I like the following three sentences as illustrating the sorts of different things we ideally want a complete philosophy of language to capture:
1) “Snow is white”
Sentence (1) is the sort of sentence that has received the most attention from philosophers. To get a satisfactory account of that sentence, we’ll want something that not only assigns the sentence truth conditions, but which explains why the sentence has the truth conditions that it does. Sentences (2) and (3) don’t have truth conditions, and yet, they are unquestionably part of language. Even though sentence (2) and sentence (1) differ in terms of whether or not they have truth conditions, they seem to share a different feature: they can be used insincerely. A complete account of language would ideally assign something we could call “sincerity conditions” to those sentences, and also explain why the sentences have the sincerity conditions they do.
What I find appealing in the family of approaches to linguistic theorizing labeled “expressivism” is that, in assigning mental states to sentences, they make something like sincerity conditions the foundation of their philosophy of language. This is an interesting project because the resulting account is structured in a way that enables us to explain the truth-conditions of sentence (1) in terms of its sincerity conditions. Sentence (1) expresses the belief that snow is white (and thus, to sincerely utter sentence (1), you need to believe that snow is white), and the belief in question has truth conditions. Those truth-conditions are inherited by the sentence. Sentence (2) expresses that one is in pain (and thus, to sincerely utter sentence (2), you need to be in pain). Since being in pain does not have truth-conditions, there are no truth-conditions for the sentence to inherit. We now have a nicely packaged explanation of truth-conditions in terms of sincerity conditions, and the account unifies sentences like (1) with sentences like (2) under the umbrella of a single approach to semantic theorizing.
Of course, sentence (3) does not seem amenable to this treatment, as it cannot be used insincerely. It is true that if I say “hello” in a happy tone, I might lead you to believe I am pleased to see you, even in a case where I am not pleased to see you, and thus mislead you, but that does not make my “hello” insincere. Why can’t I be insincere in saying “hello”? It seems like this is because there is no distinctive mental state expressed by “Hello”. While “ouch” is a display of pain, and “Snow is white” is a display of belief, “hello” is not a display. Rather, in saying “hello”, I greet you. So, this throws a wrench in the view that served to unify (1) and (2) so well. Because that view can’t address (3).
So, if we want to preserve the nice explanation we had of sentences (1) and (2), we would need to embed them within a further account, which captures “hello” as well as “ouch” and “snow is white”. Such an account (in order to preserve the structure of the explanation already on the table), would need to assign something to sentences such that, from those assignments, we could reconstruct the sincerity conditions of (1) and (2).
Now, the obvious account of what goes on with (3) is that it is used to greet someone. So a natural theory is that competent speakers know that “hello” is used to greet someone. Can we embed the expressivist account of “ouch” into this framework? We’d have to say that competent speakers know that “ouch” is used to display pain. And for sentence (1), we’d need to say that competent speakers know that (1) is used to display belief that snow is white. But note that this is only a natural account of (1) if we are trying to preserve the account of (1)’s truth conditions we liked before. If we were simply to ask what speakers know about (1), it is far more natural to say they know that (1) is used to claim that snow is white. So, even though we technically can reconstruct the earlier account within this “knowledge of uses” approach, it is unclear that there is good theoretical basis for doing so.
I am left with a sort of uncertainty about how well we can informatively unify our treatments of (1), (2), and (3). One might say, “ok, but ‘hello’ is sort of an outlier. Why not just give treat things like that as sui generis linguistic behaviors?” This is where sentences like, “Go to the store” and “Is Thomas at home?” come in. If we extend our picture in the way necessary to account for “hello”, we don’t need to assign individual mental states to questions and commands. And this is good because it is hard to see what mental state would be a sincerity condition for commands and questions (though it is not impossible to make the case that such sentences have sincerity conditions).*
That is about enough rambling on this topic for now. I will probably have some follow-up posts soon.
*For what it is worth, I am inclined to think that commands cannot be insincere, and lean that way for questions, but am less confident about the latter (cases that spring to mind as insincere commands seem, to me, more accurately described as “reluctant commands”.)