J.L. Austin insisted that utterances of performative sentences are neither true nor false. If I say, “I promise to dine with you” my utterance has no truth-value. Presumably this implies that it expresses no proposition (though it is clearly meaningful), since if it did it would have to be either true or false. It is not, as Austin puts it, a constative. Performative sentences thus belong with interrogatives and imperatives, despite their declarative grammar. Some people have contended against Austin’s position, claiming that such utterances do have truth-value, being generally true. They say that the utterance is true precisely in case the speaker is making a promise: if I say, “I promise to dine with you” this utterance will be true if and only if I (thereby) promise to dine with you. After all, “You promised to dine with me” will be truly uttered by you in the circumstance that I made the utterance in question. Thus, it is contended, performatives are constatives and do express truth-evaluable propositions; no special category needs to be created for them. Who is right?
There seems to be something correct in both positions. Let us assemble more data, which Austin would have approved. If performatives can be true we ought to be able to prefix them with “It’s true that”: so can I say, “It’s true that I promise to dine with you”? That sounds distinctly odd and not equivalent to the embedded performative. I don’t promise to dine with you by uttering such a sentence. Thus we have a breakdown of the usual equivalence of “p” and “It’s true that p”. I doubt that anyone has ever uttered such a sentence with the intention of making a promise or for any other reason. Compare “It’s true that I name this ship Bertha” and “It’s true that I hereby make you man and wife”. These may not be nonsense but they are close to it. They violate some sort of linguistic rule. But there are sentences in the vicinity that suffer no such defect and which cloud the issue. Thus there is nothing wrong with the following: “It’s true that I promised to dine with you”, “It’s true that I will promise to dine with you”, “It’s true that I ought to promise to dine with you”, and “It’s true that in saying ‘I promise to dine with you’ I thereby promise to dine with you”. We might even tolerate “It’s true that I am promising to dine with you”. And of course there is nothing amiss with “It’s true that you promised to dine with me” or even “It’s true that you are promising to dine with me” (uttered while I am mid speech act). The only one of these sentences that raises hackles is the present tense performative case: this is the one that I cannot prefix with “It’s true that”. Thus it can be true that I promised but I can’t build this locution into my promising explicitly—I can’t say, “It’s true that I promise to dine with you”. It is as if the truth cannot be said but only shown. This is very different from the case of ordinary assertion where I can happily add the truth prefix. It’s puzzling.
Austin focused on the question of truth and performatives, but actually the issue arises more broadly. Consider “I know that I promise to dine with you”: this too has an odd ring, in contrast to “I know that I promised to dine with you”, along with the future tense and deontic variants (“I know that I ought to promise to dine with you”). The performative stands out as uniquely resistant to the epistemic prefix. And yet isn’t it true that as I make a promise I know I am promising? You can certainly say, “He knew he was making a promise” or even “He knows he is making a promise”, but I can’t say, “I know I promise”; and similarly for performatives of naming and marrying. So it is not just “true” that interacts oddly with performatives; “know” does too. Austin might respond to this by saying that performative utterances are neither known nor unknown (by the speaker)—they are not candidates for knowledge. Others may retort that many sentences containing the relevant verbs that do admit of “know” (e.g. “He knows full well that he promised to dine with me”). It is only the performative use of the verb that rejects the epistemic prefix.
What about other types of embedding? Consider negation: “It’s not the case that I promise to dine with you”. Again, this is a very odd sentence—is it some kind of negative performative used to decline to make a promise? You might say to me, “Promise to dine with me” and I might reply, “I won’t promise to dine with you”, but I won’t reply, “It’s not the case that I promise to dine with you”. What does that even mean? It doesn’t mean the same as “I promise not to dine with you” which has the look of a regular performative—I have made a promise by uttering it. And yet I can obviously fail to make promises. Nor is there is anything amiss with “It’s not the case that I ought to promise to dine with you”. Again, it is the performative case alone that declines negation. Imagine if instead of saying “Thank you for carrying my bag” I say, “I don’t thank you for carrying my bag”. Is this an attempt at a negative performative or just an inept way of saying “I’m not grateful for your carrying my bag”?
It doesn’t end there, for consider: “Necessarily I promise to dine with you”. A linguistic monster indeed—what could it possibly mean? We can insert necessity all over the place with these verbs, but not there: I can say “Necessarily I will promise to dine with you” to express my belief in fatalism, or “Necessarily I ought to promise to dine with you” to express my deep moral convictions; but necessitating the performative itself would be a bizarre move in the language game. The same is true for “Possibly” or “It is contingent that”: we can’t put these in front of the performative either, but we can for other uses of the same verb. This is all grist to Austin’s mill because it confirms his doctrine that performative utterances are not statements at all but performances. If they were statements they could be true, could be known, could be negated, and could be necessitated; but instead they are acts performed by uttering words—acts of promising, naming, marrying, and thanking (and not acts of stating or asserting). If I promise to dine with you, I have performed an act like shaking your hand, and such acts are not true or false. If I could promise by some method other than saying “I promise”, then there would be no temptation to suppose that promising is a kind of stating, since it need not be linguistic at all. Promising, greeting, thanking, marrying, and so on are not inherently linguistic acts—they could in principle be performed non-linguistically. We can talk about these acts and thereby speak truly or falsely, but the acts themselves aren’t true or false—though, as Austin reminds us, they can be performed more or less felicitously.
So is it just wrong to suggest that performatives have truth-value? True, I can’t sensibly say, “It’s true that I promise to dine with you”, but does it follow that my speech act can’t be assigned a truth-value? When a person fails to name a ship by performing the ceremony, because he lacks the authority to name ships, isn’t it false that he named a ship? Can’t we say that his utterance “I name this ship Bertha” expressed a falsehood, since he failed to name the ship Bertha? That sounds reasonable enough, inescapable even, but we can’t convert this into permission to prefix performatives with the truth operator. So there is still something odd about performatives: even if they can be assigned truth-value, they differ from ordinary statements or constatives in that we can’t bring them within the scope of “it’s true that”. In fact, they also differ from non-indicative sentences in that these sentences really can’t be assigned truth-value (they don’t even look like statements). We can’t say, “It’s true that shut the door”, but we also can’t assign the truth-value True to “Shut the door” (when it has been shut). Imperatives shun truth altogether, while performatives tolerate it within limits. So performatives really do belong in a linguistic class of their own–puzzlingly so. Constatives are true or false and accept the truth operator; imperatives are not true or false and resist the truth operator; but performatives can be true or false while rejecting the truth operator (and other operators). A performative utterance is a statement-like speech act without being a genuine statement, so it has an ambivalent relationship to the concept of truth. What Austin really discovered is that the dichotomy between statements and non-statements is too simple: for some utterances are a bit like statements and a bit not like them. We shouldn’t operate with a dualism of the declarative and the non-declarative speech act, because performatives are genuine hybrids; they are neither one thing nor the other. They are a special class of sentences, but with affinities to other classes of sentence. Austin was basically right in his dispute with the levelers, but he exaggerated the distinctness of the performative utterance. Ironically, he was too wedded to a dichotomy in types of speech act. We need a trichotomy: declarative, non-declarative, and performative.