A Simple Theory of Overt and Covert Dogwhistles

Abstract

Politicians select their words meticulously, never losing sight of their ultimate communicative goal. Sometimes, their objective may be that of not being fully understood by a large portion of the audience. They can achieve this by means of dogwhistles; linguistic expressions that, in addition to their literal meaning, convey a concealed message to a specific sub-group of the audience. This paper focuses on the distinction between overt and covert dogwhistles introduced by J. Saul (2018). I argue that, even if the distinction successfully captures a genuine divide within the category of dogwhistles, the account proposed by Saul to explain the distinction is unsatisfactory. In response to this state of affairs, I illustrate how the distinction between overt and covert dogwhistle can be refined and illuminated by incorporating it into the 'Simple Theory' of dogwhistles advanced by J. Khoo (2017).

Keywords:
Dogwhistles; Political communication; Hate Speech; Khoo

Introduction

Contemporary political discourse is, in many respects, more akin to poetry than to ordinary speech. The politician selects his words meticulously, never losing sight of his ultimate communicative goal. Surprisingly enough, this goal might be specifically that of not being fully understood by a large portion of the audience. He might, in fact, strategically aim to communicate a concealed message to a restricted segment of the audience without arousing any suspicion of having done so. Linguists and philosophers refer to this phenomenon by figuratively adopting the term 'dogwhistle'.¹ Providing a unitary and perspicuous analysis of the nature of dogwhistles is, however, no easy task. As an initial approximation, we can simply conceive of them as linguistic expressions that, in addition to their literal meaning, can transmit a message accessible only to a specific sub-group (henceforth target audience).

That of dogwhistles, however, is not a unitary category. Indeed, following the taxonomy introduced by J. Saul (2018), dogwhistles can be classified along two distinct dimensions: intentionality and overtness. On the first dimension, dogwhistles are classed as either 'intentional' or 'unintentional'. Intuitively, a dogwhistle is intentional only when its production is a conscious act of the speaker; if, on the other hand, it is issued accidentally, then the dogwhistle should be classified as unintentional.

On the second dimension, dogwhistles are qualified as either 'overt' or 'covert'. This latter distinction, however, is far more challenging to capture in a simple formulation. Indeed, cutting to the very core of this distinction is-in a sense-the ultimate goal of the present contribution. Nonetheless, as a first sketch, we can postulate that a dogwhistle should be classed as overt only if the target audience is fully aware of receiving a concealed message from the speaker; otherwise, the dogwhistle is to be regarded as a covert one.

Since these two dimensions of classification are not mutually exclusive, the upshot is that every dogwhistle can be assigned to one of the four categories shown in the table below:

Fig. 1

The present paper buys into the idea that philosophy progresses through piecemeal contributions that, when (and if) put together, lead to significant septs forward. In what follows, I shall in fact focus exclusively on the distinction between overt and covert dogwhistles, illustrating how it can be refined and substantiated by incorporating it into the "Simple Theory" of dogwhistles advanced by Justin Khoo (2017). To this end, I shall begin by showing that the distinction between overt and covert dogwhistles introduced by Saul (2018) successfully captures a genuine divide within the category of dogwhistles; a distinction that is reflected in the psychological and linguistic effects they elicit. This will be achieved by presenting and discussing two examples of dogwhistle-one overt (§1.1) and one covert (§1.2)-drawn from the contemporary political scene. Yet, the theoretical framework in which Saul embeds the overt/covert distinction-I turn to argue in §1.3-is somewhat underdeveloped and, as a consequence, inadequate to properly clarify the nature of the distinction.

I shall therefore proceed to demonstrate that the Simple Theory of dogwhistles presented by Khoo (2017) can be successfully expanded to explain Saul’s distinction-this will constitute the central theme of §2. To achieve this goal, I will start by presenting and motivating the approach adopted by Khoo (§2.1). Ultimately, this discussion will set the stage for implementing the expansion that will enable Khoo’s theory to accommodate and clarify the distinction between overt and covert dogwhistles.

1. Overt and Covert Dogwhistles

The purpose of the present section is twofold. On the one hand, I shall suggest (i) that the distinction between overt and covert dogwhistles proposed by Saul (2018) is a perspicuous one; a distinction that, as it were, cuts at joints of the phenomenon of dogwhistling. On the other hand, however, I will maintain that (ii) the theoretical framework in which Saul couches the distinction is significantly underdeveloped and, as such, unable to adequately illuminate the nature of the distinction.

In further detail, in order to establish (i), I shall adopt a bottom-up approach. Accordingly, I will start by presenting two concrete examples of dogwhistles (one overt in §1.1, and one covert in §1.2) drawn from contemporary politics. The purpose of such examples is to illustrate that it is indeed possible and, most importantly, worthwhile to detect a contrast between overt and covert dogwhistles. In pursuit of (ii), I will then turn to present a brief overview of Saul’s own proposed solution to account for the distinction. In this context, I shall contend that Saul’s proposal, although correct in outline, is underdeveloped. More precisely, I will argue that Saul’s account merely records the distinction between overt and covert dogwhistles, without really developing a theory to substantiate it.

1.1 Overt Dogwhistles

Let us begin our discussion by examining a concrete example of overt dogwhistle taken from Italian contemporary politics. In 2018, the Italian Minister of Foreign Affairs Matteo Salvini, in response to a press article reporting on the various attacks he received, posted the following words on his Twitter page:

“Tanti nemici, tanto onore!”

[Many enemies, much honour!]

(Salvini 2018; my translation)

By virtue of what factors-let us turn to enquire-the sentence can be considered an example of dogwhistle? The answer is that the Tweet communicates radically diverse messages to different sub-groups of the audience. Additionally, one of these messages is-in some relevant sense-hidden or concealed (i.e. it is not directly accessible to the rest of the audience). More specifically, although Salvini’s use of the motto 'many enemies, much honour!' might at first appear entirely innocuous, it can plausibly be argued that the phrase is deliberately tailored to win the favour of a circumscribed segment of the electorate. In effect, Salvini’s Tweet mimics a slogan consistently utilised (even if not coined) by the dictator Benito Mussolini, the father of fascism.² Moreover, the connexion between Salvini’s post and fascist propaganda is reinforced when one takes into account the very peculiar date on which the post was published. The Tweet, in fact, was published exactly on the 29^th of July; the date of Mussolini’s birth.

Regardless of Salvini’s original intention, these factors result in the effective transmission of a message to the extreme right-wing of the electorate-those people who are most likely to appreciate the connexion between Salvini’s Tweet and fascist rhetoric. The 'hidden' piece of information delivered-directly available only to the 'trained ears' of a sub-group of the audience-is that Salvini declares to be sympathetic to the ideology embraced by this particular group. It is worth remarking that, in this case, those who can 'decode' the hidden message are consciously aware of being the target of Salvini’s dogwhistle. They know, to put it differently, that Salvini is addressing them, and that he is doing so by endorsing their ideology. This awareness makes Salvini’s Tweet an example of overt (rather than covert) dogwhistle.

Pushing our analysis one step further, we can observe that Salvini could not have communicated this message of allegiance in a fully explicit manner. In effect, he could not because-in terms of Tali Mendelberg’s theory (2001)-there is a social norm in force that rules out the possibility of publicly sharing the ideology of fascism.³ In fact, as Mendelberg’s work highlights, dogwhistles are a powerful tool precisely because they allow a speaker to violate one (or more) existing social norms while retaining the possibility of plausibly denying having done so.⁴ This particular feature-henceforth plausible deniability-appears to be one of the most distinctive properties shared by both overt and covert dogwhistles.

In further detail, plausible deniability can be schematically unpacked as follows: first, assume that there is a social norm N that regulates the behaviour of people living in a given community. In particular, there exist social incentives that prompt members of the community to conform their behaviour to the norm N. The norm N-suppose further-prohibits that a member of the community says that φ Consider a hypothetical linguistic expression D that, by virtue of some mechanism, manages to send different messages to different members of the community. In particular, when a speaker uses expression D, it can be thought either that he meant that ψ, or that he meant that φ. Now, imagine a scenario in which a community member uses expression D in a public speech. Plausibly, some members of the audience will understand that ψ, while some others will understand that φ Yet, if the speaker were accused of violating norm N (i.e. if he were accused of saying that φ), he could plausibly deny having done so by saying that all he wanted to do was to say that ψ. In fact, as mentioned, the expression D does mean either that ψ, or that φ.

Thus, returning to our example, Salvini can dismiss the accusation of sympathising with fascist ideology (which would constitute a violation of an extant social norm) just by saying that all he wanted to communicate is the circumstance that the more enemies one has, the more honourable he is. Such a dismissal-we can add-is particularly plausible precisely because the motto 'many enemies, much honour!' seems to possess a special duplicity that allows it to simultaneously deliver different messages.

1.2 Covert Dogwhistles

Let us now turn to discuss an example of covert dogwhistle. This time, the example comes from contemporary American politics, where the term 'inner city' has begun to be perceived in a very particular manner. In effect, despite its dictionary-meaning being 'densely populated, high-crime, urban area', empirical evidence (White 2007) suggests that the term acts as a dogwhistle for 'African American'.⁵ Thus, thanks to plausible deniability, a politician can strategically adopt the term to publicly oppose (or promote) a policy on racist grounds without occurring in the social disadvantages of doing so. Consider, for example, the statement that the 2012 vice-presidential nominee Paul Rayan made on a radio programme:

“We have got this tailspin of culture, in our inner cities in particular, of men not working and just generations of men not even thinking about working or learning the value and the culture of work.” (BBC News 2014)

In this case, those in the audience who were able to 'hear' the dogwhistle, understood Ryan to be propagating the racial stereotype that African Americans are lazy. Moreover, as discussed above, Ryan still retained the possibility of plausibly denying that he did so. Indeed, since the word “inner city” does mean 'densely populated, high-crime, urban area', he can deny having meant anything more than that.

Plausible deniability, however, is neither the only nor the most surprising effect covert dogwhistles produce. Indeed, as documented by White (2007), the term 'inner city’ can be used to connect a subject’s pre-existing racial attitudes with his decisional processes (e.g. favouring a particular political stance). Using Saul’s terminology (2018, p. 367), we can say that dogwhistles like this one are able to bring to salience racial attitudes that would otherwise be irrelevant to the subject’s decisional processes. Such attitudes, it is worth emphasising, need not be understood as beliefs or, more generally, as propositional attitudes. It is in fact possible (and plausible) to construe the attitudes brought to salience by covert dogwhistles in terms of emotions such as resentment or fear. In general, determining the exact nature of the relevant attitudes (or predispositions) is a very interesting task that, unfortunately, goes beyond the scope of this paper. What is of special importance for present purposes, however, is that dogwhistles appear to generate this surprising effect-henceforth pre-existing attitude salience-in a wholly subliminal manner. In effect, as Mendelberg shows (2001, pp. 191-208), if the subject of a covert dogwhistle consciously realises that his or her decisions are driven by racial considerations, the impact of the dogwhistle entirely dissolves. In this regard, it is interesting to observe that in order for a covert dogwhistle to stop influencing the audience’s decisional processes, it is necessary for those who 'hear' the dogwhistle to accept (at least superficially) the social norm that conflicts with their pre-existing racial attitude. For, if this were not the case, it would be rather difficult to explain why the effect disappears as soon as the subject becomes aware of having fallen victim to a dogwhistle.

Generalising, then, covert dogwhistles appear to be able to bring to salience those pre-existing attitudes that are incompatible with extant social norms. This particular effect, moreover, is an exclusive characteristic of covert dogwhistles.⁶ Indeed, although both overt and covert dogwhistles can convey different messages to different sub-groups of the audience and ensure plausible deniability, only covert dogwhistles appear to be able to produce pre-existing attitude salience. This asymmetry is particularly significant not only because it testifies that the overt/covert distinction is not artificial, but also because it establishes a theoretical desideratum for a comprehensive theory of dogwhistles. Ultimately-I suggest-any theory of dogwhistles that seeks to fully capture the variety of this puzzling phenomenon should offer a convincing account of the asymmetry between the effects produced by different types of dogwhistle.

1.3 Saul’s Proposal

We now turn to examine Saul’s proposal for theoretically substantiating the distinction between overt and covert dogwhistles. Preliminarily, it is relevant to notice that, although Saul extensively discusses numerous examples of both overt and covert dogwhistles, she argues that whilst overt dogwhistles can be understood using existing categories such as ambiguity and implicature, the same is not true for the case of covert dogwhistles (Saul 2018, p. 371). Accordingly, on a theoretical level, Saul’s efforts seem to be chiefly directed towards the refinement of the notion of covert dogwhistle(Saul 2018, pp. 371-378).⁷ The distinction between overt and covert dogwhistles is therefore drawn by elucidating the nature of covert dogwhistle in the first instance, and leaving the overt ones to be defined in opposition.

Furthermore, Saul’s treatment of covert dogwhistles is not uniform. Her proposal, in effect, expounds the functioning of covert intentional dogwhistles differently from that of covert unintentional ones. More specifically-Saul proposes-covert intentional dogwhistles should be understood in terms of perlocutionary speech acts. In general, perlocutionary acts are 'consequential effects upon the feelings, thoughts, or actions of the speaker, or of other persons' (Austin 1962, p. 101). A covert intentional dogwhistle-Saul contends-is a special kind of perlocutionary act; one that can be characterised as a 'covert perlocutionary speech act'. Such a speech act, more precisely, is one that 'does not succeed if the intended perlocutionary effect is recognized as intended' (Saul 2018, p. 377). As Saul herself notices, however, classifying covert unintentional dogwhistles as covert perlocutionary speech acts would be an obvious mistake. Indeed, according to the above definition, a covert perlocutionary act succeeds only if the audience does not recognise the perlocutionary effect intended by the speaker. Thus, where the perlocutionary effect is not intended by the speaker in the first place-as in the case of unintentional dogwhistles - ‘‘there can be no question of the act failing if the speaker’s intention is recognized” (Saul 2018, p. 377). As a result, on Saul’s theory, covert unintentional dogwhistles are merely treated as speech acts which generate particularly unfavourable perlocutionary effects (Saul 2018, pp. 377-378).

The main problem with Saul’s proposal-I maintain-is not in what it tells us about the nature of dogwhistles, but in how much it tells us. Firstly, although employing Austin’s notion of perlocution to characterise dogwhistles is a sensible move, in itself, it does nothing to get us closer to explaining dogwhistles’ workings. In effect, by invoking the notion of perlocution Saul correctly suggests that the nature of dogwhistles should be illuminated by focusing on the causal effects they bring about. Her account, however, is surprisingly silent on exactly how the 'perlocutionary effects' produced by dogwhistles should be explained and systematised. Secondly-and most importantly-the appeal to the notion of perlocution per se is hardly helpful in drawing a clear demarcation line between overt and covert dogwhistles. The only specification that Saul offers in this regard is that the perlocutionary effect generated by a covert dogwhistle-unlike that generated by an overt one-"occurs outside of consciousness" (Saul 2018, p. 371). Additionally, even in the case of the treatment she offers to covert intentional dogwhistles, the term 'covert' remains essentially undefined. In fact, as we have seen, Saul proposes to characterise covert intentional dogwhistles as perlocutionary acts that succeed just in case the audience does not recognise the intention of the speaker. Yet, Saul says very little about how it can actually be the case that the audience fails to recognise the speaker’s intention-which is precisely what makes the perlocutionary act a covert one.

The upshot is that Saul’s proposal, although correct in outline, lacks the theoretical tools to properly explain the distinction between overt and covert dogwhistles. This, however, is not to say that the distinction itself should be rejected. On the contrary, as the examples presented in §1.1 and §1.2 testify, the distinction that Saul introduced encodes a genuine partition within the class of dogwhistles. That this genuine distinction does exist, however, we only know from the examples; not from Saul’s theory.

2. Expanding Khoo’s Simple Theory

In the present section, I shall turn to suggest that the 'Simple Theory' of dogwhistles advanced by J. Khoo (2017) can be naturally expanded to capture-just as simply-Saul’s distinction between overt and covert dogwhistles. To this end, I shall begin by introducing and motivating Khoo’s Simple Theory. In this regard, we will see that Khoo’s theory skilfully avoids difficulties that prove fatal to rival approaches.⁸ Ultimately, this discussion will pave the way for implementing the expansion that will enable Khoo’s theory to accommodate the distinction between overt and covert dogwhistles.

2.1 Overview

Let us begin our discussion by asking what makes a dogwhistle capable of conveying a multiplicity of messages through a single linguistic item. A prima facie reasonable hypothesis might be that it is some peculiarity in the meaning of the expression itself that is responsible for this multiplicity. Following up on this hypothesis, one could argue that dogwhistles are either semantically multi-dimensional (Stanley 2015) or ambiguous.⁹

In the first case, the idea appears to be that every dogwhistle has at least two separate meanings. It is precisely this ambiguity-the story continues-that explains how different sub-groups of the audience can interpret expressions like 'inner city' so differently from one another. Moreover, interpreting dogwhistling phenomena as cases of lexical ambiguity allows for a neat account of dogwhistles’ plausible deniability. Indeed, when a speaker uses an ambiguous expression E, he can plausibly deny having intended to convey one of the two meanings of the expression (say, E ₁ ) by simply declaring that what he really meant to convey was the other meaning (E ₂ ). Appealing as it is, however, this account of how dogwhistles work faces serious objections. Indeed, as Khoo (2017, pp. 40-42) establishes by means of several linguistic diagnostics, dogwhistles do not exhibit the linguistic behaviour that is characteristic of ambiguous expressions. As an example, consider the ambiguous expression 'religious leader'. This expression can either mean 'head of some religion' or 'leader who is pious'. Crucially, this ambiguity allows for the possibility of affirming without contradiction that someone is a religious leader and, at the same time, not the head of any religion:¹⁰

(1) Michael is a religious leader who is not the head of any religion.

As Khoo shows (Khoo 2017, p. 41), however, dogwhistles do not display the same linguistic behaviour. Indeed, supposing that the expression 'inner city' is ambiguous between the two meanings 'densely populated, high-crime, urban area' and 'African American', it should be possible to non-contradictorily affirm something like:

(2) # Michelle is an inner-city waitress who comes from, works, and lives in the suburbs.

But, Khoo suggests, sentences like (2) are about as close as one can get to an outright contradiction. Whilst this may not be conclusive evidence, it nevertheless speaks against the idea that the functioning of dogwhistles could be illuminated by appealing to the notion of ambiguity. Moreover-and more generally-an ambiguity approach appears to be ill-suited to explain dogwhistles like Salvini’s 'Many enemies, much honour!'. Indeed, to say that the phrase 'many enemies, much honour!', in addition to its literal meaning, can also stand for 'I am sympathetic to fascism' seems hopelessly bizarre.

A second meaning-centred approach is the one propounded by Stanley (2015). In broad outline, Stanley’s account hinges on that idea that, given a particular utterance, we can draw a distinction between two kinds of content: at-issue content and not-at-issue content. Given a particular utterance, its at-issue content is what the speaker puts forward as the main point of his production; what he (generally) proposes his interlocutor to accept. Not-at-issue content, on the other hand, is not openly available for acceptance (or denial); on the contrary, it is what the speaker expects his interlocutor to have already accepted. Thus, for example, by uttering 'James took up golf', a speaker commits himself not only to asserting that James has started golfing (at-issue-content), but also to asserting that he was not golfing before (not-at-issue content). Yet, this second proposition-unlike the first one-is not the main point of the utterance; it is rather its presupposition.

Reverting to dogwhistles, Stanley’s idea is that expressions like 'inner city', in addition to their literal meaning (i.e. their at-issue-content), carry a not-at-issue content that is responsible for the surprising effects they bring about. To see exactly how, let us focus again on the expression 'inner city'. According to the view in question, the content of this expression can represented along the following lines:¹¹

(At-issue content) _'inner-city': Densely populated, high-crime, urban area.

(Not-at-issue content) _'inner-city': Urban area mainly populated by African Americans.

Hence-the story goes on-it is in virtue of such multi-dimensionality that when a politician affirms, e.g., that he is opposing a policy that will help inner-city workers, he communicates both that the policy he is opposing will help workers living in densely populated, high-crime, urban areas (at-issue-content) and that the policy will primarily help African-American workers (not-at-issue content).

Furthermore, this approach too can boast a convincing explanation as to why dogwhistles are plausibly deniable. To use the same example, thanks to the multiple dimensions of content introduced by the expression 'inner-city', when a politician opposes a policy on the grounds that, e.g., it unfairly benefits inner-city workers, he can plausibly explain his stance in either of two ways:

1. He is opposing the policy because it unfairly benefits workers living in densely populated, high-crime, urban areas.

2. He is opposing the policy because it unfairly benefits African-American workers.

Hence, if accused of opposing the policy on racial grounds, as in case (ii), the politician can deny doing so by simply explaining his opposition as in (i). Additionally, the denial is especially plausible because the racial content of the expression 'inner-city' is not-at-issue.

As Khoo (2017, pp. 45-46) ¹² shows, however, the linguistic behaviour of dogwhistle expressions strongly speaks against Stanley’s proposal. Indeed, a distinctive feature of conventional not-at-issue content is that it is non-cancellable. Thus, a speaker cannot cancel his commitment to the not-at-issue content of a sentence by simply attaching the negation of that content to the sentence:

(3) # James took up golf, but he’s been golfing for years.

Crucially, however, there appears to be no oddity when a speaker attempts to disavow his commitment to the alleged not-at-issue content of a dogwhistle expression like 'inner-city':¹³

(4) The policy benefits inner-city workers, most of whom are white.

Examples like this suggest, contra Stanley, that dogwhistle expressions simply do not introduce any kind of conventional not-at-issue content. In addition-and independently of the above argument-if dogwhistle expressions possessed their alleged not-at-issue content conventionally, it would be especially difficult to explain why only a specific sub-group of the audience (and not the entire audience) is able to 'hear' them.

Given this result, it might appear reasonable to attempt to illuminate the functioning of dogwhistles characterising the not-at-issue content they introduce as a Gricean conversational implicature. According to Grice’s seminal theory (1975, pp. 44-45), indeed, conversational implicatures depend on special features of the conversational context, and are not determined by the conventional meaning of the expressions contained in the sentences that trigger them. Roughly speaking, we can say that what is conversationally implicated by a given utterance is what the audience needs to assume the speaker to be communicating in order for his/her production to be understood as cooperative. Consider, for example, the following exchange:

X: "Are you coming to the match tomorrow?"

Y: "I have a work meeting."

In this example, speaker Y is conveying the message that he cannot go to the match without explicitly saying so. According to Grice’s theory, the inference that allows X to understand that Y cannot go to the match can be reconstructed-or, more precisely, 'worked out'-in the following terms: on a literal level, Y’s utterance would seem to be uncooperative, as it appears to contravene the conversational maxim that requires contributions to be relevant. However, under the assumption that Y is indeed acting as a cooperative speaker, X is able to infer that Y must have intended to communicate that a work meeting will prevent him from going to the match.

The same mechanism, according to T. Marques (2020, pp. 125-126), could be deployed to explaining dogwhistling phenomena. As J. R. Torices (2021, pp. 329-333) has argued, however, a conversational implicature story about dogwhistles does not seem viable. Without entering the details of Torices’ criticisms, consider again Salvini’s Tweet 'many enemies, much honour!'. In this case, it is rather difficult to see how the dogwhistled content (something like 'I am sympathetic to the ideology of fascism') could be worked out by leveraging the presumption that Salvini is a cooperative speaker. Such a presumption, in effect, appears to play virtually no role in reconstructing how the audience is able to reach the hidden message behind Salvini’s words. However, the possibility of working out conversational implicatures through the assumption that the speaker is being cooperative is, according to Grice (1975, p. 50), one of their distinctive traits. The fact that Salvini’s dogwhistle cannot be readily worked out in this manner, consequently, speaks against the possibility of representing it through the notion of conversational implicature.¹⁴

In general, there appear to be good reasons to reject not only each of the specific proposals presented above, but also the very idea of expounding the functioning of dogwhistles by focussing on the content of the expressions that give rise to them. Content-driven approaches can in fact encounter pressing difficulties due to the centrality they assign to the notions of content/meaning. Indeed, in the literature there are examples of dogwhistles involving neither words nor sentences, but rather images and other non-linguistic entities.¹⁵ Whilst it is undoubtedly reasonable to appeal to the notion of content/meaning when dealing with words and sentences, it does not appear equally natural to do so with, e.g., images. Subsequently, any content-driven account that aims to fully elucidate the nature of dogwhistles, should at the very least supply a convincing elucidation of the notion of 'content/meaning of an image'.

In brief, the difficulties encountered by content-driven approaches strongly suggest that going to the heart of the matter may require a change of perspective. That is precisely what the Simple Theory of dogwhistles advanced by J. Khoo (2017, pp. 47-52) has to offer. Khoo’s theory successfully sidesteps the problems encountered by the approaches discussed, by focusing not on the content/meaning of dogwhistle expressions, but exclusively on what goes on in the mind of those who 'hear' such expressions. In effect, according to Khoo, the meaning of the linguistic items (if any) employed to generate a dogwhistle is just their literal meaning. What is genuinely responsible for producing the surprising effects discussed in §1 is the audience’s pre-existing doxastic state. There is no unusualness in the meaning of expressions like 'inner city' or 'many enemies, much honour!'-Khoo’s story says. What is crucial is the hearers’ pre-existing beliefs that inner cities are mostly populated by African Americans and that 'many enemies, much honour!' is a fascist motto, respectively.

The key-idea, to be more precise, is that the words and sentences used to produce the typical effects of dogwhistles can be better understood as 'trigger expressions' which, in the presence of an appropriate pre-existing belief, initiate an inferential process in the mind of the hearer. Hence, to put it schematically, if the speaker utters that x is F and an hearer has a pre-existing belief according to which everything that is F is also G, then the hearer will infer that x is G. Notice, however, that only those hearers that already believed that every F is also a G will conclude that x is G. According to Khoo’s Simple Theory, it is therefore the pre-existing doxastic state of each member of the audience that explains why only a sub-group is able to 'hear' the dogwhistle. In other words, the duplicity of a dogwhistle expression is not an intrinsic feature of its content; rather, it is the result of an interaction between that content and the hearer’s prior beliefs. Lastly, like its rivals, Khoo’s theory too can boast a very convincing explanation of dogwhistles’ plausible deniability: if the speaker wants to deny having intended to convey a hidden message, he can simply invoke the literal meaning of the expression he utilised to generate the dogwhistle. In fact, since in Khoo’s account the literal meaning is the only meaning of a dogwhistle expression, the denial of an additional hidden message is incredibly plausible.

2.2 A Simple Expansion to Khoo’s Simple Theory

We have seen that the theory advanced by Khoo offers a remarkably simple explanation of the functioning of dogwhistles. Yet, as it stands, the theory does not explicitly account for the distinction between overt and covert dogwhistles introduced by Saul (2018). Still, as we saw in §1, this distinction captures a genuine difference between two kinds of dogwhistles; a difference that is reflected in an asymmetry between the effects they generate. In what follows, I shall therefore focus on the possibility of combining Khoo’s theory with Saul’s distinction to illustrate how it is possible to amend the former, making it capable of accounting for the latter.

First, let us notice that there is no apparent reason to think that Khoo’s Simple Theory and Saul’s conception cannot be consistently conjoined.¹⁶ On the contrary, Khoo (2017, p. 47) himself seems to hint at something akin to the distinction between overt and covert dogwhistles, without, however, following up on this insight.

What I intend to show, however, is not only that Khoo’s theory can be consistently conjoined with Saul’s distinction, but that the core intuition around which the Simple Theory is built accommodates the distinction in a neat and intuitive way. In other terms, the idea is that we can frame the difference between overt and covert dogwhistles by focussing on the pre-existing doxastic state of the audience. The suggestion, in particular, is that Khoo’s theory is on the right track in focusing on the audience’s pre-existing beliefs but, in order to account for the asymmetry between the effects of different types of dogwhistles, it needs to be amended by looking at the higher-order propositional attitudes of the hearers.

More precisely, the difference between overt and covert dogwhistles-I maintain-can be sharply framed in terms of what the heares know themselves to believe.

To better see the point, consider again the way in which Saul’s distinction is customarily introduced. A dogwhistle-the story goes-is classified as 'overt' only if the target audience is aware of receiving a concealed message from the speaker; otherwise-the story continues-the dogwhistle is said to be 'covert'. If we now combine this intuitive elucidation with Khoo’s insight that the functioning of dogwhistles should be unfolded by inspecting the audience’s pre-existing doxastic state, the following explanation naturally presents itself: in the case of an overt dogwhistle, the hearers not only have a specific pre-existing belief, but they also know that they have it. This is the reason why those targeted by overt dogwhistles can recognise the speaker’s intention of communicating directly with them. Conversely, in the case of a covert dogwhistle, the hearers have a pre-existing belief but-crucially-they do not know they have it.

To see the proposed account in action, let us begin by re-visiting the case of overt dogwhistle presented in §1.1. We observed that Salvini’s public employment of the motto 'many enemies, much honour!' constitutes a vivid example of overt dogwhistling. Salvini’s Tweet, we noticed, is in fact capable of transmitting a message that only a circumscribed sub-group of the audience can decode. Moreover, we plausibly supposed that the members of the relevant sub-group are well aware of being the addressees of Salvini’s concealed message. As discussed in §2.1, Khoo’s theory seeks to explain the behaviour of dogwhistles by focusing on the pre-existing doxastic state of the audience. Accordingly, the Simple Theory (correctly) predicts that only those members of the audience who believe that 'many enemies, much honour!' is a fascist motto can grasp the hidden message conveyed by Salvini’s words. What Khoo’s theory alone cannot not tell us, however, is whether Salvini’s Tweet should be classified as a covert or overt dogwhistle. It is exactly at this point that the proposed amendment comes to the rescue of Khoo’s account.

As mentioned, the suggestion is that, if we want to understand whether Salvini’s dogwhistle should be classified as overt or covert, we need to attend not only to what the hearers believe but also to what they know themselves to believe. Specifically, the idea is that if they are aware of possessing the pre-existing belief which-according to Khoo’s theory-allows them to 'decipher' the speaker’s coded message, then the dogwhistle is overt. Conversely, if they merely have the relevant pre-existing belief, but do not know they have it, the dogwhistle should be classified as covert.

In the case of our example, the proposed account seems to yield the expected result, identifying Salvini’s Tweet as an overt dogwhistle. Indeed, following the proposed heuristics, the suggestion is that those who can 'hear' Salvini’s dogwhistle are aware of being the target of his message precisely because not only do they possess the pre-existing belief that the motto 'many enemies, much honour!' is fascist, but they also know they possess this belief. Hence, when confronted with the public utterance of the phrase 'many enemies, much honour!', they consciously realise that by choosing that specific motto, the speaker is manifesting his sympathy towards the ideology of fascism.

Furthermore, the proposed framework intuitively accounts for how covert dogwhistles seem to act below the level of consciousness and, connectedly, how their effect vanishes when the hearer becomes aware of them. To appreciate the point, let us focus again the pre-existing attitude salience effect that covert dogwhistles generate. As we have seen, empirical evidence (White 2007) suggests that covert dogwhistles like 'inner city' are able to bring one’s pre-existing racial attitudes to bear on his decisional processes. Moreover-and crucially-it has been observed that this effect completely vanishes when the hearer becomes aware of being victim of a dogwhistle.

To see how this phenomenon could be explained, let us start by considering the familiar case of a politician who publicly opposes a policy (P) claiming that it unfairly favours inner-city workers. As we saw earlier, the politician might use this statement to communicate to a restricted segment of the audience that policy P unfairly favours, more specifically, African-American workers. In terms of Mendelberg’s theory (2001), the politician can pursue this strategy to gain voter approval on racial grounds without violating the social norm that requires one not to be openly racist. According to Khoo’s theory, the expression 'inner-city' is capable of generating this surprising effect by leveraging hearers’ pre-existing belief that inner cities are predominantly populated by African Americans. Specifically-Khoo proposes (2017, pp. 47-48)-only those members of the audience who possess the relevant pre-existing belief can hear the politician’s dogwhistle. Accordingly, only those audience members who believe that inners cities are inhabited by African Americans come to believe that policy P favours, specifically, African-American workers.

Notice that, already at this point in the discussion, we are in a position to offer a minimal explanation as to why the term 'inner city' is able to bring a subject’s existing racial attitudes to bear on his decisional processes [as documented by White (2007)]. Effectively, if a person has prejudices against African Americans, coming to believe that a certain policy favours African-American workers could conceivably lead them to oppose that policy (or support a politician who opposes the policy). What we have not yet explained, however, is exactly what makes the politician’s statement an example of covert rather than overt dogwhistling. And, in parallel, why if the hearers were to learn that the politician’s statement was racist, they would cease to be influenced by it. This is where the proposed amendment to Khoo’s Simple Theory comes directly into play.

According to the account offered, in order to illustrate why the politician’s statement constitutes a case of covert dogwhistling, we need to examine the pre-existing belief responsible for generating the dogwhistle. As suggested by Khoo (2017, p. 47), in the case described, the pre-existing belief in question is, approximately, that inner cities are predominantly inhabited by African Americans. Building on this explanation, the idea is that, if this belief were consciously endorsed by the hearers, it would be rather difficult to illustrate how the politician’s dogwhistle could act "outside of consciousness", as covert dogwhistles are supposed to do (Saul 2018, p. 371, emphasis in original). And, at the same time, it would be equally problematic to explain why the covert dogwhistle ceases to be effective when the deceptive intention of the speaker is acknowledged. The suggestion, then, is that the pre-existing belief responsible for the production of a covert dogwhistle is not consciously endorsed by the mind of the hearer (although, plausibly, it is available to his mind for endorsement). In other words, the members of the audience to whom the politician’s dogwhistle is addressed believe that inner cities are inhabited by African Americans, and yet they do not consciously endorse this belief.

This framework allows us to draw on the resources offered by Mendelberg’s theory (2001) to schematically illustrate what happens when a covert dogwhistle ceases to be effective in influencing the decisional processes of a subject S. We can first frame the mental state of those audience members who 'hear' a covert dogwhistle in the following terms:

1. They possess prior predispositions that are inconsistent with a an existing social norm N that governs their social behaviour (e.g., 'don’t be racist');¹⁷

2. They accept (at least superficially) the social norm N.

According to Khoo’s Simple Theory, for S to be subject to a dogwhistle, he must possess a specific pre-existing belief φ that allows him to inferentially reach the (racial)¹⁸ message δ sent by the speaker. Furthermore, in accordance with the suggested elucidation of the overt/covert distinction, in order for S to be subject specifically to a covert dogwhistle, he must also be unaware of possessing the pre-existing belief that allows him to infer δ. Namely, S must believe that φ but, at the same time, fail to know that he believes that φ.

Now-as specified in (i) and (ii)-despite knowingly accepting the social norm N, the subject possesses some existing (racial) predispositions that are inconsistent with N. Moreover, since we are dealing with a covert dogwhistle, we also know that S’s pre-existing belief φ has the power to bring such (racial) predispositions to bear on S’s decisional processes. Indeed, by allowing the subject to receive the speaker’s racial message, believing φ can induce S to engage in behaviour incompatible with the social norm N (e.g., making voting decisions on the basis of racial considerations). However, when S becomes aware of the racial nature of the speaker’s message, his prior (and conscious) acceptance of the social norm N overrides the covert effect of the dogwhistle. Accordingly, in order to maintain a behaviour compatible with the norm, S disconnects his decisional processes from his belief that φ.

All of that-the proposed account correctly predicts-does not apply to overt dogwhistles. In fact, when a subject S' is targeted by an overt dogwhistle, he already knows that he possesses the preexisting belief ψ that allows him to understand the hidden message conveyed by the speaker. Consequently-as we saw when discussing the case of Salvini’s Tweet-S' is already consciously aware of the social unacceptability of the speaker’s concealed message. Of course, should he wish to do so, S' might still decide to engage in behaviour incompatible with a social norm N' on the basis of the message dogwhistled by the speaker. In this case, however, revealing the nature of the speaker’s hidden message to S' will not result in a change in his decision-making processes for, crucially, S' already knows what the real nature of the speaker’s message is.

The simple modification proposed to expand Khoo’s theory successfully explains the different behaviour of overt and covert dogwhistles observed in §1. However, it is also vital that it succeeds in preserving, so to speak, what both types of dogwhistle have in common. As we have seen, both covert and overt dogwhistles allow the speaker to communicate a concealed message to the audience while retaining the possibility of plausibly denying having done so. As illustrated in §2.1, Khoo’s Simple Theory provides a coherent and intuitive explanation of this phenomenon. Indeed, from the perspective of Khoo’s account, if the speaker wants to deny having intended to convey a hidden message, all he needs to do is to invoke the literal meaning of the expression he utilised to generate the dogwhistle; which, according to the Simple Theory, is the only meaning of that expression. Now, the expansion of Khoo’s theory proposed in this paper in no way undermines this explanation. In fact, as pointed out, what enables Khoo to explain dogwhistle’s plausible deniability so straightforwardly is the shift in focus from the content of the expressions employed to produce dogwhistles, to the doxastic state of the hearers who perceive them. Hence, since the proposed expansion accounts for the distinction between overt and covert dogwhistles precisely by focussing on what the hearers know themselves to believe, the shift in focus proposed by Khoo is thoroughly honoured. The explanation of plausible deniability put forward by Khoo is therefore fully preserved when the Simple Theory is expanded as suggested.

Just as the expanded version of the Simple Theory retains the virtues of Khoo’s original proposal, it inherits many of its objectionable aspects as well. In this regard, a general criticism that can be raised against Khoo’s theory (and, as a consequence, against its augmented version) concerns its inherently propositional nature. As we have seen, in fact, the Simple Theory relies on the audience’s pre-existing beliefs to explain the functioning of dogwhistles. Yet, belief is typically considered a propositional attitude (Schwitzgebel 2006); i.e. an attitude that a subject has towards a certain proposition. Khoo’s account, as a consequence, appears to portray dogwhistles as a propositional phenomenon. This view, of course, stands in contrast to any approach that seeks to illuminate the nature of dogwhistle through associations. According to this latter proposal, what allows dogwhistles to produce their surprising effects is not the hearer’s pre-existing belief that, for example, inner cities are predominantly populated by African Americans. Rather, what is crucial is that for a circumscribed sub-group of hearers the concept of inner city (or the term 'inner city') primes specific racial attitudes.¹⁹

Firstly, let us notice that both the inference-driven approach suggested by Khoo and an association-driven approach reject the idea that dogwhistles arise from some peculiarity in the meaning of the expressions used to produce them.²⁰ On the contrary, they both aim to illuminate the nature of dogwhistles by focusing on what happens in the mind of the hearer. The difference is that, according to Khoo, dogwhistle effects stem from an inference involving a pre-existing belief; whereas, according to an association-driven account, they are the result of association mechanisms. Hence, although initial appearances might suggest otherwise, the commonality between the two approaches is significant. So much so that an association-driven theory is able, just as much as Khoo’s Simple Theory, to evade the objections that-as discussed in §2.1-prove fatal to content-centred approaches.

Determining which of these two approaches best captures the phenomenon of dogwhistling is an interesting and challenging matter. At least partly, the difficulty comes from the fact that the issue lies right at the borderline between three different disciplines: philosophy, psychology, and linguistics. What seems most plausible, however, is that the matter cannot be settled through the exclusive use of armchair methods. Effectively, as Khoo (2017, pp. 50-51) points out, the theoretical differences between inference-driven and association-driven approaches lead them to yield substantially different empirical predictions. For instance, an association-driven theory (but not an inference-driven one) should predict that when a politician merely mentions the word 'inner city', the same dogwhistle effects should be registered as when he actually uses that word (Khoo 2017, p. 51).²¹ Effectively, given that according to an association-driven account it is the concept of inner city (or the word 'inner city' itself) that primes the hearer’s racial attitudes, it would be difficult to see why the mere mention of the term should not suffice to bring the relevant racial attitudes to salience. This, of course, is not to suggest that an association-driven approach is lacking in credibility. There is absolutely nothing incredible or absurd about this empirical prediction. Quite the contrary, the point is precisely that, owing to the different predictions they yield, the soundest way to contrast inference-driven and association-driven approaches appears to be through experimental research. Such an empirical task, however, exceeds the modest scope of the present contribution.

Let us finally consider an additional difficulty that the suggested way of expanding Khoo’s theory appears to generate. The proposed approach commits us to a conception of belief according to which it is possible to believe a proposition without being aware of doing so. This view is apparently in contrast with many traditional axiomatisations of epistemic logic that, following Hintikka’s seminal work (1962), accept that if a subject S believes a proposition φ, he also knows that he believes φ:²²

$\left(KB2\right)B_S\phi \to K_S{B}_S\phi .$

This principle-often called the principle of positive introspection-is meant to capture one of the fundamental connections binding the concept of belief to that of knowledge. The thought, to put it more discursively, is that our beliefs are introspectively accessible to us and, consequently, that we can attain knowledge of our own doxastic state. Yet, while this may seem a very reasonable assumption when theorising about ideal epistemic agents, the same does not necessarily hold true for non-idealised epistemic subjects as ourselves. Indeed, many philosophers have suggested that the word 'belief' generally functions as an umbrella term whose semantic field encompasses quite different phenomena. Notably, when attributing a belief to a subject, one can either refer to a proposition consciously endorsed by the subject (occurrent belief), or to a piece of information merely available to the mind for endorsement (dispositional belief) (Schwitzgebel 2006). Observe, however, that whilst it might seem reasonable to accept the principle (KB2) when dealing with occurrent beliefs, it is not at all clear that the same could be so lightly done in the case of dispositional beliefs. Now, returning to the point, it would appear that in the specific case of covert dogwhistles, the pre-existing beliefs that enable the audience to decode the speaker’s message are better understood as dispositional beliefs, rather than occurrent ones. Indeed, were these beliefs consciously endorsed by the subject, there would be no point in drawing a distinction between overt and covert dogwhistles in the first place. Yet, if the foregoing considerations do not stray far from the truth, there is no reason to expect a priori that the pre-existing beliefs that give rise to a covert dogwhistle will comply with the intuition that (KB2) is intended to capture.

Conclusion

The primary purpose of the present paper was to illustrate how Saul’s distinction between overt and covert dogwhistles (Saul 2018) can be refined and illuminated by embedding it into the 'Simple Theory' of dogwhistles advanced by Justin Khoo (2017).

We begun our discussion by showing that Saul’s distinction successfully captures a genuine divide within the category of dogwhistles-a distinction that is reflected in the different psychological and linguistic effects they generate. This point has been established by analysing two concrete examples of dogwhistle (one overt, and one covert) drawn from the contemporary political scene. We then proceeded to consider Saul’s own proposal to explicate the overt/covert distinction. In this context, I argued that Saul’s solution, although correct in outline, merely records the distinction between overt and covert dogwhistles without really developing a theory to substantiate it.

Ultimately, we turned to explore the Simple Theory of dogwhistles advanced by Khoo (2017). As we have seen, Khoo’s theory sets out to explain the functioning of dogwhistles by focusing on the doxastic state of the members of the audience. Along the same lines, I have maintained that Khoo’s Simple Theory can be expanded to accommodate and clarify the distinction between overt and covert dogwhistles. Specifically, we saw that the different effects engendered by these two types of dogwhistles can be easily explained by focusing not only on what the hearers believe, but on what they know themselves to believe. In the case of overt dogwhistles-the suggestion is-those who can decipher the speaker’s coded message not only have a specific pre-existing belief, but they also know they have it. Conversely, in the case of covert dogwhistles, the hearers do have a pre-existing belief, but-crucially-they do not know they have it.

References

• AUSTIN, J. L. How to Do Things with Words. Cambridge: Harvard University Press, 1962.
• BBC NEWS. "Paul Ryan clarifies 'inarticulate' remarks on inner cities". (BBC News: US & Canada), < (BBC News: US & Canada), https://www.bbc.co.uk/news/world-us-canada-26564949 >, 2014. [accessed 27 July 2022]
  » https://www.bbc.co.uk/news/world-us-canada-26564949
• GRICE, H. P. "Logic and Conversation". In Syntax and Semantics 3: Speech Acts, C. Peter, and J. L. Morgan (eds.), New York: Academic Press, pp. 41-58, 1975.
• BLOME-TILLMANN, M. "Conversational Implicatures (and How to Spot Them)". Philosophy Compass, 8. <https://doi.org/10.1111/phc3.12003>, pp. 170-185, 2013.
  » https://doi.org/10.1111/phc3.12003
• HENDERSON, R., MCCREADY, E. "How Dogwhistles Work". In Arai, S., Kojima, K., Mineshima, K., Bekki, D., Satoh, K., Ohta, Y. (eds) New Frontiers in Artificial Intelligence. JSAI-isAI 2017. Lecture Notes in Computer Science, vol 10838, Springer, Cham. <https://doi.org/10.1007/978-3-319-93794-6_16>, pp. 231-240, 2018.
  » https://doi.org/10.1007/978-3-319-93794-6_16
• HINTIKKA, J. Knowledge and Belief: An Introduction to the Logic of the Two Notions. Ithaca: Cornell University Press, 1962.
• KHOO, J. "Code Words in Political Discourse". Philosophical Topics, 45(2). pp. 33-64, 2017.
• MARQUES, T. "How can philosophy of language help us navigate the political news cycle?". In: E. Vintiadis (ed.). Philosophy by Women: 22 Philosophers Reflect on Philosophy and Its Value, New York: Routledge, pp. 122-130, 2020.
• MENDELBERG, T. The Race Card: Campaign Strategy, Implicit Messages, and the Norm of Equality. Princeton: Princeton University Press, 2001.
• PANNULLO, A. "Salvini cita Lui: “Molti nemici, molto onore”. Ma la frase non è del Duce…". (Secolo d’Italia) < (Secolo d’Italia) https://www.secoloditalia.it/2018/07/salvini-cita-lui-molti-nemici-molto-onore-ma-la-frase-non-e-del-duce/ >, 2018. [accessed 27 July 2022]
  » https://www.secoloditalia.it/2018/07/salvini-cita-lui-molti-nemici-molto-onore-ma-la-frase-non-e-del-duce/
• SAUL, J. "Dogwhistles, political manipulation, and philosophy of language". In New Work on Speech Acts, D. Fogal, D. W. Harris, and M. Moss (eds.), New York: Oxford University Press, pp. 360-383, 2018.
• SCHWITZGEBEL, E. "Belief". The Stanford Encyclopedia of Philosophy (Fall 2019 Edition), E. N. Zalta (ed.) <) https://plato.stanford.edu/archives/fall2019/entries/belief/ >, 2006. [accessed 27 July 2022]
  » https://plato.stanford.edu/archives/fall2019/entries/belief/
• SALVINI, M. [n.t.]. (Twitter) < (Twitter) https://twitter.com/matteosalvinimi/status/1023500457229070336 >, 2018. [accessed 27 July 2022].
  » https://twitter.com/matteosalvinimi/status/1023500457229070336
• STANLEY, J. How Propaganda Works. Princeton: Princeton University Press, 2015.
• TORICES, J. R. "Understanding dogwhistles politics". Theoria, 36(3), <https://doi.org/10.1387/theoria.22510>, pp, 321-339, 2021.
  » https://doi.org/10.1387/theoria.22510
• WARREN, B. "Ambiguity and Vagueness in Adjectives". Studia Linguistica, 42, <https://doi.org/10.1111/j.1467-9582.1988.tb00789.x>, pp. 122-172, 1988.
  » https://doi.org/10.1111/j.1467-9582.1988.tb00789.x
• WHITE, I. K. "When Race Matters and When It Doesn’t: Racial Group Differences in Response to Racial Cues". American Political Science Review, 101(2), <doi: 10.1017.S0003055407070117>, pp. 339-354, 2007.
  » doi: 10.1017.S0003055407070117

• 1
  It is worth observing that dogwhistles, owing to their ability to deliver a 'ciphered' message to a sub-group of the audience, are occasionally referred to as 'code words'. Nonetheless, there seems to be no general consensus on the precise usage of these two expressions in the literature. J. Khoo (2017), for instance, appears to use the two expressions 'code word' and 'dogwhistle' quite interchangeably. Other authors, however, appear to use the term 'code words' to refer exclusively to the linguistic elements involved in the production of dogwhistling effects [See, for instance, (Stanley 2015, pp. 125-177)]. According to this latter terminological choice, code words are only a particular instance of the broader category of dogwhistles. Indeed, although many of the examples discussed in the literature revolve (at least superficially) around the use of linguistic elements, it is important to stress that dogwhistles can also be produced by means of a combination of linguistic and visual elements [See (Mendelberg 2001)].
• 2
  The motto can almost certainly be attributed to the German leader Georg von Frundsberg (1473-1528) [See (Pannullo 2018)]. Subsequently, however, the slogan 'Molti nemici, molto onore!' has been employed and popularised by the fascist propaganda.
• 3
  According to Italian law (LEGGE 20 giugno 1952, n. 645, Art. 4), supporting the ideology of fascism is an actual crime.
• 4
  It is worth emphasising that in (Mendelberg 2001) the term 'dogwhistle' is virtually absent. Mendelberg’s research focuses more specifically on what she calls 'implicit racial appeals' [See (Mendelberg 2001, p.11)]. Despite their more specific focus, however, Mendelberg’s studies appear to be able to shed light on crucial aspects of the functioning of the whole category of dogwhistles. Khoo, for instance, explicitly introduces his general account of code words using Mendelberg’s theory as a "backdrop" (Khoo 2017, p. 38). The possibility of extending Mendelberg’s considerations to the entire category of dogwhistles, however, is obscured by the fact that her distinction between 'implicit' and 'explicit' appeals does not coincide with that between overt and covert dogwhistles. Admittedly, since Mendelberg’s studies focus on 'implicit racial appeals', one might expect her considerations to apply directly only to covert dogwhistles. Yet, while it is true that the examples primarily discussed by Mendelberg fall within the category of covert dogwhistles, it is also true that the way she defines the category of implicit appeals suggests that it should include both covert and overt dogwhistles. In specifying the relevant notion of implicitness, Mendelberg writes: "What exactly is the difference between implicit and explicit messages? First, consider what makes an appeal explicitly racial. By my definition, a racial appeal is explicit if it uses racial nouns or adjectives to endorse white prerogatives, to express anti-black sentiment, to represent racial stereotypes, or to portray a threat from African Americans. […] Implicit racial appeals convey the same message as explicit racial appeals, but they replace the racial nouns and adjectives with more oblique references to race. […] Implicit racial appeals discuss a nonracial matter and avoid a direct reference to black inferiority or to white group interest" (Mendelberg 2001, pp. 8-9). This suggests that, according to Mendelberg, a racial appeal is implicit if it conveys a racial message without containing linguistic expressions whose content is directly racial. However, both covert and overt dogwhistles can be used to send a racial message without resorting to the use of explicitly racial linguistic expressions. Indeed, the ability to convey a message implicitly (i.e. without using linguistic items that directly express it) is a trait shared by overt and covert dogwhistles.
• 5
  It is worth emphasising that, given the distance in time of the cited empirical results concerning the use of the expression 'inner city', it is questionable whether this phrase is still able to give rise to the effects described above.
• 6
  See (Saul 2018, pp. 361-371), (Mendelberg 2001, pp. 197-199).
• 7
  This, however, is not to say that Saul is exclusively concerned with covert dogwhistles. On the contrary, as previously noted, the very idea of introducing the distinction between overt and covert dogwhistles is Saul’s great merit. Despite this, Saul emphasises (Saul 2018, p. 371) that existing linguistic categories are unable to adequately explain the behaviour of covert dogwhistles specifically. To address this shortcoming, Saul devotes special attention to the refinement of the notion of covert intentional dogwhistle as what she calls a 'covert perlocutionary speech act'(Saul 2018, pp. 376-378). This approach is briefly presented in the present section.
• 8
  This should not suggest that Khoo’s theory is entirely unproblematic. The main objection directed against Khoo’s theory [See (Khoo 2017, pp. 50-51), (Henderson & McCready 2018, pp. 233-234), (Torices 2021, pp. 334-335)], is that it erroneously predicts that the effects of a dogwhistle expression D persist even if it is substituted with a co-extensional expression C. This objection, however, does not directly collide with the adjustment of the theory that this article intends to propose. Moreover, there are dogwhistle theories [e.g. (Henderson & McCready 2018)] that, while preserving the central idea of Khoo’s account, manage to escape the above objection. The amendment proposed in this article can be trivially extended to such theories. A second objection—more relevant for present purposes—accuses Khoo’s account of not being able to readily explain why the effect of a covert dogwhistle vanishes when the speaker’s deceptive intention is revealed [See (Torices 2021, p. 335)]. As we shall see in §2.2, however, the expansion to Khoo’s Theory proposed in the present essay puts Khoo’s Theory in a better position to deal with this phenomenon.
• 9
  As indicated by Khoo (2017, p.40, footnote 14), some passages in (Mendelberg 2001) might suggest that she subscribes to an ambiguity approach. For example, she notes that "[i]mplicit racial appeals can be generated with words alone" (Mendelberg 2001, p. 9) and subsequently she writes that "[t]he final 'A' factor in my theoretical framework is the ambiguity of the racial content of the message"(Mendelberg 2001, p. 125) and "[i]mplicit racial messages are ambiguous racial cues [...]. This ambiguity may be the source of their power" (Mendelberg 2001, p. 126). However, it must be underlined that Mendelberg’s research primarily targets a very specific kind of ambiguity; namely, the one generated by the combined use of linguistic and visual elements. As a consequence, the attribution to Mendelberg of an ambiguity approach is not entirely justified.
• 10
  The example is taken from (Warren 1988).
• 11
  Here, I am following the phrasing presented in (Khoo 2017, p. 44).
• 12
  A different argument to the same effect is presented in (Henderson & McCready 2018).
• 13
  The example is discussed in (Khoo 2017, p. 46).
• 14
  For a more general and comprehensive discussion of the relationship between dogwhistles and conversational implicatures, see (Torices 2021, pp. 329-333). For the present case, however, the conclusion that the dogwhistling effects of Salvini’s Tweet are not due to a conversational implicature could be strengthened by means of the so-called non-detachability test. Conversational implicatures are normally non-detachable(Blome-Tillmann 2013, p. 173): BLOME long as the conversational context is left untouched, substituting the expression that generates a conversational implicature with a truth-conditionally equivalent expression preserves the implicature. This condition, however, is clearly not satisfied in the case of Salvini’s Tweet: if, instead of 'Many enemies, much honour!', Salvini had used the truth-conditionally equivalent expression 'The number of enemies one has is proportional to how honourable he is', his Tweet would have lost its dogwhistling effects.
• 15
  See (Mendelberg 2001, pp. 3-8).
• 16
  This point is explicitly noted—but not explored—by Torices (2021, p. 335).
• 17
  As pointed out in the initial characterisation of covert dogwhistles (see §1.2), it is worth remarking that the relevant predispositions need not be understood as beliefs (or, more generally, in terms of propositional attitudes). The proposed account, in effect, allows leeway for understanding such predispositions in terms of emotions such as resentment or fear.
• 18
  To link the preset schematic explanation to the examples of covert dogwhistling discussed earlier, I decided to consider a case in which the message δ inferred by the hearer is racial. This, however, is not to say that all covert dogwhistle are necessarily racial. In effect, the schematic explanation provided is in no way dependent on the racial nature of the chosen example.
• 19
  An account of covert dogwhistles in terms of associations can be found in (Torices 2021, pp. 335-336). Torices’ proposal, in terms of simplicity, is in no way inferior to Khoo’s. In broad outline, Torices suggests to characterise covert dogwhistles as 'attitude-foregrounders' (rather than inference-triggers, as in Khoo’s theory). Specifically, according to this account, covert dogwhistles are able to bring to salience hearers’ pre-existing (racial) attitudes by means of associations. Overall, as Torices (2021, p. 336) points out, this approach bypasses one of the main problems of Khoo’s Simple Theory, i.e. the fact that it prima facie predicts that the effects of a covert dogwhistle should persist even when the expression used to generate it is substituted with a co-extensional one. Additionally, Torices’ approach natively incorporates an explanation as to why, when the speaker’s deceptive intention is revealed, the effect of a covert dogwhistle vanishes [See (Torices 2021, p. 336)]. On the contrary, as we have seen (pp. 21-22 of this paper), Khoo’s Simple Theory has to be expanded to enable it to offer an explanation of this phenomenon.
• 20
  See (Khoo 2017, p. 51).
• 21
  For additional empirically testable differences between inference-driven and association-driven accounts see (Khoo 2017, pp. 51-52).
• 22
  Since knowledge is normally taken to entail belief [(KB1): $K_S\phi \to B_S\phi$ ], the conflict extends also to the axiom of doxastic logic (Axiom 4) according to which if a subject S believes that φ, then S believes that he believes that φ [i.e $B_S\phi \to B_S{B}_S\phi$].

• 23
  Article info CDD: 401

Publication Dates

• Publication in this collection
  22 Dec 2023
• Date of issue
  2023

History

• Received
  30 July 2022
• Reviewed
  11 Mar 2023
• Accepted
  29 Mar 2023