What is written below has been known in the field since the mid-1970's, especially from a Precision Teaching and Direct Instruction perspective. What is new is the explicit addition of an Emergence stage of operant evolution and inculcation. That addition now gives us a model for accomodating the multiple sources of control (Skinner, 1957, Chapt, 9).
Phase 1 is acquisition, which typically, but not always, shows slow progress that requires lots of ASRs (many more if not precisely controlled as through, say, the delivery of Learn Units).
Phase 2 is generalization and stabilization through schedule thinning.
Phase 3 is retention and sustainability through fluency and non-fluency programming across contexts and reinforcers.
Phase 4, the current addition, is the emergence of new, higher-order operant structures, or "repertoires" that possess their own functional and feature characteristics that, in large part, allow causal relations that are local to the emergent structure alone.