That's right, I'm talking about the Y and U keys from the qwerty keyboard layout. I'm going to practice as you suggest and see if I can't find some method of comfortably pressing that key without too much stretching. The main candidates with a little quick experimentation seem to be shifting my right hand's resting position to the left or using my middle finger for as many finger-rolls as I can.

You're also probably correct that my right index finger has never really moved like that before. It may just take a little time to get used to.

Hopefully one of those will be comfortable enough to get back to my original qwerty speed without any undue strain!

@davkol: The typematrix keyboard is sort of interesting, but I'm deeply attached to my IBM Model M! I'm typing on it right now, I just love the thing. 30 years old and still working perfectly.

@DraymaR: Thanks for the tips. I've kept my wrists straight for typing, I'm fairly concious about it since I get pain in my wrists if I don't after extended periods of typing. I'll try that finger curling/armpushing technique and see if it works for me.

I've noticed that I started typing 'nk' with my middle and index finger as well. Some others are 'ln' 'lk' for which I do the same thing.

That's an impressive bag of tricks! I used to use the ergonomic emacs keybindings and it's actually a really similar idea to your extended mappings except usable only from within emacs. I'd have to tweak it a little I think, especially since I already use capslock for another meta-mode like set of keybindings (for controlling my tiling window manager).

There's not loads going on on the RHS of Qwerty.  So I think you'd naturally gravitate to the top row, where four common vowels lay.  You'd also transposed left a column, which is quite sensible when you look at the layout.

Some people type with a curve, some people type flat.  Look at Sean Wronga's videos!  Having your fingres flat doesn't meld well with the home row.   But would probably favour Qwerty to some degree.

When learning Dvorak, because it actually uses a substantial amount of the right hand side of the keyboard - my right hand found it alien, because it had to work.  For my right pinky - it was like learning to walk.  I'd used it rarely.

I tried out your bag of tricks today, it seems like a really cool idea. I think for it to be most effective it should be bound to something at least as convenient to press as capslock, but that leaves me with a dilema. Where do I  put my bindings for my window manager?

I previously rebound capslock to the super key and used it along with some vim like keybindings for moving windows around and resizing them. I also had applications bound to several different keys.

My options at this point seem to be leaving the keybindings on the actual super key (the windows key) which is inconvenient to press (much moreso than left alt or ctrl),  or the right ctrl and alt (which are also difficult to press, and not present on my laptop). I still require left control and alt for emacs since it uses those heavily for keybindings so those are out. I think I might just use emacs up and down (C-n C-p) because those are actually decently placed in colemak as opposed to in qwerty. The bindings are also quite prevalent and can be found in most applications.

Also, one thing that might be useful to people that don't use gnome is a mention of how to enable your bindings from the commandline.

For the us varient:

setxkbmap -model pc104 -layout us -variant cmk_ed_us

setxkbmap -option lv5:caps_switch_lock,misc:extend

You're welcome!

As for making the modifier keys work on normal letter keys, that's something I can't tell you.

I did puruse the xkb protocol specification a little and didn't find anything that would point to the conclusion that it isn't possible.

I'm guessing you've already tried using xev to try and diagnose the problem, but in case you haven't. Using those letter-modifiers does report something interesting:

KeyPress event, serial 32, synthetic NO, window 0x2600001,
    root 0x150, subw 0x0, time 10503788, (363,665), root:(364,690),
    state 0x30, keycode 41 (keysym 0xffe3, Control_L), same_screen YES,
    XKeysymToKeycode returns keycode: 37
    XLookupString gives 0 bytes: 
    XmbLookupString gives 0 bytes: 
    XFilterEvent returns: False

KeyRelease event, serial 32, synthetic NO, window 0x2600001,
    root 0x150, subw 0x0, time 10503828, (363,665), root:(364,690),
    state 0x30, keycode 41 (keysym 0xffe3, Control_L), same_screen YES,
    XKeysymToKeycode returns keycode: 37
    XLookupString gives 0 bytes: 
    XFilterEvent returns: False

KeyPress event, serial 32, synthetic NO, window 0x2600001,
    root 0x150, subw 0x0, time 10503828, (363,665), root:(364,690),
    state 0x30, keycode 41 (keysym 0xffe3, Control_L), same_screen YES,
    XKeysymToKeycode returns keycode: 37
    XLookupString gives 0 bytes: 
    XmbLookupString gives 0 bytes: 
    XFilterEvent returns: False

KeyRelease event, serial 32, synthetic NO, window 0x2600001,
    root 0x150, subw 0x0, time 10503868, (363,665), root:(364,690),
    state 0x30, keycode 41 (keysym 0xffe3, Control_L), same_screen YES,
    XKeysymToKeycode returns keycode: 37
    XLookupString gives 0 bytes: 
    XFilterEvent returns: False

KeyPress event, serial 32, synthetic NO, window 0x2600001,
    root 0x150, subw 0x0, time 10503868, (363,665), root:(364,690),
    state 0x30, keycode 41 (keysym 0xffe3, Control_L), same_screen YES,
    XKeysymToKeycode returns keycode: 37
    XLookupString gives 0 bytes: 
    XmbLookupString gives 0 bytes: 
    XFilterEvent returns: False

KeyRelease event, serial 32, synthetic NO, window 0x2600001,
    root 0x150, subw 0x0, time 10503872, (363,665), root:(364,690),
    state 0x30, keycode 41 (keysym 0xffe3, Control_L), same_screen YES,
    XKeysymToKeycode returns keycode: 37
    XLookupString gives 0 bytes: 
    XFilterEvent returns: False

KeyRelease event, serial 32, synthetic NO, window 0x2600001,
    root 0x150, subw 0x0, time 10504600, (363,665), root:(364,690),
    state 0x30, keycode 66 (keysym 0xfe13, ISO_Level5_Lock), same_screen YES,
    XLookupString gives 0 bytes: 
    XFilterEvent returns: False

Note how close the time of the events are. The modifiers are being "pressed" and "released" rapidly just as you would normally expect from letter keys when they are held down. You probably need to condionally control their repeat rate or somehow change them to act more like modifiers. I'm really not sure if it can be done, but the answer probably lies deep in the xkb and x window protocol specifications.