I think some posters have failed to understand the value (literally and figuratively) of simulation. The Royal New Zealand Air Force conducts basic and advanced helicopter training on the AW109LUH (twin engine, glass cockpit), for pilots and loadmasters going on to fly the AW109LUH, NH90 or SH-2 Seasprite operationally. The effective use of a full motion simulator allows all the things GoodGrief mentions to be trained to a high standard (and without risk). Some 50% of all basic training is conducted synthetically. Of note, however, since none of the operational types have piston engines or manual throttles (they are all FADEC), training in collective/throttle correlation is somewhat redundant. Similarly, because of the nature of FADEC systems, none of the aircraft allows the manual use of throttle to control direction with a seized or failed tail rotor - the simulator allows appropriate recovery techniques to be trained and conducted safely and realistically. Finally, the direct operating cost of the simulator is less than the dry lease cost of an R44 in NZ. Thus the cost of teaching hovering is very reasonable (and as a measure of the efficacy of good simulation, every RNZAF helicopter pilot trainee receives some 6 hours of simulator training before their first aircraft flight and every student over the past six years has been able to start, taxi, take-off, hover and land the aircraft to a rudimentary but safe standard on their first attempt).
By its considered calculation, the RNZAF could not conduct their rotary training cheaper if the R44 was used for basic training before transitioning to the AW109 for advanced training. The R44 would only save some 15-20 hours off the AW109 course and the cost of ownership (or lease) and operation of a second fleet would be considerably more than the single type system.