Here's my attempt at an answer. Note that I'm just a biochemist and a physicist would probably be able to answer the question more completely:
Plane polarized light can be considered to be the superposition of two counter-rotating electrical field oscilations. Molecules which have symmetry elements which prevent them from being chiral will interact equally with each electrical field component. However, chiral molecules will interact more strongly with one electrical field component. Therefore, the two components travel at different velocities through an optically-active (chiral) environment. Since one component arrives at the observer later than the other, the plane of the light appears rotated.
Now, it is probably true that different molecules in different orientations will interract with PPL differently. However, in a liquid, the molecules will be fairly randomly oriented, so the rotation of the PPL light will essentially be averaged out. Just as one would expect a coin to land on heads 50% of the time if flipped a million times, one would expect the same rotation of PPL when passed through a million molecules in random orientations.