How long does it take to learn Colemak?

How long does it take to learn Colemak?

How long it takes to learn Colemak is a range, not a number. Public logs put the return to your old QWERTY speed anywhere from about three weeks to a full year. The thing that predicts your result is not the calendar. It is total hours of deliberate practice, and whether you keep QWERTY for work while you learn.

colemak.com does not publish a figure. Its FAQ answers the question in one line: "Each person will have a different learning experience, and it depends on many factors." It then points you at the forum. That is the honest answer, so the rest of this page is what those forum threads and published typing logs actually report, with names attached, so you can calibrate against people whose starting conditions look like yours.

If you have not committed to Colemak yet, the timeline is the wrong question to open with. The choice between Colemak, Dvorak and staying on QWERTY is a separate decision with a separate answer. This page owns only the learning curve.

Why does nobody publish a number?

Four variables move the result more than the layout does, and no two learners share all four.

• Practice hours per day. Ten minutes a day and two hours a day reach the same milestone at wildly different dates. Every "six weeks" claim you read is really an hours claim with the hours removed.
• Cold turkey or gradual. Deleting QWERTY from your day is the fastest route and the most expensive first fortnight. Keeping QWERTY for work protects your output and stretches the timeline, because both layouts compete for the same finger positions.
• Your starting speed. A 90 words per minute (WPM) typist has further to fall and further to climb back. Someone who never touch typed at all often ends up faster on Colemak than they ever were on QWERTY, because they are learning to touch type for the first time and the layout is incidental to that.
• Speed chasing or accuracy first. Practising speed on top of a 90% accuracy rate trains the errors, so the plateau arrives early and then stays.

colemak.com's Learn page does give process advice, and it reads as advice rather than as a timeline. It suggests "practicing for 10 minutes every day for a couple of months, and switching when you've reached two thirds of your previous typing speed." The same page warns that "The first 10 days are going to be difficult." Neither line is a promise about when you get your speed back, and the FAQ deliberately refuses to make one.

What does the Colemak learning curve actually look like?

Phase one is a collapse. On day one you will type at roughly a sixth of your normal rate. Touch typing is position memory rather than letter memory, so when the positions move the memory has nothing left to point at. You are not slow because Colemak is difficult. You are slow because you are hunting for keys again, which is what you did before you ever learned to touch type.

[
  {
    "label": "Start",
    "colemak_wpm": 12,
    "prior_qwerty_wpm": 75
  },
  {
    "label": "Day 6",
    "colemak_wpm": 24,
    "prior_qwerty_wpm": 75
  },
  {
    "label": "Day 17",
    "colemak_wpm": 40,
    "prior_qwerty_wpm": 75
  },
  {
    "label": "Day 24",
    "colemak_wpm": 50,
    "prior_qwerty_wpm": 75
  }
]

Those 4 points come from Annie Cherkaev's public log "colemak is for quitters", written after a cold turkey switch from 75 WPM on QWERTY. She started at 12 WPM, reached 24 by day six, and sat around 50 by day 24. She also counted the price: "I spent a total of 24 hours (!) doing typing tutors, and an additional 17 hours writing." That is roughly 41 hours of typing packed into 24 days. Read it as one person's curve with its practice budget attached, not as a schedule you will match.

Phase two is deliberate typing. Somewhere around 20 WPM the letters come back to you, but you have to ask for each one. The word arrives in your head, then you place it letter by letter. This phase costs more when you write code than when you write prose, because you are already holding a model of the program in working memory and the keyboard is now taking a share of it. Cherkaev's marker for the end of it is precise and worth stealing: by day 17 she "typed notes on a video call without missing what was being said."

Phase three is the plateau. Most people stall between 40 and 55 WPM, and stay there long enough to doubt the whole project. The cause is that progress has gone uneven. The common letters are automatic, while rare letter pairs, capitals and punctuation-dense code are still deliberate. If you spent phase two racing the WPM counter, there is a second cause on top: an accuracy debt, where every fast run is followed by a backspace run, so your net rate stops climbing even though your raw rate does not.

Phase four is the crossover, if it comes. Your old QWERTY speed was built over years, so matching it inside a few months means you are learning faster than you originally did. Be clear about the payoff. Crossing your old QWERTY number is not guaranteed and it is not why most people stay. Colemak's argument is finger travel and hand alternation, which is comfort over a long day. If the number is what you want, a month of accuracy drills on QWERTY is a cheaper way to get it.

What did people actually report?

Every figure below is from a source you can open and read yourself.

• Annie Cherkaev, "colemak is for quitters": 75 WPM on QWERTY, cold turkey, about 50 WPM after 24 days and roughly 41 hours of typing.
• Colemak forum, "40-60ish WPM Qwerty": the user bph, coming from about 55 WPM on QWERTY, wrote that it "took me 6 months to hit ~55wpm" and reports "about 70-80wpm now on a good day".
• Colemak forum, "Typing Speed Progress": Tony_VN summarised the room as "Main consensus here is that you will get back to your Qwerty speed in three months (~90 days) but with more comfort."
• Colemak forum, "How many days of practice until basic proficiency": aubug reported 46 WPM after a month from a 50 to 60 WPM start, off "13 hours or so" on one tutor site. boli reached half of a previous QWERTY speed in "roughly 3 weeks". simonh switched during a week off work at "at least a couple of hours a day" and was comfortable enough by the time work resumed.
• colemak.com's Learn page links a log of "from 6 to 74 wpm in 31 hours" as inspiration. That is a best case from someone practising hard, and the page presents it as inspiration rather than as an average.

The pattern under all of it is hours, not weeks. Two people who both say "three weeks" can be 6 practice hours apart or 60, which is why the week counts disagree so violently. Convert any claim you read into hours before you compare it to your own plan.

Cold turkey or gradual: which should you pick?

Cold turkey means QWERTY leaves your day entirely. It is the fastest route to automatic typing, because every keystroke you make becomes practice instead of only the ones inside a tutor. It also means one or two weeks where your output drops to a level you would not accept from yourself. simonh's approach on the forum is the sane version: "I decided to switch during a week off work to give myself a bit of time to adjust, and I think that helped a lot." Book the quiet week first, then switch.

Gradual means Colemak at home and QWERTY at work. Your output survives and the timeline stretches, because the layouts interfere with each other while both are half learned. The Colemak FAQ is direct about the cost: "While beginning to learn the layout you are likely to experience a fall in typing speed in QWERTY as you adjust." It is equally direct that the interference is temporary: "Most people who have mastered both layouts can switch easily between keyboard layouts on a regular basis."

There is a middle route. Tarmak, by the forum user DreymaR, is a set of five transitional layouts. Each step moves three or four keys, so you relearn a small group at a time instead of all 17 at once, and you keep working throughout. The thing that makes it work is a gate rather than a clock: "you should aim for about 97% accuracy at 40 WPM before you move on to the next step." People who skip that gate carry the error rate into every later step and stall at the end with no obvious cause. Reported Tarmak timelines run from about nine days to several months, which is the same spread as everything else here.

Decide the variant before you start, because changing it afterwards costs a second relearn. Colemak Mod-DH, which shifts a few keys off the index finger stretch, costs nothing extra if you learn it from day one and costs real weeks if you move to it in month four.

What happens to vim, tmux and shell shortcuts?

Colemak moves 17 keys and leaves everything else where QWERTY put it. Only two letters change hands: E moves to the right hand, and P moves to the left. Q, W, A, Z, X, C, V, B, M and H do not move at all, and the layout "keeps almost all of the punctuation keys in their QWERTY positions to ease the transition from QWERTY".

That is why copy and paste survive. The FAQ states it plainly: "Colemak keeps the Ctrl+Z/X/C/V in the same place." Ctrl+A, Ctrl+B, Ctrl+W and Ctrl+Q are untouched for the same reason, which means tmux is unaffected: its default prefix is Ctrl+B, and B does not move. GNU screen's Ctrl+A is safe too.

The shell is where it starts to hurt, because readline (the line editing library bash uses) binds by letter, so any binding on a moved letter moves with it. Ctrl+A still jumps to the start of the line under your left little finger. Ctrl+E, which jumps to the end, now sits on the physical QWERTY K key under your right middle finger, so a pair you used to hit with one hand is split across two. Ctrl+R for reverse history search is on the physical QWERTY S key. Ctrl+U is on QWERTY I, Ctrl+K is on QWERTY N, and Ctrl+D, the key that logs you out of a shell, is on QWERTY G. None of these break. They simply are not where your hands go.

vim is the worst case, because hjkl was chosen for its shape and the shape is gone. In Colemak, h stays on the QWERTY H key. j sits on the QWERTY Y key, k on the QWERTY N key, and l on the QWERTY U key. In standard touch typing all four of those are right index finger keys, so the cursor row stops being four keys under four fingers and becomes one finger reaching across three rows. You will feel this within an hour of your first :e.

You have two honest options. Learn the new positions and accept a slower index finger, which costs nothing to maintain and keeps every tutorial correct. Or remap the motions to h, n, e and i, because those four letters land on the physical QWERTY H, J, K and L keys in Colemak, so your fingers do exactly what they always did. The cost lands on n, e and i, which lose repeat-search, end-of-word and insert and need new homes, and on every :help page, plugin default and colleague's config still saying hjkl. Pick one in week one and write the choice down, because switching this in month three is a second relearn stacked on the first.

One more key your hands will miss. The stock layout has no Caps Lock: an extra Backspace occupies that position, and on Linux the us(colemak) XKB (X keyboard extension) variant maps that key to Backspace as well. Terminal users usually already own that key, because remapping Caps Lock to Ctrl or Escape is the first thing many people do to a new machine. So you have a conflict to settle on day one. To see which behaviour you got, press the key in a text field: if it deletes the character to the left, the variant claimed it. An XKB option such as caps:escape takes it back, and setting up the Colemak layout on Linux covers where to declare that so it survives a reboot.

What about the server console that is still QWERTY?

SSH (secure shell) carries characters, not key positions. Your own machine turns the key you pressed into the letter t before anything leaves it, so every SSH session on every server behaves exactly like your local terminal. No server needs to know you use Colemak, and no server can be configured wrongly for you over SSH.

Every other console is a question mark. A provider's web console, usually VNC (virtual network computing) or a serial console in a browser tab, may pass your keystrokes through your local layout or hand raw key positions to the remote machine, and a fresh Linux install is QWERTY. You find out which at the exact moment your server will not boot. Test yours on purpose while the server is healthy: open the console and type a short sentence containing e and p, because those are the two letters that change hands and they expose the problem immediately.

On the server itself, the layout you set inside a desktop session does not follow you to the text console. localectl status prints VC Keymap and X11 Layout on separate lines, and they can disagree, so a machine that is Colemak in your session is still QWERTY on the virtual console you drop to after a failed boot. The same applies to a full disk encryption passphrase prompt, which runs from the initramfs (the small early boot filesystem) and uses whatever keymap that image was built with. Test that once deliberately, before an unplanned reboot tests it for you.

Two habits make all of this survivable. Keep enough QWERTY alive to type a password and one systemctl command, which most people retain without effort once the interference phase passes. And do not relabel your keycaps: colemak.com's Learn page advises against it, and the mechanism is simple. A keyboard you can read is a keyboard you will look at, and every glance down is a repetition your fingers did not get, so relabelling slows the exact memory you are trying to build.

How do you know you are making progress?

Measure accuracy before speed. colemak.com's Learn page sets the bar at "at least 97% accuracy without hesitating" before moving on, and that is the same gate Tarmak uses between steps. Cherkaev's turning point matches it: "From about 28 wpm I switched to focusing on accuracy rather than trying to increase my speed." Accuracy converts into speed later. Speed built on errors converts into a plateau.

Count hours, not days. Two hours a day for two weeks is 28 hours of practice. Ten minutes a day for two months is 10 hours. The first person will be far ahead of the second, and neither of them has a week number that means anything to you. Log your hours next to your WPM and your own curve becomes readable within a fortnight.

If you are still under 30 WPM after 30 focused hours, look for the usual cause before you blame the layout: you are looking at the keyboard. Position memory forms from the hand finding the key without help, so a glance down replaces the repetition you were about to earn. The fix is uncomfortable and fast. Cover your hands, slow down to the speed where you are accurate, and let the speed come back on its own.

FAQ

How long does it take to learn Colemak?

Published reports run from about three weeks to a year to get back to your previous QWERTY speed, and colemak.com refuses to give a figure because the spread is that wide. The clearest data points: Annie Cherkaev's cold turkey log reached about 50 WPM from a 75 WPM QWERTY baseline in 24 days and roughly 41 hours of typing; the forum user bph, starting near 55 WPM, wrote that it "took me 6 months to hit ~55wpm"; Tony_VN described the forum consensus as "three months (~90 days)". Convert those into practice hours before comparing them to your plan, because hours, not weeks, are what actually differ.

Will learning Colemak ruin my QWERTY typing?

Temporarily, yes, and the Colemak FAQ says so directly: "While beginning to learn the layout you are likely to experience a fall in typing speed in QWERTY as you adjust." It happens because both layouts are competing for the same finger positions while neither is automatic. The same FAQ says the effect passes: "Most people who have mastered both layouts can switch easily between keyboard layouts on a regular basis." Keep enough QWERTY to type a password and a recovery command, because you will need it on a machine that is not yours.

Does Colemak break vim's hjkl?

It scatters it. h stays on the QWERTY H key, j moves to the QWERTY Y key, k to the QWERTY N key, and l to the QWERTY U key. All four are right index finger keys in standard touch typing, so one finger now covers three rows. The common fix is remapping the motions to h, n, e and i, because in Colemak those letters sit on the physical QWERTY H, J, K and L keys, so your hand moves exactly as before. The cost is that n, e and i lose their own commands, and that every plugin default and help page still assumes hjkl.

Do Ctrl+C and Ctrl+V change in Colemak?

No. The Colemak FAQ states that "Colemak keeps the Ctrl+Z/X/C/V in the same place", and Ctrl+A, Ctrl+B, Ctrl+W and Ctrl+Q do not move either, which is why tmux keeps its default Ctrl+B prefix and GNU screen keeps Ctrl+A. The shortcuts that do move are the ones on the 17 relocated keys: Ctrl+E is now on the physical QWERTY K key, Ctrl+R on QWERTY S, Ctrl+U on QWERTY I, and Ctrl+D on QWERTY G. Caps Lock is a separate problem, since the stock layout puts Backspace there and you may already be using that key as Ctrl or Escape.

Should I go cold turkey or keep QWERTY for work?

Go cold turkey if you can book a quiet week for it, because every keystroke then counts as practice and the recovery is fastest. Keep QWERTY for work if your typing output is the job, and expect the timeline to run in months rather than weeks. The middle route is Tarmak, five transitional layouts that move three or four keys at a time, gated on "about 97% accuracy at 40 WPM before you move on to the next step". Whichever you pick, decide your variant first, because moving to a different Colemak variant later means relearning keys a second time.