These events (Air France and Colgan) made me think about the way that stall recovery has been taught over the years. During initial training in light aircraft (in the early 1970s) I was taught that the important thing during recovery from a full stall was to unstall the wings first (by pushing forward to reduce AoA) then to use power to help accelerate back to a normal speed and to reduce the height loss.
As I moved up through larger and heavier aircraft types, this technique seemed to work just as well, until I checked out on the Boeing 737 in 1987. During simulator training we concentrated on recoveries at the stick-shaker (i.e. before the full stall) and the technique here was to slam the thrust levers fully forward and let the plane fly itself out with the attitude held at 5° nose up. This seemed fair enough, given that we hadn't actually stalled.
In 1997 on the Boeing 747 however, this technique was even used to recover from a full stall. A typical exercise would have us stalled in a holding pattern at about 8,000ft with the autopilot engaged. I remember being shaken quite strongly by the buffeting for 10-20 seconds as we waited for the thrust to accelerate us out of the stall with the nose at about 10° and asking, 'why don't we just take the autopilot out and lower the nose?' The answer was that the emphasis then was on minimising the height loss.
Why the emphasis on minimising height loss? I believe that the FAA Practical Test Standards (PTS) may hold the answer. From memory, these were quite prescriptive in specifying a maximum loss of height during a demonstration of a stall recovery. If this loss of height was exceeded, the manoeuvre was failed. So, most pilots' training for the 'Check Ride', emphasised the technique for minimising loss of height (rather than getting the wings flying again first).
Three accidents helped to change things: the Turkish Airlines 737 at Amsterdam, the Air France A.330 over the Atlantic and the Colgan Air Dash 8 at Buffalo. In all cases the aircraft stalled at a reasonable altitude and never recovered before ground/sea impact. Analysis shows that the aircraft could have recovered in the height remaining if only the pilot flying had taken positive steps to reduce the AoA and unstall the wings. There was plenty of 'height to lose' for these aircraft but it's a fair bet that (for Turkish and Colgan) both pilots' training resulted in them hanging on grimly with the stick-shaker rattling away, waiting for the engines to 'fly them out'.
(In the case of Colgan, it is possible that the PF thought that there was tailplane icing, in which case pulling back on the stick was a reasonable response).
Boeing and Airbus and FlightSafety reissued their guidance on stall recovery and, guess what? The emphasis is now on reducing the AoA, even if this involves losing height! We have come full circle.