In her final pitches to voters, Hillary Clinton has been arguing that much of what Donald Trump’s says amounts to "a dog whistle to his most hateful supporters" — as she put it in North Carolina last week.
Meanwhile, reacting to one of Trump’s final ads, which suggests Clinton is a tool of "the global special interest," Josh Marshall of TalkingPointsMemo, finds it "packed full of anti-Semitic dog whistles."
The phrase "dog whistle" has been around for years. It's political shorthand for a phrase that may sound innocuous to some people, but which also communicates something more insidious either to a subset of the audience or outside of the audience’s conscious awareness — a covert appeal to some noxious set of views. Given Trump’s racially charged campaign, and the support he has attracted from fringe groups, including the KKK, it’s not surprising that the phrase has featured so prominently in the 2016 political lexicon.
To be sure, many people believe there is no shortage of overtly offensive content in Trump’s crystal-clear statements — whether he’s suggesting that the typical illegal immigrant is a rapist or stating outright that American Muslims know about terror attacks in advance. Additionally, the philosopher Jennifer Saul has argued that Trump has moved beyond the dog whistle into other forms of barely disguised bigotry.
From "inner cities" to Pepe the frog
Still, every couple of weeks we see a new accusation of dog-whistling leveled against Trump and his supporters — think of "bad hombres," Pepe the frog, "law and order," "inner cities," "America First." One recent and much-discussed example comes from the October 13 speech in which Trump accused Clinton of "meet[ing] in secret with international banks to plot the destruction of US sovereignty."
While many people might hear "international banks" quite literally, or maybe as an allusion to Clinton’s ties to foreign financial interests in general, anti-Semites hear something very different. After all, the supposed existence of a cabal of international Jewish bankers working to undermine US democracy is a recurring theme in American anti-Semitism, from Henry Ford’s The International Jew to Reddit troll-conventions. Trump’s choice of language serves as a signal that he is one of them.
Or at least, that’s what many commentators have alleged. The problem is that it’s hard to establish whether a piece of speech or writing is a dog whistle. Indeed, it’s not obvious what evidence could, in principle, settle a dispute over whether some expression is or isn’t one. They are, by their nature, sneaky things.
Every now and again, a politician might, in a moment of candor, fess up. (David Kuo, a White House staffer under George W. Bush, reports — referring to speeches by Bush — that "we threw in a few obscure turns of phrase known clearly to any evangelical, yet unlikely to be noticed by anyone else." Lee Atwater’s infamous remarks on how to imply the n-word without saying it also come to mind: "You say stuff like, uh, forced busing, states’ rights…")
But accusations of dog whistling are generally met with exasperated denials. Some commentators, like the blogger Slate Star Codex, have even concluded that the concept of the dog whistle is often too vague and open to abuse for it to be useful. (He was especially dubious that Ted Cruz was dog-whistling at anti-Semites when he spoke of "New York values," as opposed to attacking liberal social mores, a staple of Republican rhetoric.)
I disagree. Dog whistle denialists’ concerns are real, but we can answer them by getting a better sense of what types of dog whistles there are and how they might work. They fall into at least three families — semantic, contextual, and stereotype-dependent — each of which depends on a characteristic mechanism and is disclosed by a characteristic type of evidence.
How to think about statements that have multiple meanings
Of course, there are some relatively boring ways that a single utterance or act of writing can be used to mean different things to different addressees. Suppose I’m leading a guided meditation and I say, "You are on a beach." I’m not using "you" to refer to all of the people as a group, and saying that they’re all on the same beach. I’m addressing each person separately, asking her to envision her own beach. Or suppose I record a voicemail message saying, "I can’t make it to the phone now." I’m not using "now" once and for all to refer to one particular time. Rather, I’m using it to refer to something different for each person that hears the message.
In both of these examples, it’s common knowledge among people listening to me that they will interpret what I’m saying differently, and that all of these interpretations are correct. Part of what makes dog whistles interesting, on the other hand, is that they lack this element of common knowledge.
Semantic dog whistles work by exploiting different linguistic conventions among different subsets of a speaker’s audience. It’s the stuff of spy movies. My accomplice and I agree before the mission that she will disable the cameras when I say "salt." At the crucial moment, I say "Please pass the salt" to my mark. The mark hears it, correctly, as a request for salt; my accomplice hears it, correctly, as a directive to turn off the cameras. As in the guided meditation and voicemail cases, I use a single utterance to perform different speech acts for different addressees. But in the secret little linguistic community made up of me and my accomplice, "salt" has a special meaning.
Moving from secret codes to contextual cues
Take the use of the word "coincidence" as a dog whistle. In the recent kerfuffle over the "Coincidence Detector" app, many news readers learned that internet anti-Semites use "coincidence" to mean, roughly, a Jewish conspiracy. Until the story broke, this subculture could go around calling people and events "coincidences" with impunity – pointing to an anti-Trump article by a Jewish author, for example — maybe confusing readers who didn’t share their views, but not raising any hackles.
That said, I would conjecture that most dog whistles don’t work in this way. Dog whistles only work as long as most people don’t know about them. But "coincidence" has been ruined as a dog whistle by a single news story. Semantic dog whistles depend on secret codes, but it seems that those codes are pretty easy to crack.
I would argue that most dog whistles do not depend on a secret code. Rather, many derive their efficacy from features of the context, broadly speaking, in which they’re most at home. In particular, an expression might make an effective dog whistle because of a) the linguistic constructions in which it is especially likely to appear, b) the perceived character of its typical users, or c) the interaction types in which it typically occurs.
Some words and phrases acquire a kind of emotional charge from their collocates, linguist jargon for the verbal company they keep. Linguists call this emotional charge "semantic prosody." Consider the phrase "women and children." On the face of it, an English language learner might think this is simply a noun phrase referring to a group of people. But look at the larger phrases or sentences in which this phrase is actually embedded. The first three hits in a random selection from the online Corpus of Contemporary American English show that it is typically used to refer to this group of people as the victims of some form of mistreatment – "the connection between women and children and poverty"; "the violence, especially directed at women and children"; "the injured — a majority of them women and children." Because of the contexts in which it occurs, the phrase is saturated with the emotions we feel about innocent victims of violence and injustice.
The case of "Barack Hussein Obama"
"Women and children" has this type of emotional resonance for all speakers of standard American English. It is likely to be used in victim-related contexts across genres — national print and TV journalism, especially — that are regularly consumed by most American English speakers.
But there might be other words and phrases whose semantic prosody varies across varieties of English to which different people are differentially exposed. I think the right’s use of the president’s full name — "Barack Hussein Obama" — falls into this category (to the extent that it isn’t a shamelessly overt attempt to make people think Obama is a Muslim). Among people who consume media that mentions Muslims primarily in contexts that stoke fear and distrust, the name "Hussein" will evoke those attitudes. Among people whose media diets (and personal experiences) contain more positive representations of Muslims, the name "Hussein" might remind them of the late Iraqi ruler, but perhaps won’t carry the same sort of emotional weight.
Other contextual dog whistles work because of the type of person they are typically used by. They are tools of what the sociologist Erving Goffman called impression management, or what Aristotle called ethos— the shaping of an audience’s perception of your character for persuasive purposes. Some expressions are used only (or primarily) by certain groups, and so their use can signal membership in that group. Sometimes it’s common knowledge that a group owns an expression — think of "crony capitalism" (the left) or "nanny state" (the right).
But often only the group members themselves know that they own the expression. In her 2014 study of religious dog whistles, political psychologist Bethany Albertson used fake political campaign messages to compare how people reacted to overt religious appeals and to covertly religious uses of the phrase "wonder-working power" —which appears in the evangelical hymn "There is Power in the Blood" and is known primarily by evangelicals. Overall, nonreligious subjects disliked the overt religious appeals when they encountered them in political messages, but didn’t mind the use of "wonder-working power." Religious subjects, on the other hand, appreciated both the overt and the covert religious appeals.
"Wonder-working power" seems to be a signal of group membership for evangelicals, which, for the most part, only evangelicals can hear. As the legal scholar Ian Haney-López suggests, this is likely the mechanism at work in Hillary Clinton’s recent adoption of insider anti-racist lingo like "implicit bias" and "systemic racism."
Yet other dog whistles might work because they are typically used in certain sorts of interactions. Sometimes speakers will use a certain expression, or even a certain language, in order to "reframe" an interaction — that is, to encourage their audience to categorize the interaction in a certain way, and to act accordingly. Sociolinguists call these devices contextualization cues.
Code-switching doctors and "welfare queens"
One example can be found in Frederick Erickson’s Talk and Social Theory: a medical intern is giving a report to his supervisor about a patient he’s just examined. They go back and forth, in dry, technical language, until the intern mentions the patient’s $30 a week marijuana habit. The supervisor then smiles and asks, "How much is that?" It’s clear enough what he’s up to. With the switch to casual speech, he’s trying, at least for a while, to turn a formal, work-related interaction between medical professionals into a more informal interaction between people who are not just medical professionals and colleagues, but potential weed purchasers. A contextualization cue becomes a potential dog whistle, though, when the cue is picked up on differently by different people.
A last group of dog whistles works by an entirely different mechanism — stereotype activation. If the stereotypical F is a G, by disparaging F's, perhaps one can disparage, or at least appeal to audiences who dislike, G's. I think that this is the mechanism underlying the use of "welfare queens" on the right (although, tellingly, I can’t prove it).
For most people, the stereotypical welfare recipient is black, and so politicians can disparage black people, or appeal to anti-black racists, by disparaging welfare recipients. Relatedly, politicians can defend policies favoring an unpopular group by systematically replacing reference to that group with reference to a related group that enjoys a positive stereotype. I suspect this is the mechanism underlying the use of "small business" (as opposed to, say, "international corporations") by capitalists and plutocrats in both parties.
This is a different sort of dog whistle from those above. It works not by the conscious communication of a particular message to a subset of one’s audience, but by the (perhaps unconscious) activation of a stereotype that is likely to be shared by much of one’s audience. To demonstrate with any certainty that an expression is a stereotype-dependent dog whistle, however, we’d better turn to the methods used by psychologists who study these sorts of things — implicit attitude tests and semantic priming experiments, for example, which allow us to identify when people have subtle negative reactions to certain words and ideas, even if they are unaware of those feelings.
We can see that while it might be difficult to establish whether a politician is blowing a dog whistle on any given occasion, it’s not impossible. Exactly what it takes to find out one way or the other depends on the type of dog whistle at work. While we should perhaps use the concept of a dog whistle more cautiously than some headline writers, we shouldn’t throw it out altogether.
Indeed, we need the concept in order to make sense of our political discourse. The stakes are high. Think of the women and children.
The Big Idea is Vox’s home for smart, often scholarly excursions into the most important issues and ideas in politics, science, and culture — typically written by outside contributors. If you have an idea for a piece, pitch us at email@example.com.