Sincerity in bulk

Abstract

This paper is concerned with situations in which a speaker issues many speech acts at the same time. A common example is the publication of a large text such as a book containing many distinct assertions. It is argued that these cases present a challenge for speech act theory related to how we are to understand sincerity. With reference to the well known paradox of the preface, it is argued that sincerity of such bulk speech cannot be understood as a simple conjunction of the sincerity of the encoded acts. A proposal is given according to which sincere bulk speech requires the speaker explicitly and as precisely as possible mark any subsets of her communication about which she has doubts.

1. INTRODUCTION

Some of our communicative acts are, to put it plainly, rather long. Sometimes they are presented to the world all at once as letters, articles, and books containing dozens, hundreds, and even thousands of distinct but interrelated sentences. At other times, they come not written but still presented en masse, as when one gives a talk or delivers a speech. These forms of communication are central to our epistemic, moral, and political lives: scientific papers, newspaper articles, and political addresses are all instances of this genre.

Presently, one of the most fruitful theoretical frameworks for analysing social and epistemic dimensions of communication is speech act theory. Its utility arises at least in part from the fact that it allows us to integrate the study of language and communication with action theory. By taking communication to be a kind of acting, we are able to study a speaker's intentions, beliefs, and other mental states, alongside normative questions about their commitments and responsibilities. The framework has thus shed light on a wide array of topics including silencing, protests, social institutions, and social contract theory.¹ Recently as well, there has been growing interest in speech coming from groups and institutions—speech often performed in bulk, and via a spokesperson or written statement.² It is desirable, therefore, to properly integrate books and speeches into this larger picture.

Yet, as we shall see, these sorts of bulk communication present some subtle challenges for a speech act theoretic framework. This is because the central unit of analysis in speech act theory is what Austin (1975) called the illocutionary act.³ Illocutionary acts are, roughly, functions or uses of language—things we employ language to do. Some common examples include promising, apologising, asserting, testifying, ordering, warning, asking and requesting. Illocutionary acts are performed by making utterances (or in Austin's terminology locutionary acts) although there is no strict correspondence between the utterance used and the illocutionary act it encodes in a particular instance. For example,“shut the door!” might be an order, a request, or a warning; “I'll be back” might be a promise, a threat, or even simply an assertion about the speaker's anticipated whereabouts. Determining what illocutionary act is being performed in a given case and deciding what to do with it (whether to believe it, how to respond etc.) is a complex task for audience members and theorists alike. Systematising and analysing this communicative unit is the task of speech act theory.

One of the troubles with bulk speech is that it seems to contain many illocutionary acts which are not created and delivered one at a time but rather are offered together. Moreover, the utterances encoding the constituent illocutions may not be written sequentially, and they likely have been revisited and revised by the author—perhaps many times. Thus these communications raise the relatively understudied question of illocutionary compositionality: how do illocutionary acts combine with one another?⁴ A straightforward hypothesis about the composition of speech acts is, of course, that each individual sentence encodes a distinct stand‐alone illocutionary act. Doing this allows us to say that many individual speech acts are being performed while holding the rest of our theory fixed. A research article or scholarly book is, on this understanding, nothing more than a long string of assertions sprinkled, perhaps, with the occasional question (for future study) or promise (of future results).

The aim of this paper is to bring to light one way in which this understanding of speech act compositionality is sometimes inadequate and to suggest a way to handle the problem. The problem relates specifically to the question of whether or not what has been said in cases of bulk speech is sincere. I will show that we must be cautious about treating these sorts of communications as simple conjunctions of distinct speech acts. We cannot simply detach one embedded speech act—an assertion, say—from a text and treat it as pragmatically the same as if it had been issued in isolation. Instead, we need to consider the presentation of these large scale communications as a communicative act in its own right, and one having a distinctive sincerity condition at that.

I will press these concerns by considering a new spin on the well‐known paradox of the preface. This paradox is of interest here because it presents the kind of scenario which challenges the simple picture of speech act compositionality. That said, it should be emphasised that I am not attempting to solve (or dissolve) the paradox as it traditionally gets used within epistemology and logic. Rather, the point is to learn some new lessons from an old paradox—lessons applicable to a rather different domain of philosophical thought. To that end, I will use the paradox of the preface as a core example while asking what it takes for a speaker to sincerely perform a bulk speech act such as publishing a book. I will consider a number of different options for a sincerity condition underlying such an act, with reference to how the original paradox of the preface has been handled. I will argue that some intuitive approaches fail because they focus on the individual speech acts encoded in bulk speech at the expense of the act as a whole. On the other hand, other approaches fail because they focus on features of the bulk speech as a whole at the expense of the individual encoded acts. My proposed sincerity condition seeks to steer a path between these extremes.

2. SINCERITY AND THE PREFACE

An important concept developed in speech act theory, and one of the reasons its study has been so fruitful in domains such as ethics, is that of felicity. Felicitous (or “happy”) speech acts are those which function smoothly whereas infelicitous speech acts go, in some way, awry. Austin (1975) considered there to be two broad failure modes of speech acts: misfires and abuses. In the case of a misfire, something goes wrong with the set‐up of the speech act such that it does not, in an important sense, go off. This can arise for instance with institutional speech acts such as marrying when various prerequisites are not satisfied (the speaker is already married; there is nobody for the speaker to marry, etc.). Likewise misfires can occur if the speaker lacks the authority or standing to perform the kind of speech act they are attempting.⁵

Abuses, on the other hand, are cases in which the speech act goes off but is, we want to say, importantly defective. It is an abuse in the sense that the speech act is being used inappropriately and perhaps even deceptively. One of the most widespread kinds of illocutionary abuse is simple insincerity. Insincerity arises when a speaker performs a speech act that they are not properly committed to, or which expresses a mental state they lack (Green, 2009; Searle, 1969). Perhaps most obviously, it is insincere to assert a claim you do not believe or to make an “empty” promise. Other examples of insincerity might include asking a question you already know the answer to, issuing a warning when you consider there to be little or no actual danger, and inviting someone to your party when you do not actually want them to attend.

When we examine bulk speech, we are faced with how to make sense of sincerity. What if a politician intends to fulfil some of her promises but not others? What if she wants to follow through on all of them, but realises that this is infeasible and that some of the promises must go unsatisfied? To what degree is what she has said, then, sincere? These questions suggest that in order to get a good handle on how speech acts combine, we should examine more closely what occurs at the level of sincerity. It is to this that we now turn.

In a short paper from 1965, David Makinson describes what he calls the paradox of the preface. He notes that authors commonly write in the preface of a book that there may be—indeed probably are—errors therein, and that they take full responsibility for any such errors. Makinson observes that there is a tension between how the author believes all the individual statements in his book, yet also believes that some of them must be false.

The argument he gives can be fairly closely reproduced as follows:

• A1.In the course of his book a writer “asserts and believes” all of $s_1,\dots ,s_n$
• A2.Therefore he believes that all of $s_1,\dots ,s_n$ are true
• A3.He also “asserts and believes”: s _n+1 that there are errors in his text
• A4.Therefore he believes that at least one of $s_1,\dots ,s_n$ is false
• A5.But the beliefs in A2 and A4 are inconsistent
• A6.Therefore the author cannot rationally hold the beliefs in A2 and A4 at the same time.

This is almost always presented as a paradox about the logic of rationality and belief, although structurally analogous arguments have been presented concerning the logic of intentions and obligations (Goldstein, 2016; Immerman, 2019; Shpall, 2016). The idea is that the author is being rational when he believes all the claims he writes in his book. At the same time, to believe that there are potential mistakes in a book is not only plausible but reflects appropriate, perhaps even praiseworthy, epistemic humility. Consequently, the author arrives at a state where “he is being rational even though he believes each of a certain collection of statements, which he knows are logically incompatible.” (Makinson, 1965, p. 205, emphasis in original). However, to self‐consciously hold inconsistent beliefs is widely considered irrational. The paradox is that the author seems to be entirely rational at each step but arrives at a point where he is committed to something that is apparently irrational.

We are interested here in the paradox for somewhat different reasons. Specifically, it is of interest to us because it captures how communicating by way of such things as long texts can bring in unique complications. Moreover, it displays the way in which speakers, at some level, are aware of these issues and actively try to address them.

Since the connection to speech acts is not immediately obvious, we shall outline it here. Notice that Makinson's argument makes use of the fact that there is a tight connection between the attitude of belief and the speech act of assertion. More specifically, on many views, sincere assertion requires that the speaker believe what she is asserting (or that some other even stronger doxastic condition obtains). Implicit to premises A1 and A3 is that the writer is asserting sincerely in this sense. This presupposition allows us to move seamlessly from the matter of the author's assertions to the matter of his beliefs. Lost along the way in this move is the simple observation that this can also be read as a story about speech acts.

Here is another way to present the story, one which makes explicit the presupposition noted above⁶:

• B1.In the writing a book an author asserts all of $s_1,\dots ,s_n$
• B2.These assertions are all sincere
• B3.Therefore he believes all of $s_1,\dots ,s_n$ are true
• B4.He also asserts s _n+1: that there are errors in his text
• B5. s _n+1 is sincere
• B6.Therefore he believes there are errors in his text
• B7.Therefore he believes that at least one of $s_1,\dots ,s_n$ is false
• B8.But the beliefs in B3 and B7 are inconsistent
• B9.Therefore the author cannot sincerely make this combination of assertions

Stated this way, the preface paradox is a puzzle to do with sincerity. The problem is that the author seems to be simultaneously making a bunch of assertions that express conflicting beliefs, and that are therefore not simultaneously satisfiable. Moreover, the author is well aware of this. He knows that not everything he has said can hold at once. But despite this, his speech acts do not strike us as insincere or indeed in any other way infelicitous.

The paradox so understood needn't be about assertion at all. Other kinds of speech act may also be expressed en masse by being grouped into a large text and then presented as a whole. These too may be performed either sincerely or insincerely, where sincerity depends on the speaker having a particular mental state appropriate to the speech act they are performing (Green, 2009; Searle, 1969; Searle & Vanderveken, 1985). While asserting will call for the speaker to believe some proposition, demanding will call for her to desire a particular state of affairs, promising will call for her to have appropriate intentions, and so forth.

Adding to the complexity, bulk speech may be a mixed package in the sense that it may express more than one kind of speech act and thus invoke more than one kind of sincerity condition. The heady mix of mental states potentially entailed by sincerity conditions on different speech acts can be expected to have complicated logical interrelations. Not all will be compatible in the sense of simultaneous satisfiability. Some of these combinations will trigger problems due to the sort of out‐and‐out contradiction we have already seen, and some will depend on other various sorts of practical and pragmatic incompatibilities.⁷ In general, we should expect this most often when the relevant mental states are, in some way, responsive to the outside world. Thus, assertives and commissives are the most common hosts to these problems; for not every belief will turn out to be correct and not every intention will be acted upon. On the other hand, pure expressives, which display mental states such as gratitude, do not see these same problems—or at least, such scenarios are much more rare.

Consider the “calendar paradox” (Shpall, 2016) which adapts the preface paradox to intentions (see also Goldstein (2016) for related cases). In the calendar paradox, an agent fills out his monthly calendar in a manner he believes consistent. At the same time, he “is bothered by a skeptical thought. No matter how strenuous his exertions, it is highly unlikely that he will successfully execute the plan in its entirety.”(p. 809) The paradox arises because, according to the principal of agglomerability of intentions, it seems perfectly rational for the agent to combine his distinct intentions into one large intention (Bratman, 1987). It also seems rational for the ambitious agent to believe that, realistically, he will not achieve everything he sets out to do. But this means that he has formed an intention that he does not believe he will satisfy. This would appear to be irrational.

The calendar paradox can be easily adapted to a speech act context if the agent simply shares his calendar with others. Even more pointedly, he could issue a series of promises to carry out the various tasks on his calendar. Realising that he has filled up his calendar, the agent might include the caveat that something will probably go unfinished (though he knows not what). If he is sincere, this puts him in an analogous position to Shpall's agent. He is expressing a series of intentions alongside the belief that he will not satisfy their agglomerate.

To sharpen the issue somewhat, it seems intuitive that there is a form of insincerity in communicating information about your mental states that conflicts with your conscious beliefs about your mental states (Stokke, 2014). Given this, the final statement of these paradoxes—the explicit observation that there is something false in the book, that not all the month's appointments will be met—does not even seem necessary for the bulk speech to be insincere. Even if he does not assert them, the speaker harbouring such skeptical beliefs suffices for us to say that his bulk speech is insincere. This is because his speech would plausibly lead the audience to believe that he believes in the conjunction of his claims, or is committed to the agglomerate of intentions. And this would be misleading.

3. SOME POTENTIAL SOLUTIONS

Why not simply say that sincerity of a collection of speech acts is simply a matter of each individual speech act being sincere—regardless of the degree to which the author has confidence in their overall coherence? We could then say that the author is being sincere since he believes everything he writes and, crucially, does not know where the inevitable error lies. He believes each individual claim and does not disbelieve any claims in particular. On this view, the sincerity for bulk speech would require satisfying the following:

No Insincere Encoded Speech: the speaker satisfies the sincerity condition for each encoded speech act.

Our author believes of his book that something in it is false, though he knows not what. That is, given that assertions aim at expressing truth, he knows there is somewhere in his text a defective speech act. Nonetheless, he believes of each claim that it is true. We can therefore say, given the No Insincere Encoded Speech condition, that he is speaking sincerely because he satisfies the sincerity condition for each of his assertions. Analogous situations arise whenever a speaker performs a number of speech acts, believing some to be defective (but remaining ignorant of which). For instance, it might be that the speaker believes that her speech acts do not form a coherent whole, or that they are not all satisfiable together. This kind of situation is far from unusual in the cases of commissive speech acts such as promises. A sincere promise involves having a particular intentions (and for it to be fully happy, these must be acted upon). But a speaker might sign an agreement that commits them to a number of future actions, despite not intending to actually perform all of the acts. There might even be good reason for this such as that they are running up against constraints set by time or their own capacities. In this case, the individual sincerity conditions on the original agreement would call for a number of intentions that the speaker believes are not, taken all together, satisfiable. Nonetheless there are certainly situations in which this combination of speech acts would seem appropriate and unproblematic. The fact that she may simply not know which promise is going to have to go unmet seems, as with the author's ignorance of where his error lies, somewhat exculpatory.

All that said, it seems strange to disregard the speaker's own beliefs about the ultimate likely success of his communication as a whole. To see why this is, consider the following case: a sloppy researcher produces a book containing many claims. She believes each of these claims. However, she also knows that her research methodologies are poor. Her past books have been roundly debunked and there is a good chance that this one will be too. Indeed, the text may even be riddled with contradictory claims where she believes each possibility equally. In other words, she believes each individual assertion, but she believes that many—indeed most—of them will turn out false.

Similarly, recalling the calendar paradox, an enthusiastic but indecisive socialite might issue a statement promising to attend many overlapping events—uncertain which one she will actually be interested in attending on the day. Here the fact that she does not know which commitment will be acted up does not absolve her.

Both the indecisive socialite and sloppy researcher seem to be insincere in a way that Makinson's author is not. Yet they are not guilty of violating the No Insincere Encoded Speech condition. Each individual speech act is, it would seem, fine. The problem only arises when they are taken together.

Our sloppy researcher is not being sincere in the way that Makinson's author is. She is putting out a collection of assertions that she fully expects to be rife with errors. The fact that she cannot identify where those errors will be and genuinely believes each potentially erroneous claim does not get her off the hook. She is very clearly misleading her audience in some way. And similar cases can be constructed for other collections of speech acts. If, like the socialite, you sign onto many commitments while believing that you will fail to meet most of them (though you know not which) then you are similarly misleading your audience. These cases suggest that what it is to be sincere when issuing collections of speech acts is not wholly reducible to the sincerity of the constituent acts taken individually. There must be some wholistic criterion.

A plausible option is based on the solution to the original preface paradox proposed by Hannes Leitgeb. Leitgeb is sensitive to some of the problems we have been discussing:

Even for a couple of consecutive utterances, one may still take the speaker perfectly seriously in expressing their conjunction. However, as illustrated very nicely by the Paradox of the Preface, one should no longer do so once an author makes a great many of consecutive utterances, as it were, in one fell swoop, and when the context is such that the attention is directed toward this “mass utterance” taken as a whole: The linguistic acts of publishing a bulk of data or a book come with different conversational implicatures than the linguistic act of saying a single sentence (or saying one sentence after the other, but with the focus just on one at a time). When one utters a handful of declarative sentences, the default is to mean their conjunction; not so in the case of “mass utterance,” where the sheer amount of information displayed signals the author's fallibility. (Leitgeb, 2014, p. 13, all emphasis in original)

This seems in broad strokes correct. Taking a large number of linguistic acts together should be expected to have different pragmatic effects from performing them individually or in small sets. Communication is a subtle and context sensitive activity. There is no reason to assume that the norms active in simple conversation are the same as those active when publishing a large text or batch of data. In light of these observations, Leitgeb's suggested solution to the preface paradox is to say that in such cases the speaker is actually asserting something to the effect that “the vast majority” (p. 12) of what they have said is correct, or that “the frequency or statistical probability of a claim in the book being true is high” (p. 12). Presumably, sincerity about a mass utterance would be a matter of believing this assertion about the text as a whole, rather than each individual assertion.

This is an intuitive approach when considering instances of the paradox that deal only in assertions (as the original). It would need some tweaking to accommodate bulk speech featuring illocutionary acts other than assertion, but this is tractable. One could say, for instance, that the speaker believes that the vast majority of the speech acts will be satisfied: asserted content is true, intentions are ultimately acted upon, gratitude expressed is also felt, and so on. The sloppy researcher fails this modified statistical sincerity condition, as does the indecisive socialite.

That said, the statistical sincerity condition derived from Leitgeb's solution to the paradox still leaves much to be desired. One issue is simply with its applicability to real, rather than ideal, cases. Actual bulk communications do not generally contain a set of logically (probabilistically) independent speech acts as the solution presupposes. Instead, they are better understood as collections of interdependent speech acts, sometimes following an argumentative structure, and often featuring various sorts of cross‐referencing and dependencies. Moreover, not all encoded acts are equally important. It stands to reason that we might be significantly more concerned about the author's confidence in certain of the statements (for instance the main theses, or foundational premises) and less concerned with his confidence in others (a joke in a footnote). Some statements are more important than others; some claims are argumentatively load‐bearing.

Some of these issues could be addressed by including appropriate weights for the various embedded speech acts. Thus, sincerity in producing such a mass utterance would require that the speaker believe that a vast majority of the encoded illocutionary acts, weighted appropriately according to importance and relevance, will be satisfied.

However, even so patched, such as statistical sincerity condition fails to block at least two clear forms of illocutionary abuse. The first is a variety of speaker equivocation. As has been observed in Cevolani and Schurz (2017), Leitgeb's solution to the preface paradox is compatible with the author believing none of the individual claims (and, they argue, even entails this). This problem also befalls the sincerity condition derived from his solution. Thus, when challenged, the author can demur about individual claims so long as he still believes sufficiently many of them must be true.

The second potential abuse is what I'll call rhetorical swamping. Someone like the sloppy researcher could simply pad her writing with drivel she knows to be true. Including enough of this will satisfy the statistical sincerity condition, regardless of whatever nonsense is going on in the actually substantive parts of the book. Should important claims be weighted more heavily, she could simply add more irrelevancies. It is clear, I think, that this practice should be considered insincere (not to mention the fact that it would violate Gricean maxims such as quantity and relevance). However, the statistical condition does not rule it out.

A particularly severe version of the swamping problem is that the speaker can intentionally include false or otherwise deceptive statements so long as these are outnumbered by those for which she is in good standing or the falsehoods are considered sufficiently unimportant. The reason Makinson's author is not guilty of this is, of course, that he does not know where the inevitable errors lie and he has certainly not intentionally planted them (or so we assume). But if he had produced a book in which he had smuggled in some lies, this would be an act of deception even if the majority of the book's claims were perfectly honest. Indeed, in some cases the deception might be considered more vicious because of the use of the rest of the text as a mask to conceal the known falsehoods.

To summarise: the problem with No Insincere Encoded Speech was that it was insufficiently holistic. What has become clear is that the a statistical solution is too holistic: it calls for a belief about a property of the whole communication without requiring any sort of commitment to the individual encoded speech acts. The challenge is to find condition that sits comfortably in the Goldilocks zone between these two extremes. The remainder of the paper will be concerned with that task.

Let us call a speech act defective if it is in some way unhappy. Likewise, let us call a collection of speech acts defective if they cannot all be happy together. Issuing speech acts or combinations of speech acts that one believes to be defective is, it would seem, itself insincere. This suggests the following condition:

No Defective Subset: The speaker believes of each subset of speech acts issued together that it is unlikely to contain a defective speech act or combination of speech acts.

Intuitively, the idea is that sincerity over a large body of speech acts requires that the speaker have a high level confidence in both her individual encoded acts and combinations thereof. The condition entails that a sincere speaker believes of any one particular embedded speech act that it is unlikely to be defective and thus that she satisfies the sincerity condition for that speech act. So it captures what was required in the original No Insincere Encoded Speech condition. What it adds is the requirement that the speaker should also not believe of any larger subsets of the text that they contain defective speech acts, nor defective combinations of speech acts—for instance pairs of speech acts that are mutually incompatible.

Speaker equivocation and rhetorical swamping are both ruled out by this condition because in each case the speaker lacks the required confidence in the individual the encoded acts. At the same time, the sloppy researcher and indecisive socialite run afoul of the condition because they believe of various subsets of encoded acts that they are almost certainly defective even if they are not certain which specific individual acts will go amiss. The socialite is aware of blatantly incompatible promissory combinations, while the sloppy researcher knows that she is likely making a great many defective assertions. Indeed, she knows that it would require incredible luck for this not to be the case.

Unfortunately, however, this condition does not quite work for the example of Makinson's author. In fact, the author now appears to be in the position where he cannot sincerely produce his text given his uncertainty. He is in the uncomfortable position where while he is confident in each individual act encoded in his text, and most of the various combinations thereof, he has doubts about the combinations of all claims taken together. This uncertainty about the full text is entirely reasonable given those risks involved in speaking in bulk. After all, there is simply an increased potential to miss something or make an error in these situations. We are again back to saying that his very reasonable concern makes his overall communication insincere.

Moreover, this condition also does not help us in cases where a speaker has reason to have some concern about some smaller part of her communication. According to No Defective Subset, they cannot felicitously include this subset in their bulk speech. But in practice, because of the logical and argumentative complexity of these communications, it is often infeasible to excise every such weak point from the whole. These issues together mean that No Defective Subset diagnoses a huge amount of unproblematic bulk communications as insincere.

4. THE POINT OF THE PREFACE

At this point, it is worth taking a step back and considering why (besides cheekiness) Makinson's author might chose to include his prefatory comment at all. Thus far we have assumed that he is simply reporting on his personal epistemic quagmire: he is telling us that he is in the awkward (indeed, potentially paradoxical) position of believing all these claims despite knowing that at least one is probably wrong. Let us consider, however, the possibility that such a statement serves a distinctive communicative purpose—one that is an important ingredient in felicity presenting speech in bulk. The idea is that a sincerity condition on bulk speech requires that speakers identify what, among everything they have said, they may consider a potential source of errors. With this thought in mind, and guided by the insights gained in the earlier discussion, we are prepared at last to state the sincerity condition we endorse:

No Unmarked Defective Subset: The speaker believes of each unmarked subset of speech acts issued together that it is unlikely to contain a defective speech act or combination of speech acts.

What this means is that we do not demand perfection of speakers but we do demand transparency. The speaker is expected to mark the parts of their communication that they are worried about with the maximum of precision that they can manage. This, as we will discuss in more detail shortly, is precisely what Makinson's author is up to. Similarly, an author with one speculative chapter in a text that considers to be on an otherwise entirely secure epistemic footing, should note this and thereby alert the reader to the changing epistemic status of what they are reading. In general, a speaker will be on the hook for making explicit those combinations of encoded acts for which she has any doubt. Often these will be contiguous (for instance, the contents of the speculative chapter); but one could also have worries about how well some non‐contiguous parts mesh. If so, the condition demands that these too be marked. In general, then, the condition requires very simply that the speaker make explicit where the potential problems lie.

To get a sense of what this condition calls for, consider the case of a different, and sleepier, socialite. Like the indecisive socialite, the sleepy socialite makes a great many commitments over a busy week with the knowledge that he almost certainly will not be able to satisfy them all. However, unlike the indecisive socialite, he is pretty confident that if he misses anything it will be one of the evening events. He just gets tired sometimes. When asked by a friend what his plans are, he does not explain this. What he says instead is that he will try his best to make it to everything, but that he will probably miss one or two events. His friend goes ahead and makes contingency plans for his various potential absences throughout the week. Later, she learns that he only really had concerns about those things happening in the evening. Irritated, she complains that he should have told her that. She is, I think, justified in this frustration. Her friend did not lie, but he told a kind of half‐truth.

Let us revisit some of the other examples we have seen. Consider the sloppy researcher: she knows that the text as a whole is likely to contain many errors. However, flagging the text as a whole (as Makinson's author does) would be misleading. It would communicate that she does not suspect smaller subsets of the text when in fact she believes the text to be rife with errors. Because of this belief, she is on the hook to flag almost any reasonably sized subsection of her text. In other words, if the sloppy researcher were to present her book sincerely, she would have to aggressively pepper it with flags, warnings, and hedges.

The condition also avoids the problems of rhetorical swamping and speaker equivocation that we saw befall the statistical approach. The problem of rhetorical swamping was that a speaker could conceal speech acts that they believe would not satisfied by including them amongst a great many that are unproblematic. This is defanged since no matter how many other speech acts are included, the speaker would still be required to mark those she knew or suspected to be flawed.

The problem of speaker equivocation was that the speaker could sincerely issue a bulk speech act while not actually believing of any of the specific encoded acts that it would be satisfied. All they would need to believe would be that a great many of them will be satisfied. Thus, the author might not believe any of the particular statements in his book so long as he has confidence that enough of them will just happen to turn out true. But when we take into account the No Unmarked Defective Subset condition, this possibility is also dissolved. The author is required to mark any claims and combinations of claims he has doubts about. So, as with the sloppy researcher, the equivocator's text would have to peppered with hedges and flags.

In closing, let us consider one last time Makinson's author. Notice that because of this sincerity condition, not marking any part of a bulk communication carries with it an implicature that you are very confident in what has been said. Similarly, pointing to more specific weak spots in such a communication carries the implicature that there are not others (else they should be marked too). But what Makinson's author does do is mark the text as a whole. This represents an appropriate degree of epistemic humility (we all make mistakes) while also communicating overall confidence in the text. Yes, there may be an error somewhere, but he does not know of any particular weaknesses the reader should be aware of. The reader is assured that if the author was only really worried about, say, a speculative chapter, he would have said that. And this is because withholding such information, while not explicitly a lie, would nonetheless be misleading. It is this misleading quality which reveals that there is an underlying insincerity in such an act. So one of the reasons for including the sorts of prefatory comments Makinson's author makes is simply to avoid committing an infelicity that misleads the reader.

FUNDING INFORMATION

European Union's Horizon 2020 research and innovation programme. Grant Number: 740922

Paterson, G. (2022). Sincerity in bulk. Ratio, 35, 214–224. 10.1111/rati.12343