How hard is each script to type? Every keyboard here, measured

What can honestly be measured

"Hard" is not one thing, and most of what people mean by it cannot be counted. How quickly a script feels natural depends on whether you read it already, how much you practise, and what you are used to. None of that is in this repository.

What is in here is the keyboards. Every key map on this site was read from the layout file for that keyboard rather than transcribed, and every practice bank is a list of real words in the language. From those two things four questions have countable answers: how many keys a layout defines, how many of them you would have to learn, how much of the language's own text those keys reach, and how many distinct characters the language needs.

Those add up to a fair proxy for the part of difficulty that is about positions rather than about you. They say nothing at all about how fast you will get, and the last section says why.

Every layout here, counted

The table is ordered by how much of the language's own practice words the three unshifted letter rows reach — the chart a reader sees under the lessons, and the part of the keyboard you learn first.

At the top are the alphabets that fit. Russian, Hebrew and Korean reach everything: their inventories sit inside the three rows with room to spare, so the chart is the whole keyboard and there is nothing hiding on another level.

At the bottom are the abugidas. A consonant carries vowel signs, the combinations multiply, and the inventory outgrows the rows — so the chart shows the letters you press most and the rest live on the shifted level. That is not a fault in the layout. It is arithmetic.

What the ordering actually shows

Three patterns come out of it, and none of them is about which languages are difficult.

Alphabets fit and abugidas do not. Every layout reaching all or nearly all of its language is an alphabet with roughly a keyboard's worth of letters. Every layout at the bottom is a script where one written unit is a consonant plus a vowel sign, which multiplies the inventory past what a level of keys can hold.

The Cyrillic layouts cluster high because Cyrillic alphabets are small. Russian's 33 letters fit the letter rows with keys to spare, which is why the whole family sits near the top whatever else differs between them.

And the scripts with the most characters are typed by composition rather than by layout. Korean reaches everything only because its keys produce jamo that assemble into syllables; Chinese, Japanese and Ethiopic are not in the table at all, because no keyboard maps their characters one to a key.

What none of this tells you

It does not predict your speed. The number of positions decides how long the first fortnight feels awkward, and after that what matters is practice, which the table knows nothing about.

It does not measure the script's difficulty as a script. Reading fluency, spelling conventions, whether the language is one you know — all of that is larger than anything here and none of it is in a key map.

And it says nothing about how well any of this is supported outside a keyboard. Fonts, rendering, line breaking and search are separate problems, and some of the scripts near the top of the table have worse software support than some near the bottom.

What the table is for is the one question it can answer honestly: before you start, how many positions is this going to be? For an alphabet, about thirty. For an abugida, about thirty that you will use constantly and as many again on a level you reach for less often.