There are broadly speaking two conceptions of how we perfrom speech act. One, the Austinian one, is that speech acts are purely conventional. So to promise is simly to utter the words ‘I promise…’ because there is a convention in English that says that saying ‘I promise…’ counts as making a promise. The other view is that (at least some) speech acts are performed via a kind of Gricean intention. This is the view that Strawson defended in his famous paper ‘Intention and Convention in Speech Acts’. On this view what makes something a promise is the intention that the speaker has in uttering it. In other words if I intend to be making a promise (and other conditions are met) then I count as making a promise.
Now, one thing I have noticed, is that when communicating via email it is easy to misinterpret what someone has ‘said’ (hence emoticons), which can lead to a quick escalation of tensions. How is this possible? It seems that the only way that this is possible is if the Strawsonian conception of speech acts is right. If performing a speech act is a purely convention act then there should be no question of whether a certain token, say of ‘I promise…’, is a promise, or a threat, or a guarentee, or what.
One may think, ‘ah, but there are different conventions governing that sentence type’ then one still needs to know what convention the ‘speaker’ intends to be conforming to. Either way the purely conventional nature of speech acts is brought into question.
Richard Brown 
That seems a bit quick. It’s not as though speaking in person gives us direct access to the other person’s intentions. Rather, we track a variety of clues including tonal intonation, body language, broader conversational context, etc. Let X be the complex (disjunction of conjunctions) of clues that provide good evidence that a speaker has an intention to promise. Couldn’t the conventionalist simply say that X is the convention which amounts to promising? What matters is the actual behaviour, not the hidden intentions that may or may not be behind it.
The problem with email, on this picture, is simply that the conventions are not yet sufficiently worked out. You’re right that we have trouble discerning the writer’s actual intentions. But here’s another thing we have equal trouble discerning: whether the writer’s actual behaviour is evidence of certain intentions (regardless of whether they actually have the intentions or not).
So the difficulty we have in determining whether a promise has been made over email is not evidence that actual intentions matter for promising.
What if I promise: Never to promise?
Hi Richard,
I may be missing what your point is, but as I understand the conventionalist’s position, the intentions of the speaker DO NOT matter AT ALL. So, we don’t use behavior to infer the speakers intentions at all on this view. So, as an extreme example, take the Malasian convention that a man can divorce his wife simply by saying ‘I divorce you’ three times; in fact this can be done by text message! There was a case a few years back of a couple where the husband said ‘I divorce you’ three times IN HIS SLEEP. When he awoke he disavowed any intention of wanting to divorce his wife, but the council of elders held that the convention had been met, and that the intention of the man was irrelevant. So, they were legally divorced. So, it seems to me that the admission that actual behavior is evidence for intentiosn is already to give up the game to the intentionalist…
Hi Philo,
Thanks for the comment.
I suppose that your example is an example of what is called ‘Moore’s paradox’ (his version was ‘p but I don’t believe p’)…this is in general an example of what we might call a ‘speech act paradox’…but I don’t see that it has any bearing on the issue here…
Right, it’s also the case that the speaker’s intentions don’t matter at all on the view I described. Rather, what matters is that behaviour X is performed. (And it just so happens that X is typically good evidence of certain intentions. But it doesn’t matter: one might do X in their sleep, and it would still count even though completely unintentional in this case.)
OK, but the behavior that the conventionalist normally has in mind is just an utterance, so then I take it your suggestion is to amplify thie notion of the behavior that matters? This would mean that the relevant convention isn’t a linguistic convention; but that seems odd!
Right, but it’s not odd. If one “promises” something in a really sarcastic tone of voice, and blatantly waving one’s crossed fingers about, then it doesn’t seem to matter what words you utter: the conventions for promising are clearly not being followed here.
Yeah Richard, you may be right that this was a bit quick…I’ll have to give it some more thought. Thanks for the comments!