Lyman, as has been discussed before, there isn't an AoA gauge available for reference on the instrument panel. Pages of discussion on that, as I am sure you recall, over the course of the BEA "reveal" process and on the competing philosophy of what to do with the precious square feet of display area available to the crew.
Someone posted this further, up, IIRC bubbers.
The captain that can hand fly comes up in the last minute and sees something he has seen before.
Full stall, full back side stick and it is too late.
I don't know if it was too late at that point or not.
What the crucial moment was that would have allowed a nose down, restore airspeed, unstall, recover via pullout at an altitude of > 1.0 feet has been calculated by some sharper folks than me. A best guess. It may have been past that point that Captain Dubois arrived, or it may have been X seconds previous.
I question whether or not he'd "seen that before" in an A330.
My rejoinder to you is due to the general point that A330 full stall practice isn't in the training regime (based on 2-3 years of discussions on this). Thus, "stall it to get various data points" was not done during development since it is not a cert requirement. (Probably).
Until the AF 447 crew became test pilots unwillingly, some of the stall and post stall characteristics of an A330 were unknown.
How do you train someone (Captain Dubois) when the info to train him with doesn't exist? Further that lack of data points, the flight sims can't be soundly programmed to give "what it does when stalled" training.
This leaves any pilot, not just Captain Dubois, lacking a chance to be in a full stall situation in a training scenario. Granted, stall prevention is the general training focus, for good and valid reasons.
To sum up: I don't think he'd "seen that before" and thus was playing catch up from the moment he entered the cockpit. Had what he saw, as you suggest, been something he recognized as a stalled A330 -- something "he'd seen before" -- my estimate is that he'd have directed Bonin to make stall recovery control inputs rather than the directions he did give him.
It is also my estimate that CVR transcripts would have included some rather forceful language, to include such
bon mots as "merde, we're stalled, get the &$^# nose down!" or words to that effect.
That's an estimate, and we can't ever know.