The polarity will not affect the internal construction of the transformer windings but only with the routing of leads to the bushings.
I am ignorant of the secondary rotation, is this the same as angular displacement between the primary and secondary?
If your transformer are connected with subtractive polarity where correspondingly marked terminals for the primary and secondary windings opposite each other, I don't see how that would affect any rotation.
The apparent paradox of the question is what happens to a connected motor when this change is made, with no other changes, such that each secondary voltage with respect to a fixed chosen secondary lead now appears to be opposite in phase to the primary voltage. Consider two different ways of analyzing the problem:
1. If you shift the phase of all secondary voltages (relative to some constant reference signal) by 30 degrees, it is clear that neither the amplitude nor the phase rotation direction change. Now increase the phase change to 60 degrees, then 90 and so on. Eventually you reach a phase change of 180 degrees on all three phase leads and yet the direction of motor rotation has not changed. Seems simple and very hard to misinterpret, therefore probably correct.
2. Alternatively, we know that reversing the position of two phase leads as connected to the motor will reverse the direction of rotation. Reverse a second pair and you will go back to the original rotation. Make a third reversal and you are in opposite rotation again. The argument now goes that reversing the polarity of all three phase leads is equivalent to interchanging leads three times, and so should cause opposite rotation. I know that this argument is wrong, and the problem now becomes to determine what is wrong with the model I just started out with.
I have some pretty clear ideas about this, in particular the realization that if I switch the polarity of one secondary winding it is NOT the same as reversing two of the leads connected to the motor (and quite spectacular in result). Let's see what others have to say!
(As math textbooks often say, the proof is left as an exercise for the reader.)