shines as strongly as the power of words to foment conflict.
We would not want to rule it out of court through definition. Restricting the notion of power to adversarial cases would bring the danger of doing just that, making not only the notion of an individual having power over themselves, but also the more important notion of a group having power over itself, incoherent. Words can help grant a group such self-efficacy.
57 Just as we have described the instrumental power of words, we could give a similar instrumental description of the power of a gun. We can also see the gun not merely as a device for wielding the power of an individual, but as having a certain power in and of itself, the power to cause pain or death. Similarly, we can see the power of words not as a power they give to speakers, but as an intrinsic power to effect change in others.
All words have intrinsic power in the sense that they can effect change in people, but the powers are idiosyncratic. Let us say that exigence is the intrinsic power of a communicative practice to affect participants that is hard to resist, independently of the intention of the speaker (or anyone else).
The word “elephant” has the exigent power to make English speakers think of elephants.
Jokes have the power to make us cringe or laugh. Slurs, as we will discuss in section 10.2, have many exigent powers. They have the power to draw attention to an ideology, they have the power to make a person’s position within that ideology visible, and they have the power to affect people’s feelings, and, in particular, to hurt people. All of these are hard to resist.
The power of words is not always the power of the speaker. For when words transmit ideology, the speaker may not understand what they are doing. That is certainly how Klemperer characterizes the insouciant uses by ordinary Germans, including Jews, of phrases that were part of the cocoon of Nazi ideology. We suppose the same was true of many Hutus repeating words seeded into the ground of their life by propagandists. But we suppose also that to the extent that any word is a container at all, it is like a piece of just-recovered driftwood, heavy with the ideology of the waters from which it has been dragged. A theme of this volume is that all our words are reflective of broader ideologies, and every speaker takes part in a social process of maintaining and transmitting ideology with every word they use.
If so, then every one of us is engaged in part of a process of persuasion every time we open our mouth or text a friend, an act that plays a tiny role within the grand argument. This grand argument is no more visible to us than is the forest visible to an ant. A way of looking at this is that language is not just an instrument of power of an individual, wielded in order to achieve that individual’s goals, but is also an instrument of power of a community of practice.
The use of language by community members gives the community power over both its own members, and potentially, as we discuss in chapter 11, over those outside the community.
In the systematic or central approaches to persuasion, the persuader conveys the virtues of the position they wish to advocate explicitly and directly, without any attempt to hide what they are doing. These approaches exemplify what we call straight talk, and they are obviously the paradigmatic methods assumed in enlightenment-inspired approaches to deliberative democracy and free speech, as we will later discuss. (Straight talk, in which the intentions of speakers are transparent, is discussed more fully in chapter 8.) Heuristic persuasion, on the other hand, tries to persuade people without giving them the cognitively taxing task of evaluating complex information, instead offering them heuristic shortcuts. For example, Freud’s nephew Edward Bernays famously helped persuade women to make a collective shift in their disposition to smoke and in their acceptance of other women’s tobacco habits, by arranging for attractive, well-dressed society-type women to walk up and down Fifth Avenue puffing Lucky Strikes. Bernays helped Lucky Strike wield power surreptitiously. He offered a heuristic route, the same route later offered by thousands of smoking Hollywood film stars and by the iconic cigarette advertising icon, the Marlboro Man, caught by the camera lighting up on horseback, perhaps with lasso in hand, the epitome of rugged cool, and a resonant image. Instead of evaluating the pros and cons of smoking, you can just reflect on the behaviors of people you want to be like, people who resonate with you. Then use the following standard heuristic: do what the cool kids do.
Peripheral persuasion is, similarly, any approach to changing someone’s mind, or their behavior, that doesn’t appeal directly to rational deliberation, and instead uses emotional or indirect methods to sneakily effect a change.
We can understand persuasion in the tobacco advertising cases as involving a peripheral route. The route is peripheral because it is not driven by epistemic attunement, but by emotional and social attunements, exploited in order to drive a dispositional attunement to smoke a particular brand of cigarettes. The visceral potency of dehumanizing language, the you-knowyou-want-to-laugh-with-me tug of a sexist joke, and the subtle allure of an ideology portrayed indirectly through the actions of the sympathetic characters of Casablanca, all exemplify peripheral persuasion. The indirectness of these methods, designed to induce a passive change in attunement without need for conscious reflection, means that in none of these cases is there the transparency of communicative intention that is seen in what we term straight talk. Peripheral persuasion, including any approach to persuasion that depends on tempting the audience to substitute heuristic shortcuts for rational deliberation, is then a special case of hustle. Hustle is an instrument of soft power. The nondeliberative or semideliberative nature of attunement change is central to the development and transmission of ideology, and to the functioning of language that mediates these processes, that is, political language. In the next chapter, we consider these themes in both greater depth and greater breadth. We will discuss the significance of classic work in psychology and sociology for the model of resonance and attunement we have developed in order to bring out how change of attunement works, the process we term harmonization. An important theme of the developments there, in common with the literature on persuasion, is the nondeliberative nature of much attunement change at both individual and group levels.
1. Lewis, Amini, and Lannon, A General Theory of Love, 63–64.
2. Curtiz, Casablanca.
3. Our discussion (and section heading) is informed by Stephen McVeigh, “You Must Remember This: Casablanca at 75—Still a Classic of WWII Propaganda,” The Conversation, November 24, 2017, accessed March 1, 2023, https://theconversation.com/you-must-remember-this-casablanca-at75-still-a-classic-of-wwiipropaganda-87113.
4. There is a circularity here that we will brush up against again in this chapter, because the practices that help develop and maintain ideologies, and the norms of maintaining those practices, are themselves components of ideologies. But this circularity is, we take it, inherent to what we are trying to model, and is not in and of itself a shortcoming of the account.
5. Jaeggi, Critique of Forms of Life, 35.
6. Jaeggi, Critique of Forms of Life, 55.
7. A more general statement of predictability here would be as follows: given some class of models of an agent’s behavior, predictability is a measure of the extent to which the best models that take into account the presence of a feature of context more tightly match the agent’s behavior than the best models that do not take into account the presence of this feature.
We might alternatively use explanation rather than predictability, so that attunement would depend on whether a behavior was explained by the presence of the object of attunement.
This brings in a rich but contentious literature on the nature of explanation. We note that, for better or worse, it is difficult to provide a measure of whether an agent is attuned to a tautology, or in cases of physical laws that hold universally or in all situations where an observation might easily be made. Thus, it would be difficult to say whether anyone was attuned to the proposition expressed by 1 + 1 = 2, although it would be possible to evaluate attunement to the contingent and practice-dependent proposition that “1 + 1 = 2” is a tautology. Similarly, while finding out whether a creature was attuned to gravity would require observations of its behavior in zero-gravity contexts, it would be easier to figure out whether someone was attuned to a statement of a gravitational principle (say a statement of the inverse square law), e.g., by asking them.
8. Complications arise when stating what it is for an ideology like Catholicism to be present in the context, because of the dangers of circularity: the presence of Catholicism might be taken to be evidenced only by the presence of adherents, i.e., people who are attuned to it.
We are not sure there is a way around this circularity (which ultimately may apply also to the issue of the presence of the concept of power tools too).
9. Here we see why a notion such as mutual information might be helpful in modeling predictability. Using a notion like mutual information would establish both that the presence of a sander was correlated with an agent having a certain behavioral disposition, and that the presence of that behavioral disposition was correlated with the presence of a sander. This combination would constitute a reason to describe an agent as having a sanding disposition, and not, say, a door-stopping disposition that happened to involve an electric sander.
10. Gibson, “The Theory of Affordances.” 11. Jaeggi, Critique of Forms of Life, 62.
12. See Marques and García-Carpintero, “Really Expressive Presuppositions,” in which they argue that “pejoratives in general and slurs in particular trigger normative, non-propositional presuppositions that require speakers and hearers to share some specific reactive attitude” (140).
13. James, The Varieties of Religious Experience.
14. Jaeggi, Critique of Forms of Life, 67–73 (section 1.2), drawing on Georg Simmel, usefully distinguishes between fashions and forms of life, by appeal to such factors as the temporary nature of fashion, and the greater normative charge associated with forms of life. Our use of “ideology” encompasses both fashions and forms of life.
15. See Bourdieu, Outline of a Theory of Practice; The Logic of Practice. Two further authors who influenced us relatively early on in the development of our approach to practice are the anthropologist Alexander Duranti and the philosopher Olúf1mi Táíwò. In The Anthropology of Intentions, Duranti provides a clear grounding of his position on intention against a background of work in philosophy of language (roughly what we term the content-delivery model), and provides arguments for the noncentrality of intention in his studies of speech practices in Samoa. Táíwò (“Beware of Schools Bearing Gifts: Miseducation and Trojan Horse Propaganda”) argues that updating via privately held beliefs alone will not be a satisfactory account of the way propaganda works—one must be sensitive to “collective epistemic resources,” as well as practices that people maintain independently of their beliefs.
Our hope is that the framework we provide is adequate to this concern.
16. An excellent and broad scholarly discussion of practice-based approaches is Rouse, “Practice Theory.” For discussion more focused on methodologies for observing and describing practice, see Streeck and Mehus, “Microethnography: The Study of Practices.” 17. Haslanger, “What Is a Social Practice?” 18. Jaeggi, Critique of Forms of Life, 56.
19. Jaeggi, Critique of Forms of Life, 61.
20. Jaeggi writes, “Just as some people apply grammatical rules correctly without being able to explain them, one can perform aspects of driving without having learned a corresponding rule—even though these rules can in principle be explicated. From these aspects of driving must be distinguished … those which are so closely bound up with experience that corresponding (formalized) rules cannot even be found or the rules would be far too complex to be illuminating or communicable at the practical level. Here it is not a matter of rule knowledge having become implicit or only existing implicitly; rather, these rules cannot even be formulated on account of the specific circumstances of implementation. Even the simple sequence ‘carefully release the clutch until the gear engages and then depress the gas pedal’ is a procedure that one can ultimately learn only by trial and error—hence only in practice—even if the sequence may be clear in principle in advance” (Critique of Forms of Life, 107).
21. Haslanger comments, “On one interpretation of practices, they are constituted by rules, and agents participate in the practices by intentionally following the rules, or at least (implicitly) taking the rules to provide reasons for their action. There are two standard concerns about this conception of practices. First, not all practices are constituted by rules (and, some would argue, we need a prior conception of practice to explain what it is to follow a rule)” (“What Is a Social Practice?,” 234). Haslanger’s second concern relates to the requirement of “intentionality” in the interpretation of practices she is considering, not the use of rules. Haslanger’s discussion of intentionality jibes well with Jaeggi’s, and it is in line with the view that we ourselves offer, i.e., the view that performing a practice does not depend on an intention to perform that practice. Later, she comments further on the variable significance of rules for practices: “Practices fall along a spectrum from explicitly coordinated behavior that is rule-governed, intentional, voluntary (e.g., games), to regularities in patterns in behavior that are the result of shared cultural schemas or social meanings that have been internalized through socialization and shape primitive psychological mechanisms governing cognition, affect, and experience (e.g., body comportment, verbal inflection)” (239).
22. Haslanger, “What Is a Social Practice?,” 239–40.
23. The works of Jaeggi and Haslanger we focus on here share a similar vision and common intellectual roots, but appeared in the same year, presumably without either author having access to the other’s work.
24. Much discussion of language centers around its function in cognition rather than communication. In some traditions, the goal of the study of language is taken to be primarily an investigation of a mental rather than social phenomenon. This is the case, for example, for the minimalist program in linguistics, which focuses on using language data to reveal a computational system common to humans (Chomsky, The Minimalist Program). For discussion of the related (but distinct) view that human language should be primarily considered as an internal mental system for the representation of thought, rather than a social system for communication, see Hinzen, “Narrow Syntax and the Language of Thought”; Reboul, “A Cognitive View of Language Evolution.” The question of whether observable human language should be regarded as a mere surface manifestation of an underlying mental phenomenon is again related to (but distinct from) the idea that there is a language of thought, which may or may not mirror spoken language. Here Lev Vygotsky’s suggestion that external language comes to dominate thought processes as a child develops (and which thus emphasizes the social as an influence on the mental), and Jerry Fodor’s hypothesis of “mentalese,” a form of mental representation that at least shares features with external language, are well known: Vygotsky, Mind in Society; Fodor, The Language of Thought. 25. Jaeggi highlights the possibility of practices being performed without an intention to do so when she talks of the “active-passive” character of practice. She writes, “To assert that practices have purposes and that they are structured by those purposes is not to posit that they are based on intentions that are fully known,” and later in the same passage, she continues, “Practices are to a certain extent subject-independent patterns of action that are still not entirely transsubjective; or, to put it in more concrete terms, they arise as it were through subjects and yet exist prior to them (and their intentions) and hence cannot be reduced to the intentions of the subjects concerned” (Critique of Forms of Life, 85).
26. Here, let us also tangentially note that the communities of practice that must be modeled in the realm of political language are no longer solely human, but involve computational agents too.
Computational propaganda is a fast-growing emerging field of study. See, e.g., the essays in Woolley and Howard, Computational Propaganda.
27. Lewis, “Scorekeeping in a Language Game”; Stalnaker, “Assertion.” 28. We leave as an open question whether a more standard notion of common ground could be used in our account. Although it’s not clear to us whether such a move would be enlightening, here are some reasons to think it is possible: (i) it is not inconceivable that relevant aspects of, e.g., the weather could be thought of as a special set of propositions; (ii) propositions could be probabilistic, and this might allow for the gradient effects we are interested in; and (iii) on some accounts, such as Stalnaker’s, attitudes are themselves dispositional, so there is no sharp difference between a class of disposition-based theories of common ground and a class of attitude-based theories of common ground. However, it is relevant here that while common ground is standardly entirely symmetric, our notion of collective attunement is not: people can be collectively attuned to different aspects of the same practice. Collective attunement does not require that actors have the same dispositions, only that their collective behavior can be seen as a manifestation of dispositional attunements to the same features of context, for example, the same practices.
29. Stalnaker, “Pragmatic Presuppositions”; Stalnaker, “Assertion.” 30. “Idealized” because it’s not clear to us the degree to which Stalnaker is committed to all of these claims about the common ground.
31. One of Stalnaker’s arguments that presupposition should be explained pragmatically rather than semantically is that “the constraints imposed by a statement on what is presupposed seem to be a matter of degree” (“Pragmatic Presuppositions,” in Context and Content, 54). This does not, as far as we can tell, entail a thesis about the discreteness or not of the common ground, and the assumption that the common ground is discrete is typically made in presentations of the framework.
32. In “Intellectualism and the Objects of Knowledge,” Stalnaker argues forcefully against the view that one can as a rule verbally articulate one’s beliefs.
33. Durkheim, Rules of Sociological Method, 51.
34. Portner, “The Semantics of Imperatives,” “Imperatives and Modals.” See also Charlow, “Clause-Type, Force, and Normative Judgment,” and Marques and García-Carpintero, “Really Expressive Presuppositions.” This tradition is influenced by the pioneering work of Gibbard, Wise Choices, Apt Feelings.
35. Bonilla-Silva, Racism without Racists.
36. We analyze presupposition in chapters 4 and 5, harmony in chapter 3, and accommodation in chapters 6 and 7.
37. Our use of “communities of practice” is set within a tradition whereby the term “speech community” refers to an arbitrarily large group who speaks something that can usefully be described as the “same” language. As previously noted in section 1.8, the way we use the term “community of practice” is more general than the use of the phrase in much work in sociolinguistics and theory of organizational learning. Whereas we allow groups of arbitrary size to be communities of practice, provided they distinctively share practices, there is a somewhat standardized use that is restricted to local settings. Eckert offers the following characterization of a community of practice: “A community of practice is a collection of people who engage on an ongoing basis in some common endeavor: a bowling team, a book club, a friendship group, a crack house, a nuclear family, a church congregation.” She goes on: “The value of the notion communities of practice to sociolinguistics and linguistic anthropology lies in the fact that it identifies a social grouping not in virtue of shared abstract characteristics (e.g. class, gender) or simple co-presence (e.g. neighborhood, workplace), but in virtue of shared practice. In the course of regular joint activity, a community of practice develops ways of doing things, views, values, power relations, ways of talking” (Eckert, “Communities of Practice,” 683).
Etienne Wenger writes as follows: A community of practice defines itself along three dimensions: What it is about: its joint enterprise as understood and continually renegotiated by its members How it functions: the relationships of mutual engagement that bind members together into a social entity What capability it has produced: the shared repertoire of communal resources (routines, sensibilities, artifacts, vocabulary, styles, etc.) that members have developed over time.
Communities of practice also move through various stages of development characterized by different levels of interaction among the members and different kinds of activities.… Communities of practice develop around things that matter to people. As a result, their practices reflect the members’ own understanding of what is important. Even when a community’s actions conform to an external mandate, it is the community—not the mandate—that produces the practice. In this sense, communities of practice are self-organizing systems. (Wenger, “Communities of Practice: Learning as a Social System,” 2) 38. Eckert and Wenger, “Communities of Practice in Sociolinguistics,” 583.
39. We do not mean to imply that all who opposed the Oakland Schoolboard’s policy were ignorant of the linguistic status of African American Vernacular English as a language variety in its own right, rather than being in some sense a degraded form of English, or what some might call “street slang.” Some undoubtedly opposed the use of African American Vernacular English within the school district precisely because they recognized that it is a stigmatized language variety, and perhaps worried that so long as that variety was set in a positive light in schools, students would continue to use it in contexts where they might draw that same stigma onto themselves.
40. Althusser, On the Reproduction of Capitalism: Ideology and Ideological State Apparatuses.
41. We are grateful to Justin Khoo and Timothy Williamson, discussions with whom helped us see the importance of these issues.
42. Carroll, Through the Looking-Glass, 259.
43. Although we do not follow this line, we are sympathetic to Quill Kukla’s suggestion that ideology is not merely a social construct, but also includes the spaces that society physically constructs. As Kukla says, Ideologies are built into practices and the material environment; they are not primarily or essentially “ideas in the head.” Swanson says that an ideology “is a cluster of mutually supporting beliefs, interests, norms, values, practices, institutions, scripts, habits, affective dispositions, and ways of interpreting and interacting with the world” (forthcoming, 6). I like this list but would add even less idea-like phenomena such as buildings, aesthetic products, street signs, spatial divisions such as gates and hedges, and the like. (“Slurs, Interpellation, and Ideology,” 9; see Swanson, “Slurs and Ideologies,” for reference) 44. Parsons, The Social System, 349. We are grateful for the extensive discussion of both pejorative and non-pejorative notions of ideology found in Eagleton, Ideology: An Introduction, 1– 32; and Gerring, “Ideology: A definitional analysis.” 45. For further discussion of the complex history of definitions of the term ideology see Gerring, “Ideology: A Definitional Analysis.” 46. Kukla, “Slurs, Interpellation, and Ideology,” 9.
47. Tirrell, “Genocidal Language Games.” 48. Tirell, “Genocidal Language Games.” 49. The fact that Habyarimana was a Hutu may have enhanced the strategic usefulness of his murder, since it made it easier for Hutu rebels to blame his death on Tutsis.
50. As discussed in section 10.2, some linguists and philosophers have argued that slurs’ derogatory meaning or effects are best explained in terms of presuppositions. See, e.g., Cepollaro and Stojanovic, “Hybrid Evaluatives”; Marques and García-Carpintero, “Really Expressive Presuppositions”; Schlenker, “Expressive Presuppositions.” 51. Brown and Levinson focus on politeness strategies as ways of managing speech that would otherwise threaten face. Positive politeness is opposed to negative politeness, which involves distancing: deferential, impersonal language and lack of imposition. We are using the term positive politeness in a slightly more general way than Brown and Levinson, since someone telling a joke is not lessening the effects of a specific face-threatening act. However, a joke, and unprompted politeness more generally, can be seen as mitigating potential future threats to face. Manifesting common group identity places the interlocutors in a position where they can potentially say things to each other that they could not otherwise, such as requesting help. See Brown and Levinson, Politeness, 101–29.
52. In Elisabeth Camp’s work (e.g., “Imaginative Frames for Scientific Inquiry”), she argues that metaphors involve adopting a perspective—describing someone as a pig thus involves adopting a perspective that makes sense of this description.
53. Freud, Jokes and Their Relation to the Unconscious, 122–23.
54. Chaiken, “The Heuristic Model of Persuasion”; Petty and Cacioppo, “The Elaboration Likelihood Model of Persuasion.” 55. Lukes, Power: A Radical View, 42.
56. Bachrach and Baratz, Power and Poverty.
57. Lukes contrasts his own conception of power with Hannah Arendt’s. As he notes, Arendt’s is set at a group level: “the human ability not just to act but to act in concert” (Arendt, On Violence, 44).
To analyze words as having the power to coordinate and effect joint action would be to take a leaf from Arendt’s work. But note here that Lukes objects strongly to Arendt’s approach and others of its ilk as “revisionary persuasive redefinitions of power which are out of line with the central meanings of ‘power’ as traditionally understood” (Lukes, Power: A Radical View, 34). Attractive as it might be, a revisionary persuasive redefinition of power does not appear to us to be essential to our project. It suffices for our purposes merely to note that quite varied conceptions of power are found in the literature, especially as regards whether it is defined at the level of a relation between individuals, or whether it is defined at the level of the group.
CHAPTER THREE Harmony Conversing together is never comparable with the transfer of material. In the understander, as in the speaker, the same thing must be developed from the inner power of each; and what the former receives is merely the harmoniously attuning stimulus. Hence it is also very natural for man to re-utter at once what he has just understood.
—WILHELM VON HUMBOLDT1 WORDS HAVE POWER. They have power by virtue of their resonances.
Resonance is key to a speaker’s power over others. The resonant properties of words affect people emotionally and socially, and they shift their attunements. Here we get to the crux of what we seek to analyze in this chapter. We seek to analyze the effects that communicative acts have on people, the way this plays into group dynamics, and how this can have political implications.
Harmonization is our term for the process by which systems of attunement change. These changes can be at the level of systems of attunement of individual people, or at the level of collective attunements of a group or larger community. At both the individual level and the group level, a further distinction can be made, between (i) short-term changes in attunement activation, what might be called the immediate impression that a stimulus makes on someone, and (ii) more stable changes with predictable long-term effects. Examples of attunements becoming more or less activated in the moment include an attunement to the practice of using a certain word being activated by someone using that word, or an attunement to being shamed in some particular way being activated by something found to be shaming in just that way. Harmonization can also involve the wholesale creation of new attunements, or changes in the strength of existing attunements, as when a new word or a new concept is acquired, or when dispositions change such that the tendency to frame things in one way is partly replaced by a tendency to frame them in a new way.
Harmonization is an adaptive process. It is a matter of changing so as to accord with features of the environment. If that environment involves a certain action being performed, then we are in alignment with what is normal in the environment (and hence in alignment with the action) if our state develops in accord with the practice of which the action is an example.
We are, in this technical sense, in accord with an insult someone has hurled at us when we get riled up by it, since that is a normal part of societal practices of insulting people. But here it becomes clear that being in accord with the practice to which we are attuned does not imply being in individual harmony, or at least not positive harmony. When insulted, we are in internal discord. As the musical metaphor suggests, the dissonant harmony induced by the insult leaves a tension that must be resolved, seeking a new and more consonant harmony.
The most important drivers of harmonization in this book (especially chapter 6, where we connect harmonization to standard notions of accommodation) will be those related to identity and group cohesion. We feel consonance when we are in tune with those around us or those with whom we identify, and dissonance emerges when we are not in tune with them.
It challenges our identity if we feel that we do not fit in with those with whom we identify.
These issues, of identity and group cohesion, become particularly acute in the context of crowd behavior. The behavior of crowds calls for a theory of collective harmony, whereby the search for consonance and the avoidance of dissonance at a group level drives the behavior of groups.
So far, we have defined what the resonances of words are within a practice, and we have said what it is for someone to be attuned to something, but we have not explicitly put these together to talk about the different ways in which words can resonate with someone, and the power that words can thereby have over people. Relatedly, we have said little about the nature of the emotional and social power words have, or about how words come to have such powers. Further, we have focused on the resonances of individual things, and the attunements that people or collections of people have to individual things, including individual practices, but we have not talked about how the different attunements people have relate to each other. That is, we have not talked about the properties of systems of attunement. Ideologies are systems of attunements and their attendant practices. So, to talk about the transmission and development of ideology, which is the central goal of an account of the politics of language, we must in this chapter talk about how systems of attunements cohere and how they change in response to an external stimulus, whether that stimulus be the provocations of politicians, the banter of a plumber, or the chanting of a protest group.
We begin by discussing ways in which standard assumptions in what we have termed the content-delivery model of communication fall short. At this point we turn to prior literature, introducing prior work that will guide our own theoretical development. With this in hand, we describe some of the properties that a theory of harmonization of attunements must have so as to avoid the pitfalls faced by the content-delivery model. This is initially set at the level of individual agents, but in the last two sections of the chapter, we shift to the level of collective harmonization, that is, changes of attunement at the level of groups and communities.
3.1. Comprehension in the Content-Delivery Model The traditional content-delivery model of communication, as we have characterized it, looks at language in terms of transport of neutral information that is packaged in a bundle of words. It is a straw man, insofar as many philosophers of language and linguists would certainly agree that aspects of it are at best idealizations. And yet it has been unclear what the alternatives to this way of looking at communication are. Also, and to us more worryingly, the model has tended to lead to a focus only on certain types of data for which it is a good first approximation. We have already suggested that the dichotomy of active speaker and passive listener is problematic, arguing that it is better to start with a model of communicative interaction in which one-way conversation is a special case. (Herb Clark and Deana Wilkes-Gibbs call this the “literary” model: “speakers refer as if they were writing to distant readers.”2) We begin the analytic work of this chapter by considering what the content-delivery model, which tackles a single direction of information flow, suggests about the role of the hearer, and why that is problematic in its own right, especially in the context of political language and hate speech.
This will lead us to a way of looking at the role of the hearer in terms of three processes that flow naturally from the attunement-to-resonance paradigm. The standard model allows for only one of the three, and even in that case prior scholars provide us with good reasons to doubt that the standard model gets it right.
As discussed in chapter 1, the content-delivery model can be seen as a special case of Michael Reddy’s conduit metaphor, a special case in which the message has been packaged, as if words are little boxes for meanings.
The hearer’s role is to unwrap the packaging to find the content inside, and then, as if it were a surprise gift, make a judgment: can I accept this gift, and, if so, where can I put it? That is, if I accept what I have been given, how then should I integrate it into my system of beliefs?
Metaphorical it may be, but the metaphor is as ingrained in analytic philosophy as it is in broader contemporary culture. Because the metaphor is so ingrained, it is not seen as problematic that there are clear sociological divisions between subfields that focus on the comprehension problem, and those that focus on integration of the message, how a hearer accepts and internalizes what has been communicated. Within philosophy, the former issue, comprehension, is studied mostly in philosophy of language and the latter, integration, in epistemology. The problems resulting from this disciplinary division stem partly from the division itself and partly from standard assumptions in each of the subfields. Here are the problems: 1. Neutrality of integration: The cleavage implies that processes of comprehension and integration are independent. Formal work in epistemology commonly uses logics for representation of the message meaning. Integration is then analyzed in terms of a sterile artificial representation of the content of the message, abstracting away from the original form of the message. A consequence is that it is not the form of the message that affects its integration, but only the supposedly neutrally characterized information that is “contained” in the message.
2. Short-termism: The cleavage is allied with a focus on comprehension or integration of a single utterance, rather than on processes that involve long chains of interaction, perhaps spaced over multiple encounters between multiple people who might not themselves share common communicative goals.
3. Passive transmission: Comprehension is seen as a deliberative process of intention recognition. This biases study toward the analysis of how people use communication to progress toward goals. It is thus illsuited to shed light on processes that occur reflexively (as is the case for much emotional and social resonance), and ill-suited to shed light on processes by which ideology is passively transmitted, let alone processes by which ideology evolves.
4. Propositionality: Within epistemology, the integration of communicative acts has primarily been studied in terms of how people change their degree of belief (their credence) in a proposition as they receive new information. So, the study of integration has almost exclusively concerned propositional knowledge. Models developed in this vein provide little insight into how language can produce emotional reactions or incite action.
5. Deliberativity: Although there has been a push in epistemology against idealized models, the field has been dominated by a picture of belief revision as deliberative reasoning, as if the hearer was a highly (perhaps even perfectly) rational detective or scientist carefully and unemotively sifting through evidence with no restriction on the complexity of reasoning employed. Real people have emotional, social, and cultural biases toward particular outcomes and have limited attention. There is typically no manifest chain of reasoning that can be reliably identified as explaining why someone has been affected by an utterance in a particular way.
6. Noncompulsion: The deliberative nature of the process assumed in the integration phase implies that hearers are free to make a rational decision about what to accept. On this view, we might think that a primary reason people would be hurt by an utterance (other than simply because they are upset that the speaker is behaving badly toward them) is that they rationally accept that some proposition that has been asserted is true, and the truth of that proposition is itself hurtful. That is, one would be hurt when one learns bad news.
Similarly, the only type of utterance that would make one feel good would be an utterance that conveyed good news that one decided to accept. No explanation is provided for language that directly triggers emotional reaction, in a way that circumvents processes of rational deliberation.
Let us use the term deliberative uptake to mean a multistage process like that assumed in the content-delivery model, consisting of comprehension, in which the meaning of an utterance is identified, followed by integration, in which a decision is reached deliberatively to accept the message and update one’s mental state accordingly, or to reject it. The above set of six problems suggests that the way speech affects hearers is not well modeled as deliberative uptake.
Some of the ways in which deliberative uptake is inadequate are consonant with issues we have already raised in previous chapters. In particular, the process of uptake cannot be based on forming and then integrating a representation that is emotionally (or otherwise) neutral. But we also take the above six problems to show that even if the mental representations assumed were not neutral, deliberative uptake could still not be the sole process by which utterances affect hearers. Based on these shortcomings, we will describe an alternative model of how speech affects people in our resonance and attunement framework, a model in which a variant of deliberative uptake (minus the assumption of neutral representations) is supplemented with additional processes by which people change during and after conversational interactions.
Before proceeding, we should make clear that although the assumptions of the content-delivery model are, in varying degrees, common to much analytic philosophy, they are by no means standard assumptions in cognitive psychology. Much of our presentation in this chapter will be based on well-known psychological research. One important literature that we will not draw direct comparisons with is the psychological literature on reading comprehension.3 Relative to the body of philosophical work we are critiquing, the intended domain of application of the psychological work is much more tightly circumscribed, its empirical base is much stronger, and its psychological foundations are far more nuanced.
While we take inspiration from models of reading comprehension and will briefly allude to some further connections below, a crucial difference is that the readingcomprehension literature focuses only on factors affecting individual comprehension. We emphasize in this chapter the limits of approaches to meaning that assume something like a deliberative uptake model in order to emphasize that such models cannot possibly capture interactive aspects of communication, and are not well-suited to capturing effects that occur at a group level. In this, the reading-comprehension literature in psychology is, by design, limited, albeit that, as a result, the reading-comprehension literature bears a limited resemblance to less psychologically sophisticated models of meaning in much of philosophy.
3.2. Cognitive Dissonance When one is deciding between alternative courses of action, knowledge, it might be plausibly thought, is the only true coin of the realm.4 If so, it makes sense to assume that when people are faced with clear disconfirmatory evidence, they simply drop false beliefs in order to better decide how to behave.
Now, imagine a human-like species for which the pipeline ran in reverse. These hypothetical animals would use their own behavior as a driver of what to believe, sticking with beliefs that conform to their behavioral tendencies, and refusing to accept evidence that ran counter to those tendencies. Surely a species displaying such irrational tendencies would quickly become extinct? Apparently not: the work of the psychologist Leon Festinger suggests that humans are in some circumstances just such unfortunate creatures. Or at the very least, we humans are very stubborn.
In contemporary epistemology, fragmentation is the splitting of representations of knowledge or belief into multiple self-contained segments. A primary motivation for the introduction of fragmentation into epistemic theorizing was the problem that contradictions raise for some formal frameworks in epistemology and the theory of meaning. At root, it is the idea that our minds are compartmentalized. Freudian psychology can be understood as resting on a particular version of this doctrine, and it’s now a familiar part of everyday conceptions about how the mind works, how we deal with difficult life situations such as trauma. When we hear a story of a Nazi prison guard who in returning to his family transitions immediately from brutal abuser to gentle parent, it is nowadays natural to think of this in terms of compartmentalization, by which the guard maintains two personalities without experiencing tension between them. It is important here that, on the view we are describing, it is really personalities that are being maintained and not merely personas, public presentations of personality.
5 In mainstream epistemology, the equivalent of a system of attunements is systems of belief or knowledge. The technical usage of fragmentation within epistemology involves splitting the representation of knowledge (or belief) into multiple self-contained segments. It arises as a reaction to wellknown problems with some of the most widely studied frameworks in epistemology, frameworks that take classical logic as their basis. Of these problems, the most relevant is that in classical logics anything can be derived from a contradiction. In classical logic, contradictions are like exploding mines: as soon as a contradiction exists within a system of propositions, classical logics allow all other propositions to be derived. It follows that as soon as someone believes things that contradict each other, they are also predicted to believe every proposition. The trouble is that it seems plausible that all of us at some time or other maintain beliefs that are inconsistent with each other, say thinking that you rushed out of the house late for work in the morning without having time to do anything else, and thinking that you let the cat out as you usually do after your morning coffee.
Yet everybody can distinguish between things they believe and things they don’t believe, even people that believe they let the cat out at 8 a.m. while also having a distinct memory of their morning actions that does not include anything to do with the cat. More generally, there is simply no evidence that having contradictory beliefs impairs our ability to reason and draw conclusions in the way that these classical models suggest would be the case.
Fragmentation has been suggested as a possible solution to this problem.6 The idea is that in a situation where someone has inconsistent beliefs, their mental state should not be characterized as a single logically inconsistent representation of the world, but rather as two or more individually consistent representations, say one in which you let the cat out before going to work as usual, and another in which you didn’t. This solves the problem, because now neither representation justifies fantastical inferences.7 In the late 1950s, Leon Festinger discussed situations in which people are faced with contrary cognitions, using the term “cognitions” with something like the breadth of our term “attunements,” covering not only beliefs, but attitudes and behavioral inclinations more generally. For him, as for us, logical contradiction between beliefs is a special case of cognitions (or, as we would say, attunements) that are in tension. To describe this tension, he introduced a term that has entered everyday talk of human mental life: cognitive dissonance.
8 Its contrary, cognitive consonance, occurs when one cognition follows from another. The predictive value of the theory comes from the hypothesis that much human behavior is driven by the need to minimize dissonance. This minimization can result in stark departures from what we would normally describe as rational reactions to new evidence, especially when the level of dissonance created by that evidence is so high as to threaten one’s psychological well-being.
We can put the issue Festinger tackled like this: when people experience high levels of dissonance, something has to give, and the question is, what?
Festinger’s answer, which is the lynchpin of a decades-long research program, allows that sometimes the dispositional and emotive tail will wag the epistemic dog: The maximum dissonance that can possibly exist between any two elements is equal to the total resistance to change of the less resistant element. The magnitude of dissonance cannot exceed this amount because, at this point of maximum possible dissonance, the less resistant element would change, thus eliminating the dissonance.9 Festinger can be seen as saying that as dissonance increases it will eventually result in change at the weakest link, the cognition (or, for us, the attunement, although not necessarily a cognitive attunement) that is held least dearly: let’s call this the weakest-link hypothesis.
In some cases, we might reasonably interpret a failure to accept what should be compelling evidence as a sign of an underlying pathology. For example, suppose an addict is faced with both expert warnings and direct experience of disastrous effects on others of the addiction.
The addict might deny the evidence or minimize the extent or dangers of the addiction, and deny the seriousness of the behaviors they are engaged in so as to maintain their addiction, because facing those issues would introduce dissonance. In such a case it might be said that the addiction is, by its nature, more compelling than the compelling evidence. This is then a first example in which the contrariness of attunements extends across dimensions: the highly compulsive dispositional and emotional attunements of the addict are in conflict with evidence that might otherwise lead to cognitive attunements. These cognitive attunements would be veridical in the sense that if the addict had those attunements they would be better attuned to reality. Unfortunately, reality may be no match for the addiction.
The idea that beliefs can have a genealogy formed from our fears, confusions, and weaknesses, rather than reason, has, of course, a long philosophical pedigree. David Hume’s work on religious belief is an obvious example. 10 Hume argues that humans have religious beliefs because they promise an easy resolution of frightening mysteries. On Hume’s view, we have religious beliefs not because they are rational, but because they are comforting.
Festinger, like Hume, was also struck by the case of religious belief.
This emerges in his extensive focus on cult members. Cult members are able to maintain systems of beliefs that appear patently absurd to outsiders, even in the face of what appears to be irrefutable disconfirmatory evidence.
One might have thought that when a cult’s raison d’être concerns a prophesy of a great event, and that event does not transpire, that would quickly lead to disenchantment among cult members. As Festinger discusses, on the basis of carefully researched historical and contemporary examples, religious and spiritual cults do not simply evaporate when the Second Coming, alien invasion, or end of the Earth fails to materialize on the appointed day.
Rather, they often get stronger as a result.
A cult is not a set of propositions, but a set of people. A cult member is heavily invested in this community and in the way of life that binds them to that community—a community of practice and its attendant ideology.
Within the cult, the members find like-minded people who are supportive of their lifestyle and actively reinforce their beliefs, but outside they face opprobrium. Thus, cult members are trapped in an epistemic dead end.
Suppose the cult’s central beliefs are critically challenged by events. The stronger the challenge, the more unbridgeable the gap between insider beliefs and outsider beliefs, the greater the likely opprobrium from outsiders, and the greater the probability that outsiders may not merely disdain but actively shame insiders. It is entirely in line with the weakestlink hypothesis that cult members should be unlikely to accept the disconfirmatory evidence at face value, for that would require a massive change in lifestyle and community, a change that has been made harder by the new evidence, not easier. Indeed, it may even be that the stronger the disconfirmatory evidence, the higher the cost of accepting it. If so, cult members are in a truly absurd position, since their social and emotional incentives run precisely opposite to what would seem to be an a priori epistemic principle, that a rational agent should be more likely to accept a proposition the stronger the evidence for it.
Festinger observed that facing damning counterevidence, cults often turn in on themselves, so that group members saturate themselves in a pool of like opinions, the cult members becoming ever more dependent on each other in the face of a hostile world. Sometimes this inward turn redoubles strength of conviction to such an extent that the group is able to follow it with an outward turn, attempts at proselytization becoming fervid. Festinger explains this latter effect in terms of the conversion of outsiders to the cause satisfying a need to reduce dissonance, by increasing the pool of those from whom cult members gain approbation.
The cases Festinger studied in most detail involve failures of cult prophecies. However, their morals apply more generally. The work suggests that members of a cult may be unable to recognize the failings of their leader, because recognition of such failings would create dissonance with a much larger set of practices and commitments. Again, it may even be that the more egregious the leader’s behavior, the harder it will be for cult members to accept that behavior as a failing. If so, the oft publicized unusual behaviors of cult leaders (owning gold Bentleys, sexual antics, etc.) are not mere peccadilloes. Like external, real-world counterevidence, they may actually function indirectly to increase group cohesion. That is, the peculiarities of group leaders, a certain expectation-confounding otherworldliness, is a feature rather than a bug. Similar arguments apply to the fact that the cult way of life, in terms of daily practices, is often far from the mainstream. This is not an accident. The more unlike the mainstream, the greater the dissonance group members would feel in returning to that mainstream. The weakest-link hypothesis then suggests that the abnormality of everyday living practices within a cult is part of a psychologically sound strategy for maintaining cult integrity, since it leads to cult members having greater resistance to awkward facts than would be the case if their everyday life stayed close to outside norms.
Cognitive dissonance theory predicts that people should not be easily swayed by evidence that runs counter to practices that are central to their way of life, even very strong evidence.
This implies that the epistemic state of a cult member can be quite strange. Here let us observe that fragmentation theory offers a path toward representing that epistemic state, since it allows people to simultaneously maintain contradictory beliefs. We note this not in order to propose a direct development of cognitive dissonance theory, or indeed of fragmentation theory, but simply as a way of underlining the fact that the two bodies of work are compatible. Later in the chapter, we will suggest a development of the resonance and attunement framework that combines ideas from both. But first, let us introduce one more big idea, an idea that in some ways is in tension with cognitive dissonance theory.
3.3. Nondeliberative Uptake As described above, the content-delivery model suggests that when a hearer encounters a claim, they engage in a process of deliberative uptake: first the hearer must identify the proffered proposition without initially taking a stance on its verisimilitude; then, through a process of contemplation, the hearer must pass judgment on the truth of the proposition, either adding it to their stock of beliefs, or rejecting it. In a classic paper on how people process reportative evidence, the psychologist Dan Gilbert traces to Descartes the idea that comprehension is followed by a separate process of judgment as to the merit of the claim.11 However, Gilbert sheds doubt on the idea that belief change is an inherently deliberative process in this way, and contrasts deliberative uptake with a view that he traces back to Spinoza.12 On this second view, which we will term nondeliberative uptake, when an idea first enters our minds, it enters as a belief, and only later might we decide to reject it. Note here that though we use the term “nondeliberative uptake,” Gilbert’s model does not strictly concern whether uptake is deliberative, but the timing of any deliberation. His hypothesis is that people initially accept claims, and only later, as circumstances permit, consider more carefully whether they have, as it were, bitten off more than they can chew.
The idea of people too readily accepting what they hear, of being gullible, is an old one, familiar from antiquity. “This shows how little trouble most people take in their search for the truth—they happily resort to ready-made opinions,” writes Thucydides after citing several common misconceptions.13 Despite widespread recognition of the fact that all of us are to at least some extent gullible, the nondeliberative model of uptake remains challenging. It threatens what for many is presumably a deeply held conviction that we have free will in making up our minds, and do not simply accept the things people say to us automatically.
However, in some cases, the nondeliberative view has immediate intuitive plausibility.
Consider visual illusions. We observe something (say a straw apparently bent at an unexpected angle at the point it enters a glass of water) and only slowly recognize that the appearance is illusory (the straw is in fact straight). We will turn shortly to another class of cases, more central to our enterprise, where a process like the one Gilbert described is intuitive, or perhaps even rather obvious.
Whether Gilbert’s model is intuitive or not, he offered a host of experimental evidence that human behavior is better modeled as nondeliberative rather than deliberative uptake. Much of the evidence in favor of the nondeliberative model depends on the ways that belief update changes when people are distracted or under pressure. The deliberative uptake model predicts that if the process of thinking about a proposition is interrupted or restricted by cognitive load, people should end up equivocal about the truth of the proposition, for they will not have had the opportunity to pass judgment affirming or denying it. On the other hand, the nondeliberative uptake model predicts that in the same circumstances, people should be likely to believe the proposition, even if it is inconsistent with other evidence they have previously been exposed to. To summarize a range of different experiments discussed by Gilbert: the nondeliberative uptake model seems to better fit the facts of how people under cognitive load change their minds in the face of new information. For example, people who are distracted in some way, and who are exposed to a proposition inconsistent with their prior beliefs, show a clearly increased tendency to manifest belief in that proposition afterward compared to exposure under lower cognitive load. This suggests, for example, that a great way to get people to believe peculiar things would be to broadcast on talk radio while they’re driving.
Subsequent literature has made it clear that we should adopt a somewhat circumspect attitude toward the nondeliberative uptake model. Contra Gilbert’s predictions, people sometimes shift to a deliberative stance, particularly when confronted with claims that challenge beliefs with obvious significance to them; that is, people can become epistemically vigilant.14 The psychological literature here echoes a theme found in Stanley Cavell’s interpretation of Wittgenstein on criteria of judgment. Using a notion of attunement related to our own, although it centers on attunement to others rather than to our more general notion of attunement to arbitrary features of context, Cavell writes, Appealing to criteria is not a way of explaining or proving the fact of our attunement in words (hence in forms of life). It is only another description of the same fact; or rather, it is an appeal we make when the attunement is threatened or lost. Official criteria are appealed to when judgments of assessment must be declared: Wittgensteinian criteria are appealed to when we “don’t know our way about,” when we are lost with respect to our words and the world they anticipate.15 On behalf of Wittgenstein, Cavell here suggests that when the attunements we have to the words of others are consonant with our forms of life, there is no need for deliberation in which reason (“criteria”) is made explicit. In such cases, the understanding of others, our attunement to their words, amounts to what we have termed nondeliberative uptake. It is only when the speaker wishes to change our attunements in a way that clashes with our forms of life that we resort to the business of deliberation and challenge. We might say that ideological anomaly promotes vigilance, or, in our terms, that dissonance triggers deliberation.
Even if a strong claim that initial uptake with assertions is always nondeliberative is unwarranted, it’s fair to say that a strong claim that initial uptake with assertions is always circumspect would be equally unwarranted. Based both on Cavell’s Wittgensteinian considerations and the later psychological evidence, we have arrived at a tempered version of Gilbert’s conclusions. As regards change of belief resulting from presentation of reportative evidence, uptake is sometimes relatively nonreflective, but can become circumspect when dissonance is experienced.
We will talk more in this chapter about when dissonance is experienced, but to cut a long story short, that will typically be when attention is drawn to clashes between attunements, especially attunements that are strongly held, as will be the case for core ideological commitments. To get even further ahead of ourselves: to get someone to nondeliberatively accept propositions that clash to some extent with core ideology, it would be necessary to avoid drawing attention to the fact that you are doing so, to hustle, to sneak those ideas into the message rather than simply asserting them. We postpone discussion of a mechanism by which this is commonly achieved, presupposition, until the next chapter. Our interest in this volume goes beyond changes in belief, belief being just one of many types of attunement. So let us now consider how change of other types of attunement works.
First, consider perceptual attunement, which we touched on implicitly in our earlier discussion of the bent appearance of a straight straw entering water. In paradigmatic cases of people sensing what is in their environment, no conscious deliberation is required to become aware of a stimulus and develop a percept. Indeed, the attunement model is an intellectual cousin of the so-called direct perception model introduced by the psychologist James Gibson,16 whereby perception is understood not in terms of a process of forming percepts that must then be integrated into understanding, but in terms of an informational connection between what is happening in the world and the state and behavior of an agent.
We have little to say about perceptual processes in general, but note that comprehension itself may be thought of as a perceptual process. On our somewhat Gibsonian view, the way people are affected by an utterance is not so much like the way Sherlock Holmes contemplatively follows a trail of clues, but the way an ant reflexively follows a trail of pheromones. The ant is in tune with its environment and incredibly sensitive to subtle differences about which it has no power to introspect.
Second, for dispositional attunement, while we may develop dispositions through conscious application of thought, this does not seem to characterize many examples of mimicry or learning how to perform a task like riding a bike. The immediate and noncontemplative nature of mimicry will be important in this volume, forming a central theme in our discussion of group behavior (in chapter 6).
Third, let us turn to emotional attunement, which we will consider at greater length. Emotions are often opposed to beliefs precisely because they are not thought to be deliberative, and feeling is not usually regarded as a species of “rational” thought. The question that Gilbert considered for belief was whether there is a reflective stage between comprehending a declarative sentence and coming to believe what it describes. For emotion, then, the question must be whether there is a reflective stage between comprehending an utterance and reacting to it emotionally. Although we must be careful not to make unwarranted empirical claims, it seems somewhat obvious that while emotional response can follow reflection, much emotional response is more immediately reactive. Intuitively, once one understands that someone has said something complimentary (or insulting, or shocking, etc.), one can only dissociate from one’s immediate emotional response slowly. That is, we tend to feel first and ask ourselves questions later.
17 Words can therefore hurt us by a direct, causal process. Like sticks and stones, words have a power that can bypass the free will of the hearer to evade the harm, or at least make the harm difficult to avoid. In the context of the model that we have introduced in the last two chapters, this should come as no surprise. We have used a metaphor of resonance precisely because it analogizes communication to a physical process of energy transmission, such that a suitably “tuned” hearer will, by virtue of their attunement, be affected in accord with whatever is the conventionalized effect of the message.
We hypothesize that there are speech acts for which a primary function, and in some case the entire raison d’être, is that their conventional resonances include emotional effects on the hearer in particular contexts, whether positive emotional effects or negative emotional effects. For example, as regards the speech act of apologizing following a minor faux pas, one important function is the attendant effect of mollifying the addressee. Suppose that someone apologizes to another person who has just been mildly and not entirely intentionally discomfited by their actions. If the addressee is fully attuned to the practice of apologizing, and is also suitably attuned to the details of what happened, it would follow that they will be at least somewhat mollified, for that is part of what it means to be attuned to the practice.18 The practice of complimenting is often instantiated through the performance of what appears to be an assertion. Yet its raison d’être is not providing information, but making someone feel good.
Someone who has no tendency whatsoever to feel good when complimented, or, for that matter, to feel bad when insulted, is then not fully attuned to the practices. If nobody had such tendencies, the practices of complimenting and insulting would not even exist as such.
In the last chapter, we defined what it is for someone to be attuned to something. To repeat: an agent is attuned to something to the extent that their state and behavior predictably evolve in accordance with its presence in the agent’s context. In the case of attunement to a practice, this means that the agent’s behavior aligns with this practice. Thus, since it is normal after someone has just insulted you to feel bad, being attuned to the practice of insults thus implies that you are subject to this resonance, and, contrarily, if you completely lack any tendency to feel bad, then (lucky for you) you are not fully attuned to the practice of insults. (If you “get” the insult but do not feel hurt, then you have what we have termed a second-order attunement to that insult.) Similarly, but perhaps more tendentiously, since it is somewhat normal in the community for people to come to believe something that has just been asserted, someone attuned to the practice of assertion should then have a predilection to come to believe assertions.
To say that people have such tendencies or predilections does not imply that they have no choice but to feel bad when insulted, or to believe what is asserted. It also doesn’t imply that uptake is nondeliberative, for the fact that someone’s state has a tendency to evolve in this way does not in and of itself determine whether their state tends to evolve this way through conscious reflection, or through some other mental process. Furthermore, the difficulty of learning how other people’s mental states are evolving while they are engaged in a practice presumably makes it difficult for temporally fine-grained requirements on the evolution of mental states to become conventionalized as parts of the resonances of communicative practices. However, if the development of a certain mental state is a resonance of a practice in certain types of context, then we should expect the mental states of community members engaged in the practice to evolve accordingly by default in contexts of those sorts. In contexts where the opinion of the insulter matters, then, an insult should hurt by default, and in contexts where the opinion of an asserter is trusted, an assertion should be accepted by default. It would at least be unsurprising if such default behavior were implemented through something like the nondeliberative uptake process that Gilbert describes. This means that, at least in certain contexts, people’s states evolve semiautomatically in line with whatever is normal for the practice. In such contexts, it would be mentally effortful to fail to withhold belief in an asserted claim, and mentally effortful to avoid feeling bad when insulted. A deliberative uptake model, by contrast, is suggestive of similar mental effort for acceptance of a claim as for rejection of a claim, and were the deliberative uptake model extended to emotional impact, it would imply equal mental effort to feel bad when insulted as to maintain one’s composure. The idea that an utterance can affect people’s emotional state immediately, as they comprehend it, and without reflective deliberation, will be important as we consider in this chapter and later in the volume the functioning of hate speech, and of slurs in particular.
The takeaway from our consideration of nondeliberative uptake is simple: while some of the effects of communication on hearers are doubtless mediated by reflective deliberation, others are clearly not. The question, then, is which effects are more deliberative, and which are more, to use Gibson’s term, direct. Put in terms of modern psychology, the hybrid position we advocate resembles what are known as dual-process models, in which some processes are automatic and easy, and others are reflective and effortful, as in our brief discussion of models of persuasion in chapter 2.
19 But here we make no commitment to such processes being psychologically distinct or independent, and we merely take it that there is abundant psychological evidence that some processes tend to be more reflective and effortful, and others less so. From this point of view, we should not be asking whether it is Gilbert’s Spinozan- or Cartesian-influenced model that best describes how communicative acts affect people, but rather which uptake processes are more deliberative, and which are more automatic. We will be returning to this question in the next chapter, where we study a large class of communicative effects not explicitly considered by Gilbert (or, for that matter, Spinoza and Descartes), namely, the effects of presupposition, roughly the collectively assumed background behind an utterance.
As will be seen, a wide range of evidence suggests that much presupposition processing, especially the processing of presupposed framings, is highly automatic. While Gilbert may have been only half right as regards uptake with assertions, the evidence suggests he might have hit the mark regarding uptake with presuppositions.
3.4. Individual Harmony Our discussion in the last two sections concerned what prior literature tells us about how people adapt to new information, how people re-attune. We drew the following conclusions: (i) Communicative effects are not, or at least not always, a matter of deliberatively weighing evidence, and at least some uptake may occur semiautomatically.
(ii) The processes that determine how our cognitive attunements change inherently involve noncognitive attunements, such that logical consistency cannot be the prime factor.
Attunement change is based on a more holistic property of systems of attunements, so that, as we put it above, the emotive and dispositional tail has the potential to wag the epistemic dog.
(iii) If we sometimes accept information without fully considering its implications, then we must sometimes end up with contradictory beliefs, which implies that our beliefs must in some way be fragmented. Generalizing, there at least are several ways we can deal with the problem that would occur when attunements are at odds with each other: we can drop or fail to develop some of the attunements, or we can compartmentalize, avoiding dissonance by fragmenting the space of attunements into subspaces in which there is less conflict.
We now discuss one way these insights might be brought together into an account of harmonization, the special way that groups of attunements evolve. Our starting place is the goal of harmonization: harmony. We will consider two subtypes of harmony: individual harmony and collective harmony. The first is discussed in this and the following sections, and collective harmony is introduced in sections 3.7 and 3.8.
We begin with individual harmony, defined as follows: Individual harmony: What is experienced emotionally when one is aware of how one’s attunements relate to one another, a sense of consonance (positive harmony, or just harmony when this will not cause confusion) or dissonance (negative harmony).
Consonance: The experience of manifest coherence of systems of attunements.
Dissonance: The experience of manifest incoherence of systems of attunements.
This then leads to a derivative definition of harmonization: Harmonization: The process by which groups of attunements evolve in order to bring about positive harmony.
One way in which the musical metaphor seems to us apt is that it suggests that people may at times find a limited degree of dissonance acceptable, and even in a certain way a positive, provided it is followed by a suitable resolution. This comports with our intuitions about human behavior. For example, people may actively seek the discomfort of learning so new that it does not fit with their preconception, which may involve experiencing a lack of coherence for a while, with the hope that this will eventually bring them to a new stage of harmony in which there is greater consonance.20 Our definitions of consonance and dissonance involve several further notions, referring to (i) coherence and incoherence, (ii) the idea of a system of attunements, and (iii) the idea that coherence or incoherence can become manifest.
Although we further discuss coherence in the context of narrative, in section 3.5, for the most part we leave it unanalyzed. What we do want to stress is that coherence is not just logical consistency of sets of propositions, and indeed it cannot be, because we do not take attunements to be purely propositional. For example, a dispositional attunement to play a competitive sport may cohere with a cognitive attunement to the effect that playing this sport is healthy, as well as cohering with emotional attunements whereby sporting performance brings pleasure, but conflict with a cognitive attunement such as a belief that playing this sport is not a wise use of limited time, a dispositional attunement to sit at home and watch this sport on TV, and an emotional attunement involving a fear of losing.21 The bulk of this section is taken up with exploring the broad idea that fragmentation creates subsystems of attunements that have a tendency to be activated together, and across which coherence or incoherence might be sensed. For current purposes, it suffices that coherence and incoherence among attunements will normally only become manifest when those attunements are simultaneously activated. Conscious higher-order reflection on one’s attunements could also lead to evaluation of attunements being coherent or incoherent, but as we understand the psychological literature, such conscious processing is usually not taken to be the source of felt consonance or dissonance.
The resonance-based framework revolves around association, and the approach we now take to developing an account of systems of attunement, and thence harmonization, centers around a consideration of what it is for one attunement to be associated with another.
The prestige of association as an explanatory tool in psychological work has swung backward and forward for centuries. The proposal developed in this volume is not strictly associationist, insofar as we don’t make any assumption at all that attunements themselves are just a matter of association: in principle, an attunement could involve arbitrary algorithmic processing. We certainly make no claim that association is the be all and end all of mental processing. Rather, we think, as so many have found before us, that for a relatively simple idea, the mechanism of association has remarkable power. This power explains why association remains an important theme in modern psychology, albeit tempered with many other mechanisms.
We think that the power of association is underutilized in most contemporary theories of the semantic and pragmatic function of language.
Our theory of how words carry meaning is somewhat associationist, with the crucial distinction that the associations we are interested in are regularities shared by members of communities of practice and would not exist if they were not sustained by feedback loops within such communities.
What we assume about human psychology is, in line with our account of word meaning, about to get a little more associationist. Just as Hume considered that thoughts can become associated with each other when the sense stimuli that trigger the thoughts regularly co-occur, so we will consider under what conditions attunements can become associated with, or dissociated from, each other.
On the view we have developed, people associate a word with its resonances. That is, the resonances of communicative actions just are associations, albeit associations created and perceived collectively by the members of a community of practice. To be attuned to something is to associate different things with each other as a group, for example, to associate the different behaviors and effects that constitute a practice with each other, and to conceive of that grouping as a single practice. Some of these associations are of the body as much as they are of the mind. We might say that a word is, in part, a certain type of association between the physical behaviors that produce it, the prior contexts it is used in, and the functions it has, as well as with other behaviors that, taking a more interactive stance, tend to be performed by interlocutors when the word is used.
In terms of how they relate to systems of attunements, we suggest that fragmentation and association can profitably be seen as two sides of the same coin, a single property: we propose that fragmentation is no more and no less than dissociation. To see what this means, start with the idea that attunements can be active or inactive, depending on whether we are performing an activity that depends on those attunements. When someone is playing chess, their attunements to chess will be active, and when they are not thinking about chess at all, those attunements will be inactive. For two attunements to be associated is for there to be a pattern of coactivation.
However, conversely, there are patterns of deactivation, by analogy (and perhaps it is something more than mere analogy) to neural connections, which can be excitatory or inhibitory. We suggest that fragmentation derives from the fact that attunements do not each exist in a vacuum, but rather are associated or dissociated with each other in the sense that activation of one attunement tends to lead to activation or deactivation of others to varying degrees. Pedaling a bike is tightly associated with steering the bike and maintaining balance, but less associated with facts about how bike riding works in terms of the physics of balance and the mechanics of the bike and the body, and still less associated with the practices of playing chess, frying fish, or voting. This is just to say that biking dispositions are somewhat compartmentalized relative to other attunements. More generally, if there is any systematicity at all to the associations between attunements, that is, if there is any systematicity to the patterning of which attunements are commonly active together and which are rarely active together, then fragmentation is inevitable within our model of attunement.
To switch to a more holistic level, consider again the Nazi prison guard who goes home to their family and is a loving, tender parent. How does compartmentalizing work so effectively, so that their brutality in one context does not bleed into the other? From the point of view of an associationist model like ours, it would be more surprising if such a person failed to compartmentalize. The guard is engaged in different practices at different times, and our model offers no intrinsic reason for there to be a strong connection between the two. Of course, as onlookers, we see both patterns of behavior simultaneously, and the contradiction between brutality and tenderness is manifest to us. But to think this way is to fail to see through the prison guard’s eyes. In one context, they focus hawklike on untrustworthy objects of disdain and hatred who must be subjugated, and in another they proudly gaze upon beloved and delicate children to be nurtured.
Our model of fragmentation depends on different attunements being differentially active.
When we talk about levels of mental activity for different attunements, we are implicitly talking about attention and how it changes. Leaving aside for simplicity the difference between conscious and unconscious mental activity, the attunements that are active help define our attention. The spotlight of attention consists in active attunements and all the things that play a role in the particular application of those attunements, for example, the combination of bike-riding attunements and the bike and the road and the feelings in our body as we ride. Attention is in and of itself an important topic. It doesn’t take a political savant to recognize that attention is exploited by manipulative political messaging, and is relevant to the evolution of public concern even when not directly exploited.
Earthquakes lead only to a temporary increase in attention to earthquake safety, and although we are unaware of any direct empirical comparison, we suspect that collective attention to the misdemeanors of politicians has an even shorter half-life. One reason for this is that politicians and their media representatives actively distract us, bringing new things to our attention, so that we no longer attend to what they wish to distract us from. To model attention, whether on a more classical view or in the attunement framework, some machinery must be added. We simplify by amalgamating the notions of attention and mental activation, which in psychology are distinct notions. This simplification is not intended to represent a theoretical stance, but rather a placeholder for future work. Let’s adopt the following recipe: whatever your favorite model of a mental state (say, a set of propositions, or a set of attunements), add to it an activation function. The activation function maps each of the constituents of the mental state to a number, a level of activation. If your favorite model of a mental state was pairs of propositions and their credences, then the new model would consist of a function from pairs of propositions and credences to activations. Our favorite model of a mental state is a system of attunements, so we now have a function from attunements to a pair consisting of (a) a level of attunement and (b) a level of activation. An oversimplistic illustrative implementation of attention, for any of the above variant models of a mental state, would simply map unattended objects to 0, and objects in the spotlight of mental attention to 1.
Before going on, let us point out here that the moves we have made so far are not at all radical. Within philosophy, at least philosophy of mind, discussion of attentional processes has been on the rise in recent years,22 and in work on attitude shift and framing in political science, one of the dominant formal models for the last four decades uses attention in a way quite similar to our own. We are thinking here of the expectancy value theory of attitude developed by Martin Fishbein, partly in collaboration with Icek Ajzen.23 In that model, an individual’s attitude toward an issue is a weighted sum of evaluations of particular attributes and the salience of those attributes. On this view, persuasion operates not necessarily by presentation of new facts, but by use of framings that shift attention, directing salience toward the attributes for which evaluations are in line with the framer’s goals.
The moves we will now make, and indeed predecessors of our terminology, are also found in the reading-comprehension literature. In the landscape model of reading comprehension, developed by Paul van den Broeck and colleagues, conceptual representations are activated at different levels determined by what was previously read and integrated and background knowledge; the mental state of the reader is seen as a vector mapping concepts to levels of activation, a vector that changes dynamically as text comprehension proceeds.24 In this and other reading-comprehension models, the representation of the text that is formed is affected by the requirement that a particular reader has for textual coherence, and the effort the reader puts into the search for a representation that is globally coherent.
In our terms, readers seek a harmonious understanding of the text, and they search through a space of possible representations when faced with dissonance. But note here that while some models of reading comprehension focus on active and strategic integration of material into their mental representations of the text, others focus more on passive, automatic processes. In one branch of the literature, these automatic processes are described using a resonance metaphor: new representations are formed partly as a result of resonances with previously activated conceptual representations.25 Let us now add a little geography to the mental landscape of our harmony model, by assuming the existence of an inter-attunement distance that associates any two attunements with a positive value, the distance between them.26 The inter-attunement distance represents the level of association between the attunements, that is, the degree to which activation of one attunement is likely to lead to activation of the other. This creates something like neighborhoods of thought, that is, sets of attunements that tend to be simultaneously activated to some degree because the interattunement distances are low. The landscape of attunement for a person is the sum of their neighborhoods of thought. Our model now consists of two functions: a function from attunements to a pair of a level of attunement and a level of activation, and a function from pairs of attunements to distances.27 This immediately creates the possibility of attunements that are in some sense inaccessible, an intuition that is basic to Freud’s tripartite model of the mind, and that Freud explained with a suggestively spatial metaphor in a 1933 lecture: Symptoms are derived from the repressed, they are, as it were, its representatives before the ego; but the repressed is foreign territory to the ego—internal foreign territory—just as reality (if you will forgive the unusual expression) is external foreign territory.
28 As already stated, we are not attempting to theorize the difference between conscious and unconscious processing, and we will not invest our model with foreign territories as such. We merely take the proximity between two attunements to be a measure of the probability that activation of one attunement will lead to activation of the other.
29 Given that this chapter builds on an associative theory of resonance and attunement, it is natural that the inter-attunement distance be given a Humean interpretation: proximity between attunements corresponds to strength of association, or, in contemporary psychological terminology, the degree to which attunements prime each other.
A simple implementation of the inter-attunement distance might map the distance between any two concepts concerning, for example, family to zero, any two sports-related concepts to zero, and any pair of a family thought and a sports-related thought to one. This would mean that familial thoughts were all interrelated, sporting interests were all interrelated, but family thoughts were unrelated to sporting interests. That would yield exactly two neighborhoods of thought. A minimalist model of attention could then consist in a stipulation that if some attunement is activated at level 1, then all attunements at distance 0 from it are activated at level 1, and all attunements at distance 1 are activated at level 0, that is, that only one of those neighborhoods could be activated at a time, like street lights in a county experiencing power shortages. Hey presto! We have created a model of an agent with a two-track mind.
The agent has three cognitive states: total inattention (emptiness of mind), thinking about sports, and thinking about family.
A more sophisticated model might not only have gradient distances and activations, but also allow flexibility in whether attention was focused on a narrow region of mental space, or was somewhat diffuse and dispersed over a larger region. In such a model, the landscape of attunement would not absolutely determine where an agent’s attention was directed, but would rather provide defaults for how attention would evolve absent mindful control. It would then require effort to focus on only one of two attunements with a strong association, or to simultaneously activate two attunements that were not tightly associated. That is, we might relate patterns of attention to processing cost. A high cost would be incurred by focusing attention simultaneously on highly disconnected regions or on only a small part of a highly connected region, and a low cost to a pattern of activation in which only a cluster of highly connected attunements were activated, and no other highly connected attunements were activated.
Let us apply some of these ideas in terms of the examples we have introduced in the chapter, starting with the prison guard who lives a double life. For the prison guard to see what we see would require an act of reflection as regards the relationship between different practices they engage in, and such reflection would inevitably be painful. That is, it would create dissonance. Nazi prison guards who managed to maintain double lives with equanimity were, we speculate, precisely those who avoided performing such inconvenient acts of reflection, and thus were easily able to maintain the boundaries between mental compartments. The boundaries arose naturally as a reflection of the intrinsic compartmentalization of the guards’ different roles in life, and reflection is precisely what their discipline mitigated against, discipline instilled in them through propaganda, training, and workplace and family practices. We are obviously not experts on prison-guard psychology. We intend this example of the sort of account of a real-world behavior that might be given in the framework we have developed only as an illustration and an intuition pump.
What it illustrates is that fragmentation is hardly a surprising addition to our fundamentally associationist model. It is rather a natural consequence. And there are two related intuitions that we hope the example pumps, one concerning dissociation of practices, and the other concerning avoidance of dissonance.
Earlier, we noted a problem that a model of fragmentation imposed on top of a classical logical representation faces, namely the problem that there is no clear basis on how sets of propositions might be fragmented. In our terms, the prison-guard example suggests that at least one (on reflection, unsurprising) source of separation between groups of attunements is the separation of practices in someone’s life. On the view we have developed, it would be unsurprising if a professional logician were far from logical when cooking or arguing with their spouse, because there is a real-world separation between the practice of being a logician and the practices of being a cook or a spouse. Although a logician might well reflect on the relationship between the two, there is nothing forcing them to do so. At this abstract level, which leaves aside the important moral differences between the two cases, the difference between the philosopher and the prison guard is simply the degree of dissonance that simultaneously embracing the different facets of their life might bring.
This takes us to the second intuition suggested by the prison-guard example. Our discussion of the separation of practices might be thought to make an implausible assumption about the prison guard. A skeptic might ask how someone who spends so much of their lives behaving in such dramatically opposed ways could possibly be so unreflective that they do not become aware of, and perhaps then obsessed by, the (to outsiders) obvious tension between their different “forms of life.” This is where the concept of cognitive dissonance comes to the fore, although our framework introduces a possibility that was not present, or at the very least not explicit, in Festinger’s work. Put in everyday terms, our hopefully somewhat commonsensical suggestion is that a prison guard might fail to reflect expansively on the tension between their different forms of life precisely because to do so would be painful. In terms of our framework: preferring consonance, the guard avoids a mental activity that would bring dissonance, and so conveniently allows the two practically separated parts of their life to also remain mentally fragmented.
Before considering the ramifications of this reasoning for our framework, let us note that we have brushed upon a question that has filled volumes. Are the perpetrators of great evils, and in particular heinous war crimes, in some sense deeply bad people before they commit crimes against humanity, or are they merely “ordinary men” who become twisted by propaganda and the extraordinarily violent and perverse situations in which they find themselves?30 We do not take a position on this issue here, beyond noting that converging lines of historical and psychological research suggest that it is common for people to have the potential to adopt inhumane practices in situations where they have extreme power over others.31 What we do want to note is that the type of compartmentalization that emerges in our framework mirrors a twist on this question, a Freudian twist now so familiar in psychoanalytic work as to perhaps seem unremarkable. For we might also ask whether the perpetrators of great evils are sometimes still ordinary people in other walks of life. The geography of attunements at least allows the possibility that the same person be depraved in one form of life, and saintly in another. Returning to the role of cognitive dissonance in our framework, recall that Festinger argued for a position whereby people’s epistemic attitudes can be swayed by avoidance of dissonance, as when someone fails to accept evidence that would challenge their dispositions and desires. But in the cases Festinger discussed, the issue was discussed on the assumption that acceptance of evidence is a binary: one accepts the evidence or fails to accept it, and, more generally, one either does or does not believe things. In terms of attunements, this would amount to either having an attunement or not having it. Our attentional account of systems of attunement brings a further possibility.
Since harmony is an experience of the relationship between attunements, one can avoid felt dissonance by suppressing some subset of the attunements that would otherwise be in tension, and not only by eliminating or avoiding attunements. If suppression meant permanent suppression, such that suppressed attunements were never activated, there would be no practical difference from Festinger’s model. But suppression need not be permanent or total. Leaving aside Freud’s well-known claim, cited above, that repressed ideas or tendencies can still be psychologically active, suppression in our model can be temporary. Fragmentation allows that two sets of attunements can be in complementary distribution: one set is active only at times when the other set is suppressed, and vice versa.
Thus, there are at least two ways of avoiding dissonance: first, there is the Festinger approach, of abandoning the weakest attunements that are causing felt dissonance, and of refusing to take on new attunements that would cause such dissonance, and second, there is the option of pushing attunements that would combine problematically into separate domains of the attunement landscape, such that they are rarely or never coactivated, with the consequence that it is easy to ignore the fact that they are in tension.
Politics is rife with contradiction and hypocrisy. A politician might fight for freedom of speech and simultaneously propose legislation that curtails it, for example, by banning books or legislating against issues being discussed in the classroom. The question is one of why supporting a strategy that simultaneously suggests support for antagonistic goals, undermining the ideals it advocates, does not lead to cognitive dissonance.
Part of the answer is that neither fighting for freedom of speech nor legislating what educators can say is propositional. They are behaviors.
There is little tension between the behaviors for one who does not wish to reflect on the relationship between the ideals behind them.
3.5. Narrative Harmonization Attunements can be structured in complex ways, ways that are not completely captured simply in terms of coactivation or inter-attunement distance. Most obviously, attunements to simple practices can jointly constitute attunement to a complex practice. Attunement to the practice of speaking a language is like this, consisting of attunement to many simpler subpractices. A related type of structure is that imposed by a narrative or plan. A plan relates attunements to each other by enabling goals to be met by organized performance of separate actions, so that the plan calls for joint or sequential activation of attunements.
Narratives can provide much richer structure than this, determining not only courses of action, but also ways of understanding the world and our place in it. Our goal in this section is to show in outline how narrative structure may be a source of coherence for attunements, and hence a driver of harmonization.
Jack Balkin lists some of the things “we use narratives for,” clearly for him a partial list: 1.
Remembering events in temporal sequences.
2. Ordering and organizing the past.
3. Explaining human action in terms of plans, goals, and intentions.
4. Understanding our own selves and motivations through autobiography.
5. Giving causal explanations of events.
6. Creating expectations about the future.
7. Internalizing expectations about how to behave in social situations and interact with others.
8. Providing scripts that tell us how to understand social situations, engage in social conventions, and assume social roles.
9. Creating notions of what is ordinary and extraordinary, expected and unexpected, canonical and deviant in social life.
10. Accounting for deviations from what is ordinary, expected, or canonical. 11. Creating social myths and shared memories that unite groups we are part of, frame their experience of contemporary events, and produce shared expectations about how the group is supposed to behave.32 Hearing a commotion in a restaurant and looking over to see one patron with wine on their clothes and another in the act of leaving the restaurant, we rapidly construct a narrative as to what occurred. We may be right and we may be wrong, but we certainly use our ability to access a store of narratives to rapidly interpret a small amount of evidence in a rich way.
This is not always a good thing. At time of writing, there is a US national news item of a familiar type: a Black pastor, the Reverend Michael Jennings, was watering a vacationing neighbor’s petunias and hydrangeas while they were on vacation, and another neighbor called the police on him because they took him to be engaging in suspicious activity. Police approached him, and, even though he had a watering hose in his hand and identified himself as a pastor watering his neighbor’s plants, they arrested him. For both the person who found the pastor’s actions suspicious and the police, the narrative that came most readily to mind in seeing a Black man in the yard of a nice house involved criminality.
33 We often relate narratives by uttering a series of sentences.34 But, following Rachel Fraser, we reject thinking of narratives as characterizable in terms of a list of sentences.35 Fraser rejects the set-of-sentences model, essentially because narratives have to be holistically coherent, in a way that cannot always be captured by a string of sentences. It is because we approach interpretation with narratives that we can make coherent sense of a string of sentences—not the other way around.
The coherence brought by a narrative, then, is not simply logical consistency.
36 When attunements are structured according to a narrative, the coherence of those attunements will depend on how well they match the narrative, and how coherent the narrative is. Narratives involve events and actors. The events are often sequenced, or at least are usually connected in some way, and the actors have characteristics of importance to the story that may include, for example, taxonomic kind, personality, personal circumstances, societal functions, or identity. Both events and actors may be given valuations, as when an event is calamitous, or an actor is a hero or villain. Let us say that a narrative frame is an abstract template that consists of (i) a set of principal actors that have particular characteristics and relationships with each other, (ii) a set of connected events that involve those actors and locations and lead to particular changes affecting the actors, and (iii) optionally, valuations of some of the actors, behaviors, or events. An instantiation of a narrative frame is then a narrative in which the abstract events, actors, and locations of the narrative frame are identified with particular events, actors, and locations. To be cognitively attuned to a narrative frame is to have a disposition to see groups of events, actors, and locations as instantiations of that frame. Importantly, such a disposition can bring with it a tendency to reason and act based on the narrative frame.
Suppose someone experiences dissonance among attunements because a disposition to act in a certain way is not justified by what they know about the world. Then narrative instantiation can potentially enable harmonization via a backfilling process (the narrative equivalent of abductive reasoning in logic). That is, the person experiencing dissonance fills in their understanding of the world on the assumption that the world conforms to the narrative. Just as when someone observes smoke and then infers abductively the presence of fire, one may find oneself alone and then draw solace from the idea that one is on a hero’s quest. What we will term narrative harmonization is a change in a system of attunements based on the supposition that real-world characteristics of individuals, relationships, and events match the characteristics found in a narrative, and that behaviors portrayed positively in the narrative are normatively desirable.
To take a simple example, consider the narrative frame of the saying “The early bird gets the worm.” Suppose that someone desirous of some prized object acts too hastily to obtain it, and fails, thus producing dissonance. They may bring coherence to their action by mapping themselves onto the early bird actor in the narrative frame, the prized object onto the worm, and their disposition to act quickly onto that of the early bird. Within this framing, their attunements are seen as coherent at a high level, even though in this particular case the prize was not obtained.
Since narrative harmonization can involve treating a narrative frame as normative, it can be used to justify behavior using the narrative. But it is possible that someone’s attunements to a narrative go further than justification of behavior using the narrative, and that they actually behave according to what the narrative suggests is desirable. If someone has a tendency to behave by analogy with characters drawn from a narrative frame, we can say that the individual is dispositionally attuned to that narrative frame. Much education and indoctrination are based on the idea that stories serve as models for behavior and misbehavior. The idea is presumably that attuning people dispositionally to the narrative frame of a holy book, or for that matter a book of folk tales, is a more effective way to inculcate in them a set of ideological practices than simply telling them what to do.
Let us note here that in invoking narrative frames, we do not claim to have provided a theory of when attunements are coherent, but rather to have related the question of when attunements are coherent to the question of when narratives are coherent. Although we will not attempt to provide a general account of narrative coherence, it is broadly clear what the components of such an account might include. Narrative coherence presumably depends in part on the ease of identifying causal connections between the events in the narrative, and the degree to which the interplay of actors within the events is explained by their characteristics and relationships. However, we suppose that a theory of narrative coherence would be a psychological theory, and so might also invoke gestalt principles involving the organization of the narrative. For example, a theory of narrative coherence might involve the presence of clear parallelisms or contrasts, such as the implied parallelism between the early bird and other birds implicit in the proverb, the clear contrast in how early the different birds were, and the clear contrast as regards who got the worm. A narrative may suggest a plan of action, but it may also serve to help us understand what is happening without suggesting a future course of action.
It is important to realize that the story does not need to be made explicit, because our culture comes replete with a stock of narrative frames, and a single word may have sufficient resonance to invoke one. For example, on February 15, 2022, the lieutenant governor of Texas, Dan Patrick, sent out the following tweet, which, since it refers in an ugly way to faculty at the University of Texas at Austin, and since it was followed up at a press conference with a threat to dismantle the University of Texas tenure system, happens to hit close to home for us: I will not stand by and let looney Marxist UT professors poison the minds of young students with Critical Race Theory. We banned it in publicly funded K-12 and we will ban it in publicly funded higher ed. That’s why we created the Liberty Institute at UT.
37 Dan Patrick’s use of a poisoning metaphor immediately conjures a narrative frame in which anybody can identify the main roles: academics = villains, students = victims, Patrick = savior. In the framing of this tweet, critical race theory obviously has a role familiar from history, mythology, and fairy tales: poison. The use of narrative frames allows persuasive rhetoric to be effective without defining its terms. In the example, the term “critical race theory” has no clear definition, and it has never received a clear definition from those on the right attacking it, though it is clear that their use presupposes a far broader notion than is employed in literature on critical race theory in legal scholarship. In the mouths and tweets of right-wing politicians and journalists, it is not a term that refers to a series of concepts with which a reasonable person might agree or disagree; it is a role in a narrative. We discuss the demonization of critical race theory, and its relationship to racist politics, in chapter 11.
Plans and narratives provide alternative ways in which attunements are structured, but they also provide alternative ways in which tension between attunements can be introduced or dispelled. For example, someone may feel tension when faced with a need to perform an action that takes them further from their desired goal, but that tension could be somewhat resolved when the action is seen as part of a larger plan that will eventually take them to that very goal. That is, a plan can explain why one is doing something undesirable, by justifying it as a necessary subgoal of the original goal.
Narratives are more general than plans, since they intertwine combinations of actions more deeply with ideology, and with issues of identity, character, and society. Narratives can be aspirational, helping to set goals for an individual by offering models for behavior. A narrative can create tension by suggesting that an individual is diverging from standard behavior, or, contrarily, dispel tension by explaining behavior as conforming to a standard template.
People can be led to perform extreme acts that would otherwise create unsustainable personal tension, such as acts of war, if they see these acts as part of a heroic story in which their actions play a positive role in establishing a greater good.
Narratives and plans provide organizational frameworks that group attunements together.
Relatedly, narratives and plans affect attention. They direct attention by providing default expectations of sequencing and colocation of events. They thus drive processes of fragmentation, or potentially defragmentation, as when we start to see something in one part of our lives as a metaphor for something in another, and thus as exemplifying a common narrative frame. Narratives can also influence attention by enhancing vividness of particular situations or events. Like stereotypes, and often in tandem with stereotyping, narratives provide schemas that enable people to better focus on and remember certain things, or at least to focus on how the narrative depicts those things. The narrative can tell us which tensions to focus on, for example, the tension created by a perception of threat or loss.
Likewise, the narrative can also bring about harmony, by drawing our attention away from dangers or inconsistencies, and toward things that are consonant with our self-image and well-being.
Harmony, then, need not require our beliefs to be explained scientifically or logically, but rather to be integrated, to be given a meaning in the context of some general structure. If so, then harmony can be achieved not only by drawing valid inferences, the method usually associated with logic and science, but also by assimilation to patterns and schemas. To recognize something as fitting into a pattern is not to deduce anything, but merely to categorize it. Narrative harmonization is sufficiently powerful that it might allow an entire form of life to be categorized, to fit comfortably into the interwoven patterns of a complex tapestry of stories.
3.6. Priming Hate The associationist model we have sketched provides a way of thinking about an important process for the politics of language, namely priming. By exposing people to words that are associated with practices, and perhaps inciting them to repeat or chant those words, attunements to the associated practices may themselves become activated. Mere exposure to the right words may help get an audience into the right frame of mind, a frame in which they are attuned to a demagogue’s speech or a crowd’s behavior, and perhaps are even somewhat pliant. This is not to say that such processes are always problematic. Practices of ritual chanting of scriptural verses and mantras, whether individually or en masse, are widely felt to be beneficial, simultaneously combining the apparent opposites of being a grounding discipline and spiritually uplifting.
As we have noted, Lieutenant Governor Dan Patrick’s invocation of critical race theory is rhetorically functional even though in its right-wing usage the term lacks clear definition. The phrase does not function primarily to help the speaker convey a precise proposition, but to raise to prominence a web of associations. This web of associations helps form a tightly connected region of attunements, a segment of an ideology. In a similar vein, but with less ideological density, consider the function of Donald Trump’s repeated use of epithets, such as “Crooked Hillary,” to label his political opponent Hillary Clinton. The obviousness and superficiality of his strategy in no way blunted the power of the relentless jab to prime an association between Clinton and crookedness. It is hard to believe that anyone who experienced the 2016 US election campaign would not, as a result of both this phrase and the constant chanting of “Lock her up!” exhibit at least some priming effect whereby after hearing the word “crooked,” the word “Hillary” was primed, and vice versa.
To take another much-discussed example, consider the significance of priming for group stereotypes. Such stereotypes are involved in a wide range of race and gender vocabulary, including racial and gender epithets.
The model of attention we have outlined suggests that once language has activated an attunement to a certain categorization, perhaps using a racial or gender stereotype, further attunements might then become activated. A hearer who is exposed to a labeling of someone using a term for which there is a strong stereotype would activate inferential attunements that involve tendencies to draw further conclusions about the individual who has been so categorized, as well as further dispositional and emotional attunements. These in turn might lead to ways of treating that individual and ways of feeling about that individual that fit within the ideology of the hearer. Thus, we might say that a stereotype is not merely a set of default properties; it is a doorway into an ideology and to all the emotions and practices that belong within it. In all these cases of priming, our claim is not that a hearer is necessarily powerless to resist associations that have been seeded by a history of occurrence of words in certain contexts, but that it will cost a hearer effort to keep a clear head. For example, it does not follow from our account that those who experienced the 2016 election had no choice but to develop a dispositional attunement to treat Hillary Clinton as the less ethical of the two candidates, but it came to require at least some effort to keep a clear head and to recognize that the association between her name and criminality was planted in our heads independently of any clear evidence about her behavior.
Priming is reflexive activation of one attunement by another, an association between attunements. Political persuasion is partly a matter of strengthening the associations between some things, so that associations with other things will be weakened. For example, in the aftermath of massshooting tragedies in the United States, right-wing politicians reliably mention mental-health problems, as well as “good guy with a gun” rhetoric, distracting from one of the primary causes of the tragedies, namely, the ease with which bad guys, mentally ill or not, can get guns. Similarly, if voters are able to rapidly recall a narrative regarding some politician’s stance on crime, that could reduce their tendency to think about economic or welfare stances when voting.
The obverse of priming, where activation of one attunement suppresses activation of others, is a natural consequence of our basic assumptions. This is what creates the possibility of what might be termed distractive priming, whereby the fact that processing of one thing has been facilitated may mean that processing of something else is attenuated. The psychological connection between priming, on the one hand, and attention and distraction, on the other, is a complex one. Attention and distraction are inherently short-term processes, while priming effects can be long lasting, and can affect processing without having a direct effect on attention. Indeed, the fact that someone is primed to process something in a particular way can potentially reduce the amount of attention that the processing requires.
We thus resist making strong claims about exactly how priming relates to attentional effects.
What we can say is that politicians both distract people in the short term, moving the conversation away from topics that they would rather people not think about, and also use distractive priming as part of longer-term strategies.
Here, let us draw a connection with nondeliberative uptake, which we use not to refer to Gilbert’s theory of hearer belief revision per se, but to the idea that some effects of communication are effortless and semiautomatic.
One effect of communication is the impact on attention, and this impact is for the most part beyond deliberative control, as illustrated by the wellworn “Don’t think of an elephant!” conundrum (to which we previously alluded in our discussion of the exigent power of words, in section 2.6).
Politicians seek not only to drive attunement in a desired direction, but to control attention and prime certain ways of reacting to what people attend to. Thus, a theory of political language is also a theory of political psychology, and must analyze attentional and associative processes at both short and long timescales.38 3.7. Mass Coordination For some purposes, it’s better to characterize a gas thermodynamically than in terms of its component molecules. More generally, sometimes the best way to look at a complex system is holistically, at the level of the system rather than at the level of its constituent parts. In work on communication, Chomskyan linguistics has focused exclusively on language as a function of individual human psychology, while, in contrast, much work in sociolinguistics and communication studies is set at the group level. As will hopefully have been clear from the earlier chapters of this volume, we see value in both levels of description. While we defined what it is for an individual to be attuned to (the practice of using) a word, we also distinguished individual attunement from collective attunement, and we analyzed the conventional resonances of words as collective attunements across a community of practice.
But so far in this chapter we have only described harmonization as an individual-level process.
In this section, we motivate the idea of considering harmonization at a group level, by considering cases in which the behavior of groups seems special, a thing in its own right that is perhaps most naturally analyzed, at least in part, at the level of the group rather than at the level of the individual. We begin with a simple biological example to pump intuitions, and then move to considering two more “big” ideas from prior literature, in fact from over a century ago: contagion and collective effervescence. These ideas illustrate the interest in group-level description, and also provide inspiration for development of our notion of collective harmony.
To prime the intuition as to how a theory of collective harmony might deliver value that a theory of individual harmony cannot, let us consider a relatively extreme case: the behavior of a colony of ants engaged in foraging activities in your kitchen. The ants have recently discovered a way of getting from their nest, accessed via a crack in the wall behind the sink, to your well-stocked larder. Without even realizing that a bidirectional highway of ants stretches across your kitchen, you wander bleary-eyed into the room, slide a chair out from under the table and slump into it. In the process you squash some ants and block the foraging line. Although it might be possible to argue that at the level of the individual ants there is a sort of dissonance, because they don’t have a clear pheromone-laden path to follow, this level of analysis is neither clearly motivated nor clearly revealing of what then happens: individual ants have little clue what is going on at a global level. Yet, in the course of just a few minutes, a new foraging line develops, circumnavigating the chair you pulled out, and reconnecting your larder to the place the ants call home. It is relative to the level of the ant colony that we can give a functional explanation: (i) the ants were collectively engaged in an evolutionarily optimized foraging activity; (ii) the changes you unconsciously imposed on their environment disrupted this activity; (iii) collective behavior was then suboptimal insofar as the nest was not being actively stocked; (iv) they collectively reoptimized until they determined a new locally minimal path between nest and larder, thus returning to optimized foraging activity as before.
The collective change of behavior of the ants in your kitchen is extraordinary. It is remarkable that they can coordinatively solve a logistically complex problem using only very simple chemical markers to communicate. The communication processes of ants are powerful, but rely on simple heuristics: innate and attunements, if you will. It is by now somewhat well understood scientifically how some species of ants optimize for their environments, even though the detailed functioning of the individual ant as a computational system remains a hard scientific problem.
Collective change can be explained without reference to individual ants having a detailed representation of the organization of the space they were negotiating, or dispositions that would allow them to precisely navigate that space in the absence of other ants. While we could talk of individual ants reharmonizing their dispositional attunements in reaction to dissonance created by lack of clear pheromonal signposting, what we want to suggest is that the standard explanation (in terms of processes of pheromone trail reinforcement combined with semistochastic variation in individual ant behavior) is really a story at the level of collective harmony. It is the ant collective that deals with the dissonance introduced by your sleepy entrance into the kitchen, and it does so by reharmonizing its collective behavior so that the ants are collectively attuned to their changed environment.
CONTAGION Gustave Le Bon’s 1895 Psychologie des foules39 introduces the idea of contagion as a process within crowds, using it to compare the effects of the crowd to those of hypnotic suggestibility: Contagion is a phenomenon of which it is easy to establish the presence, but that it is not easy to explain. It must be classed among those phenomena of a hypnotic order.… In a crowd every sentiment and act is contagious, and contagious to such a degree that an individual readily sacrifices his personal interest to the collective interest.
Though Le Bon does see crowds as bringing forth a positive and altruistic side of human nature, he harps on the idea that masses of people are driven by emotion rather than pure reason, stating that the crowd’s “collective observations are as erroneous as possible, and that most often they merely represent the illusion of an individual who, by a process of contagion, has suggestioned his fellows.”40 Le Bon’s work was, for a while, influential. Its political significance is seen in the mutual admiration between Le Bon and Mussolini, who in one interview is reported to have said of Le Bon, “I don’t know how many times I have re-read his Psychologie des foules. It is an excellent work to which I frequently refer,” and in a later meeting, made extensive reference to Le Bon both for political justification and for his explanation of political method, specifically the idea that the crowd is irrational and driven by emotion.41 Le Bon’s contemporary Gabriel Tarde also spoke of contagion, in his 1890 Les lois de l’imitation (The Laws of Imitation). Tarde used the term in a much broader sense than Le Bon, to refer not merely to the spread of emotion and behavior in a crowd, but to the spread of arbitrary elements of culture within and between societies. He thus essentially invented what Richard Dawkins later termed cultural memes, although Tarde’s phrase chose sociale (social thing) was clearly not as contagious as Dawkins’s meme.
42 Tarde saw the contagion of culture (he also talks of diffusion) as being explained in terms of a fundamental drive toward imitation, which in turn he saw as an automatic, unconscious, nondeliberative process. As he writes, memorably, “Society is imitation and imitation is a kind of somnambulism.”43 We do not think that society is just imitation, or that all imitation is unconscious in the way that Tarde is suggesting, but we do think both that imitation is central to ideological transmission, and that the process is often somewhat nondeliberative—more Spinozan than Cartesian, if you accept Dan Gilbert’s historical framing. Here let us note that in an important series of articles, Lynne Tirrell, taking a cue from Victor Klemperer, exploits the metaphor of disease contagion in her analysis of toxic language.44 The point here is that ideologically problematic speech practices not only spread like diseases, but they harm like diseases too, even though the harmful effects on different groups are quite distinct.
Sometimes toxic speech is debilitating due to the pain it causes, and sometimes toxic speech causes harm to those who encounter it only insofar as the attunements to the discriminatory ideology it induces are themselves intrinsically problematic.
COLLECTIVE EFFERVESCENCE In his 1912 The Elementary Forms of the Religious Life, Émile Durkheim gave a theoretical interpretation of religious life in an aboriginal Australian society that was based on the ethnographic work of other scholars. The groups Durkheim discussed were totemists, investing objects with mystical power, and each of the groups he studied in central and northern Australia regarded things that named or characterized their group as having particularly strong magical powers. If you can imagine a nation that worshiped its own flag, or a gun or a cross, or in which, to take an example from earlier in this chapter, the incantation of words like freedom and equality could leave people spellbound, then you perhaps get the basic idea of totemism, albeit totemic objects may be particulars rather than abstract types, and can be more directly causally efficacious in the totemists’ lives. Durkheim’s work is an important antecedent for our account of harmony, and at a general level, Durkheim’s account fits naturally with our own. First, as has become common in anthropological and sociological work, practice occupies center stage: there is discussion both of prohibition of practices as undesirable or taboo, and of positively approved or required ritual. Second, he emphasizes (and here he is reacting to an individualism that already pervaded much intellectual thought at the time) that the right level of analysis for religion is not the individual but the collective, in line with his general views about the subject matter of sociology. As he states in his introduction, Religious representations are collective representations that express collective realities; rites are ways of acting that are born only in the midst of assembled groups and whose purpose is to evoke, maintain, or recreate certain mental states of those groups. But if categories are of religious origin, then they must participate in what is common to all religion: They, too, must be social things, products of collective thought.45 The aspect of Durkheim’s account that most obviously inspires our own is his introduction of the notion of collective effervescence. One understands the term in Durkheim’s work primarily by example, but we would define it based on his work as a charged emotional state of activity that reigns when the individual wills of a group of people is collectively engaged in an activity that captures their entire attention and channels it into the cocreation of a single joint event. Alternatively, this definition can be rendered in terms of the account developed in this chapter.
Collective effervescence: A state in which behavioral and emotional harmony within a close-knit group dominates the collective attention of that group to the exclusion of anything else. Given that collective effervescence involves a peculiarly narrow focus of joint attention over a sustained period, it might be said that it is analogous to a flow state in the sense of the psychologist Mihaly Csikszentmihalyi.46 Note that we have taken the liberty of not including the presence of religious ritual into the definition of collective effervescence, although all of Durkheim’s examples are of religious ritual. Thus, our definition allows that collective effervescence might spring up among sports fans, ravers, and political protesters: it needs only that there are simultaneously activated collective behavioral and emotional attunement, serving to galvanize joint activity. So, on this broad interpretation, there is collective effervescence during Orwell’s Two Minutes Hate, a daily ritual in the novel Nineteen Eighty-Four in which the masses scream with rage at the principal enemy of the state: In a lucid moment Winston found that he was shouting with the others and kicking his heel violently against the rung of his chair.
The horrible thing about the Two Minutes Hate was not that one was obliged to act a part, but, on the contrary, that it was impossible to avoid joining in. Within thirty seconds any pretence was always unnecessary. A hideous ecstasy of fear and vindictiveness, a desire to kill, to torture, to smash faces in with a sledge-hammer, seemed to flow through the whole group of people like an electric current, turning one even against one’s will into a grimacing, screaming lunatic.47 If there is collective effervescence in the distinctly dissonant harmony of the hate-filled space of Orwell’s fiction, there is equally collective effervescence during the singing of “Happy Birthday” at a children’s party, young eyes glowing, the sound of smiles in voices more enthusiastic than tuneful, candles on the cake flickering. What all these occasions have in common is that a group of people brought together develops selfreinforcing, short-term collective dispositions that dominate their attention.
In both cases, it is hard to resist being drawn into the group activity, and individual will is submerged.
While Durkheim’s work still seems fresh and relevant to us, let us also point to a new series of studies that explicitly targeted collective effervescence across a range of multiday secular mass gatherings (e.g., festivals like Burning Man, where there is an ethos of self-growth and community engagement). The studies can be seen as offering support to Durkheim’s ideas, but perhaps the greater importance of the work might rest in the transformative power that the participants reported, with positive prosocial effects such as increased generosity. While the net positives and negatives of the gatherings are difficult to evaluate, the work at least suggests a positive path in a society where alienation is rife. We might say that whereas Orwell’s vision, very much in line with Carl Schmitt’s fascist philosophy, was of feelings of alienation channeled into hatred for a common enemy, the more hopeful path offered by these studies involves methods to overcome personal alienation by channeling attention toward common humanity.
48 3.8. Collective Harmony We now define harmony at the group level by close analogy with individual harmony at the level of individuals.
Collective harmony: The emotions jointly experienced by members of a group due to the way different group members’ attunements relate to each other. This may be an experience of collective consonance, when there is manifest coherence of attunements, which implies a high degree of collective attunement, or collective dissonance, when there is manifest incoherence of attunements.
The special properties of collective harmony, and, we hope, its distinctive explanatory value, derive from the fact that a central term in the definition, manifest coherence (or incoherence), comes to have a special significance in the group setting. We will consider what it is for attunements to be coherent, and what it is for that coherence to be manifest, in turn, and sketch how the definition can be applied to group behavior and attitudes.
Coherence is a matter of whether different attunements are mutually supporting. For individual harmony, we focused on support relations that might be across highly dissimilar attunements, for example, cognitive attunements like a belief in the value of riding a bike and dispositional attunements like the ability to ride one. At the group level, one person’s attunements can be supported by the presence of others with the same or similar attunements. So, coherence of attunements across a group is partly a matter of the extent to which attunements are shared, a notion that doesn’t make immediate sense at the level of the individual.49 Likewise, what it is for coherence to become manifest is quite different at an individual and group level. At the individual level, it is to be understood as involving an introspective psychological process that can involve attunements of any type. At the group level, evaluation of coherence is mediated. Attunements become manifest through behavior.
We cannot directly evaluate coherence of attunements, but only evaluate the coherence of what we see, hear, taste, smell, and feel, attunements projected onto the big screen of embodied action and sensation. Nonetheless, the impression made by the behavior of others is not only one of intellectual comprehension, but of empathic identification. The expression “mindreading” is suggestive, but it does not adequately capture this. It is not emotion reading but emotion mirroring that is at stake.50 At least for some emotions, it is experience of what others are experiencing, and thence also motivation, a drive to do what others are doing. And this we take to be uncontroversial. There is a large literature on the topic of emotional contagion and its physical basis and motivational consequences, much of this literature taking emotional contagion to be a secondary effect of physical mimicry of emotional presentations.51 It is apparently not just laughter that can be infectious.
We can pick out a further property of the manifestation of coherence of attunements that is distinctive at the group level: simultaneity, or, at larger time scales, the property of being contemporaneous. People demonstrate the similarity of their attunements by activating them in similar contexts, and people cannot be in a more similar context than when they are present at the same place and time. Density and size of a crowd are factors here: it is obvious that a behavior will become highly manifest when displayed in unison by a huge number of copresent people, or by a large number who are very close to the observer.
So, immediacy is crucial to collective harmony. At scales of days and above, and at the physical scale of villages and above, this presence of attunements changing in synchrony with each other underlies fashion, the fact that a mode of dress, thought, and expression must be à la mode in order to fit. Gabriel Tarde’s discussion of contagion as a slow process of cultural diffusion was largely set at these time scales.52 On the other hand, Le Bon’s crowd contagion and Durkheim’s collective effervescence both involve the rapid activation of attunements, and sometimes rapid development of new attunements seen in new behaviors, simultaneously deployed through conspicuous behavior. The behavior is often conspicuous not only because it may be raucous and exaggerated, but because it is highly distinctive, in that it is a type of behavior that would not be seen on other occasions.53 In our terms, Durkheim’s collective effervescence consists in highly cohering and distinctive attunements being activated synchronously within a group that shares a strong bond of common identity.
The effervescence serves to reaffirm that identity and provide a potentially transformative experience by providing participants with a situation in which they can develop and experience attunements that they otherwise would not.
Synchrony of action is central to our understanding of Durkheim’s collective effervescence.
In the case of chanting the same chant, clapping and marching to the same beat, synchrony involves a pattern of repeated simultaneity. But we should note that synchrony does not simply mean simultaneity, and in other cases it is more complex. Call and response patterns involve careful timing, as do crowd reactions to a central performer. In sports, synchrony is seen in everything from the careful timing of a single throw and catch combination to the extraordinary synchronization of a larger play in a team sport. We might say that in all these cases, a combination of strong joint identity, high synchronization, and great distinctiveness of behavior leads to something akin to the collective effervescence that Durkheim described.
Crowd contagion is naturally understood in these terms. A situation in which attunements of the group are immediately manifest is one in which those attunements can spread rapidly. If the group shares a relevant identity, sharing attunements will produce consonance, leading to powerful feedback loops in which any attunement that is shared by a portion of the group can become rapidly shared by larger portions to whom the attunement is manifest. For some attunements, like basic emotions of panic and excitement, the shared identity of cohumanity may suffice to support rapid spread and mutual reinforcement.
Although for some purposes merely having the common identity of being human, or perhaps “being there,” can trigger contagion effects, as when an arbitrary crowd gathers around a spectacle, for some collective attunements, narrower identities are relevant. These might include collective attunements that are peculiar to the specific environment, such as attunements to playing sport for a sports team, or attunements to familial habits of eating and arguing and celebrating in family settings, or attunements to a specific vocabulary of protest peculiar to a rally of a particular political stripe. A crowd will have a stronger tendency to contagion when they share identity. But here the feedback loops become more complex.
In the case of a rally or protest, part of the logic of the event is that development of the identity goes hand in hand with synchronous display of attunement. The more people jump up and down or shout the same slogans or songs in time with each other, the stronger the common bond of joint identity becomes, and the stronger the consonance created by further synchronous action.
In a crowd, and especially in ritual settings, feedback loops encourage the formation of collective emotions and behaviors that would not result from individual attunements of group members in other circumstances. In other words, in these settings, some aspects of individual agency are lost or transformed. This is an idea that is quite clear in the work of Tarde, Le Bon, and Durkheim. The idea has resurfaced in various overlapping strands of contemporary psychological work, notably in the concept of diffusion of responsibility, seen in John Darley and Bibb Latané’s explanations of the bystander effect (whereby people are less likely to take responsibility to act in an emergency situation the more other people are present), and in the concept of identity fusion introduced by William Swann and Michael Buhrmester.
54 As Herbert Kelman and Lee Hamilton argue, diffusion of responsibility played a central role in the perpetration of Nazi war crimes: The Nazi extermination program was carried out by a vast bureaucracy in which many functionaries—from Adolf Eichmann down to junior clerks—sat at desks, shuffled papers, arranged train schedules, and carried out a variety of other tasks without having to consider the final product of their efforts. The perception of personal causation was reduced not only by the dissociation of each functionary’s contributory acts from the human consequences of those acts but also by the diffusion of responsibility within the bureaucracy. The more people are involved in an action, the less likelihood that any one of them will see herself or himself as a causal agent with moral responsibility.
55 The bystander effect does not depend on people having a common social identity; the original work by Darley and Latané in fact suggested that people’s tendency to dissociate themselves from moral responsibility to act when others are present was not highly correlated with at least one type of identity, namely gender identity. Rather, the bystander effect can be seen as a passive failure to take significant individual action when in any group situation. It is not so much groupthink as group avoidance of potentially dissonance-yielding action.
If diffusion of responsibility depends little on social identity, identity fusion goes in the other direction: it can be thought of as particularly strong type of identity-based psychological bond between people in a group, although in some ways identity fusion is also somewhat independent of what are often thought of as social identities, for example, those based on gender, age, occupation, or sexual orientation. Central to social identity are aspects of self-image and self-presentation that serve as exemplars of some group. For example, someone for whom being American is an important social identity might see their predilection for burgers as natural. However, that same person might also see their enjoyment of French New Wave cinema as idiosyncratic. Or if one has a Catholic social identity, one might find oneself to be in some ways a bad Catholic, without thinking there is anything bad about Catholicism.
So, one’s personal identity is distinct from one’s social identities: the social identities inform the personal identity, but are not confused with it.
Identity fusion, on the other hand, involves seeing oneself and a group as not entirely separate. Someone whose identity is strongly fused with the United States would see a harm done to them as a harm done to the United States (i.e., as anti-American), and, vice versa, would take a harm to the United States as being very much like a personal insult. Further, it is quite possible for identity fusion to take place with a group with whom one feels no strong social identification, as when a group coalesces for some arbitrary reason not connected with social identity, and the primary thing shared is the experience itself. Based on extensive studies of people’s behavioral predilections and how those predilections relate to their degree of identity fusion, it has been suggested by Bill Swann, Michael Buhrmester, and other researchers that identity fusion is a far more powerful motivator for actions on behalf of a group than is social identity (see fn. 54 on p. 163 in this section for references). Someone who merely feels that they fit in well with a national ethos is less likely to shoot or take a bullet for their nation than is someone who doesn’t clearly distinguish themselves and their country.
The potential of the concept of identity fusion to explain examples of apparently “selfless” action makes the concept important for our project, and for any work on political persuasion.
We do not claim to offer a new analysis of identity fusion, but we do see such processes of submergence of the self as being naturally described within the framework of collective harmony that we have laid out in this section. If identity fusion is a possible end point of harmonization with a group, then at least part of what must have happened when such a point is reached is that one’s idiosyncratic individual attunements become deactivated, relative to strongly activated collective attunements of the group. As we have seen, at least in some settings, like those at large political rallies, the consonance gained by people jointly activating attunements can lead to strong positive feedback loops.
Under these conditions, if attunements and attention are pushed in a certain direction, collective harmonization may overwhelm individual preferences. And the thing about large political rallies is that they are carefully oriented to give just such a push. Attendees are offered hats to wear and signs to wave, insurgents are planted in the crowd to act enthusiastically in line with the behavior that is desired, the stage is set up as a focus of attention, perhaps with huge Big Brother screens to bolster the effect of amplification and spotlights. And most of all, after a string of somewhat mediocre henchmen speak, the great leader emerges. The leader is a peculiar and extreme person, at once dominant and full of machismo, and at the same time almost a caricature of himself, so odd that one cannot take one’s eyes off him. It is no accident that the great charismatic leaders are often such strange people. As we commented in our brief discussion of spiritual cult leaders, above, this strangeness is a feature, not a bug. For it is one way of being sure that they will be magnets for attention. And attention, when it is joint among thousands of people, becomes a driver of collective harmonization, leading to a condition of collective effervescence in which the active attunements of any individual are the active attunements of the group. It is a scary situation to behold as an outsider. For what one sees if one is sufficiently far removed, perhaps watching a decades-old newsreel, is thousands of spellbound people chanting and moving together like puppets, as if the demagogue has somehow stripped them of their free will.
Let us consider here the role of the ideas we introduced in sections 3.2 and 3.3, fragmentation, cognitive dissonance, and nondeliberative uptake.
That the processes we have described in this section are largely nondeliberative is fairly clear from the literature on the topic: it is generally agreed that emotional contagion is nondeliberative, and there is a similar case to be made for many of the other processes underlying collective harmonization within a crowd environment.
As regards cognitive dissonance theory, recall Festinger’s weakest-link hypothesis, which suggested an explanation of cases where people’s emotional and dispositional attunements appear to drive their epistemic attunements. According to this principle, people have a tendency to do whatever is easiest to avoid dissonance. Thus, people will tend to suppress whichever attunement is most conveniently suppressed in order that they do not feel conflicted. Suppose that a participant at a political rally has a cognitive attunement that runs counter to the sentiment expressed on the stage before them, say a belief in the equality and goodness of all set against the speaker’s blatantly racist rhetoric. Someone who focuses on such a righteous belief will face extreme dissonance.
Which will be easier: ceasing to applaud and scream with everyone else, or ceasing to focus on the belief? By the simple expedient of attending to the rally and not attending to the belief, the immediate feeling of cognitive dissonance is avoided. The attendee has not necessarily ceased to hold the righteous beliefs, but they have compartmentalized them; that is, they have fragmented their mental state in such a way that the conflicting attunements are not simultaneously activated. This allows their attention to stray away from what, from our external moral perch, they ought to have focused on, a belief in what is good, and hence a conclusion that the activities around them are not. Fragmentation is what makes this move possible, for absent the possibility of fragmentation, which consists only in dissociating the belief from actions that conflict with it, the attendee would have faced a more blatant choice with a high personal cost: stop participating in the rally, or stop believing in equality and goodness of all.
This defensive strategy of compartmentalizing is complementary to a separate effect of propaganda that we have discussed, for example, in the context of the state declarations prior to the US Civil War (outset of chapter 1). Propaganda may lead to a reframing of what equality consists in, so that cognitive attunements are adjusted; such basic notions as equality and freedom can be subverted so as not to be discordant with the racist overtones, or overt racism, of right-wing rhetoric and action. We suppose that whereas changes of attention that suppress a strongly held cognitive attunement so that its relevance is not apparent may happen in the blink of an eye, transformations of such attunements are commonly slower. This is surely a process that happens more at the time scale of the cultural diffusion processes discussed by Tarde, not in the context of crowd dynamics discussed by Le Bon.
3.9. What Resonates and Why In chapter 1, we argued for the development of a theory of meaning based on resonance rather than content. After summarizing prior literature on resonance in communication, we started to develop our own account. That account went in a different direction from most prior literature in the following way. Prior scholars have typically been interested in what makes something resonant, or when something resonates with an audience, for example, what makes a framing of a political goal resonant for a target audience within a movement or for potential recruits. We have not defined what makes an action resonant. Rather, we defined what the resonances of an action are, using that as the basis of a theory of meaning intended to be rich enough to model broadly political and sociolinguistic phenomena. In following this path, we left aside the question of what makes something resonate with someone. We have outlined a theory of meaning but said nothing about what makes meanings special. If you like, our theory of meaning says nothing about what makes something feel deeply meaningful.
Although not normally put in such vaguely spiritual terms, that is an issue of central import in analysis of political messaging. The framing of a political debate is not arbitrary, but shaped by the question of what makes meanings special to people, that is, the question of which frames resonate with which audiences in which contexts.
There is a reason why we could not state back in chapter 1 what makes something resonant, what makes its meaning special. There we presented a highly idealized model of the resonances of individual signals within an interaction that made no reference at all to the broader culture of the agents involved in the interaction. However, prior literature makes it clear that the question of what makes something resonant is deeply entwined with the cultural context of those involved. For example, Deva Woodly argues for a theory of resonance that depends crucially on “common sense” cultural background. As she puts it, Resonant arguments, and the frames that they combine into, are able to influence people’s political understandings and social imaginations more forcefully than other kinds of information and evidence because they inhabit a special discursive space in which background notions, common logics, and new ideas are aligned in a harmonious way.
56 We agree. In our terms, the notion of what it is for something to resonate with an audience cannot be analyzed at the level of the associations of the individual signal, and not even at the level of what it is for an individual to be attuned to that signal, but rather must be analyzed in terms of how the associations of the signal relate to broader systems of attunement of that audience, for it is only at that level that the question of whether attunements are “aligned in a harmonious way” arises. Adapting the intuition in the quote from Woodly to our own framework, we arrive at the following definition of the extent to which something is resonant, or equivalently, how much it resonates: Degree to which something resonates: Something resonates (positively) for a group or individual to the extent that it induces increased (positive) harmony for them.
The above definition of the extent to which something is resonant is not limited to the question of when frames resonate, but applies to arbitrary events or objects. Seeing someone give away their last gummy bear might resonate, as a type of narrative resonance, if you will. A picture of a rainbow might resonate. And so might a peculiar man with a small mustache. Nonetheless, we will focus in this section on the question of when frames resonate.
To take a standard example from the framing literature, much political discourse seeks to explicate issues in terms of civil rights. Indeed, in political science literature, “civil rights” is often used as a label for what Benford and Snow term a “master frame,” because it is not issue specific, but rather is generalizable to multiple areas.57 As Benford explains, Typically, once a social movement fashions and espouses a highly resonant frame that is broad in interpretive scope, other social movements within a cycle of protest will modify that frame and apply it to their own cause. For example, once the US civil rights movement of the 1950s and 1960s experienced a number of successes based on an equal rights and opportunities frame, several other movements, including the American Indian, women’s, gay and lesbian, Chicano/a, and Gray Panthers, adopted and proffered a similar frame to their specific movement campaigns.58 Civil rights framings are resonant for those people for whom those rights form central organizing principles in their ideology. For these people, a framing that casts an issue in terms of civil rights might produce a sense of individual harmony if it connects together disparate attunements into a way that makes their system of attunements more coherent. To take the example of same-sex marriage in the United States, someone who feels dissonance at the idea of an institutional change, and thus specifically dissonance at the idea of legal and civil change to recognition of same-sex marriages, may be able to overcome that dissonance if they can come to see same-sex marriage not as a divergence from historical precedent and prevailing ideology, but rather as an expression of the central features of the ideology that shaped their modern nation. Looked at this way, it is the past in which rights were distributed unevenly that is anomalous, and a future in which a wider range of people have access to those rights is consonant.
What makes a framing resonant is often not just its ability to tap into a wealth of prior attunements, but to do so effortlessly, so that it is processed with little conscious deliberation. Sticking with same-sex marriage and related LGBTQ+ issues, a good example is the use of the phrase “Love is love,” which at the time of writing has been spread widely around the United States on “We believe …” yard signs proclaiming a range of values perceived to be under threat following the election of Donald Trump in 2016. The tautology “Love is love” might seem provocatively asinine, but that is part of the point. Who could disagree? Or, to tie into our discussion of deliberativity in this chapter, how much does one really need to think before agreeing?
The resonance and hence power of the adage “Love is love” derives from the fact that it allows LGBTQ+ issues, and same-sex marriage in particular, to be understood in terms of collective attunements that have become central to contemporary US ideology. The first collective attunement, discussed at length in Woodly’s analysis of the same-sex marriage movement, is the idea of romantic love as the central component of a good marriage, an idea that is not globally recognized at present, and was not dominant in the United States until relatively recently (within the last century). Part of the power of the adage is that some of those who are opposed to same-sex marriage are among those most likely to subscribe not merely to the view that love is central to marriage, but that romantic love has no place outside of marriage, or at least of courting with marriage as a goal. The second collective attunement is the idea that romantic love does not answer to any higher principle, but is entirely and inherently a matter of personal choice. It is relevant here that romantic love has since ancient times been seen as inherently transgressive. In literature and mythology around the world, romantic love is often in tension with other prerogatives.
For this reason, the mere fact that someone might find same-sex union to be in tension with their ideology does not necessarily mean that they cannot recognize same-sex love as being an expression of “true” love. These two collective attunements then tap into the civil-rights frame. If marriage is the idealized state for those in a state of love, and love is entirely a matter of personal choice, it follows from attunements to personal freedom that marriage among those of an age to make such choices should be unfettered by further institutional restrictions. The simple slogan “Love is love” is resonant in large part because it connects unobjectionably to collective attunements that are core parts of the ideology of a wide range of people at whom the message is targeted. We have defined what it is for something to resonate positively in terms of harmony and its attendant feeling. However, the point of a framing in politics, say a civil-rights framing of same-sex marriage issues, is not just to make individual people feel good, if it even achieves that. The point is to persuade people, to cause them to change attitudes and behaviors to which they were resistant. It is central to the mechanism of persuasion that by tying an issue, say same-sex marriage, to broader ideology, to what Woodly refers to as “common sense” and Jaeggi refers to as “forms of life,” the crafters of political messages create the potential for dissonance. The process is not complex, but let us spell it out.
Suppose there is some goal idea (or behavior, or emotional attitude) to which a person targeted by a messaging campaign is not attuned. Now suppose that this targeted individual shares other collective attunements, for example, accepting principles that they and most people around them see as common sense or obvious, and that the message crafter finds a framing that establishes a way to link the goal idea to those collective attunements. If the target recognizes that the goal idea is natural in the light of the connection to those collective attunements, for example, because the goal idea can be seen as an expression of shared principles, then they will feel a tension.
Harmonizing by accepting the goal idea can then resolve this tension. That process is straightforward enough, basically similar in structure to the way any argument works to convince someone of something that they have not previously embraced, except that it applies to arbitrary attunements and not merely to belief in propositions. But we can now go further, to the group level.
By adopting the goal idea, or other intended attunement, a targeted individual will have attunements that are in alignment with members of the political movement for which the message crafter is advocating. They can thus derive harmony not only through consonance of attunement of the goal idea with their broader ideology, but also through consonance of their attunement with attunements of others. At the same time, they may feel dissonance with others who do not share attunement to the goal idea.
Shifting someone’s mind on an issue may therefore be part of a process of peeling them off from one group and aligning them with another. This has strategic implications. It might be that the best political strategy is not always to hit an issue head on, where it will create immediate dissonance if recognizably associated with a group with which the targeted individual does not feel social alignment, perhaps to the extent of preventing effective communication. It might be better to start with less obvious behaviors and issues that can indirectly incentivize. For example, start by getting them into church, and only then work on aligning their faith with the rest of the congregation. Or bus them to the rally, and then get them shouting and jumping and wearing the same hats and T-shirts. Or, if you want people to align with attitudes or behaviors of a group that stereotypically drinks a lot of lattes, you might find it easiest to start by giving away some free latte vouchers, and only move onto civil rights and “Love is love” framings later.
Here it is significant that something can be resonant for a group first, and for individuals only secondarily. A chant, a slogan on a hat, or a march along a certain route might resonate with a group in part because the group finds it easy to become behaviorally aligned while chanting, while wearing the hat, or while marching. That might lead secondarily to resonance for the individuals, and to identification with the crowd. At least sometimes, it is only thereafter, in a very indirect process, that the meaning of the chant or slogan, or the significance of the route of the march, or what was witnessed along the route, itself take on special significance for the participants. It is not just that the behavioral tail can wag the cognitive dog, but that what resonates with a pack of dogs can drive the individual dogs secondarily, through processes of collective effervescence that we have described.
To our knowledge, the most theoretically sophisticated extant analysis of frame resonance other than Woodly’s is that of Terence E. McDonnell, Christopher A. Bail, and Iddo Tavory, who take their philosophical inspiration less from Aristotelian rhetoric, as she does, and more from the pragmatist tradition of Dewey, James, and Peirce.59 McDonnell et al. agree with the bulk of prior work that embedding a frame in the cultural milieu is crucial for frame resonance; they take this factor to have been overemphasized and to have encouraged a view of resonance as a fixed property of a framing or other cultural object, independent of who is using the frame or being exposed to it, and independent of the particular context in which the frame is used. What they take to have been underemphasized, drawing on the pragmatist tradition, is what the frame does for people in particular contexts, the functional load it bears in helping people work through problems experienced in particular ways at particular times, and in helping motivate action appropriate to those contexts. As they say, “Cultural objects are not relevant unless employed to solve a problem,” so we should see “resonance as an experience emerging when affective and cognitive work provides actors with novel ways to puzzle out, or ‘solve,’ practical situations.” They go on (making explicit their debt to Peirce), Resonance is a specific kind of experiential effect (or interpretant), emerging at the same time that actors come to see the world in a new light.… It is thus only through the effect of signs that meaningmaking is completed, and such effect cannot be encapsulated by an analysis of cultural objects but must also take into account the habits of thought and action through which an interpreter experiences such an object.60 Resonance, on this view, is not a fixed property, but something that occurs as people “puzzle out” (a phrase they draw from Peirce) a problem, come to see aspects of the problem in a new light, and imbue aspects of the problem situation with new meaning. Furthermore, this does not happen only at an individual level: Instead of focusing on patterns of interaction among individuals, resonance can also be studied via analysis of interaction among groups, organizations, or other collective actors within broader social arenas such as the public sphere.… We can see resonance occur on a macro level as different organizational actors come together and converge on a cultural object—as in the unification of Gay and Lesbian social movement organizations around “queer” identities.61 As we have made clear, we view resonance as being usefully analyzed at a collective rather than individual level. We also agree with McDonnell et al. as regards resonance being more than just cultural embedding, and having a problem-solving character. This is reflected in our definition. Our definition does not refer to a fixed state of a frame or other cultural object, but rather to its ability to induce harmonization.
Let us make the relationship to problem-solving explicit. A problem is something that causes dissonance, a clash between a desire to change or understand things, and other attunements that mitigate against such change or sense-making. On our definition, a framing will be resonant for an individual or group facing such a problem precisely when it provides a way to overcome such dissonance, for example, by reinterpreting the situation in such a way as to produce consonance between attunements.
We have not discussed what psychological processes determine the intensity of feelings of dissonance or consonance, and we do not claim to have the relevant psychological expertise, but it is consistent with the account we have proposed (i) that harmonization, especially at the group level, need not be an instantaneous process, but can occur gradually and as the result of a string of interactions, and (ii) that the fact of there being an extended process that people engage in is a contributor to the strength of consonance experienced (or indeed to the build-up of dissonance in case no resolution to the problem is found). This latter point, as regards the importance of process, relates to the broader intuition that we gain more satisfaction from solving problems that we have attended to for a while than from those that we solved more or less instantaneously, since in the latter case we perhaps didn’t see them as “real” problems at all. This idea fits in naturally with the musical metaphor: sophisticated composers do not simply provide consonance at all times, but rather exploit discord, resolving tension only once the dissonance has become manifest and perhaps even unnerving. A feeling of harmony is not a long-term state, but a reaction to change.62 The examples we have given above are largely concerned with political framings and with what is commonly called the cultural resonance of those frames. Even though they take embedding in preexisting culture to have been overemphasized, McDonnell et al. see themselves as contributing to the development of a theory of cultural resonance. Yet in all the cases we have discussed, and in accord with McDonnell et al.’s account, resonance might result not only because of broad properties of a culture, but from individual peculiarities of an experiencer. Something is personally resonant with someone when it induces harmony by manifesting consonance of attunements that are distinctive to that individual. By contrast, cultural resonance, the greater concern of this book, and of the prior literature on political framing, is harmony resulting from consonance with the distinctive attunements of an ideology, the distinctive collective attunements of a community of practice. 3.10. Conclusion In this chapter, we have developed an account of harmonization at both the individual and group levels. It is designed to help describe both political processes at different scales, a range that includes personal interactions, social media, and mass rallies. The resulting framework has properties that are quite unlike those found in more standard content-delivery models of communication. To end, let us consider one by one what we presented as the central relevant features of the content-delivery model, and the ways in which our analysis of harmonization changes the picture.
1. Neutrality of integration: In the content-delivery model, the form of the message is just packaging. But harmonization is strongly affected by form: people may become attuned to forms of collective activity before they fully develop matching cognitive attunements. A political rally, to state the obvious, is more about form than content.
2. Short-termism: The content-delivery model has led to a focus on comprehension and integration of a single utterance. Harmonization can be both faster and slower. There are important communicative effects that are faster than deliberative integration (or non-integration) of a message, but there are also much longer-term processes, whereby consonance or dissonance may develop slowly, and whereby the effects of multiple concordant communicative interactions can slowly drive re-attunement and fragmentation across large subsystems of attunements.
3. Passive transmission: Whereas the content-delivery model centers on a deliberative process of intention recognition, we have emphasized the nondeliberative character of many communicative processes. In particular, people do not have full control over how they will react to hateful speech, to the excitement of a group, or over what political messages will resonate with them.
4. Propositionality: We have emphasized the central role that change in dispositional and emotional attunements plays in communication. The power of words rests in large part on their ability to drive emotion and action. As we have seen, the power to shape beliefs may follow from that power, because harmonization does not give priority to cognitive attunements. Our Festinger-influenced model allows that rational deliberation over propositionally represented evidence may sometimes play second fiddle.
5. Deliberativity: We do not deny the existence of processes of belief revision as deliberative reasoning, but inspired by Gilbert’s research, as well as by classic work on collective effervescence and group contagion, we have emphasized that both many short-term effects of communication, and many longer-term effects of harmonization, are largely nondeliberative.
6. Noncompulsion: The nondeliberativity of many aspects of harmonization implies that people can be affected by communication in ways they would not wish. The immediate resonances of hateful speech may include the driving of attention and felt pain. A mass rally may semiautomatically arouse duplication of activation patterns within attendees, and a compulsion to harmonize with the group.
To close this chapter, let us briefly consider an important example in which the concepts we have discussed are important, in particular the notion of dissonance. Festinger’s best-known work on dissonance concerned cults, but the term “cult” is negatively charged. Those inside a cult would not describe it as such. Applying the term “cult” to a group also suggests that a group is small enough that its ideology can clearly be distinguished from that of the mainstream. Nonetheless, there are times when a broad majority of people collectively exhibit behavior that fits well with the cults that Festinger described.
Let us briefly consider the way people think and behave with respect to the environment in a time of dramatic climate change. It turns out that being told regularly, in the language of science, that the climate is in a death spiral that will kill and displace millions or billions of people, has remarkably little effect. People of all stripes continue to vote, consume, exploit, and argue in much the same way as before. To many people, it is those worrying about climate change who appear to be in a cult-like group, the climatechange alarmists. After all, what is science if not a sort of cult?
It is more convenient to believe that these alarmists have a secret agenda, say destroying our way of life, or drawing attention to themselves, and thus cannot be trusted, than it is to believe that they are right. It is more convenient because accepting that the scientists are right would imply change in behavior on a massive scale, a radical reorganization of economic priorities at a global scale that would challenge core assumptions of the dominant consumption and growth-based ideology. What humans are now doing in continuing to act as before while the planet burns around us is structurally highly similar to the way cult members successfully bury their collective heads in the sand and refuse to accept what should be powerful evidence. The big difference, of course, is that a small cult behaves the way it does in part because of how it is seen by outsiders, and this factor cannot play the same role when the group being considered is not small, but makes up a sizeable portion of humanity. Yet the dynamics are the same. Change of behavior creates more short-term dissonance than would acceptance of fact, and so behavioral inertia wins out, and humanity plunges ever onward in the direction of catastrophe.
Does Festinger’s theory, or the account of harmony and resonance we have developed, teach us anything about how climate-change messaging should operate? There is no low-hanging fruit here, no easy one-step solution that we can offer, but we can make some observations.
First, if it is true that part of the problem of convincing people of the significance of climate change is that the need for them to change their own way of life produces dissonance, then it follows that messaging should not necessarily emphasize personal sacrifice except possibly when part of a heroic narrative. The message may often be that action on climate change is what is needed in order for people to maintain their lives, rather than that dealing with climate change begins with people disrupting the patterns they are used to.
Considerations of dissonance suggest that calls to change individual patterns of consumption may therefore be counterproductive.
Second, work on framing in protest movements suggests that there often are resonant frames to be found even in the face of stubborn long-term problems, frames that can allow people to understand those problems in a new light. These frames do not come out of nowhere. A movement that remains stuck with a limited selection of frames that resonate with only a subset of people is unlikely to be successful in growing and winning over converts, as seen, for example, in Deva Woodly’s extensive discussion of US fair-wage movements, whose extensive local successes have not been duplicated at a national level.63 When people actively explore different framings, they sometimes discover new framings that work. It is not for us to say whether it would be more effective to use a social-justice framing, for example, a framing in terms of the rights of unborn children to life on a clean planet, or a framing in terms of the rights of the planet itself as an actor rather than as a physical system, or any other framing. It is a creative exercise to explore and develop new framings and narratives, and an empirical question what frames will work with what groups and when they should be deployed. What we take from the traditions this chapter builds on is that the development of ways of thinking and talking about climate change is of comparable import to the development of scientific understanding of the problem. For without a way to bring collective harmony on the issue to a large enough global community, the political will for change will remain lacking.
1. Humboldt, On Language, 57.
2. Clark and Wilkes-Gibbs, “Referring as a Collaborative Process,” 3.
3. See McNamara and Magliano, “Toward a Comprehensive Model of Comprehension,” for an insightful and integrative discussion of the reading-comprehension literature. We thank Eyal Sagi for discussion on this topic.
4. See, e.g., Hawthorne and Stanley, “Knowledge and Action,” for a development and defense of this view.
5. This is not to say that our account implies just this explanation of the hypothetical prison guard’s psychology, which should be seen as an illustration of the power of the approach rather than as a diagnosis. The account allows other explanations. For example, if the prison guard is able to understand the prisoners as a threat to the motherland (and hence even to his own children’s future), or, within his discriminatory ideology, as not worthy of decent treatment, then he may be able to reconcile his different dispositions at home and work.
Note that the definition of a discriminatory ideology merely implies that outgroups are “inherently deserving of less than equal treatment or resources,” but later in the volume, in section 10.3, we will introduce the notion of a genocidally antagonistic social group, a group whose ideology is such that the outgroup is viewed not merely as worthy of unequal treatment, but as deserving destruction.
6. See, for a first-rate recent discussion, Elga and Rayo, “Fragmentation and Logical Omniscience.” Also recommended, with a wide range of perspectives and much new development, is the excellent collection of essays in Borgoni, Kindermann, and Onofri, The Fragmented Mind.
7. The use of fragmentation in classical approaches to belief offers a line of solution for the problem of contradiction and reopens problems for which standard models had solutions.
That is, the solution creates a research program, that of determining how systems of propositions can be fragmented, how an agent with fragmented beliefs should modify their fragmented belief systems when faced with new information, and how agents with fragmented beliefs should reason. Our adaptation of fragmentation to systems of attunement can be seen as suggesting one direction that this research program might take.
8. The primary reference for the theory of cognitive dissonance remains Festinger, A Theory of Cognitive Dissonance. The other classic reference is the extended case study in Festinger, Riecken, and Schachter, When Prophecy Fails. Although the heyday of cognitive dissonance theory within psychology has long since passed, it continues to be influential.
Not only is the concept firmly entrenched in everyday folk psychology, but it continues to be used in scholarly work. For a recent overview, see Harmon-Jones, Cognitive Dissonance. A good starting point is the introductory chapter of the volume by Eddie Harmon-Jones and Judson Mills (“An Introduction to Cognitive Dissonance Theory and an Overview of Current Perspectives on the Theory”).
9. Festinger, A Theory of Cognitive Dissonance, 28.
10. Hume, The Natural History of Religion.
11. Gilbert, “How Mental Systems Believe.” Gilbert’s paper has been influential in both psychology and philosophy. Our discussion draws on Gilbert and on work by philosophers Andy Egan (“Seeing and Believing”) and Eric Mandelbaum, whose dissertation, “The Architecture of Belief,” recapitulates and extends Gilbert’s argument in crucial ways.
The Cartesian view appears centrally in Descartes’s exposition of the method of doubt. The sixth of his “Principles of Human Knowledge” reads (in part), “We can refrain from admitting to a place in our belief aught that is not manifestly certain and undoubted, and thus guard against ever being deceived” (Descartes, Selections from the Principles of Philosophy, 36–37).
12. In the relevant passages of Spinoza’s Ethics, he explicitly opposes Descartes’s view, especially with regard to Descartes’s explanation of why people sometimes perform errors in judgment. Spinoza’s positive view is famously difficult to disentangle. However, we can see the inspiration for Gilbert’s reading from the fact that Spinoza (i) says one way we form ideas is by comprehending utterances (“from the fact of having read or heard certain words we remember things and form certain ideas concerning them,” notes to Proposition XL, Book II, The Chief Works of Benedict de Spinoza, 113), and (ii) says that someone who has an idea, or at least a true idea, can’t help but believe it (“He, who has a true idea, simultaneously knows that he has a true idea, and cannot doubt of the truth of the thing perceived,” Proposition XLIII, Book II, The Chief Works of Benedict de Spinoza, 114). Taken together, these imply that when you comprehend an utterance, you believe it—at least when the utterance is itself true. Jonathan Bennett, to whom Gilbert refers, provides a detailed and far more sophisticated argument for attributing something like Gilbert’s view to Spinoza—see Bennett, A Study of Spinoza’s “Ethics,” especially 159–62.
13. Thucydides, The Peloponnesian War, book 1, 12.
14. See Hasson, Simmons, and Todorov, “On the Possibility of Suspending Belief.” While Gilbert had shown that people under cognitive load have a tendency to process anything they comprehended as if it were true, Hasson and Todorov showed that the effect is only clear when the claim does not provide obviously usable information.
15. Cavell, The Claim of Reason, 34.
16. Gibson, “A Theory of Direct Visual Perception.” 17. We don’t wish to argue our case for what is fundamentally a question of empirical psychology on the basis of seventeenth-century philosophy, but we should be remiss not to note that it is hardly surprising that something like Gilbert’s Spinoza-influenced model of nonreflective uptake should be plausible in the case of emotion. Emotion was central to Spinoza’s project. His rejection of Cartesian dualism and his deterministic view of human nature implied that (bodily) emotional reactions and the functioning of reason should be of a piece. We should also note that as regards emotion, Descartes certainly did not argue for emotional reaction being essentially the result of reflection, so although there is a rift between the views of Spinoza and Descartes on emotion, it is not of relevance here.
18. Some might prefer to see the mollifying effect of an apology as a perlocutionary effect, in J.
L. Austin’s sense. We do not reject the concept of perlocutionary effects, but we do reject the possibility of drawing a line between illocutionary force and perlocutionary effects. In our view, the extent to which an effect is a conventional resonance of an action is linked to the regularity with which the effect is consistently associated with the action. That makes it an empirical question to what extent a mollifying effect is conventionally associated with apologizing. Mollifying is a resonance of apologizing just to the extent that apologizing tends to mollify. Likewise, making an interlocutor feel bad is a conventionalized resonance of the practice of insulting people to just the extent that insulted people feel bad.
19. While the various dual-process models that have been proposed are not without their detractors, it would be difficult to overstate the ubiquity of such models in modern psychology, especially cognitive and social psychology. A commitment we are carefully avoiding is to what is sometimes known as a strong dual-process model, which would imply a clear distinction of fast and automatic processes from slow and deliberative processes, perhaps even a neurophysiological modularization. We take the issues of how distinct the processes are, and how they are implemented in the brain, to be empirical, and clearly beyond the scope of this book. The narrower idea we are suggesting—that Gilbert’s model be seen in terms of dual-process theory—is not original, but it is found in a classic analysis of people’s tendency toward acquiescence, i.e., people’s bias to respond to information questions affirmatively: see Knowles and Condon, “Why People Say ‘Yes.’ ” For an overview of dual-process theory in cognitive psychology, see, e.g., Evans, “In Two Minds: DualProcess Accounts of Reasoning.” For social psychology, see Gawronski and Creighton, Dual Process Theories. The clearest nontechnical introduction to dual-process theory is Kahneman, Thinking, Fast and Slow. A defense of some controversies surrounding dual-process models is Evans and Stanovich, “Dual-Process Theories of Higher Cognition.” For a philosophical introduction to and analysis of dual-process theory, see Samuels, “Dual-Process Theory as a Theory of Cognitive Kinds.” Further extended discussion of dual-process theories in a philosophical context is found in literature on implicit bias. Since implicit bias is by definition not easily accessible to processes of rational deliberation, the very idea of implicit bias is suggestive of a dual-process view. For extensive discussion of the dual-process view and alternatives to such a view, in the context of implicit bias, see the introduction and many excellent chapters in the first volume of a collection edited by Michael Brownstein and Jennifer Saul that kick-started interest in implicit bias in philosophy: Brownstein and Saul, Metaphysics and Epistemology.
20. Although Festinger does not talk of harmonization in this way, he does allow that people can sustain a degree of cognitive dissonance. Indeed, it is an important theme of his work that only massive dissonance can drive a radical change in beliefs or behaviors.
21. To be clear, it’s not that we are claiming the relevant attunements could not possibly be rendered in propositional form, and logical methods used to analyze coherence. Standard logical methods are just not a great fit to the problem of analyzing coherence, e.g., because we take it to be a scalar notion.
22. See, e.g., Seemann, Joint Attention; Watzl, “The Philosophical Significance of Attention”; and Eilan et al., Joint Attention.
23. Ajzen and Fishbein, Understanding Attitudes and Predicting Social Behavior. A helpful overview of framing theory that discusses the expectancy value theory model is Chong and Druckman, “Framing Theory.” 24. Linderholm et al., “Fluctuations in the Availability of Information during Reading.” We thank Eyal Sagi for discussion of the reading-comprehension literature. Models in this literature and associated literature on discourse analysis are more thoroughly and computationally articulated than is our own proposal, and a natural line of development of the harmony model would use this prior work as a starting point.
25. O’Brien and Myers, “Text Comprehension,” as discussed also in section 1.3, fn. 34.
26. Our talk of a mental landscape rather obviously recalls the notion of “mental space” developed by Giles Fauconnier. There are strong similarities between the two. In particular, he targets types of reasoning, for example, reasoning involving analogical mappings, that are not well accounted for in standard logics. The classic reference is Fauconnier, Mental Spaces. Another body of work that puts a spatial metaphor to work in analyzing conceptual structure is due to Peter Gärdenfors, e.g., The Geometry of Meaning.
27. Alternatively, we could use a function from attunements to a triple of a level, an activation, and a location in an abstract mental space, from which distance could be calculated. This would then bring up a question of what the geometry of mental space is like, how many dimensions it needs, and what axioms it follows. There is no special reason to assume a Euclidean two- or three-dimensional space, except that it is so conceptually familiar as to be the obvious starting point for work on this sort of model.
28. Freud, “The Dissection of the Psychical Personality,” 57.
29. Using a standard technical approach, one might take the distance between two attunements to be the negative log of the probability that one will activate the other, so that a distance of zero implies that when one is activated, the other is always activated, and that as the distance tends to infinity, the probability of simultaneous activation tends to zero.
However, using formal definitions of this sort would be overstating the power of the landscape model, which must remain metaphorical, at best a useful source of intuitions. For it is far from clear that activation tendencies of attunements can in fact be modeled using a Euclidean space. To see why, suppose that attunement A has a high probability of being activated when exactly one of attunement B and attunement C are active, but that attunements B and C have a significant probability of coactivation. There is no way of representing these dependencies by placing A, B, and C in a spatial arrangement, and predicting coactivation solely on the basis of probabilities defined as described. Thus it might be better to think of the landscape metaphor not as directly representing the probability of coactivation between attunements, but rather as representing one source of constraints on coactivation, and allowing that there might be additional factors determining how attunements interact.
30. See Browning, Ordinary Men.
31. We are thinking of two controversial lines of work. The first is one of the most influential, and highly criticized psychological studies ever run, the Stanford Prison Experiment: Haney, Banks, and Zimbardo, “Interpersonal Dynamics in a Simulated Prison.” Leaving aside the deep issues of scientific ethics that the experiment immediately raised, we should note that the specific conclusions usually drawn, to do with conformation to social roles, have been questioned—see, e.g., Le Texier, “Debunking the Stanford Prison Experiment.” Further, the study did not have a large sample size, and it is unclear to what extent the behavior of the participants reflected only their own understandings of the social roles that they were given (prisoner or guard), rather than the expectations of the experimenters. The second line of work involves Stanley Milgram’s equally terrifying and famous experiments showing that people will obey authority figures who ask them to perform acts that harm others. The general conclusion that somewhat randomly chosen people can easily be manipulated into acting oppressively is by now well demonstrated.
32. Balkin, Cultural Software, 189.
33. See Eduardo Medina, “Alabama Pastor Is Arrested while Watering Neighbor’s Flowers, Video Shows,” New York Times, August 31, 2022, https://www.nytimes.com/2022/08/31/us/black-alabamapastor-arrested-flowers.html.
34. Arthur C. Danto argues that histories must incorporate what he calls “narrative sentences,” ones that encode, informationally, a future perspective on past events (Narration and Knowledge, chapter 8, 143–81). We suggest a stronger thesis, that narratives cannot be reduced to narrative sentences.
35. Fraser, “Narrative Testimony.” She summarizes her approach as follows: “Just as the Stalnakerian model of conversation takes simple testimony to be a conversational technology aimed at opinional co-ordination, I take narrative testimony to be a conversational technology which aims at perspectival co-ordination” (4028). There is much overlap between our approach and hers. If it is allowed that perspective can be cashed out in terms of attunements, as we do in section 6.4, then narrative harmonization is akin to her notion of perspectival coordination via narrative testimony. 36. To say that we are not the first to emphasize differences between narrative coherence and logical consistency would be to make a massive understatement. Plato made the contrast between rhetoric and rational argumentation in a passage of the Gorgias that is famously dismissive of both sophists and rhetors: “as cosmetics is to gymnastics, so is sophistry to legislation, and as cookery is to medicine, so is rhetoric to justice,” a passage that culminates in the aphorism that “[rhetoric does] in the soul what cookery does in the body” (Plato, Gorgias, 33–34). From Aristotle on, logic and rhetoric have been seen as complementary, and the contrast between them has arguably been a pillar of Western thought. A more recent line of work begins with Walter Fisher’s Human Communication as Narration, where he develops a model of “Narrative Rationality” in which narrative forms are seen as a source of “good reasons” in and of themselves, and not merely as aesthetically pleasing or emotionally laden persuasive techniques that must play intellectual second fiddle to logical argumentation. For critical and scholarly discussion, see Stroud “Narrative Rationality.” 37. Dan Patrick, @DanPatrick, Twitter, February 15, 2022, accessed March 1, 2023, https://twitter.com/DanPatrick/status/1493694009600053250?ref_src=twsrc%5Etfw.
38. We thank Eyal Sagi for discussion.
39. Le Bon, The Crowd, 10.
40. Le Bon, The Crowd, 31.
41. Nye, The Origins of Crowd Psychology, 178.
42. Dawkins introduces the term meme in The Selfish Gene, describing it as “a unit of cultural transmission or a unit of imitation” (192, emphasis in original), and as a shortened form of “mimeme,” making the connection with imitation, and hence the similarity with Tarde’s work, particularly striking. However, Dawkins appears to have been unaware of his predecessor, whose work has enjoyed a renaissance only more recently. Tarde’s conception of cultural evolution is seen clearly in passages like this one: “In the beginning of societies, the art of chipping flint, of domesticating dogs, of making bows, and, later, of leavening bread, of working bronze, of extracting iron, etc., must have spread like a contagion; since every arrow, every flake, every morsel of bread, every thread of bronze, served both as model and copy. Nowadays the diffusion of all kinds of useful processes is brought about in the same way, except that our increasing density of population and our advance in civilisation prodigiously accelerate their diffusion, just as velocity of sound is proportionate to density of medium. Every social thing, that is to say, every invention or discovery, tends to expand in its social environment, an environment which itself, I might add, tends to selfexpansion, since it is essentially composed of like things, all of which have infinite ambitions” (Tarde, Laws of Imitation, 17).
To dig somewhat further back historically, Tarde observes that Pliny the Younger had used of contagion in this way, talking of the contagion of religious belief. The context is a letter to Emperor Trajan regarding the dangers posed by the spread of Christianity (as opposed to worshiping Trajan himself and the Roman pantheon). Pliny explains that “the contagion of this superstition has spread not merely through the free towns, but into the villages and farms.” He continues, on an optimistic note, “Still I think it can be halted and things set right” (“Letter to Trajan”).
43. Tarde, Laws of Imitation, 87. Tarde himself takes inspiration from Adam Smith in recognizing the importance of imitation.
44. Tirrell, “Toxic Speech: Toward an Epidemiology of Discursive Harm”; “Toxic Speech: Inoculations and Antidotes”; “Discursive Epidemiology.” 45. Durkheim, The Elementary Forms of Religious Life, 9. 46. Csikszentmihalyi, Flow. Csikszentmihalyi characterizes flow as an “optimal experience” in which an individual is willingly absorbed by whatever they are doing, i.e., their attention is undividedly focused on one activity. He also talks of flow, not just coincidentally, in terms of “inner harmony” (217, and elsewhere). So collective flow would involve an outer or collective harmony and the joint attention of a group, freely given and focused on a single group activity.
47. Orwell, Nineteen Eighty-Four, 133.
48. See Yudkin et al., “Prosocial Correlates of Transformative Experiences.” Note that the presence of hallucinogenic drugs at festivals was a predictive factor in the studies, but far from being the sole explanation of the effects observed. Schmitt’s more troublesome vision is expounded, for example, in The Concept of the Political.
49. We have been assuming that there is no sense in which an individual can have the same attunement twice, but for completeness let us note that it is logically possible to make sense of the idea of a single person having shared attunements, provided they are shared across distinct aspects of the individual’s life. A version of the landscape model described above might allow that different clusters of attunements both include some particular attunement, without mutual activation across the systems. Thus, the prison guard might have an attunement to vigilance when with their children, and an attunement to vigilance when with prisoners, and yet without vigilance in the first role activating other attunements associated with the second role. More generally, the sharedness of attunements across aspects of a person’s life would be a matter of the extent to which they have the same personality when playing different roles. This consideration then naturally suggests an analogy whereby we consider the question of the extent to which different individuals sharing attunements might come to have not merely a shared “identity” but also a shared personality, but we refrain from speculating further as to the analogies to be found between individual psychology and group psychology.
50. In The Theory of Moral Sentiments, 4–5, Adam Smith discusses empathic emotional response in the context of seeing someone on the rack: “By the imagination we place ourselves in his situation, we conceive ourselves enduring all the same torments, we enter as it were into his body, and become in some measure the same person with him, and thence form some idea of his sensations, and even feel something which, though weaker in degree, is not altogether unlike them. His agonies, when they are thus brought home to ourselves, when we have thus adopted and made them our own, begin at last to affect us, and we then tremble and shudder at the thought of what he feels. For as to be in pain or distress of any kind excites the most excessive sorrow, so to conceive or to imagine that we are in it, excites some degree of the same emotion, in proportion to the vivacity or dullness of the conception.” 51. Much of the large and diverse literature on emotional contagion focuses on the process by which one person is, as the disease metaphor would have it, infected by another. A somewhat standard definition of primitive emotional contagion is “the tendency to automatically mimic and synchronize facial expressions, vocalizations, postures, and movements with those of another person and, consequently, to converge emotionally” (Hatfield and Rapson, “Emotional Contagion,” 153–54).
Some more recent literature identifies relatively small but statistically significant effects in big data studies of emotional contagion in social media. A by now somewhat infamous study shows the effects in a study of Facebook users (problematic because the feeds of these users were manipulated without their consent): Kramer, Guillory, and Hancock, “Experimental Evidence of Massive-Scale Emotional Contagion.” Emotional contagion effects on the Twitter platform are discussed in Ferrara and Yang, “Measuring Emotional Contagion in Social Media.” 52. This is not to say that Tarde did not consider contagion at fine time scales. His discussion of the development of a “group mind” in a crowd hinged on the possibility of rapid imitation and repetition, and foreshadowed recent work in social psychology on identity fusion, briefly discussed below.
53. Both the distinctiveness and the synchrony of actions during episodes of collective effervescence is memorably described by Durkheim: On every side one sees nothing but violent gestures, cries, veritable howls, and deafening noises of every sort, which aid in intensifying still more the state of mind which they manifest. And since a collective sentiment cannot express itself collectively except on the condition of observing a certain order permitting co-operation and movements in unison, these gestures and cries naturally tend to become rhythmic and regular; hence come songs and dances. But in taking a more regular form, they lose nothing of their natural violence; a regulated tumult remains tumult. The human voice is not sufficient for the task; it is reinforced by means of artificial processes: boomerangs are beaten against each other; bull-roarers are whirled. It is probable that these instruments, the use of which is so general in the Australian religious ceremonies, are used primarily to express in a more adequate fashion the agitation felt. But while they express it, they also strengthen it. This effervescence often reaches such a point that it causes unheard-of actions. The passions released are of such an impetuosity that they can be restrained by nothing. (The Elementary Forms of Religious Life, 217–18) 54. Darley and Latané, “Bystander Intervention in Emergencies”; Swann et al., “Identity Fusion.” The theory of identity fusion is further developed in Swann et al., “When Group Membership Gets Personal.” See also Whitehouse and Lanman, “Ritual, Fusion, and Identification”; Swann and Buhrmester, “Identity Fusion.” For discussion of diffusion of responsibility in the bystander effect, we thank Eyal Sagi.
55. Kelman and Hamilton, Crimes of Obedience, 165.
56. Woodly, The Politics of Common Sense, 97.
57. Benford and Snow, “Framing Processes and Social Movements,” 618–19.
58. Benford, “Master Frame,” 366.
59. McDonnell, Bail, and Tavory, “A Theory of Resonance.” 60. McDonnell, Bail, and Tavory, “A Theory of Resonance,” 3.
61. McDonnell, Bail, and Tavory, “A Theory of Resonance,” 8.
62. Let us note a desideratum of a theory of resonance, as seen in McDonnell, Bail, and Tavory’s discussion: A pragmatist approach enabled us to distinguish between resonance and Durkheimian moments of “collective effervescence” [Durkheim, The Elementary Forms of Religious Life] or Collins’s [Interaction Ritual Chains]. Rather than the heightened emotions one feels when engaging in ritualized interactions with objects, an experience that draws attention to and reinforces group beliefs and commitments, resonance produces heightened emotions as people come to novel solutions. Resonance, then, is closer to Dewey’s (1934) notion of “having an experience” [Dewey, Art as Experience]—resulting from a process of “undergoing” that leads to consummation when an unexpected solution is found. (“A Theory of Resonance,” 9) The desideratum, then, is that a feeling of resonance be separated both from feelings of collective excitement, and from the sense of consonance with a community of practice experienced by those performing established behaviors ritualistically, i.e., in a highly standardized way. It should be noted here that our definitions of collective effervescence and resonance are related but distinct, since we define collective effervescence in part in terms of maintenance of joint attention, which does not play a direct role in our definition of degree of resonance. Although we do not analyze Collins’s model in the same depth, we discuss it further in chapter 6, where we consider the process of ideological transmission in terms of accommodation, in particular accommodation of practices. 63. Woodly, The Politics of Common Sense.
PART II Presupposition and Ideology
CHAPTER FOUR The Psychology of Presupposition In this world, while “race talk” is by no means unknown … racism is located as well, and very importantly, in an undercurrent of presuppositions that provide no moments of awkwardness or embarrassment for participants, and that permit White privilege to be taken for granted.
—JANE HILL1 4.1. Common Ground and Common Enemies It is intuitively obvious that when people communicate, they take a lot for granted, that they assume, that is, they presuppose, many things that their interlocutor is probably already familiar with. These presuppositions form the very fabric into which our communicative acts are woven, without which communicative connection would be difficult to sustain. To switch metaphors, the most basic role of presuppositions in our resonance-based theory of communication is analogous to the role played by a carrier signal, the frequency you turn the dial to if you have an old-fashioned radio. On this analogy, the physiological and neurological makeup of humans is like a tunable radio or walkie talkie, and the conventions of a particular language are like a setting on the dial, which determines the frequency of the carrier signal. When we talk, we presuppose a common setting. Presuppositions fix the basic conventions of communication, sufficiently to provide a stable background relative to which the modulations of the primary intended message can be recognized.
The carrier-signal analogy is hopefully illuminating, and yet it might suggest too limited a role for presuppositions. Presuppositions don’t stop at the level of shared community practices, but go arbitrarily deep, including shared values when talking to someone who shares a taste in music or game shows, shared humanity when conversing with another human, and even some level of shared dispositions and attunement to reality when we talk to a pet or the latest gizmo on our smartphones. Unless we are poets or academics studying communication, we typically become conscious of these carrier signal-like presuppositions only when they fail, as when an accent or an unexpected locution suggests: you’re not from around here, are you?
Beginning with the work of Gottlob Frege and Peter Strawson, the literature on presupposition made a significant break from the intuitive idea that people presuppose things (a break that Robert Stalnaker later suggested had been a mistake). Instead of talking about people presupposing things, much of the linguistic and philosophical work on presupposition suggests that it is people’s words or utterances that do the presupposing.
Thus, for example, the act of saying “Sarah is at home” presupposes the identifiability of Sarah from use of the word “Sarah.” This move ushered in a rich line of work on how presuppositions are encoded in language, driven by the discovery of a set of phenomena that could not easily be explained in models of language that did not distinguish between presupposition and other types of meaning.
Of particular relevance for the current project is the phenomenon of presupposition projection, which will play a major role in this chapter.
Presupposition projection is the remarkable ability of presuppositions that are triggered deep within a complex utterance to pop out, the presupposition behaving much as if it occurred without the complex material around it. For example, suppose someone says, “Most of us doubt that Sarah is at home.” The speaker is clearly not committed to the claim “Sarah is at home,” and yet does appear committed to the identifiability of Sarah, which occurs inside this embedded clause. Here is a similar naturally occurring example. When the journalist and broadcaster Walter Cronkite said, “It is doubtful that the awesome image of this bomb will ever—or should ever—be erased,” he was denying something, and yet the presupposition that a particular mutually identifiable image of a bomb was awesome is clearly something that he was committed to, and took to be uncontroversial.2 Such examples illustrate the way that presuppositions allow people to reveal and share their commitments without making an overt claim that they hold those commitments. This ability to say things without being obvious means that presuppositions have a special role in persuasion, and more generally, in ideological transmission, and that is why we are devoting part II of the book to them.
Because presuppositions tend not to be the focus of attention in a communicative interaction, they can fly beneath the radar. For that reason, presupposition is often exploited in communication, as a way to manipulate someone by using the presuppositional resonances of an action to shift someone’s attunements without them realizing it. As Edward Herman and Noam Chomsky commented in an interview: “There is a firm elite consensus on the legitimacy of state violence—in fact, it is a simple presupposition, which is much more insidious than assertion.”3 Scholars such as Marina Sbisà, Rae Langton, Caroline West, and Mary Kate McGowan have focused on presupposition as a central mechanism by which speech enables ideological transfer.
4 In this chapter, we use examples from experimental psychology and literature to illustrate these effects, and thus justify these authors’ choice, and our own, of focusing on presupposition as a central mechanism of hidden ideological transfer. We will also demonstrate the importance of a related process, accommodation, which is the standard term for the way people adapt to presuppositions.
Accommodation is typically a subconscious process that, as we will later argue, is heavily influenced by social, emotional, and dispositional attunements that are difficult to account for in standard models. We will suggest that accommodation is simply harmonization driven by social cues in the speech context. In cases of both presupposition and accommodation, we rework these notions here and in subsequent chapters to fit into a model that centers their role in ideological transfer.
This part of the book, then, develops an account of presupposition, and the related process of accommodation, in terms of the notions of resonance, attunement, and harmony presented in part I. The literature on presupposition is large. An academic career could be founded merely on writing handbook articles about it. The purpose of this chapter is to introduce presupposition and show its significance for our project, setting the stage for the coming theoretical development.
Let us make a distinction among types of resonance that will help make sense of the division of labor between this chapter and the next, with the caveat that it is not, for us, an important theoretical distinction, but rather one that is practically useful in introducing presupposition to the reader. The distinction, which is gradient, is one of transparency, and thus relates to the notion of hustle. The division of labor simply involves beginning with relatively more obvious cases where a resonance is transparent, and then turning to the more subtle presuppositions that we take to be central to ideological transmission and change.
Any resonance that is highly salient for a practice is, in the sense intended, transparent, in that it is clear that it is associated with the practice. For example, saying “What she ate was a sandwich” makes the sandwich salient, and does not draw attention to the female who did the eating, which is backgrounded, but the existence of a female whose eating is under discussion would presumably seem obvious to the participants. So, some resonances are not merely transparent but highly salient, some are not salient but are transparent, being readily discernible given the grammatical form of a construction, commonsense knowledge about meaning and use, or dictionary definitions, or other factors that make the resonances culturally salient.
There are also resonances that are less readily discernible, neither salient nor transparent.
For example, “The cat is on the sofa” has a readily apparent resonance that its use co-occurs with situations in which a cat is salient.
This salience arises by virtue of practices involving the sentence’s constituent words and practices of combining them. But the sentence also has as a resonance that the speaker is within a certain community of practice, namely English speakers, and that in this community there are certain practices of using the word “cat.” This resonance is not one that would normally be salient among conversational participants when the sentence is produced. On the view we will develop, both the somewhat salient resonance that there is a cat, and the (typically) less salient resonances concerning the practices involved in producing it, are presuppositions. In neither case would it presumably be obvious to most English speakers that they are specifically presuppositional resonances, and indeed this classification is not one that we necessarily expect other academics who study linguistic meaning to agree on.
In this chapter, we will be concerned primarily with presuppositional resonances for which it is somewhat evident that they are resonances, although, crucially to many of the cases we will discuss, it is not so evident that they are presuppositional. These are the examples that are clearest for introducing and motivating the concept of presupposition. In the next chapter, we will turn to presuppositional resonances that are significantly less obvious, cases where the resonance itself is not necessarily evident, and the question of whether it is presupposed or not is not something normally considered.
Presupposition theory revolves around the explanation of distinctive patterns of inference and around the technical machinery needed to model such inference patterns. Our presentation does not focus quite so directly on inference but on influence: we focus on the use of presupposition to persuade. In the next two sections, we consider classic work in psychology through the lens of presupposition theory. We discuss two lines of experimental work: Elizabeth Loftus’s work on leading questions in eyewitness testimony and their effects on long-term episodic memory (section 4.2); and equally famous work by Amos Tversky and Daniel Kahneman on prospect theory, which relies on differential effects on people’s choices resulting from linguistic framing in terms of gains or losses (section 4.3). In both lines of work, presuppositions are crucial to the way that experimental subjects are manipulated. In the Loftus experiments, presuppositions guide subjects into a certain understanding of how to interpret what they have seen, and in the Tversky and Kahneman experiments, presuppositions guide subjects into seeing the choice before them in a certain light.
The presuppositional constrictions considered in sections 4.2 and 4.3 include many grammatical devices: definite descriptions, questions, temporal subordinate clauses, and verbs expressing change. In section 4.4, we consider what might at first appear to be an innocuous part of the grammatical machinery of language: pronouns. We start by considering literary cases in which authors use pronouns to establish common ground.
In these cases, the author places strong presuppositional demands on the reader, effectively jumping into the middle of a story as if the reader already knew who and what it was about.
We then give examples of a similar, widely discussed trick in political persuasion, using an us versus them pronominal distinction to present as if already established a common ground of who is a friend, and who is an enemy. All of the many presuppositional constructions we will consider demand collective attunement. When that attunement is not present, people usually harmonize with their interlocutors without even realizing it, sliding effortlessly right onto their wavelength. Hearers don’t so much read the speakers’ minds as meld with them.
A mind meld might seem magical, but what we are describing requires no special powers.
People’s ability to establish connection, what we have termed harmonization, is no more fantastical than is the ability of musicians to play in tune with each other.
We pay attention to what a group around us is doing and do something that matches what they do. Perhaps an even stronger statement can be made: as we pay attention to those around us, and unless we actively resist it, we become who they are, or who they seem to be. That is a strong claim, and yet such realignment of identity is precisely what the propagandist seeks.
Collective attunement extends what is seen in most work on common ground in philosophy of language and linguistics. One way it extends the standard notion is by incorporating social perspective and emotion, rather than being limited to factual information. Here an unexpected vista comes into view, a surprising connection between the dry mental accountancy seen in standard theories of how we calculate the reference of pronouns and the stirring of passions of hate speech. Both are used strategically to establish common ground. Yet no common ground is stronger, or more motivating, than hatred of a common enemy. This chapter takes us on a journey from presuppositions that draw people into particular beliefs, to presuppositions that draw people into groups that are cleaved along antagonistic societal borders.
4.2. A Collision of Language and Psychology You witness a multiple automobile accident and are soon being interviewed by a police officer on the scene. You’re asked, How fast were the cars travelling when they smashed into each other?
“35 miles per hour,” you respond. Your friend, who also witnessed the collision, is being separately interviewed. Your friend is asked, How fast were the cars travelling when they hit each other?
Your friend says, “25 miles per hour.” The higher answer you gave could have as much to do with bias in the question as with the actual speed of the vehicles. Specifically, the word “smash” in the question you were asked probably had a different effect than the word “hit” in the question your friend was asked. Yet you would deny that there was any such effect: as far as you are concerned, seeing is believing, and you are just reporting what you saw. A week later, you and your friend are each contacted by an insurance claims adjuster, who asks, among other things, “Did you see any broken glass?” You think you might well have seen some, but your friend is confident there wasn’t any shattered glass. You should reflect on the fact that despite what you remember seeing, your friend may well be right. You may remember glass although there was no glass. A biased question like the one you were asked a week ago may have had a greater impact on your long-term episodic memory than did the colliding cars.
Such effects were first reported by psychologists Elizabeth Loftus and John Palmer, 5 and they have been reproduced robustly in many further experiments. The demonstration of the effect of wording on judgments of speed and on long-term memory of the event is part of a seminal group of studies by Loftus, studies in which she showed that eyewitness testimony is unreliable and highly susceptible to the methods used to obtain that testimony. This work led to substantive changes in the US criminal-justice system. The wording of the questions above is verbatim from their experiments, in which subjects saw a video of a car crash. The difference of 10 m.p.h. is roughly the average difference found in those experiments for the different wordings of the question, and people who are asked the “smashed” question are about twice as likely to later report having seen glass than those who are asked the “hit” question.
What is the nature of the bias in the automobile collision questions? In the original paper, Loftus and Palmer didn’t discuss this, beyond saying that the verb smash “biases [the] response,” that the interviewer “supplies a piece of external information, namely, that the cars did indeed smash into each other,” and that the questioner “is effectively labeling the accident a smash,” and hence having an effect on memory representation. However, in a paper the following year in which she considered a wider range of data, Loftus identified the relevant linguistic mechanism: presupposition.
Although Loftus says little about the nature of presuppositions, we understand her idea as follows: the presuppositions of utterances are aspects of the wording that reflect assumptions of the speaker. The apparent assumptions of a questioner as reflected in the presuppositions of the question can change the hearer’s representation of an event under discussion without the hearer realizing that this is happening. The standard view in linguistic theory at the time that Loftus was writing (although it was also challenged at around the same time) is that presuppositions can be conventionalized: when a linguistic expression is conventionally associated with presuppositions, the expression is called a presupposition trigger.
The most widely discussed example of a presupposition trigger is the definite article “the,” a definite description being said to presuppose existence of the entity described. The opposition between “the” and “a,” which lacks this existential presupposition, is the basis of some of the experimental designs Loftus used to show that presupposition was crucial in biasing recall of observed events. Subjects were shown videos and then asked questions of the form “Did you see X?” where X contained either a definite description or an indefinite description. So, for example, some subjects were asked “Did you see the broken headlight?” which presupposes the existence of a broken headlight, while others were asked “Did you see a broken headlight?” which does not carry this presupposition.
In numerous variants of this setup, it was found both that the presence of a presupposition trigger led to greater initial false responses, and that the presupposition trigger led to greater rates of false memory a week later.
6 Returning to the original example in this section, how is presupposition at play in the smash/hit case? Although we will expand upon the explanation in terms of resonance, the basic analysis is easily stated in terms of standard presupposition theory and two well-studied classes of presupposition trigger: wh-questions and temporal subordinate clauses. All wh-questions (“how” being considered a wh-question despite its spelling) are standardly analyzed as carrying presuppositions. Asking the question “Why did you pay?” uncontroversially presupposes that you paid, as do “When did you pay?” and “How did you pay?”7 Likewise, temporal subordinate clauses, which can be headed not only by “when” but also by “before,” “after,” “until,” and “while,” are all standardly analyzed as carrying presuppositions. Thus, the sentences “The shopkeeper smiled before / after / when / until / while you paid” all uncontroversially carry the presupposition that you paid.8 Thus even if the experimental stimuli had not been in the form of a question, there would have been a presupposition. Given the presence of constructions standardly analyzed as presupposition triggers, it is clear that any question of the form “How fast were the cars traveling when they X-ed?” will be associated with a presupposition that the cars X-ed. For one group of subjects X is “smashed into each other,” and for the other it is “hit each other.” So, whatever the meaning differences between “smash” and “hit,” those differences are transformed into presuppositions in the experimental stimuli at hand.
Consequently, if we can figure out what those lexical meaning differences are, we will then know what the differences are in the overall presuppositions of the questions.
The basic difference in lexical meaning is clear. Although “smashed” and “hit” can describe an overlapping set of circumstances, “smash” is used only to describe high-energy collisions, whereas “hit” is used to describe collisions independently of the energy of the collision. You can hit the ground running, but if you smash into the ground, you’re likely to stay there; “violently smashed into” is common, as we found in simple textual searches on the large corpus of books made available by Google Ngram, whereas “softly smashed into” appears not to have occurred in any published book prior to this one. Likewise, neither “slowly smashed into” nor “delicately smashed into” occur even once in a truly gigantic text collection. By contrast, it’s easy to find cases of “delicately hit” and “slowly hit.”9 Similarly “the feather/pillow smashed into” don’t occur, but “the plane/bullet smashed into” are (unfortunately) not uncommon. The fact that feathers, pillows, planes, and bullets all hit things provides clear evidence that “hit” is not restricted to either low-energy or high-energy collisions, while the fact that “smash” combines with planes and bullets but not with feathers and pillows demonstrates again that “smash” is exclusively used for high-energy collisions.
Let us note in passing another difference between “smash” and “hit” that is of interest for the current project, although our suspicion that it is relevant to the original Loftus and Palmer results must be seen as purely speculative.
This concerns what we have called resonance. While neither “hitting” nor “smashing into” an object implies a change in the object, smashing prototypically has a significant effect.
Imagine that you’re strapped for cash and buying a car, and you learn of two that are for sale, one that another car smashed into, and one that another car hit: which do you think will have the lower price? Furthermore, a direct object of “smash” (i.e., X in “smash X” rather than “smash into X”) is substantially changed as a result: it is, in fact, smashed. We can hit something over and over precisely because hitting it may have little effect. But how many times can you smash something without putting it back together in the middle? Even though “smash into” lacks the entailment of partial destruction that “smash” combined with a direct object has, the resonances of the latter form might get activated even by the former, so that having conceptualized an event as involving something smashing into something else, we also come to represent it, associatively, in terms of things getting smashed. More generally, both communicative actions like uttering “smashed into” and noncommunicative actions like smashing into something have resonances, and those resonances may depend on arbitrary contingencies concerning the practices they instantiate, or similarities between these practices and others as perceived by an observer.
It is clear that there are significant differences in the implications and resonances between “smash” and “hit.” Furthermore, as argued above, their positioning in the experimental stimuli (within a temporal subordinate clause that is itself embedded in a wh-question) implies that all these differences become differences in presuppositions in the case at hand.
Whereas in the hit case what is presupposed is a collision between cars, in the smash case what is presupposed is a collision that is associated with high-energy and substantial structural changes in those vehicles.
Biased question experiments like those run by Loftus and colleagues demonstrate that presuppositions are particularly effective at influencing people’s memory of an event, even overriding what they have directly perceived. Why is this? Three relevant factors are widely discussed in the literature on presupposition: first, presuppositions are normally taken to be uncontroversial; second (and now we must introduce some standard terminology), presuppositions project; and third, presuppositions are accommodated. Of these, the apparent uncontroversiality of presuppositions, discussed at least since Stalnaker’s classic work on presupposition,10 is self-explanatory. If Jason says, “The rain is getting heavier,” his use of the definite article presupposes the existence of rain, and he seems to take this proposition to be less controversial than the new information he’s presenting, namely that it’s getting heavier. It’s clear why uncontroversiality might be relevant for the effects Loftus observed: one would expect experimental subjects to be less circumspect and less careful when they integrate information that is linguistically marked as uncontroversial, than they would be in integrating other information, and thus the subjects might not take as much care to check presupposed information against their prior memory representations as they would with other information. In terms of our discussion in chapter 3, it is plausible that presuppositions are commonly subject to nondeliberative uptake.
In the classical examples of presupposition, interpreters regularly see them as commitments of the speaker, even when they are embedded in a larger construction that would otherwise prevent inferences about the speaker’s commitments. This phenomenon is called projection.
Questions are an example of an embedding construction that blocks the speaker’s commitments. For example, when you ask, “Did Jason come to the party?” you’re clearly not committed to Jason having come to the party. Two other construction types that also typically block speaker commitments are negation and possibility modals, words like “maybe” and “could.” Normally these block commitments. If you say, “Maybe Sam likes tennis” or “Sam doesn’t like tennis,” or, for that matter, “Maybe Sam doesn’t like tennis,” you’re not committed to Sam liking tennis.
Now, consider this example from an MSNBC article: “Maybe Trump doesn’t realize that his Opportunity Zones policy has been exposed as a sham.”11 The writer appears committed to the proposition that Donald Trump’s policy has been exposed as a sham. The commitment of this proposition remains even under the words, “maybe” and “doesn’t,” which, as we have seen, typically block commitment. Why? Because realize is a presupposition trigger, and the propositional complement of realize (“his Opportunity Zones … sham”) is presupposed, and thus somewhat immune to the commitment-blocking effect of the modal and the negation.
Presuppositions can also project from questions, despite the fact that the primary function of questions is to find out about commitments, not to make them. For example, consider the following question from Trump: Was Andy McCabe ever forced to pay back the $700,000 illegally given to him and his wife, for his wife’s political campaign, by Crooked Hillary Clinton while Hillary was under FBI investigation, and McCabe was the head of the FBI??? Just askin’?12 The question is whether Andy McCabe was forced to pay some money, so clearly Trump is not committed to him having paid this money. Indeed, it presumably being a rhetorical question, Trump in fact seems committed to the opposite. However, there are a number of presuppositions in the question. You can only pay something back if you had it in the first place, and the question presupposes that McCabe received this money. There are a number of other presuppositions, including the propositions that there was $700,000 illegally given to McCabe, that McCabe has a wife who had a political campaign, that Hillary Clinton is appropriately referred to as Crooked Hillary Clinton, that Clinton was under FBI investigation, and that McCabe was the head of the FBI. The question form appears to introduce a reasonable issue for discussion (perhaps suggesting that after answering it negatively, we might ask ourselves Why not?), but Trump uses the question to present all this presupposed information, which projects from the question, as if it were uncontroversial. The paradigm Loftus introduced employs a sneakiness similar to Trump’s question: presuppositions embedded in questions project, and are accepted as reasonable by a significant percentage of subjects, who are perhaps distractedly making a good-faith effort to answer the question that has been asked. They wouldn’t even be able to answer the question directly if they didn’t treat the presuppositions as true. Hearers do what’s needed in order to coordinate with the speaker, for example, by answering their questions if they possibly can. This coordinative adjustment is a part of the process of collective harmonization discussed in chapter 3. We can also understand it in terms of what the philosopher Donald Davidson terms principles of charity, although Davidson himself does not distinguish clearly between presupposed and nonpresupposed information.13 Davidson suggests that in order to understand an interlocutor, we use a principle of coherence, whereby we understand them as having an internally consistent worldview, and a principle of correspondence, whereby we do our best to take what they are talking about to correspond to aspects of the world that we ourselves are sensitive to, or at least would be sensitive to if in the speaker’s position. Thus, if someone alludes to the rain, to a time when cars smashed into each other, or to the $700,000 illegally given to someone, we will assume that these descriptions are part of a rationally coherent way of looking at the world, and that they are parts of the world that we ourselves could detect if we have not already done so. When someone asks a question, we instinctively interpret the question charitably, accepting that the things apparently referred to in the question really exist, and using these charitable assumptions as the basis for proceeding with the conversation in accord with the speaker’s designs.
In later work, Davidson describes the charitable impulse of speakers in terms of “a policy of rational accommodation,” that is, a policy of adapting to interlocutors so as to understand them.14 This terminological shift is apt.
Accommodation, which has been a central notion in presupposition theory since David Lewis introduced the terminology forty years ago,15 is the process that yields the most significant difference between presupposed and nonpresupposed information. Lewis and others working on semantics and pragmatics have since understood accommodation in a broadly similar way to Davidson, as a process by which hearers adapt to the demands of the speaker. In our terms, hearers adjust some of their attunements to match those of the speaker in a process of harmonization. Although Lewis sees accommodation at play in a number of processes, the example that he discusses at greatest length and that has taken a firm hold in subsequent literature involves accommodation of presuppositions. The idea is straightforward: when someone says something that presupposes something to be true that the hearer did not previously believe, the hearer can accommodate by adjusting their beliefs in line with the presupposition. So, for example, Trump’s eighty-five million Twitter followers could (before he was banned from that social-media platform) accommodate to him by accepting, erroneously, the existence of a massive illegal payment to McCabe.
Although our own analysis of accommodation as a special case of harmonization differs somewhat from those in most Lewisian literature, we are in agreement with prior scholars as regards a simple point that is crucial to the current discussion: accommodation of presuppositions is a different process than regular uptake of the primary speech act a speaker has made.
In particular, standard analyses suggest that, when a speaker makes an assertion, this is seen as a negotiable proposal for addition to the common ground. However, the standard view is that presuppositions are presented as nonnegotiable. If presuppositions must be accommodated, this tends to occur in a relatively automatic way that is not subject to extensive deliberation.16 If presupposition accommodation typically happens in an automatic way when it happens, then it is not subject to deliberation.
Let us however distinguish the automaticity of accommodation from its rapidity. The gradual acquisition and improvement of skills has a large nondeliberative element. When someone learns to hit a jump shot in basketball, they train and improve gradually though nondeliberately and automatically via practice. Accommodation can also work this way, particularly in the accommodation of practices (e.g., speech practices). The claim that accommodation is automatic and nondeliberative does not entail anything about its rapidity.
The term accommodation was not used in this way when Loftus and colleagues published the original studies discussed above, but the paradigms they created depend on experimental subjects’ accommodating unsatisfied presuppositions. These and further studies have demonstrated that accommodation of presuppositions can lead to people developing false memories, with no awareness that those false memories arose through the use of language that leads them to accommodate false information.
Describing the series of experimental paradigms she created, Loftus explains that “information was introduced via presuppositions in questions, a technique which is effective in introducing information without calling attention to it.”17 That is, presupposition is a mechanism that can allow information to be smuggled into a person’s mind, as it were, below the radar, nondeliberatively. In short, it is no surprise that foundational theorists such as Langton have centered this notion. It is one ideal communicative mechanism for ideological transfer.
4.3. Valence Framing It is apt, writing in the midst of the Covid-19 pandemic, that we next turn to an only mildly adapted version of the famous “Asian Disease” problem developed by psychologists Daniel Kahneman and Amos Tversky, 18 which involves an experiment designed to probe the effects of framing. The very framing of a disease as “Asian” has become an intensely political matter during the pandemic, what would be described as a question of issue framing or attribute framing in the political-science literature on framing that we have commented on at various points in part 1 of the book (sections 1.3, 3.4, and, especially, 3.9). The experiments we now turn to were among a group of studies playing a pivotal role in developing the overlapping fields of experimental and behavioral economics, and that are now standard fodder for introductory psychology courses. The research emerged in a separate tradition from other work on framing we have discussed, in psychology rather than political science. What is at stake in the experiments is not some particular attribute of a problem, like the country of origin of a disease, but rather the question of whether something is set in a positive light, in terms of potential gains, or in a negative light, as a loss. This glass half-full versus glass half-empty difference is sometimes termed valence framing.
19 We now present the experiments and the standard analysis, briefly set that analysis in the context of the resonance framework introduced in part I of the book, and then turn to the question of how presupposition can help make sense of some aspects of Kahneman and Tversky’s reasoning about the “framing effect” (as they termed it) that they observed. We begin with the experimental stimuli, reproduced from the original study.
20 Context: Imagine that the United States is preparing for the outbreak of an unusual Asian disease, which is expected to kill 600 people. Two alternative programs to combat the disease have been proposed.
Assume that the exact scientific estimates of the consequences of the programs are as follows: Loss-framing decision problem: i. If program A′ is adopted, 400 people will die. ii. If program B′ is adopted, there is a one-third probability that nobody will die and a two-thirds probability that 600 people will die.
In the original study, 78 percent of subjects chose the option that preserved a one-third probability of nobody dying. However, it will come as no surprise that the wording of the choices is crucial. In a second condition, subjects saw the same initial context, but wording in terms of lives saved rather than lost. Gain-framing decision problem: i. If program A is adopted, 200 people will be saved.
ii. If program B is adopted, there is a one-third probability that 600 people will be saved and a two-thirds probability that no people will be saved.
This reversed typical preferences, as a clear majority (72 percent) in this condition preferred the safe option, a guarantee of lives being saved. These experiments, and the attendant effects, are among the most reproduced in psychology, although many studies have differed in important respects from the original.21 The stimuli are designed in order to reveal departures from the predictions of classical decision theory, based on the assumption that people maximize expected utility. The experiment is designed so that all four options have the same expected utility, since in each option there is an overall expectation that 200 people will live and 400 will die.22 Thus, expected-utility theory suggests that in both conditions, subjects will be indifferent to the options, predicting a roughly 50/50 split in both conditions, and certainly no reversal of preferences from one condition to the other.
Kahneman and Tversky explain the data using one part of what they term prospect theory, a theory in tune with our project insofar as it focuses on the context-sensitive nature of human decision-making. Options looked at from one angle can seem very different than from another; they present quite different prospects (“prospect” coming from the Latin prospectus, “view”). The central idea relevant to the preference reversal in the Asian Disease problem is that people have diminishing sensitivity to increasingly large gains or losses. This idea is essentially as old as the above method for calculating expected utility, both being products of the Swiss polymath Daniel Bernoulli, over three centuries ago.23 In standard decision theory, it shouldn’t matter to your choices on simple decision problems whether you’re a billionaire or a pauper. But think about the difference between how happy you’d feel if someone gave you a free ticket in a lottery and you won nothing, versus how you’d feel if you won a hundred dollars. A pretty big difference, right? Now think about the difference between how happy you’d feel if you won a million and how you’d feel if you won $1,000,100. Not so much difference! That is, as the gains grow big, we become increasingly indifferent to small further gains.
Contrarily, whereas a loss of $1K would count for most of us as a pretty bad day at the races, few of us would say there’s a great deal of difference between a day when we lost $50K and a day when we lost $51K. Of course, in each case the difference, $1K, is the same. We might say that the prospect of that $1K is very different at a distance of $50K than it is right up close.
The effects follow immediately. Each time a big number is presented (600 saved or 600 dying), that number is effectively discounted slightly more than the smaller numbers (400, 200, or 0). To see the effects of discounting in an oversimplified way, try replacing every occurrence of 600 by 500 in the problems, but leave everything else the same. It is immediately clear that in the loss framing, option ii, the risky option, has higher expected utility, because what was 600 deaths risked has become 100 deaths less awful. Similarly, in the gain framing, the risky option suddenly has lower utility than it had before, because the 600 saved has been replaced with 500 people saved. Hence in this case we get a preference reversal: as Kahneman and Tversky put it, people are risk averse for gains.24 We can see Kahneman and Tversky as having revealed a problematic idealization in classical economic theory. The traditional, idealized model takes any object of desire, whether it is people being alive rather than dead, or a sum of money, to be equally desirable whatever the situation. But as we have seen, this idealization does not match intuition. A hundred dollars is a nice little gain, but has less value to you when judged as part of a gain of a million: on top of a million it seems remote and unimportant. What Kahneman and Tversky tell us is that value is not absolute, but is relative to some reference point. And moving that reference point may even turn a perceived gain into a perceived loss or vice versa. This, they claim, is what happens in the Asian Disease problem. They say that framing the decision problem using the word “save” sets up a reference point in which the default state is people dying, so that people tend to view the problem in terms of possible gains, whereas framing the problem using the word “die” sets up a reference point in which the default state is people being alive, so that people tend to view the problem in terms of possible losses. Kahneman and Tversky’s modeling is convincing. However, their use of the term “framing” labels a phenomenon without providing any insight into how it relates to the lexical semantics of the expressions as conceived of in linguistics, philosophy of language, or lexicography. Neither does their notion of framing relate in any way that’s been made explicit to standard notions in the theory of how language is used, pragmatics. It is not at all obvious where in the theory of meaning Kahneman and Tversky’s notion of framing resides.
Let us now consider in general terms how their notion of framing might relate to resonance, attunement, and the standard concept of attention, which played a big role in our discussion of harmonization, and then turn to the issue of more immediate import to the current chapter, the relevance of presupposition.
There is a lot going on in the frames used in the Asian Disease problem, and even more going on in the many other variants in the literature that we are not considering here. At the very least, we can reframe the framing discussion in our own terms, by saying that the difference between the conditions is one of resonance: the saving lives wording has different resonances than the people dying wording. In these terms, probably the most important difference between the wordings concerns the fact that people have different emotional attunements to uses of “saving lives” than to uses of “people dying”; that is, the first has an emotionally positive resonance, and the second has an emotionally negative resonance. We can also say, relatedly, that the two wordings have different attentional effects.
Whereas “saving lives” draws attention to a positive aspect of the outcome, “people dying” draws attention to a negative aspect of the outcome.
Neither of these differences in resonance have to do with reference point setting per se.
They have rather to do with how people feel about the outcomes, and what people focus on.
Let us now consider what Kahneman and Tversky’s analysis that the two framings induce different reference points amounts to. The mechanism by which “save” sets up a reference point in which the default state is people dying is not mysterious. It’s presupposition. You can only save money that you would otherwise have had to spend, you can only save time if you expected a task to require it, and you can only save people’s lives if they were otherwise going to die.
Describing an action as “saving” presupposes an expected loss, although certainly the presupposition is not as transparently worn on the sleeve of the word “save,” the presupposition trigger, as with many other cases discussed in this chapter. It’s a testament to the siloed nature of academia that the framings in the Asian Disease problem stimuli have not, to our knowledge, been analyzed explicitly in terms of presupposition before.25 We know this is a case of presupposition rather than some other aspect of the meaning, because the standard diagnostic for presupposition, the property of projectivity, tells us. To use this test to figure out whether an expression presupposes some feature of context, we check whether we would normally expect that feature to be present in both simple positive uses of the expression, and in uses where the expression is embedded in a way that would tend to suppress a speaker’s commitments, say because the utterance expresses a denial, a question, or a hypothetical, and the ordinary meaning is thus not asserted but questioned, denied, or merely hypothesized. (For detailed discussion of projectivity, see section 5.4.) So let’s test the claim that uttering “save X” presupposes the expectation of loss of X (or at least damage). First, it’s clear that this holds for a simple positive use of “save X.” We see this in the names of thousands of organizations using a “Save the X” template, everything from “Save the Amazon” to “Save the Zambizi,” names that only make sense under the assumption of a clear and present danger to X. Now we can look at embeddings. Here are three headlines using the verb “save,” with embeddings involving negation, a question, and an expression of possibility: The Pandemic Won’t Save the Climate.26 GM/UAW Strike: Will the Strike Save the Cadillac CT6?27 Pilsen Residents Say Landmark Designation Might Save Buildings, But It Won’t Stop Gentrification.28 It is clear that in the context assumed by the writers and editors, there is severe danger of damage or loss to the climate, the Cadillac CT6, and buildings in Pilsen, respectively. This illustrates the projective behavior we hypothesized, and so, based on its projective behavior, “save” is a presupposition trigger, triggering a presupposition of expected loss. Hence, the two options in the gain-framing condition for the Asian Disease problem presuppose that the context is one in which mass death is anticipated. All lexical predicates that denote a change of state or the lack of it presuppose some initial state, so that, for example, “stop smoking” and “keep smoking” both presuppose prior smoking. So, no diagnostic testing is required to demonstrate that the two constructions involved in the framing of the Asian Disease problem, “X’s life will be saved” and “X will die,” clearly have a further common initial state presupposition, namely that X is alive. But unlike in the “save” case, the verb “die” is not associated with a lexical presupposition that someone was expected to live or expected to die.
There is only a normal default assumption of continuity, which implies that living things are expected to continue to live unless there are reasons to expect otherwise. However, this isn’t part of the meaning of “die.” The contrast between “save” and “die” in this respect is illustrated by fact that the first version of the following constructed example is odd (as indicated by the hash), while the second version is fine: There had been nothing threatening the patient’s life, and … a. # fortunately, the doctors saved him.
b. unfortunately, he died.
One further nuance is that even leaving life-and-death decisions aside, saving something is stereotypically good, while the expected state that saving prevents is bad. We can add to this obvious fact that the inherent goodness of saving, an aspect of its positive affective resonance, is presupposed. We can see this by comparison with the verb “prevent.” That the two verbs are different is clear: if someone is sick, you can prevent them from getting better, but you cannot save them from getting better. That this difference stems from the verbs’ presuppositions is evidenced by how the presuppositions of “save” project, for example, in questions. The nonsense question “Did David prevent Jason from being splogged?” tells us little about whether being splogged would have been a good thing, whereas the question “Did David save Jason from being splogged?” clearly implies that sploggings are best avoided. Thus, to save is inherently good in just the sense that it involves prevention of an expected outcome that is presupposed to be bad. Recall that we said that the Kahneman and Tversky account was incomplete. The problem is that their account includes only a partial explanation of what linguistic properties of the experimental stimuli cause subjects to set the reference points they use in making their decision. What Kahneman and Tversky say is that one stimulus pair involves a framing in terms of gains, and the other involves a framing in terms of losses. This is fine as far as it goes. But we can now say more.
As a preliminary, note that if we are right, then there’s a sense in which the two experimental conditions are not symmetrical, because one framing carries more relevant presuppositions than the other. Specifically, the save condition is the more semantically interesting case, in that it has the rather special presuppositions just discussed, whereas die just carries the normal presupposition of any change of state verb (that the entity in question is in a state such that the transition is possible, i.e., in this case that the subject is alive). Further, although your own moral stance may suggest otherwise, as regards lexical meaning the badness of death is at most a presupposed affective resonance, and not a presupposition that something dying is inherently bad. If you’re a vermin exterminator, the more cockroaches die, the better. So, let’s begin with the assumption that it’s the special context in which you’re trying to save human lives in a pandemic that makes mass death a bad thing.
The story goes like this, then. First, since people dying presupposes that they are alive, in the same way that any transition presupposes its start state, the loss framing tends to push the reference point as one where people are alive. Since human deaths are bad, the stimuli in the loss framing do indeed express losses relative to that reference point. Let us note in passing that this setting of the reference point is somewhat in tension with the description given to all subjects in both conditions, which in our slight variant is “a disease that is expected to kill three million people worldwide.” Second, since “save lives” presupposes a strong expectation of death, and describes actions that will potentially lead to outcomes other than what is expected, the gain framing sets up a reference point in which people are expected to die, and in which the outcomes are reversals of this expectation.
Since death is both bad in this context and presupposed to be bad by “save lives,” that reference point is seen as having lower expected utility than what is achieved by taking either of the life-saving options. Thus, the second formulation of the disease problem is a framing in terms of gains. In both cases, the framing effects are produced by presuppositions embedded in the experimenter’s language. We can then understand uptake by experimental subjects as presupposition accommodation. Experimental subjects harmonize so as to have a short-term collective attunement with their interlocutor, implying a disposition to reason about the problem in a similar way. In one experimental condition, collective attunement implies a disposition to reason about the problem as if it involves potential gains, and in the other condition it implies a disposition to reason about the problem as if it involves potential losses. If we take someone’s perspective in this experiment to mean the reference point used for conceptualizing the problem, then what the results of the experiments suggest is that there is a fairly high level of attunement to the experimenter’s proffered perspective.
This is not to say, by the way, that we are claiming that subjects require a complex theory of mind to perform this experimental task. People simply have tendencies to harmonize, tendencies that need not be the results of deliberation. But here let us note that we are not alone in claiming that deliberation is not essential to the process of developing common ground.
For example, Elisabeth Camp takes a similar line in discussing quite different phenomena: she has argued in her explanations of both metaphors and slurs that subjects simply have a tendency to accept the perspective that’s offered to them.29 Analyzing the Asian Disease problem in full would require further discussion of exactly how the above presuppositions interact with the numeric quantifiers in the examples (a type of interaction that one of us has studied30), as well as a more thorough discussion of the specific properties of the context set up for both the save stimuli and the die stimuli. There is, furthermore, a much broader experimental paradigm involving reference point setting through framings, and that is itself part of an even broader set of paradigms involving interactions between attention and decision-making.
What you pay attention to affects how you weigh different things in your judgments and decision-making. As discussed in the last chapter, attention is shifted by communication when we harmonize with our interlocutors and develop shared attention with them. We described these attentional effects of communication in terms of collective harmonization, suggesting that processes of collective harmonization can be studied at radically different time scales, anything from the instantaneous mimicry and mind reading of conversational interlocutors to the larger time scales of cultural diffusion.
Joint attention of the sort relevant to the current discussion drives shortterm collective attunement, the shared attunements we have at the level of fractions of a second or minutes.
These shifts in attention are an important vehicle for propaganda, precisely because the cause of the shift is often not apparent to us. That the shift occurs at all, and especially that it’s typically beyond our conscious control, demonstrates how subtle a power others can have over us, which is why manipulation of attention is widely used in persuasion.
Even a moderately comprehensive discussion of how existing psychological results of reference-setting paradigms, what is usually called grounding, relate to our project would require a book in itself. The larger literatures on attention and joint attention would require much more. So, we desist from further such discussion here. It suffices for our immediate enterprise that we have shown how the framing terminology used by Kahneman and Tversky and in much following literature in cognitive psychology and behavioral economics can be related to the presupposition terminology more standard in linguistic semantics and pragmatics. As we saw, this terminology is not unknown in cognitive psychology, since it was already employed by Loftus and followers. The smash/hit and save/die paradigms reveal that a single linguistic mechanism underlies two quite distinct cases by which bias is transmitted to a hearer without their becoming aware of it. That mechanism is presupposition accommodation.
4.4. Marking Our Common Ground “It was now lunch time and they were all sitting under the double green fly of the dining tent pretending that nothing had happened.” That is an odd way to start a section. What makes it even odder than typical sentences in this book is obvious: you can’t tell to whom “they” refers, or, for that matter, when “now” is, or recover anything more about the identity of the dining tent. Words are triggering presuppositions of a rich common ground in which various entities are joint objects of attention, and the presuppositions are unsatisfied.
The pronoun and definite description presuppose entities jointly available for reference, and the use of past tense and the unresolved temporal marker “now” presuppose a commonly identifiable time interval.
These constructions presuppose objects of such high salience that the speaker’s strategy assumes a full description is not necessary for coordination. And yet, the quote is the beginning not only of this section, but also a short story, by Ernest Hemingway.
31 Presuppositional attunement is a central mechanism behind the formation and maintenance of us-them distinctions. We will concentrate in this section on the presuppositions of pronouns. By focusing on the presuppositions of pronouns, the way their use calls for joint attention to a shared referent, we can shed light upon how linguistic mechanisms can be used to bond and to divide.
To utter a pronoun in a sentence is to signal that a presupposed object of joint attention plays a certain role in the proposition the sentence expresses.
The sentence “She disagrees, doesn’t she?” would be odd out of the blue, because it is neither clear who “she” is, nor what claim she disagrees with.
Presupposing joint attention to a female individual and a claim would be fine in a context where you had just asserted, “Everyone says the queen should abdicate in favor of her son.” In that case, to say “She disagrees, doesn’t she?” would be to suggest that the queen disagrees with the claim that she should abdicate. Insufficient context is available when the sentence is used out of the blue for much sense to be made of it.
Absent an appropriate context, you could imagine simply accommodating the existence of a female and a claim so that the sentence conveyed merely that some girl/woman disagreed with something.
Similarly, you can accommodate referents that you conceptualize merely as whoever the hell the speaker is thinking of. These strategies provide you with a way of interpreting the sentence, but they miss the point: you still don’t know which things the speaker was presupposing to be centers of joint attention. In addition, you haven’t managed to discover the proposition that the speaker expressed, merely a proposition that approximates what the speaker expressed and yet isn’t about the same objects.32 The presuppositions of pronouns are a little different than those of other constructions we’ve discussed in this chapter. As should be clear, they’re much harder to accommodate, a fact discussed in earlier literature and confirmed in empirical work.33 If the speaker presupposes that two cars smashed into each other, it’s obvious what to accommodate: that the cars smashed into each other.
Similarly, if the speaker presupposes a way of thinking about a decision problem in terms of a reference point where some number of people are expected to die: just accommodate that reference point. But if the speaker presupposes joint attention on something and you’ve no clue what it is, you can’t accommodate by attending to that thing.
Here are your options: either give up, as the reader may have done with “She disagrees …,” or else take a leap of faith. The leap of faith amounts to accommodation by guesswork, filling in the referent as best you can, perhaps hoping that its identity will become clearer as discourse proceeds.
But the speaker is asking a lot of you.
Sometimes, the speaker, or writer, does ask a lot of you. Here’s a tweet that’s circulated as we’ve been writing this book, a quote from F. Scott Fitzgerald: They were careless people.… They smashed up things and creatures and then retreated back into their money or their vast carelessness, or whatever it was that kept them together, and let other people clean up the mess they had made.34 It was cited in the New York Times as having taken on a “new resonance.”35 The general nature of that resonance is of interest for us, but we want to focus on just one question: who does “they” refer to in the tweet? It is notable that what has been elided from the original Fitzgerald quote actually provided referents: the original began “They were careless people, Tom and Daisy—they smashed up things.” It certainly might be argued that the fictional Tom and Daisy are still the referents of the tweet, and doubtless some of those reading the tweet remembered the book, which they might well have read in high school. But what gives the tweet a new resonance is presumably the fact that the quote can be seen not as about the fictional Tom, Daisy, and their ilk in the 1920s, but about their counterparts in today’s world. But who, exactly, is being talked about? It’s presumably not the group referred to by “they” in the idiomatic template “They don’t call me X for nothing,” if indeed that “they” can be said to refer to a group at all, rather than being a sort of expletive place-holder. Neither is it the same group referred to by “they” in the slightly less idiomatic but not uncommon “They finally caught [description or name of criminal],” where the reference of “they” is vague. Given this vagueness, it is not obvious that it makes sense to ask what the pronoun refers to. It might be better to say that the pronoun marks a role, a role that is presumably occupied by a societally relevant group like the local or national police.
The sentence does not clearly mean anything more than the passive “[description or name of criminal] was finally caught.” The “they” of the tweeted quote does not include among its referents the thousands of people who retweeted it. Neither is it the same “they” as in the following National Rifle Association ad from 2017. Here we’ve highlighted relevant occurrences of “they” and “their,” as well as “we” and “our”: They use their media to assassinate real news. They use their schools to teach children that their president is another Hitler. They use their movie stars and singers and comedy shows and award shows to repeat their narrative over and over again. And then they use their ex-president to endorse the resistance.
All to make them march, make them protest, make them scream racism and sexism and xenophobia and homophobia. To smash windows, burn cars, shut down interstates and airports, bully and terrorize the law-abiding—until the only option left is for the police to do their jobs and stop the madness.
And when that happens, they’ll use it as an excuse for their outrage. The only way we stop this, the only way we save our country and our freedom, is to fight this violence of lies with the clenched fist of truth. I’m the National Rifle Association of America, and I’m freedom’s safest place.36 Just as we noted in chapter 2 that when the line “Play it, Sam” is uttered in Casablanca, the referent of the pronoun “it” is not explicit, so it is that in this advertisement the referent for “they”/“their” is not explicit. We do learn that “they” have a rightful president, obviously Donald Trump. We are to infer that Trump is not another Hitler. “They” also have an ex-president, Barack Obama. We are told that “they” use Obama for the nefarious purpose of endorsing the resistance. And why do “they” use stars, shows, and an ex-president? Answer: “to make them march,” etcetera. For the intended audience of the ad, only one thing matters about the referent of any of the third-person pronouns: it’s not “us.” The “we” of the final paragraph are the right-minded Americans who recognize the danger “they” pose. These people’s country and freedom are apparently in danger, for the NRA is offering to “save” them. As an aside, although “save” is not the object of discussion in the current section, it is no coincidence that the same word is being used in an NRA ad as in an experimental stimulus designed to introduce bias, a word with a clear presupposition of incipient danger.
The NRA ad above is not unusual as regards its use of an antecedentless they to prime common attunement to an unidentified evil group. We have not made a quantitative estimate, but, impressionistically, the device is common in propagandistic discourse we have studied. Here are two “Qdrops,” that is, messages posted by Q, creator of the QAnon movement, a personage who, fittingly for the world’s number-one conspiracy theorist, may not exist: Their need for symbolism will be their downfall.
Follow the Owl & Y head around the world.
Identify and list.
They don’t hide it.
They don’t fear you.
You are sheep to them.
You are feeders. Godfather III.
Q37 Why are we here?
Why are we providing crumbs?
Think MEMO.
BUILDING THE ARMY.
Not convinced by this spreading?
You, the PEOPLE, have THE POWER.
TOGETHER you are STRONG.
APART you are weak.
THEY WANT YOU DIVIDED.
THEY WANT RACE WARS.
THEY WANT CLASS WARS.
THEY WANT RELIGIOUS WARS.
THEY WANT POLITICAL WARS.
THEY WANT YOU DIVIDED! LEARN!
FOR GOD & COUNTRY—LEARN!
STAY STRONG.
STAY TOGETHER.
FIGHT, FIGHT, FIGHT.
This is more important than you can imagine.
Q38 For an uninitiated reader of Q’s many musings, the lack of explicit antecedents for the pronouns would seem to be the least of your interpretative problems. But this is to miss the point of these QAnon posts, which are supposedly revelatory leaks of classified information from a well-placed governmental source. With their crazy swerves between Koanic profundity, telegraphic urgency, and surrealist absurdity, Qdrops are to leaked government secrets what the sound of one hand clapping is to music.
To borrow an expression from Gertrude Stein, there’s no there there: it’s all resonance and no facts. Indeed, to say there are no facts is an overstatement: it is hard even to find clearly stated propositions. Lack of explicitness, juxtaposition of suggestive, emotive imagery, epigrammatic imperatives, and puzzling allusions are all central to the QAnon practice. What is demanded of the reader is not uptake of information, but the embrace of contradiction and omission, and preparedness to see whatever is unexplained as something that has been intentionally hidden. What is presupposed is not just a particular collection of facts or referents for pronouns and other gnomic nominals (“the owl & Y head,” “crumbs,” “MEMO”) but collective attunement to a deeply paranoid way of thinking and talking. And once you’ve accepted this presupposed common ground, once you’ve been, as QAnon adherents put it, “red pilled,” your world will never be the same again.
Let us note in passing an interesting complication that we will not discuss in detail here. As the reader may have observed, the Qdrop involves a sophisticated three-way distinction not covered by the standard us-them dichotomy, a distinction between us, you, and them. Q seems sometimes to appeal directly to the reader, and sometimes to appeal to the reader indirectly, as part of the (presumably inclusive) second-person pronoun plural collective we/us. It is perhaps of relevance here that the identity of the anonymous Q is of an insider-outsider, hiding within the body of federal government, distinguished from other government officials by their revelatory understanding, but also distinguished from their legion supporters, who are taken to be outside the government machine, and not privy to the special access that Q has to what is really happening in the circles of power. The combined use of you and we/us is resonant with Q’s nuanced position, at once one with their audience, and simultaneously a person apart, able to address them from a distant perch.
Returning to a very different but equally subtle use of an unresolved pronoun in an alternate world, why was the “it” in the Casablanca line “Play it, Sam!” not given an explicit referent?
As we discussed in chapter 2, this device manifested common ground between the characters of the movie. But unresolved pronouns also serve to pull an audience into a narrative. Adding to the Hemingway quote with which we started this section, here are some more literary first lines that use this device (with emphasis added to the unresolved pronouns):39 “He was born with a gift of laughter and a sense that the world was mad.” (Raphael Sabatini40) “He—for there could be no doubt of his sex, though the fashion of the time did something to disguise it—was in the act of slicing at the head of a Moor which swung from the rafters.” (Virginia Woolf41) “He was an old man who fished alone in a skiff in the Gulf Stream and he had gone eighty-four days now without taking a fish.” (Ernest Hemingway42) “Deep one night he was trimming his nose that would never walk again into sunlight atop living legs, busily feeling every hair with a Rotex rotary nostril clipper as if to make his nostrils as bare as a monkey’s, when suddenly a man, perhaps escaped from the mental ward in the same hospital or perhaps a lunatic who happened to be passing, with a body abnormally small and meagre for a man save only for a face as round as a Dharma’s and covered in hair, sat down on the edge of his bed and shouted, foaming,—What in God’s name are you?” (Kenzaburō Ōe 43) “Thanks to his rare talent for keeping a diary over an extended period of time without missing a single day, he was able to cite the exact date his vomiting started and the exact date it stopped.” (Haruki Murakami44) “It was a wrong number that started it, the telephone ringing three times in the dead of night, and the voice on the other end asking for someone he was not.”45 (Paul Auster46) “It was like so, but wasn’t.” (Richard Powers47) “They shoot the white girl first. With the rest they can take their time.” (Toni Morrison48) As a literary device, the unresolved pronoun presents the reader with a conundrum. On the one hand, someone is speaking to you as if there were already sufficient common ground for you to resolve a pronoun, but on the other hand, there is no such common ground. The device biases you, as it were, to read on, for there is no other way to resolve the pronoun and the tension it has created.49 So in part the unresolved opening-line genre is parallel to what we see in some online clickbait: “Use this one simple trick to …,” but the literary device, unlike this type of clickbait, offers at least a hint of a conversation joined midway, as if the reader had just asked the author to tell them more about somebody’s life.
Authors of fiction are connoisseurs and controllers of context. To stretch the terminology of low and high context slightly from its standard anthropological use (where it is used to talk about differences in culture), a literary author can present a low-context situation, in which there is little assumed common ground, as if it were a high-context situation, in which there is a great deal of common ground. Authors manipulate context not only to seduce readers into the web of their stories, but also to portray everyday situations in which there are clashes of assumed context. A beautiful and very relevant example is found in Flannery O’Connor’s short story “Everything That Rises Must Converge.” The young central protagonist, Julian, and his mother, both White, have just stepped onto a bus in the US South shortly after racial integration, and we focus on Julian’s mother (already salient): She sat forward and looked up and down the bus. It was half filled.
Everybody was white. “I see we have the bus to ourselves,” she said.
Julian cringed. “For a change,” said the woman across the aisle, the owner of the red and white canvas sandals. “I come on one the other day and they were thick as fleas—up front and all through.” “The world is in a mess everywhere,” his mother said. “I don’t know how we’ve let it get in this fix.”50 Flannery represents a high-context Southern White culture, deeply infused with racism, in which everyone present understands tacitly what everyone else is saying. The use of pronouns, which is just one way Flannery constructs that high context, is striking. When Julian’s mother says “we have the bus to ourselves,” the reader immediately realizes that she can’t mean that the two of them have the bus to themselves, for the bus is halffilled. She means there are only White people on the bus. Thus, that single, simple utterance, apparently not directed at Julian but thrown out into the air of the bus, presupposes an us-them distinction in which Blacks are the other, not named, but referred to in a way that is at once implicit and shockingly blunt. The second woman’s rejoinder refers to the unintroduced other, Black people, with a third-person pronoun, and, in case there was any doubt at all about the deeply racist attitudes being portrayed, uses the standard dehumanizing trope of comparing out-group members to a parasitic infestation—the truly awful “they were thick as fleas.” Julian’s mother is clearly in agreement with the sentiment.
Her repetition of “we” in the final utterance offers solidarity, albeit while sharing discomfort and accepting collective blame.
The antecedentless pronouns in the NRA ad (and likewise in the Qdrops) are functionally similar to the literary cases. The use of pronoun in the ad for which no referent has been made explicit has a strong presupposition, the presupposition that in comprehending the ad you have as objects of attention the same sets of individuals as Dana Loesch, the narrator.
In every one of the above literary cases, the author soon makes entirely clear who is being referred to, but in the NRA ad, there’s simply no clear reference, as in most propagandistic cases with which we are familiar.
We are never granted the resolution of a full noun phrase describing the group in question, just the hammering repetition of awful properties the group has. You hear the hammer strike but have to imagine the wall it is demolishing. That is rhetorically powerful. In work that was influential as Nazi ideology took hold in prewar Germany, Carl Schmitt argued that the distinction between friend and enemy is central to politics. In creating a political formation, entity, or group, one must appeal to a “they” against which “we” are to be compared.
Schmitt writes, The enemy is not merely any competitor or just any partner of a conflict in general. He is also not the private adversary whom one hates. An enemy exists only when, at least potentially, one fighting collectivity of people confronts a similar collectivity.
51 The NRA ad is a political ad, and the Qdrops are seeding alienation from the existing body politic—they call for the formation of a political alliance, by opposing it to another political group, a “they” against which “we” are to be compared. Such propaganda calls on the friend-enemy distinction to simultaneously resolve the “us” and “them” pronouns. But resolution does not feel effortful. We don’t need the referents for the NRA’s antecedentless “they” to be named because, to the extent that there is any definite reference, we can effortlessly figure out who “they” is supposed to refer to: it is, like “He Who Shall Not Be Named,”52 the common enemy.
Recall here the discussion of the term “critical race theory” in chapter 3.
We suggested there that this term, in its right-wing usage, lacks clear reference to particular concepts or practices, but is rather a role in a narrative, like the flag of the enemy. It is meant to evoke a narrative of existential threat. The fact that the term “critical race theory” works so well to evoke existential threat evidences the fact that it is not causing the intended audience to worry excessively about the exact definition of the term, even though there are, as yet, no psycholinguistic studies of whether use of the term causes processing difficulty. The “they” of the NRA ad, or of the Qdrop examples, has a similar function to “critical race theory.” Again, despite its unclear reference, it does not appear to be hard to swallow. The pronoun marks only otherness, identifying a role in an easily recognizable narrative frame as being filled by an indistinct horde of out-group foes. In this narrative frame, what is under attack is a form of life that is presumed to be shared by a collective that includes the speaker or writer and other right-minded people who wish to protect American values and practices from liberal extremists. This collective is the “we” of the third paragraph of the NRA ad. To the extent that “we” has any clear reference, it is partly defined by opposition to “they.” By positing a “they,” “we” are united. But just like “they,” it is perhaps better to think of “we” not as a specific grouping but as a role in a story: the good guys. The audience is presupposed to consist of good guys.
The use of such us-them labels helps build group cohesion in two ways.
First, the mere use of a pronoun with no explicit referent presupposes collective attunement.
Secondly, the vile properties ascribed to the outgroup and the positive motives or beliefs ascribed to the in-group reaffirm the intended audience’s feeling that the group they are part of is the right one.
For you to have the same sets of individuals in mind as Loesch does, it must be exceedingly obvious what those sets are. If you’re hooked by the NRA strategy, attuned to the right things, then you’ll feel like your attention is on the same things as Loesch, and you won’t worry too much about the details, for instance about exactly which people are in the extension of they and which aren’t. The NRA they is a maximal group of bad actors, the other, if you will, and whatever groups you hate must be subsumed within that maximal group. The group is understood as homogeneous, as acting in concert to tear down everything you love, with differences within that group of others minimized or seen as irrelevant. Likewise, the in-group is homogeneous: if you’re one of us, that’s all we need to know.
If you don’t or won’t share Loesch’s values, the ad will probably make you bristle. Assuming common ground is a way of welcoming those who can accommodate into the in-group, but it can be just as powerful in alienating outsiders. In this, Loesch’s style of speech is welcoming for its target market, but it is also divisive, separating those with her from those against her. It is a paradigm case of making friends by identifying common enemies. And the presence of common enemies, we speculate, is such a strongly motivating type of common ground that it is ideal for directing attention. A focus on a common enemy can distract from inconvenient details, like the fact that Q never evidences any coherent political philosophy, or the fact that the NRA represents large corporations that have a vested interest in particular political policies, and can drive attention to a commonality between in-group members that can help forge a political movement, or a social identity. In this, there is a surprising confluence between the bonding effects of in-group uses of slurs to label a common enemy (though not reclaimed uses of slurs—see chapter 11), and the use of us-them pronominal distinctions.
This is a surprising fact about language, for at first blush it would be difficult to identify linguistic categories that have more obviously different functions than pronouns and slurs.
Politics, as they say, makes strange bedfellows.
1. Hill, The Everyday Language of White Racism, 47.
2. Walter Cronkite is quoted in Senate Resolution 68, “To Establish a Select Senate Committee on Technology and the Human Environment,” 404.
3. Mullen, “The Propaganda Model after 20 Years: Interview with Edward S. Herman and Noam Chomsky,” 12. As the interview makes clear, this is in line with what Herman and Chomsky had said in their milestone work on propaganda, Manufacturing Consent. However, in Manufacturing Consent, Herman and Chomsky do not use the terminology of presupposition theory, but rather talk of “assumptions.” For example, in discussing the framing of the Vietnam war in terms of US victimhood justifying military action as retribution rather than as being inherently aggressive (in the introduction to the 2011 edition of the book), Chomsky and Herman state, “It is compelling evidence of the propaganda service of the mainstream media that throughout the war they accepted this basic propaganda assumption of the war managers, and from that era up to today, we have never found a mainstream editorial or news report that characterized the U.S. war against Vietnam, and then all of Indochina, as a case of aggression” (Manufacturing Consent, xxx). At a more general level, the developments in the current volume are compatible with the themes explored in Manufacturing Consent, which excels in providing historical evidence for the systematicity of processes that have yielded a relatively high degree of political conformity in the United States, conformity as regards central tenets of foreign policy, largescale military actions, and fundamental organization of the economy.
Although we do not pursue this agenda here, it would be natural to study in our own model Herman and Chomsky’s seminal analyses of the framings used by politicians and mainstream media, the way in which much mainstream media tends to fall into line with regard to certain types of government policy, and, resulting from these, the ways in which public attention is effectively controlled, or at least limited. Framing, already discussed in chapter 2, will again be discussed in section 4 of the current chapter. The broader processes of alignment they consider would, in our terms, be mechanisms by which collective harmony is reached, with journalists and editors both individually and collectively dispositionally attuned to accept the easy story that governmental organizations, advertisers, and the businesses that own the presses provide, and to avoid the dissonance of reporting on stories or angles that would run counter to their own interests. Attention is an important part of our account of harmony, but we stop far short of attempting to describe in detail complex collective attentional processes at a national level as Herman and Chomsky do.
4. Presupposition and accommodation play a central role in, e.g., Sbisà, “Ideology and the Persuasive Use of Presupposition”; Langton and West, “Scorekeeping in a Pornographic Language Game”; Langton, “Blocking as Counter-Speech”; McGowan, Just Words.
5. Loftus and Palmer, “Reconstruction of Automobile Destruction.” 6. The headlight example is from Loftus and Zanni, “Eyewitness Testimony.” There have been studies showing that the issues are complex. Zanni and Offermann, “Eyewitness Testimony,” showed that the definite article did not lead to higher false memories for subjects high on a neuroticism scale, while Singer and Spear, “Cleft Constructions,” show that at least in cases involving recent textual antecedents, subjects are as sensitive to erroneous information presented through some types of presupposition trigger as they are to information presented more directly.
7. While one can respond “I didn’t” to any of these three questions, this is usually taken by presupposition theorists to be a denial of a presupposition rather than an answer to what has been asked. Note here that all of the following are at least mildly infelicitous: “#I don’t know whether you paid, but why/when/how did you pay?” (The fact that “If you paid, why/when/how did you pay?” sounds better is exactly what is predicted on standard analyses of presupposition, since this is understood as asking a question that only needs to be answered if the assumptions in the antecedent of the conditional is true.) Note that although questions are, as stated, standardly analyzed as carrying presuppositions, the evidence is indirect, since standard tests for presupposition rely on projection tests, which cannot easily be applied to questions. However, if indirect questions like the whyquestion in “Mary wondered why you paid” are used, projection tests (discussed below in the main text) become easier. An argument based on this methodology would run as follows: (i) consider the pair “Mary wondered why you paid” and “Mary didn’t wonder why you paid”; (ii) observe that both carry an implication that the addressee paid (and similarly for when and how); so (iii) this behavior under negation is strong evidence for the presence of a presupposition carried by the question.
8. The standard argument that temporal subordinate clauses carry presuppositions depends again (for this is standard in presupposition theory) on projection tests. In the case of “The shopkeeper smiled before / after / when / until / while you paid,” the standard approach is to consider embedding the entire sentence in an environment like that provided by a possibility modal. We then construct “Perhaps the shopkeeper smiled before / after / when / until / while you paid,” and note that while it does not entail that the shopkeeper smiled, it still carries the implication that the addressee paid. This is evidence that while it is not presupposed that the shopkeeper smiled, it is presupposed that the addressee paid. Similarly, we can embed in the antecedent of a conditional: “If the shopkeeper smiled before / after / when / until / while you paid, that makes me suspicious.” Again, the implication that the addressee paid survives, evidencing the presupposition. The presuppositions of temporal clauses have been recognized for a half-century, a thorough early discussion being found in the dissertation of Heinemaki, “Semantics of English Temporal Connectives.” For a somewhat more recent discussion of presuppositions of a subset of temporal clauses, see, e.g., Beaver and Condoravdi, “A Uniform Analysis of ‘Before’ and ‘After.’ ” 9. Our evidence for differences in frequency of such constructions comes from Google Ngram searches. These were performed at https://books.google.com/ngrams. A sample search is generated by the following query: https://tinyurl.com/mr37m78c. The corpus used for these searches is so large that the number of words in it is the same order of magnitude as a person might be exposed to in an entire lifetime, hundreds of millions, we estimate.
10. The idea that presuppositions are taken for granted and uncontroversial is discussed in Stalnaker, “Pragmatic Presuppositions,” in Stalnaker, Context and Content, 135–48. Taking Stalnaker’s work as a starting point, Scott Soames built uncontroversiality into his definition of “utterance presupposition”; see his “How Presuppositions Are Inherited.” 11. Steve Benen, “The Problem with Trump’s Proof That He’s Helped the Black Community,” Maddowblog, MSNBC, June 3, 2020, accessed March 1, 2023, https://www.msnbc.com/rachelmaddow-show/problem-trump-s-proof-he-s-helped-black-com munity-n1223426.
12. Trump’s tweet of September 12, 2020, is cited by Michael S. Schmidt, “Comey and McCabe, Who Infuriated Trump, Both Faced Intensive I.R.S. Audits,” New York Times, July 6, 2022, accessed March 1, 2023, https://www.nytimes.com/2022/07/06/us/politics/comey-mccabe-irs-audits.html.
13. Davidson, “Three Varieties of Knowledge,” 158. 14. Davidson’s shift of terminology is explicit in his 1984 Lindley lecture, appearing in Davidson, “Expressing Evaluations.” 15.
Lewis, “Scorekeeping in a Language Game.” 16. The description of presuppositions as nonnegotiable is found, for example, in AnderBois, Brasoveanu, and Henderson, “Crossing the Appositive/At-Issue Meaning Boundary,” and in two works by Sarah Murray: “Evidentiality and the Structure of Speech Acts,” and “Varieties of Update.” 17. Loftus, “Leading Questions and the Eyewitness Report,” 571–72.
18. Tversky and Kahneman, “The Framing of Decisions,” 453.
19. Chong and Druckman, “Framing Theory.” For an application of valence framing to electoral issues in political psychology, see, e.g., Bizer and Petty, “How We Conceptualize Our Attitudes Matters.” 20. Kahneman and Tversky, “Choices, Values and Frames.” The context and framing conditions are presented using the text from the original study, but with our headings and numbering. The design is between-subjects, so subjects had only seen one of the two framings when making their choice.
21. Of particular note are the reproductions in the Many Labs Replication Project, which used materials very close to the originals, and the 1998 meta-analysis of Kühberger, recently updated and extended by Steiger and Kühberger. Overall, these studies have shown that the effects are robust in close replications of the original, although the effect size is smaller than found in the original study, and somewhat robust in studies that vary significantly from the original. See Klein et al., “Investigating Variation in Replicability”; Kühberger, “The Influence of Framing on Risky Decisions”; Steiger and Kühberger, “A Meta-analytic Re-appraisal of the Framing Effect.” 22. Assuming that (negative) utility is proportional to the number of deaths, the expected utility of option (i), a one-third probability of 0 dying (i.e., 600 saved), and a two-thirds probability of 600 dying (0 saved) gives ⅓ * 0 + ⅔ * 600 = 400 deaths expected (200 expected to be saved). This obviously gives the same as the expected utility of option (ii), which is presented as a guaranteed 400 deaths (again, 200 saved).
23. The ideas emerged in Bernoulli’s solution to the St. Petersburg lottery problem: Bernoulli, “Exposition of a New Theory on the Measurement of Risk.” Technically, the property of classical decision theory that Kahneman and Tversky are arguing against is the property of decision problems being invariant under addition and subtraction of a fixed sum to all options, a property that is already lacking under Bernoulli’s analysis.
24. We note that our characterization of the problem suggests a hypothesis that, to our knowledge, has not yet been directly tested in the large literature on this problem: perhaps what is discounted is the numeric values themselves, not the utilities of the different options.
That is, perhaps people don’t merely have skewed perceptions of the value of 600 deaths (or lives) relative to 400 deaths (or lives), but rather have skewed perceptions of the quantity 600 relative to the quantity 400. This together with a standardly observed additional effect of negativity dominance, that whatever people feel bad about has heightened salience or significance relative to what they feel good about, would yield the standardly observed S-curve of value, steeper for negatives than positives, at the heart of Kahneman and Tversky’s model. If this were true, then in an affectively fairly neutral issue like how far away an unnamed object is, we should still see similar effects. Thus it would be predicted that option (i) will be preferred in the following judgment: “Which is further away: (i) something definitely 400 miles away, or (ii) something with a ⅓ chance of being 0 miles away, and a ⅔ chance of being 600 miles away?” We speculate that if there is any such effect, it is smaller than the effects Kahneman and Tversky hypothesize.
25. One of the most influential discussions of the crucial features of the Kahneman and Tversky framing effect, Levin, Schneider, and Gaeth, “All Frames Are Not Created Equal,” does not mention presupposition as a possibly relevant factor, and neither do the metastudies cited earlier (see fn. 21 on p. 194). Moxey and Keren (“Mechanisms Underlying Linguistic Framing Effects”) discuss the Asian Disease problem in the context of an account of presupposition denial, but do not explicitly discuss the role of presupposition in the Asian Disease problem. Note that we must emphasize that we are not claiming that presuppositional setting of the reference point is the only factor creating what Kahneman and Tversky term the “framing effect.” This hypothesis would be difficult to maintain given that some close variants of the experiments, including some used in Kahneman and Tversky’s classic studies, use wordings that do not have such strong reference setting presuppositions, e.g., wordings that talk about the number of people who are alive rather than the number of people saved. 26. Headline, Foreign Affairs, May 7, 2020, accessed March 1, 2023, https://www.foreignaffairs.com/articles/2020-05-07/pandemic-wont-save-climate. The subhead, tellingly, is “Don’t Expect the Clear Skies to Last.” 27. Headline, Automobile, September 17, 2019, accessed March 1, 2023, https://www.automobilemag.com/news/general-motors-gm-uaw-strike-plants-cadillac/.
28. Headline, WBEZ Chicago, October 29, 2020, accessed February 23, 2023, https://www.wbez.org/stories/pilsen-residents-say-landmark-designation-might-save-building s-but-itwont-stop-gentrification/31fd3199-a5ca-4c72-9c45-e2b54a64feb1.
29. For example, Camp explains, “An author or narrator also presents the facts of the fictional world from a certain perspective, which she expects us to share” (“Two Varieties of Literary Imagination,” 117, emphasis added). On the subject of slurs, Camp writes, “The automatic nature of semantic understanding in general, along with the fact that perspectives are intuitive cognitive structures only partially under conscious control, means that simply hearing a slur activates an associated perspective in the mind of a linguistically and culturally competent hearer” (“Slurring Perspectives,” 343).
30. Beaver, “When Variables Don’t Vary Enough.” 31. Hemingway, “The Short Happy Life of Francis Macomber,” 263.
32. In a situation where someone eavesdrops on a conversation that they are not intended to be a full party to, the hearer has no choice but to accommodate anything and everything presupposed by the speakers, or at least accommodate that the speakers take the context to satisfy the presuppositions. An everyday situation where this would occur would be overhearing a conversation in an elevator, and hence David Beaver and Henk Zeevat refer to the process the hearer must go through as “elevator accommodation” (Beaver and Zeevat, “Presupposition and Accommodation,” 12–13). One can accommodate unresolved presuppositions in an elevator that one could not stomach in a normal conversation.
33. Tonhauser et al., “Toward a Taxonomy of Projective Content.” 34. Natalie J. Ring, @HistoryCounts, Twitter, October 2, 2020, accessed February 19, 2023, https://twitter.com/historycounts/status/1312092169612726273?lang=eu.
35. Ian Prasad Philbrick, “A ‘Great Gatsby’ Quote Takes on New Resonance,” New York Times, October 7, 2020, accessed March 1, 2023, https://www.nytimes.com/2020/10/07/books/great-gatsbyquote-trump.html.
36. Dana Loesch, “The Violence of Lies,” National Rifle Association ad, December 12, 2018, accessed October 12, 2020, https://www.youtube.com/watch?v=169zQ1g-Ul0.
37. QAnon post is numbered 184 and dated November 21, 2017, https://qanon.news/Q#.
38. QAnon post is numbered 563 and dated January 19, 2018, https://qanon.news/Q#.
39. In briefly surveying the first lines of several hundred novels in English, mostly from wellknown authors, we found no prominent examples with completely antecedentless female pronouns.
We can only speculate as to why antecedentless male pronouns are apparently much more common, but will resist the temptation to do so.
40. Sabatini, Scaramouche, 3. 41. Woolf, Orlando, 13.
42. Hemingway, The Old Man and the Sea, 5.
43. Ōe, “The Day He Himself Shall Wipe My Tears Away,” 1.
44. Murakami, “Nausea 1979,” 151.
45. We take it that the opening “It” is an expletive pronoun in a cleft structure of the form “it was X that Y” rather than an unresolved pronoun, though it’s not clear that there’s any fact of the matter.
Similar comments apply to Richard Powers’s opening line. 46. Auster, City of Glass, 7.
47. Powers, Galatea 2.2, 3.
48. Morrison, Paradise, 3.
49. The device of opening a novel so as to present the impression of common ground when in fact it is lacking can also be achieved using expressions other than unresolved pronouns.
In each of the following cases, we can think of the missing information in terms of an implicit question: “You better not never tell nobody but God” (Alice Walker, The Color Purple, 1).
Question: Tell what?
“In the late summer of that year we lived in a house in a village that looked across the river and the plain to the mountains” (Ernest Hemingway, A Farewell to Arms, 3). Question: Which year is “that year”?
“Many years later, as he faced the firing squad, Colonel Aureliano Buendía was to remember that distant afternoon when his father took him to discover ice” (Gabriel García Márquez, One Hundred Years of Solitude, 1). Question: Later than what?
50. O’Connor, “Everything That Rises Must Converge,” 10. Fittingly, the first word of the story is a pronoun. Although the referent is revealed later in the same sentence, it’s only by linkage to Julian, who himself has not been previously introduced: “Her doctor had told Julian’s mother that she must lose twenty pounds on account of her blood pressure, so on Wednesday nights Julian had to take her downtown on the bus for a reducing class at the Y.” 51. Schmitt, The Concept of the Political, 28.
52. The phrase “He Who Shall Not Be Named” entered popular culture from J. K. Rowling’s Harry Potter novels, where it is used to refer to the archenemy, Voldemort. In the magical world of the novels, words have the power to invoke more than shared reference, so the reader is being asked to imagine that there might be very special reasons for using a pronoun rather than a name.
CHAPTER FIVE Presupposing Practice If the practices of the members of the same group or class are more and better harmonized than the agents know or wish, it is because, as Leibniz puts it, “following only [his] own laws,” each “nonetheless agrees with the other.” The habitus is precisely this immanent law … laid down in each agent by his earliest upbringing, which is the precondition not only for the co-ordination of practices but also for practices of co-ordination, since the corrections and adjustments the agents themselves consciously carry out presuppose their mastery of a common code.
—PIERRE BOURDIEU1 OUR JOB IN THIS CHAPTER is to provide a theoretical description of presupposition in terms of resonance that can illuminate its role in persuasion specifically and in the maintenance and transmission of ideology more generally. That burden of explanation will not be carried by presupposition alone, but through a combination of presupposition and accommodation. An action carries presuppositions, and it is the accommodation of these presuppositions that results in many cases of unobvious transmission, maintenance, and even creation of ideology.
However, we defer detailed discussion of accommodation to chapter 6, where accommodation is tied to the broad processes of psychological and social alignment introduced in chapter 3, harmonization.
The work of this chapter is at once conservative and technically radical. We analyze a standard concept, namely presupposition, in a novel way, developing an approach that differs from others in the huge prior literature on the topic in at least two respects. First, whereas presupposition is standardly categorical, so that something either is or is not presupposed, for us presupposition is a matter of degree, as is the case for all resonances. Second, whereas presuppositions are standardly taken to be propositions, for us presuppositions are resonances, which may or may not be propositional.
It is true that various prior theories extend presupposition to other potential objects beyond propositions, such as questions under discussion.
But these are treated as extensions of a basic notion of propositional presupposition. On our model, by contrast, presupposition is not, in the first instance, a relation to propositions at all.
Each word belongs to a speech practice (usually more than one). Using a word is a manifestation of a speech practice. Manifesting a speech practice by using a word simultaneously presupposes, among other things, that very practice. On our account, then, it might be said that the most basic thing presupposed is a practice. Presupposing practices, including speech practices, is a core example of presupposition, and not an afterthought.
Both the gradience and the nonpropositionality of presuppositions have immediate consequences. In models where presupposition is categorical, there is little room for nuance, and it is hard to make sense of the observation that some presuppositions project more vigorously than others.2 This emerges naturally in a gradient theory of presupposition. In models where presuppositions are propositional, accommodation involves change in beliefs.
But if what is presupposed is a practice or an emotion, then accommodation can become a change in behavior or emotion.
We begin this chapter by rehearsing some of the properties of the model introduced in part I and by showing what is missing. Then, in section 5.2, we plug the gap with a probabilistic account of presupposition that straightforwardly extends our probabilistic theory of resonance, setting that probabilistic proposal against a background of existing accounts of presupposition. In the remaining sections of the chapter, we first discuss various consequences of adopting the new model of presupposition (sections 5.3–5.5), and then, in section 5.6, segue into the discussion of accommodation that follows in chapter 6.
5.1. What More Do We Need from a Theory of Resonance?
Instead of thinking of meaning in terms of content, as if words were little packages with the meaning wrapped up tightly inside, we have suggested using a metaphor of resonance, a metaphor already employed in fields such as social-movement theory and advertising. On this view, the communicative function of language is not to transfer packets of information, but to establish connection. We went on to make a terminological distinction between attunement and resonance. Whereas things resonate, only animate beings can have attunements, which involve their dispositions, attitudes, or emotions. Things resonate by virtue of people being attuned to them. That is, something carries resonance by virtue of people recognizing or creating patterns of connection between that thing and anything else, including their own behavior or emotions. For example, a favorite cup may carry a cocoa resonance and various associated feelings because that’s what you like drinking from it: you are attuned to a relationship (which you created) between the cup and cocoa, and similar cups may not resonate with you as strongly. However, the most important resonances for us are those attached not to individual habits, but to community practices. Practices of drinking cocoa may carry resonances of cold winter nights or of spiritual connection to ancestors, depending on the community that practices them.
A falling apple recalls Newton, where a bite of the apple evokes Eve, Snow White, or Steve Jobs. But what of specifically communicative actions or, more generally, interactions? They are special because resonance, on our view, is not merely incidental for a communicative interaction: it’s the whole point. To see a practice as communicative is to understand it as having a function of creating a dependency between the attunements of individuals, to produce connection.
Approaching communication in terms of the connection created by collective attunement provides a setting in which to study nonideal communicative practices, for example, lying and insulting. Both lying and insulting yield a mixture of what we might term positive and negative connection, positive because they create common attunement as to what the speaker is publicly committed to, and negative because both tend to produce misalignments of attitudinal, emotional, and dispositional attunement. Indeed, they not only tend to produce such a mix of alignment and misalignment, but they are also intended to produce them.
In developing the resonance-based framework, we simplified in many ways. For example, we set aside the details of how, for example, emotional and similarity-based resonances of actions might be studied, and we focused on what we termed associative resonance, which measures the extent to which some feature of context co-occurs with an action, that is, the extent to which that feature tends to be present in the contexts in which the action takes place. We showed how this idea is related to a prominent approach in the literature, from the philosopher David Kaplan, who identifies the meaning of an expressive action like saying “ouch” with whatever is common to its contexts of use.
The notion of context we need is inherently “shifty,” in that small changes of contextual anchoring can dramatically change what it is appropriate to say. It does not suffice for a felicitous use of “ouch” that there is, as Dickens’s Mrs. Gradgrind put it, “a pain somewhere in the room”3 ; the pain must be the speaker’s. That is, we need a notion of the context of an action such that we can differentiate between the local situation of that action and other nonlocal situations. Such a notion of context would enable us to say, for example, what the (local) state and (nonlocal) history of the speaker is like for any particular instance of the practice of saying “ouch.” To take another example, consider the telling of a well-known, off-color joke. The local situation includes immediate properties of the particular joke-teller and audience, while the nonlocal situation includes not only more distant histories of those individuals, but also more general properties, like any tendency for tellers of this joke and their audiences to be prejudiced. In looking at the general properties of a practice, we should expect features of interest to be such things as the time in which it is being instantiated, and the attunements, group identities, and social relationships of the interactants. Thus, features of the context of an action must be indexed to the action they are being related to as regards time, place, and the identities of actors involved in various roles.
While the need for indexicality is a common feature of our proposal and Kaplan’s, a big difference, following a pattern that will now have become familiar to our readers, is that we suggested that the relationship between actions and features should not be black and white.
That is, instead of thinking of a dichotomous notion of meaning whereby a feature of context either is or is not part of the Kaplanian meaning of an action, we suggested a scalar notion of resonance. On the model we proposed, something is an associative resonance of an action to the extent that it is more likely given that the action occurred than if it had not occurred.
In chapter 2, we introduced collective attunement as a variant of the standard notion of common ground, and in chapters 2 and 3, we showed how many social and political phenomena might be modeled in terms of mechanisms that produce a shift of individual or collective attunement. So, political persuasion consists, in part, in the use of resonant messages that lead people to harmonize around attunements activated by the politician. In introducing presupposition in chapter 4, we presented a number of cases suggesting its centrality for shifting attitudes. So political persuasion must consist, in part, in the use of messages that presuppose the ideas and perspectives that the politician favors. It would appear, then, that we are faced with an embarrassment of riches, two distinct approaches to modeling political persuasion and ideological shift more generally. The goal of the current chapter is to show that these two approaches are not merely complementary, but describe different aspects of the same broader process of cultural transmission.
Here, let us return to a big theme of this book, hustle, which in turn can be related to the well-established psychological literature on persuasion (see section 2.6). As we defined it in chapter 2, persuasion consists in shifting someone’s attunements in a way that would have been in tension with their prior attunements. One strategy for persuading is to use a straight-talk strategy, overtly suggesting reasons for adopting a new attunement.
However, doing so is difficult, because persuasion is needed precisely when the new attunement is in tension with prior attunements, producing recalcitrance. Literature in psychology and communication studies suggests that, instead, persuasion often works through indirect routes that induce change of attitude through methods other than directly explicating reasons for the change. Whereas in deliberative presentation of reasons the intentions behind the argument are explicit at every step of the way, this is not the case for these alternative, indirect forms of persuasion.
To take a famous example, an ad campaign that seeks to convince people to buy a certain brand of cigarettes through presentation of a rugged cowboy smoking does not make manifest the intention to convince in this way; it does not make manifest that you should smoke that brand because that will imbue you with characteristics of the cowboy. Quite generally, indirect routes to persuasion are not based only on straight talk, but involve hustle.
That is, indirect persuasion involves at least some communicative actions in which the persuader is not overtly manifesting their intention to change someone’s mind, even if it is clear that the persuader is more broadly engaged in such an act. It is obvious that a politician holding a baby in the crowd is hoping to attract more votes, but as soon as one cynically recognizes the act of holding the baby as motivated by electoral considerations rather than prosocial loving tenderness, the persuasive value of the act is diminished.
When we see a cowboy smoking or a politician holding a baby, we are not seeing assertions; we are just seeing people performing practices. The performance is crafted so as to appear incidental, as if we just happen to have caught the individual in the act of smoking or cradling, an act that is not portrayed as exceptional for the individual as normal. In a sense that we will attempt to make clearer in this chapter, the very act of smoking or cradling presupposes the appropriacy and structure of those practices, as well as the normalcy of the cowboy or politician performing them. We are not asked to see the acts as normal; we are simply presented with the acts, and their normalcy is presupposed. Thus, nobody tells us to take on board the gentle resonances of the scenes we observe; we simply do so as part of our regular perceptual function of becoming better attuned to the world around us.
We will not make too much of the role of presupposition in cases like the smoking cowboy or the cradling politician. Invoking presupposition here perhaps adds little analytic understanding to these particular cases beyond what the resonance framework might already offer. But the examples serve to focus attention once again on a repeated theme of this volume, the wide range of ways that persuasion operates without assertion.
Nonassertive persuasion is central to propaganda and to political hustle more generally, and presupposition is one of the most well-studied mechanisms by which people change their attitudes independently of assertion. The examples we presented in chapter 4 were chosen to illustrate the function of presupposition in indirect persuasion, often in the context of speech acts such as questions in which no information is asserted at all.
To see why the model we introduced in part I of the book does not suffice to do the theoretical work we need, note two interrelated properties of associative resonance. First, it is timeless. The associative resonances of the practice of ordering an espresso at a cafe include the speaker needing a boost, the addressee being a barista, the speaker being perhaps a little sleepy before the act, the speaker receiving a small cup of concentrated liquid caffeine, and, hopefully, the speaker being somewhat livelier afterward. Some of these are states prior to the act, some after, and some overlap. The notion of associative resonance does not distinguish. The second property is closely related: our correlational notion of associative resonance is noncausal, so the mere fact that a practice has a certain resonance does not tell you whether that resonance should be thought of as a cause of the action, an effect of the action, or merely something that happens to co-occur with the action for independent reasons.
The work that is needed in this chapter, what is missing so far from our account of associative resonance, hinges on the oft-repeated fact that correlation is not causation. What we need to do is to distinguish the things that a communicative action is correlated with from those it causes. The features of context with which a communicative act is merely correlated will be what we identify as the presuppositions (or, more fully, the presuppositional resonances of the act). Correspondingly, the things that a type of act tends to reliably cause are the equivalent of what in speech act theory is termed the illocutionary and perlocutionary effects of the act, although these will not be a focus of the chapter. It is the presuppositions of the act that drive the process of accommodation, while the standardly recognized effects of the act are manifested through a separate process, roughly what is standardly termed uptake.
Having located presupposition within the resonance framework, we will spend the remainder of this chapter tying the resonances of interactions more tightly to the attunements of the interactants and studying the properties of the model of presupposition that results.
Specifically, we argue that the model better accounts for the complex inferential behavior that prior empirical work on presupposition has revealed than do existing models of presupposition, and furthermore that the scalar nature of the resonance framework provides a novel perspective on a question that has played an important role in prior literature on presupposition, the question of whether presuppositions are conventionalized. 5.2.
Presuppositional Resonance We now build both on standard notions of presupposition and on the cluster of concepts that J. L. Austin referred to as conditions on the felicity of speech acts, sometimes called preconditions in later literature, although for him preparatory conditions were a subset of a larger set that might be generically referred to as felicity conditions.
4 Before proceeding, a short recap of relevant parts of the long history of work on presupposition is in order.
In the modern era, work on presupposition dates back to Gottlob Frege and Peter Strawson.
They suggested the idea of presuppositions as necessary conditions for a sentence to be meaningful. According to this view, “I have to get my sister from the airport” would be meaningless absent the existence of a sister. Of the many later evolutions of the idea, two, both originating in the early 1970s, are especially relevant here. Robert Stalnaker’s pragmatic presupposition centers on the notion of common ground that we have already discussed, and we will return to it shortly. The other is the notion of presupposition in dynamic semantic models, first suggested by Lauri Karttunen. Karttunen suggested that we think of sentential clauses as updating a context, with contexts being sets of propositions. A presupposition is then a proposition that must be in the prior context in order for a sentence to be used to perform an update. For example, only contexts containing the proposition that the speaker has a sister could be updated with “I have to get my sister from the airport.” So, whereas in the Frege/Strawson model there is no temporal priority between what is presupposed and the content expressed, the dynamic model has an explicit procedural interpretation; presuppositions, as might be anticipated on etymological grounds, are prior conditions.
The bulk of formal work on presupposition focuses on the presuppositions of sentences or utterances of sentences, rather than on presuppositions associated with arbitrary speech acts. The model we will propose can be seen as a development of this line of work, but it also shares much in common with Austin’s notion of the preconditions of a speech act.
The preconditions of a speech act include both propositions that a speaker and hearers must believe in order for a speech act to be appropriate, and also social requirements, for example, the condition that the speaker is legally or otherwise empowered to perform the act in question. As Austin writes, making the connection to presupposition explicit: We might say that the formula “I do” presupposes lots of things: if these are not satisfied the formula is unhappy, void: it does not succeed in being a contract when the reference fails … any more than [the statement that “John’s children are all bald” made when John has no children] succeeds in being a statement.5 Putting Austin’s observation into our terms, the act of marrying two people by declaration, and the act of referring using a possessive description (“John’s children”) are alike: they are both communicative practices. Each practice is shaped by a history of use—the extension of the practice, as we termed it in chapter 2, that has yielded conventional patterns of interactive behavior. The pattern involves certain properties that contexts have when the practice is used, and certain ways that performance of the practice changes the context. In one case, when the context involves a person playing a certain role within the larger practice of a marriage ceremony, the effect of performing the subpractice of saying “I do” is to advance the ceremony in such a way as to enable the enactment of the marriage itself shortly thereafter. In the other case, when the context makes some guy called “John” identifiable as such, and if this guy has children in the (possibly separate) context under discussion, the effect of the act is to make those children a topic of conversation. Further things can then be said of them, for example, that they are bald, or that they need to be picked up from the airport. What is needed is an account of the presuppositions of communicative practices that is general enough to be usefully applied in both cases.
Stepping away from the historical backdrop, our immediate goal is to give a theory of the presuppositions of communicative practices that accounts for a range of properties of presuppositions, including (i) their ability to leak out of embedded contexts, that is, their projectivity; (ii) the fact that they seem to circumvent the ordinary process of information update, often being accommodated without attracting the degree of reflection, awareness, or public deniability that we might expect for standard uptake of a speech act; and (iii) their iceberg-like combination of vastness and near invisibility. This last property of presuppositions derives both from the fact that presuppositions help determine our way of seeing the world (and it’s hard to see a telescope while looking through it), and from the fact that they are not usually a focus of attention in a communicative interaction.6 We seek to explain these properties.
The approach we now develop hinges on the solution to a conceptual problem alluded to in the first section of this chapter, closely related to issues raised in the final section of chapter 1. Put bluntly, we must deal with the fact that the notion of resonance does not distinguish between the background of an action and its effects. What we must now pinpoint, then, is the background of a practice, the things that typically populate the context of the actions comprising an instantiation of that practice before they take place, and without which the actions would not even function as an exemplar of the practice to which it belongs.
Our recipe for extracting the background is roughly this: start with all the correlational resonances of the practice and remove any resonances that are effects of the actions. At least, that would be the recipe if the resonances were simply a set of features. However, there is no set of features that are the resonances of an action. Resonance is a matter of degree. So, what we will do is subtract the degree to which some feature is an effect of an action from the degree to which something is a resonance. It follows that what we need is a measure of the degree to which something is an effect of an action.
Let us simplify drastically by leaving unanalyzed the nature of causality.
Having made this enormous shortcut, the problem turns out to be easy: the degree to which something is an effect can just be identified with the probability that a feature is caused by the action. We’ll call this the effect probability. We can now define a dynamic notion of the presuppositions of practices, which we will term the presuppositional resonance of a practice for a feature, as the difference between the associative resonance and the effect probability.
Our theory of presupposition in terms of associative resonance, itself defined in chapter 1, is then given by these three equations relating a practice to a feature: 1. Associative resonance = p(feature| instantiation of practice) − p(feature) 2. Effect probability = p(instantiation of practice-caused feature) 3. Presuppositional resonance = Associative resonance − Effect probability These simple equations give us a numeric measure of the extent to which a feature is a presuppositional resonance of a practice, that is: Presuppositional resonance = p(feature| instantiation of practice) − p(feature) − p(instantiation of practice-caused feature) We have defined the presuppositions of a practice, but not the presuppositions of an action, that is, a token or instantiation of a practice.
This is complicated by the fact that an action can simultaneously instantiate multiple practices, as when uttering “I would like some more cake” simultaneously instantiates the practice of speaking English, the practice of combining a subject and a verb to make a sentence, the practice of asserting, and the practice of requesting. For the moment, let us characterize the presuppositions of actions that instantiate multiple practices in this way as being exactly the presuppositions of the practices that are instantiated (combined using standard probability theory).
In thus characterizing the presuppositions of individual actions, we are temporarily sidestepping two difficult and important issues that have already come up in our discussion of the resonances of an action in a community of practice (section 2.5). First, the question of whether an action instantiates a practice is itself a biased question, assuming a dichotomy.
Actually, it’s a matter of degree: an action may be taken to instantiate a practice even if it differs substantially from prior examples of a practice, or it may be taken to partially instantiate a practice, and in either case it is far from clear what the presuppositions should be. Second, the question of whether an action instantiates a practice is one that reasonable people may disagree on. Who decides? Is it a matter of the speaker’s intentions? Do other interlocutors or an audience get a say in the matter? An observing linguist or anthropologist?
We will not address this issue in this chapter, save to note that it gets to the heart of why many theorists have adopted practice-based accounts in the first place, namely to produce a theory of action that is not centered on the intentions of actors. Our account of presupposition is inspired by the dynamic accounts of presupposition and speech act models we’ve discussed. Despite this inspiration, our definition of the numeric measure of presuppositional resonance will perhaps be seen by practitioners in that domain as a radical departure (perhaps even an overly radical departure), since it doesn’t closely resemble any of the standard definitions of presupposition in the literature. Most obviously, in being scalar, it departs both from both standard dynamic notions of presupposition and the standard notion of a precondition of a speech act. By contrast, the standard notions are categorical: it’s taken to be a yes/no question whether a given utterance has a certain presupposition, and similarly for the notion of the precondition of a speech act. As in our discussion of collective attunement in chapter 2, a comparison with the more standard approach might be made by introducing an approximation to our scalar notion. Thus if we set a threshold probability, we could say that a feature of context is categorically presupposed by a practice if its presuppositional resonance reaches the threshold.
There’s another somewhat related way that our account differs from many prior accounts: on the view we have proposed, presuppositions are not necessary conditions for an act, but tendencies. It may be of solace to any conservatively minded readers that necessary conditions for an action to take place will always have positive presuppositional resonance.
Well, almost always! In fact, there’s an interesting type of case where something that on traditional accounts would come out as a necessary condition, and hence as a presupposition, has zero presuppositional resonance. That is the case of tautologies. On standard accounts, an utterance of “The cheese factory is smelly” would presuppose that seventeen is a positive number, simply because the latter is a tautology, and hence “necessary” in classical logics. Our model does not have this property. The fact that seventeen is a positive number, although it may be true in every context, is not a presuppositional resonance of any practice. This follows directly from the fact that it is not a resonance at all: its probability does not get boosted by any action, because its probability is always 1.
In chapter 2, we took the reader through an account of common ground as collective attunement, and common ground is the central concept in Stalnaker’s theory of pragmatic presupposition. Readers might then reasonably have expected our account of presupposition, like Stalnaker’s, to take our adapted version of common ground to be the main ingredient in our definition of presupposition. Instead, we adopted a definition of the presuppositions of a practice that doesn’t mention common ground (even indirectly, as collective attunement). This might seem a significant departure from a large subset of existing accounts, but it is not quite as significant as it might appear to be, because our definition sneaks collective attunement in through the back door. The back door in question is the fact that resonance is a very rich notion.
Let’s go back to the presupposition literature. We can think of presuppositions in some accounts as providing a relation between sentences, so the sentence “The king of France is bald” presupposes the sentence “There is a French king.” In other accounts, presupposition can be thought of as a relation between utterances and propositions, where an utterance is a production of a sentence by a particular speaker in a particular context. The importance of using a notion of utterance presupposition rather than sentence presupposition is manifest in the case of a sentence involving a first-person pronoun, like “my” in “my husband.” An utterance of “My husband is late again!” would presuppose the proposition that the speaker of that utterance has a husband, and it would be hard to express this presupposition cleanly in terms of the presuppositions of the decontextualized sentence alone, since the sentence is, as it were, separated from the speaker’s husband, if not completely divorced. Having observed the importance of an utterance-based notion of presupposition in cases where a sentence is indexical, that is, indexes the utterance situation, it then becomes apparent that the issue was present with the classic “The king of France is bald.” On a natural reading, that sentence implicitly indexes a time at which it is uttered (although there may be another, less obvious reading where the sentence spells out a generic claim about French kings, rather than referring to someone reigning at time of utterance). In “Pragmatic Presuppositions,” Stalnaker goes in a different direction. For him, presuppositions aren’t used to describe properties of sentences or utterances: pragmatic presuppositions are attitudes of people, a relation between a person and a proposition that holds when the proposition is in the common ground of the speaker and whoever the speaker is talking to.
So according to Stalnaker, it is at best unhelpful to talk of an utterance of “My husband is late!” presupposing that the speaker has a husband, and we should instead say that the speaker has, when making the statement, presupposed that they are married. We agree with Stalnaker that a notion of common ground is central to understanding presuppositional phenomena. Where we disagree is that, unlike Stalnaker, we see considerable utility in considering the presuppositions of utterances, or more generally, of behaviors, or, more generally still, of certain classes of behaviors, that is, practices. Although we will give a more detailed argument when we discuss the nature of accommodation below, the intuitive reason why we hold this view can be summed up as follows: there exist properties of language that are clearly conventionalized, since they identifiably recur with different occurrences of the same communicative practice (e.g., utterances of certain words), and which are needed to explain the communicative effects of those practices, and yet which have these effects somewhat independently of the psychology, attitudes, and expertise of the person saying them. A clear example is racial slurs: like grenades, they are still dangerous when thrown by a child who doesn’t understand what they are doing. When a child uses a slur, a lot is presupposed, including the structure of the practice itself. More generally, the resonances of actions are only partially determined by the people performing them, and the conventions responsible for such effects of words and other communicative actions are presupposed not by the users, but by the uses.
How then, does common ground, which we have generalized to collective attunement, bear on presupposition? Presuppositional resonance encodes all the prior resonances of a practice. Given that interactants are involved in every instantiation of the practice, and that each set of interactants has collective attunements, it follows that the distinctive regularities in those attunements will be part of the presuppositional profile.
For example, the presuppositional profile for a telic verb like “stop” might register a tendency for interactants to be collectively attuned to a preexisting process, so that when someone says, “Mary stopped smoking,” there might be not only a presupposition that Mary smoked previously, but also a presupposition that this is in the common ground. Similarly, if there is a distinctive tendency for people who use the expression “caffè mocha” to have an above-average collective attunement to the location of the nearest Starbucks coffee house, then that will be represented in the measure that the presuppositional profile derived from performances of the “caffè mocha” act attributes to the collective attunement of interactants to that location. To take a more politically relevant case, given that there is a distinctive tendency for people who use the word “inner city” in certain contexts to be collectively attuned to anti-Black racist ideology, then that will be in a presuppositional resonance of the practice of saying “inner city” in those contexts. Therefore, certain uses of the word “inner city” will constitute evidence of the political leanings of the interlocutors. Although the presuppositional profile is not defined in terms of common ground, it nonetheless yields a sort of normalized common ground, a pattern of distinctive collective attunements prior to an interaction. Let’s call the attunement profile of a practice the subpart of the presuppositional resonances that involves collective attunements of interactants.
As previously, we can give a discrete approximation of the typical common ground of a practice as the set of features of context that are above some threshold (e.g., 50 percent) in the attunement profile. Thus, we arrive at a variant of Stalnaker’s pragmatic presupposition.
It’s like his notion insofar as it involves the common ground of the interactants, but it is unlike his notion because it is not the actual common ground of those involved in any particular interaction, but is rather related to what an observer might discern about the common ground (qua collective attunement) on the basis of knowledge of the extension of the practice. The extension of the practice includes all sorts of things that somewhat consistently hold in the contexts in which the practice is performed. These include the collective attunements of interactants.
Echoing classic work on externalism in philosophy of language from Hilary Putnam, one consequence of attaching presuppositions to actions and practices rather than to people is that interactions involving novices can carry similar ideological presuppositions to interactions involving experienced practitioners.7 Consider kindergarten name-calling, say one kid trying to get a rise out of another using a homophobic slur. It might be that neither kid has a clear sense of the sociological background of the slur (a situation that one of the authors experienced in the schoolyard). Despite lacking understanding, the name-caller is somewhat attuned to a practice of homophobic slurring, and the name-caller’s action has unfortunate, or even painful resonances. Without knowing exactly what they are doing, the kid invokes a homophobic ideology, a resonance presupposed by the practice of using that term. Clearly, if the slur were only used by kids who knew nothing of the history of the practice, the practice would over time transmogrify. We would no longer be able to say that collective attunement to a homophobic ideology distinctively attached to the practice. But if kids use the practice as part of a broader community of practitioners attuned to the practice, then the presupposition will be maintained. Furthermore, the name-calling little kid has just taken a step toward mature attunement to the ideology, albeit that full attunement is going to require both practice and observation of the community of practitioners, plus, as we will discuss in the next chapter, some affinity to that community.
We’ve seen that practices carry presuppositions. But the reverse is also true: practices can be presupposed. Similarly, the ideologies in which those practices are collective attunements can be presupposed. How can a practice be presupposed? Recall that the extension of a practice is just a history of interactions understood as belonging to the same category. They are understood as belonging to the same category because they bear similarity to each other, because certain features tend to recur among these interactions, and collectively these features are distinctive. Features can recur in at least the following ways, all of which may overlap: (i) there can be a tendency for the practice to take place in environments bearing that feature or in which that feature is found at a distinctively high rate; (ii) there can be a tendency for interactants to themselves bear that feature; or (iii) there can be a tendency for the feature to be a collective attunement of the interactants. Let’s first consider a simple case of geographical restriction of a tendency, and then consider how one practice may be presupposed by another practice.
Suppose that there is a distinctive tendency for people speaking with a certain accent to be in Rome. Speaking with this accent then has a high presuppositional resonance with the location Rome, which is presumably why we would label it a Roman accent. There are also tendencies for the interactants performing the practice of speaking with that accent both to be located in Rome at the time of the act, and to have a history of location in Rome, especially while young. Finally, there is a tendency for those speaking with the accent to be attuned to Rome, dispositionally, emotionally, and attitudinally. Thus, speaking in the accent is correlated with behaving in a way that is sensitive to the geographical organization of Rome, having strong emotions with regard to Rome and all things Roman, and having a set of distinctive attitudes about Rome, for example, beliefs, hopes, and regrets, including metalinguistic knowledge of the Roman accent itself. Furthermore, one would expect these to be collective attunements of Romans.
What practices do practices presuppose? They have presuppositional resonances both with practices that occur at similar locations and times, and with practices with a heavily overlapping set of practitioners. Thus, practices of male friends drinking beer at a pub might have had, or still have, presuppositional resonances for practices of telling sexist jokes, and vice versa. But equally, the practice of telling sexist jokes might have resonances with male practices of behaving toward women in ways that make them feel uncomfortable, for example, the practice wolf-whistling, the practice of ogling, or practices of commenting on women in a way that reflects sexist stereotypes about their appearance and demeanor, prioritizing sexual attractiveness and quiet submissiveness over intelligence, independence, and forcefulness. Similarly, the practice of telling sexist jokes may resonate with discriminatory hiring and professional advancement practices if, as we suppose is the case, those who tell such jokes tend to belong to subcommunities in which the rate of these further sexist practices is elevated.
Practices carry presuppositions of collective attunement at very different levels of temporal granularity. At one end of the spectrum are patterns of short-term dispositions and mental states. The various practices involved in playing a complex board game (e.g., moving a piece in chess) commonly presuppose short-term collective attunement of the players to the state of play. Similarly, the subpractices of driving, for example, taking a left turn, presuppose collective attunement of drivers in the immediate vicinity both to the local road layout and to each other’s planned trajectories and vehicular maneuver capacities. Likewise, linguistic practices presuppose short-term collective attunement. At a level that for most people is consciously inaccessible, the practice of uttering a phoneme presupposes collective attunement to the prior phoneme, since that affects its production.
The practice of making a contribution in a conversation presupposes collective attunement to whose turn it is. Even those who interrupt will normally be so attuned, for people do not interrupt in the same way as they hold the floor. Let us return to an example that we discussed in the last chapter (section 4.4), the practices of using pronouns. Pronoun use presupposes the ability to rapidly develop joint attention on an entity with suitable number and gender characteristics. This presupposition often holds in the context in which a pronoun is used precisely because prior speech acts have introduced a suitable entity into discourse, making it jointly salient. But as we have also seen, pronouns are often used in cases where the presupposition of collective attunement to the referent of a pronoun is not met by virtue of mention in the immediate discourse context. In such a case, the speaker may be exploiting the standard conditions under which a pronoun is used to give the impression of collective attunement, the impression that the hearer is being treated as a confidante, while simultaneously engaging the audience by forcing them to do some thinking. Who is this “he” or (apparently less commonly) “she” the author is talking about? Or, in more political speech, who are “they”? More critically, are you one of “us”? The spectrum of collective attunement presuppositions runs from short time scale attunements that must be analyzed in terms of attention and other real-time mental processes to long-term attunements that concern ideological convictions and identity. The presuppositions of pronouns run the gamut, with the innocuously hidden and yet painfully sharp presuppositional resonances of us-them distinctions unwinding a strand of barbed wire through humanity, and forcing the hearer to choose an ideological side. More generally, even an utterance of a short word or phrase can be rich in presupposed ideological resonances.
An “Oy vey!” resonates with very different ideological overtones than a “Mamma mia!” Neither of them have the vile ideological resonances of a “Heil Hitler!” or, to take an example from Lynne Tirrell’s work, of the Kinyarwandan “Inzoka!” (“snake!”) used as a specifically racial taunt amid the crumbling social strife that presaged the Rwandan genocide.8 Our account of slurs, to be presented in chapter 10, will not be a purely presuppositional account, because understanding the practice of slurring must involve what slurs do as well as what they presuppose. Nonetheless, presupposed resonance plays a central role in that account, and slurs provide clear cases of presupposed and frequently dehumanizing ideology, whether that ideology is racist, sexist, homophobic, or, for that matter, liberal-phobic.
Standard accounts of presupposition, for example, as necessary conditions for interpretation, have limited value in explaining such effects.
5.3. Conventional Meaning As we discussed back in chapter 1, David Kaplan identified the meaning of an expressive term (a class that includes racist and sexist slurs, as well as his favorite examples, “ouch” and “oops”) with the set of contexts in which the term is used. Kaplan was using a standard approach to meaning whereby a set of contexts is equivalent to a proposition that might be spelled out as a big conjunction, each conjunct representing something that holds in every single one of those contexts, and such that the conjunction together entails anything that’s true in every single one of them. We noted that using this idea brings with it a problem: since the contexts of use of the term all have the property that they include the convention by which the term expresses what it does, it follows that the Kaplanian meaning includes the convention itself. Yet surely, when you utter “ouch,” you are not expressing the fact that there is a convention that one says “ouch” when in pain, albeit your utterance might convey evidence to this effect to someone unfamiliar with the convention. The problem is easily resolved in terms of presupposition.9 In this section, we’ll make the solution explicit by exposing more clearly the difference between what is presupposed and what is expressed, briefly discuss how expressive meaning relates to compositional meaning, and show how in both cases the model we have proposed leads to an account of conventionality that is somewhat different from those in the literature.
“Ouch!” If the message just expressed is not the conjunction “speaker pain is expressed by ‘ouch,’ and I’m in pain,” then what is it? An utterance of “ouch” presupposes the convention and expresses the pain. But how do we know that? Apart from being obviously silly, why is it wrong to say the reverse, that is, that “ouch” presupposes pain and expresses a convention?
This question is answered straightforwardly by the account of presupposition above. There is a high tendency for interlocutors to be collectively attuned to the convention prior to use of “ouch,” but there is not such a high tendency for interlocutors to be collectively attuned to the pain prior to utterance. Thus, the first attunement but not the second is presupposed. As for what is communicated by the expressive, it is natural to extend our account in terms of the change in collective attunements. So, we might say that the message communicated by a signal is what interactants tend to become more collectively attuned to by virtue of the signal being used. (Those familiar with dynamic semantics will immediately see that this is just a variant of a fairly standard dynamic account of meaning and presupposition, with roots in the work of Irene Heim and Lauri Karttunen, although it also has much in common with Robert Stalnaker’s pragmatic model of assertion.10) Let us shift from expressives like “ouch” to larger, compositional sentences. Consider an utterance of “Either it’s raining or it’s snowing.” What is expressed by the word “raining,” what is expressed by the first disjunct, and what is expressed by the entire utterance? We will say that in all three cases ideas are expressed.
“Raining” expresses an idea corresponding to an abstract property that a situation might have; “it’s raining” expresses an idea corresponding to a proposition about a particular situation at the time of utterance; and the entire disjunction expresses an idea corresponding to yet another proposition, a slightly weaker one. Of course, such an utterance would normally be used not merely to express an idea but also to perform an assertion. Whenever you assert something, you simultaneously instantiate the practice of expressing an idea and the practice of asserting it. On this view, whereas the first disjunct involves only an expression (or description, or presentation) of a way that things can be, the full utterance is involved in two distinct acts that involve its meaning, the expression of a certain idea, and the assertion that the idea corresponds to reality.
Uttering “it’s raining” is very different from uttering “ouch.” The first expresses an idea, and can be used to assert it, while the second expresses pain, which is not the right category of thing to be asserted. Nonetheless, they come with overlapping presuppositions, both presupposing not merely the specific practices needed to use them, but the much larger family of practices involved in speaking English. And to the extent that people who say “it’s raining” or “ouch” tend to share cultural tendencies different from people who say “il pleut” or “aïe,” those cultural tendencies, everything distinctively involved in anglophone ideology, will also be presupposed to at least a mild extent. When we observe people saying “il pleut” or “aïe,” we obtain a lot of evidence about them and their circumstances. Unlike someone who says “it’s raining” with a British accent, they are vanishingly unlikely to have eaten baked beans on toast for dinner last night. Yet someone who says “il pleut” does not thereby signal what they did or didn’t eat for dinner last night; that is not part of the message standardly encoded by this signal. Such information is a presuppositional resonance of the signal.
This takes us to a tricky issue: to what extent are presuppositions part of the conventional meaning of an expression? Some might take the fact that the presuppositional resonances for “ouch” and “aïe” encode information not only about the practices of expressing pain but also about what you ate for dinner as being a reductio ad absurdum of the claim that these resonances have any intrinsic connection to conventions of language. And indeed, perhaps it is. However, we should like to suggest that the absurdity rests not in the concept of presuppositional resonance, but in the idealization that there is a small, finite set of things that are part of the conventional meaning of an utterance of a word.
We don’t think that conventionality of the bulk of speech practices is black and white, but rather take it to vary continuously, so that some aspects of a practice, including a speech practice, may be more conventionalized than others. Note that although we have exemplified the noncategoricity of conventions in terms of differing degrees of presupposition (an issue that will become important in the next section), we believe it applies equally to the principal message communicated by a signal. In the domain of meaning, this vagueness is a natural concomitant of our assumptions about practice, for if the extension of a predicate is determined by its history of practice, then there should be borderline cases for which conventions less clearly determine its correct application than for central examples. To take classic cases from Eubulides and Wittgenstein, respectively, it is not clear exactly what counts as a heap, nor what counts as a game. 11 This is not to say that categoricity is not an essential part of grammar and communication, a point to which we shall return. The meanings of “heap” and “game” certainly exhibit the hallmarks of categoricity, as do judgments of grammaticality in many cases. Our point here is not to claim that there are no categories, or that there is no categoricity, but rather to point out that in our descriptions of linguistic conventions we often make simplifying assumptions.
The idea that there is a binary fact of the matter as to whether a particular action instantiates standard speech practice, that is, that grammatical conventions define completely strict categories for which one can state sufficient and necessary membership conditions, is an idealization. It is a useful idealization. It underpins generative linguistics, allowing syntacticians to equate grammars with sets of strings. But it is not an idealization that has clear empirical backing, and it is not an idealization that holds sway throughout linguistics. In studies of language change, as in the subfields of historical linguistics and sociolinguistic variation, it is standard to look at changes in progress as involving tendencies that are in the process of freezing into standard practice, or that continue to shift without fully stabilizing, or that are in competition with other practices without there being a uniformly preferred choice.
Certainly there are ways of modeling such processes in terms of standard methods with categorical generalizations, for example, by considering people’s knowledge of language to consist in mixtures of generative grammars (e.g., by assigning a probability to each grammar).12 However, to model people’s knowledge of language this way and allow every speaker in a community to have different mixtures is really to accept that the idealization of categorical grammatical conventions fails, and to offer a model that makes sensible conservative use of existing methods. If knowledge of language is knowledge of grammar mixtures, then that is as much an argument against exclusively categorical conventions as a spectrum of increasingly fishy merpeople would be an argument against exclusively categorical speciation.
5.4. Projection Having discussed the relationship between the message communicated by a signal and its presuppositional resonance, we now pivot to an account of presupposition projection. That is, in this section we will discuss the distinctive behavior of presuppositions in utterances that involve embedding one construction inside another. The nondiscrete model of presupposition (and collective attunement more generally) that we have adopted has more flexibility than standard approaches. We will argue that this gives it a significant advantage in accounting for the complex properties of projection that have emerged in recent linguistic work. As we have noted, projectivity is of interest not only as a curious linguistic phenomenon, but also for its particular significance in an account of the politics of language, since it helps explain why presuppositions are important vehicles for hustle.13 Presuppositions allow manipulative people to subtly bias questions, and, more generally, to affect hearers’ attitudes and dispositions without performing the type of assertive act that would make the intention to effect such change explicit. But for the moment we will focus on the abstract patterns of presupposition projection in their own right, and on how those patterns may be explained.
In section 4.2, we considered the presuppositions of the following Trump tweet, which among other things carries the presupposition that $700,000 was illegally given to Andy McCabe: Was Andy McCabe ever forced to pay back the $700,000 illegally given to him and his wife, for his wife’s political campaign, by Crooked Hillary Clinton while Hillary was under FBI investigation, and McCabe was the head of the FBI??? Just askin’?
What makes this an instance of presupposition projection is that a presupposition seems to jump out of a complex construction rather than being affected by that construction in the way that ordinary nonpresupposed meaning would be. Let’s start with where the presupposition comes from, the presupposition trigger. Triggers are normally thought of as grammatical constructions, for example, a nominal phrase headed by the word “the.” We see the grammar of a language as a practice, a collection of communicative actions seen as exemplifying a whole. Words have many facets. You may look at printed words and see something physical, as physical as the book or screen you’re reading from. But those word tokens embody a part of the broader grammatical practice. It is at this level, as embodying grammatical subpractices, that we can talk of a word or construction being a presupposition trigger.
The presuppositional resonances of a practice can be thought of as measuring how presupposed a particular feature of context is: the more strongly a feature is distinctively present in the contexts where the word occurs, the greater the degree to which it is presupposed. How does this explain projection? Rather than immediately attempting to resolve the general issue, let us first get our terminology straight and consider a narrower framing of the question. In the previous chapter, we said that certain constructions, like negation and questions, block commitments, so that, for example, whereas someone who uttered “It was raining” would be committed to a certain fact about previous weather, someone who uttered the question “Was it raining?” obviously would not be. Another term used in the literature is nonveridical. A linguistic construction that embeds or modifies others is nonveridical if the truth of sentences involving that construction does not depend on any proposition expressing material they embed or modify being true. Questions and negations are said to be nonveridical because they can be felicitously used independently of whether any sentence they modify expresses something true, or something the speaker takes to be true. The notion of nonveridicality allows us to give a narrow definition of projection in line with most prior work on the topic: Propositional projection: A construction is a trigger for a projective proposition if both unembedded uses of the construction and uses of the construction in nonveridical environments provide evidence of speaker commitment to the truth of that proposition.
For example, an unembedded use of “It was raining” commits the speaker to (i) grammatical and lexical practices such as those involving use of the word “rain” and its gerundive form “raining,” (ii) there being a mutually identifiable past interval and location under discussion, and (iii) there having been rain then and there. Which of these commitments survives embedding in the more complex construction “Perhaps it was raining”?
Clearly, the first and second commitments survive, and not the third. Why?
Why does the act of asserting the basic sentence carry many of the same presuppositions as asserting the modal variant?
In a celebrated 1974 paper, the linguist Lauri Karttunen inverted the concept of presupposition projection.14 The term “projection” suggests that presuppositions jump somewhat explosively out of constructions; we might think of Karttunen’s proposal as the idea that patterns of presupposition related inference result not from how presuppositions project out of constructions, but rather from how features of context tunnel in. To say that a presupposition of a construction A projects when embedded in a larger construction B is then to say that both A and B are normal utterances in contexts where the presupposition holds, as if features of the outer context in which B is uttered have tunneled in to become part of the context in which A occurs. As regards the clause “it was raining,” the argument would go as follows:15 (i) expressing a thought using that clause is appropriate in contexts in which interlocutors speak English and in which a past interval of time is salient; (ii) this is true also of the local context in which the clause occurs when embedded under a modal; (iii) by default, we should expect the local context created by the modal to be the same as the global context of utterance as regards these features; (iv) therefore, the contexts in which the modalized variant can be appropriately used should be the same in relevant respects as contexts in which the nonmodalized version can appropriately be used; and thus (v) projection is observed: presuppositional properties associated with the unembedded clause are also associated with the embedded clause.
The Trump tweet above involves embedding in a question construction rather than under a modal. Let’s just go through why this is a case of presupposition projection, simplifying the tweet slightly for the sake of clarity. Consider the unembedded clause in (1). If (1) was uttered, the speaker would be committed to $700,000 having been illegally given to Andy McCabe. But this would equally be the case if (1) was negated, modalized, or turned into a question, as in (2), (3), and (4), respectively.
Thus this speaker commitment has the hallmark of presupposition: it projects.
1. Andy McCabe was forced to pay back the $700,000 illegally given to him.
2. Andy McCabe was not forced to pay back the $700,000 illegally given to him.
3. Perhaps Andy McCabe was forced to pay back the $700,000 illegally given to him.
4. Was Andy McCabe forced to pay back the $700,000 illegally given to him? So, given that the commitment that $700,000 was illegally given to Andy McCabe is a presupposition, we can now say why it projects. First, definite descriptions are commonly used in the practice of referring. Successful use of a description results in a context in which there is joint attention on an object, and this joint attention then allows further practices to take place that depend on such attention, for example, acts of predication.16 Contexts in which acts of referring occur involve the distinctive presence of something matching the description well enough to be identified, so the presence of such an object is a presupposition. The context created by embedding the clause containing the description in a question, as in the original tweet, does not differ from the global context of tweeting in any relevant way. Therefore, the speaker’s commitment must hold in the global context of the act of tweeting. Trump has thus succeeded in conveying his commitment to McCabe’s shadiness inside a question, as if McCabe having received such money were a widely accepted and unremarkable feature of context, and not something new or controversial, something one might expect to be asserted rather than presupposed. As it happens, Trump has a penchant for presupposing bad things about Andrew McCabe. We leave it to the reader to unpack the presuppositions of the tweet that follows. (Rubric: one point each for explaining the role of the telic verb “continue,” which presupposes that the pre-state holds, and the role of embeddings involving the modal verb [“let”], negation, and imperative mood.) Republicans, don’t let Andrew McCabe continue to get away with totally criminal activity. What he did should never be allowed to happen to our Country again. FIGHT FOR JUSTICE!17 Neither is Trump the only one in his orbit using embedded presupposition triggers to communicate untruths. In case you need yet another exercise in unpacking presuppositions, here is former New York City mayor Rudy Giuliani embedding the presupposition trigger “find out” under the negative adverb “never” and a presumably rhetorical question at a press conference on November 19, 2020, dedicated to undermining confidence in the results of the national election that had just taken place: “Are you seriously going to want me to take seriously the secretary of state of Michigan when the secretary of state of Michigan never bothered to find out that the votes in her state were being counted in Germany by a Venezuelan company?” Presupposition projection turns out to be a messy business.
Presuppositions are normally differentiated from ordinary meaning (what is sometimes termed at-issue meaning) using evidence from projection, but there is now a long literature presenting examples in which projection does not occur. For example, the factive verb “realize” is normally taken to be a presupposition trigger, because it is associated with projective inferences.
The basic inference is seen in the fact that (1) implies the truth of the factive complement, the inference to the truth of (4). What makes this inference projective is that it is also seen in examples (2) and (3), in which “realize” is embedded under negation and in the antecedent of a conditional, respectively.
(1) Jason realizes that I’m in a bar right now.
(2) Jason doesn’t realize that I’m in a bar right now.
(3) If Jason realizes that I’m in a bar right now, he’ll text me to get back to work.
(4) The speaker is in a bar at the time of utterance.
In general, factive verbs (thus, not only “realize,” but also, e.g., “know,” “regret,” and “discover”) are usually analyzed as presupposition triggers.
But now consider this example, introduced by Lauri Karttunen a halfcentury ago:18 (5) If I realize later that I have not told the truth, I will confess it to everyone.
Example (5) does not imply that the speaker has not told the truth, but only that the speaker leaves open that possibility, so in this case the factive complement of “realize” does not project.
It was the variability of projection data, as seen in examples like (5), that led Stalnaker to propose his pragmatic account of presupposition. In advancing a pragmatic approach that he argued was more appropriate to such inconstant inferences, he avoided talk of presupposition triggering. He suggested that presuppositional inferences should be derived not by considering the conventional meaning of expressions to determine what is presupposed and what is not, but rather by considering the facts pertaining to a particular utterance and the conversational situation in which it takes place, and in particular what our knowledge of the conversational situation would lead us to take to be in the common ground of interlocutors.
On our reading of his early papers on presupposition, Stalnaker offers there a strong methodological hypothesis: although it is possible for meanings of expressions to have conventionalized presuppositional requirements, there is in fact no need to postulate such requirements, since presuppositional phenomena can be explained entirely in terms of speaker presuppositions. On this view, there is simply a tendency for people to use factive verbs when the factive complement is in the common ground of the interlocutors, and this is not a fact about the meaning of factive verbs per se. It’s only a tendency, so there are counterexamples, as in Karttunen’s example (5).
We certainly agree with Stalnaker that there is such a tendency. Indeed, the notion of presupposition we have developed is like Stalnaker’s insofar as it allows such tendencies, rather than depending on necessary conditions for interpretation, as would be the case in some semantic theories of presupposition. Where we disagree with Stalnaker is as regards the significance of this fact. Stalnaker assumes that conventional meaning is categorical—there is a conventionalized connection between a term and something else; it must manifest in every context. We reject this as an idealization about conventional meaning, one that we do not think should be built into our models.
On our view, tendencies to draw inferences are part and parcel of the history of the practice of using words, and the conventional meaning of those words is given by nothing besides history. In general, our attunementbased model allows for part-time presuppositions, in which weak conventions might in some cases lead to only weak evidence that a presupposition holds in the context of utterance, while allowing that sometimes it doesn’t.19 This much is in the spirit of the clarifications Stalnaker has offered regarding his account in more recent work: Claims about what sentences presuppose should be understood as claims about what cooperative speakers can normally be expected to be presupposing when they use those sentences. A presupposition “trigger,” on this way of thinking about the phenomena, is an expression or construction that signals, for one reason or another, that a certain presupposition is being made.20 Our agreement with Stalnaker here is partial. What we agree with is his suggestion that presuppositional effects reflect tendencies among speakers.
We do not, however, think that “claims about what sentences presuppose should be understood as claims about what cooperative speakers can normally be expected to be presupposing.” We think that claims about what sentences of a given type presuppose are claims about the typical contexts in which those sentences arise, or, more generally, that claims about what communicative actions drawn from a given practice presuppose are claims about the typical contexts in which those practices are performed. One type of property found in contexts in which practices are performed is the presence of collective attunements of various sorts among the interlocutors.
As we suggested in section 2.4, collective attunement plays much the same role in our framework as common ground plays for Stalnaker, and common ground is precisely what Stalnaker takes speakers to presuppose. But we have emphasized that the collective attunements we are interested in are neither limited to being beliefs, nor limited to the speakers involved in the interaction, and indeed may not specifically be attunements of any of the individuals directly involved.
The presuppositions of an action provide evidence as to the collective attunements of the community of practice that gives the action its communicative significance. These presuppositions include arbitrary attunements, and not merely beliefs: the bundle of attentional tendencies, dispositional attunements, emotional attunements, and cognitive attunements that make up an ideology. Claims about what sentences presuppose should be understood in part as claims about the ideology of the relevant community of practice within which the sentence has meaning.
Given that we are not only interested in propositional commitments of the speaker, but interested in arbitrary features found in the context of communicative acts, let us propose a generalization of the notion of propositional projection defined earlier in this section: Projection of resonance: A construction is a trigger for a projective resonance if both unembedded uses of the construction and uses of the construction in nonveridical environments tend to carry that resonance to a significant extent.
Note that we do not limit this to presuppositional resonances. In this part of the book, we are focused on presuppositional resonances, because we believe they are central to the transmission of ideology and to ideological evolution and conflict. But it is not only presuppositional resonances that project. Of particular importance when we come to discuss hate speech in detail in chapter 10 will be the projection of attentional and emotional resonances, and these resonances can be effects of communicative actions rather than presuppositions. Consideration of hate speech has proven problematic for standard theories of presupposition projection in part because certain features associated with hate speech, slurs in particular, are what has been termed “hyper-projective.”21 When a presupposition is said to be projective, the nonveridical embedding environments that are considered include, inter alia, negation, conditionals, questions, and imperatives. One type of environment that is usually not used as a diagnostic for presupposition is what we might call metalinguistic environments, such as indirect speech reports, quotative environments, and abstract discussions in which aspects of linguistic constructions are discussed in the abstract.
Metalinguistic environments generally have been assumed to be what Lauri Karttunen termed “plugs” to presupposition projection.22 So, for example, suppose David says the following: Jason said, “My sister is late!” It has generally been held that David would be not committed to Jason having a sister, despite this being a presupposition associated with what Jason purportedly said. Likewise, theorists working on presupposition would typically assume that in giving an example sentence, like that above, the authors of a book do not thereby become committed to the presuppositions triggered within the example sentence. In fact, Jason does not have a sister, although the fact that we feel a need to clarify this might give pause for thought.
The details of how presuppositions project in metalinguistic environments remain underexplored, perhaps because projection in such environments has been assumed to be so limited that they could not usefully be used as diagnostics for presupposition. Absent much data on projection of presupposition from metalinguistic environments, we will only note that our framework allows that there could be partial projection, in which a presuppositional resonance is weakened by embedding in a metalinguistic environment, but not eliminated. What is empirically clear is that some resonances of slurs project quite strongly from metalinguistic environments.
This is in fact the reason that we, like many authors, avoid even mentioning the N-word (as opposed to referring to it indirectly with the awkward locution “the N-word”). Hyperprojective resonances are those that are clearly not plugged, but project from metalinguistic environments.
Hyperprojectivity: A construction is a trigger for a hyperprojective resonance if both unembedded uses of the construction and uses of the construction in metalinguistic environments (including indirect speech reports and quotation) tend to carry that resonance to a significant extent.
The contemporary scholarly discussion of presupposition projection kicked into high gear after Terence Langendoen and Harris Savin introduced what they termed “the projection problem for presuppositions” a half century ago. They defined this as the problem of relating “the presuppositions and assertions of a complex sentence … to the presuppositions and assertions of the clauses it contains.”23 Let us generalize. The projection problem for resonance is the problem of relating the resonances of a complex communicative event to the resonances of its parts. Of particular interest to us later in the book will be the resonances of complex communicative events involving embedded hate speech.
5.5. Categoricity The resonance-based model of presupposition does not make a clean distinction between what is categorically presupposed and what is not categorically presupposed. We can imagine that some might view this as a shortcoming. The flexibility of the model would, at the very least, be unnecessary if there were clear evidence that in fact there are clear categorical differences, that is, evidence that the grammar makes a sharp distinction between presupposition triggers and nonpresupposition triggers.
In that case Occam’s razor might suggest a simpler account would be a better account.
Continuous models are often hard to work with analytically, so why bother? But maybe there is a bigger problem. Could it be that the type of continuous model we are proposing is not merely methodologically awkward, but is in a deeper sense a poor fit for human speech practices, because those speech practices involve categorical distinctions where our model doesn’t? We will now consider the question of whether grammars have clear categories of presupposing and nonpresupposing constructions, what we will term the question of presuppositional categoricity. But first it is helpful to think about what categorical differences in grammars look like more generally, itself a specific instance of the age-old question of what conceptual categories of any kind are like.
Categoricity has been much studied not only in abstract philosophical and mathematical terms, but also experimentally in cognitive psychology, especially in the area of categorical perception. Categorical perception is often studied in terms of the emergence of perceptual boundaries between categories, this emergence leading to two properties that we may think of as the hallmarks of categorical perception. First, people make sharp judgment changes at transition points between categories. So, if someone is exposed to a sequence of stimuli that gradually change (say a sequence of computationally morphed cartoons starting with dog-like images, but becoming increasingly cat-like), we tend to see a sudden transition in which images people will confidently describe as cats, rather than a gradual decrease in confidence of the images’ dogginess and increase in confidence of the images’ cattiness.
Second, people show much greater ease distinguishing two stimuli that fall on either side of a category boundary than they do distinguishing two stimuli in the same category. Thus, if shown two doggy pictures a few minutes apart, people may not be confident in saying whether it was the same picture, whereas if people are shown a dog picture and then a cat picture a few minutes apart, they are much more likely to decisively and confidently identify them as being different, and that may be so even if the first two images were more physically distinct than the second two.
The effect that it can be hard to tell the difference between two things in the same category if you haven’t explicitly learned to make such distinctions is known as within-category compression. It’s responsible for the fact that if you led a sheltered childhood and then met people of an ethnicity you were not exposed to, you might at first have found yourself embarrassingly unable to distinguish them from one another.
As it happens, one of the classic studies in categorical perception was a study of perception of language; we are thinking here of the highly influential work of Alvin Liberman and colleagues24 on the perception of consonants, work that created a standard paradigm in acoustic phonetics. As background, consonants (and speech sounds more generally) can be thought of as existing in an acoustic space involving combinations of frequencies distributed over time. For example, both “ba” and “da” correspond to an initial sudden burst of sound energy across the spectrum, followed by a disappearance of much of the high-end energy and a rapid transition into the base frequency and formants characteristic of the production of an “ah” sound. The sounds “ba” and “da” are very obviously distinct as regards the mechanics of production (the first involving lip closure, and the second a closure of the tongue against palate), and, introspectively, we feel that the two sound very different.
However, from a physical point of view, the acoustic differences are subtle enough that it needs a little practice to tell the difference just by looking at spectrograms. Furthermore, the question of how to specify the difference at a technical level is a difficult one, because people produce slightly different sounds from each other when saying what we might think of as the same thing, and, indeed, vary from one production to the next, and also because hearers will identify a range of sounds as being “ba” or “da.” That is, if we think of the acoustics of “ba” and “da” in terms of acoustic space, they are not points but zones in that space, with a range of physical event types that are heard as “ba,” and a range that are heard as “da.” What does it mean, then, to say that these sounds are categorical in nature? The experiments run by Liberman’s team involve hearers identifying sounds produced by a speech synthesizer, with examples of sounds drawn from an area of acoustic space that includes canonical “ba,” “ga,” and “da” sounds, but which also includes examples of sounds that are intermediary between these. The experiments, which involved both making judgments on individual sounds and making decisions as to whether two sounds were distinct, demonstrated both the above hallmarks of categoricity: sharp boundaries emerge with rapid changes in category identification across those boundaries, and high within-category compression, it being much more difficult to tell the difference between two “ba” sounds than between a “ba” and a “da,” even if the former pair of sounds were somewhat more physically distinct than the latter.
Let us return now to presupposition: is there reason to think that there is a categorical difference between constructions that bear a presupposition and those that don’t? There are certainly clear cases of both, but there is no evidence of presuppositional categoricity analogous to that found in work on categorical perception. In fact, on the contrary, recent work on presupposition projection is highly suggestive of there being degrees of presuppositionality, and perhaps even a continuous space. This work, in part joint with one of the current authors, involves people’s judgments as to what follows from an utterance or text, in particular judgments regarding inferences that involve projection, cases involving embedding in a nonveridical environment of a construction thought to be associated with a particular feature of context.
The studies involve either constructed experimental stimuli or texts extracted from a corpus, and in either case subjects perform judgments as to whether projection occurred. What has emerged in this work is, first, that there is great variance in constructions standardly identified as presupposition triggers as regards the strength of projection effects, and furthermore that this variation is observed somewhat independently of the embedding environment. The literature on presupposition might be taken to suggest that in the relatively simple cases studied in the experimental work, people should have sharp judgments on whether a presupposition projects or not, but that is not the case: confidence in a projective inference doesn’t appear bimodal at the individual or group level. Neither do we observe a bimodal split between presupposed and nonpresupposed inferences when looking across different constructions.
For example, if we consider the presuppositions standardly associated with the factives “know” and “discover” (i.e., the presupposition that the complement is true), we find that projection is significantly stronger for “know” than for “discover.” If we then compare both of these with the presupposition standardly associated with the construction “stupid to” (as in “Jason was stupid to coauthor with David,” presupposing that Jason did coauthor with David), we find significantly stronger projection for “know” and “discover” than for “stupid.” And if we compare these with a standardly analyzed presupposition of “only” (“Only Jason cried,” presupposing that Jason cried), we find that projection in this case, while still clearly present, is weaker than for any of the other three.25 Furthermore, it turns out that typically projective inferences are found even in cases that are not standardly taken to be presuppositional, for example, for the complements of the verbs “believe” and “say.” Consider the sentences “He doesn’t believe that he lost the election,” and “He won’t say that he lost the election,” both which involve embedding a nonfactive verb under negation.
When someone hears such sentences, there is a significant chance they will conclude that the speaker is in fact committed to it being the case that whoever is under discussion lost the election. Summarizing: constructions standardly analyzed as presuppositional seem to have varying presuppositional strengths (as measured by projective inferences), and even constructions standardly analyzed as nonpresuppositional have properties that could be described as weak presuppositions insofar as they sometimes lead to typically presuppositional inferences.
We do not wish to argue that a clear case has been made that presupposition is an inherently noncategorical phenomenon. Rather, we would seek to put the shoe on the other foot: categoricity of presupposition is an idealization that may be helpful for formalization, but is not yet rooted in empirical research, there being no analogue in this area of the work in categorical perception discussed above. There is, at present, no clear evidence that it is a categorical matter whether a construction carries a presupposition or not. Until such evidence is found, we would suggest that a sound research strategy is to be open to noncategorical models and to recognize that in assuming categoricity one is making a convenient idealization that so far lacks direct support.
Whether a theoretician chooses to make an assumption of categoricity will presumably depend on their immediate goals, for example, the particular slice of data they are looking at, but one thing should be clear from the discussion in this section: assuming categoricity will not always make the linguist’s life easier. If your goal is to analyze the sort of projectivity data discussed in this section, you would be hamstrung attempting to produce variable projectivity results in a categorical model.
On the other hand, a resonance-based account along the lines we have sketched could readily be applied to such data. Indeed, the results on projectivity just discussed suggest a potential methodology for quantifying the presuppositional resonances of different constructions, whereby a feature of context would have a presuppositional resonance strength for some construction that is proportional to the degree of projectivity found empirically, once other factors have been controlled for. We will not pursue such an approach here. Our goal in this section has been more modest: to clarify what categoricity of an account of presupposition would mean, and establish the plausibility of a noncategorical model of presupposition along the lines we have proposed.
5.6. What Else Is Missing?
We have now set forth the main elements of our view on how meaning relates to context, a view that we hope might support analysis of aspects of social and political discourse that are not well described under standard idealizations in the theory of meaning. We started the chapter with a missing piece. We had suggested that many of the effects that concern us in this volume depend on presupposition, but we had not said what the presuppositions of an action are like. This lacuna was particularly pressing, given that in the first part of this volume we developed a theory of meaning in terms of resonance and attunement, notions that are not discussed in any prior model of presupposition.
Our treatment of presupposition generalizes standard approaches in two ways. First, we treat context dependence as intrinsically scalar, allowing different degrees to which an act is revealing of (or dependent on) aspects of its context. Second, the types of presupposition we consider are not typically discussed. In particular, an act may presuppose social practices, including the communicative conventions of the act itself.
Let us return to the explanations of the effects of biased questions in two lines of experimental work in cognitive psychology, Elizabeth Loftus’s studies of recall and memory, and the empirical studies on which Daniel Kahneman and Amos Tversky founded prospect theory. In both cases, the experimental effects hinge on the framing of the question. In Loftus’s work, framing a question in ways that evoke properties or constituents of scenes that were not present causes people to “remember” the scene differently. In Kahneman and Tversky’s work, framing a decision problem in terms of gains causes people to make different choices than if the problem were framed in terms of losses, an effect that would not be predicted in a simplistic application of classical utility theory to human decision-making.
Loftus herself has advocated for the role of presupposition in the memory paradigm we discussed, and we argued that presuppositional analyses are appropriate for both the memory and prospect theory paradigms.
Quite generally, while it is possible to offer a framing explicitly, for example, in negotiation (“Do you think we might put it this way …?”), it is far more common for a framing to be presupposed. What does it mean for a framing to be presupposed? A framing is a communicative practice, a practice of using certain expressions for some purpose. In the cognitivepsychology examples, the framing expressions are taken to provide a particular conceptualization of an issue or an event, for example, a conceptualization of a vehicle collision as involving high energy if the verb “smash” is used, or a conceptualization of the effects of a health program as offering a positive gain if it is described in terms of lives “saved.” The wellworn example of pro-choice versus pro-life framings for abortion can be seen as providing a conceptualization in this way, but it also suggests that frames are something more. Each of these abortion framings is a broad rhetorical practice that doesn’t merely describe a state of the world in a particular way, but offers a key dichotomy around which abortion debates can be structured. Much of the literature on framing concerns complex rhetorical practices used to do things other than describe situations or outcomes; for instance, a frame may serve to motivate protest, unify a group, or deride opponents.
Having recognized that frames do not merely describe, but have other functional effects too, we can see that the description of frames in the cognitive-psychology cases as merely offering ways of conceptualizing an issue is unnecessarily limiting. It might be that a relevant effect of using the word “smash” is to engender excitement, and it might be that there is a dramatic difference between the emotional resonances conjured by talk of “lives saved” as against “lives lost.” Further, it might be that there are dispositions associated with thoughts of gain that are quite different than dispositions associated with thoughts of loss, for example, a disposition to play it safe in the first case, and a disposition to gamble to avert disaster in the second. On the model we have described, frames can not only presuppose particular conceptualizations, but can also carry emotional and dispositional resonances as part of their presuppositional profile.
Given these considerations, we feel that our account of presupposition not only explains core aspects of examples we have discussed, like the ability of the framing presuppositions to project from within questions, but also offers some novel directions for further study. And yet the story so far is incomplete, for there is one central component of the explanation of the effects of frames in the previous chapter about which we have said very little: accommodation. That is, we have provided an account in which the presuppositions of the experimental stimuli in the experiments we discussed there can be modeled, but we have left completely open the question of how hearers adapt to those presuppositions.
In standard models of communication as information exchange, accommodation is what allows presuppositions to be informative. In the next chapter, we will suggest that a generalized notion of accommodation can not only account for standard cases of informative presupposition, including the adaptation subjects apparently perform in response to deviously framed experimental stimuli, but can also play a central role in describing the transmission of ideology, ideological conflict, and the crystallization of separate groups.
These are big claims, and they may seem fanciful to those who are familiar with the literature on presupposition in philosophy of language and formal semantics/pragmatics. Here a crucial difference between prior work on presupposition and our own becomes significant. Prior work focuses on presuppositions about the world or about what has been said earlier in a discourse, and in these models, accommodation, as noted, explains how these presuppositions can be informative, effectively allowing repair of faulty contexts in which there is a mismatch between interlocutors’ beliefs. Once we move to considering presupposed norms and practices, and once we allow that accommodation involves not only modification of beliefs, but also modification of behaviors and dispositions, and once we also consider the motivations for accommodation in terms of joint behavior at the level of groups of arbitrary size, accommodation becomes something far more remarkable. It becomes an engine of societal change. To get a sense of what we take to be the importance and breadth of application of accommodation, note that we will suggest that it can provide a way of understanding aspects of what Judith Butler termed performativity, which she develops as part of an anti-essentialist model of categories such as gender.
26 Butler’s notion of performativity has had an impact on many literatures and is a central notion in current third-wave sociolinguistics, building on a long line of work in sociology that begins with Erving Goffman’s use of the term performance. (Butler’s use of performativity is presumably influenced by both J. L. Austin and Goffman, but it is clearly unlike Austin’s in that it is not restricted to a narrow set of sentence types.
Goffman’s theatrically inspired notion of performance27 is broad, encompassing all aspects of the way someone acts during an extended encounter.) Folk models of social identity, often used at least tacitly in scientific and humanistic scholarship, are essentialist in the sense that they treat categories such as gender and ethnicity as fixed, and regard people’s behaviors as allowing them to mark their identity relative to these fixed categories. Butler’s seminal work swept the rug from under this view, explaining how people’s behaviors can dynamically redefine or create the very social categories they differentiate. Our view borrows heavily from this perspective. As we will develop the account, accommodation might be said to be not only informative but also formative, contributing to the cascading development of new social categories, behaviors, and groupings in overlapping chains of successive social interactions.
1. Bourdieu, Outline of a Theory of Practice, 80–81.
2. Tonhauser, Beaver, and Degen, “Gradience in Projectivity and At-Issueness.” 3. Dickens, Hard Times, 234.
4. Austin, How to Do Things with Words.
5. Austin, How to Do Things with Words, 51.
6. What we refer to here as “presupposition” is related to other standard concepts like “tacit knowledge” and “implicit assumptions” or to Camp’s “perspective” (e.g., in “Two Varieties of Literary Imagination”). Against this background, the idea that complex systems of presuppositions can be virtually invisible should not be contentious, since the idea that complex assumptions provide a lens through which one sees the world and yet are themselves invisible until revealed by careful investigation has long been central to the entire enterprise of philosophy.
7. Putnam, “Meaning and Reference.” 8. Tirrell, “Genocidal Language Games.” 9. It will have occurred to some readers that perhaps the problem we identify with Kaplan’s proposal could be solved in terms of his distinction, in separate work, between character and content. The idea would be that the character of an expressive is a function from contexts to a Kaplanian meaning. So, in contexts in which there is a convention that “ouch” expresses pain, the Kaplanian meaning of “ouch” is then the set of all contexts in which “ouch” would appropriately express pain. Perhaps this could be made to work, but it would certainly require a careful slicing and dicing of contexts. At any rate, our point here is not to “fix” Kaplan’s proposal, but rather to use the problem to illustrate how we differentiate what is at issue from what is presupposed. Our approach could even be seen as building on Kaplan’s account of indexicals, as it has a similar two-dimensionality. On our view, it is only relative to a context that a given behavior can be seen as instantiating a practice, and only relative to a context that practice can then have whatever effects it does (say, expressing a proposition, promising something, or slurring).
10. Heim, “On the Projection Problem for Presuppositions”; Karttunen, “Presuppositions and Linguistic Context”; Stalnaker, “Assertion,” “Pragmatic Presuppositions.” 11. Wittgenstein, Philosophical Investigations, 3. It remains contested whether Eubulides is the originator of the paradox of the heap, and no text of his survives.
12. The mixtures of grammars approach can be found, for example, in the work of historical linguist Tony Kroch (e.g., “Reflexes of Grammar in Patterns of Language Change”). Such methods are also common in computational models of language acquisition, with much work using probabilistic context-free grammars (which allow each grammar rule to be associated with a probability).
13. We are far from being the first to see presupposition as playing a central role in manipulative language. For example, in Sold on Language, Julie Sedivy and Greg Carlson argue that presupposition is an important device in advertising.
14. Karttunen, “Presuppositions and Linguistic Context.” Karttunen’s model of presupposition is closely related to Stalnaker’s. Both propose to explain presuppositional data in terms of the contexts in which utterances occur, and both allow that these contexts will be updated dynamically as conversation proceeds. However, Stalnaker’s model cannot properly be described as an account of presupposition projection, because the pragmatic model of presupposition he develops does not assume that presuppositions are conventionally associated with particular constructions, i.e., presupposition triggers. If there are no presupposition triggers, it doesn’t make much sense to ask what happens when presupposition triggers are embedded. A further difference between Karttunen’s work and Stalnaker’s is that while Karttunen details exactly how local contexts in which presupposition triggers occur may be different from the broader, global context in which a complete utterance occurs, Stalnaker does not attempt any such detailed description, and indeed in much of his work he seems committed to there not being a useful notion of local context that differs from the global context of utterance.
15. The Karttunen-derived explanation of projection suffices for our purposes here, but we note that alternative explanations might be compatible with our general approach. In particular, we are thinking of an explanation in terms of at-issueness as proposed by a group including one of the current authors (see Simons et al., “What Projects and Why”). The idea would be that the history of practice is used not merely to determine what features of the context are normally present when a given construction is used, but also what issues are under discussion. Projection would then tend to occur when a construction is used that indicates that some feature holds in the context, but when the question of whether that feature holds in the context is not itself under discussion.
16. Definite descriptions have various uses other than referring, e.g., in “Jason is not the only author of this book,” where the description “the only author of this book” is nonreferential. Such examples are used by Coppock and Beaver in “Definiteness and Determinacy” to motivate associating definite descriptions with a presupposition weaker than is normal in other approaches, a presupposition that there is at most one object satisfying the description. 17. Donald J. Trump, @realDonaldTrump, Twitter, November 10, 2020, accessed February 20, 2023, https://twitter.com/realDonaldTrump/status/1326194143132082178?ref_src=twsrc%5Etfw.
18. Karttunen, “Some Observations on Factivity,” 64. The example is his (25b). For naturally occurring data involving nonprojection of factive presuppositions, see Beaver, “Belly Button Lint Colour.” 19. Note that the possibility of part-time presuppositions in our model makes the account in some respects like that of Gerald Gazdar’s Pragmatics, Implicature, Presupposition, and Logical Form.
This is hardly surprising, since we and Gazdar owe a common debt to Stalnaker’s pragmatic model of presupposition. However, our model differs from Gazdar’s in that we allow some constructions to trigger presuppositions more strongly than others. In contrast, for Gazdar every use of a trigger introduces a defeasible preference for the presupposition to be accepted, and there is no difference in the strength of this preference from one trigger to the next.
20. Stalnaker, Context and Content, 94. We note in passing that a much stronger view than that expressed in this passage, namely the view that presupposition theory could get by only with mention of speaker attitude and without mention of conventionalized properties of linguistic expressions, would lead to a contradiction in Stalnaker’s account. The problem concerns cases of informative (what he later terms anticipatory) presupposition. Let us make the strong assumption that knowledge of language does not involve knowledge distinguishing presuppositions associated with particular constructions from other types of content. Now suppose that a speaker utters a sentence like “I don’t have time to feed the cat” to someone who they realize doesn’t know they have a cat. On the strong view suggested by Stalnaker’s early work, these cases involve a pretense by the speaker that the presupposition is in the common ground, when in fact it is clear that this is not so. The question then would be, How is the speaker manifesting the pretense? Since the only relevant behavior being performed by the speaker is uttering a certain expression (“the cat”), the pretense must consist in uttering that expression. But in that case, uttering the expression constitutes acting as if the presupposition in question is in the common ground.
The use of the expression, in other words, must signal that the speaker has a certain presupposition, and a speaker’s knowledge of the expression must include the knowledge that the expression provides such a signal. But this is inconsistent with our initial assumption. Therefore, knowledge of language must involve knowledge distinguishing presuppositions associated with particular constructions from other types of content.
21. Camp, “A Dual Act Analysis of Slurs,” 39.
22. Karttunen, “Presuppositions of Compound Sentences,” 174.
23. Langendoen and Savin, “The Projection Problem for Presuppositions,” 55. Note that their statement of the problem uses the term “assertions” to mean something like entailments, since they do not appear to be taking sentences or clauses to be performances of assertive speech acts. Their statement of the problem can be related naturally to Frege’s principle of compositionality, that the meaning of a larger unit must be a function of the meaning of the parts and their mode of composition. Langendoen and Savin can be seen as saying that if Fregean compositionality is right, then it must apply to presuppositions too, and the projection problem is the problem of determining how composition affects presuppositions.
24. Liberman et al., “The Discrimination of Speech Sounds.” 25. The cited results are found in Tonhauser, Beaver, and Degen, “Gradience in Projectivity and At-Issueness.” For a corpus study of projection involving a wider range of constructions and embedding environments, see de Marneffe, Simons, and Tonhauser, “The CommitmentBank.” 26. Butler, Gender Trouble.
27. Goffman, “Presentation of Self in Everyday Life.
CHAPTER SIX On Parole All the world’s a stage, And all the men and women merely players: They have their exits and their entrances; And one man in his time plays many parts.
—WILLIAM SHAKESPEARE, AS YOU LIKE IT, ACT 2, SCENE 7 Each society demands of its members a certain amount of acting, the ability to present, represent, and act what one actually is. When society disintegrates into cliques such demands are no longer made of the individual but of members of cliques. Behavior then is controlled by silent demands and not by individual capacities, exactly as an actor’s performance must fit into the ensemble of all other roles in the play. The salons of the Faubourg Saint-Germain consisted of such an ensemble of cliques, each of which presented an extreme behavior pattern. The role of the inverts was to show their abnormality, of the Jews to represent black magic (“necromancy”), of the artists to manifest another form of supranatural and superhuman contact, of the aristocrats to show that they were not like ordinary (“bourgeois”) people.
—HANNAH ARENDT1 6.1. Vox Populi A dictionary clings to language by the tail. Communicative practice rushes forth, caring not for ideals of stability of form or determinacy of meaning.
There is no fact of the matter as to where a communicative practice is going. There is no fact of the matter as to where it’s been; there is just a history of instantiation of which no individual has ever experienced more than a tiny slice. If ideologues seem to succeed in tying a practice down for a generation or two, what is left in the hands of their heirs is hollow ritual.
Practice must live free to remain vital. The power of political communication rests on its ability to draw attention, drive attunement, and differentiate adversaries. This is achieved through an ever-evolving mixture of repurposed spandrels (to hijack a term from evolutionary biology2), new enough to catch the eye and ear, and old enough to be recognized. To understand the politics of language is not merely to understand the catchphrase du jour, but to understand the processes that make such phrases effective, and the processes that make them rot.
Although this volume is entitled The Politics of Language, our goal is not to detail the role of some particular set of linguistic constructions found in political rhetoric. The words and phrases of any particular political moment are in constant turmoil. Even when some phrases, like “law and order” or “states’ rights” recur, they recur with new resonances. Though in no way improved, neither a Confederate flag nor a noose hanging from a tree mean quite the same as they once did. The already borrowed “Make America Great Again” of the Trump presidency will never be the same “Make America Great Again” again. A word repeated is not the same word but an echo of the original, with all the resonances it has acquired along the way: plus c’est la même chose, plus ça change. Jorge Luis Borges takes this idea as far as it can go in his story Pierre Menard, in which Menard is imagined not to copy Cervantes’s Don Quixote, but to reauthor fragments of it from scratch centuries after Cervantes. The new version is word for word identical with the original despite being written independently, centuries later. As Borges tells us, “The Cervantes text and the Menard text are verbally identical, but the second is almost infinitely richer.”3 Sometimes there are functional motivations for performing a communicative action in something other than a way that prototypically exemplifies prior practice. Such a motivation arises whenever there is a need to distinguish between one signal and another. A peacock’s best bet for impressing the peahens is not to have tail feathers indistinguishable from those of his fellow peacocks, but to have bigger and brighter tail feathers than any of the others. The extra tail-feather production can only be achieved if the peacock is in excellent condition, and even then it will be a drain on resources. But if there wasn’t a price, it wouldn’t be costly signaling. Thus the practice of tail-feather signaling, once started, does not remain fixed, and neither does the cost: an evolutionary tail-feather fashion race ensues.
Or suppose, to use an example from sociolinguist Penny Eckert’s seminal Detroit high-school study, consider a teenage girl who has a need to mark her membership in an anti-establishment grouping, the Burnouts.4 The Burnout pronunciation of “fight” sounds a bit like “foyt,” and black makeup is in. Should she pronounce “fight” just like the average Burnout, or might she have reason to push it to more of an extreme? Would she aim to look like an average member of the group, or would she be tempted to dab just a bit more makeup on, you know, so everybody really knows who she is? The distinctive accent and dark makeup may not win favors with the teachers, or elite jobs down the road, but if there wasn’t a price, it wouldn’t be costly signaling. Thus do vowels and cosmetics evolve.
Or consider a political candidate trying to stand out from all the politicians that have gone before, much despised for having looked after their own and not the bulk of people in the country. One way of differentiating oneself would be to break norms and push boundaries of what is acceptable. Such a strategy will garner enemies. But here, again, the price is the point.
Clearly, there are sometimes reasons to perform a practice not as it has been typically performed before, but in such a way as to clearly distinguish performance of the practice from performances of variant practices. This is closely related to the point Ferdinand de Saussure was making when he remarked that “in language there are only differences,” a central theme of his work.5 The practice of producing the vowel in “fight” needs to be distinguished from the practice of producing the vowel in “feet,” just as Burnouts need to distinguish their practice of producing the vowel in “fight” from the practices others use when pronouncing “fight.” Each individual performance must strike a fine balance. It must be sufficiently like other performances of the practice to be seen as an instance of the class, and yet sufficiently different from performances of other practices so as to minimize the possibility of confusion.
The need to differentiate one practice from another places limits on a norm-based account of practice, for in a norm-based account an optimal action would be one that epitomizes the practice it instantiates. Our point here is that a communicatively optimal action is often not one that epitomizes a communicative practice, but rather one that most clearly instantiates the practice. Clarity of performance demands not merely faithfulness to a prior practice, but distinction from others.
Saussure’s dictum applies at the level of differences in speech practices marking differences in social identity. The consequences of the social need for distinctiveness for speech practices can be divided in two.
First, there is selective pressure to perform in a key that is incompatible with the keys of groups from which one’s own group is distinguishing itself.
Recognition will be easy if members of one group dress or wear their hair or speak in ways that out-group members wouldn’t, or if they all manifest beliefs that are antithetical to the out-group creeds. Further, faking group membership will be harder under the same conditions, when playing in the key of the group risks censure by the rest of society. For some, to wear extravagant makeup might or might not be as hard it would be for an Ephraimite to pronounce the “sh” of “shibboleth,” but to wear such makeup would be to take a public risk, albeit one that many others have chosen to take. The example suggests that the same features that play into the key of a social group, and thus support collective harmony within that group, may also in some circumstances be sources of dissonance.
There may be dissonance for individual group members, who could belong to multiple overlapping groups, each tuned to distinct keys, and there may be dissonance caused in society more broadly.
Second, there is pressure for the key of a social group to evolve. This is true partly for reasons given above: drift is inevitable unless explicit work is done to maintain conventions, and regular change in practices makes them harder to imitate. But there is a further reason: however distinctive a behavior is, it can always be made more distinctive by performing it in a way that is less like the behaviors of those in other groups. It is furthermore quite possible that after regular exposure to an attunement, or because of change in other attunements, out-group members will cease to be surprised by it, and cease to attend to it, thus bleaching its value as a marker.
Here let us make a brief excursus to examine another important scholarly connection and its significance for political propaganda.
Saussure’s focus on differentiation in language is mirrored by the more general analysis of individual and cultural differentiation developed a couple of decades later by the anthropologist Gregory Bateson on the basis of his studies of the Iatmul people of Papua New Guinea. Bateson defines what he terms schismogenesis as “a process of differentiation in the norms of individual behaviour resulting from cumulative interaction between individuals.”6 The processes we exemplified above are what he terms complementary schismogenesis, when behaviors of individuals or groups coevolve to become increasingly differentiated. He opposes this to symmetrical schismogenesis, in which accentuation of the same pattern of behavior occurs simultaneously in opposed individuals or groups as a result of pressure to outdo competition. A classic example he gives is of ever more extreme boasting behavior among males in a group he worked with, but examples of similarly unstable evolution of progressively more extreme behavior are found in many arms races, both metaphorical and, unfortunately, literal.
Bateson’s concept of schismogenesis is not merely analytic, but has seen strategic use in the political sphere. During World War II, Bateson’s roles included application of schismogenesis to sow discord in Japan through the use of “black propaganda.”7 This involved Bateson operating a fake Japanese radio station from Burma and Thailand, for example, transmitting variants of Japan’s own wartime messaging that were sufficiently exaggerated to undermine confidence in the veracity of the original. One of us has defined undermining propaganda as a “contribution to public discourse that is presented as an embodiment of certain ideals yet is of a kind that tends to erode those very ideals.”8 Black propaganda goes further.
It is a contribution to public discourse that is misrepresented as from a committed group member yet is of a kind that tends to erode that very group. The insight that Bateson and other subversive political propagandists have used is that societies are systems that are at best only in an unstable equilibrium. In terms of the resonance metaphor, this system is prone to fracture when energetically stimulated.
The engineering iconoclast Nikola Tesla famously claimed to have shaken buildings almost to the point of collapse with a carefully tuned pocket-size oscillator, and he is variously reported to have boasted that he could have toppled the Brooklyn Bridge or the Empire State Building, and that he did cause an earthquake.9 The application of schismogenesis to black propaganda can then be understood by physical analogy with the type of machine Tesla claimed to have produced, or the smaller-scale phenomenon of wine glasses shattering when exposed to a carefully pitched voice. Black propaganda is an attempt to induce catastrophically disruptive tumult within a group using apparently innocuous messaging that resonates with members of the group. Despite its apparent mildness, the messaging progressively shakes the group from the inside, driving it away from equilibrium and making it ungovernable.
Similarly divisive messaging strategies appear to be a mainstay of contemporary social media disinformation campaigns.10 It is perhaps helpful to understand the mechanisms of such campaigns, as Bateson did, as simply leveraging existing social processes to sow division in a target group. It immediately becomes clear that while in certain types of propaganda campaign the goal is to always stay on message, we should not be surprised to see that in others neither truth nor consistency is central.
This latter characteristic is most obvious in the so-called firehose of falsehood approach to propaganda exemplified in the disinformation campaigns of Putin’s Russia.11 We return to these broad issues of political group dynamics in the next section, but for now we will focus on consideration of the processes of individual and group adaptation. These processes support the communicative fabric of communities of practice, but provide the substrate for disruptive propaganda as well as for constructive government and social development.
In this volume, we do not focus on truth, but on resonance, and we do not study language as a fixed construct, but as a collection of societal practices in constant tension and development. The model we have been building up provides a way of looking at evolution of meaning that is rather different from that found in more standard models of meaning, whether that of Richard Montague or J. L. Austin. These models analyze meaning in terms of a single, determinate set of rules. It’s not that Montague and Austin did not recognize the possibility of change or its societal importance, but that they abstracted away from change in order to lay bare the structure of the beast on their dissection table, frozen, as it were, in time. In studying the social and political workings of language, we think it important to start with the idea that the functions of expressions of ordinary language do not hew to fixed rules, but are determined by the living conventions of practice. The extensions of practices change with each instantiation, and what counts as a distinct practice can change as rapidly as our recognition of what counts as a distinct community of practice. The fission of a practice into distinct subpractices (say, a “fight” subpractice and a “foyt” subpractice) is a natural concomitant and indeed integral part of the fission of communities into distinct subcommunities in which processes of schismogenesis are inevitable, and can sometimes be accelerated by active social engineering.
In our version of a Wittgensteinian practice-based account of conventional communicative function, the meaning of an expression is not given by a lemma in a dictionary, or even by a short list of lemmas. Rather, there is a history of usage, what we have termed the extension of a communicative practice, and the functions the expression has had on different occasions of use form a cloud of resonances. Interlocutors can coordinate on a function on a particular occasion of use to the extent that the functions the expression seems to have had when used in comparable contexts are sufficiently narrowly circumscribed for the purposes at hand.
To adapt an example from a well-known discussion of naming, from psychologist Roger Brown, if an adult uses “dog” in talking to a child, they might successfully coordinate on reference to the animal before them, and not its collar or the chew toy in its mouth, even if the child also uses “dog” to refer to arbitrary large quadrupeds including cows and horses.12 Similarly, to use Quine’s famous skeptical argument for the impossibility of reliable translation, one can imagine successful coordination on the reference of a use of “gavagai” (by a speaker of an unknown language when in the presence of a rabbit) being at least equally robust.13 Such coordination does not depend on the presence of identical conceptual representations among interlocutors. It merely depends on there being sufficient similarity in cognitive and other attunements to serve the immediate purpose, whether that commonality arises as a matter of genetic predisposition, similarity of experience, or congruent acculturation.14 Communicative practices at once serve as pivots enabling coordination, and yet shape-shift. In the remainder of this chapter, we will study how communicative practice evolves in terms of the resonance model of practice and presupposition laid out in the last five chapters, focusing on how the process of accommodation used in the account of presupposition can be generalized and explained.
Our discussion of accommodation addresses a gap in our account of resonance. The work of Rae Langton, Caroline West, and Marina Sbisà, and much other work besides, has suggested that accommodation is the mechanism by which the presuppositions of communicative actions wield their power, as well as being the process underlying the effectiveness of much manipulative language in politics, advertising, and everyday life.15 We agree, and will go somewhat further, suggesting that accommodation is a central mechanism by which multiple linguistic interactions over time lead to ideological change. It is the place in the theory of meaning where slow processes of adaptation are accounted for.
Standard accounts model accommodation as a process in which a sharp single change takes place to fit the perceived needs of an interaction. On this view, accommodation is not a good candidate to explain long-term effects of the ideological resonances of language. We aim to improve on this situation, by providing a way of thinking about accommodation whereby it becomes clearer how small changes within individual interactions can lead to long-term ideological change, whether that change is for better or for worse.16 Accommodation does more than plug a gap in our resonance-based model. The term “accommodation” is used in two highly divergent bodies of work. One is in analytic philosophy of language and formal pragmatics, and the other is in sociolinguistics, social psychology, and communication studies. The two bodies of work do not involve any visible signs of significant scholarly cross-fertilization, such as cross-citation. Indeed, at first glance, these lines of scholarship appear to have little more in common than a coincidence of nomenclature, although it is at least the case that both approaches take accommodation to be an interactional, adaptive process. In this chapter and the next, we will suggest that there is value in viewing these two bodies of work on accommodation as part of a single generalized picture of adaptation in interactional situations. Just as we can speak of behavioral attunement and cognitive attunement, so we can speak of behavioral accommodation and cognitive accommodation, and these two types of accommodation, roughly speaking, correspond to the types of accommodation found in the two literatures.
6.2. Accommodation as Harmonization The view that we develop in this chapter is that in considering the evolution of practice, accommodation is the engine of change. It is not merely a special type of comprehension. Accommodation is a force shaping both our individual performances and the practices and ideologies that come to define the groups we are aligned with. So how will we analyze it? Here’s the soundbite: Accommodation is harmonization to context. Large societal and cultural structures must be understood in terms of individual human interactions. The large structures are not only reflected by those local interactions, but indeed can be said to consist in local interactions and to be transformed by them. This idea is at the heart of the adaptation of Durkheim’s work to the study of conversational interaction by Erving Goffman, a shift in focus made completely explicit in Thomas Scheff’s conception of microsociology, 17 and developed further in Randall Collins’s interaction ritual theory.
18 Collins summarizes (in the abstract of a central paper) his view in a way that recalls both Durkheim’s account of rituals and Festinger’s account of dissonance: Individuals continuously negotiate [social] coalitions in chains of interaction rituals in which conversations create symbols of group membership. Every encounter is a marketplace in which individuals tacitly match conversational and emotional resources acquired from previous encounters. Individuals are motivated to move toward those ritual encounters in which their microresources pay the greatest emotional returns until they reach personal equilibrium points at which their emotional returns stabilize or decline.19 How do individuals “tacitly match conversational and emotional resources acquired from previous encounters”?
How can a negotiation take place implicitly? Our answer is that a history of practice is encoded in the presuppositional resonances of conversational actions, and that the implicit parts of the negotiation involve adaptive moves of accommodation, driven by preferences for individual and collective harmony. We will not engage closely with Collins’s proposal in this volume, but if our interpretation of his model is on the right track, then what his account demands is a theory of interactional accommodation. In the rest of this chapter, we will explain what we think such a theory must involve and then apply it to one of the hallmarks of the contemporary political scene: polarization and the formation of echo chambers.
A slur, a sexist joke, or a casual dismissive gesture can change future attitudes and behaviors. We have suggested that the presuppositional profiles of denigrating acts like these include discriminatory and oppressive ideologies. Spread and acceptance of these ideologies takes place when exposure to the acts leads observers to internalize the behavior of typical interactants, deepening their attunement to the ideologies that are prominent in the presuppositional profiles of the act. Each interaction might have only a small effect on an audience member, whose grasp of the act will be incomplete. On a first exposure to a slur, a watching kid may grasp no more than that they’ve observed a practice of someone expressing heightened negative emotion toward members of a group, perhaps with the group itself not being clearly delineated as far as the kid is concerned. Nonetheless, over the course of subsequent exposures to this and related behaviors, and with the help of bootstrapping achieved through mimicry, and thus learning by doing, the kid can develop the discriminatory dispositions they have observed. The mechanism whereby attunement to presuppositions is deepened is accommodation. Extending slightly the soundbite at the start of this section, here’s a definition: Accommodation: Accommodation is harmonization triggered by the perceived context of communicative interactions.
According to this definition, accommodation is a process of adaptive change in a conversational setting. But not just any change. First, it is specifically harmonization.
Second, it is harmonization to the perceived context of interactions.
If accommodation is harmonization, then accommodation is a psychosocial process whereby a system of attunements changes so as to yield increased consonance for an individual or group. Our use of “harmonization” implies that people can accommodate to different degrees.
This is in line with our model of presupposition, which is also scalar, but contrasts with the all-or-nothing models of accommodation that are standard in analytic philosophy and theoretical semantics/pragmatics. As we shall see later in the chapter, the process of finding consonance can also yield a certain type of dissonance, so the claim that accommodation yields consonance is nontrivial. Note also that the definition of accommodation does not limit it to being a psychological process of just one agent, and does not limit it to being a process in the head of a hearer, both of which are common assumptions in prior work. On our view, accommodation can be an adaptation of any participant in an interaction, or an observer of the interaction. In particular, speakers may accommodate as they speak. Indeed, accommodation is not limited to individuals, but can be analyzed at a group level, whereby what is harmonized is not an individual’s attunements, but a group’s collective attunements.
Accommodation at the group level might involve the people of an entire nation harmonizing in a way that is triggered by the contexts they observe in TV sitcoms or election campaigns.
A pliant population accommodates by coming to speak and act in ways that are partly determined by the models set in front of them, using the mannerisms of sitcom stars and smoking with the calm style of a Marlboro cowboy.
We say in the above definition that accommodation is triggered by the context of an interaction. Why do we emphasize context? Because this is what distinguishes accommodation from uptake. By context, we are referring to the situation in which an interaction takes place, and not to the new situation created by the predictable effects of conversational moves within the interaction. Consider an instance of the command “Hand over the money!” uttered in a bank. The context includes physical features, such as the location and the visible appearance of the interlocutors, and various sociocultural features, like the cultural significance of the bank, the roles of those who work in it and visit it, and a mass of social practices that are taken by default to be shared among those in this space, including speech practices. According to the model of presupposition developed in the last two chapters, the complex communicative action of issuing the command can supplement this context. But how does this supplementation work? That is precisely where accommodation comes in to provide the missing link in the story so far.
Recall that a feature of context is a resonance of a communicative practice to the extent that an occurrence of the action changes the probability of that feature (section 2.5). A bank robber demanding “Gimme the money!” instantiates many different practices, including the practice of speaking English, the practice of definite reference, the practice of using an imperative to command, and the practice of robbing a bank. All of these have resonances.
Some of these resonances are features of the context that results from the communicative action, features that are sufficiently associated with instantiations of the practices in question that they can be seen as somewhat predictable effects. Examples are the hearer recognizing what the speaker is asking for, the hearer becoming publicly committed to giving the speaker what they have demanded, the hearer actually giving the speaker what they have demanded, and the hearer getting scared.
The presupposed resonances of a practice are features of typical contexts in which the actions take place. Here, presuppositional resonances include the presence of money to be handed over, the ability of the hearer to provide cash without the normal paperwork, and a power relationship sufficient to coerce the addressee into performing this distinctive financial transaction. The fact that someone has performed practices with certain presupposed resonances provides a clue as to what the context is like, or at least what the speaker takes the context to be like. A hearer, if they are attuned to these same practices, can then leverage the presupposed resonances of a communicative action to adapt to the situation.
The hearer changes the way they are attuned to aspects of the context reflected in the communicative action directed at them. This adaptation is accommodation. In the simplest cases that we consider first in this chapter, the adaptation monotonically increases collective attunement between interactional participants. Their individual harmonization, as each interactant adapts to the presuppositions of the other’s actions, is part of a broader process of collective harmonization taking place as a conversational interaction or relationship unfolds. A well-studied special case of such collective harmonization is the process known as grounding, whereby interactants develop shared understanding of their environment and shared ways of talking about issues of interest to them within that environment. Grounding is then a process involving layered accommodation of presuppositions, although it might also involve metadiscourse, explicit negotiation about the assumptions to be made and the speech practices to be adopted.
It is important to realize that while accommodation, on our view, always involves harmonization, harmonization is more general: it can include explicit negotiation, and it can also involve reactions to features of the environment that are neither communicative nor social. Suppose you are attuned to swans being white and observe something that appears to be a black swan, and that you then gradually adapt, modifying in some way your beliefs or your understanding of your perceptions. This is harmonization, but it is not accommodation, since it was not triggered by the perceived context of a social interaction.
6.3. Beyond Scorekeeping In the midst of our discussion of presupposition in the last two chapters, we also introduced the logico-linguistic tradition of work on accommodation, which, at least as regards use of the term “accommodation,” dates to David Lewis’s foundational 1979 paper “Scorekeeping in a Language Game.”20 This tradition takes accommodation to be a repair strategy, whereby a hearer’s beliefs change in order to enable comprehension of what a speaker has said, or in order to reflect what a speaker appears to have assumed is common ground.
Accommodation played a central part in our explanation of various phenomena. Although in some cases we invoked accommodation to clarify phenomena that are not normally explained in terms of accommodation, even there our use was quite standard from the point of view of work on presupposition. For example, our explanations in chapter 4 of framing effects both in Kahneman and Tversky’s work on prospect theory, and in Elizabeth Loftus’s work on memory, depended on experimental subjects accommodating framing presuppositions.
Our account of accommodation is, we freely admit, more powerful than is needed to account for the bulk of data on presupposition accommodation in linguistic theory. Harmonization is a leviathan, and the standard wellworn examples, as they are normally described, are small fry in the sense that they require neither attention to social and emotional factors, nor a general theory of dissonance and mental landscapes in which compartmentalization and self-deception are possible. In essence, most of the work for these cases is done by the way people recognize context, and very little work is done by harmonization.
Let us recap the explanations of classic psychological experiments we presented in chapter 4. Suppose that an utterance, involving a strategically deployed how-much question and when-clause, presupposes that two cars “smashed” into each other. The hearer knows that “smash” is an expression standardly used in a context of a high-energy collision, and simply has to take the same perspective as the speaker appears to have, recognizing that the current context is the normal type of context in which the expression is used. Harmonization here consists in nothing more than blithely adopting the perspective of those with whom you are engaging.
The explanation is similar for many standard cases in the logicolinguistic tradition. When a speaker uses the description “my sister,” a hearer can then recognize, simply adopting the same perspective as the speaker, that they are in the sort of context in which such a description would be apt, that is, one in which the speaker has a sister. Unless the hearer previously had reason to doubt the existence of said sister, there need be no complex chain of reasoning of the form “Wow, that’s weird. The speaker is acting as if they believe that they have a sister. I wonder why they have this belief? That my interlocutor would not only have this belief but also seem to take it for granted that I have this belief is creating dissonance for me, for it reflects a difference between our beliefs. How can I resolve it? I know, I’ll adapt my beliefs, and henceforth will believe that the speaker has a sister.” No, what normally happens is that by virtue of their competence in the same speech practices, which includes an implicit understanding of the presuppositional resonances of the utterance, the hearer somewhat automatically develops exactly the attunements that the speaker’s act presupposed. That is, and though we don’t think it controversial, we are hypothesizing that most accommodation of this sort is nondeliberative.
Similarly, the speaker may not have reasoned in a terribly complex way when using the phrase “my sister,” and may not have considered the probability that the hearer would be familiar with the sister’s existence. The fact that a speaker would assume things with which the hearer is not familiar indicates not that they probably did some complex reasoning about what pretenses they could get away with, or what would be the optimal way to convey the desired information to the hearer, but rather exactly the opposite. The speaker may have done no careful audience design, and may not have deliberated at all about how to describe their sister in the way that, say, the joint authors of an academic monograph might deliberate about what the audience could be expected to know. The upshot of these considerations is that the type of harmonization needed to account for the effects of presupposition in these cases is very restricted: it is merely collective harmonization of interactants. This collective harmonization centers on a preference for the consonance produced by speaking and thinking like others in the group you are in. We do not need a very complex model of attention, or a complex model of reaction to psychological dissonance, in order to account for such cases.
Most speakers do not function like the writers discussed previously, in the final section of chapter 4, who strategically began novels with unsatisfied presuppositions as to the referents of pronouns. When a speaker says something with presuppositions unknown to their interlocutors, we should not by default conclude that they are masters of sophisticated audience design, but rather that they have probably engaged in no conscious audience design at all. It is not a trivial fact that both speakers and hearers tend to adopt fairly lazy communicative processing strategies. Interlocutors assume that their communicative practices are sufficiently robust that they will manage to jointly coordinate on many of the same features of context, even without explicitly discussing or reasoning about those features. This fact is closely related to the more general phenomena of mimicry to be discussed in the next section, a good default approach to developing coordination, and it is not particularly controversial. However, unlike most speakers, both experimental psychologists designing stimuli and novelists composing the first line of a novel are skilled manipulators. Thus, they don’t merely act in accordance with a context, but can design a context that has no prior existence. It’s because of experimental psychologists and novelists, as well as propagandists, that we have been careful to include in our definition of accommodation the word “perceived.” In accommodating, people don’t necessarily harmonize to the real-world context in which we are situated, but can harmonize to a perceived context, whether recognizing that they are doing so (as perhaps may be true in the case of a mysterious pronoun opening a novel), or not (when successfully duped by psychologists and propagandists).
Unconscious adoption of the perspective of an interlocutor might be said to be on the ground floor of a theory of harmonization, and does not in and of itself demand the towering edifice above it. Indeed, many of the more interesting cases, the cases for which upper stories of the building are needed, are those in which accommodation does not take place, in which an audience questions a presupposition or refuses to accept it, even in the face of the dissonance this creates in the conversational situation. Although it is far from being a default, such accommodation failures are not uncommon in political discussion, or, more generally, in conversations between people with strong affinities to groups that are in conflict. In such cases, the refusal to accept framings or go along with tacit assumptions might potentially be the basis of fruitful negotiation, but refusal can easily lead to communicative breakdown.
This is a topic we will be returning to throughout the chapter, and especially in section 6.6, where we consider the role of systematic conversational breakdown in cases of extreme political polarization.
As should be clear, accommodation is typically subconscious. Or, at least, most accommodation is not sufficiently attended to that the subject in a psychological experiment would later report that their judgments depended on accepting assumptions implicit in the experimental stimuli.
Most subjects in experiments we have discussed presumably neither attend to bias in the framing of the stimulus questions, nor consider alternative framings. As we have noted, for Lewis, accommodation occurs in the blink of an eye. There is something right about this, and something wrong.
What’s right about it is that since accommodation typically occurs as an autonomic and relatively nondeliberative process, the actual time spent adapting to a particular speaker may be short, and the amount of time we are aware of adapting to the speaker may be negligible or zero. This lack of attention to accommodation is precisely the reason presuppositions make such excellent vehicles for hustle, and thus explains why the manipulative use of presupposition is a key strategy for politicians and advertisers alike.
What’s wrong with Lewis’s blink-of-an-eye claim is that it assumes accommodation to be an all-or-nothing thing: some feature either is or is not in the conversational score.
Accommodation may happen rapidly, but it may also be the case that what happens rapidly is a small change, and that the net effect of a lot of small changes is in effect a gradual process, a process that appears more like slow adaptation to prevailing attunements of a community of practice, acclimatization to its ethos and mores, than like a sudden volte-face.
To repeat part of the passage from Lewis we have cited earlier: “If at time t something is said that requires presupposition P to be acceptable, and if P is not presupposed just before t, then—ceteris paribus and within certain limits—presupposition P comes into existence at t.” This is clearly an unhelpful way to think about the process of accommodating practices. If we have practices as core examples, central cases of accommodation must be regarded as taking much longer, say the length of a childhood, or of an extended propaganda campaign.
In the examples that motivate the standard theory, accommodation of a proposition can reasonably take place in real time as we process an individual utterance.
Suppose Elly says, “I am picking my sister up from the airport,” and Allie accommodates the proposition that Elly has a sister. This sort of process is so rapid that it is hard to find evidence of extra time being taken to accommodate presupposed material above and beyond the time taken to process similar expressions when the presupposition is already known to the hearer. Contrast this with the types of cases that are central for us, for example, when executives describe people solely in terms of their work efficiency and their interlocutors increasingly treat those too old or infirm to work with ever more disdain. The time course of this process might be measured not in dozens of milliseconds but in dozens of PowerPoint presentations. To be sure, there might be rapid and unconscious periods of adaptation during each PowerPoint presentation, but each of these episodes feeds into a much longer-term process of harmonization that might lead the hearer to become gradually inculcated in a particular technocratic way of thinking, or might eventually lead to such dissonance that they eventually cut themselves off from their in-group, and leave for greener pastures.
Here we hope the reader is beginning to see how viewing accommodation as harmonization might constitute a major shift in presupposition theory. Whereas previously accommodation of presuppositions was understood as a sort of short-term online belief revision process, we are suggesting that it involves not only belief revision, but also behavioral and emotional adaptation, and that these processes of adaptation might take place across a wide range of time scales.
Furthermore, a process of adaptation that takes place across multiple sales meetings and business lunches is not simply a process in which one person produces an utterance with a presupposition, and another reacts to it. Accommodation is a social process, and groups can accommodate, collectively harmonizing to the jointly perceived contexts of multiple conversational interactions. It is a process in which multiple hearers are affected by multiple presuppositions in multiple utterances by multiple speakers, and in which many of the hearers will themselves also play the role of speaker. That is, accommodation becomes an interactive process involving feedback loops. Thus accommodation need not simply be a reshaping of attunements to better fit a context, but can become an evolving process of group change that, while triggered by a particular misalignment with context, ends up yielding a succession of collective re-attunements that have no simple relationship with the original misalignment. We should expect this process to be deeply affected by the social organization of the group within which accommodation is occurring, because, for example, power relationships and factors such as identity and role will affect who pays attention to whom, and thus the dynamics of who adapts to whom and in what way.
Consider the rippling effects of accommodation to the presuppositions of a new meme and its associated metadiscourse, say the #metoo movement that began in 2017. Following a spate of reporting on victims of sexual misconduct, and in many cases criminal sexual assault, a single Twitter posting encouraging people to post their own victimhood stories under the #metoo hashtag inspired an explosive cascade of responses. The phenomenon included tens of millions of #metoo social-media posts within days. Such effects need to be seen holistically, at the level of a large community, not at the level of individual interactions.
There is thus a parallel to individual accommodation at the level of groups. Since such processes lead to change in collective attunements, we could naturally call what is occurring collective accommodation. It is collective accommodation not merely to a new speech act, but a new speech practice. That practice evolved, and continues to evolve, as does the community of practice within which the practice exists, each use of the hashtag adding to the extension of the practice, and members of the ever-changing and increasingly global community accommodating as they perceive differences between their attunement to the practice and what they perceive to be the broader collective attunement. What is collectively accommodated may be collective attunements that are not even analyzable at the individual level. For example, one might optimistically hope that #metoo led to a reduction in the level of certain types of oppression, but the disposition to oppress is itself not a disposition at an individual level. Virality of a meme and oppression are like war, famine, and mass extinction.
They can only be understood at a systemic level.
In the rest of this chapter, we discuss the process of accommodation at a group level. We argue that harmonization helps us understand the accommodation of types of attunement not considered in the standard Lewis-Stalnaker type models we have discussed so far, and we further describe how accommodation might be usefully seen as central to grouplevel processes of ideological transmission and change.
6.4. Tuning In to Others We have at various times in this volume talked of perspectives, and of how people shift perspectives, or take the perspective of another. These notions will be of particular import in chapter 7, where we discuss the concept of neutrality. But what is a perspective? In terms of our model, a perspective on a set of features of context is a distinctive system of attunements to those features. Similarity of perspectives on a set of features for a group of people then equates to the group having what we have termed collective attunement to those features.
Given this way of explicating perspective, we can then say what adopting someone’s perspective consists in. To adopt some individual or group’s perspective on a set of features of context is to accommodate toward them as regards attunements to those features, harmonizing sufficiently that one’s attunements to the features are similar to theirs.
Extending this definition of perspective shift, we might say that to empathize is to temporarily adopt someone’s perspective, especially by harmonizing with and attending to emotional attunements and behaving in accord with them. The fragmented nature of the landscape of attunements allows such temporary perspective-taking to occur without greatly impacting an agent’s other attunements. We can, in the moment, take on aspects of someone’s thoughts and feelings, even their personality, but compartmentalize, much as does a method actor.
The method actor inhabits their role not only mentally, but physically, and yet in an instant can return to their more familiar persona, like Sherlock Holmes miraculously whipping off his disguise before a befuddled Watson.
We will now argue that our definition of accommodation as harmonization to context implies that it can be manifested not merely passively and epistemically, as a Lewisian change in cognitive attunements, but performatively, as short-term change in behavior within an interaction.
How so? Well, if accommodation is harmonization to the context of communicative interaction, then it follows that it can involve reactions to arbitrary features of that context.
What are the relevant features of context?
They are features of context presupposed by the utterance, and features that are salient to the interaction. This could potentially include the physical space within which the interaction is held, or the medium of communication, but the most important features are features of the interactants themselves. Accommodation might be a response to any feature of an interactant, including the identities they are seen as representing, the way they are sitting, and the way they are talking. But what response? Since harmonization is a process of finding consonance across a system of attunements that is not restricted to cognitive attitudes, accommodation could include an emotional response or a dispositional response.
Furthermore, a dispositional response might be long-term or short-term, and a short-term dispositional change might be registered by a perceptible alteration in behavior. Thus, accommodation could produce an emotionalbehavioral response to someone, say, smiling, relaxing, or lighting up a cigarette, or becoming visibly or audibly angry, tense, or threatening. Let us term behavioral adaptations to conversational context behavioral accommodation.
Behavioral accommodation is not studied in the logicolinguistic tradition but in a separate psychosocial tradition of work that was initiated in the early 1970s by Howard Giles, a tradition now called Communication Accommodation Theory.
21 The primary area of theorizing in Communication Accommodation Theory is the effect of the social relationship between interlocutors on their interactional behavior, considering factors such as status and ethnic or gender identity. For the most part, the focus is on the degree to which behaviors converge or diverge, although some work considers more complex social behaviors, like the telling of jokes. Seen at a high level, both the logico-linguistic and social traditions treat the default and most common types of accommodation as coordinative, whereby interactants come to think alike, act alike, and see themselves as alike. Thinking alike, in the sense of having common beliefs, is the notion of accommodation prominent in the logicolinguistic tradition; acting alike, and seeing ourselves as alike, are central to the psychosocial tradition. The psychosocial tradition also considers the case of interactants who behaviorally manifest their differences. We will suggest that such cases are in fact not exceptions to the generalization that accommodation is coordinative, but we postpone that discussion to the next section. Our goal in the remainder of this section is to describe the social basis of coordinative accommodation.
It seems reasonable to postulate that the reason we accommodate to others is that we care what they think about us. William James’s description of our sensitivity to the views of others, and the sense in which that feedback shapes how we see ourselves, remains apt: A man’s Social Self is the recognition which he gets from his mates.
We are not only gregarious animals, liking to be in sight of our fellows, but we have an innate propensity to get ourselves noticed, and noticed favorably, by our kind. No more fiendish punishment could be devised, were such a thing physically possible, than that one should be turned loose in society and remain absolutely unnoticed by all the members thereof.… Properly speaking, a man has as many social selves as there are individuals who recognize him and carry an image of him in their mind. To wound any one of these images is to wound him. But as the individuals who carry the images fall naturally into classes, we may practically say that he has as many different social selves as there are distinct groups of persons about whose opinion he cares. He generally shows a different side of himself to each of these different groups. Many a youth who is demure enough before his parents and teachers, swears and swaggers like a pirate among his “tough” young friends.22 It seems safe to say that our feelings of consonance, or dissonance, are hugely dependent on our perceptions of the behaviors of others. Which others? In the above quote, James simply says that it is others “about whose opinion [one] cares.” But whose opinions do we care about? Psychological work on personal and social-identity theory23 suggests that people’s social behavior can be thought of in terms of a balance between (i) conformity to groups to which one feels affiliation, (ii) the constraints of social roles which one plays, and (iii), as emphasized in Social Distinctiveness Theory, 24 competing needs for individual selfdetermination and distinctiveness within one’s in-groups. This idea of optimal social distinctiveness is central to Thomas Scheff’s Microsociology.
The close relationship between Scheff’s model and our own is apparent in the following passage, in which he relates his model to Émile Durkheim’s notions of mechanical solidarity (similarities of experience and practice, enhanced by joint ritual, and weakening in modern society) and organic similarity (social and work-functional complementarity and interdependence, increasingly dominant for large, modern, urbanized populations):25 Too little distance produces overconformity, just as too much produces underconformity. Some balance seems to be necessary to allow for the system to maintain coherence, a mixture of tradition and change.
Although at first sight the level of differentiation seems similar to the distinction Durkheim made between mechanical and organic solidarity, the overlap is slight. Durkheim’s distinction is strictly behavioral; it does not encompass the main elements in [psychiatrist Murray Bowen]’s idea [of social bonds within families], which concern internal as well as external states. Organic solidarity concerns cooperative behavior. Optimal differentiation in Bowen’s sense concerns intellectual and emotional attunement, mutual understanding as well as behavioral interdependence.26 As the earlier passage from James suggests, pressures for similarity and differentiation may be felt through observation of the behavior of others with whom we interact. Such feedback might be in the form of explicit praise or criticism. But it might also take the form of nonverbal behavioral accommodation such as semitransparent feedback given by bodily orientation, gesture, and facial expression, or still more subtle feedback based on the level of attention we receive. One way in which someone’s behavior manifests their attention to us is via its responsiveness to our own behavior, even if that response has no inherent emotional valence. If someone catches a ball you have thrown across a crowded football field, or catches a glance that you have shot across a crowded room, then they are paying attention to you. They care about you in at least a very limited sense. Note here that if you are static, then you do not provide a clear opportunity for your interlocutor to accommodate to you, since absence of major movement is a default. If it is important, whether for coordinative or emotional reasons, to know that people are paying attention, and if responsiveness provides a way of gauging attentiveness, then it would be adaptively beneficial to keep shifting, so that each shift offers an opportunity for response. If every time one person smiles the other smiles, and when one crosses their arms the other crosses their arms, and when one tells a joke the other laughs, then they know they have each other’s attention.
From a young age, people take cues about how to behave from those that represent the groups with which they are affiliated. Among the earliest studies demonstrating this effect systematically were the “Bobo doll” experiments conducted sixty years ago by the psychologist Albert Bandura and collaborators, building on a preexisting line of experimental research in imitative learning.27 Kids aged three to five got to watch how adults interacted with large wobbly dolls: some adults were violent toward the dolls, while others were gentler.
When the kids were later left alone with the dolls, their play tended to mimic the actions they had witnessed, acting aggressively toward the dolls if that’s what they’d seen, and more gently if they’d witnessed adults doing so. Crucially, the extent to which they mimicked certain behaviors depended on what can be interpreted as an ingroup/out-group variable: kids were more likely to reproduce verbally aggressive behavior by an adult of like gender.
28 Leaving aside the extent to which we can draw strong conclusions directly from these old experiments, what is clear is that Bandura’s study has been hugely influential in work on how aggression spreads, and on mimicry in the absence of reward. As regards effects of in-group status, there have now been many studies showing differential preferences for imitation of behavior by in-group members.29 To put these results in our terms, without making any strong claim that we are doing more than redescribing them: (i) people (and especially children) have a tendency to accommodate to interactional behaviors they have witnessed by harmonizing with those behaviors, by developing dispositions to behave in the same way; and (ii) people are more likely to accommodate to in-group members than out-group members because the need to harmonize with a group is dependent on the affinity felt toward that group. It is far from surprising that people are disposed to act like those they perceive to be of their kind; birds of a feather, after all, have been known to flock together. What makes humans special in this regard is that both the feather and the flocking are cultural artifacts, and that a single individual can wear many feathers, each of which marks membership of groups that themselves each flock in many ways. For humans to be birds of a feather is for them to jointly accept membership and trappings of a category that humans themselves have constructed. These constructed categories may be loosely based on naturalistic concepts, like race and gender, or not, as in ideological groupings or football team allegiances. For humans to flock, in the broad metaphorical sense we intend, is not merely for them to gather together physically, but to exhibit arbitrary common behaviors. To flock is to form a community of practice.
Cultural flocking produces feedback loops: our behaviors come to define our categories, the wearing of the feathers that mark out the community itself being a community behavior. To take an example that has been a focus of feminist academic study for many years, gendered behaviors help define what it is to have a certain gender.
30 Similarly, to be a redneck is no longer to be a fair-skinned person who labors under a Southern sun, but to exhibit whatever dispositions and attitudes have come to loosely define a clustering of people who are unlikely to self-identify as rednecks at all. Even without self-identifying as rednecks, the clustering of behaviors and attitudes that out-groups dismissively refer to as stereotypical of rednecks may be strong enough that it brings people together at the church, the concert, the racing circuit, and the polling booth.
Category-defining behaviors, and hence the categories themselves, are dynamic. If highly visible people in a group develop a tendency to use social media in a new way, or party, parade, or pray in a new way, others in the group will have a tendency to accommodate those new practices, that is, fashions. And once the practices become identified with the group, part of the group’s standard social scripts, the group’s identity will in turn become identified with the practices, producing a cycle of renewal. We see in this dynamicity an abstract similarity with the suggestion above that dynamically shifting behavior can be adaptively advantageous in verifying attentiveness and thereby underpinning coordination.
At the group level, too, change itself can be a positive adaptive trait, and not merely an inevitable effect of drift. If social-identity markers are static, then they might be easily appropriated by other groups or mimicked by outsiders wishing to join a club that the members might rather keep exclusive. If social-identity markers evolve dynamically, like a changing password at the door, then the club will be harder to get into. Stay trendy.
While natural properties such as age and skin tone clearly play a role in providing perceptually accessible markers for many groupings, what we’re talking about here is the question of what cultural devices serve to identify the members of social groups. That is, we are interested in what attunements people in groups need to have in order to be recognized by each other. Although arbitrarily complex protocols are possible, a simple principle is obvious: the more distinctive attunements members of a group share, the easier it will be for them to recognize each other. To return to an earlier metaphor, the distinctiveness of these attunements is analogous to a musical key: you recognize others in your groups when they play in keys that you like to play in, that is, when they behave in a way that manifests dispositions consonant with yours.
The term “key signature” is apt, but a key is more than a distinctive squiggle: it both identifies and defines a group, helping to make membership of the group and the individual attunements that form the key meaningful. There is, then, a collective harmony in a social group when there is a clearly defined and recognized key collectively accommodated by the group that conveys meaning to the group members. The key is a pattern of group-appropriate attunements, such that individual group members can manifest their membership by displaying attunements in that key, such that other group members are perceptually attuned to the meaning of the key, and such that when group members perceive someone playing in that key, what they perceive is a fellow group member. A given performance in this key may or may not make group identity apparent or salient for out-group members. A key may be difficult to reproduce because it is a form of costly signaling and the cost is too high for many; it may be difficult to reproduce because it keeps changing, as we just suggested; or it may be difficult to reproduce because it has the quality of a shibboleth—virtually imperceptible or practically irreproducible for those not in the know.
Parts of the key may be invisible to those not paying attention, as with a key ring or carabiner hanging in a particular way to signify gender identity. The choice of hanging keys for this function itself carries suggestive resonances that go beyond a mere lexical coincidence with our use of the term “key.” And parts of the key may be literally invisible: you see (say) a local shopkeeper cordially greeting a police officer, but as the two shake hands, each of them sees a fellow mason.
What is important here is that accommodation is a process that can cause things to become both inwardly and outwardly meaningful.
Accommodation supports feedback loops of group adoption of some styles, and group avoidance of others. But meaningfulness as we are using it here is a slippery concept, and so let us say a little more about how we are using it.
In talking of meaningfulness, we are not talking about the meanings of formal semantics as developed in the Western analytic tradition from Frege on, or at least not just about these post-Fregean formal semantic meanings.
It is clear that in ordinary talk the word “meaningful” is ambiguous: it certainly implies symbolism, significance, or richness of resonance, but can also be used in a valuative sense. For something to be meaningful in this sense is akin to it being resonant as we defined it in section 3.9. There we said that something resonates (positively) for a group or individual to the extent that it induces increased (positive) harmony for them. But we want to suggest that this only partially captures what we are trying to get at by meaningfulness in this section, which has to do with identity.
Meaningfulness: Something is inwardly (/outwardly) meaningful for some individual or group to the extent that it resonates for them (/for others) in a way that activates attunements that are distinctive of their identity.
The definition is intended to allow for people have complex or perhaps even multiple identities. Crucifixes might be inwardly meaningful to someone qua their Catholicism, but a particular crucifix might be especially meaningful to them because they have worn that particular object for many years, or because it has other important personal associations connected with its history. To see what we mean by outward meaningfulness, consider the ambiguity of the old proverb “the clothes make the man.” It can be understood both to mean that your clothes signify your character to others, and that your clothes are part of who you are (cf. the similarly ambiguous “you are what you eat”).31 The first is outward meaningfulness, and the second inward meaningfulness.
The definition of meaningfulness also allows that it is not merely objects that are meaningful.
Practices, too, can be meaningful. In the case of clothes, it might be that the wearing of certain types of clothes is more meaningful than the clothes themselves. More generally, any behavior can become meaningful. An example of central relevance for us is that talking in a certain way can be meaningful, whether talking in a way that is a hallmark of origin, gender orientation, class, or education.
Note also that although we have talked in this section about what is meaningful in a positive way, our definition of what it is for something to resonate (in section 3.9) allows something to resonate negatively when it leads to dissonance for a group. Hence, inward and outward meaningfulness can also be negative, as when something makes negative stereotypes associated with the group salient. Racist imagery and paraphernalia involved in racist practices can be meaningful in this negative way. Perhaps the most shocking examples are duplications of paraphernalia that has been involved in violence against members of the group, such as nooses, torches, and clothing typical of lynchings in the United States. A racist might tell a gas-chamber joke knowing exactly how it might be meaningful to any Jews who hear it, the way in which it will resonate negatively, the dissonance it will cause.
Let us return to the idea suggested by William James that we depend for our own sense of harmony on the attention and responsiveness of others with whom we are affiliated. Other than by performing explicit acts of reassurance, how can you provide evidence to someone that you are paying attention? And what is the simplest way of being responsive to them?
Here’s a general way of achieving both, already hinted at by examples we have considered: do whatever they do, mimicking any aspect of their behavior that isn’t independently governed by the needs of your role in the interaction. If they cross their legs, you cross your legs; if they talk fast, you speed up; if they smile, you smile; if they say it was “absolutely awful,” then as you nod and make a pained expression, say “absolutely awful!” The chameleon effect, that is, short-term behavioral mimicry extending to arbitrary behaviors such as foot shaking that have no conventionalized communicative significance, is known to be largely automatic and unconscious.32 Nonetheless, such mimicry logically requires a basic level of attention to your interlocutor’s behavior and constitutes the simplest and most flexible response strategy that establishes such attention: any time you either don’t have a conventional response to your interlocutor’s behavior and don’t have the time to compute a new response, repeating what they did will not only show that you were attending at some level, but will also show by direct demonstration that you are attuned to their behaviors. This type of mirroring also serves to make interlocutors feel good, a case of what is described in the pragmatics literature as positive politeness. By acting the same way that someone else acts, you show implicit approval of that way of acting, and register your similarity and hence closeness to them, mitigating the negative effects of any demands you may place on them.
As regards coordination, mimicry of speech practices amounts to adopting common communication protocols, and a background of bodily norms against which to evaluate gestures. The robustness of communication depends upon people not simply using a one-size-fits-all set of conventions for every aspect of communication, but rather adapting as needed. In an environment where someone is not able to hear you clearly, it may be important to speak slowly and loudly, and when speaking to a child, it may be important to use basic vocabulary. A good general way of adapting to the needs of your interlocutor is to adapt your speech to be more like theirs.
There’s no better way of being sure they will be familiar with the words you use than by choosing words that you have heard them say, and a decent way of choosing topics they will be interested in hearing about is choosing topics you’ve heard them discuss. Indeed, the associative workings of the mind mean that often both you and your interlocutors will be primed by the words and topics already mentioned, as well as by other aspects of the commonly accessible speech situation, meaning that there will be a happy coincidence between what one of you finds easy to produce and the other finds easy to comprehend. Furthermore, since speakers of the same language in the same speech situation will tend to exhibit similar priming effects, they will have a natural tendency to use similar words. Thus priming supports convergence of speech behavior, while simultaneously supporting success in coordination.
The upshot of all these considerations is that there are functional pressures for behavioral convergence of group members not only at the level of identity marking, but also at the level of individual conversational exchanges. This is a central plank of Communication Accommodation Theory, initially focusing on accent patterns in speech.
In some of his earliest work on the topic, Giles recognized the complementary possibility of divergent behavioral accommodation (to be discussed in the next section), and he explained convergent behavioral accommodation in terms of the need for social approval: If the sender in a dyadic situation wishes to gain the receiver’s social approval then he may adapt his accent patterns towards that of this person, i.e. reduce pronunciation dissimilarities—accent convergence. On the other hand, if the sender wishes to dissociate himself from the receiver (maybe because of unfavourable characteristics, attitudes, or beliefs), then there may exist tendencies opposed to the receiver, i.e. emphasize pronunciation dissimilarities —accent divergence.
33 Giles’s earliest studies showed convergent behavioral accommodation among Welsh speakers of English, who, when positively disposed toward their interlocutor, used speech features that were more typical of the English of Wales when talking to other Welsh people, and speech features that were more typical of the English of England when talking to English people.
That is, various speech features converged with those of the interlocutor, whether Welsh or English.34 Similar results were found in an experimental study of Canadian French-English bilinguals, in which experimental subjects were presented with recordings by a nonpresent speaker and then asked to record a description for the same speaker. The more the recording involved perceived adaptation to the style of the subject, the more the subject expressed approval of the speaker, and the more the subject adapted their own speech to the perceived style of the recording they’d heard.35 Let us define convergence and divergence in a way that encompasses accommodation across logico-linguistic and social traditions: Convergence/divergence: Accommodation is convergent toward a second party (an individual or group) when it results in a monotonic increase of the level of collective attunement with that party, divergent when it results in a monotonic decrease of the level of attunement with that party among some system of attunements, and mixed when there is a combination of convergence and divergence to the same party.
On this definition, accent convergence or divergence would involve changes in a system of short-term dispositional attunements governing active speech production, changes that result in a net increase or decrease in the similarity of relevant phonetic features (say, features of a vowel affected by placement of the tongue during production).
In the decades following Giles’s introduction of accommodation as a socially motivated behavior among interactants in a speech situation, Communication Accommodation Theory has been vastly extended. It now includes matching of arbitrary aspects of communicative style and posture, for example, gesture, gaze, bodily position, code-switching, word-choice, syntactic construction, and speech rate, and even performance of complex acts such as telling jokes or making personal disclosures, and other behaviors that can be understood in terms of positive politeness. Many instances of behavioral accommodation have been documented in political contexts, for example, when politicians speak with a more Southern accent to sound like children of the South, when they drop the “g” in “-ing” to sound folksy, 36 or when they dress down and roll up their sleeves to look more like the people they’re speaking to.
Studies of word use show that across a very wide range of conditions, people “style match” lexically, meaning that as an interaction proceeds they increasingly use a common vocabulary.
37 Since this vocabulary includes functional words that correlate with the syntactic constructions being employed, this same data is also indicative of matching at the level of syntactic complexity and construction choice. Notably, lexical stylematching effects are found even in somewhat adversarial situations, such as business negotiations, hostage negotiations, and police interviews.38 Of course, the fact that a situation is adversarial does not imply that there is no desire at all for mutual social approval, and certainly doesn’t imply that there aren’t other social benefits to be gained from coordination, which is essential to successful negotiation. In non-adversarial situations, there is evidence (though we must be careful in moving from correlation to causation) that lexical style matching is functionally expedient: groups engaged in problem-solving tasks in which style matching occurs tend to be more productive.39 Style matching is positively correlated with whether people who have just met will go on to form a romantic relationship, and is correlated with measures of how well a relationship is going.40 The literature on mirroring and mimicry includes study of the underlying neurological processes involved. It is not our goal here to study the neurological underpinnings of mirroring behaviors, but we think it’s worth considering whether there might be general principles underlying those behaviors, a reason why mirroring makes sense as a part of normal communication. Let us suggest briefly a line of explanation that is complementary to explanations in terms of marking of politeness or group identity, although it rests on principles that are used in Peter Burke’s development of Identity Theory.
41 Burke adapted an approach from Perceptual Control Theory42 in which perception is modeled in terms of a feedback loop. Rather than just passively waiting to receive signals, the perceiving animal actively adjusts its physical state and perceptual processing so as to maintain constancy of the stimulus, so that at least some aspects of perceptual processing become a special case of the broader biological process of homeostasis. Simple examples of perceptual feedback are the dilation of the pupil in response to reduced light or the tracking with eyes or head of a moving object so as to maintain a constant position in the (highest acuity part of the) visual field.
For identity, Burke cites an example that had previously been studied experimentally, the example of people who perceive themselves as dominant engaging in more dominant behaviors when feedback suggests they are perceived as submissive.
Burke’s model can be seen as cashing out the idea we began this section with, from William James, that our social selves are reflections of the views of others on which our positive self-esteem depends, so that failure to reflect what we aspire to causes dissonance. This is made explicit in the following passage from Michael Hogg, Deborah Terry, and Katherine White: The perception that one is enacting a role satisfactorily should enhance feelings of self-esteem, whereas perceptions of poor role performance may engender doubts about one’s self-worth, and may even produce symptoms of psychological distress.… Distress may arise if feedback from others—in the form of reflected appraisals or perceptions of the self suggested by others’ behavior—is perceived to be incongruent with one’s identity. According to Burke …, identities act as cybernetic control systems: they bring into play a dissonance-reduction mechanism whereby people modify their behavior to achieve a match with their internalized identity standards.43 Let us speculatively suggest how a feedback model like Burke’s, and like that found in a host of technologies since the invention of negative feedback control almost a century ago, might apply to mirroring within an interaction. As well as observing others, people self-monitor. Suppose then that, following general principles of biological systems, people wish to maintain a relatively constant perceptual environment, due to the higher processing load brought about by perceptual variation, and the greater difficulty of identifying meaningful signals when set against a variable background. In that case, they will make whatever behavioral changes they can to control and reduce such variation. It immediately follows that whenever there is a discrepancy between the signal they obtain by selfmonitoring and the signal they maintain by monitoring another, they will feel a pressure to reduce that discrepancy, and will adjust their behavior until the perceptual difference vanishes. So if you hear yourself talking faster than your interlocutor, you will feel a pressure to slow down, and if you perceive a difference between the sound or effect of the words you are using and the sound or effect of the words someone else is using, you will feel a pressure to make your words more like theirs. If someone raises their voice, you talk louder, and vice versa … “Why are we whispering?”44 Independently of whether anything like the feedback mechanism we have described is a useful way of thinking about the types of mirroring behaviors that form the simplest cases of behavioral accommodation, it gets at an important general point. Convergent behavioral accommodation, which is a process of harmonization within a group of conversational interactants, might not merely be polite in-group marking, but is also functionally advantageous. So far we’ve suggested some low-level reasons why it might be functionally advantageous, related to the ability to recognize signals: behavioral convergence reduces distracting variation, makes the communicative situation more predictable, and provides a baseline from which potentially communicative discrepancies in behavior can be judged. But there is also a higher-level functional reason for behavioral convergence that is higher-level insofar as it concerns complex interactional practices that are used to perform arbitrarily complex communicative actions, and not merely duplicated individual behavior indicative of identity and rapport. The higher-level function is the coordinative development of the communicative practices themselves.
Theoretical linguists and analytic philosophers have long been enamored by generative models of language, models that allow a finite formal system to produce an infinite number of forms, and express an infinite number of meanings. As if the infinities in such models were such a big deal! The problem is that the infinities in question are not nearly big enough.
Consider the fact that no two adult speakers have the same lexicon.
If Jason uses a word in a way that is unfamiliar to David, David’s supposed ability to give meaning to an infinite number of forms will not be of direct use, since he won’t be able to give meaning to the form in question. Looked at this way, infinite generativity could be as much a handicap as a benefit, for it is a corollary of infinite generativity, and one that on the basis of personal experience seems entirely plausible, that there are an infinite number of sentences in Jason’s grammar that David cannot completely understand.
Generalizing, not only does every speaker’s system of grammatical attunements differ slightly from that of every other speaker, but every interaction places novel demands on the interactants. These demands cannot be met by deployment of a fixed tool-for-every-occasion Swiss Army knife bequeathed to us by our forebears, however ingenious, but rather require processes of dynamic optimization of communicative practices.
These processes have been studied extensively in work of psychologists of language and sociologists and sociolinguists in the Conversation Analysis tradition, work that focuses not on the infinity of forms and meanings that an individual speaker can produce or comprehend, but on the patterns of interaction seen in individual conversations. An important result from that work (although it seems obvious after the fact) is that rather than simply using a common stock of conventions, people innovate new conventions during their conversations.
For example, when faced with a novel task that involves selecting shapes, people collaboratively develop efficient descriptive language, starting off with longer descriptions, and honing in on shorter, name-like descriptions.45 Both explicit and implicit negotiation of meaning are seen in studies of dialogue.46 Explicit negotiation might involve people proposing a practice, perhaps saying what they mean or what they are going to mean by an expression, but explicit negotiation may also be part of a repair sequence.
After someone has signaled failure to understand, there might be a corrective move with clarification or substitution of an alternative expression, and sometimes metadiscussion of how expressions will be used.
Negotiation of meaning is implicit when a use of an expression that is novel to some audience member is unannounced, and the novelty is not explicitly acknowledged. In such cases, what grounds47 the conversational move— that is, establishes that there is joint understanding of what move has been performed—is at most a general acknowledgment of understanding, perhaps a back-channeling nod or confirmatory vocalization, and perhaps merely the absence of a furrowed brow or other manifest signs of confusion.
Accommodation is harmonization to context. When the context is a task you are trying to complete, dissonance would be created if your language did not allow you to perform that task efficiently. The dissonance is avoided by dynamically altering dispositional attunements to match the needs of the context, for example, by developing a disposition to refer to a salient object using a name or short description. Since the bespoke language behaviors people use in response to short-term conversational needs are adaptations that are not entirely predictable on the basis of broader community practice, they involve behavioral accommodation.
Any novel communicative behavior extends the prior history of language practice with an instance that diverges from what the hearer has previously observed. The hearer must then find a consonant way of adapting to the novel behavior in the conversational context, and this adaptation is accommodatory. The hearer both accommodates to the particular instance of a novel behavior by finding a consonant understanding of the behavior, and changes their dispositions to understand and possibly also produce the behavior they have observed. The hearer presumably also accommodates cognitively, recognizing how the new practice functions, and adapting to any further presuppositional resonances of the new behavior. The bottom line is that the process of implicit negotiation of meaning, and more broadly implicit negotiation of novel language practices, is one of accommodation by interlocutors. We use the term collaborative language accommodation to refer to collective harmonization around novel language practices, whether primarily implicit or involving overt metalinguistic moves of negotiation. The takeaway from this discussion is that the robustness and flexibility of human language rest not on the infinite generativity of fixed individual grammars, but on a social trait of humans: accommodational adaptability. In what follows, we will suggest how the alliances and group rivalries of the social and political world depend in large part on people’s choice or ability to engage jointly in collaborative language accommodation.
6.5. Marching to Your Own Drum The Communication Accommodation Theory literature is replete with examples of people not accommodating to their interlocutors, or accommodating in a way that is somehow suboptimal. While all such behaviors have sometimes been labeled nonaccommodation, most writers currently speak even of people whose communicative adaptations result in increased behavioral distinctiveness from their interlocutors as accommodating, hence our talk of divergent accommodation. The literature also includes many examples of people underaccommodating, so that adaptations to the interlocutor can seem perfunctory, or overaccommodating, as when a caregiver adopts a style that reflects in amplified form the slowness or limited active vocabulary of an elderly person, or misaccommodating, which can happen when someone adapts to their stereotypical preconception of how their interlocutor speaks rather than adapting to actual speech behavior.
48 Our definition of accommodation does not require that interactants converge as regards thought or behavior. However, we suggest that accommodation always involves convergence, even if it produces divergence among interlocutors in a particular speech situation. In terms of our above definition of convergent, divergent, and mixed accommodation, divergent accommodation with one party should then involve convergent accommodation to another. So our claim that accommodation always involves convergence does not entail that all divergent accommodation is in fact mixed accommodation (although this is also an interesting hypothesis), because mixed accommodation, as defined, involves simultaneous convergent and divergent accommodation to the same party.
Consider again the paradigmatic example from Communication Accommodation Theory: a Welsh person is in conversation with an English person, finding the latter (or what they represent) to be irksome. In such cases, Welsh people sometimes speak with more distinctively Welsh English speech than they would otherwise, thus diverging from their interlocutors. Similarly, people often adopt a speech style that emphasizes the characteristics of their own gender identity when speaking to someone presenting as having a distinct gender identity. While it is logically possible for someone to diverge from their interlocutors without doing so in a way that conforms to any existing pattern, recent work in Communication Accommodation Theory suggests that divergent accommodation from an interlocutor is best seen as involving convergence to some other attractor.
This attractor would be a group toward which accommodators (in the above example, the speakers) have greater affinity than they do to those they diverge from, and with whom they wish to mark their common identity.
One of many overviews of Communication Accommodation Theory puts the idea as follows: Convergence is defined as a strategy through which individuals adapt their communicative behavior in such a way as to become more similar to their interlocutor’s behavior.
Conversely, the strategy of divergence leads to an accentuation of differences between self and other. A strategy similar to divergence is maintenance, in which a person persists in his or her original style, regardless of the communication behavior of the interlocutor. Central to the theory is the idea that speakers adjust (or accommodate) their speech styles in order to create and maintain positive personal and social identities.
Giles (1978) also invoked Tajfel and Turner’s (1979) social identity theory of intergroup relations (SIT), and [Speech Accommodation Theory] thereafter has largely (but not solely) relied on the framework of SIT to explain the motives behind the strategies of divergence and maintenance. Why should one choose to appear dissimilar to another? Referring to similarity-attraction theory alone would mean that the motive driving divergence or maintenance behaviors would be to appear dislikable, or at least that the speaker’s need for social approval is low. Invoking the intergroup context, SIT explains the adoption of these strategies through the desire to signal a salient group distinctiveness so as to reinforce a social identity.
49 Although at least some aspects of accommodation may be based in perceptual and behavioral mechanisms that require no attention to or awareness of social function, the idea that divergence is a performance of social identity rather than an arbitrary marking of distinctness suggests to us a generalization: Social Alignment Hypothesis: Accommodation by an interactional participant promotes confidence in the nature of their attunements to identities or to groupings with which coordination is valued.
Convergence is the base case in which only the interactional group and its subparts need to be considered. In this case, accommodation by an interactant increases their attunement (or perception of attunement) to the group of other interactants. Thus the collective attunement of the interactional group is increased. According to the hypothesis, when an agent marks their distinctness from an interlocutor by diverging from them, we need to consider at least two groups. The agent reinforces their identity as a member of another group by convergence to that group, that is, by adopting dispositions, emotions, or attitudes that are distinctively associated with the second group.
Corrollary of Social Alignment Hypothesis: Divergent accommodation from some individual recognized as belonging to a group occurs to promote confidence in membership of an alternative group or commitment to a contrasting identity. Nonaccommodation (maintenance), seen as stubborn refusal to adapt, similarly promotes confidence in affiliation to an alternative group or commitment to an contrasting identity.
Note that divergence of behavior from another interactant need not connote nonmembership in their group(s), but could equally indicate difference of positioning within the same group, for example, within the same family.
This is why we say that accommodation provides confidence in the nature of attunement to identities or to groupings, rather than simply indicating the fact of attunement. The fact that someone behaves so as to mark distinctive identity within a group, or to mark membership of a subgroup, need not indicate any lack of affiliation to the larger group, and need not indicate a lack of cooperativeness. A theory of accommodation must allow for the fact that groups have structure and that individuals have positions within that structure, positions that can be demarcated by distinctive behaviors. A theory of accommodation that uniformly predicted ever more blurring of distinctions of behavior within groups would not be a very good theory.
The social alignment hypothesis leaves open the possibility of mixed accommodation, whereby someone accommodates convergently and divergently with respect to the same person or group. Consider two people with different gender identities in conversation. They may simultaneously exaggerate style features associated with their own gender identities, for example, the pitch of their voice and their body posture, and yet style match on other features, such as speech rate or word choice, to the extent that such matching is perceived as gender neutral. So their pitch and body-posture accommodation is divergent (though converging to their own gender grouping), and their speech rate and word-choice accommodation is convergent.
Such simultaneous convergence and divergence might equally take place with respect to accommodation of attitudes. For example, suppose two mildly antagonistic interlocutors from different social groupings are discussing who they do and don’t hang out with, and one of them says, “I would never be seen hanging out with my loser goody-two-shoes brother!” The addressee might convergently accommodate the belief that the speaker has a brother, and that the brother is somewhat inclined to doing what society, school, or parents expect of him, and yet draw further conclusions at odds with the speaker’s beliefs, for example, that the brother is probably a decent guy and not a loser at all, and perhaps that, contra what the speaker seems to assume, the fact that someone is a goody-two-shoes is no reason not to hang out with them.
We speak of an increase of confidence in attunement, rather than of an increase in attunement simpliciter. This allows for two twists. First, there is the idea considered earlier in the chapter (section 6.1) that by performing an exaggerated form of a practice, by being, as it were, more Catholic than the pope, an actor might better manifest their affinity to a group than by performing a more standardized version of the practice. By performing in a way that overshoots the standard practice of what we can call the target group, that is, the group they are manifesting alignment with, the actor increases the confidence of any spectator in judging that they belong, or wish to be seen to belong, to the target group as opposed to any other salient group, for example, the group to which an interlocutor they are diverging from belongs. Second, and more generally, our reference to confidence allows that people might converge with what they perceive to be the behavior of another interactant or group rather than with the actual behavior. So, some cases in which accommodation is not fully convergent, for example, cases of underaccommodation and overaccommodation, can then be understood not as someone behaving in a way that connotes difference, but rather as someone behaving in a way that connotes similarity. The behavior represents the actor’s best attempt to behave as they think the individual or group they are mimicking behaves.
A different way of putting both twists is that accommodation need not produce convergence to the attunements of a group, but to the attunements of a stereotype of a group. To return to a Wittgensteinian point, a social practice does not have a single standardized form, but rather consists in a history with much variation, and, furthermore, any one individual has only a small window on the history of a practice they are attempting to produce.
So the best anyone can possibly do in producing that practice is perform it in a way that is recognizably like the way they take it to be currently appropriate to perform that practice, and it is inevitable that what they will converge to is not any particular way of performing the practice (for no one way of performing is privileged above others), and is not an ideal version of the practice (there being no ideal), but rather is something like a stereotyped version of the practice that the actor has interpellated. One phenomenon discussed in Communication Accommodation Theory that the social alignment hypothesis seems at first to say nothing about is true nonaccommodation, or maintenance, in which an actor does not adapt their social behavior or speech style in any way in reaction to the particular circumstances of an interaction. Such failures could result from a lack of adaptability, including simple insensitivity (as when someone fails to perceive an acoustic feature that has significance for their interlocutor but not for them), but could also, on occasion, be better seen as performative stubbornness. But what does it mean to say that an actor does not adapt in any way? To say this is to suppose that there is a fact of the matter as to how the actor would have behaved had the situation been in some way normal, and to claim that the actor has not adapted to the difference from normality. But there is no normal. There is just a sequence of interactions that someone experiences in the course of their life, and behaviors that they exhibit in these situations.
Only someone suffering from an extreme clinical condition, such as death, can truly be said not to have adapted to the social situation in which they are located, and we might question to what extent in such extreme cases a person is a full participant in a social interaction.
Everyone else adapts, if only to a small degree, to every new situation. To the extent that the adaptation is small on some measure, that is because only a small adaptation is needed for the actor to maintain harmony. Such cases show not that someone is failing to behave in a way that increases confidence in their identity, but rather that someone is adapting to their interactional situation precisely in the way that maximizes confidence in their identity.
When there is social pressure to adapt, and when an actor is taken to have the power to adapt, that actor’s resoluteness and consistency of selfpresentation become a strong indicator of their confidence and investment in their distinctive presentational style. So, when in this book we talk of nonaccommodation, what we mean is not failure to accommodate simpliciter, but failure to accommodate to some significant subset of the attunements of some particular interactant.
On this view, the very act of maintaining a speech style is a form of accommodation to a group, manifesting our alignment with and positioning within that group, that is, our social identity. Indeed, far from expecting people to mimic their interlocutors faithfully, we expect people to actively maintain their own speech styles, physical deportment, and even attitudes.
People who mimic too much are often seen as ridiculing those they imitate, or else as being Janus-faced, shiftily vacillating, capricious and shapeless, lacking spine.
Contemporary individualist Western mores demand of people that they “be themselves.” But to be yourself, socially at least, is to manifest that you are being yourself, and to manifest a way of being is to use a style that is recognized as a style. But that suggests that even something as supposedly individualistic as being yourself might involve accommodation to some combination of social groupings; it is in the context of other social groupings that ways of being carry any public meaning. We might then say that your accommodation to a social situation is the way in which you adapt to that situation, and if someone appears inflexible or somewhat constant across a wide range of social situations, then we might say that they perform with a relatively narrow band of accommodation to their interlocutors, and are constantly adapting their performance in such a way as to maintain high confidence in their affiliation to a fixed complex of identities.
The suggestion, then, is that we are always accommodating, always performing our social identity with greater or lesser adaptation to the key of those around us. This is not a new thought—to return to Shakespeare, “All the world’s a stage.” Yet Shakespeare’s metaphor might be taken to imply that people have preassigned roles, and that in performing these roles, we follow preexisting scripts determined by the role. Judith Butler famously suggested a far more radical view of performance in developing her strongly anti-essentialist view of gender and other identities as socially constructed. For Butler, not only does the performance of gender establish one’s gender identity, but, furthermore, it is the history of performance of gender identity, and the associated practices that assimilate these performances into a wider web of meaning, that are constitutive of the gender identity itself.50 These ideas, worked out in detail and applied in her epochal Gender Trouble51 and other works, are already clear in a classic 1988 essay, “Performative Acts and Gender Constitution: An Essay in Phenomenology and Feminist Theory,” and indeed are completely explicit in the stunning introduction to that work: Gender is in no way a stable identity or locus of agency from which various acts proceed; rather, it is an identity tenuously constituted in time—an identity instituted through a stylized repetition of acts.
Further, gender is instituted through the stylization of the body and, hence, must be understood as the mundane way in which bodily gestures, movements, and enactments of various kinds constitute the illusion of an abiding gendered self.
… If gender is instituted through acts which are internally discontinuous, then the appearance of substance is precisely that, a constructed identity, a performative accomplishment which the mundane social audience, including the actors themselves, come to believe and to perform in the mode of belief.
… In opposition to theatrical or phenomenological models which take the gendered self to be prior to its acts, I will understand constituting acts not only as constituting the identity of the actor, but as constituting that identity as a compelling illusion.52 As should be clear, our understanding of identity performance, and of practice more generally, is strongly influenced by Butler, although we make no attempt here at a detailed discussion of her program, or a detailed reckoning of what we owe. We will also not explore Butler’s powerful idea that recognition of the performative aspect of gender norms provides a way of undercutting and contesting gender norms, a call to action that she frames at the end of her essay as a warning. She cautions that if the performative nature of gender “is mistaken for a natural or linguistic given, power is relinquished to expand the cultural field bodily through subversive performances of various kinds.”53 6.6. Echo Chambers As we have defined it, a community of practice is an ideologically bound social group. It shares a distinctive set of collective attunements that harmonizes across dimensions, the ideology being precisely these collective attunements. To be a social group, its members must feel identified with the group and the ideology. To be ideologically rich, the group’s collective attunements must include a wide range of attitudes and practices that determine many aspects of how individuals should live their lives and how they should behave socially and politically. Ping-pong players do not form an ideologically rich group. Catholics do.
All ideological groups are harmonic attractors. By this we mean that entering the group resolves some dissonance, and leaving the group produces dissonance. Note that this allows the ideology to be far from any ideal of harmony: it may be that members feel dissonance because of the behaviors of other members of the group, behaviors that may or may not be consonant with the ideology, and it may be that there are contradictions within the collective cognitive attunements (i.e., the belief system) of the ideology, or misalignments of cognitive attunements and emotional or dispositional attunements. What makes the group a harmonic attractor is not that the collective attunements of the group are perfectly consonant, but that leaving the group would produce more dissonance than staying in it.
There is nothing inherently bad about richly ideological communities of practice. They can give life meaning, at least in the restricted sense of meaningfulness that we offered in the last section. Richly ideological communities of practice range from villagers tied together by common heritage and, perhaps, occasionally, willful ignorance of other ways of life, to protest groups tied together by common cause against injustice. They also include cults of many different flavors and degrees of connectedness.
To use the term “cult” is, of course, to make a strongly negative normative valuation. To call a group a cult is to say that it is an ideologically rich grouping for which the collective attunements are reprehensible, or even abhorrent. For example, it may be manifest to an outsider that the cult’s belief system is manifestly inconsistent, that the group is poorly attuned to reality, that the emotional attunements of members are so strong as to prevent normatively desirable deliberation, or that practices of the group are unhealthy or immoral.
Some cults, like the contemporary QAnon movement, are labeled with the term “conspiracy theory.” While there is some merit to the term, to which we’ll return, it is misleading. Calling it a “theory” suggests that what is central to the ideology is a certain set of propositions believed by adherents. If anything, QAnon is not a conspiracy theory but a conspiracy practice, a practice that includes not only various discursive practices but also pseudo-investigative processes for revealing any crumb of evidence, all of which will increase consonance of group members. Some of those discursive practices are for belittling or bamboozling opposition and interpreting counterclaims or negative evaluations directed at the group as evidence of the desperation of outsiders. One prominent set of discursive practices involves the Qdrop, whereby an enigmatic message apparently from the leader Q is released, the message following a template of Delphic inscrutability and pseudoprofundity that is in equal parts guru and Goebbels, an impenetrable call to action to protect us from them, abounding in juxtaposition of highly resonant terms that have no clear extension or connection. The extent to which the discursive practices of Q diverge from mainstream society speech practices is best seen in examples of Qdrops, like this one that we discussed earlier: Symbolism Will Be Their Downfall Their need for symbolism will be their downfall.
Follow the Owl & Y head around the world.
Identify and list.
They don’t hide it.
They don’t fear you.
You are sheep to them.
You are feeders.
Godfather III.54 Given the centrality of distinctive practices in any cult, we make a mistake when we focus on untenable beliefs that cult members apparently hold. It matters not that neither Scientology nor QAnon forms what regular enlightenment-style thinkers would consider a logically consistent set of ideas, for an ideology need not comprise a set of logically consistent ideas.
It needs to be a strong harmonic attractor. Manifestly, both QAnon and Scientology are exactly that. Whatever incongruities we see within the group, their practices mean that members are able to live with any dissonance they feel, and they would experience greater dissonance in transitioning out of the group, into a world that is manifestly hostile to them, than they would by remaining within the fold. They can, furthermore, avoid strong feelings of dissonance if they are able to maintain a landscape of attunements, in the sense of section 3.4, in which inconsistent attunements are separated into separate neighborhoods of thought, and this separation, or fragmentation, can be supported by narratives that explain the appearance of inconsistency. For example, the narrative that the enemy intentionally sows confusion can suggest that it is not worth paying attention to otherwise irremediable inconsistency. Thus the inconsistency may be, to use a standard spatial metaphor for the organization of thought, pushed to the back of one’s mind.
The further point we wish to make about the term “conspiracy theory” relates to the fact that it reflects two-way disdain. When we label someone a “conspiracy theorist,” we are negatively evaluating that person’s own negative evaluation of the actions of some other societally important group.
The term “conspiracy theory” correctly captures both the mistrust felt for some other group and the validation of that mistrust in terms of an analysis of that group’s actions as self-serving. Conspiracy theories are a paradigmatic example of us-them politics. They often go hand-in-hand with hatred: when people hate a group, they interpret the group’s actions as selfserving, and such interpretations justify or amplify the hatred.
It is important to realize that it is social and emotional attunements, especially attunements that might eventually motivate actions in support of a cause, that are of greatest fundamental import politically, not the strange and often inconsistent description of the world that provided the ostensible import of the conspiratorial rhetoric. In this regard, conspiracy theorizing is just one of many associated discursive practices that are politically significant because they engender strongly pro-in-group and strongly antiout-group sentiment and drive attunement toward behaviors that go beyond the bounds of accepted democratic practice.55 But it is also important to realize that when we bandy around the label “conspiracy theory,” we may ourselves be playing a role in us-them politics, accepting that there is an unbridgeable epistemic gulf between us (the clear thinkers) and them (the confused). “Conspiracy theorist” is a slur. That is not to say that the terms “conspiracy theory” or “conspiracy theorist” should not be used, but that we should recognize that to use them is not merely to label, but also to take both epistemic and moral high ground.
Let us say that an antagonistic ideological social group (or, equivalently, an antagonistic community of practice) is a community of practice for which a central feature of the ideology is a strongly negative collective emotional attunement to some other social group. While there have always been such ideologies, Carl Schmitt made explicit the idea that a negative view of others is a distinctively positive or even necessary element of effective government.
56 What he characterized as a friend-enemy distinction was obviously a fundamental plank of Nazism, the Nazis being a paradigm case of a strongly antagonistic, richly ideological social group.
In a polarized political environment, the dominant ideological groupings differ on the most salient social and political issues and practices, and those ideologies are central to identity. It follows that most people will either be well aligned or sharply misaligned on the most salient social and political issues and practices. The resulting social effervescence, in which those with whom one is aligned express aligned views, leads in turn to increasing crystallization of attitudes, by which we mean in-group similarity of attitudes across multiple issues, in-group similarity of discursive and other practices, and widespread inflexibility of both attitudes and practices. We see examples of such crystallization now, for example, in studies showing correlations between attitudes toward race and immigration and attitudes toward climate change, issues that are, on their face, somewhat orthogonal.57 When does a society become politically polarized? The mere presence of antagonistic ideologies is not sufficient.
On the contrary, a nation can become largely united in its antagonism toward an outside group or power or toward an internal group that constitutes only a small minority. This was Schmitt’s point. Polarization involves the growth of an ideologically rich grouping that is collectively strongly antagonistic toward a significant group of others in the same society.
Political polarization occurs when the targets of antagonism themselves represent or are allied to a politically powerful group. In an extremely polarized situation, for some large ideological grouping, the target of antagonism is everyone who is not part of the grouping.
As such a situation is approached, a high level of mutual antagonism between groupings, always present to some extent in a twoparty democratic system, is inevitable, a nadir of us-them politics.
The philosopher C. Thi Nguyen distinguishes between epistemic bubbles and echo chambers.
58 An epistemic bubble forms when the voices of certain groups or institutions are excluded by omission—there is just no access to them. Echo chambers arise when “other relevant voices have been actively discredited.”59 In an echo chamber, you may hear other voices, but you will not give them any credence, as they have been discredited as sources of information. Conspiracy theories are characteristic mechanisms by which to discredit sources of information, and so are central to the formation of echo chambers. Echo chambers, as Nguyen describes them, pose a more serious social epistemic problem than epistemic bubbles. As Nguyen points out, an epistemic bubble can be “burst” by simply introducing omitted information. To dismantle an echo chamber, you have to unravel or bypass the system by which legitimate sources of information are discredited.
How might such a chamber develop? An echo chamber forms when only the practices of insiders resonate strongly inside a group. The problem of echo chambers (as opposed to mere epistemic bubbles) is not merely that other voices cannot be heard or accepted, but that they do not resonate at the right frequency. Yet it is not just voices that are drowned out.
In the echo chambers of extreme ideological groupings, many practices may cease to resonate, including characteristically democratic ones—practices of treating people equitably, or practices of respecting scientific or other kinds of institutionally expert advice.
Looking at echo chambers this way suggests the sense in which impermeability to outside views is not merely failure of uptake. Our model suggests that the following is a fairly clear route to the development of echo chambers: Antagonistic Echo Amplification 1. Ideologically distant groups have incommensurable practices, values, and emotions.
2. In a dialogically healthy environment, such gulfs can be at least partially overcome by joint negotiation of meaning (whether implicit or explicit).
3. But such negotiation of meaning depends on collaborative language accommodation.
4. Behavioral accommodation to the practices of others, a crucial step in collaborative language accommodation, is known to be limited by antagonism between groups. 5. So, in the presence of at least one ideological grouping displaying somewhat generalized out-group antagonism, communicative coordination between members of that group and others will be rare.
6. The social alignment hypothesis predicts that an emotionally polarized environment will become an epistemically and dispositionally polarized environment.60 7. It follows that once a significant ideological grouping displays generalized out-group antagonism, a negative feedback loop will set in, amplifying the echo of collective attunements within the in-group, and causing ever-increasing divergence from other societal groupings.
Note here that while a tendency for divergent accommodation away from out-group interlocutors would exaggerate the speed of the breakdown of deliberative democratic processes, the high probability of nonaccommodation in the face of a disdained out-group member suffices for our argument.
Consider the implications of this argument for the literature on framing.
Absent accommodation, frames and modes of reasoning used by other parties will not be adopted, because they will not resonate (in the sense defined in section 3.9). There will thus be no uptake for ideas couched in those frames or justified by such reasoning. Divergent accommodation of language practices may exaggerate this process further, say, for example, if one party offers pro-choice argumentation, and the other uses pro-life frames, with no possibility for compromise. The problem then is that even when voices that use other frames are heard, as is indeed common in public debate, an ideologically polarized audience will be unable to fully grasp the message of their out-group, and a fortiori that message cannot be taken up.
Consider arguments made by George Lakoff to the effect that if Democrats wanted to persuade people, they should push back on Republican framing with their own frames.61 We find much to like in Lakoff’s discussion, and certainly agree that framing practices are strategically central in politics, as discussed in sections 3.5 and 3.9.
However, if we are right about the role of nonaccommodation in polarization, then Lakoff’s policy conclusions were at a minimum perilous, and possibly even counterproductive. We agree with Lakoff that some sort of pushback on frames is needed. The problem is that if this were implemented simply by having Democratic politicians insist on using preferred Democratic framings, then they would fail to get their message across at all in a polarized environment. If you use your own frames, and those frames do not resonate for the other side, then the other side is unlikely to accommodate. You may as well not have spoken.
Indeed, you may have unwittingly encouraged divergent accommodation, in which case you have only promoted the very polarization you sought to overcome. One way to be heard by someone who is deeply opposed to you is to find a messenger with whom they empathize, a medium that they’ll tune into, and talk to them in their own language. Yet talking to someone in their own language is not so easy. A further conclusion suggested by our argument, though again we might be seen to be stating the obvious, and not offering a concrete way to put it into effect, is that a root cause of runaway echo chamber formation is generalized antagonism, for it is intergroup rivalry that makes incommensurable positions unbridgeable. There is, then, little prospect for voices to be not just heard, but also understood, so long as enmity reigns. If there is to be reconciliation of attitudes, it must go hand in hand with a restoration of empathy for those with different identities.
Restoring empathy and trust, especially when confronted with mechanisms such as conspiracy theories that would need to be unraveled in order to do so, is complex and difficult. Fortunately, such a personal approach is not the only way to address the problem of echo chambers. A social identity is a pattern of attunements, attunements to particular communities of practice, and attunements shared with others in that community. Each of us has several, indeed many, social identities. Each one of us is a member of multiple organizations—a citizen of a country, a member of a synagogue or mosque, a player on a softball team. Instead of going to war with frames, or one by one trying to unravel personal mistrust between individuals, one could try appealing to another social identity, one that is not defined by the problematic echo chamber. Somebody who may be opposed to one politically may share a social identity as a fellow vegetarian, or a fellow athlete, or even a fellow citizen, and can be appealed to by messages that activate shared social identities rather than opposed ones.
Ideally, in times of emergency, a healthy democratic country is one whose citizens come together across these different social identities, their identity as fellow citizens being paramount. Good leadership in a democratic country activates shared identity, so that what people have in common becomes more meaningful than what sets them apart. This is a prerequisite if every voice is to be heard, the fundamental principle of democracy itself.
1. Arendt, The Origins of Totalitarianism, 84–85.
2. Gould and Lewontin, A Critique of the Adaptationist Programme. They observe that certain architectural structures (such as the semitriangular spandrels found at the intersections between domes and supporting columns) emerge out of engineering need, but then become architectural motifs in their own right, even though the original need has vanished. They use this as a metaphor for biological traits that have functions quite unlike those that originally led them to evolve.
3. Borges, Labyrinths, 42.
4. Eckert, Jocks and Burnouts.
5. Saussure, Course in General Linguistics, 120.
6. Bateson, Naven, 175.
7. Becker, “The Nature and Consequences of Black Propaganda.” 8. Stanley, How Propaganda Works, 69.
9. Biographer James O’Neil writes as follows: “So powerful are the effects of the telegeodynamic oscillator,” said Tesla in reviewing the subject in the thirties, “that I could now go over to the Empire State Building and reduce it to a tangled mass of wreckage in a very short time. I could accomplish this result with utmost certainty and without any difficulty whatever. I would use a small mechanical vibrating device, an engine so small you could slip it in your pocket.” (The Life of Nikola Tesla, 144) 10. Guess and Lyons, “Misinformation, Disinformation, and Online Propaganda”; Jayamaha and Matisek, “Social Media Warriors.” 11. Paul and Matthews, “The Russian ‘Firehose of Falsehood’ Propaganda Model.” The firehose of falsehood differs in many ways from the black propaganda of earlier years, for example, because it does not uniformly target an enemy but rather targets Russians and foreigners alike, and because at least part of its function seems to be to create a situation in which it is hard to know what information can be trusted or which sources are credible, thus effectively blanketing populations in such a layer of disinformation that democratic decision-making is infeasible. However, there are strong similarities with black propaganda, e.g., in the use of professional Internet trolls and bots that infiltrate online discussion groups and social media, presenting themselves as in-group members while deploying subversive messages.
12. Brown, “How Shall a Thing Be Called?,” 14–15 and 18–19.
13. Quine, “Translation and Meaning,” 28–33.
14. The question of the extent to which linguistic practices build on a common genetically inherited substrate is a matter of extreme controversy. We do not take a stand here. Nothing we say about the development of attunement toward practices depends on a particular level of innate shared disposition, or indeed on any significant level of predisposition toward specifically linguistic practices. An example of evidence of linguistic sensitivity in young infants, suggestive of the interpretation that significant aspects of human linguistic competence is in some way already built in at time of birth, is found in Waxman and Markow, “Words as Invitations to Form Categories.” 15. Langton and West, “Scorekeeping in a Pornographic Language Game”; Sbisà, “Ideology and the Persuasive Use of Presupposition.” 16. Michael Barnes also rightly argues that accommodation must be gradual in order to account for the phenomena it is being used by Langton, Sbisà, and others to explain. However, Barnes does not supply a model of accommodation that is gradual; see his “Presupposition and Propaganda.” 17. Scheff, Microsociology.
18. Randall Collins, Interaction Ritual Chains.
19. Randall Collins, “On the Microfoundations of Macrosociology,” 984.
20. Lewis, “Scorekeeping in a Language Game.” 21. Giles’s introduction of what is now known as Communication Accommodation Theory began with his thesis, “A Study of Speech Patterns in Social Interaction.” A standard overview is Giles, Coupland, and Coupland, “Accommodation Theory.” Note that both Giles’s work and the recent discussion of accommodation in psychological and computational analysis of dialogue (e.g., Doyle and Frank, “Investigating the Sources of Linguistic Alignment in Conversation”) is substantially predated by the use of the term “accommodation” popularized by Jean Piaget in his developmental work in the 1950s. For Piaget, accommodation concerns children’s adaption of their mental schemas to fit features of the environment. Piaget is perhaps a direct antecedent of both Giles’s and Lewis’s usage of accommodation, since Piaget explicitly mentions accommodation in social interaction, commenting, e.g., “accommodation to the point of view of others enables individual thought to be located in a totality of perspectives that insures its objectivity and reduces its egocentrism” (The Construction of Reality in the Child, 356–57).
22. James, The Principles of Psychology, 294. James goes on to distinguish one particular social self: “The most peculiar social self which one is apt to have is in the mind of the person one is in love with. The good or bad fortunes of this self cause the most intense elation and dejection— unreasonable enough as measured by every other standard than that of the organic feeling of the individual. To his own consciousness he is not, so long as this particular social self fails to get recognition, and when it is recognized his contentment passes all bounds” (294). That is, inner harmony and relationship harmony can sometimes be hard to distinguish. 23. Good introductions to Social Identity Theory and Identity Theory are, respectively, Tajfel et al., “An Integrative Theory of Intergroup Conflict”; Burke, “Identity Control Theory.” We strongly recommend the overview of both in Hogg, Terry, and White, “A Tale of Two Theories.” 24. Brewer, “The Social Self”; Hornsey and Jetten, “The Individual within the Group”; Leonardelli, Pickett, and Brewer, “Optimal Distinctiveness Theory.” 25. Durkheim, The Division of Labor in Society.
26. Scheff, Microsociology, 5. Note that although the notion of attunement mentioned in this quote plays a similar functional role in Scheff’s theory as it does in ours, the two notions are defined quite differently. Scheff’s attunement is a relation between people who may be attuned to each other to a greater or lesser degree if they are jointly paying attention to the same things and have shared understanding. On our view, attunement is a relation between an individual (or group) and anything else, with attunement to another individual (or group) being a special case. Thus for us, people being attuned to each other implies that they have common attunements to other things, which is not how Scheff uses the term. He does however talk of “attunement of thought and feeling” (Microsociology, 106), so our talk of cognitive and emotional attunements can be seen as an extension and to some extent precisification of Scheff’s terminology.
27. Bandura, Ross, and Ross, “Transmission of Aggression.” 28. For levels of physical aggression, kids in general took their cues more from male investigators than female investigators. Whether this is because aggressive females were seen as acting against stereotype, or because female investigators are generally mimicked less in such experiments, or whether it is just statistical happenstance, we leave open.
29. For adults, see, e.g., Yabar et al., “Implicit Behavioral Mimicry.” Imitation effects are strong among those facing ostracism from in-group members: Lakin, Chartrand, and Arkin, “I Am Too Just Like You.” For infants and children, studies showing effects of group membership on mimicry include Watson-Jones, Whitehouse, and Legare, “In-Group Ostracism”; Buttelmann et al., “Selective Imitation”; Howard et al., “Infants’ and Young Children’s Imitation of Linguistic In-Group and OutGroup Informants”; Genschow and Schindler, “The Influence of Group Membership on CrossContextual Imitation.” 30. The classic work taking this anti-essentialist approach to gender is that of Judith Butler, in particular her book Gender Trouble.
31. The most natural interpretation for “the apparel oft proclaims the man,” Pollonius’s advice to his departing son, in Shakespeare’s Hamlet, is that the clothes signify the nature or status of the wearer, but the sophisticated use of clothing to represent character in Hamlet is open to deeper analysis. Brillat-Savarin’s eighteenth-century “Dis-moi ce que tu manges, je te dirai ce que tu es” (Tell me what you eat, and I’ll tell you who you are), although ambiguous, suggests that your eating habits are diagnostic of, and perhaps central to your identity, while Feuerbach brought out a distinctly metaphysical interpretation in his later punning variant “Der Mensch ist, was er isst” (A person is what they eat). (Our thanks to Hank Southgate for discussion.) 32. Chartrand and Bargh, “The Chameleon Effect.” 33. Giles, “Accent Mobility,” 90, emphasis in original.
34. It is tempting to describe this situation, somewhat colloquially, as Welsh speakers speaking with less of a Welsh accent when speaking to English people than when speaking to Welsh people.
This description, however, builds in a presupposition that the dialect spoken in Wales involves an “accent” relative to so-called “Received Pronunciation” paradigmatically spoken in England, which is then presupposed to be neutral. To talk of one dialect being more neutral than another, is, however, to make a political statement, and to presuppose such neutrality is to leave the same politics perniciously implicit.
35. Giles, Taylor, and Bourhis, “Towards a Theory of Interpersonal Accommodation.” 36.
Purnell, Raimy, and Salmons, “Defining Dialect, Perceiving Dialect, and New Dialect Formation.” 37. A good starting point to what is termed Language Style Matching (LSM) in the literature, but which focuses on lexical similarity, is Niederhoffer and Pennebaker, “Linguistic Style Matching in Social Interaction.” 38. For a study of negotiation in an experimental setting, see Ireland and Henderson, “Language Style Matching, Engagement, and Impasse in Negotiations.” For an analysis of police-negotiation transcripts, see Taylor and Thomas, “Linguistic Style Matching and Negotiation Outcome.” Police interview data is reported in Richardson et al., “Language Style Matching and Police Interrogation Outcomes.” For a final case of adversarial lexical style matching, consider the fact that participants in a presidential debate style match, and that the degree of style matching is predictive of audience approval: Romero et al., “Linguistic Style Matching in Presidential Debates.” See also CarreraFernández, Guàrdia-Olmos, and Peró-Cebollero, “Linguistic Style in the Mexican Electoral Process.” 39. Gonzales, Hancock, and Pennebaker, “Language Style Matching.” 40. For the importance of lexical style matching in speed-dating interactions, see Ireland et al., “Language Style Matching Predicts Relationship Initiation and Stability.” A later speed-dating study, by McFarland, Jurafsky, and Rawling (“Making the Connection”), revealed the significance of three types of mimicry (laughter, function word use, and speech rate), but the picture is complicated both because the men and women do not value mimicry in the same way, and because the models include many other factors. Correlative evidence of a relationship between lexical style matching and relationship health is found in three archival studies in Ireland and Pennebaker, “Language Style Matching in Writing.” 41. Burke, “Identity Processes and Social Stress.” 42. For Perceptual Control Theory in the words of its somewhat iconoclastic primary developer, see Powers, Behavior: The Control of Perception.
43. Hogg, Terry, and White, “A Tale of Two Theories,” 257.
44. The process of harmonization, and hence accommodation as we use that term, is akin to what Jean Piaget termed equilibration (Piaget, The Development of Thought). Piaget’s discussion of this process reveals that he sees it as a high-level extension of regulatory processes that operate in perception-behavior feedback loops that are biologically fundamental: “Self-regulations are the very nature of equilibration. These self-regulations come into play at all levels of cognition, including the very lowest level of perception” (Piaget, “Problems of Equilibration,” 10). We will not attempt a detailed mapping of Piaget’s triad of assimilation, accommodation, and equilibration into our own use of the terms harmonization and accommodation, and also leave it open how our discussion of imitation might be related to Piaget’s extensive work on imitation in infants.
45. Wilkes-Gibbs and Clark, “Coordinating Beliefs in Conversation.” They give the example of a subject in an experiment involving the description of shaped objects who initially identified a shape by saying “All right, the next one looks like a person who’s ice skating, except they’re sticking two arms out in front,” and later refined the description to “the ice skater” (“Coordinating Beliefs in Conversation,” 28).
46. A good general discussion of the processes involved in multiple types of communicative coordination is Pickering and Garrod, “Toward a Mechanistic Psychology of Dialogue.” See also Clark and Wilkes-Gibbes, “Referring as a Collaborative Process”; Clark and Gerrig, “Understanding Old Words with New Meanings”; Larsson, “Ad-hoc Semantic Systems,” “Grounding”; Healey, “Expertise or Expertese?”; Hawkins, “Coordinating on Meaning in Communication.” 47. Clark and Brennan, “Grounding in Communication.” 48. One of several good overviews of the relevant literature is Gallois, Ogay, and Giles, “Communication Accommodation Theory.” For extensive annotated references to both the Communication Accommodation Literature and to semantic literature in the Lewisian tradition, see Beaver and Denlinger, “Linguistic Accommodation.” 49. Gallois, Ogay, and Giles, “Communication Accommodation Theory,” 123. The term Communication Accommodation Theory (CAT) has become standardized in place of Speech Accommodation Theory (SAT), as the purview of the work has extended from early work centering on speech styles to a broader consideration of communicative style and performance. The references in the quotation are to Giles, “Linguistic Differentiation between Ethnic Groups”; and Tajfel and Turner, “An Integrative Theory of Intergroup Conflict.” 50. The thought that communities of practice ground social kinds, which are every bit as part of the natural world as other kinds, is explored and defended in detail in the analytic tradition by Sally Haslanger, in Resisting Reality: Social Construction and Social Critique.
51. Butler, Gender Trouble.
52. Butler, “Performative Acts and Gender Constitution,” 519–20.
53. Butler, “Performative Acts and Gender Constitution,” 531.
54. Q, posted on November 21, 2017, https://digitalsoldiers.info/symbolism/.
55. In focusing on the social role of political messaging rather than on belief formation, we follow a line that is clear in Hannah Arendt’s The Origins of Totalitarianism, and which has recently been developed by Megan Hyska (“Propaganda, Irrationality, and Group Agency,” “Against Irrationalism in the Theory of Propaganda”). Drawing on an Arendtian analysis of propaganda as primarily functioning to create or destroy social groups, she makes a distinction between propaganda as “group-forming speech” and indoctrination as “group-addressing speech” (“Propaganda, Irrationality, and Group Agency,” 231). We take it that this is a distinction of function rather than a clear separation of practices, since it is not evident to us why a particular message or systematic messaging strategy could not serve both roles. We would hesitate to classify Qdrops as being exclusively group-forming or group-addressing.
56. Schmitt, The Concept of the Political.
57. Cf. Dan Kahan (“Ideology, Motivated Reasoning, and Cognitive Reflection”) on polarization.
The empirical studies he reports on suggest that alignment of attitudes within groups is largely driven by preferences for commitment to ideas that signal membership of and loyalty toward those very groups. As he writes, “As a form of ‘identity self-defense,’ individuals are unconsciously motivated to resist empirical assertions … if those assertions run contrary to the dominant belief within their groups” (408).
58. Nguyen, “Echo Chambers and Epistemic Bubbles.” We do not adopt Nguyen’s terminology here, so for us an echo chamber need not involve active policing of outside views.
59. Nguyen, “Echo Chambers and Epistemic Bubbles,” 141.
60. That is, when in-group members converse primarily among themselves, there is a welldocumented tendency for their attitudes to become more extreme by accommodating only to each other, and if anything divergently accommodating on those occasions when they are in contact with others. This tendency is seen in Festinger’s work, discussed in section 3.2. More recent studies are summarized by Cass Sunstein as follows: We can sharpen our understanding of this problem if we attend to the phenomenon of group polarization. Found in many settings, it involves like-minded people going to extremes. More precisely, group polarization means that after deliberating with one another, people are likely to move toward a more extreme point of view in the direction to which they were already inclined. With respect to the Internet, the implication is that groups of people, especially if they are like-minded, will end up thinking the same thing they thought before—but in more extreme form, and sometimes in a much more extreme form.
Consider some examples of the basic phenomenon, as studied in more than a dozen nations. (a) After discussion, the citizens of France become more critical of the U.S. and its intentions with respect to economic aid. (b) After discussion, whites predisposed to show racial prejudice offer more negative responses to the question of whether white racism is responsible for certain conditions faced by African-Americans. (c) After discussion, whites predisposed not to show racial prejudice offer more positive responses to the same question. (d) After discussion, a group of moderately pro-feminist women become more strongly pro-feminist. (e) Republican appointees to the federal judiciary show far more conservative voting patterns when they are sitting on a panel consisting solely of Republican appointees; and Democratic appointees show far more liberal voting patterns when they are sitting on a panel consisting solely of Democratic appointees. (Sunstein, “Democracy and Filtering,” 59) We are assuming that the tendencies Sunstein discusses extend to dispositional and emotional attunements, and not merely cognitive attunements.
61. Lakoff, Know Your Values and Frame the Debate.