Accelerating from a stand, the traction motors can exert a maximum power output, that can only be sustained for a short period of time, above the continuous rating for the motors. That's why class 47s could burn out their traction motors with a concerted effort at starting a train from a stand, as the maximum power output is exceeded. As Peter Sarf says above, acceleration soon drops off as speed increases, as does the maximum power output of the traction motors as the exerted power falls into line with the continuous power rating that can be sustained for an extended period of time.
It’s the ratio of the gearing that governs the maximum rail power that can be exerted in starting a train: A low gearing gives for good acceleration but a lower top speed, as the maximum rail power is a greater percentage above the continuous motor rating, making it easier to lift a train off the mark. A high gearing gives a slower acceleration, but a high top speed. A class 08 shunter for example is able to haul a forty wagon HAA rake, despite it having only 350hp at its’ disposal, as it is very low geared, with a starting tractive effort THREE TIMES that of the continuous rating it can sustain. But of course the locos are limited to just 15mph. A class 91 electric locomotive, with a 140mph top speed, has a high gearing that means it accelerates slower off the mark (with a comparative train), as the maximum power output is not much higher than the 6090hp continuous rating.
Distributed power seems to give better acceleration than a single locomotive with the same power output, as distributed power spread over a number of traction motors on more axles gives a bigger tractive effort at the railhead.
Bascially if you plot power output against acceleration over time you end up with curved graphs, not straight lines.