A question answering system which permits a computer-assisted instruction (CAI) student greater initiative in the variety of questions he can ask is described. A method is presented to represent the dynamic processes of a subject matter area by augmented finite state automata, which permits efficient inferencing about dynamic processes and provides a satisfactory deep structure for paragraph generation. A CAI system dealing with meteorology is described which uses this automation model to represent the processes in meteorology. Examples of the inferencing techniques using both the automaton model and the semantic network are given. A sample session with the system is included in the appendixes. (Author/RH)