Value-based interpretationism

Abstract

In this paper I sketch a novel interpretationist account of linguistic content that has important consequences for thinking about intentionality. I solve the challenge presented by a foundational indeterminacy of reference argument to the effect that the meaning of linguistic expressions is radically indeterminate. Happily, my solution doesn’t require positing natural properties as “reference magnets”. Non-deflationist rivals to interpretationist metasemantics include various kinds of causal theories such as Fodor-style asymmetric-dependence accounts and Millikan-style teleosemantics. These accounts face their own indeterminacy challenges that have yet to be solved, as well as similar indeterminacy of reference challenges at the level of individual words. Moreover, they must contend with our apparent reference to things that we are not plausibly causally acquainted with. If indeterminacy of reference arguments can be answered and we can avoid dubious use of natural properties, an interpretationist account like my own will be the strongest contender..

Appendix A

1.1 Performative speech acts

Another way for some semantic theories to have more value for a given linguistic community concerns performative speech acts. Speech acts are what we do with our utterances. Some are characterized as performative, others conative. The latter category includes our assertive utterances that communicate information or beliefs: asserting, stating, claiming, etc. Speech acts in the performative category, on the other hand, include promising, adjourning, appointing, christening, etc. A popular test to see whether an utterance is a performative speech act is to ask whether it makes sense to confirm or deny the speaker’s utterance. For example,

• Speaker 1: I (hereby) order you to leave the room.

• Speaker 2: No, that’s not true. (Or yes, that’s true).

To be clear, this test is not meant to show that performative speech acts don’t have truth values or truth conditions. It does, however, point to an interesting feature of some performative speech acts: they appear to be self-verifying, or self-guaranteeing, in a way that ordinary assertions or statements are not.

I’ve been fairly neutral between various types of interpretationist theories while discussing value-based eligibility thus far. Now I turn to a development of the constraint best suited for a Lewis-style metasemantics that takes for granted that coarse-grained belief and desire content has been settled and maps sentences to coarse-grained possible worlds truth conditions if a certain convention of truthfulness and trust prevails in the linguistic community. With care, my case may be extendable to other settings, but it is most compelling for mental-first approaches to content. Since most philosophers favor mental-first views of intentionality, and Lewis-style versions of interpretationist metasemantics are the most popular among the varieties of interpretationism, what follows will nonetheless be of interest.

To start, a performative utterance is typically characterized as an utterance of a performative sentence, where a performative sentence is usually a present-tense indicative description of the speaker as performing an act named by the performative verb in the sentence, for example, “I hereby adjourn this meeting” and “I promise to come to the meeting”. Sentences in the interrogative or imperative mood (“Shut the door on your way out!”) are in some sense uttered performatively too, but in the literature “performative sentence” is reserved for the indicative sentences I’ve described. I’ll use these terms in the same way here and set imperatives and interrogatives aside in the following discussion, as it is in part the fact that these performative sentences are in the indicative mood that these sentences have the special features that will interest us here.

Note too that I’m following speech act theorists in characterizing performative utterances as direct performative speech acts, i.e., speech acts which fall under the type of illocutionary force associated with the grammatical mood of the sentence uttered (indicative, interrogative, or imperative), and which are not performed by way of another speech act. (An indirect speech act would be, for example, requesting someone to close a window by saying “It’s really cold in here”). Despite not meeting the first condition on direct speech acts, performative speech acts with indicative sentences are considered direct because they explicitly name the act performed, and it is widely regarded as implausible that there is a direct conative speech act, such as assertion, performed in these cases. This is why the performative “test” mentioned above is a useful heuristic. I by no means deny that there are indirect performative speech acts. But indirect speech acts can be accounted for with some account of pragmatic implicature and lack the features that will interest me here. There are some philosophers, namely Bach and Harnish (1979, 1992), who deny that there are any direct performative speech acts at all. They hold that performative speech acts are always indirect and occur by way of a conative (assertive) speech act. I’ll set this view aside and suppose that the majority opinion that there are some direct performative speech acts is correct.¹¹

Many conditions must be met before a performative utterance can successfully bring about a new social fact. For example, to successfully appoint someone Secretary of State in the U.S., you must have the right sort of political power, and you can’t be mid-monologue in a theatre performance. Uttered in the right contexts, performative sentences make new social facts obtain that did not obtain before. There are numerous examples of such facts: two people are married, various individuals are appointed, ships are christened, employees resign from their jobs, promises are made, etc. These social facts are phenomena that have better and worse explanations. The best explanation of these facts will be those that cite the fact that there was a performative utterance in the right context. Yet, as I’ll now argue, performative utterances are not possible on permuted interpretations of the language. And if this is right, we’ll have identified another way interpretations have value which can serve to increase or decrease value-based eligibility.

As before, it will be instructive to speak of the permuted semantic values for expressions. I’ll speak of “the image of x”—“φ(x)” for short—to talk about the semantic value of some expression on the permuted interpretation of the language. We know that when Obama utters “I hereby appoint you Secretary of State” sincerely, and in a suitable context to Clinton, he utters a sentence with the same truth conditions even on a permuted interpretation. But ‘I’ will not pick out Obama, but φ(I). Likewise ‘Secretary of state’ picks out φ(secretary of state), and ‘appoint’ picks out φ(appoint).

My basic argument is this. The success conditions for direct performative speech acts include (among other things) the condition that the subject of the performative verb in the uttered sentence must denote the performer of the act. I argue that if this condition is not met, putative performative utterance events will, at the very least, fail to be the performative speech acts we expect them to be. To illustrate, take Obama’s utterance that establishes Clinton’s appointment. (Let us pretend a Secretary of State appointment is not subject to congressional approval). The success conditions for this performative act include Obama’s having the authority which comes with being the elected president, among other laws and political conventions. In our example of Obama’s appointing utterance, φ(I) is not Obama but someone (or some thing) else altogether on a permuted interpretation, and this individual does not have the power to make appointments that only the elected U.S. president can make. So performative utterances typically will not succeed on a twisted interpretation if I am right about the success conditions for performatives; in performative utterances beginning with “I (hereby) X”, where X is a verb that can be used performatively, φ(I) will not pick out the performer of the performative utterance. Worse, if Obama succeeds in performing a speech act at all with his utterance on a twisted interpretation, it is φ(appointing), not appointing. Consequently, if it succeeds in establishing a social fact via performative utterance at all, it generally will not be the social fact we expect.

Why hold that the success conditions for direct performative speech acts include the condition that the subject of the performative verb in the uttered sentence must denote the performer of the act? Before I present the core argument showing why we need this condition, let me first argue that we can’t explain how a performative utterance successfully constitutes some performative speech act just by determining whether the truth conditions have been satisfied. An utterance of, for example, “I (hereby) promise to come” has truth conditions which are satisfied only if the speaker in fact successfully promises to come with her utterance. Because the utterance event constitutes the speech act of promising, we will be unable to explain how performative utterances sometimes succeed and sometimes misfire. When I utter “I hereby appoint you Secretary of State” to my cat, it misfires for reasons that include the fact that I don’t have the authority to make such appointments, and that cats can’t hold office. So, I fail to appoint him Secretary of State with my utterance, and this is why the truth conditions won’t be satisfied. But if it is the satisfaction of the truth conditional content itself that makes an utterance a successful act of promising or appointing, the explanation of successful performative acts is viciously circular: the utterance is a successful promise or appointment if the speaker in fact successfully promised (appointed) with her utterance, thus satisfying the truth conditions of the uttered sentence; and these truth conditions are satisfied only if the utterance was a successful promise or appointment. We can’t determine whether there has been a successful performative utterance by checking to see if the truth conditions are satisfied—we can’t know if they are satisfied until we know that it was a successful act of promising, appointing, etc. There must be success conditions for performative speech acts that we can distinguish from the truth conditions of the uttered sentences.

Consider the permuted interpretation of my utterance, “I hereby declare that the United States is at war”. Suppose I stand next to President Obama in the appropriate government building and at an appropriate time for making such declarations and utter the sentence. My utterance certainly misfires and does not succeed in declaring war. And compare this to a scenario where standing in that government building at the appropriate time Obama utters “Callie hereby declares that the United States is at war”. For directly referential expressions like ‘the United States’, contents don’t change from one context to another—the content is always the referent, in this case, the United States. For indexical expressions, that stable context-invariant content is some sort of primary intension or Kaplanian character such that ‘I’, for example, always refers to the speaker. Moreover, the compositional rules are the same for both utterances on a permuted interpretation. And given a suitably large base of sentences conventionally mapped to propositions by the first step of interpretationist metasemantics, those rules will ensure that the portion of the uttered sentence ‘the United States is at war’ has just the truth conditions we would expect. Our utterances will have the same content (secondary intension) in that circumstance on the permuted interpretation, though the character or primary intension of ‘Callie’ and ‘I’ will differ. Now let us ask why Obama’s utterance in this particular case fails to bring it about that war has been declared. We cannot point to the truth conditions to explain why both utterances fail in light of the argument of the previous paragraph. It is true that for both utterances the truth conditions are not satisfied, but the reason they aren’t satisfied is that the utterance misfires—i.e., the success conditions do not obtain for performative acts of declaring the U.S. is at war. So the best explanation of why Obama’s utterance fails will explain why the success conditions are not satisfied for Obama’s utterance. These other facts that do the explanatory work are plausibly the facts about Obama’s having the political power to make declarations of war that I do not have. But citing these facts when there is a permuted interpretation of the language won’t suffice. In both cases the subject of the performative verb does not refer to me. Uttered by me, φ(I) picks out some other individual, and φ(Callie), uttered by Obama, also picks out some other individual on a permuted interpretation. Whichever individual is picked out by these terms on the twisted interpretation, they are highly unlikely to have the right authority to successfully make such declarations. But this point does not yet show why the subject of the performative verb must be ascribed to the performer of the performative act in order for a direct performative speech act to succeed. To reach that conclusion, we must appreciate why, if Obama had instead said “I hereby declare that the United States is at war” (when the background political facts are in place), that the utterance would have been a successful declaration of war. The difference would merely be a switching of the subject of the main verb, so it must be the semantic value of the verb’s subject that explains why the success conditions are not satisfied if ‘Callie’ is replaced by ‘I’. This, in combination with the fact that I do not have the right authority to declare war, indicates that the subject of the performative verb in the sentence must denote the performer of the act if we are to explain why Obama’s utterance did not succeed.

This argument supports my contention that the success conditions for direct performative speech acts include the condition that the subject of the performative verb in the uttered sentence must denote the performer of the act, a condition which in turn implies that if there’s a permuted interpretation of the language, there will be no successful performative utterances.¹² So, if some social facts obtain because of a performative utterance, it must not be the case that the selected interpretation assigns the intuitively incorrect semantic values to names, pronouns, and expressions denoting speech act types. It would take us too far afield to defend the claim that the best explanations of many present social facts cite successful direct performative speech acts, but I hope it is evident that this claim is extremely plausible. Philosophers who write on the subject of performative speech acts certainly take the view for granted: Searle (1995, 2010), for example, argues that performatives are essential to the creation and continued function of all kinds of social institutions. I take the upshot of this to be that the selected interpretation in fact assigns the intuitively correct semantic values to names, pronouns, and expressions denoting speech act types, and interpretations which deviate with respect to those expressions must not have been eligible interpretations.

Let us get clear on the sort of value at stake with social facts and performatives. It seems that both creating social facts and being able to create them in a certain way may be instrumentally valuable, but not intrinsically so. Above I suggested that if candidate interpretations both do maximally well with respect to the fit constraint, they could only differ in non-instrumental value. Do we have an argument here for the contrary? I don’t think so. Although a permuted interpretation would prevent successful performative utterances, we surely would never be able to discover that there is a permuted interpretation which would prevent those utterances from being successful performative speech acts, except perhaps through philosophical investigation of the sort we are engaged in presently. Matters would go on in the same way with respect to our beliefs and behavior, linguistic or otherwise.

We might even imagine a super-permutation argument to exploit this. For example, we could maintain that every utterance is of some particular speech act type, and then extend the interpretationist account by maintaining that what is selected is a pair consisting of a semantic theory and a speech act function. The speech act function will assign speech act types to utterances. The crucial question is what the inputs to this function should be. It seems that the input cannot just be a sentence (type), e.g., ‘I promise to come’, because the same sentence can on some occasions be used performatively, and others non-performatively. For example, someone could be describing things they do on a regular basis and say something like “I call you each week. I promise to come and see you. Then I never actually come”. The speaker does not promise anything with that utterance; it’s an assertion describing a routine. Yet on other occasions that sentence might be uttered by the speaker to make a promise. No speech act function would be the intended or correct function. To circumvent this complication, we might have the input include context facts, where the context facts include the sorts of facts that tend to differ between performative and non-performative uses of a sentence. So, the input to the speech act function should be pairs of sentences and centered possible worlds with < speaker, time > centers.¹³ Although this is a rough sketch, we can see that if we do not modify the fit constraint in any way then without an eligibility constraint this setup will ensure there are many equally good candidates for selection which assign unexpected speech acts to our utterances.

If a permutation argument like this can be constructed, it’s doubtful that it is instrumental value that permuted interpretations cost us when it comes to performative speech acts. After all, everyone will nonetheless believe that the relevant (intended) speech acts succeeded, and that’s all that’s needed for these utterances to cause the small- and large-scale cooperative activity that is so beneficial for a community. So, while I think it is still appropriate to describe the situation where a permuted interpretation is selected as one where various social facts we believe to obtain do not, it is not instrumentally valuable for them to obtain. While it may seem equally clear that we do not want to claim that it’s intrinsically valuable for them to obtain either, we can claim that it’s non-instrumentally (but not intrinsically) valuable for these facts to obtain insofar as members of the linguistic community will be overall more reasons-responsive and in that sense more substantively rational, if we don’t have to interpret them as having all kinds of false beliefs about promises, appointments, and so on. The kind of non-instrumental value at stake here is just the kind of value we considered in the last few sections.

More could be said about performatives and speech acts generally in connection with interpretationism and indeterminacy arguments, but the foregoing demonstrates another way in which some interpretations will do better than others when it comes to value-based eligibility.