User:Dyelsey1/Final project

From LING073
Jump to: navigation, search


The code for my keyboards is available on my public github.

For the OSX keyboard, I used SIL's Ukelele.

For Windows, I used Microsoft Keyboard Layout Creator.


For each operating system, I have included 2 keyboards. For Windows and OSX, this includes a Russian-y layout and a custom layout. For linux, this includes a Russian-y layout and a phonetic layout. I have also included bash scripts for installation for OSX and Linux.

For the Russian-y keyboard, I tried to keep the keyboard layout as close as possible to the regular Russian layout. This is to allow users who are already familiar with the Russian layout get started faster. Judeo-Tat has a few different letters than Russian and I moved things around accordingly.

For the custom layout, I took my Judeo-Tat corpus and found letter frequency counts and letter successor frequency counts. The code for this is here. To run this, you need to be in my corpus directory, which is a private repo on Swarthmore's institutional github server in order to not violate licenses. This told me which letters were most popular and which letters were likely to follow other letters. After some research on Dvorak vs. Qwerty keyboards, I found that an optimal keyboard layout would have the most common keys on the "home" row (asdf on an english keyboard). It would also have different hands typing letters most commonly found next to one another. For example, your left hand could be typing one letter while your right hand is already moving to type the next letter. Both of these things increase words per minute (wpm). Keeping those in mind, I tried to layout my optimized keyboard accordingly.

The optimized keyboard is also better than the Russian keyboard because Judeo-Tat naturally has different sounds than Russian so although the Russian layout may be good for Russian, it may also be horribly inefficient for Judeo-Tat.

Judeo-Tat (Russian layout)
Judeo-Tat (Optimized Layout)
Judeo-Tat (Phonetic layout


Both these keyboards have the problem that they have a learning-curve to get started. Typing is largely based on muscle-memory so in order to get a good test would be difficult without having an experienced user for both keyboards to have a "type-off". In my own findings, however, I found that my optimized layout was much better for typing Judeo-Tat because I found the most used letters more accessible and overall had a much more pleasant experience. Being as I do not natively know either layout, I experienced the same learning-curve for both, and thus was able to get a feel for both keyboard layouts from the stance of someone simply typing the language.