Even in Quine's understanding, I wouldn't call them trivial or meaningless, but incomplete and therefore vacuously true.
Every 'should' statement, for instance, is true for someone crazy enough to believe it: 'You should kill anyone who owns a ShiTzu' is true for someone who worships the great ShiTzu-avenging godling of New Berlin. If you develop an acute schizophrenia tomorrow, that person might be you. (He might be you now, so let me inform you that I do not own a ShiTzu.)
Such a statement can only become false when a given moral principle is attached afterwards as a premise or tacitly presumed by the speaker; say, religious relativism, or obeying the law. At the point the omitted condition is adequately grounded, the statement is rendered conditional and not modal.
This remains a position other than realism or fictionalism: since this clarification is ultimately intended, direct modal statements are statements about reality, this reality rather than some extended one, and they not fictions. (That does not mean that modal statements about fictions are not about fictions. But we make indicative statements about fictions, too.)
The same is true of the statement's inverse -- they are both true, and do not conflict. There is no problem with 'You should and you should not do X.' It simply indicates your moral premises are not clear enough yet. So the mechanism of box and diamond operators in modal logic just tracks what is and is not grounded yet, relative to a single given missing premise, and therefore cannot lead to broader deduction. This makes it a much less useful tool than it might seem.
The notion of moods captures the fact that humans never really intend to clean up after themselves in this regard. We just rate the odds of guessing the condition wrong, and continue. The mood marker (would, should, could, etc.) besides conveying this intention to remain elliptical, indicates the likely form of the condition. But it does not really have meaning in and of itself, and does not even prescribe the condition's nature entirely. One can clarify what initially appears to be a moral issue by providing facts, for instance, which displace the applicability of some given premise other than the missing one(s).
Grammatically this interpretation of modal statements as vacuously true due to missing premises is now called 'suspending control', and is one modern interpretation of the grammar of the subjunctive and related constructions. In this model, when you use subjugation with a mood like the classical English subjunctive or the old Greek 'optative', you are making the statement for which a condition exists which controls the meaning of your statement, but you are admitting that you are not certain of exactly what that condition is.
This whole framing makes modality elliptical, rather than nonsense.