Consider some moral claims (they might be prima facie or ultima facie--it does not matter for what I am doing here):
- It is wrong to intentionally kill innocent people.
- It is wrong to intentionally go against the terms of one's promises.
- It is wrong to intentionally appropriate things that belong to someone else.
- It is wrong to intentionally engage in sexual relations with someone one is not married to. (Some will want to say that this prohibition only applies to one if one is married, but I think it applies in general. But this won't matter)
- It is wrong to intentionally say what one believes is not true.
Here, item (5) stands out as not quite parallel to the others. In all the others, the subjective state of the agent enters in through the term "intentionally". But in (5), subjectivity enters twice, once through the "intentionally" and a second time through the "does not believe". I want to suggest that things would be neater if instead of (5), we took the basic form of the moral prohibition in question to be:
- It is wrong to intentionally say what is not true.
I think one can derive (5) from (6). If one is intentionally saying what one believes not to be true, then one is acting in a way that one believes will accomplish the intentional saying of what is not true. But it is wrong to act in a way that one believes will accomplish a forbidden thing if one succeeds. Hence, if (6) is true, one is acting wrongly in intentionally saying what one believes not to be true.
Observe that for (1)-(4) there are doubly subjectivized variants. It is wrong to kill innocent people, but it is also wrong to kill those that one believes to be innocent people (it is wrong to shoot at a deer if one mistakes it for an innocent person). And so on. But the doubly subjectivized variants are secondary, derivative from (1)-(4) and a principle about the wrongness of doing what one believes will accomplish a forbidden thing if one succeeds.
Suppose we take (6) to be the basic form of the moral prohibition, and (5) to be derivative. Then we can say that in the primary case--the case of the liar who not only says what she thinks is false, but what is actually false--what we have is an offense primarily against truth, and only secondarily against sincerity. And I think it is right to see lying as primarily opposed to the value of truth. This, I think, also fits well with the examples (especially the first) in this post.
Seeing the duty to avoid false speech as grounded in an obligation of truth also makes it plausible that we should not speak if we do not have good reason to think that what we are saying is true.