However, it is useful to model the system as a particle (hole) that moves. After all, even in semiconductors you are at a macro-level in terms of electrons -- trillions of electrons moving in/out of their valance band. You're really never even talking about specific electrons or holes, but rather statistics. After all, with any current, the specific electrons that flow from one terminal might not reach the other terminal. There isn't any charge accumulation, so statistically the same number of electrons flow into/out of the terminals.
But it is nicer to keep this view and think of both electrons and holes as flowing particles.