From before the stall, until the moment it hit the Atlantic, the airspeed indications were correct and proper!
FGD135,
You need to look at the graph appendix in the actual report.
From ~24,000 feet on down there were only two very, very short periods where computed airspeed was not NCD. Even prior to that there were periods of NCD interspersed with inaccurate values. NCD on a round dial would have looked similar to what I posted earlier. It was an air data computer problem, not an airspeed indicator problem.
As a matter of interest, if you were to encounter icing induced unreliable airspeed in whatever you fly, what is your benchmark for categorically stating, "airspeed is now reliable again". This is somewhat problematical if you've ever encountered it, round dial or tapes, until you can unequivocally correlate it with pitch attitude and power setting or AOA if so equipped. It's easy to fly thru it, less so to be sure it's over, especially if one induces as many flight path variables as they did.