As a scientist, I would like to congratulate Steven Tripp for an elegant experimental design and report. However as a "teacher" I would suggest that the hypothesis being tested is largely irrelevant because I don't think the various media should be used to present the same information for the same purpose.
I teach a senior undergraduate unit in forest measurement and modeling--lots of theory, statistics, mathematics, and pragmatics. My teaching approach uses a variety of resources: including paper, overhead transparencies, scribbles on blackboards or whiteboards, the melodious sound of my voice (taped and live), computer projections of graphics and animations, Powerpoint slide-shows (also available outside the lecture times), and field-based practical exercises. I go to all this trouble because I think that the different resources (media) present different information that is appropriate to different learning outcomes. For example, an audio presentation (i.e., the spoken part of my lecture) allows the students to hear the passion or conviction I have in an idea whereas the printed notes emphasize instead the logic of the idea. The animations and computer projections emphasize movement and change far more elegantly than my blackboard scribbles, but those scribbles emphasize the development of the concept from its basic parameters. The practical exercises develop problem solving and integration skills more than the lectures, but the lectures do something else. And so on...
Thus, I believe that different media can be used to encourage different learning outcomes. The experiment devised by Steve (and the other research he references) is only allowing for one type of learning outcome--rote learning or memorization. I think he demonstrates his point pretty well--paper is better than oral in the nominated case. Of course, if the learning outcome had been to see if the student's beliefs or actions had changed as a result of his/her study, I could easily believe that a well spoken reading would have more impact than the "dry words" on the page--his would again demonstrate the point that the media is important.
I think that many of the "media is not important" findings are truly confounded by the attempt to have a simple, single learning outcome that can be compared. The selection of this outcome will have dramatic effects on the results--and rote memorization of phrases isn't a really useful learning outcome. If you "want" to prove one media is better or worse, you just need to select an appropriate learning outcome measure--or pick one that is not appropriate to either media to get a "no significant difference!"
At the same time, any attempt to present exactly the same information via two media is difficult. For example, Steve tries to equalize the written and oral presentation of information by setting times and computer interfaces. However, as mentioned above, a lot of information can be provided by the way a speaker speaks--the tone and inflection, masculine or feminine voice, etc. Even the removal of tone and inflection (through computer synthesis) provides its own information content. If you have ever listened to a convincing or entertaining speaker and then read a verbatim script of what was said, it is likely that you will be astounded at what is not in print. To a lesser extent, formatting of the written word can change the information content (e.g., look at English poetry written as prose).
The real questions recognize that there are different types of information. The challenge then is to ensure that the correct types of information can be expressed by the different media.