Practice that actually works — instead of “once more from the top”

Playing through is not practice

The commonest form of practice is also the least effective: play the piece from beginning to end, stumble at the hard parts, carry on, repeat the whole thing. What already works gets consolidated further, and what does not work gets consolidated too — as a stumble.

A piece typically has three to five places where it catches. Those places may be ten per cent of the playing time but account for ninety per cent of the progress available. Work on them in isolation and you will get further in half an hour than in five run-throughs.

The section size is smaller than most people assume: not the page, not the system, but two to four bars plus one bar either side. The connecting bars belong to it, because otherwise the passage works alone and collapses again in context. That transition is often the actual problem.

Error-free repetition

Repetition consolidates what actually happened, not what was intended. A movement that went slightly wrong ten times and succeeded on the eleventh was practised wrong ten times and right once. That is why slow practice is not a concession to limited ability but the only way to repeat the correct movement at all.

From this follows a hard rule that works surprisingly well: count only the error-free repetitions. On a mistake, the count returns to zero. Five clean repetitions in a row is the day’s target for a passage — and anyone noticing they never get past two has the diagnosis in hand: the tempo is too high or the section too long.

Equally important is knowing before each repetition what you are attending to. A run-through without a point of attention is idling. The sentence can be mundane — “this time the thumb stays down” — but it has to be fixed beforehand, not afterwards.

For the same reason self-observation pays: what exactly goes wrong? “The passage doesn’t work” is not a diagnosis. It only becomes usable as “the fourth finger arrives late” or “the position shift in bar three slips”. Only that precision allows targeted work instead of repetition — and usually finding out takes two minutes and saves half an hour.

A phone recording helps here. What you hear while playing is not what actually sounds; attention is on the hands, not the ears. Thirty seconds of recording reveal unevenness of touch, tempo drift and notes held too short, all of which are reliably missed live.

Spacing beats volume

The same amount of practice time works differently depending on how it is distributed. Four sessions of fifteen minutes across four days achieve more than one hour on a single day. The reason is that memory consolidates on retrieval rather than on encoding — and you can only retrieve what has faded in between.

In practice that means twenty minutes daily rather than two hours at the weekend. And within a session, three passages in rotation rather than one passage straight through. It feels worse, because each switch loses a little ground — and that very resistance is the effect.

Sleep plays a special role. What was practised in the evening is often measurably better the next morning without any practice in between. It is therefore worth playing a difficult passage slowly and cleanly once more in the evening — and then stopping.

The last repetition of the day is the one that sticks most strongly anyway. Putting a piece away after a failed run-through is the worst possible way to end a session.

From practice

A practice log of three lines helps, no more: what was today’s goal, what tempo was reached, what is next tomorrow. It takes thirty seconds and changes practice more than any method, because it forces a decision before playing.

With children it is better to replace time with numbers. “Practise for twenty minutes” produces twenty minutes of presence at the instrument. “Play these four bars five times without a mistake” produces four bars that are secure — and often takes only six minutes. Tasks with a checkable ending work; tasks with a duration do not.

One misunderstanding comes up regularly: people believe a passage is “learned” once it has succeeded once. Learned means it succeeds cold, without a run-up, first thing the next day. That test is incorruptible, takes ten seconds and spares you the surprise in performance.

And a final point that gets overlooked: breaks are part of practice. After roughly twenty to thirty minutes of concentrated work, accuracy measurably drops. Carry on then and you consolidate inaccuracies. Five minutes on your feet and the next block is productive again.

Common mistakes

Always starting from the top. The beginning of a piece thereby gets played ten times as often as the ending. In many students the first page is brilliant and the last unfinished — a pure distribution problem. Remedy: start with the last section and work backwards.

Practising only what is enjoyable. What sounds good sounds good because you can already do it. Practice time belongs to what does not work — with one deliberate run-through at the end as a reward.

Taking tempo as the measure of progress. Faster is not better if accuracy falls. The more honest measure is the number of error-free repetitions at an unchanged tempo.

Too much at once. Three pieces in progress is sensible, seven is not. With too many building sites none gets enough repetitions to consolidate, and the impression of standing still is then not an impression but accurate.