I think that it's much easier to simply observe how they use such a word, rather than making blanket statements. If someone says "I'm tired of seeing fucking niggers on the bus all the time!" then they're clearly being callous, racist, and insensitive. If, on the other hand, someone says "let's have an intelligent conversation about usage of the word 'nigger,'" they're clearly not; but the former statement corresponds almost exactly to "I'm tired of seeing fucking black people on the bus all the time!" and the latter to "let's have an intelligent conversation on the use of the n-word." In other words, being racist and using that word are orthogonal; the reason that blatant racists are more likely to use such a word is that they presumably have less ability to see what is socially acceptable or morally right, and that they have less restraint. In other words, racist people's higher likelihood of using such a word is purely incidental, not a direct correlation. An analogy might be getting a higher insurance premium because you live in city X and actuaries notices that city X has a higher-than-average rate of crashes. Even though city X is quite large, and you live towards the north side, but all the crashes happen on the north side because it's a poorer part of town so driving education is worse.
This already is what we do. We observe whether someone just uses the word or if they show the respect of at least referring to it as “the n-word” when there is a meta-context of discussing it.
That’s no different than what you propose, except that there’s far less risk of either incidentally triggering someone for whom the word itself materially carries a weight of trauma, and far less risk of making it seem like the word is OK to say so long as you do mental gymnastics about the context of how you said it (something which people who want to use it racially would like to do to avoid punishment).
Allowing people to scapegoat their clearly racist usage of the word, under the pretense that it’s OK as long as the intended usage is not meant to be offensive, is particularly terrible, as it shifts the blame of “being offended” onto the class of people for whom it is a racial slur. It’s tacit racism endorsement 101.
What you describe seems like it contributes nothing in the way of actually bolstering free speech or better identifying slur usage from discussion usage. The current way already serves those purposes better, and without the extra risks that your described point of view carries with it.
