lalboard VS kb-layout-evaluation

Good eye! W.r.t the key placement, I was actually inspired by the lalboard, which is an open source version of the DataHand.

Drawing inspiration from the vertical actuations of JesusFreke's lalboard, the keys above and below the home row are positioned such that actuating them involves the continuation of a single motion (i.e. extending or curling the fingers) rather than two separate motions (extending then pressing down or curling then pressing down, as is the case with a conventional flat keyboard layout). Less muscle memory + shorter travel distance = quicker and more comfortable typing!

It also incorporates some novel features: * Drawing inspiration from the vertical actuations of JesusFreke's lalboard, the keys above and below the home row are positioned such that actuating them involves the continuation of a single motion (i.e. extending or curling the fingers) rather than two separate motions (extending then pressing down or curling then pressing down, as is the case with a conventional flat keyboard layout). Less muscle memory + shorter travel distance = quicker and more comfortable typing! * I've yeeted keys that I find exceed a certain (low) threshold of difficulty to press on a standard ortholinear layout. These include the (QWERTY) n and b keys, as well as all the keys outwards from the pinky column. * To make up for this lack of keys, the left side of this keyboard features a powerful thumb cluster with five keys (four of which are vertically actuated: one by the base of the thumb (actuated by curling the thumb inwards), two by the tip of the thumb (actuated by extending the thumb upwards), and another by the thumb knuckle (actuated by moving the thumb outwards)) that allow for fast level-shifting. * The right side incorporates a trackball mount with a mechanism (consisting of a switch mounted underneath the trackball) that allows the trackball itself to act as a "switch" that can be pressed down to both enable mouse movement and activate the mouse button layer. This means that you don't have to press any buttons to enter the mouse layer before using the trackball -- the action of pressing down the trackball (and thus activating the trackswitch) shifts into the mouse layer. I refer to this collective assembly as the "trackswitch". * Both the trackswitch and trackball sensor have fully parameterized mounts with mounting mechanisms that allow their distance from the trackball to be micro-adjustable. * Fully parameterized case mounts for the arduino micro and pro-micro MCUs.

Maybe build something like https://github.com/JesusFreke/lalboard

Have you ever looked are https://github.com/JesusFreke/lalboard ? Adopting thumb clusters from there might work.

On the hardware side you could use switches like in the lalboard which is modeled after the DataHand. All keys use a pair of magnets to provide the clickiness and key return force, and an IR LED and phototransistor for detecting a keypress. It would be a large change on the hardware side but good to be aware of as an option as this design is reminiscent of those two keyboards.

Maybe you can get Portuguese bigrams frequency and run this: https://github.com/bclnr/kb-layout-evaluation I don’t know how different it is from French or Spanish.

I wrote a layout analyzer at some point and that was the reasoning.

And indeed quantitative studies tend to show that Colemak can be very good for other languages, for instance French: https://github.com/bclnr/kb-layout-evaluation (although you need to modify it slightly to add accented characters to reasonable spots). Interestingly enough, French variant of Qwerty (Azerty) is much worse than Colemak or anything.

A while ago I coded a little comparison between layouts to help me with that decision, you might find it useful: https://github.com/bclnr/kb-layout-evaluation

> Some forward-thinking people have set about evaluating key layouts objectively on metrics like these.

[...]

> What If You’re Bilingual?

On that topic some might be interested in this little analysis:

https://github.com/bclnr/kb-layout-evaluation

These scoring systems already exist. Here's the first one I found: https://github.com/bclnr/kb-layout-evaluation