Rollercoasters are not Fun for Mary: Against Indexical Contextualism

Abstract

Indexical contextualism (IC) is an account of predicates of personal taste (PPTs) which views the semantic content of PPTs as sensitive to the context in which they are uttered, by virtue of their containing an implicit indexical element. Should the context of utterance change, the semantic content carried by the PPT will also change. The main aim of this paper is to show that IC is unable to provide a satisfactory account of PPTs. I look at what I call “pure” IC accounts and show that because they fail to respect empirical data regarding disagreements where neither person is at fault, known as “faultless disagreements”, they must be rejected. I then go on to consider what I call IC “plus” (IC+) accounts. Such accounts attempt to account for the faultlessness of such disagreements using a simple indexical semantics, whilst introducing some extra ingredient to account for the disagreement part. I focus on two main versions of IC+: Gutzmann’s (in: Meier, van Wijnberger-Huitink (eds) Subjective meaning: alternatives to relativism, De Gruyter, Berlin, 2016) expressivist account, and López de Sa’s (in: García-Carpintero, Kölbel (eds) Relative truth, Oxford University Press, Oxford, 2008; Erkenntnis 80(Supp 1):153–165, 2015) presuppositional account. I discuss some internal worries with these accounts before going on to some final remarks about IC/IC+ in general. I conclude that neither IC nor IC+ can provide a satisfactory semantics for PPTs.

1 Introduction

Indexical contextualism (IC)¹ is an account of predicates of personal taste (PPTs) which views the semantic content of PPTs as sensitive to the context in which they are uttered, by virtue of their containing an implicit indexical element. Should the context of utterance change, the semantic content carried by the PPT will also change. The main aim of this paper is to show that IC is unable to provide a satisfactory account of PPTs.

First, I will consider what I call “pure” IC accounts, wherein PPTs are given a simple indexical semantics. As has been pointed out by many,² one severe consequence of pure IC is its inability to account for so-called “faultless disagreements” that seem to arise from uses of PPTs. That is, pure IC theories are unable to give an account of disagreements where neither speaker is at fault. It will be shown how pure IC is only able to account for the disagreement part or the faultlessness part-not both.

Pure IC proponents are happy to dismiss faultless disagreements as not being cases of genuine disagreements to avoid this worry. I show, however, that the contextualists are not in a position to simply ignore the phenomenon. I place importance on empirical findings collected by Solt (2018) and others to demonstrate that ordinary speakers take faultless disagreements to be a real phenomenon. Hence, IC cannot ignore faultless disagreement and must have an account of it. Because they are unable to do so, I suggest that pure IC must be rejected.

After my discussion of pure IC, I will consider what I call Indexical Contextualist “Plus” (IC+) theories, where the faultlessness aspect is accounted for using a simple indexical semantics, but the disagreement aspect is accounted for by an extra ingredient. The focus will be on two versions of IC+: Gutzmann’s (2016) expressivist account, and López de Sa’s (2008, 2015) presupposition of commonality account.

I conclude that, to some extent, IC+ succeeds where pure IC accounts fail. They respect empirical data, by attempting to account for faultless disagreement. However, there are some internal worries about the extra ingredient that these approaches put forth to account for disagreement. Firstly, we will see that PPT-sentences do not appear to behave like expressives under projection, thus it seems that PPT content is not of the same kind as expressive content. Secondly, through presupposition detection tests, we shall see that PPTs do not appear to carry a presupposition of commonality with them as claimed.

I will end the paper with some final remarks which raise problems for both pure IC and IC+ accounts. I conclude that neither IC nor IC+ can provide a satisfactory semantics for PPTs.

2 Predicates of Personal Taste and Faultless Disagreement

Broadly construed, predicates of personal taste (PPTs) are words that express our tastes, paradigm examples being ‘fun’ and ‘tasty’. The most remarkable feature of PPTs is the personal aspect they carry with them. There seem to be no matters of fact whether ‘this lager is tasty’ or ‘the Big Dipper is fun’. Rather, it is down to each speaker’s own preferences whether such things are the case. The personal aspect of PPTs appears to produce puzzling disagreements, called faultless disagreements. As an illustration, consider the following examples:

• (1) John: The Big Dipper is fun!³

  Mary: No, it’s not!

• (2) John: The Young Ones first aired in 1982.

  Mary: No, The Young Ones did not first air in 1982.

In both examples, two speakers are expressing apparently contradictory propositions—Mary is denying the content expressed by John. However, (1) and (2) diverge when we consider whether either speaker has uttered anything false. In (2), since The Young Ones first aired in 1982 Mary has uttered something false. We get the intuition of a disagreement which is explained by the presence of a contradiction between the two speakers, and we also get the intuition of one speaker being ‘at fault’—one of the speakers has expressed a false proposition.

The intuitions differ in (1). Here we have two contradictory sentences, yet we do not feel that either speaker is ‘at fault’ for expressing a false proposition. Regardless of a feeling of a contradiction, we’re compelled to say that the propositions expressed in (1) are both true.⁴

A satisfactory semantic theory must address the intuition of faultless disagreement in paradigm examples like (1).⁵ If an account is to capture faultless disagreement, it must explain both the faultlessness part and the disagreement part. If an account is to reject such a disagreement, it must give us a good explanation for why we have this intuition in the first place.

3 Semantic Relativism

Debates surrounding PPTs often focus on two opposing approaches—contextualism and semantic relativism. Semantic relativists claim that they can best account for PPTs as they give a neat explanation of faultless disagreement. Contextualists often present their account as a response to relativism. Because of this, it will be fruitful to establish some background. My focus will be on Lasersohn’s (2005, 2017) relativism.⁶

Lasersohn uses a Kaplanian semantics (see Kaplan 1989) in building his semantic relativism. Following Kaplan, we distinguish between the context of utterance and the circumstance of evaluation. The context of utterance determines the proposition, as is clearly illustrated with indexicals. The linguistic meaning of the indexical ‘I’ is something like ‘the speaker of the utterance’. Whenever ‘I’ is used the speaker of the utterance is picked out of the context of utterance. This results in the speaker being part of the proposition. As such, if ‘I’ is uttered at a different context of utterance a different content will be expressed (if the speaker is different at the context of utterance). Once the proposition is determined, truth-values are assigned at the circumstance of evaluation. For Kaplan, circumstances of evaluation consist of world and time parameters. Thus, the truth of a proposition will depend on how things are at the world and time at which it is evaluated.

Lasersohn keeps with the distinction of the context of utterance and circumstance of evaluation, however, it should be made clear that he does not take PPTs to behave in the same way as indexicals. The content of the PPT will stay the same in all contexts of utterance—‘fun’ will have the same meaning for both John and Mary. The main tenet of Lasersohn’s account is the introduction of the judge parameter into the circumstance of evaluation. As a result, it will be relevant to the truth-conditions which judge assesses the proposition involving PPTs.

This account explains paradigm examples of disagreement in a straightforward manner. When John and Mary disagree, their utterances express the same content with respect to the meaning of the PPT. Mary’s utterance expresses the denial of John’s utterance and thus the two propositions expressed are incompatible with one another. Both speakers have uttered something true because the judge parameter picks out different judges for each of their propositions. The truth of John’s proposition is evaluated with John as the judge, and the truth of Mary’s proposition is evaluated with Mary as the judge.

(3)	John: The Big Dipper is fun.	[true, relative to John as the judge]
	Mary: No, it’s not!	[true, relative to Mary as the judge]

For Lasersohn, the judge parameter is only a feature of the circumstance of evaluation and not the context of utterance. Although there may be a default coincidence between the speaker of the utterance and the judge of the circumstance, this is wholly contingent and there is no obstacle to evaluating a proposition as true relative to a judge that is distinct from the speaker who uttered that proposition. This allows Lasersohn to explain how we can speak truthfully relative to tastes other than our own. For example, if I am trying to cheer up a child, I might utter:

(4)

Gobstoppers are tasty.

[true, relative to the child as the judge]

Even if I do not find Gobstoppers tasty, I know the child will. Thus, (4) can be true relative to the child as the judge, but false relative to me as the judge. Lasersohn calls perspectives where we take someone other than ourselves to be judges exocentric, and ones where we take ourselves to be judges autocentric (Lasersohn 2005, p. 670).

Lasersohn’s relativism allows each speaker to have uttered something true. In (3), both speakers have expressed true propositions because the judge parameter has picked out each of them as judges. Consequently, the faultlessness of faultless disagreement is accounted for. Neither speaker is at fault, because neither speaker has uttered anything false.

The disagreement aspect of faultless disagreement is explained by the presence of a contradiction. We have a disagreement because one speaker contradicts the other. To ensure a contradiction we must have stability in content between both utterances—one must be the negation of the other. If this condition is not met, then there will be no genuine disagreement, rather two speakers will be ‘talking past one another’, as illustrated below:

• (5) John: The Big Dipper is fun [for John]!

  Mary: No, it’s not fun [for Mary]!

• (6) John: The Big Dipper is fun₁.

  Mary: No, it’s not fun₂.

In (5) and (6) there is no genuine disagreement about the content, rather the speakers are ‘talking past one another’. Essentially, they are talking about different things.⁷

4 Pure Indexical Contextualism

The purpose of this section is to give some background on what I call ‘pure’ Indexical Contextualism (IC). I call such accounts pure IC accounts, as PPTs are explained by a simple indexical semantic framework, and thus faultless disagreement is also explained (or explained away) using a simple indexical semantics.⁸

A clear example of a pure IC account is provided by Cappelen and Hawthorne (2009, henceforth C&H), who propose that the interpretation of a PPT is dependent on a direct reference to an individual/group.⁹ They start with the predicate ‘filling’ (later using more traditional PPTs) and claim that “on an occasion of use, a predication of ‘filling’ to some item will tacitly relate that item to a particular individual or group”, where “‘That is filling’, as made by X, where ‘that’ refers to Y, will express the proposition that Y is filling for X” (Cappelen and Hawthorne 2009, p. 103).

The indexicality in their account lies in the use of a PPT picking out the relevant individual/group for whom the taste applies. Thus, if John is taking an autocentric perspective and utters “The Big Dipper is fun” the proposition expressed is “The Big Dipper is fun [for John]”. Similarly, when Mary, speaking autocentrically, utters “The Big Dipper is not fun”, what she is expressing is “The Big Dipper is not fun [for Mary]”.

Just like relativist accounts, IC allows for exocentric perspectives. This explains different types of disagreements, especially those concerning tastes that are other than those of the speaker. Consider two cases that C&H mention. Firstly, we might have “correction” cases (Cappelen and Hawthorne 2009, p. 110). For example, Mary might be reminding John that he actually did not find the rollercoaster fun. The propositional content of such disagreements is represented below:

• (7) John: The Big Dipper is fun [for John].

  Mary: No, it’s not fun [for John].

Mary might go on to point out that John complained about how bored he was during the whole ride, thus justifying her claim that the Big Dipper is not fun for John. In such cases, there is a clear contradiction, however, one person will be at fault as it will either be true or false whether the Big Dipper is fun for John.

Another example of an exocentric disagreement is that of a group disagreement (ibid. pp. 110–111). John’s and Mary’s disagreement concerns whether the Big Dipper is fun for the group, as represented below:

• (8) John: The Big Dipper is fun [for this group].

  Mary: No, it’s not fun [for this group].

Again, there is clear disagreement in the sense that two contradictory propositions are expressed. However, there will be a matter of fact whether the Big Dipper is fun for the group, thus the faultlessness intuition is not present in such cases.¹⁰

Although both cases secure disagreement, they are not cases of faultless disagreement. The difference between these examples and our paradigm example is that in (7) and (8) both speakers are not taking autocentric perspectives. It seems that faultless disagreements only arise when both speakers are speaking autocentrically.

It appears that, under IC, faultless disagreement is not attainable. Firstly, to capture the faultlessness intuition both speakers must be able to express a true proposition. If this is the case then, as noted in Sect. 3, the propositions expressed by John and Mary would be consistent, as demonstrated below:

• (9) John: The Big Dipper is fun [for John]!

  Mary: No, it’s not [fun for Mary]!

We see that faultlessness is captured as both Mary and John have expressed true propositions. However, contradiction is no longer present as the propositions expressed are compatible. Consequently, IC fails to secure disagreement of the sort illustrated in our paradigm examples.

To capture disagreement, we need to make sure that the propositions expressed are determined by the same context of utterance. In ‘correction’ and ‘group’ cases, the PPT is indexed to the same individual which allows for disagreement in terms of a contradiction. There, however, we saw that although disagreement is captured, faultlessness is not.

4.1 A Pure Indexical Contextualist Response

It seems that pure IC is not able to account for faultless disagreement. However, the proponents of pure IC do not think this is problematic as they do not take faultless disagreement to be a real semantic phenomenon. Glanzberg, defending a version of pure IC, writes:

From a traditional, non-relativist, point of view, this idea [of faultless disagreement] is prima facie absurd: if two propositions express disagreement, one must fail to be correct […] I have not argued for this conclusion directly, but I have argued that sober reflection on the semantics of predicates of personal taste gives us no reason to accept any notion like faultless disagreement¹¹

(Glanzberg 2007: 16)

The quote above tells us that such an account simply rejects the possibility of faultless disagreement. If we can have a rich semantic profile of PPTs without the notion of faultless disagreement, there is no reason for us to accept such an ‘absurd’ phenomenon.

C&H also reject faultless disagreement; however, they provide an explanation of why we might intuitively accept such a phenomenon. C&H claim that the relativist over-estimates the data and we do not have genuine disagreements when both speakers take autocentric perspectives. This, of course, would mean that there is no space for faultless disagreement within such a framework. C&H write: “we do not ourselves find anything very compelling about the purported intuition of faultless disagreement. Cases where the sense of no fault runs deep are ones where the sense of disagreement runs shallow” (Cappelen and Hawthorne 2009, p. 132). I take the claim that with faultlessness ‘disagreement runs shallow’ to mean that there is no actual disagreement in cases of faultless disagreement.

C&H put our mistaken belief in faultless disagreement down to semantic blindness (Cappelen and Hawthorne 2009, p. 118). We are blind to the fact that our standards are different and thus fail to see that the propositions expressed are actually consistent, rather than contradictory. We project our own standards of taste onto those we disagree with. Such an approach is used in support of contextualism in epistemology.¹² Semantic blindness can be used to explain disagreements over knowledge claims. For example:

• (10) John: I know I have hands

  Mary: You don’t know that!

Let epistemic standards for John be appropriate to an ordinary conversation, where the standard of knowledge is low. Take Mary’s epistemic standard to be set high, such as is appropriate to an epistemology seminar. In this case, there would be no disagreement as both speakers are merely taking different standards into consideration, thus they are not talking about the same thing. This is illustrated below:

• (11) John: I know I have hands according to a low epistemic standard.

  Mary: You don’t know that according to a high epistemic standard.

In cases like these, there is no genuine disagreement. The speakers merely get the feeling of disagreement as both John and Mary are blind to the fact that two different standards of knowledge are at play. Once we draw their attention to the low/high epistemic standard divide, they will agree that their respective uses of ‘know’ have different meanings due to differing standards of knowledge. After this, they would, or should, retreat from the claim that there is a disagreement between them (or at least engage in a discussion under the same epistemic standard).

An example used by C&H (2009, pp. 117–118) to convey this is one of an Arizonian hearing a Bostonian proclaim that it’s hot:

• (12) Bostonian: This is hot!

  Arizonian: This is not hot!

The Arizonian’s standard for hotness is very different to that of the Bostonian’s, however, “this is not recognized by the Arizonian’s language faculty and, owing to misjudgements about semantic uniformity, some disagreement judgements are accepted when they ought not to be” (ibid., p. 118). Thus, because the standards for hotness are vastly different, yet this is not recognised by the Arizonian, she accepts this as a case of a genuine disagreement when in fact she is mistaken in doing so. Should the Arizonian realise there are two wildly different standards at play, she would not think that she is disagreeing in (12).

Similarly, when John and Mary appear to be disagreeing from autocentric perspectives they are mistakenly attributing their own standards of taste to the other speaker. If this were to be reflected on, the semantic blindness would disappear. Both John and Mary would agree that there is no genuine disagreement between them. Thus, faultless disagreement is not a genuine phenomenon, rather we are simply mistaken about our intuitions.

5 Empirical Data

In this section, I present empirical data to show that it is implausible for IC accounts to simply reject faultless disagreement. Ordinary speakers readily accept such a phenomenon. If IC is to provide an empirically adequate semantic for PPTs, it needs to respect the intuitions of ordinary speakers. I consider two experiments and then go on to show that the semantic blindness response is not strong enough to cast doubt on the correctness of the faultless disagreement intuition. This section will conclude that IC must provide an explanation for faultless disagreement and since it is not able to, it ought to be rejected.

I will first consider an experiment carried out by Solt (2018), which suggests that ordinary speakers recognise the difference between faultless disagreements and disagreements where one speaker is ‘at fault’. Solt tested an idea presented in Kennedy (2013), where he discusses two types of subjectivity. Orthodox gradable adjectives (GAs) like ‘tall’, ‘rich’, etc., can be subjective as the standard changes from context to context. However, there is a crucial difference in subjectivity between these GAs and PPTs, which is illustrated in comparative form constructions:

• (13) Mary: Falafel wraps are tastier than burgers.

  John: No, burgers are tastier than falafels.

• (14) Mary: Frank is taller than Freddie

  John: No, Frank is not taller than Freddie.

When we put PPTs in the comparative form, faultless disagreement is still present; it seems that both Mary and John can have uttered something true in (13). After all, there are no objective facts to tell us whether falafel wraps are tastier than burgers. It seems that the type of subjectivity exhibited by PPTs stays even if we switch paradigm disagreements from positive form to comparative form. In (14), we see a disagreement, but this is not faultless. Either Mary or John will be ‘at fault’, for there is a fact whether Frank is taller than Freddie. Thus, the kind of subjectivity exhibited by GAs like ‘tall’ is different from the kind of subjectivity present in PPTs like ‘tasty’.

Solt (2018) looked at 35 different types of GAs (including our paradigm example ‘fun’) in which the following dialogues were given to the participants:

(15)	A: John and Fred look similar but John is taller than Fred.
	B: No, Fred is the taller one of the two.
(16)	A: The vase on the table is more beautiful than the one on the bookshelf.
	B: No, the vase on the bookshelf is more beautiful.
	(Solt 2018, p. 63)13

For each of the 35 GAs, a similar dialogue representing disagreement was given and the participants were asked to make a choice whether one of the speakers was right, or whether this is a matter of opinion. In other words, they wanted to know whether participants thought that the disagreement was about ‘fact’ or ‘opinion’.

The data demonstrates that for orthodox GAs like ‘tall’ [as in (15)], 98% of the participants said that the disagreement fell in the ‘fact’ category, whilst only 4% percent claimed that disagreements over PPTs were over fact and not opinion (ibid., p. 65). From this, we see that there is a clear disparity between judgements from the participants between faultless disagreement and disagreement where one speaker is ‘at fault’.

The significance of such experiments is that those who dismiss faultless disagreement outright, as Glanzberg does, fail to account for this evident disparity in judgments between those disagreements which are down to matters of fact and those that participants take to be faultless. Experiments such as these give us a reason to demand a story of why we have faultlessness intuitions in the first place.

Similar results are found in Foushee and Srinivasan (2017).¹⁴ However, instead of having the disagreements in comparative constructions, they were presented in their positive form. Again, the participants judged disagreements to be faultless in cases of subjective adjectives with the explanation given being that of speaker’s opinion. Interestingly, the intuition that faultless disagreement is present was found in cases of relative adjectives (such as ‘tall’ and ‘big’), however, the explanation for why participants found such disagreements to be faultless differed. Instead of citing that the disagreement came down to speaker’s opinion, the explanation was that it was down to speaker’s experience (Foushee and Srinivasan 2017, p. 382). It was also clear, that the participants did not find disagreements to be faultless in cases of absolute adjectives (such as ‘spotted’ and ‘striped’). The overwhelming explanation for why this is the case involved participants citing that it was to do with the object property itself, rather than anything to do with the speakers.¹⁵

It appears that paradigm cases of faultless disagreements are accepted by ordinary speakers. Since no direct argument against faultless disagreement is given, we can dismiss accounts like Glanzberg’s that refuse to acknowledge faultless disagreement as, at least, incomplete. As we saw, C&H do provide us with an explanation of why we have this intuition about disagreements. If they are genuine autocentric discourses, then they are not really disagreements, but we mistake them as such due to semantic blindness. However, the semantic blindness thesis does not put C&H in any better position than Glanzberg, for two reasons.

Firstly, we might accept semantic blindness in the case of epistemic standards (and maybe in the case of ‘hot’), but it is nowhere near as convincing in the case of PPTs. When it comes to tastes it is appropriate to have authority over your own tastes and project your own standard, for that is the point of predicates of personal taste. When two people disagree over tastes, they are not oblivious to the fact that the other speaker has her own tastes. We saw in the case of epistemic standards, once we make it clear to the speakers that two different standards are at play, they might just back down and accept this as true. However, this need not (and often is not) the case with PPTs. People already know they have different standards of taste. If we point out that the agents engaged in a disagreement have different standards of taste, and more precisely that they have different standards in assessing a particular object (such as The Big Dipper) they can ‘stick to their guns’ and carry on with the disagreement. This point can be illustrated with an amended example taken from Zeman (2016a, p. 102)¹⁶:

• (17) John: The Big Dipper is fun.

  Mary: So, you’re projecting your standard of taste on both of us?

  John: No, I’m not, I know how much you hate this rollercoaster!

Insisting that there is some standard which would result in one speaker being wrong, appears to beg the question against proponents of faultless disagreement. If we insist that one would back down from a disagreement over whether the Big Dipper is fun or not, we’re insisting that there is one standard in that context that is ‘correct’. That is, it will either be true or false within a particular context whether the Big Dipper is fun. But this is precisely what is at issue; we are trying to provide an explanation for judgments over taste which are subjective. That is, we are trying to provide a theory that explains how we can have a disagreement when two different standards are at play.

The second reason to doubt that the semantic blindness thesis is a good explanation for faultless disagreement (in the light of linguistic data) arises when we think of who is making the judgements about the disagreements in the experiments. We are asking participants external to the disagreements to observe the conversations and then comment on what sort of disagreement they think it is. To use semantic blindness as an excuse for the intuition of faultless disagreement fails to consider this point. We should expect the participants observing the disagreements to be semantically blind as well if semantic blindness is the cause of our mistaken belief in faultless disagreement. That is, they would have said that there is a matter of fact whether the vase is more beautiful, or whether falafels are tastier than burgers, given such a high percentage of participants accepting faultless disagreement. Thus, even if the semantic blindness thesis is correct for some cases, it still fails to explain participants’ intuitions about faultless disagreement here.

Further, it is not only the case that the participants are not subject to semantic blindness, but they also see that no speaker could be objectively correct. Participants see that when speakers are engaged in disagreements over PPTs, each speaker is ‘locked’ in their own tastes/subjectivity and this is precisely the result we want. The reason why PPTs are interesting is because of the speaker dependent subjectivity that is exhibited by them. To try and discard this would be denying the essential feature of PPTs.¹⁷

In light of the linguistic data, it seems that IC cannot simply reject faultless disagreements. If they do, they fail to explain important data arising from PPTs. The contextualists have also not provided a satisfactory argument against the intuition of faultless disagreement. We saw in the previous section, that pure IC accounts do not have the tools to account for faultless disagreement within their frameworks,

Of course, what C&H and Glanzberg reject is the intuition that apparent faultless disagreement really does boil down to disagreements arising from contradictory propositions. It is open to them to modify IC by supplementing it with an alternative account to explain faultless disagreement in a non-propositional (non-doxastic) way. In the next section, we will consider IC+ accounts which attempt to do exactly that. We will see that although such IC+ theories can account for the empirical data they still do not provide a satisfactory semantic profile for PPTs.

6 Indexical Contextualism+

The above considerations present serious worries for pure IC accounts. The main issue is that they are not in a position to reject faultless disagreement or to explain it away. Contextualists have been aware of this and have proposed different ways of dealing with the challenge from faultless disagreement. The views I will be considering in this last section I will call ‘Indexical Contextualism+’. The reason for this terminology is that the proponents of such views tend to combine an indexical semantics as discussed above with some additional ingredient.

The main strategy of IC+ accounts is to reject the rigid notion of disagreement discussed in section one, where the disagreement was explained in terms of the literal truth-conditions of the propositions contradicting one another. Instead, IC+ puts forth some extra ingredient¹⁸ where the disagreement is explained in a non-truth-conditional way arising from conflicting attitudes or desires or can be explained by some pragmatic process, yielding non-literal propositional content. IC+ approaches respect the linguistic data presented in section four, for the data only specifies that ordinary speakers accept faultless disagreement but do not specify whether the disagreement is propositional or non-propositional.

In what follows, I will consider two versions of IC+. First, I will discuss Gutzmann’s (2016) expressive account, where disagreement is explained in terms of deontic force. Second, I discuss López de Sa’s (2008, 2015) account where disagreement is explained in terms of presupposition of commonality. I will end this section with some final remarks that apply to IC/IC+ more generally.

6.1 Indexical Contextualism Plus Expressive Content

Gutzmann’s theory uses as a foundation Kaplan’s (1999) use-conditional semantics. Roughly, just like we have truth-conditions which tell us when a sentence is true or false, we also have use-conditions which tell us when it is appropriate/inappropriate to use a certain piece of language. Most notably, a use-conditional semantics can be applied to expressives. Conditions can be given for when one is using an expressive felicitously. For example:

• (18) That damn Kaplan got promoted.

  (Kaplan 1999, p. 5)

We can say that (18) is true if at the actual world Kaplan got promoted, whilst the utterance is felicitous if the speaker/judge of the context dislikes Kaplan. Should I have positive attitudes toward Kaplan, yet utter (18), the truth-conditions will be the same, but my utterance is infelicitous as the context I am in is not a context in which it is appropriate to use ‘damn’. Gutzmann (2016, p. 38) makes use of the judge parameter in a similar way to Lasersohn. However, an important difference is that the judge is placed in the context of utterance rather than the circumstance of evaluation. The judge will affect the content of the proposition rather than the truth-value of the proposition.^19,20

Because the expressions in examples like (18) involve both truth-conditional and use-conditional meanings, they are ‘hybrid’ expressions (Gutzmann 2016, p. 37). Gutzmann builds on this insight to account for PPTs which he also considers to be hybrid expressions. Gutzmann attempts to account for faultless disagreement in the following way:

While the faultlessness is established by the subjective truth-conditional content of a PPT-utterance, the disagreement happens on a use-conditional layer that, for PPTs, is normative and hence up to dispute. (Gutzmann 2016, p. 35)

Thus, in paradigm examples, faultlessness is accounted for by the fact that both speakers have uttered something true:

• (19) John: The Big Dipper is fun [for John].

  Mary: No, the Big Dipper is not fun [for Mary]

Since two compatible propositions are expressed, disagreement is not accounted for on the truth-conditional level. Instead, the use-conditional dimension accounts for disagreement, where such disagreement revolves around what should or should not count as fun in the context of utterance. We could phrase the disagreement in the following way²¹:

• (20) John: The Big Dipper should count as fun in this context of utterance.

  Mary: No, the Big Dipper should not count as fun in this context of utterance.²²

In sum, faultlessness is retained as the propositions expressed by the speakers are true and the disagreement is explained in expressive terms through deontic force: the disagreement is over whether the Big Dipper should count as fun in the context of utterance.

6.2 The Problem of Projection

The first criticism I raise will cast doubt on PPT-sentences being hybrid expressive expressions, as claimed by Gutzmann, because PPTs behave very differently to ordinary expressives with respect to projection. For our purposes, projection is the phenomenon forcing the expressive to apply to the judge of the context. Under certain circumstances projection fails. One such circumstance is if the content is embedded under an attitude verb,²³ for example, ‘believe’:

• (21) John believes that the Big Dipper is fun.

The PPT ‘fun’ fails to project onto the speaker of (21), but rather applies to John. In other words, ‘believe’ works as an operator shifting the content of the PPT away from the speaker and onto John. Thus, the deontic content that ‘The Big Dipper should count as fun’ applies to John.

This is the key difference between PPT content and ordinary expressive content. The default position of expressives is that they project. This can be clearly illustrated if we consider an example where ‘believes’ does not affect projection:

• (22) Mary believes that damn Kaplan got promoted.²⁴

Here, the expressive ‘damn’, projects not onto Mary, but the speaker of the utterance. Thus, we know that it is the speaker and not Mary who has a negative attitude towards Kaplan. Unlike with (21), we see that expressive content projects past the attitudinal verb ‘believe’.

A similar problem, which is noted by Gutzmann (2016, p. 42), is the difference in how expressives and PPTs behave under negation. When negated, PPTs lose their positive deontic force. For example, Mary’s utterance ‘The Big Dipper is not fun’, expresses truth-conditional content that ‘The Big Dipper is not fun for Mary’ as well as use conditional content that ‘The Big Dipper should not count as fun’.

We see a very different result when we consider expressives. When negated, only truth-conditional content is affected, expressive content appears to go untouched, for example:

• (23) That damn Kaplan didn’t get promoted

In (23), negation only operates on the proposition that ‘Kaplan got promoted’ and not the expressive content. The expressive content that the speaker has a negative attitude towards Kaplan is not affected by negation.

Gutzmann addresses this problem. He suggests that we need to directly link the use-conditional component of PPTs to the truth-conditional component. This is justified as follows: “the deontic force of a PPT depends on the assertion of truth-conditional content. When you do not assert that, say tofu is tasty for you, you neither make a deontic claim that it shall count as tasty” (Gutzmann 2016, p. 43). This results in a biconditional between the truth-conditional content and use-conditional content as shown below:

• (24) John: The Big Dipper is fun = The Big Dipper is fun for John iff the Big Dipper shall count as fun in the context of utterance.

This gives us use-conditions for PPTs: “one shall only assert the subjective truth-conditional content of a PPT-sentence, iff one believes that the taste judgment shall objectively hold in the utterance context” (Ibid.).

This seems to explain why the negation affects both truth-conditional and use-conditional layers of meaning. If use-conditional meaning is tied so closely with the truth-conditional content, then of course negation of truth-conditional content entails negation of the use-conditional content.

This also appears to explain the issue of projection under embedding readings as in examples (21) and (22). The expressive content ‘The Big Dipper should count as fun in the context of utterance’ applies to John because John is being taken as the judge (made explicit by ‘John believes’). Thus, the truth-conditional content expresses that ‘The Big Dipper is fun for John’, whereas the use-conditional content expresses that ‘The Big Dipper should count as fun’.

This explanation still fails to account for the difference between ordinary expressive content and the expressive content that is produced by PPTs. As we saw above, the deontic force cannot be conveyed without assertion of the truth-conditional meaning. This is true, however, for those ordinary expressive sentences which Gutzmann calls hybrid expressions. For example, the expressive content of (25) and (26) cannot be expressed without assertion:

• (25) That damn Kaplan got promoted

• (26) That damn Kaplan didn’t get promoted

What we see in these examples is that the negative attitude of the speaker towards Kaplan remains the same regardless of whether the truth-conditional content is negated.

Now, one might argue that ‘damn’ and ‘fun’ are different in the sense that ‘damn’ can be used on its own as a pure expressive to denote negative attitude, whereas it’s not so obviously the case with ‘fun’. Regardless of this, what expressive content one expresses in ‘That damn Kaplan got promoted’, is not merely some negative attitude, but a very specific negative attitude, negative attitude towards Kaplan, which could not have been expressed if the utterance was not asserted. In other words, just like the biconditional in (24) connects the truth-conditional meaning with use-conditional meaning for PPT-sentences, the same can be said about hybrid sentences involving ordinary expressives. Given these considerations, it’s apparent that Gutzmann’s account lacks sufficient explanation of the difference between ordinary expressive sentences and PPT-sentences.

The difference between deontic expressives and ordinary expressives can also be seen in cases of projection under the attitude verb ‘believe’. Cases where expressives appear to fail to project onto the speaker and instead express the attitude of an external judge are rare. A widely cited example of this can be seen in Kratzer (1999, p. 6):

• (27) My father screamed that he would never allow me to marry that bastard Webster

Here ‘bastard’ expresses the negative attitude towards Webster not of the speaker, but her father. These cases, however, are rare with ordinary expressive sentences, where speaker-oriented readings are the norm. As we have already seen, just the opposite is true of PPT-sentences. Let’s consider the example again:

• (28) John believes that the Big Dipper is fun.

Here the supposed expressive content ‘the Big Dipper shall be considered as fun’, does not project onto the speaker, but rather onto the exocentric judge—John. This appears to be the norm with PPT-sentences, which is contrary to what projection in ordinary expressive content predicts.

These examples do not conclusively show that PPTs lack any expressive ingredient in their meaning, but rather highlight that there is a big difference between expressives and PPTs which must be explained. They are enough to shift the burden of proof to Gutzmann who should precisely spell out how this expressive element is supposed to function and why PPTs differ so much from ordinary expressives under projection.

6.3 Deontic Force and PPTs

The second criticism focuses on the relationship between deontic force and PPTs. One way to attack the deontic force in relation to PPTs is to show that the biconditional as proposed by Gutzmann does not hold. To do this we could show that the truth-conditional and use-conditional contents can come apart. That is, one can reject the truth-conditions yet retain the deontic force by accepting use-conditions. This idea has been proposed by Zouhar (2019) who presents the following case (where ‘Choco’ is a name for a cake):

• (29) A: Choco is tasty

  B: You are not speaking truly, because you obviously dislike the cake. Anyway, people should definitely consider it tasty. In fact, I find it delicious.

  (Zouhar 2019, p. 15)

Here the first part of B’s utterance—You are not speaking truly, because you obviously dislike the cake—is negating A’s proposition, but it is not supposed to negate the deontic force, as B goes on to say that people ought to find Choco tasty and B herself finds it delicious. Thus, even though B is rejecting A’s truth-conditional content, she is retaining the use-conditional content and thus the deontic force.

I find this example rather convincing. However, I believe that there is room for Gutzmann to respond. He may say that what B is doing by uttering that first part of the sentence is not only negating the proposition expressed by A, but also pointing out that A has spoken infelicitously. Recall the biconditional relation between truth-conditions and use-conditions as noted in the section above: “one shall only assert the subjective truth-conditional content of a PPT-sentence, iff one believes that the taste judgment shall objectively hold in the utterance context” (Gutzmann 2016, p. 43). By challenging A’s utterance, B is not only saying that “Choco is tasty for A” is false, but (precisely because the proposition is false), B is also saying that A has used the word ‘tasty’ inappropriately. This is different from rejecting the deontic force in itself, as would be the case if B uttered ‘You are not speaking truly, Choco is not tasty’.

The deontic force of the overall sentence uttered by B is retained and is felicitous, but that’s only because B herself goes on to say that people should consider Choco to be tasty and, in fact, she finds it delicious. Simply put, B is using the PPTs ‘tasty’ and ‘delicious’ in accordance with use-conditions, but this is separate from negating A’s proposition.

This does not decisively show that truth-conditions and use-conditions are connected in a biconditional as suggested by Gutzmann, but rather, that more work needs to be done to show that they need to be kept separate. Instead of undertaking this task, I suggest showing that the deontic force itself is simply inappropriate for PPTs. Zouhar does have a suggestion on this in a footnote:

• (30) A: Choco is tasty.

  B: Yeah, you’re right. I agree that people should like its flavor.

  (Zouhar 2019, p. 16, fn, 12)

B’s response seems strange and this suggests that the normative aspect (as carried by deontic force—i.e. that Choco should count as tasty in this context of utterance) is simply not present in A’s utterance and is not up for debate. Since it’s not up for debate there is nothing one can agree or disagree with (in respect to the normative element) and that’s why it seems inappropriate for B to utter ‘Yes, you’re right’ and ‘I agree that people should like its flavour’.

There is a further case to be made in order to demonstrate that deontic force is not appropriate in relation to PPT-sentences. Take Mary who loves lager, but she does not think that in the context of utterance it should hold objectively that lager counts as tasty. Thus, it’s completely felicitous for Mary to utter:

• (31) (a) It’s okay if everyone disagrees with me, but lager is tasty.

  (b) Lager is tasty, I don’t care if you don’t think so.

  (c) I know you don’t like lager, and that’s okay, but this is tasty.

If Gutzmann is correct, then (31a–c) should be infelicitous. This is because of the condition that Mary is only to utter ‘lager is tasty’ if she objectively holds that lager should count as tasty in the context of utterance. This, however, is not the intuition that we get. Mary need not think it’s the case that lager should count as tasty objectively in the context of utterance in order to utter (31a–c). This is similar to Zouhar’s example (29), but now we are not trying to prove that the truth-conditions and use-conditions come apart, rather we are simply showing that the second part of the biconditional does not hold.

That the deontic force which counts as expressive content is somewhat misplaced within the semantics of PPTs is further exemplified when we contrast it with cases where deontic force does seem appropriate. Gutzmann’s proposal is inspired by expressivism (or emotivism) found in moral debates (see Stevenson (1973)). When it comes to moral claims—if we are to adopt an expressivist semantics—it seems appropriate that the claims made should objectively hold in the context of utterance. For example, if Mary tells John ‘stealing is wrong’, it is not just that Mary wants John to know that she disapproves of stealing, but she is attempting to change John’s interest or behaviour (Stevenson 1973, p. 19).²⁵ Thus, it seems that it would be infelicitous for Mary to utter the following:

• (32) (a) It’s okay if everyone disagrees with me, but stealing is wrong.

  (b) I think stealing is wrong, but I don’t care if you do it.

  (c) I know you like stealing, and that’s okay, but stealing is wrong.

It seems that the subjectivity exhibited by PPTs allows Mary to bypass the condition that the speaker holds objectively in the context of utterance that lager should count as tasty. There are certain contexts where deontic force is not applicable, nor would it seem appropriate. This is not the case with ethical terms, where utterances do seem inappropriate if the speaker genuinely thinks stealing is wrong.

The considerations above suggest that IC+ expressive content does not capture the semantic content of PPTs. It has been suggested that PPTs and expressives differ too much in kind—they exhibit very different behaviours under projection. Further, we saw that Gutzmann’s account runs into trouble as the use-condition that one ought to utter a PPT-sentence iff one believes that the taste judgment should objectively hold in the context of utterance, is inappropriate for PPTs. There can be cases where it is appropriate for the speaker not to believe that the taste judgement should hold objectively in the context of utterance.²⁶

6.4 Indexical Contextualism+ Presupposition of Commonality

A view that prima facie avoids issues raised about projection, is one proposed by López de Sa (2008, 2015).²⁷ Expanding on a suggestion from Lewis (1989, p. 128), that participants in a conversation presuppose that they share the same values, López de Sa proposes that PPTs trigger a presupposition of commonality. PPTs trigger a presupposition that it is part of the common ground that participants of the conversation are alike with respect to their tastes.²⁸ He goes on to explain why there is an intuition of a contradiction that is felt in paradigm examples. That is, in (33), John and Mary feel like they are actually contradicting one another:

• (33) John: The Big Dipper is fun.

  Mary: No, the Big Dipper is not fun.

This intuition is due to PPTs triggering a presupposition that the participants of the conversation are relevantly alike (López de Sa 2008, p. 305). In our case, the relevant alikeness would be that Mary and John share the same standard of fun. Because of this, there is a presumption of contradiction. Given presupposition of commonality, both John and Mary assume that they are alike regarding their tastes. It cannot be both true that ‘The Big Dipper is fun’ for the likes of John and ‘The Big Dipper is not fun’ for the likes of Mary, as the likes of John and Mary are the same (López de Sa 2015, p. 162). Thus, due to presumption of contradiction, we get the intuition of a contradiction.

We can account for the faultlessness part of faultless disagreement using the familiar indexical semantics:

• (34) John: The Big Dipper is fun [for the likes of me/John].

  Mary: No, The Big Dipper is not fun [for the likes of me/Mary]

Both propositions can be taken to be true; neither speaker is at fault for uttering anything false. López de Sa adopts a flexible notion of disagreement concerning conflicting attitudes, “the attitudes in question are non-doxastic, conative in nature—like desires, likings, or preferences” (López de Sa 2015, p. 158, original emphasis). The reason for these conflicting attitudes is because the presupposition of commonality has failed and has not been accepted into the common ground.

Note, this does not mean that a contradiction is actually expressed, merely that there is an appearance of a contradiction. Presupposition of commonality predicts that when we’re involved in a conversation, we presuppose that we are all alike, not that it actually is the case that we are all alike.

This account certainly performs better than Gutzmann’s when we consider the projection problem. As we saw, PPTs fail to project under the attitudinal verb ‘believe’:

• (35) John believes that the Big Dipper is fun

The same can be said about presuppositions, for example:

• (36) John believes that the dog outside is barking.

The presupposition entailed by ‘the dog outside is barking’ is ‘there is a dog outside’. In (36), the presupposition ‘there is a dog outside’ is not attributed to the speaker of the utterance but to John. In (35), if there is a presupposition of commonality it seems that this would not project onto the speaker. We would not want to claim that it is the speaker who presupposes that her tastes are like those participating in a conversation. Thus, if this account is successful, then not only does it give us an account of faultless disagreement, but it also avoids the troubles that we saw with Gutzmann’s account regarding behaviour under projection.

6.5 Presupposition Detection Test

In the discussion of Gutzmann above, we observed that the content carried by PPT-sentences is more like presuppositional content than expressive content. What is nice about López de Sa’s account is that it supports this intuition. However, it appears that PPT-sentences do not seem to carry the presupposition of commonality with them.

I will briefly look at an objection from Baker (2012), who points out that presupposition of commonality is not detected through linguistic tests for presupposition.²⁹ One way for detecting presuppositions is the “Hey, wait a minute” test (Baker 2012, pp. 117–118). When an utterance entails a presupposition x that the addressee is not aware of, or it is not part of the common ground, then it is appropriate for the addressee to utter “Hey, wait a minute. I had no idea that x”, for example:

• (37) John: Frank’s dog is barking again.

  Mary: Hey, wait a minute. I had no idea Frank owned a dog.

Here, since Mary was unaware of Frank owning a dog, her reply to John is felicitous. This contrasts with PPT-sentences. In (38) Mary’s reply seems inappropriate:

• (38) John: The Big Dipper is fun.

  Mary: Hey, wait a minute. I had no idea we were alike in respect to our fun judgments.

Another test for presupposition detection is the “and what’s more” test (Baker 2012, pp. 118–119). When a proposition y entails a presupposition x, it is inappropriate to utter “y and what’s more x”, because the conjunct x adds no new information to the utterance. For example:

• (39) John: Frank’s dog is barking again and what’s more Frank owns a dog.

Again, there is a sharp contrast between cases of typical presupposition and PPT-sentences:

• (40) John: The Big Dipper is fun and what’s more we are alike in our fun judgments.

It seems that in (40), new information is added, and the second conjunct does not become redundant, unlike in (39). In other words, the statement that John and Mary are alike in respect to their fun judgments is new information that was not already part of the common ground.³⁰

We have encountered similar issues as we did with Gutzmann’s account where PPT-sentences do not appear to be of the same kind as expressive sentences. PPT-sentences also appear to be of the wrong kind to carry presupposition of commonality.

6.6 Indexicality and Presupposition of Commonality

A further concern with IC+ presupposition of commonality approach is that there is an issue in explaining the connection between indexicality and the triggering of the presupposition. This is evident when we consider cases where presupposition of commonality is triggered (41) and cases in which it is not (42):

• (41) John: The Big Dipper is fun.

• (42) John: The Big Dipper is fun for me.

In (41) the presupposition of commonality is triggered by ‘is fun’, in (42) it is not—even though both are PPT-sentences. The question in need of answering is—what is the difference between (41) and (42)? The answer is that presupposition of commonality is not triggered in (42) as John makes it explicit that he is only talking about his standard of taste. Thus, it appears that the indexical element in (42) is preventing the presupposition of commonality from being triggered. This, however, is not a satisfactory answer. We must not forget that the indexical element is very much present in (41) according to IC+. The propositions expressed by (41) and (42) are identical, but still, it is only in (41) that the presupposition of commonality is triggered. We need to explain why the explicit indexical element prevents the presupposition from being triggered. The explicit indexical nature of (42) cannot be the answer, for then it seems that the role of indexicality becomes redundant in (41).

One suggestion might be that the indexical element is not individualistic. That is, when John utters (41), the proposition expressed is ‘The Big Dipper is fun for us/our group’. Thus, it would make sense for the presupposition of commonality to be triggered (that everyone in the group is alike in their judgment of fun). This would also solve the problem of the difference between (43) and (44), for the presupposition of commonality would trigger in both cases. Consider:

• (43) John: The Big Dipper is fun.

• (44) John: The Big Dipper is fun for us/our group.

In both cases, the presupposition of commonality is triggered and there is no longer a disparity between covert/explicit indexical elements. Furthermore, it would explain why the presupposition of commonality is not triggered in explicit individualistic indexical utterances, like (42). John is simply talking about his own tastes and not the group tastes.

This, of course, would bring back issues we had with pure IC accounts, where we no longer can explain faultless disagreement. For if John and Mary are disagreeing over the group standard then one of them will have uttered something true and the other something false—for there will be a matter of fact in that context whether the Big Dipper is fun for the group or not. Thus, perhaps it would be able to capture disagreement (in attitudinal, as well as, propositional, terms), but it would not give us faultlessness.

6.7 Final Remarks

Before concluding the paper, I wish to make some final remarks about IC. The critique above, although aimed specifically at López de Sa’s proposal, can easily be extended to IC more generally (whether pure IC or IC+). This critique does not rest on IC/IC+ being able to account for faultless disagreement, but rather there are internal issues with positing hidden or covert indexicals.

As pointed out by Kölbel (2004b),³¹ if IC was correct then the PPT-sentences with covert indexical elements and overt indexical elements would be equivalent, thus either (45) and (46) could be uttered in a given situation:

• (45) John: The Big Dipper is fun.

• (46) John: The Big Dipper is fun for me.

Given this, we should not be able to utter (45) and the negation of (46) (or vice versa) as this ought to be contradictory (Kölbel 2004b, pp. 303–304). Thus, the following ought to be infelicitous.

• (47) John: The Big Dipper is fun but it’s not fun for me.

John’s utterance in (47), however, is not only grammatical but also seems completely felicitous. We can further expand on Kölbel’s point by noting that when two speakers are engaged in a paradigm case of a disagreement, overt mention of an indexical adds new semantic content to the discourse. Take the following exchange:

• (48) (a) John: The Big Dipper is fun

  (b) Mary: No, the Big Dipper is not fun

  (c) John: Well the Big Dipper is fun for me.

If, as IC claims, John’s utterance in (48a) expresses a proposition that ‘The Big Dipper is fun for John’, then (48c) ought to be repetitive and redundant. However, just the opposite seems to be the case, John’s utterance in (48c) appears to add new information to the conversation. This is puzzling if (48a) and (48c) are supposed to express the same propositions.

A proponent of IC can respond to the arguments above by noting that (48a) can be understood as ‘The Big Dipper is fun for x’ where x can stand for: me, everyone, the average person, the group, etc. John’s utterance in (48c) is explained by John restricting his utterance to himself only or fixing the x to ‘me’.³²

This response on behalf of the IC proponent works against both examples, Kölbel’s and mine. John’s utterance in (45) could be understood as ‘The Big Dipper is fun for the group/everyone/the average person, but it is not fun for me’. This response would seem to explain why the utterance is not contradictory, as different indexicals are fixed.

To counter this, I would like to first note that in paradigm cases of disagreements over tastes we take, as a default, autocentric perspectives. This is not to deny that there can be exocentric perspectives, but these need to be clear from the context. It thus seems strange that John would need to clarify that he means himself in (48c). Furthermore, we can construct a dialogue where it is clear that John is talking about himself (and not the group/everyone/the average person, etc.), as in:

• (49) (a) John: The Big Dipper is fun, but I know most people do not like it.

  (b) Mary: No, The Big Dipper is not fun.

  (c) John: Well, The Big Dipper is fun for me.

Here in (49a), it’s clear that the indexical element is fixed to an autocentric perspective. If there was a covert element in the first conjunct of (49a), then John’s utterance in (49c) would again seem puzzling and ought to be redundant. However, (49c) seems completely felicitous.

A fruitful point about such exchanges is that we are not focusing on disagreement per se, but only on the propositional content conveyed. Thus, even if the disagreement turns out to be non-propositional, as is claimed, this criticism has a wider scope. It encompasses all of the IC accounts using simple indexical semantics (including those that I have not had the space to consider), as they need to explain why the propositions expressed in (48c) and (49c) seem to add semantic content.³³

7 Conclusion

We have seen that pure IC tends to reject the very possibility of faultless disagreement. Linguistic data suggest that such disagreements ought to be taken seriously as ordinary speakers readily accept faultless disagreements. After showing that semantic blindness is not a good enough explanation for our intuitions, especially in the light of the linguistic data, we considered IC+ accounts which seem to respect empirical findings. We considered two IC+ approaches and showed that they run into serious issues as PPT-sentences fail to behave like ordinary expressives, and they appear to fail to trigger the presupposition of commonality. Finally, we finished with some remarks about IC/IC+ more generally, noting that there seems to be a worrying issue of what the indexicality adds to PPT-sentences where the indexical element is covert. I conclude that IC/IC+ accounts do not provide a satisfactory semantics for PPTs.