May we explore this rotation issue further? Obviously, an unnatural yanking of the aircraft off the ground is undesirable, regardless of engine failure. And as several posts have noted here, stabilizing yaw with the assistance of visual cues is helpful before climbing into the clouds.
However, aircraft performance calculations are based on the aircraft rotating at Vr. On the CRJ, that speed is often very near or at V1. I would argue that it is nearly impossible NOT to be in the rotation as the aircraft begins the yaw towards the failed engine if we are flying the profile as specified, which requires rotating at Vr, not some arbitrary higher speed.
Sure, many airports offer plenty of margin, so being a little fast is probably not a problem in reality. But I would also argue that not rotating at the specified rotation rate at Vr is similar to deciding to discontinue a takeoff above V1; both scenarios invalidate planned data and could compromise (or eliminate) safety margins (including obstacle clearance on the climb).
So, the lessons here seem to be:
(1) Fly the profile.
(2) Rotate at Vr, but don't over-rotate or accelerate the rotation rate.
(3) Concentrate on maintaining directional control, especially while visual references are present.
Does that sound correct?