The bombardier statement in post #753 describes the event sequence that I (and many of you guys) have expected. There are however a few items within it that puzzles me.
I would have expected the flaps 5 configuration earlier in the fatal speed-drop (even before). Even if they were using non-icing speeds the clean pattern speed would have been their reference and that should be about 160 kts. I definitely agree with all of you stating that a power increase could be neglected due to some distraction but I have much harder to see that happening while changing flaps configuration, twice (to flaps 5 and to flaps 15) since in my mind this more or less automatically triggers speed awareness. What do you think?
I would also have expected a different sequence of shaker/pusher and elevator input from pilot and at a slightly higher speed than 105 given that the incr. ref speed was on. At what time did the a/c performed nz=2g, my thinking is that this is in the initial pull-up (could otherwise be later just before reaching ground) and this also means that there is still air-flow to generate lift, however this would also increase drag and AoA causing the secondary stall. The point is that the shaker/pusher system with incr. ref. speed on could have triggered while there was still good control authority and the timing of the pusher in the sequence could be a design issue especially since I believe that 20 kts increment is far too much compared to the actual aerodynamic impact in most "normal" operation icing conditions.