An alternative proof of Tataru’s dispersive estimates