Typing speed test, measured in your own script
Six languages, each with its own word list, its own keyboard and a word count that respects how the script actually works.
Now testing: Hindi · Devanagari · 600 words
Pick your language
Each one has its own word list, its own keyboard and its own measurements — not this page with the labels swapped.
How each script measures up
Each row is measured over that language's own word list, using the same grapheme segmentation a browser uses. The last three columns are the ones worth reading: they say how far a conventional character-based typing test would be off for that script.
| Language | Script | Words | Visual units / word | Codepoints / word | Ratio | Words that stack | Distinct shapes |
|---|---|---|---|---|---|---|---|
| Hindi | Devanagari | 600 | 2.55 | 4.36 | 1.71× | 90% | 310 |
| Bengali | Bengali | 600 | 2.53 | 4.21 | 1.66× | 91% | 228 |
| Gujarati | Gujarati | 600 | 2.79 | 4.93 | 1.77× | 94% | 282 |
| Marathi | Devanagari | 600 | 2.87 | 5.34 | 1.86× | 96% | 304 |
| Nepali | Devanagari | 600 | 3.01 | 5.47 | 1.82× | 96% | 340 |
| Arabic | Arabic | 600 | 3.94 | 4.02 | 1.02× | 7% | 53 |
Why we do not divide characters by five
Nearly every typing test in existence computes speed the same way: count the characters typed, divide by five, divide by the minutes elapsed. The five comes from the average length of an English word. It has been the convention since mechanical typewriters and it works fine for English.
It does not survive contact with an Indic script. In Devanagari a single visual letter is routinely three or four codepoints — क्ष is क, virama, ष — so counting characters counts keystrokes that never appear on screen as separate letters. Measured across the word lists in the table above, that inflates a Devanagari score by close to two to one. Arabic, which barely stacks at all, is almost unaffected.
So these tests report three numbers instead of one, and say which is which. Words per minute counts whole words you finished correctly — the only figure that means the same thing in every language here. Letters per minute counts visual units, using the browser's own Unicode segmentation. Keystrokes per minute counts what your fingers did, which is the figure closest to what typewriter-era exam standards were describing.
The comparison is also why words are compared whole rather than character by character. Being "half correct" on क्ष is not a state a typist can be in.
A test you can take without installing a keyboard
The other reason these pages exist. Search for a Bengali or Gujarati typing test and you will find plenty — all of which assume you already have that keyboard installed. The people searching are largely searching because they do not.
So every test here carries its language's own on-screen keyboard, taken from the keyboard page for that language, and you can click it. It is slower than touch typing and that is fine: it means you can measure where you are, on any machine, without installing anything first.
