Linguistic unity, list length, and rate of presentation in serial anticipation learning