I've never used Nastran. If I had to guess, the first would be a "quasi-static" approximation. i.e. calculate static response for every value of the input as if the input were steady at that value. Could be a reasonable approximation if the natural frequencies of the system are all much higher than the frequency of variation of the force such that the system is mostly spring controlled, mass acceleration doesn't play much role. The second would be a true dynamic calculation including effects of mass acceleration, damping etc which were neglected in the quasi-static approach.
=====================================
(2B)+(2B)' ?