There are two accounts of lying. If one asks a child, one will likely hear that to lie is to say something that isn't true, and the child is likely to count any statement she thinks is false as a lie. Adults do this, too, but reflective adults tend to grow out of this, and see a lie as a statement that the speaker takes to be false. On this more sophisticated view, a lie is opposed to sincerity, not truth. If one is mistaken about the truth, One can lie and say the truth, or one can speak falsely and not lie.
I think the more mature account is probably truer to the English word "lie". But the interesting question isn't about linguistics, but about ethics: The question is whether what morality in the first instance forbids (whether absolutely or prima facie—I think absolutely) is insincere speech or false speech.
I want to argue that what is forbidden in the first instance is false-telling. The obvious objection, of course, is that (a) one is innocent if one is convinced that the falsehood one tells is true, and (b) one is guilty if the truth one tells is something one believes to be false.
This objection, however, is one that the defender of the false-telling theory can easily handle. First, consider (b). If it is wrong to do A, it is wrong to do what one takes to be the doing of A. The person speaking a truth and believing it to be false takes herself to be speaking a falsehood, and hence she acts wrongly.
It is not obvious that claim (a) is actually correct. If I believe a falsehood p, but I have no right (moral or epistemic, say) to believe p, it is not clear that I am excused for affirming p. It is a merit of the false-telling theory that it draws one's attention to this. In any case, the false-telling theorist can say that when one sincerely affirms a falsehood p, and one has the right to believe p, then one is not culpable, though one has acted wrongly. One is like the person who handed out poison at the party, believing it to be water—it is wrong to hand out poison, but one is not culpable if one was not reasonably expected to know that it's poison.
The false-telling account has the advantage that it situates the paradigmatic case of lying as an offence against truth, rather than against sincerity, and that seems right. Moreover, it appropriately parallels the prohibition against lying with the prohibition against promise-breaking and similar objective moral rules. What is forbidden (at least prima facie) is to fail to do what one has promised. A person who thinks she is acting contrary to a promise is, of course, acting viciously, but she is not breaking her promise—she merely thinks she is, though in doing what she thinks is a breaking of a promise she does wrong. If one has fulfilled a promise without knowing one has done so, one is off the hook as far as the promise goes. And if one has innocently forgotten a promise, say due to amnesia, then one still has gone against the promise and acted wrongly, but one is not culpable. We would say the same thing about adultery. What is forbidden in the first instance is for a married person to engage in sexual relations with a non-spouse. It would be odd indeed to say that what is in the first instance forbidden is sexual relations with someone one takes to be a non-spouse.[note 1]
On an account on which what is in the first instance wrong about lying is insincerity, we need to posit an additional duty, the duty to ensure that what one speaks is not only sincere but sufficiently investigated by the speaker. To speak when one has made insufficient investigation of the matter on which one speaks is not a lie, but is nonetheless wrong, since in speaking one invites trust. But on the false-telling account, the duty to investigate the relevant matters follows from the general duty to make sure in all our actions that we have sufficiently investigated whether the action isn't wrong. For if what is wrong is the telling of a falsehood, then one must sufficiently investigate to ensure that one isn't unknowingly doing that, and this one does by investigating whether the matter is true or false. The false-telling account, thus, neatly fits with the idea that by speaking we put our authority behind something, inviting others to trust us—and we should only put our authority behind the truth.
I also think the false-telling account fits better with Example 1 in this older post.
On the other hand, the false-telling account of the wrongness of lying has the counterintuitive consequence that when we make a mistake, we have done wrong. Here I am willing to bite the bullet. In speaking falsely, I have said what was not to be said. If one thinks, as I do, that there is only one form of normativity, it is plausible that a moral misdeed was done here. One might try to press this harder—does the student who makes a mistake on an exam do morally wrong (assuming she studied hard enough)? Here, I could bite the bullet, but one might also question (I got this idea from one of our grad students in a different context) whether one makes assertions on exams.
A final reflection. It seems that if I am right, then personal virtue/vice here tracks not whether one has acted objectively rightly/wrongly, but whether one is subjectively praiseworthy/culpable. The person who speaks sincerely (at least if after sufficient investigation) is the honest person, and the insincere person is the dishonest one. That virtue/vice doesn't track right/wrong may seem implausible, at least to virtue ethicists. But the cases of promises and adultery suggest that this is indeed so—virtue does not seem to be lost in failing to keep a promise when you honestly think you've kept it but have innocently mistaken the terms of it, or by committing adultery with someone one believes to be one's long-lost spouse but who is in fact an impostor. (Nonetheless, there will likely be some trauma to the virtuous agent once she finds out what she did, and this might cause moral damage.) If I am right here, then virtue ethicists would be mistaken to try to derive rightness/wrongness facts from virtue facts.