emacs shorthand

[Rustom Mody]

This post is prompted by Xah's fervent plea to type better.  The emacs related questions are at the end.

The basic question I am addressing is "Can we type better than the usual naive use of QWERTY?"

Clearly, given the nature of human hands ('a' and 'e' are harder than hjkl on qwerty) the qwerty keyboard is suboptimal and so we have things like dvorak etc.

But dvorak makes an assumption that was necessary for mechanical typewriters but not modern computers, viz. that one key-press should correspond one-to-one to one character in the output.

There are two (that I can think of) ways to un-assume this.

1. Chording: Instead of typing one char at a time allows 3-4 to be typed simultaneously.
Two specialised hardwares exist for this -- stenotype and palantype.
Since I dont know too much of this I wont say much more.

2. Alphabetic shorthand

To understand this consider first abbrevs. Clearly if one could identify a small number of common words, one could make them into abbrevs.

The next step is to go to abbrev-patterns.  Many English words have very frequently occurring patterns like the -ing and the -tion endings.  These could be abbreviated to one letter shortforms.

See for example http://www.freewebs.com/cassyjanek/keyscriptsample.htm for an example of a system devised on these lines.

If this were done right, one could conceive of an emacs mode that takes abbrev-ed input and expands it on the fly.

So (finally!) my question:
Either we need to soup up abbrev mode to recognize sub-abbrevs inside words (like the -tion and -ing suffixes)
Or we could inline (large parts of) a dictionary as abbrev tables.

What would be the performance of option 2?  ie How well would say a five thousand word abbrev table perform?

Thanks

Links:
The first option can be studied further by googling for stenotype and palantype
The second by easyscript, keyscript, shorthand