It is obvious the C word and the I word have figured highly in the debate so far. However I think this argument can mask some of the more important issues you raise in your paper--i.e., how can we more successfully categorize the work we spend our lives producing and improve our methodologies for design and evaluation.
I think the idea of separating the navigational engagement from the instructional engagement is very important distinction because it clearly delineates access from content and understanding. The article by Galliano (1993) views interactivity as a function of the learning technological environment, it has an impact on media (technical representations), languages (presence of different symbolic systems and languages), and methods (teaching & learning processes). In seeking to quantify the axes of a taxonomy for multimedia simulated experimentation (my area of interest), I have identified three axes that interactivity impinges upon. These are media/modeling, teaching-learning process, and motivation/challenge. I'm really not sure whether interactivity is a fourth axes or is just a function of the interrelated characteristics of the other three.
You quote Laurillard (1993) in your paper but omit to mention the contribution the author makes in classifying the learning delivery process--namely the dialogue that is necessary between delivery system (teacher and/or technology) and the learner which is divided into four types: discursive, adaptive, interactive, and reflective. While the author's work is not exhaustive or prescriptive, it does point developers and users to identify appropriate media and their limitations in the delivery of learning.
I think interactivity could be considered at two levels: surface and deep. Surface interactivity is predominantly associated with navigation whereas deep interactivity engages the user in what you describe as learning engagement.
Instructional technology designers need to separate the appeal of media and technology from the nature of the teaching-learning process. This, I fear, is what we are not very good at because we do not have an effective evaluation mechanism. Gredler (1986) attempted this process and. while the paper is a bit dated with respect to technology, it attempts to classify micro computer simulation systems (very like modern multimedia systems) into a number of distinct types by separating media, instructional process, and feedback mechanisms. I attempted to update this in a paper this summer (Bates, 1995). The process is ongoing. What I'm looking for is a taxonomy that can remove some of the subjectivity from our evaluations and provide developers and users with a framework to classify and compare applications that look similar but deliver totally different levels of learning. Ideally, I'd like to be able to identify something like Alessi's (1991) five major types of CBI and then have some comparative measure of the media/modeling, teaching-learning process and the motivation/challenge incorporated in each implementation. How could your three views of interactive instruction help me with this?
Alessi, S.M., & Trollip, S.R. (1991). Computer-based instruction: Methods & development (2nd ed.). Prentice Hall 2nd. edition, pp. 9-10.
Bates, B., Leary, J.J., & Saadat, S. (1995). Virtual laboratory experimentation: A review of the state of the art and current research. Hypermedia in Sheffield 1995.
Galliani, L. (1993). The technological text: Hypertextual, hypermedial and didactic applications. In Giardina M. (Ed.), Interactive multimedia learning environments (pp. 221-225). Berlin: Springer-Verlag.
Gredler, M.B. (1986). A Taxonomy of computer aimulations. Educational Technology, 26(4), pp. 7-12.