One way to achieve a wireless, wearable, chording keyboard might be to
hack a wireless qwerty as follows.

Suppose you have a commercial, wireless qwerty keyboard. Maybe the
link is infrared, Bluetooth, Zigbee or some sort of wireless USB.
Whatever, inside are circuits sufficient to convert key presses to
something the host understands. This circuitry is likely highly
integrated, i.e., tiny, and non-programmable. You can't hack it but
you can remove it intact from the keyboard. This makes available a lot
of connections that used to go to the keys.

The way all (or most) conventional keyboards work is that the switches
are wired in a matrix. The microcontroller inside the keyboard applies
voltage to one whole row of switches at a time and then tests all the
columns to see whether any switches in that row are being pressed.
Then quickly on to the next row and so forth. There might be 7 rows
and 15 columns to cycle through, depending on the keyboard.

The microcontroller and host are programmed to handle certain chords
like shift-alt-b and control-alt-delete but if you try to press a-b-c,
all at once, you'll just get those three characters one after the
other in whatever order your fingers landed.

So there is no way to connect 8 or so keys to the excised circuitry so
that chording them looks like pressing individual keys (or small
chords) on the 101-key qwerty.

What's needed is called (I think) a crossbar switch. It's like the
original switch matrix except instead of switches it has additional
input lines so that the 8 chorded keys define intersections between
the rows and columns. So it might have 7 row inputs and 15 column
outputs to hook up where the switch matrix used to go, and 8 lines to
go to the keys to a one-hand, wearable keyboard.

Naturally you want this crossbar switch to be programmable so that you
can control which chords connect which row inputs to which column
outputs. So you want a microcontroller with at least 8+7+15=30 i/o
bits. I'm thinking the Atmel ATmega8515, which you can get in an
easy-to-handle 40-pin DIP package, might suffice:

http://www.atmel.com/dyn/products/product_card.asp?part_id=2006

Now you program the '8515 to fool the original circuitry into thinking
the qwerty switches are still there.

The obvious advantage of this approach is that whoever made the qwerty
for you has taken care of whole lot of design and engineering and made
it available at a relatively mass-produced price. You just need the
courage to rip open that beautiful new qwerty.