This is a parallel stream, because if I count with imagined pictures, then I can speak and listen to someone talking without it disturbing the process. If I do it with subvocalization, then doing other speech/language related things would disturb the counting.