Are questions requests?

Until I came up with this argument, I used to think questions were just requests for answers. But I now think this is harder to defend than I thought. Take the question:

1. What naturally stripy equine is not a zebra?

What request is made? A naive suggestion is:

2. Please name a naturally stripy equine that is not a zebra.

But unless the speaker invokes relevant authority, it is intrinsically morally permissible to do something other than what is requested. So, if (1) is just the request (2), it should be intrinsically morally permissible to say "A donkey" or to do a hundred push-ups, just as it would be intrinsically permissible to say "A donkey" or to do a hundred push-ups when requested non-authoritatively to climb a wall. Of course, rules of etiquette require that when a reasonable request is made, a failure to fulfill the request be apologized for ("Sorry, I'll do push-ups instead" or "Sorry, I'll utter 'A donkey' instead.") But this is an easily defeasible duty, clearly more easily defeasible than the duty not to lie (which I think is not defeasible, though that's more controversial).
In fact, one can argue even further that (2) isn't actually, though it seems to be, a request for the naming of a naturally stripy non-zebra equine. For the duty not to lie applies to answers to it, too. Thus, rather than (2) being a more perspicuous way to say (1), it is (1) that is the more perspicuous.
Perhaps, instead, the request is this:

3. Please assert a proposition stating what a naturally stripy equine that is not a zebra is.

And this creates a context in which the correct answer "An okapi" is an abridgment of the sentence "An okapi is a naturally stripy equine that is not a zebra", while "A donkey" is the lie "A donkey is a naturally stripy equine that is not a zebra." Of course, a non-authoritative request can be ignored. I could do push-ups instead of answering. But if I say something like "A donkey" or "An okapi", the context makes that be my assertion (just as when I am not standing on a stage, and am conversing with friends, the context makes my utterance of "Donkeys are equines" be my assertion), so the duty to avoid false assertions prohibits me from saying "A donkey", but allows me to say "An okapi".
I think (3) is a tenable move for the request account of questions. But the account may lead to something controversial. Suppose you incorrectly but justifiably believe that okapis are the only naturally stripy equines. Moreover, you don't know what the word "zebra" means. You are asked (1). You think to yourself: "The question presupposes that there is an answer to it. There is only one kind of naturally stripy equine—it's an okapi. So, whatever the word 'zebra' might mean, given the presupposition, 'An okapi' must be the correct answer." And so you say: "An okapi." But on account (3) of the answer, you are expressing the proposition <An okapi is a naturally stripy equine that is not a zebra>. But how can you express that proposition when you don't possess the concept of a zebra?
I don't think this is insuperable. Maybe it's perfectly fine to assert propositions that you don't grasp. If it is, then we have another argument that "s but I don't believe that s." But at least this is going to be controversial.
Another possibility is that we should instead take (1) to be:

4. Please assert a proposition stating what entity, considered as a kind, verifies the open formula: 'x is a stripy equine that is not a zebra'.

So when you say "An okapi", without knowing what "zebra" means, you are asserting:

5. The formula you just quoted is verified by an okapi.

It is, though, a bit awkward to take the answer to the simple question to be metalinguistic.