So, my guess (c'mon, we need a lot of opinions and especulations to make this page go back to life) is that this pilots were very comfortable with the automatization of the airplane, and not used to land it manually, without the ILS. So, they noticed too late that they were short, low and slow, and could only try to increase power and raise the nose.
I agree. However I have a problem understanding the "mechanics" of the events.
Based on the sketchy numbers of FlightAware (you can look at
http://forums.jetphotos.net/showpost.ph ... tcount=128 for a rough analysis I made with them), they were continuously slowing down for at least the last 80 seconds and the sink rate and descent gradient were continuously reduced for the last 50 seconds or so. (This means that the approach was never stabilized so, to begin with, they should have gone around by 1000ft or 500ft depending on the company policy, but that's another story).
That is strongly indicative that neither the A/T nor the A/P were engaged, which is consistent with what you've said.
However, this opens another enigma.
For the speed to decay along 80 seconds in manual flight, it means that the pilot has to be applying incraeasing amounts of nose-up inputs and pull-up force.
They could have done it with the trim ( so replace the above for very frequent nose-up trim inputs along 80 seconds), but only to a point.
The 777, wile FBW, follows a different law than the Airbus. The 777 retains the speed stability that Airbus traded for load factor stability. In an Airbus you don't need to pull up or re-trim to compensate for a decaying speed. The FBW will do it alone (to keep the load factor at 1G) very much like the autopilot would do in any plane.
Not so in the tri7. When in normal law, the autotrim effectively works like a speed selector, actually changing the speed selected in the Autopilot. The autotrim will compensate for changes in the CG or configuration, but not for changes in speed. If the airspeed decays, the trim is kept and the nose goes down, very much like in a non-FBW plane. So it would have taken manual trim input to keep the plane slowing down. And then that would have worked only to a point.
As said, in normal law the trim effectively works as a speed slector, changing the speed selected in the AP. And there is a minimum speed you can select in the AP (and hence with the trim) which is called, go figure, Minimum Selectable Speed. This speed is not much below Vref, and has a margin over the stickshaker speed.
So the pilot had to pull on the yoke to slow down to a speed below the minimum selectable speed.
Not only that, but the 777 has envelope protection. It is not a hard pilot-proof envelope protection like in the Airbus, since it's overridable with enough effort, but it is there. It works by increasing disproportionately the pull-up force needed to increase the AoA beyond a certain point (as the plane approaches the stickshaker AoA and beyond).
So, it should have been a lot of work and effort to crash this plane in the way they did.
EDIT: I couldn't attach the graph here, but posting in in JP gave me the chance to link it as an internet image