Parasitic intentions. A case against intentionalism

ABSTRACT

This paper presents a novel argument against intentionalism about demonstrative reference. The term ‘intentionalism’ is used to denote the view saying that the referent of a demonstrative expression is determined by the speaker’s intention. My argument focuses on ‘mismatch cases’, roughly, the cases in which the speaker’s intention determines a different object from the one which appears to be the referent in the light of contextual factors. The opponents of intentionalism claim that intentionalism yields simply incorrect reference predictions in these cases. Many intentionalism defenders claim to the contrary, i.e. that the relevant kind of intention fixes the referent in line with the contextual factors and thus the theory provides correct predictions. I will argue that this strategy is objectionable as the concept of the relevant referential intention is genuinely parasitic on non-intentional accounts of demonstrative reference.

KEYWORDS:

• Demonstratives
• context
• intentionalism
• reference
• speaker’s intention

1. Introduction

It is a popular idea among the philosophers of language that reference of demonstrative expressions (like ‘this’, ‘that’ or ‘that black cat’, etc.) should be explained in terms of the speaker’s intentions.¹ In short, the speaker refers to o by using a demonstrative if and only if the speaker intends to talk about o while using the demonstrative (and possibly some further conditions hold, depending on the view). I will call this view ‘intentionalism’. In my paper, I will argue that intentionalism faces a real challenge when it comes to ‘mismatch’ cases. These are the cases in which the speaker using a demonstrative intends to talk about a different object from the one which can be identified as the referent by a competent interpreter on the basis of some ‘external’ circumstances. The fact that there is a mismatch between the intention and the external factors may arise as a result of the speaker’s confusion or error. For example, the speaker can point with a finger to an object which, as it happens, is not the one she or he wants to talk about. These cases have been already regarded as problematic for intentionalism because the intuitive judgements support the claim that reference is determined by external factors, such as demonstrations, rather than what the speaker has in mind. It is a usual response of intentionalism defenders that we should identify another kind of intention, one which determines exactly the object that appears to be the referent in light of the context (e.g. Bach 1992a, 1992b; Speaks 2016). Importantly, the defenders argue that the second kind of intention has the reference-determining status, and so we get the correct predictions about the mismatch cases. My argument against this response is that the concept of the relevant reference-determining intention is parasitic on non-intentional accounts of demonstrative reference: it appeals (at least indirectly) to the factors which are treated on non-intentional accounts as reference-determining themselves. In the light of this, intentionalism has a genuine problem with the mismatch cases.²

The structure of my paper is as follows. In Section 2, I present the original argument from mismatch cases and the usual strategy of dealing with those cases offered by intentionalism defenders – the one which introduces ‘secondary’ intentions. Section 3 develops my criticism of secondary intentions. In Section 4, I anticipate a possible response of intentionalism defenders to my criticism and argue this response is not enough to defend the secondary-intention strategy; I conclude by giving a general remark about the role of intentions in the theory of reference.

2. Argument from mismatch cases and secondary intentions

As it was said, intentionalism is the view saying that the speaker’s intention fixes the referent of a demonstrative.³ But this view has many different variants. They differ with respect to how they characterize the relevant referential intention or what constraints they put (if any) on the intention to successfully secure a referent in a context. In the simplest form, intentionalism claims that a perceptually grounded intention is enough for a demonstrative to refer to something (cf. Kaplan 1989a). More complex versions of intentionalism introduce some further conditions. For instance, Bach (1987, 1992a, 1992b) claims that the relevant intention involves intending the speaker’s audience to identify the referent by means of thinking of it in an identifiable way. Reimer (1992), in turn, defends the ‘quasi-intentional’ view, according to which an act of demonstration, if involved, adds a crucial constraint to the effect that the speaker’s intention disambiguates the demonstration by determining a single object in its range.

The objection against intentionalism I want to elaborate on says roughly that the speaker’s intention can be undermined by non-intentional factors, e.g. a demonstration (e.g. Kaplan 1978, 239; Reimer 1991, Sec. 2; Stojnić, Stone, and Lepore 2013, 506). Kaplan gives a good example of this sort, which can be summarized as follows:

• Picture demonstration

Imagine I show my office to a friend and, at one moment, I point to a picture hanging on the wall behind me (without carefully looking in this direction) and say something like (1) or (2):

1. That is a picture of a great analytic philosopher.

2. That picture portrays a great analytic philosopher.

I am pretty sure that I am pointing to the picture of Rudolf Carnap which had occupied this place for a long time. However, suppose that someone has just replaced Carnap’s picture with one of Spiro Agnew, which has fully escaped my notice.

I intend to talk about the picture of Carnap, but it seems that I am actually talking about the one currently hanging on the wall. That is, I have said something false about the picture of Spiro Agnew. But intentionalism arguably yields the prediction that my utterance of (1)/(2) expresses a truth – at least in its most radical or pure form. Admittedly, not all theorists agree that (1)/(2) is false. For example, King (2014, 222–224) claims that the sentence intuitively lacks a truth-value, that is, is neither true nor false in the described circumstances.⁴ However, the intuition that (1)/(2) is false is more widely shared (e.g. Bach 1992a, 1992b; Michaelson 2013; Reimer 1992). In addition, recent findings in experimental philosophy (Rostworowski, Kuś, and Maćkiewicz 2022) confirm the prediction that ordinary users of language tend to evaluate the statements made in such situations as Kaplan’s example as ‘false’ – or, more generally, according to the prediction that the referent is fixed by the demonstration and not by the intention. It is worth noting that the participants of our study do not prefer the ‘can’t say’ answer, which intuitively fits King’s view saying that the sentence does not have a truth-value.

It is important to observe that mismatch cases do not have to feature demonstrations. Some other elements of the extra-linguistic context may indicate that the demonstrative should be interpreted as referring to a different object from the one the speaker has in mind (e.g. trying on ties, see Gauker’s 2008, 363). Also, there is another factor which guides to reference independently of the speaker’s intention, namely, the descriptive content of an expression. This can be illustrated by using an example with a complex demonstrative. It may happen that the nominal phrase in a demonstrative expression determines a different object in the context from the one the speaker wants to talk about; for example:

• Chess set⁵

I watch an old beautiful wooden set of chess pieces. At the moment, I am examining the figures of the king and queen. One figure is large, the other is tiny. Looking at them, I say to my friend:

1. This chess queen is tiny.

However, the queen is, in fact, large and the tiny figure is the king. I simply confuse the figures I am watching at the moment, that is, I mistakenly take the queen to be the king and vice versa.

This case also involves a referential confusion. I intend to talk about the perceived figure which I take to be the queen, which is in fact the king. But the complex demonstrative ‘this chess queen’ can be naturally understood as referring to the queen in the set of chess I am currently looking at and, if so, I have actually expressed a false proposition by saying (3). This interpretation gets support from the above-mentioned experimental study (Rostworowski, Kuś, and Maćkiewicz 2022). In the scenarios similar to Chess Set, people tend to respond that the sentence under evaluation is false; furthermore, they tend to evaluate the sentence as ‘true’ in an alternative variant of the scenario in which the object determined by the description (in Chess Set, the queen) satisfies the predicate (‘tiny’) in contrast to the object determined by the intention. At least prima facie these evaluations are inconsistent with intentionalism. According to it, the statement in Chess Set is about the chess king and thus expresses a truth (or: it would express falsehood if the king wasn’t tiny). Some proponents of intentionalism would say that the statement doesn’t express a proposition because – in the case of complex demonstratives – the object determined by the intention should also satisfy the description from the nominal.⁶ Since this condition is not met, the intention fails to determine the referent and so the statement lacks a truth-value. However, this prediction is not confirmed by our experimental investigations, either.

Of course, it is an issue whether the evidence obtained by experimental studies is really diagnostic or relevant to semantics, and whether the particular findings of Rostworowski, Kuś, and Maćkiewicz (2022) are indeed problematic for intentionalism. The intentionalism defenders have in fact different ways to resist them. But the point is that the mere intuition underlying the Argument from mismatch has an empirical basis to the extent that it is confirmed by the judgements of a wider group of language users.

In sum, it seems that the speaker’s intention can be undermined by some other factors, like a demonstration or the descriptive content of a demonstrative expression, at least. Let me call this reasoning the Argument from mismatch cases.

I will now discuss a response of intentionalism defenders to the Argument from mismatch cases. The intentionalist strategy generally entails stepping away from the most radical version of intentionalism and moving towards more complex variants. Yet, the underlying idea of intentionalism’s defense is quite simple: when using demonstratives, speakers can be assigned different sorts of referential intentions. In particular, in Picture Demonstration, the speaker has an intention to talk specifically about Carnap’s picture, but also an intention which he gains as a result of the belief that Carnap’s picture occupies the wall – namely, the intention to talk about the picture which is currently demonstrated. The content of this intention can be surely characterized in somewhat different ways (e.g. ‘I want to talk about the picture which I am pointing right now’, ‘I want to talk about the picture which is now hanging on that wall’, etc.) What matters is that this content determines – in one way or the other – the relevant object, i.e. the picture of Agnew. Let me call the indicated kind of intention ‘secondary’.⁷ In sum, apart from a ‘primary’ intention to refer to Carnap’s picture, the speaker has a secondary intention to refer to the picture currently demonstrated and it is the secondary intention which determines the referent of the demonstrative (Bach 1987, 49–53; 1992a; 1992b; King 2013, 298–301; Reimer 1992, 389–392; Radulescu 2019, 781–783; Speaks 2016, Sec. 7 and 8). Hence, intentionalism can yield the correct prediction that (1)/(2) is false.

Let me now consider Chess Set. In this example, the descriptive content of the demonstrative determines the chess queen, while the speaker intends to refer to the king. The secondary-intention strategy would be then to say that the speaker also has an intention to refer to whatever is the queen in the present set of chess pieces. And the speaker can be attributed such an intention – a defender of intentionalism would insist – since he wants to talk about the figure which he presumes to be a queen. While in the Picture case, there are presumably different ways to describe the content of the secondary intention, here we have one good candidate for the content of such an intention – namely, the one involving the description from the nominal. In sum, the secondary-intention strategy can be extended to cover the cases like Chess Set.

3. Against secondary intentions

I will now argue that the notion of ‘secondary intention’ is parasitic on non-intentional accounts of how reference is determined in the mismatch cases and thus the notion is theoretically superfluous. This criticism will be presented in Sec. 3.3. My argumentation requires two preliminary steps: the first is to observe that secondary intentions are claimed to be reference-determining because they rely on the features of the referent based on which it can be identified from the hearer’s perspective (Sec. 3.1); the second step is to argue that those features can be actually regarded as reference-determining themselves, by appealing to non-intentional theories of reference (Sec. 3.2).

3.1 The hearer-oriented character of secondary intentions

The secondary-intention strategy seems to be a promising route which defenders of intentionalism can take in order to account for mismatch cases. However, as I have already remarked, this route requires developments of the theory. We would expect that secondary intentions determine reference not only in mismatch cases, but in general. The ‘regular’ cases, of course, are less problematic since the primary intention determines the same object as the secondary intention. Still it may be problematic to identify the relevant ‘secondary’ intention in some cases. Furthermore, the strategy raises the question why it is actually the secondary and not the primary intention that has the reference-determining status. The defenders of intentionalism clearly owe us some explanation. I will now consider the question what makes secondary intentions significant from the semantic point of view.

To begin, let me say that the answer to the above question is by no means straightforward. Philosophers emphasize different features of secondary intentions that may be relevant. We can even be skeptical about whether a systematic explanation of why one intention ‘trumps’ another can be achieved. This is a conclusion obtained by Speaks (2016), for example. In particular, Speaks argues against the hypothesis that secondary intentions are the trumping intentions because they involve a descriptive (‘de dicto’) mode of object’s identification, in contrast to primary intentions. That is to say, their object is determined as whatever satisfies certain conditions (e.g. is a picture located in the direction of my demonstration). This property seems to be necessary for explaining why the referent in mismatch cases can be at all secured. In Picture Demonstration, the speaker might have never seen Agnew’s picture, yet he is able to think of and refer to it in virtue of the descriptive content his secondary intention appeals to. But the descriptive character of secondary intentions does not explain why they are reference-determining. A speaker may have a purely descriptive intention when using a demonstrative, yet this intention may be trumped by another perception-grounded intention. Speaks (2016, 334) offers an example in which a person chooses from among a large number of plastic balls one which contains $100. Suppose that the speaker says ‘that ball’, pointing to, let us say, ball #58. Unfortunately, another ball (say #113) contains the cash. In this case, the speaker intends – among other things – to refer to the ball which contains $100, whatever it is. However, the semantic value of ‘that ball’ is clearly ball #58 and not #113.⁸

The feature of secondary intentions which, according to many accounts, explains why they are the trumping intentions is that they are associated with features of the referent based on which it can be publicly recognized as such. This requirement takes different forms on different accounts which adopt the secondary-intention strategy, but the general idea seems to be common for such theories as the one of Bach, King, Michaelson, or Reimer. For example, King (2013, 301) says that the secondary intention (in his terminology, the ‘controlling’ intention) is always the one that the speaker sees as her ‘intended final means of executing the other intentions and wants his audience to be foremost recognized’ (emphasis added here and in the next citations). According to Bach (e.g. 1992a, 143), the proper reference-determining intention involves ‘intending the audience to identify the referent by means of thinking of it in a certain identifiable’ way. Michaelson (2013), in turn, requires a referential intention to satisfy certain ‘public commitments’ to successfully secure the referent (additionally provided that no other intention satisfying the relevant condition determines a different object).⁹ All these views predict that, e.g. in Picture Demonstration, the intention related to pointing trumps the intention of referring to Carnap’s picture because the former provides a condition for intersubjective reference’s identification, in contrast to the latter. In particular, the audience can figure out what the speaker is talking about based on a description like ‘a picture in the range of my demonstration’ and not a description like ‘Carnap’s picture’. A similar explanation can be given in the case of Chess Set.

In sum, many defenders of intentionalism believe that the hearer-oriented character of secondary intentions explains why these intentions are the ones that fix reference. This indeed seems justified. Provided that expressing a meaning serves communicative purposes, understanding the meaning must be based on intersubjective means, that is, the ones which any competent user of language involved in a situation of communication can appeal to. This concerns referential interpretations, in particular. Hence, if these interpretations have to rely on the speaker’s intention at any rate, this has to be an intention whose content is identifiable from the hearer’s perspective, and so associated with the factors like demonstrations, discourse topic, saliency, etc.

3.2 Non-intentional analysis of mismatch cases

While the indicated feature of secondary intentions justifies their role in semantics, it uncovers a weakness of intentionalism at the same time. An important step towards this observation is to establish that the hearer-oriented factors to which secondary intentions appeal in determining the referent of a demonstrative in a context can be regarded as reference-determining themselves. I will now elaborate on this claim.

Let me begin with Picture Demonstration. The intuitive response to the question how reference to Agnew’s picture is established says that it is determined by the demonstration. However, this response is too simple. The general idea that demonstrative reference is determined by demonstrations (cf. Kaplan 1989a) raises serious objections. Firstly, many demonstratives which successfully refer are not accompanied by any sort of demonstration. Secondly, demonstrations are too imprecise or ambiguous in order to determine a unique object (King 2014; Speaks 2017). The second objection actually applies to Picture Demonstration, strictly speaking, to the variant containing a bare demonstrative. We need to explain why ‘that’ refers to the picture rather than, for instance, its frame, the glass protecting it, the wall behind it, etc., which are also in the range of the demonstration.¹⁰ The defenders of intentionalism have a clear explanation. The speaker refers to the picture because he intends to talk about the picture. But there is another possible explanation. A reasonable and attentive audience can easily understand the used demonstrative as referring to Agnew’s picture, without having a (direct) access to the speaker’s mind. She can observe certain cues that indicate that reference has been made to a picture rather than anything else in the range of demonstration. The content of the uttered sentence – in particular, the ascribed predicate ‘is picture of … ’ – indicates to the audience that the statement concerns a picture and not its frame, etc. (cf. Wettstein 1984, 74–75; Reimer 1992, 380). This interpretation is rested on a charity assumption saying that the utterance is generally sensible, in particular, the speaker does not express a plain falsehood (like the claim that the frame of a picture is a picture itself).

One may raise a doubt at this point whether the proposed explanation is really non-intentional. For the speaker uses the predicate ‘is a picture’ exactly because he has an intention of talking about the picture; so, one may be tempted to conclude that it is actually the secondary intention which is involved in fixing the referent. I would agree that the reason why the speaker uses such a predicate is that he wants to talk about a picture, so the intention is in a sense explanatorily prior to the ‘sentence-content’ factor. However, this does not entail that the intention plays a role in semantics the intentionalism defenders assign to it, or at least it does not have to. The point is that the used predicate can be regarded as a cue guiding to reference in itself, quite independently of the fact whether someone has any intention when using it. To postulate that whenever an intention stands behind a speaker’s action (like an act of referring), the intention must enter the semantic analysis of the used expression would be an apparent overkilling. Arguably, the speaker’s demonstration in Picture Demonstration is also driven by his intention to talk about the picture; yet, this does not prevent us from considering demonstrations themselves as reference-determining factors. (As already said, demonstrations are insufficient, but that is another problem; the predicate used by the speaker is precise enough to determine a single object in the range of demonstration.) A different example: when I say ‘Carnap was a philosopher’, I (normally) have an intention to ascribe a certain feature to Carnap; but this does not mean that the notion of ‘intention’ should be exploited in explicating the meaning of the predicate ‘philosopher’ or determining its extension. Of course, these are rough analogies since the mechanism of demonstrative reference is more complex; but they illustrate the main point: the fact that an intention drives a demonstrative’s use as well as applications of some means guiding the listener to an object does not exclude a possibility that what counts at the semantic level are solely those means (as long as they provide properties based on which a referent can be identified).

It is important to emphasize that I am not persuading (as yet) that a non-intentional analysis of Picture Demonstration is more likely or superior than the secondary-intention analysis. The aim of this section is just to show, without going into a deeper discussion, that we can have an alternative to that analysis. And this goal can be achieved by pointing to existing non-intentional theories that account for the mismatch cases along the above lines. I will proceed to this now.

The proposed explanation can be accommodated by non-intentional theories which acknowledge that various contextual factors constitute referential relations of demonstratives; in particular, it fits those theories which equate the semantic referent of a demonstrative with an object that can be identified as the referent from the viewpoint of the hearer. Such accounts have been proposed in the literature; for instance, by Wettstein (1984) and Gauker (2008) (see also Mount 2008). Let me focus on Gauker’s proposal. According to him, the referent of a (bare) demonstrative is the object which ‘best and adequately’ satisfies various ‘accessibility criteria’. Examples of these criteria include: pointing, salience, prior reference, relevance, charity, but are not necessarily limited to these (2008, 364). In Gauker’s terms, Agnew’s picture in Picture Demonstration is thus the object being – all things considered – the best candidate for the referent in light of the accessibility criteria; in particular, the criteria of ‘pointing’ and ‘charity’ are decisive in this particular case.

Again, one may argue that the category of ‘charity’ likely involves some considerations about what the speaker intends to communicate. That is true to some extent. However, it does not entail that the referent indicated by the charity criterion is tacitly determined by the speaker’s intention. What a hearer takes to be an intention of the reasonable speaker is not the same as the actual intention of a particular speaker. The hearer does not have a direct access to the speaker’s intentions and must somehow reconstruct them based on some independent evidence. And when forming a hypothesis about what the speaker intends to refer to, the hearer can take only public factors into account, such as direct circumstances of the utterance (e.g. what the speaker demonstrates while using a demonstrative), topic of conversation, some further knowledge about the context and world, etc. It means that the ‘intended referent’ is in fact determined as the object that is the most plausible candidate for it in view of external factors. Thus, ‘charity’ does not in fact appeal to the speaker’s intentions, but it genuinely appeals to the external factors, based on which only some hypotheses about the speaker’s intentions may be entertained. Relatedly, we can observe that a prediction about the referent’s identity based on charity may differ from the prediction of intentionalism following the speaker’s primary intention. This exactly happens in mismatch cases; e.g. without any further information about the case, a charitable audience will take the speaker to refer to Agnew’s picture.¹¹

The non-intentional analysis can also cover Chess Set. In this example, the referent of ‘this chess queen’ can be identified as the object which (uniquely) satisfies the description in the domain of the context. It is important to emphasize that the notion of ‘context domain’ can be explained here without appealing to the speaker’s intentions. For instance, we can appeal to perceptual salience and say that the context domain includes only these chess pieces which are currently perceived by the speaker and his interlocutor. Such an explanation can be neatly accommodated by Gauker’s theory. His theory does not target complex demonstratives, but we can incorporate a restriction that the referent of a complex demonstrative ‘this/that F’ – the best candidate in the light of accessibility criteria – must be an F. (Alternatively, we can think of the description ‘F’ as constituting a further accessibility criterion as the description generally serves reference identification.) Either way, Gauker’s theory can account for reference in Chess Set. The criterion of ‘salience’ together with the description are decisive. Finally, in the variant of Picture Demonstration with a complex demonstrative (example (2)), ‘pointing’ plus the description do the job.

In sum, we can account for reference in mismatch cases in non-intentional terms, provided that we adopt the assumption that reference can be determined by a variety of (non-intentional) factors. Such an analysis can be offered by a theory like the one of Gauker. In the next section, I will argue that this analysis has a clear advantage over the one which appeals to secondary intentions.

3.3 Intentional vs non-intentional model: parasitic secondary intentions

In a nutshell, my criticism is as follows. We have two models of reference analysis in the mismatch cases: one which appeals to intentions (call it the ‘Intentional Model’), the other one appeals to combinations of non-intentional factors (call it the ‘Non-Intentional Model’). However, it would be an oversimplification to say that the Intentional Model explains reference just in terms of intentions. It needs secondary intentions, which are associated with the factors based on which an object can be identified as the referent from the hearer’s perspective; so, these are factors like a demonstration, salience, etc. But the contention that such factors contribute to an object’s identification is exactly the thesis of the Non-Intentional Model. In effect, the Intentional Model must tacitly adopt the Non-Intentional, as the concept of secondary intentions exploited by the former appeals to the mechanisms whose explanation requires the latter. Hence, the Intentional Model is parasitic on the Non-Intentional Model. Furthermore, if the Non-Intentional model is effective on its own, the Intentional Model seems to be theoretically superfluous.

It is instructive to make a schematic comparison of two models of reference in the tables (below) which list their referential predictions, as well as the reference-determining factors in each case:¹²

Non-intention model

Table

Case	Prediction	Factor
Picture Demonstration	Agnew’s picture	demonstration + charity + description / demonstration + description
Chess Set	chess queen	salience + description

Intentional model

Table

Case	Prediction	Factor
Picture Demonstration	Agnew’s picture	intention to talk about the object determined by demonstration + charity + description/demonstration + description
Chess Set	chess queen	intention to talk about the object determined by salience + description

As we can see, the Intentional Model gives exactly the same predictions as the Non-Intentional. The fact that their predictions overlap is not a coincidence but a rule. Secondary intentions on the Intentional Model appeal to the features of the referent based on which it can be identified in a way postulated by the Non-Intentional Model. The formulations in the ‘Factor’ column aim to reveal this correspondence. Let me explain it case by case. Consider first Chess Set, as it is more straightforward. The Non-intentional Model claims that the referent is determined by ‘salience’ + ‘description’ – that is, the referent is the chess queen in the present set of chess pieces. The Intentional Model claims that it is rather the intention of talking about the present chess queen that establishes reference. But this is actually intending the demonstrative to have as its value – so to speak – the object that is the value in the light of the external factors operated by the Non-Intentional Model in this case.¹³ So, the Intentional Model is parasitic on the Non-Intentional.

Let me now focus on Picture Demonstration. With regards to the variant with statement (1), the Non-Intentional Model claims that the referent of the bare demonstrative is determined in virtue of ‘demonstration’ and ‘charity’. The Intentional Model explains reference in terms of a secondary intention and, admittedly, this intention may be formulated in somewhat different ways. However, regardless of how we actually characterize the secondary intention, its content must be at least indirectly related to the external factors treated as reference-determining in themselves by the Non-Intentional Model. Firstly, the intention is directed towards a picture and this feature is one of the reference criteria, according to the Non-Intentional Model, which the hearer can recover from the predicate ascription. Secondly, the intention must somehow appeal to the speaker’s demonstration. Even if the speaker thinks of the demonstrated picture in a somewhat different way (i.e. not explicitly as a ‘demonstrated picture’, but, for example, as ‘the picture hanging on the wall behind my desk’), the alternative description serves the referent’s identification only to the extent it is accessible to the hearer and the only way of communicating it seems to rely on the demonstration. Altogether, the secondary intention employs the categories that are treated as reference-determining on the Non-Intentional Model. The situation in the variant with (2) is quite similar. The Non-Intentional Model claims that reference is determined by demonstration and description from the complex demonstrative. A secondary intention is the intention to refer to the object determined exactly by these factors.

Based on the above analysis, I conclude that the Intentional Model is genuinely parasitic on the Non-Intentional one. Schematically: whenever the Non-Intentional Model claims that x is determined as the referent of a demonstrative d by (a combination of) factors Fs, the Intentional Model claims that the referent of d is determined in virtue of the speaker’s intention to refer to the object that can be identified as the referent based on factors Fs. This entails that we actually know how such an identification proceeds in details, and this is exactly the stage where the Intentional Model requires the Non-Intentional one. The conclusion is problematic for intentionalism in two ways. First, if the Intentional Model tacitly assumes the Non-Intentional one, we can no longer claim that reference is determined by intentions; and the ‘parasitizing’ character of the former indicates that reference is essentially determined by non-intentional factors. Second, if the Non-Intentional Model can explain reference on its own – which, as I argued in Sec. 3.2, seems to be the case when it comes to the mismatch cases under discussion – intentions come out as redundant. Namely, we should prefer the simpler Non-Intentional Model over the Intentional one in the light of methodological parsimony.

Finally, I would note that my argument from redundancy also applies to those impure versions of intentionalism which acknowledge that intentions are supported by non-intentional factors in reference determination. In particular, factors of this sort can be regarded as constraints on intentions (e.g. Reimer 1992). However, in my view, those theories suffer from redundancy in two respects. Firstly, if the contents of secondary intentions already appeal to contextual factors, then why to postulate that these factors play an additional role in the form of external constraints on intentions? For example: the speaker’s intention to talk about a ‘picture in the direction of my demonstration’ provides enough means for securing the referent (namely, Agnew’s picture) in Picture Demonstration; so why should we add that Agnew’s Picture, the object already determined by the intention, satisfies the ‘external’ constraint, i.e. has been in fact demonstrated?¹⁴ Secondly, and more importantly, if the considerations from Sec. 3.2 are valid and external factors can explain reference on their own, the constraint-approach is directly vulnerable to the earlier-presented objection. A theory which incorporates external factors as constraints, plus requires intentions, is less preferable to a theory which incorporates the external factors only, because of redundancy. If the ‘constraints’ are sufficient to explain why demonstrative d refers to x (rather than to y, z, etc.) in context c, then there is no reason to additionally introduce ‘intention of referring to x’ into our explanation.

4. Concluding remarks: are intentions necessary?

Let me summarize the main point of this paper. I have considered different mismatch cases which are regarded as challenging for intentionalism and observed that this view may accommodate them by introducing secondary intentions. I have argued, however, that it is not enough for intentionalism just to introduce secondary intentions. A defender of intentionalism should foremost explain why the secondary intention, and not the primary, has the reference-determining status. Following Speaks (2016), we can say that it is not obvious what particular feature of secondary intentions is responsible for their trumping character. (If this is the case, intentionalism apparently must adopt some complex and non-deterministic mechanisms of weighing the importance of different sorts of referential intentions; so, at least, it is not a simple theory of demonstrative reference, as it appears to be.) Instead, I have suggested that the semantic relevance of secondary intentions may be explained by observing that they rely on the factors that guide to reference from the viewpoint of a public observer – in line with the concepts of secondary intentions offered in the philosophical literature. However, as I have argued in Sec. 3.3, these concepts imply a weakness of the secondary-intention strategy, as they tacitly adopt a non-intentional analysis of reference. This is a bad news for the advocates of intentionalism who adopt the secondary-intention strategy for the reason that reference is not essentially determined by intentions and the intentions are, in fact, explanatorily redundant.

A defender of intentionalism would surely disagree with my conclusion that secondary intentions are redundant. A possible route for her – anticipated by some considerations of intentionalism defenders – is to deny my assumption that non-intentional factors can explain reference on their own.¹⁵ For instance, consider the following passage from Bach (2017, 60–61):

These proposals [like the one of Gauker (2008) or Wettstein (1984)] invite the question of just how such contextual facts manage to determine reference. The worry is that they come into play because, and only because, the speaker intentionally exploits, and the addressee takes her to be exploiting, these facts. […] These contextual facts don’t play a semantic but merely an epistemic role.

In light of this passage, one may even think that the analysis of reference based on non-intentional factors actually requires to incorporate intentions, and not the other way round.

However, I think that this response is problematic in several respects. Firstly, non-intentional theorists do not appeal – even implicitly – to the speaker’s intention in their accounts of reference. For instance, neither Wettstein (1984) nor Gauker (2008) have exploited the notion of the speaker’s intention in their theories based on contextual cues or accessibility. On the other hand, the intentional theorists who adopt the secondary-intention strategy indeed appeal to the hearer’s perspective (and, consequently, non-intentional factors) in developing the notion of the relevant reference-fixing intention. For this reason, the Intentional Model should be accused of ‘parasitism’ but not the Non-Intentional. Secondly, I think that Bach’s contention is simply wrong when it comes to primary intentions: that is to say, it is not the case that a primary intention is reference-determining while the contextual factors are only evidence (which may be reliable or misleading) with respect to which object this intention is directed to. If that were the case, our intuitions about the correct referential interpretations would follow the primary intention and not the contextual factors in the mismatch cases. Finally, I think that the considered intentionalist reply is ineffective as I don’t see how the claim that contextual factors are evidential with respect to secondary intentions actually protects the secondary-intention strategy from my criticism. We can suppose that contextual factors are evidence indicating to the hearer that the speaker has an intention to talk about the object that can be identified as the referent in light of these factors. This still entails that we need to know how to determine the object that can be identified as the referent in light of contextual factors, if intentionalism has to be explanatory; and my criticism was that, at this point, the secondary-intention strategy is parasitic on non-intentional accounts. Hence, the criticism is not undermined by the claim that contextual factors have an epistemic role.

At the end, I want to make one remark. There is some truth in the claim that intentions are indispensable for reference. As a part of human activity, communication by means of language is intentional and in order to regard a sound or a sign one produces as bearing a linguistic meaning, we need to presume that the producer has an intention to use the sound or sign as a part of a language. This concerns referential terms, in particular. That is to say, having an intention to refer to something – by using a demonstrative or any other expression that may serve this purpose – is a precondition for expressing a meaning at all, and thus referring to something. However, it is important to realize that constituting a semantic precondition by the intention does not entail that the intention conventionally determines the referent of an expression. In order to realize this, we can consider expressions other than demonstratives, whose conventional meanings are much simpler. Consider, for instance, the expression ‘even natural numbers’. This string of symbols is meaningful in English, and thus denotes a class of numbers, only if someone produces it with a certain intention (roughly, an intention to use it as an expression of English and in order to talk about a class of numbers). However, we would not say that what determines that 0, 2, 4, 6, … are the denotation of ‘even natural numbers’ is the speaker’s intention; nor would we say that the notion of ‘intention’ must be incorporated into the semantic analysis of noun phrases in general. I believe that the same holds true about demonstratives, yet it is much harder to realize that intentions are dispensable because the semantics of demonstratives is much less straightforward than that of ‘even natural numbers’. To conclude, referential intentions are a precondition for referring with using demonstratives, but not necessarily an element of their semantic convention.

To sum up, I rephrase the conclusion that the secondary-intention strategy is not a satisfactory reply to the Argument from mismatch cases, as secondary intentions are parasitic on non-intentional analyzes. Thus, mismatch cases still pose a challenge to popular variants of intentionalism.

Disclosure statement

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

Additional information

Funding

This work was supported by National Science Centre in Poland under Grant 2017/27/B/HS1/00130.

Notes

1 This idea underlies very different theories of demonstrative reference, including the ones of Bach (1987), Gómez-Torrente (2019), Kaplan (1989b), King (2014), Michaelson (2013), Speaks (2016, 2017), or Stokke (2010), among others.

2 Although intentionalism is a ‘popular’ view in metasemantics, as I have noted at the beginning, it has been already challenged by some theorists (e.g., Gauker 2008, 2019; Stojnić, Stone, and Lepore 2013; Wettstein 1984; see also Lewis 2020 on gradable adjectives). The argument discussed in Sec. 2 has been proposed by Kaplan (1978) and others; however, to my knowledge, no one has responded to the secondary-intention defense by claiming that it is parasitic on non-intentional accounts. (I have briefly suggested this criticism in Rostworowski, Kuś, and Maćkiewicz 2022, 1046; see also Stojnić, Stone, and Lepore 2013, 508 for a suggestion in a similar vein).

3 For the sake of convenience, I usually stick to the term ‘reference’, though it might be better to talk about ‘semantic value’ of a demonstrative (see King 2014, ft. 3).For the sake of convenience, I usually stick to the term ‘reference’, though it might be better to talk about ‘semantic value’ of a demonstrative (see King 2014, ft. 3).

4 See also Radulescu 2019, 784–785.

5 This example is inspired by one set of scenarios used in the study of Rostworowski, Kuś, and Maćkiewicz (2022) (cf. ‘Game of Chess’ scenarios).

6 This is assumed by most semantic theories of complex demonstratives (e.g., Borg 2000; Braun 2008; for a broader discussion see Glanzberg and Siegel 2006).

7 I borrow the term ‘secondary’ from Reimer (1992). The distinction ‘primary/secondary’ intentions will be used accordingly in the analysis of other mismatch cases.

8 One may say that the trumping intention is always the one related to demonstration, but obviously this is only half of the story (at best) since there are cases without demonstrations, in which secondary intentions trump the primary (e.g., Chess Set).

9 There are of course important differences between the presented views. For instance, Bach insists that the speaker intends the audience to recognize the secondary intention itself (not just its object) while Michaelson argues against the relevance of the self-reflexive intention.

10 Theories of demonstrations can give some explanation in this respect. For instance, according to Devitt (2021, Sec. 12.3), there is a separate convention of referring to objects by demonstrations; the referential relation of a demonstration rests on a causal-perceptual link between the agent’s gesture and an object in the gestured area. However, this proposal raises two issues. First, a causal connection still seems to be insufficient to determine a unique object. Second, it is not clear how the account applies to Picture Demonstration as there is apparently no causal link between the speaker’s use of a demonstrative and Agnew’s Picture.

11 I owe the discussion in this paragraph to anonymous reviewers.

12 Let me stress that when it comes to the column ‘Factor’, these tables present the ingredients of each model in a simplified way. The fact how the indicated factors exactly contribute to reference determination has been explained in Sect. 3.2. Yet, the schematic comparisons help to realize the point I am making.

13 I am not saying that this is the way in which speakers conceptualize the contents of such intentions, as they presumably do not. Rather, it is a sort of ‘rational reconstruction’ of the content of secondary intentions.

14 Perhaps the latter requirement is relevant in extraordinary cases in which the speaker is hallucinating; in particular, the speaker in Picture Demonstration may be hallucinating that he is demonstrating a certain picture with his finger, while no such act of demonstration has occurred in reality. In such a case, a sole intention is not enough to determine the referent since there seems to be simply no referent secured in this case. So, the constraint-approach may not be fully redundant in the considered respect.

15 This kind of response on behalf of intentionalism has been suggested by anonymous reviewers.