Perhaps using a phone to remote control a web site powered by Squeak, IM messaging to live running Squeak images, touch tone control, or inserting phone voice mail messages into a Squeak Project files might be useful adjuncts to normal Squeaking. But, I agree with Alan, the controls are too limited for expressing, controlling, and manipulating complex concepts. (The $100 notebook would do better since it can mimic a phone.)
Cell phones might also be used in experiments in linguistics (semantics, syntax), UI design, and cognition by having one person dictate a "program" to achieve a goal and someone at a PC on the other phone trying to create what they hear (and unable to see the speaker's gestures). The first person could see the results of what the second person creates in this controlled feedback loop.
This could be varied to a) a limited vocabulary, b) adjusting the vocabulary as they progress, c) add vocabularies from different disciplines, d) restrict or require the use of metaphors and narrative techniques, and e) compare vocabularies created by different teams.
Also, take note how much is assumed, how much has to be remembered by both communicators, cultural differences of vocabulary meaning and use, varying time and vocabulary efficiencies (and misunderstandings), emotional reactions (and frustrations), and the role of abstractions and mental models.
Another cell phone experiment might include having a student roam an unfamiliar area and verbally communication with a Squeak model of the area.
In addition, cell phones tend to have very minimal resources, proprietary programs and API's, and very differing configurations making general adoption difficult.
Cheers, Darius