Nice work... I really like it and would definitely be willing to give it a shot (and even to build my own if necessary).

I think there are two important concerns that need to be addressed that I didn't notice in the post:

1) The chords used to specify a given key should be a fact-based decision, made using actual data concerning

  a) which keys are actually used most for a given target language  and

  b) which motions are actually least expensive/easiest with this system

so a sensible layout could be determined (like dvorak for classic keyboards); and

2) These things seem to keep your hands pretty busy... though I'd like to stick to only using a keyboard, and usually can 9x% of the time, I still _must_ use the mouse once in a while... so the mouse either must be integrated (as a trackball, motion sensor, or ?) or it must be easy to switch.

Have these aspects been considered/resolved?