The present analysis was begun as an attempt to develop a formal statement of the criteria used to distinguish between behaviorally and non-behaviorally stated objectives. However, as a result of the analysis the authors have concluded that to categorize descriptive terms as behavioral or non-behavioral is misleading. This paper, then, focuses instead on an alternative model for describing the language of instructional objectives which the authors believe to be more useful than the currently used behavioral model. In addition, data is presented which demonstrates the effectiveness of the alternative model in explaining and predicting the ways in which educators use their language to specify instructional outcomes. We turn first to the widely used behavioral approach to instructional objectives.Curriculum developers with a behavioristic bent have been adamant that instructional objectives should contain a description of the outcomes of instruction in terms of specific observable human behavior (Taber, Glaser, & Schaefer, 1965; Gagne, 1965). Such a demand stems largely from the current emphasis on task analysis in the development of instructional systems which carries with it the requirement that curriculum specialists describe the subject matter competencies to be acquired by the learner (see, for example, Glaser, 1967) in performance terms. The behavioristically oriented task analyst suggests that unless instructional outcomes are described in behavioral terms it is possible neither to specify when that outcome has been achieved (to measure) nor to systematically design the learning conditions most appropriate for achieving that outcome. Clearly, these recommendations and requirements rest on the assumption that human activities are themselves best described as either "behavioral" or %on-behavioral," and, furthermore, that it is possible and appropriate to distinguish between these two types of description.Many psychologists and curriculum developers have taken the position that all of what people can be observed to do is "behavior"; and, thus, that all the terms of the language used to label what people do are "behavioral." It is not clear, however, that the terms or "action words" (Gagne, 1965) most often recommended and used refer directly to observable human behavior. For example, the terms ('write," "say," '(underline," "identify," "differentiate," "classify," and ccsolve" have all been either recommended for, or used in, writing behavioral objectives. Yet, the words "identify" and "solve" certainly do not refer to directly observable behavior in the same sense as terms such as "underline," and "say." Is it that "identify" and "solve" do not actually refer to observable behavior while "underline" and "say" have behavioral referents? Although such a position may be adopted, it will lead to difficulty. The nature of language use is such that most words categorize, and, therefore, will always be something more and something less than the event or events to which they refer. The use of a word to "label" a part...