Pure Russellians are allowed not to believe

ABSTRACT

According to Pure Russellianism, if

• (1) David believes that Hesperus is a planet

is true,

• (2) David believes that Phosphorus is a planet

  is also true. It is also usually thought, by friends and foes of Pure Russellianism alike, that on it, when (1) and (2) are true,

•  (3) David does not believe that Phosphorus is a planet

 cannot but be false and because of this, many departed from Pure Russellianism. In this paper, I will show that by relying on the very explanation Pure Russellianism is usually combined with to account for sentences such as (1) and (2) and by acknowledging the well-attested linguistic phenomenon of metalinguistic negation, even if (1) and (2) are true, it is not the case that on Pure Russellianism (3) cannot but be false.

KEYWORDS:

• Russellian propositions
• Pure Russellianism
• propositional attitude reports
• the pragmatic explanation
• metalinguistic negation

Pure Russellianism is the view, first, that Russellian propositions are our objects of thought and, second, that sentences such as

(1)	David believes that Hesperus is a planet

express the holding of two-place relations between subjects and Russellian propositions. According to Pure Russellianism, if (1) is true,

(2)	David believes that Phosphorus is a planet

is true as well. It is also usually thought, by friends and foes of Pure Russellianism alike, that on it, when (1) and (2) are true,

(3)	David does not believe that Phosphorus is a planet

cannot but be false. Because of attributions such as (3), many departed from Pure Russellianism and introduced complex relations, peculiar attitudes and modes of presentation, acquaintance or familiarity, guises, ways of grasping or apprehending, vivid names, vehicles, or, to use the most recently used terminology, manners of taking propositions (Salmon 2018, 223).

In this paper, I will show that those alleged manners of taking propositions are in fact not needed to make it possible, on a Russellian account, to reject that (3) cannot but be false when (1) and (2) are true. My purpose is not to provide a general defence of Pure Russellianism or to reject manners of taking propositions. I will instead show that the story Pure Russellianism is usually combined with to account for sentences such as (1) and (2) is not obviously incorrect (§1) and it is then worth seeing where this very story takes us. I will then show that by relying on that very story and by acknowledging a very general and well-attested linguistic phenomenon involving negation, a phenomenon that has nothing to do with propositional attitude attributions or proper names in particular, we should conclude that it is not the case that on Pure Russellianism (3) cannot but be false when (1) and (2) are true. I will then argue that this solution to the prima facie problem created by sentences such as (3) perfectly respects the data, is not ad hoc and can be generalised, as it should. In particular, the solution can be extended both to attributions that do not involve proper names and to more complex, embedded attitude attributions, which have been considered to be issues for Pure Russellianism (§2). Contra a widely held opinion, the conclusion will be that we should not discard Pure Russellianism because of the truth-values it assigns to sentences such as (3).

1. Pure Russellianism, ‘Hesperus’ and ‘Phosphorus’

According to so-called Russellians and neo-Russellians, ‘in spite of all its snowfields Mont Blanc itself is a component part of what is actually asserted in the proposition “Mont Blanc is more than 4,000 meters high”’ (Russell 1904/1980, 169). Objects like Mont Blanc, you and me (rather than more exotic entities such as senses, aspects, ways of presenting, temporal phases, personae, etc.), are components of propositions as objects of thought about particular objects and as what is semantically expressed (relative to a context of use) by declarative sentences in which proper names occur as used. Pure Russellianism adds two tenets to these Russellian insights. First, at the level of attitudes, Pure Russellianism is the view that Russellian propositions are our full objects of thought, i.e. things like manners of taking propositions play no role. Second, at the level of the attributions of those attitudes, Pure Russellianism is the view that sentences such as

• (1) David believes that Hesperus is a planet

express the holding of two-place relations between subjects and Russellian propositions, rather than more complex relations.

One of the patent consequences of Pure Russellianism is that if (1) is true,

• (2) David believes that Phosphorus is a planet

is true as well: ‘Hesperus’ and ‘Phosphorus’ are co-referential proper names and the Russellian proposition that Hesperus is a planet denoted in (1) is the very same object as the Russellian proposition that Phosphorus is a planet denoted in (2). Since (1) and (2) simply express the holding of a two-place relation of belief between David and the same Russellian proposition, if (1) is true, (2) is true too.

This consequence of Russellianism is considered by many to be problematic, but Russellians advanced various well-known considerations to explain away the intuitions that (1) and (2) can differ in truth-value. As Burgess notes, when confronted with an apparent counterexample, Russellians typically ‘suggest that whatever phenomena are cited by their critics pertain to pragmatics rather than semantics’ (2005, 203) and a sketch of a pragmatic explanation is indeed what has been offered by Russellians to explain the intuitions about the truth-values of (1) and (2) away (Barwise and Perry 1983, 258–264; McKay 1981; Peacocke 1975, 127–128; Richard 1986/1987; Salmon 1986, 58–60, 114–118, 1989/2007, 198–199; Soames 1987, 2002, 228–229; Urmson 1968, 116–118). As Berg (1988, 363) Salmon (1986, 118) and Saul (1997, 106) noticed, we do use sentences such as

(4)	David believes that Hesperus is a planet, though he would never put it that way;
(5)	David believes that Hesperus is a planet, though he does not agree when it is presented to him via the very sentence ‘Hesperus is a planet’.

They maintain that (4) and (5) are attempts at cancelling what is pragmatically suggested by (1).¹ For our purposes, the details of what is pragmatically suggested by (1) are not important, and it might well in the end be something non-linguistic.² But since the cancelling is about putting a belief in a certain way or agreeing to a piece of information presented via a certain sentence, (1) is naturally taken to pragmatically suggest something along the following lines (Berg 2012, 63–70): that David would be willing to phrase his own belief using the words ‘Hesperus is a planet’ (Berg 1988, 357) or, if David does not speak English, a translation of that sentence into a language he is a speaker of (Soames 1987, 119), or that he would utter something along the lines of ‘Yes, Hesperus is a planet’, or would maybe more simply reply ‘Yes’ if we were to ask him ‘Is Hesperus a planet?’, or would accept ‘Hesperus is a planet’ (McKay 1981, 293), or take such a sentence as true (Gaskin 1997, 137). What is pragmatically suggested by (1) is of course to be taken to be different from what (2) suggests. It could be something along the lines that David would be willing to say ‘Yes’ if we asked him ‘Is Phosphorus a planet?’. Now suppose that David is a normal (Berg 2012, 57–62), stereotypical (Everett 2017, 42) speaker of English, he is sincere, reflective, etc. and replies ‘Yes’ when we ask him ‘Is Hesperus a planet?’, but shouts ‘No! No! No!’ when we ask ‘Is Phosphorus a planet?’. While what is pragmatically suggested by (1) is true, what is pragmatically suggested by (2) is false and this explains, Russellians maintain, why one might think that (2) itself, rather than what it pragmatically suggests, can be false when (1) is true: despite the truth-conditions of (1) and (2) being identical, one might think that (2) can be false even though (1) is true because one wrongly thinks that what is pragmatically suggested by the two sentences affect their truth-conditions.

Arguably, cancellability can be explained in ways different from holding that (1) and (2) differ only in what they pragmatically suggest (Recanati 1993, 332; Spencer 2006, 132–133). Yet, this seems a viable explanation.

But many object: speakers who received ‘some coaching’ are ‘often able to distinguish between the pragmatic implications and the semantic contents of utterances’ (Braun 1998, 570), but with attitude attributions ‘[e]ven professionals who are trained in semantics and pragmatics strongly resist the contention that the utterances express the same proposition’ (571. See also Davis 2005, 279; Green 1998; Recanati 1993, 338).

The objection can be resisted, though. First, even if speakers were often able to distinguish what sentences semantically express and what they pragmatically suggest, it is not clear that they always can, or that we should expect them to be able to when it comes to attitude attributions. Braun uses ‘and’, as it occurs in a sentence such as

(6)	Jules and Jim moved in together and got married

as an example of a case in which after some coaching speakers get the semantic/pragmatic distinction right. But, as Saul notices, sentences like (6) ‘are paradigm cases’ and it is ‘unreasonable to expect that all cases … should be as clear and obvious’. Moreover, she adds, ‘it’s not entirely impossible to convince people to entertain the idea that their intuitions about [attitude attributions] are mistaken’ (1998, 364).

Saul provides no data about the possibility of some coaching leading to a conversion to Russellianism, though. Moreover, in following her and then conceding that attitude attributions are not as clear cases as (6), Russellians weaken their positions. In fact, they do not need to concede this. For it is just not obvious that speakers, whether non-professionals who received some coaching (whatever that might consist in) or professionals who are trained in semantic and pragmatics, are often able to tell apart what is semantically expressed and what is pragmatically suggested. Concerning non-professionals, some think that the reverse is actually the case (for example, Richard 1997, 202) and we simply do not have sufficient data, so we should stay open to the possibility that they are not good detectors of the semantic/pragmatic divide (Crimmins 1992, 4). Moving to trained professionals, one might argue, contra Davis (2005, 279, 2017, 25), that we should just disregard their judgments, as they in fact received too much training. Moreover, if we should take their judgments into account, there are a few data that directly call into question that professionals are normally able to tell apart the semantic features and the pragmatic one. Davis argues that ‘[s]urely those of us working in pragmatics are not insensitive to the distinction’ (2005, 279). This is surely true, but when it comes to applying the distinction to various cases it seems that we are not all as good as one might have hoped. It is not just that, as Berg (2017, 109) notices in his reply to Davis, professionals regularly disagree on the truth-values of some sentences (one example for all, ‘The King of France is bald’), but they also disagree exactly on what belongs to semantics and what to pragmatics. Take

(7)	Johnny is young and he is experienced
(8)	Johnny is young but he is experienced.

Frege (1918/1956, 296), Grice (1961, 127) and then many others maintained that (7) and (8) are semantically on a par, differing only in that ‘but’ also pragmatically ‘intimates that what follows is in contrast with what would be expected from what preceded it’ (Frege 1918/1956, 296). But not everybody agrees and Bach (1999), for example, maintains that the alleged pragmatic feature of ‘but’ is really part of what (8) semantically expresses. Of course, Frege, Grice and Bach cannot all be right, but this is beside the point. What matters is that if professionals who received the relevant training are just able to see, in the particular cases, what belongs to semantic and what to pragmatics, it is a bit of a mystery that we have disagreement here. Who, among Bach, Frege and Grice do we want to say did not learn from the relevant training? The disagreement concerning ‘but’ is about a pragmatic suggestion that is traditionally classified as a conventional implicature in the Gricean framework. For our purposes, the framework we choose is irrelevant and we will take no stance on this.³ What is relevant is that one might reply here that the pragmatic suggestions generated by attitude attributions, if there are any, would not be classified as conventional implicatures within the Gricean framework (contra Thau 2002, 173–174) and then urge that, when it comes to conversational implicatures, professionals indeed just see where semantics ends and pragmatics begins. But this is not true: also in the case of conversational implicatures, things are not as transparent. Take scalar implicatures with numerals. The traditional Gricean account of an utterance of

(9)	I had three espressos

is that what (9) semantically expresses, as uttered by me, is that I had at least three espressos, while generating the conversational implicature that I had exactly three. Again, not everybody agrees. In particular, some maintain that the exactly is part of what (9) semantically expresses (Breheny 2008; Spector 2013). These debates show that it is not only in the case of attitude attributions that professionals who are trained in semantics and pragmatics disagree on what belongs to the former and what to the latter. Just as in the case of (7) and (8) (and numerals) the mere presence of professionals who strongly resist the contention that they are semantically on a par does not force us to conclude that the debate is over and the relevant phenomenon semantic, so the mere presence of professionals who strongly resist that

• (1) David believes that Hesperus is a planet

• (2) David believes that Phosphorus is a planet

are semantically on a par does not doom the pragmatic explanation Russellians suggested.

Despite the objection,⁴ the pragmatic explanation, or actually what is more aptly called a sketch of, a research programme for such explanation (Predelli 1999, 114), as we did not spell any details out, is then not obviously incorrect. Of course, this does not mean that the explanation is correct or that Pure Russellians are correct when it comes to the truth of (2), given the truth of (1). Yet, it does mean that it is worth seeing where all this can take us. In the rest of the paper, we will then simply assume that the sketch of a story we just saw Pure Russellians can tell about sentences such as (1) and (2) is correct, in order to see what that very story can tell us about attributions in which we say that subjects do not believe.

2. Adding negation

It is usually thought, by both friends and foes of Pure Russellianism, that one if its further consequences is that when

(1)	David believes that Hesperus is a planet
(2)	David believes that Phosphorus is a planet

are true,

(3)	David does not believe that Phosphorus is a planet

cannot but be false (Berg 2012, 70–71; Braun 1998; Crawford 2004a, 2004b; Heck 2014, 6; Jacob 1997, 222; Recanati 1993, 336; Salmon 1986, 100–101, 112-113, 1989/2007; Saul 1998, 372, 385; Schiffer 1987, 463; Soames 2002, 142). When (1) and (2) are true, the reasoning usually goes, the two-place relation of belief between David and the Russellian proposition that Hesperus is a planet, i.e. the Russellian proposition that Phosphorus is a planet, holds and then (3), according to which instead the very same relation does not hold, cannot but be false.

Those who defended Pure Russellianism bit the bullet (Berg 2012, 70–71; Crawford 2004a, 2004b), conceded that indeed on Pure Russellianism when (1) and (2) are true, (3) is simply false and protested that despite the fact that there might an intuition, a persistent and widespread one (Soames 2002, 142), it is not at all obvious that it is possible that (3) is not false when (1) and (2) are true (Boghossian 1994, 46), so that what Pure Russellians should do, if anything, is explain the intuition away (for an account in this direction, see Berg 2012, chap. 2). Other Russellians moved toward an Impure account (Braun 1998; Salmon 1986, 1989/2007) by holding that attributions such as (1), (2) and (3) should be taken to express the holding of a different and usually more complex relation, or introduced manners of taking propositions alongside Russellian propositions. In particular, Salmon, who thinks that we can explain away our intuition that (2) can be false when (1) is true in terms of the pragmatic explanation we saw in §1, exactly to account for the truth-value of (3), famously suggested (1986, 112, 1989/2007, 195–198) that the belief relation such as the one David is in if (1) is true is to be taken as the existential generalisation of the three-place relation BEL among subjects of attitudes, Russellian propositions and manners of taking propositions. The attribution (1) is true iff there is some manner of taking the Russellian proposition that Hesperus is a planet such that David stands in the BEL relation to it and the Russellian proposition that Hesperus is a planet. Having introduced the three-place relation BEL, on Salmon’s account, (3) can then come out true. Because we now have a three-place relation, the negation can be taken to be in different positions, so (3) has two readings. On one reading, (3) states that there is no manner of taking the Russellian proposition that Hesperus is a planet such that David BEL it and the Russellian proposition that Hesperus is a planet. This is false when (1) is true, as there is indeed one such manner. On the second reading, though, (3) states that there is a manner with which David does not BEL the proposition, and this can well be true, for example, the manner that presents Venus via ‘Phosphorus’ is such that the relation BEL does not hold among David, the Russellian proposition and such a manner. Then David does not BEL that proposition with that manner, David ‘withholds belief’ (1986, 111, 1989/2007, 197) when it comes to such a manner, and then on this reading (3) can well be true even though (1) is true. Salmon stresses that nothing along the lines of his correct account of (3) is available to Pure Russellianism:

no attempt to describe [a subject’s] complex doxastic state … can succeed using only the two-place notion of belief as a binary relation between believers and propositions. Without some relativized, ternary notion, and the resulting distinction between withholding belief and failure to believe, the attempt to describe [the subject’s] complex doxastic state with respect to the relevant singular proposition breaks down. The only thing one can say using the binary notion of belief – to wit, that [the subject] does believe the proposition … – is highly misleading at best. (1986, 112–113. See also Berg 2012, 116–122; Predelli 1999, 115n3)

In holding that on Pure Russellianism the only things one can say using the binary notion of belief are (1) and (2), friends and foes of Pure Russellianism alike are all incorrect, though. We do not need manners of taking propositions, three-place relations or the notion of withholding belief to have, on a Russellian account, that (3) can be taken not to be false even if (1) and (2) are both true. For, as we will see now, Pure Russellianism can indeed have that.

It should be recognised that it seems that we are running into absurdity in holding that on Pure Russellianism even though

(2)	David believes that Phosphorus is a planet

is true

(3)	David does not believe that Phosphorus is a planet

does not have to be false: (3) seems to simply and straightforwardly negate what (2) expresses. But when it comes to considering negation and what is pragmatically suggested by a sentence, it is well known that things are not that simple and straightforward and Pure Russellians can exploit this. In particular, Pure Russellians can rely on metalinguistic or irregular negation, a well-known phenomenon, famously brought into discussion by Horn (1989, chap. 6). Differently from what happens with regular negation, metalinguistic negation ‘is taken to be communicating an objection to some property of the representation falling within its scope … some property other than its truth-conditional semantic content’ (Carston 1996, 310); it ‘focuses not on the truth or falsity of a proposition, but on the assertability of an utterance’ (Horn 1989, 363). Metalinguistic negation is not to be taken as concerning necessarily the form of words, as in classic examples such as

(10)	You did not see two mouses.

For example, with metalinguistic negation we can object to an utterance token or type (Horn 1989, 377) falling within its scope also on etiquette grounds, as in another classic example of metalinguistic negation (Schnieder 2006, 44), i.e.

(11)	Her Majesty does not eat, she dines.

In fact, we can metalinguistically reject ‘on any grounds whatever, including the conventional or conversational implicata [an utterance] potentially induces, its morphology, its style or register, or its phonetic realization’ (Horn 1989, 363). So, and this is crucial given our purposes, the rejection can be on the grounds of what an utterance pragmatically suggests. For example, suppose I utter

(6)	Jules and Jim moved in together and got married.5

You know that Jules and Jim first got married and then moved in together. Even if you have been trained in semantics and pragmatics and you think that the order is only pragmatically suggested, in the scenario it would be very natural indeed for you to utter

(12)	Jules and Jim did not move in together and get married,

maybe adding ‘They (first) got married and (then) moved in together’. If we take the negation occurring in (12) as regular, the negation negates what (6) semantically expresses and, no matter what the order, if Jules and Jim both moved in together and got married, (6) is true and (12) cannot but be false. But if the negation occurring is (12) is metalinguistic, the negation is not negating what (6) expresses, but is grounded in what (6) pragmatically suggests. In the literature there is disagreement on various aspects of the phenomenon of metalinguistic negation (Carston 1996; Davis 2011, 2016; Foolen 1991; Geurts 1998; Horn 1989, chap. 6; Pitts 2011; Predelli 2013, 105–106). First, one might take (12), where the negation is metalinguistic, as expressing that (6) is to be objected to on the grounds of what it pragmatically suggests, and then take (12) as true, because (6) is indeed to be objected to on the grounds of what it pragmatically suggests, given that what (6) pragmatically suggests is false. But one might also instead maintain that when we use (12) metalinguistically, we are not expressing a truth-apt content, but we are rather performing a different kind of speech act, a speech act of rejection. If one goes this way, an utterance of (12) where the negation is metalinguistic is a correct speech act, as (6) is indeed to be rejected because of what it pragmatically suggests, but (12) is not truth-apt. Second, some maintain that in the case of metalinguistic rejection due to what an utterance pragmatically suggests, what is rejected is the utterance, on the grounds of what it pragmatically suggests. Others instead maintain that it is what is pragmatically suggested that is rejected. Pure Russellians need to take a stance on neither of these issues. No matter whether what is rejected by using (12) metalinguistically is (6) on the grounds of what it suggests, or rather what (6) suggests, and no matter whether or not by using (12) we produce a truth-apt content, on all accounts even though (6) is true, (12) does not have to be taken as false, but rather as true or correct, and this is all Pure Russellians need to address the objection regarding attributions concerning what subjects do not believe.

Salmon urges: ‘I am not suggesting that the case of propositional-attitude attributions is exactly analogous to the conjunction … (The issues here are quite delicate.)’ (1989/2007; 201n11. See also Richard 1990, 120). Issues here are surely delicate. Yet, I think Pure Russellians can indeed suggest that the case of propositional-attitude attributions is exactly analogous to the case of conjunction, as in this way they can provide a satisfactory account of

(3)	David does not believe that Phosphorus is a planet,

one moreover that relies on a well-known phenomenon. For Pure Russellians can maintain that the negation occurring in (3), like the one occurring in (12), can be taken both as regular and as metalinguistic. If the negation is regular, surely (3) expresses the negation of what

(2)	David believes that Phosphorus is a planet

semantically expresses and it cannot but be false when (2) is true. But Salmon is wrong to hold that the only thing one can say using the binary notion of belief is that David does believe the proposition that Phosphorus is a planet. Pure Russellians can in fact maintain that the negation occurring in (3) can also be taken as metalinguistic and if the negation is metalinguistic, (3) is not false in the case at hand, but rather true or correct. As we saw in §1, according to the pragmatic explanation we are assuming here, (2) pragmatically suggests something beyond what it semantically expresses, maybe, we saw, that David would be willing to put his belief as ‘Phosphorus is a planet’. This is false in the case at hand, as he is shouting ‘No! No! No!’ when we ask him ‘Is Phosphorus a planet?’ and would be willing to utter ‘Phosphorus is definitely not a planet, it is a hill’. But then in the scenario, despite (2)’s truth, what (2) pragmatically suggests is false and then it is correct to metalinguistically object to (2) on the grounds of what (2) pragmatically suggests, or to object to what (2) pragmatically suggests.⁶

There are various things to note. First, we should distinguish the suggestion from two other accounts in its vicinity. Heck thinks that Pure Russellians should say about (3) that ‘speakers utter it because it (sometimes) implicates things that are true and that they wish to communicate’ (2014, 6). I am not suggesting that the reason why we might utter (3) is that (3) pragmatically suggests something we might want to communicate. Rather, I am suggesting that we might utter (3) also because the negation occurring in it can be taken to be metalinguistic, correctly objecting to (2) or what (2) pragmatically suggests. We should also not be tricked by the label metalingustic negation. The suggestion is not tantamount to holding that (3) is true because it is somehow equivalent to

(13)	David does not believe that ‘Phosphorus’ is the name of a planet

or

(14)	David does not believe that ‘Phosphorus is a planet’ is true

(Cf. Davis 2005, 268, 2016, 300–304; Rieber 1997 for discussion). On Pure Russellianism, sentences (13) and (14) are surely true in our scenario, but for a different reason: they are true on a regular interpretation of the negation, since they express the not holding of relations of belief between David and the Russellian metalinguistic propositions that ‘Phosphorus’ is the name of a planet and that ‘Phosphorus is a planet’ is true, and this is indeed the case. These Russellian propositions are different from the proposition that Phosphorus is a planet, which David in fact believes. The reason why it is not the case that on Pure Russellianism (3) cannot but be false is not that (3) is equivalent to (13) or (14). Rather, (3) is not false because it is correct to metalinguistically reject (2) or what (2) pragmatically suggests.

Second, we can note some of the virtues of the proposal. In holding that on Pure Russellianism

(3)	David does not believe that Phosphorus is a planet

is true or correct in the case under discussion, we account for the truth-value of the sentence in a way that perfectly respects the data. According to Pure Russellianism, if I am right, (3) is true (if an utterance of (3) produces a truth-apt content) or correct (if it is a performance of a speech act of rejection), when the negation is taken as metalinguistic, and is false, when the negation is taken as regular. This is a welcome result indeed: it is exactly because it seems both that (3) does not have to be false even though

• (1) David believes that Hesperus is a planet

is true and that (3) cannot but be false because after all Hesperus is Phosphorus, that we have a puzzle to start with. Instead of explaining one of the pieces of data away, in acknowledging metalinguistic negation Pure Russellianism manages to account for both the apparently conflicting pieces of data, by showing that they are not conflicting after all.

Another virtue of this Pure Russellian account of (3) is that metalinguistic negation is a cross-linguistic, well-attested general phenomenon, which has nothing to do primarily with attitude attributions of proper names, so that the account is not ad hoc. Because the phenomenon is general, the explanation moreover easily generalises, as it should (Predelli 2001, 312). While the alleged problem created by sentences such as (3) is often phrased as a problem involving proper names, similar cases can be created with other pieces of language, such as ‘kechup’ and ‘catsup’, or ‘courgettes’ and ‘zucchini’ (Crawford 2004b, 224; Salmon 1989/2007, 212). Since metalinguistic negation is not tied to proper names, it can be relied on in accounting for any set of sentences of the form S believes that p, S believes that q, S does not believe that q, where p and q express the same Russellian proposition.

Acknowledging metalinguistic negation, moreover, allows Pure Russellianism to account for further more complex attributions, often presented as creating issues for Russellianism. Here is the prima facie problem (Schiffer 1987, 463): you know very well that Hesperus is Phosphorus. Also, suppose,

(15)	You believe that David realises that Hesperus is Hesperus

is true. But, Schiffer holds, according to Pure Russellianism, since (15) is true, so is

(16)	You believe that David realises that Hesperus is Phosphorus.

But, given the case at hand, it seems that also

(17)	You believe that David does not realise that Hesperus is Phosphorus

is true. Now, how can you rationally think that David both realises and does not realise what you know is the same, as you know that Hesperus is Phosphorus? Also in this case of iterated attitudes, Salmon thinks that his BEL relation and his manners of taking propositions are doing the trick (1989/2007, 215-216). But acknowledging metalinguistic negation seems in fact all a Russellian needs. Just like in the case of (3), also the negation occurring in

(18)	David does not realise that Hesperus is Phosphorus

can be taken as regular or metalinguistic. If the negation is regular, it negates what

(19)	David realises that Hesperus is Phosphorus

semantically expresses. You believe what (19) expresses, so that

(16)	You believe that David realises that Hesperus is Phosphorus

is true. If the negation occurring in (18) is metalinguistic, though, it rejects (19) itself or what (19) pragmatically suggests. Since what (19) pragmatically suggests, i.e. something along the lines that David would be willing to utter ‘Hesperus is Phosphorus’, is false, you are correct in believing that (18) is true or correct. If (18) is true when the negation occurring in it is taken to be metalinguistic, then

(17)	You believe that David does not realise that Hesperus is Phosphorus

is true as well. If instead an utterance of (18) does not produce a truth-apt content but is rather a different speech act of rejection, it is still the case that you believe that it is it correct to perform the speech act of rejecting (19) or what (19) pragmatically suggests, so something close to (17) is true. You are also fully rational in having these beliefs because, differently from what Schiffer maintains, this is not (regular negation) thinking that David both realises and does not realise the same thing.

Finally, we should consider some potential problems for the account suggested. First, one might protest at this point (Davis 2005, 278; Heck 2014): even though on Pure Russellianism

(3)	David does not believe that Phosphorus is a planet

is true or correct when the negation is taken to be metalinguistic, still Pure Russellianism makes (3) true or correct for the wrong reason, so to say, as sentences such as (3) explain David’s behaviour and it is because David does not (regular negation) believe that Hesperus is Phosphorus that we understand why he might be thrilled when we tell him that Hesperus is visible, but completely unmoved when we tell him that Phosphorus is visible.

This is an objection to Pure Russellianism that has nothing to do with (3) in particular. The same objection is in fact usually raised regarding

• (1) David believes that Hesperus is a planet

• (2) David believes that Phosphorus is a planet.

Only if we take (2) to be false, the objection goes, can we explain why David, who adores planets, is unmoved by our saying ‘Phosphorus is visible’. To address the objection concerning (3), there is nothing Pure Russellians need to say beyond what they already need to say about (1) and (2) (Berg 2012), i.e. that David’s behaviour is explicable by the fact that alongside believing what the true (1) and (2) attribute to him, and that Hesperus is Phosphorus, David also believes that Hesperus is not Phosphorus and also, suppose, that Phosphorus is a boring hill, that Hesperus’s only name in English is ‘Hesperus’, that planets are exciting, that there is one planet called ‘Hesperus’, etc. Whether or not this is a good explanation, it is part of the explanation we saw we are allowed to assume for the sake of our argument, and there is no reason to think that it breaks down when it comes to (3).

The second objection concerns metalinguistic negation in particular. While nobody defended the account I am putting forward here, the proposal has been briefly discussed, to quickly dismiss it (Davis 2017, 23; Jacob 1997, sect. 2.2; Recanati 1993, 343–345; Green 1998, 75–77). Let’s start from Jacob, who is concerned with the claim that we can take (3) as true because of what (2) pragmatically suggests. He holds that if Pure Russellians go the way suggested here, then they should provide an explanation of how ‘the insertion of a negation, yielding [(3)] from [(2)], has converted something which was merely communicated by an utterance of [(2)], not part of the truth-conditions of an utterance of [(2)], into a constituent of the truth-conditions of an utterance of [(3)]’ (1997, sect. 2.2). Jacob then abandons the proposal on the grounds that there is no such explanation. But, as we saw, an explanation is available and is moreover not ad hoc: metalinguistic negation can be taken to do the job. Davis and Recanati do consider metalinguistic negation as the explanation that Jacob thinks needs to be provided, but both maintain that it cannot work. Davis briefly urges that (3) ‘does not seem at all irregular’. Suppose that David simply has no attitude concerning whether Mars is hot. The attribution

(20)	David does not believe that Mars is hot

is true when the negation is taken to be regular, negating what

(21)	David believes that Mars is hot

semantically expresses. Davis holds that (3) ‘is much more like the perfectly regular [(20)]’ (2017; 23. See also Davis 2011, 2016), but just does not explain why Pure Russellians should agree. Maybe, Davis’s objection is supported by the thesis that speakers just see whether a negation is regular, in particular when it comes to attributions (Green 1998, 78), but this thesis is questionable. As we saw in §1, it is questionable that speakers get the semantic/pragmatic distinction right and it seems similarly questionable that they just see whether a negation is regular or not. Let’s go back to

(9)	I had three espressos.

Suppose I had exactly four espressos, you know this, you are concerned with my blood pressure and you ask me how many espressos I had. If I utter (9), it is natural for you to reply

(22)	You did not have three espressos,

maybe adding ‘you had four’. Were it transparent that the negation in your utterance is regular, we would have sorted the issue about what ‘three’ semantically expresses and what it pragmatically suggests a long time ago, but we did not. While Davis doesn’t give a reason why a metalinguistic treatment of the negation occurring in the true (3) would be incorrect, Recanati does give one such reason: ‘the metalinguistic interpretation has to be specially ‘marked’ or indicated. Nothing of the sort is the case with [(3)]’ (Recanati 1993, 344). It is not clear, though, that Recanati is correct on the need for metalinguistic negation to be specially marked. Again, even if in (22) you did not specially mark anything, it would be too big a jump to conclude that since no marking seems needed in those cases, then ‘three’ semantically expresses exactly three. Surely Recanati is right that often metalinguistic negation is specially marked, but this is the case with attitude attributions too. We might indeed find ourselves saying: David does not believe that Phosphorus is a planet. Of course he does, because he believes that Hesperus is a planet, but David does not (special marking, maybe a different tone of voice, or gesturing scare quotes with our fingers in the air) believe that Phosphorus is a planet. He thinks that Phosphorus is a hill.⁷

3. Conclusion

There might be various reasons to think that we need to reject Pure Russellianism and introduce manners of taking Russellian propositions or maybe depart completely from Russellianism. But what Pure Russellianism can say about sentences such as

(3)	David does not believe that Phosphorus is a planet
(17)	You believe that David does not realise that Hesperus is Phosphorus

should not be taken to be one of those reasons. For what many think Pure Russellians should be able to say about (3) and (17) but cannot say is in fact something that Pure Russellians can indeed say about those sentences.

Acknowledgements

Thanks to Marco Santambrogio and an anonymous referee for their helpful comments and suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Notes

1 Pragmatic import comes from a sentence’s being used in some context. When nothing relies on this, for ease, I will speak simply of sentences and not of pairs of sentences or clauses and contextual factors or indexes, or of utterances.

2 This would account for puppies and babies, if we should take them to be able to believe that Hesperus is a planet. Even if we could keep the linguistic character of what is pragmatically suggested, moreover, some details would still need fixing. For example, as Richard notes, ‘ascriptions involving demonstratives and indexicals, like “Smith believes that I am tired”, obviously carry no such implication’ (1986/1987; 260. See also Soames 1987, 119), as Smith would not put his belief as ‘I am tired’. These details do not matter here.

3 On how the pragmatic explanation might be cashed out, and in particular on whether it should be cashed out in terms of Gricean implicatures, (see Berg 1988, 363, 2012; Davis 2017; Everett 2017; Green 1998; Saul 1998; Spencer 2006).

4 For another objection and replies, (see Bezuidenhout 1997, 206–208; Davis 2005, 279n23). For still another objection, (see McKinsey 1999, 522–525). McKinsey concedes that he ‘cannot prove that no pragmatic explanation of the Russellian sort will work’ (551n9) and this is all we need here.

5 As Saul notices (1998, 364), it is almost universally accepted that (6) pragmatically suggests that Jules and Jim first moved in together and then got married, but this is not part of what the sentence semantically expresses. But nothing relies on the particular example we choose. If you think that the order of the events is semantically expressed by (6), take instead a case that you find uncontroversial.

6 Salmon holds that if what is pragmatically suggested by attitude attributions is a Gricean implicature, it should not be taken as a conversational particularized one (1989/2007, 201n11). Detecting metalinguistic negation seems to square perfectly with this, if Davis is right that metalinguistic negation can be used to reject implicatures, but not when they are particularized ones (2011, 2255). Carston is one of the few to maintain that the order of the events is semantically expressed by (6). She notices that if the order is only pragmatically suggested and (6) true, then a true (12) can only express the falsity of what is pragmatically suggested, and objects that the negation cannot do that as shown, she thinks, by the following case: ‘Consider the following exchange, where (19a), in context implicates (20):

(19) a. A: Jones has made a lot of visits to New York lately

b. B No, he hasn’t

(20) Jones probably has a girlfriend in New York

The question is whether (19b) could be taken to bear not on the explicit content of (19a), but only on its implicature (20), leaving the truth of (19a) intact. Surely the answer is no’ (Carston 1988, 172). Since this is a case of particularized implicature, the example seems unable to necessarily show something about (12), or (3).

7 While Recanati only speaks in terms of marking, it should be noted that none of the various test that, starting from Horn (1989, 392–413), have been suggested to ascertain whether a negation is metalinguistic survived scrutiny. Problems have in fact been detected for each and some take the tests as sufficient but not necessary conditions (Pitts 2011, 348–356), while others only as suggestive, being neither necessary nor sufficient (Geurts 1998, 280; Predelli 2013, 106).