The reference to the witness being sure that he saw the aircraft "tumbling end over end" and trailing sparks is an interesting one.
A tail rotor (fenestron) drive failure in a low speed, high power situation would result in the aircraft rotating (yawing) quite suddenly and rapidly, out of control. Yaw/roll coupling may also occur. We don't know exactly what might occur because this is way beyond the flight testing done by anyone, for obvious reasons.
It should also result in the pilot rapidly (possibly instinctive for an experienced pilot) lowering the collective lever to reduce the power demand on the main rotor, in an attempt to regain some sort of control. From the hover or low speed this would obviously result in a sudden and rapid descent. If this doesn't stop the yawing motion, the next step is to shut off the engines and hopefully regain a steady, autorotative state.
As Sid has posted, by "closing the throttles" of these turbine engines in the normal way will only put the engines down to an idle setting, which will disengage the drive to the main rotor. However, once the main rotor speed droops to a low figure, they will "re-engage" and therefore provide some further torque reaction, causing further aircraft rotation in yaw. The only way to prevent or stop this is to shut off the engines fully. Personally, whatever the FRC drills say, I would prefer not to arrive at the scene of a likely hard landing accident with engines running, even at idle - they are obviously a major ignition source for any spilled fuel.
The tail rotor (fenestron on an EC135) is driven by a series of jointed shafts, much like the propshaft of a rear wheel drive road vehicle. A drive shaft failure would quite likely result in some mechanical noise, possibly some sparks from metal to metal contact (depending on the exact nature of the failure) all the way to the ground, as long as the main rotors were rotating, because the two are directly geared together.
What occurred on this tragic occasion is, of course, not yet fully understood.