The ultimate goals of research on teaching are theories of teaching and these, in turn, involve the development of a critical language for the analysis of classroom behavior. A language consists of both syntax and vocabulary. Hence, both a general paradigm and a general taxonomy must be sought. The paradigm provides the syntactic structure upon which theoretical vocabulary can then be built. The position taken here is that the production of vocabulary has too frequently proceeded without adequate models to serve as syntax and has been limited by overconcern for the practical problems of classroom observation. This paper therefore concentrates on the metalanguage and methods needed in developing a critical language rather than on the details of the language itself or of its use.Over the years, a number of paradigms for research on teaching have been generated (Gage, 1963(Gage, , 1964. Gage's 1963 summary documented a growing transition from the older and more general "criteria-of-teacher-effectiveness'' terminology to a newer and 1.