Breaking the 10 WPM Plateau

The wall turns up in almost the same place for almost everybody: somewhere between 9 and 12 words per minute, after a few good weeks where the numbers went up every session. Then they stop. You practise the same amount, you feel like you are working just as hard, and the score does not move for a month. The usual diagnosis is a talent ceiling. It is not. It is a strategy that ran out of headroom, and it ran out on a schedule you can calculate in advance.

Why it is always around ten

Do the arithmetic. The word PARIS is fifty units long including the space that follows it, which is where the words-per-minute convention comes from, and it averages ten units per character. At 10 WPM a unit is 120 milliseconds, so a new character lands in your ear roughly every 1.2 seconds. At 15 WPM that gap is 800 milliseconds. At 20 WPM it is 600.

Now time the process you are running. You hear four short sounds, you count them — one, two, three, four — and you look up four dots. Counting consciously and retrieving a letter from a table takes most people between half a second and a second, on a good day, fresh. At 1.2 seconds per character that fits, barely. At 0.6 seconds it does not fit and never will, because the bottleneck is not your ears or your hands. It is the one conscious channel you have, and it does not get faster with practice.

That is the whole explanation. You did not run out of ability at 10 WPM, you ran out of counting at 10 WPM, and everyone using that method arrives at the same number — which is why the wall is so suspiciously uniform across a hobby full of very different people.

What replaces counting

Nobody hears the word cat and assembles it from three phonemes. The whole envelope arrives as one object and the meaning comes with it. That is the target state for a Morse character. Not four dots, but di-di-di-dit, one rhythmic shape with a name attached, recognised the way you recognise a doorbell rather than the way you read a table.

The catch is that a sound only becomes a shape at a speed where it is a shape. Below about 12 words per minute the elements of H are four separate events with audible silence between them and the only available strategy is counting. Practising there builds the opposite skill, very efficiently. Every hour spent copying slowly is an hour spent making the wall thicker.

Two spacings, two different jobs

Farnsworth and Koch get mentioned in the same breath so often that people assume they are alternatives. They are not. They fix different faults and they are almost always used together. The learn page covers the mechanics; what matters when you are already stuck is which knob to turn.

Farnsworth spacing keeps characters at full speed and stretches the gaps between them. Character speed 20, effective speed 9, means each letter sounds exactly as it will at 20 WPM but you get a long pause to think. The pause is scaffolding and it is supposed to come down. What it protects is the sound of the character, which is the part you cannot afford to learn wrong.

Koch's method holds the speed fixed and grows the alphabet instead. Two characters at full target speed, 90 percent accuracy, add a third. It exists to make counting impossible from the first minute rather than merely discouraged.

The practical consequence is counter-intuitive enough that people resist it: if you are stuck, raise the character speed. If you have been running characters at 13 WPM with a little extra spacing, push them to 20 or 22 and open the gaps until your effective rate drops to 8. Your score gets worse for three or four sessions, then passes where it was. Character speed carries you through the wall; effective speed is just how long you are given to think, and thinking time is what is supposed to shrink.

Work out which failure you have

Three questions, pointing at three different fixes. Can you copy isolated random characters at 25 WPM, with two seconds of silence between them, at 90 percent? If yes, your characters are sound-objects and the problem is keeping up. If no, some are still table lookups. Do you lose the next character while decoding the last? That is the chase — a buffering problem, not a recognition problem. Do you copy plain English fine and fall apart on callsigns? Then you are leaning on the redundancy of the language to guess what you missed, which works until it matters.

Five drills

Instant character recognition. Single characters at 25 to 30 WPM with two seconds of silence between them, spoken aloud rather than written. Five minutes, every session, forever. Any character that takes visible effort goes on a list and gets its own three-minute block the next day. This is the drill that finds the four or five letters quietly costing you a word per minute each.

Squeeze the spacing. Keep character speed where it is and take effective speed up one word per minute at a time. One step, four or five sessions, next step. It is the only part of this that is purely mechanical.

Copy behind. Deliberately do not write the character you just heard. Write the one before it. It feels awful for about a week and then your copy stabilises, because you have stopped trying to decide anything inside the 600 milliseconds when the next sound arrives. Every fast operator does this, most without noticing.

Whole-word sounds. A short list of things you hear every session — THE, AND, FOR, YOU, DE, CQ, TU, RST, QTH, NAME, ANT, RIG, WX, TNX, 73 — drilled until each arrives as one object rather than a spelling. Same jump as character to sound, one level up, and it is where the leap from 18 to 25 WPM comes from.

Callsigns. Mixed letters and digits, no redundancy to lean on, nothing to guess from. The hardest copy there is, and exactly the copy a real contact demands. Ten minutes of callsigns is worth an hour of plain text.

The deliberate overspeed week

You cannot get faster by practising at a speed you are already comfortable at. Once a month, take a week at three to five words per minute above your working speed and accept that you will copy 60 or 70 percent. Do not try to fix it. Just sit in it. When you drop back, your old speed will feel like it has slowed down, because relative to what you have been processing it has. This is not a trick; it is how every perceptual skill widens.

What the timeline honestly looks like

Fifteen to twenty minutes a day, most days, gets most people the full alphabet by ear in six to ten weeks. That is the easy part and the part everyone quotes. The stretch from 10 to 20 WPM is usually several months and can be closer to a year, and it is where the drop-out happens. From 20 to 25 is another long climb of a different character — by then the limit is your copying hand and your stock of whole-word sounds rather than your recognition of letters. Operators working above 30 WPM exist in every club and they all took years.

Progress does not arrive smoothly. It arrives in steps, after plateaus, never during them, which means a flat month is the normal shape of this and not evidence that it has stopped. The reorganisation from counting to hearing happens underneath a number that is not moving.

Things that feel like work and are not

Slowing down. Practising sending instead of receiving — sending is the half where you already know what the message says. Two-hour sessions on a Sunday instead of twenty minutes a day, when consolidation happens between sessions rather than inside them. Changing course every fortnight. And studying the chart harder, the most seductive of them and the most counterproductive — there is a whole dispatch on why the chart holds you back.

The other thing that works, and nobody wants to hear it, is getting on the air before you feel ready. A real contact at 13 WPM teaches you more about buffering under pressure than a week of clean practice files, and the exchange is short and scripted. Run it against the trainer until you know the shape, then go use it.