Based on my observations, I'd say it actually does alternate. My theory is that the motor speed is slow enough and consistent enough, as is the friction, that the motor consistently overshoots by half a pole when it stops. Thus, it moves in the opposite direction when turned back on. If it was moving faster, it might well be random. What's the chances of it displaying bifurcation then chaos? :-)

Given that the period of my oven is bang on 12 seconds, averaged over many revolutions, I think it's probably a synchronous motor.