How I memorize now: a spoken index

What follows is the distilled version of what I have understood about memorizing, learned the hard way. And which nobody teaches.

I can say that with some authority: I have been a conservatory teacher, and before that the student of a good number of masters, some of them very distinguished. Not one of them ever actually taught me how to memorize. It is taken for granted that you work it out on your own, by friction, and whoever manages it is considered gifted rather than informed.

Six months ago I asked myself: is it really possible that no realistic method exists? Not a list of good intentions — an actual procedure.

This is the sequel to the diagnosis, and it is exactly that: a procedure. Put another way, a recipe. An algorithm for memorizing.

The diagnosis was this: when a piece falls apart after two weeks, I have not lost the piece. I have lost the motor channel — the only one of the four that wears out quickly, and for years the only one I had built the performance on.

From which follows something that sounds absurd until you write it down:

Reviewing by playing only reinforces the channel that gets lost.

It is not that reviewing by playing is useless. It is that it repairs the part that will break anyway, and leaves untouched the part that would not have broken. Every hour spent that way is an hour spent refilling a warehouse with a hole in it.

What you need is something that survives the interruption, and that on your return is able to hook the hands back up. I call it the index.

What the index is

The index is a list of spoken sentences, one for each section of the piece. Real sentences — the ones you would say to a colleague on the phone if they asked you how that page is built.

It is not an analysis. It has nothing of the written report about it, it does not use jargon for the pleasure of using it, and it does not have to be complete. It has to be sayable: if a section will not fit into one sentence, the section is too long and must be split. The sentence limit is not an aesthetic rule, it is the measuring device for granularity.

And above all it is not a hands-review in disguise. It is a separate object, in a different language, and that is precisely why it crosses time differently.

Writing it: name the exception, not the rule

This is the rule that changed everything, and to get to it I had to throw away two badly written indexes.

A regular Alberti bass does not get named. The hands do it by themselves, it is obvious material, and writing it into the index means using space to tell your fingers something they already know.

An Alberti bass with a sixth in place of the third does get named, in writing. Because there the hand would go to the wrong place on its own: the automatism exists, and it points in the wrong direction.

Put differently: you are not describing the score. You are making the list of exceptions to the obvious. Everything predictable is executed by the motor channel with no need for addresses. The index is only needed where the predictable is betrayed.

Two practical consequences follow.

Compress by voices, not by chords. If a page has three superimposed layers, the instinct is to feel them as triple the load. They are not: they are three lines of text, and each one stays itself for a good many bars. The bass holds, the inner voice descends by step, the melody makes its shape. Three sentences cover twenty bars. Describing vertically, chord by chord, those twenty bars would become twenty list entries and the index would already be unusable.

The index has no constant step. Where the material is obvious, eight bars can be worth one word. Where it surprises, two bars are worth a whole sentence. An index with a regular step is the sign that you wrote it looking at the score instead of looking at where you would go wrong.

Calibrating it: the empty telling

Here is the part I consider the real find, and it is not a memory technique: it is a way of measuring.

The hard question is not what to put in the index. It is how fine-grained it has to be. Too coarse and it hooks nothing; too fine and it becomes a second score, which nobody reviews because it costs as much as playing.

And the right granularity is not chosen at a desk: it is measured. This is the measuring device.

With the score closed and no instrument, tell the piece out loud, section by section. Then see which of the two symptoms shows up.

The words run but you do not hear the music. You are saying empty labels — "then the second part", "then the recapitulation" — and while you say them nothing happens inside. The index is too coarse: the exception that made that moment that moment is missing. Add it.

Telling takes longer than playing. The index is too fine: you are saying things the hands already know. Take them out.

There is only one operating procedure, and it runs in the direction of subtraction: start detailed and remove something at every review. When removing one thing makes the passage go mute — you say it and nothing happens any more — that one was load-bearing. Put it back. You are done.

Reduce until the telling goes dark. The right point is the last one before the dark.

Calibrating costs two reviews. Trying to guess it by writing the perfect index first time round costs weeks, and does not work, because which exception is load-bearing is something the piece knows, not you.

Maintaining it

Maintenance is the cheapest part of all, and it is what makes everything else worth doing.

With the score closed and no instrument, tell the piece. Two or three minutes. On a train, on a plane, walking. It is not practice, it is stocktaking: you are not improving anything, you are checking what is there. Wherever you stop, that is a hole in the index — and you see it in thirty seconds instead of halfway through a performance in front of someone.

Review at increasing intervals, never at zero. One day, then three, then a week, then two, then a month. What consolidates is not the review itself: it is the review done when you were about to forget. An easy review, done immediately, teaches memory almost nothing.

If it comes out smooth, lengthen the interval. If you get stuck, go back one step — on that piece only, not on the whole repertoire.

A mature index, reviewed once a month or a little less often, crosses the summer on its own. That is the point at which the holidays stop being a threat.

Coming back after a break

This is the procedure, and the order matters. Reversing it is exactly what makes the return frustrating.

1. The index out loud, no instrument. Ten minutes. You tell the whole piece and mark where you stop. It serves to separate two things that look identical from the keyboard: the sections that are real holes and the ones that are just cold hands. You measure before you fix, instead of fixing everything at random.

2. The entrances, at the instrument. Three minutes. For each section: say the sentence, then play the first two bars, and stop. Next section. The order inside the single gesture is the important part — word first, sound second — because it is the word that has to pull the hand along, and if you play first you are training the motor chain all over again.

You are not rehearsing the piece. You are reconnecting addresses. If you feel the urge to carry on past the two bars, that address already works: move on. Wanting to continue is the signal to go forward, not to stay there.

3. The score, only where the index got stuck. That is the small part, and now you know exactly where it is.

The three lines I keep in mind

Each channel is repaired in its own language. The index is reviewed by saying it, motor memory is repaired by playing it, looking at the score repairs only your eyes. Repairing one and hoping the others sort themselves out is the most elegant way to lose a month.

Name the exception, not the rule.

What survives the summer is not the hands: it is the list of decisions.