Building a Piano Piece · Article Eleven
A piece can be played from beginning to end without the score, yet an unexpected interruption leaves the player unable to name or locate the next event. The only available response is to restart at the beginning or at an earlier familiar passage. This is commonly described as “the fingers remembering”: each action automatically cues the next, but there is no independent route into the sequence.
Motor memory is neither primitive nor wrong. Without automaticity, rapid music could not be controlled by conscious commands for each note. Auditory memory predicts upcoming sound; visual, spatial, harmonic and structural knowledge offer further support. The risk lies in staking the whole piece on one continuous chain.
Reliable memory does not remove finger memory. It creates several mutually correcting entry systems:
The hands know how to travel, the ear knows what should sound next, the mind knows where it is in the structure, visual and spatial knowledge locate the keyboard, and any one of these cues can help restart the others.
Why a motor chain fails after interruption
Continuous repetition establishes associations: one movement group triggers the next. The mechanism is efficient and vulnerable. If the only retrieval cue is “what I just did”, an interruption removes access to everything downstream.
A recent review connecting music performance anxiety with learning distinguishes purely serial memory from structurally organised performance cues. Structural, interpretive, expressive and critical technical cues can function as landmarks, allowing access to specific content rather than a mandatory restart. The review integrates different forms of evidence and cannot promise that cues eliminate stage lapses. It supports the principle that multiple locatable retrieval routes offer more possibility of recovery than one motor chain.
The classic Chaffin and Imreh case study recorded a concert pianist learning the Presto from Bach’s Italian Concerto. Formal structure organised memory, while fingering, phrasing, interpretive decisions and expressive intentions became performance cues. The study involved one expert and cannot assume that beginners remember identically. It reveals an observable expert possibility: automatic movement and conscious landmarks can coexist.
Five mutually supporting forms of memory
Motor memory
Fingers, arms and body retain practised sequences and positions. It is fast, but sensitive to interruption, a changed instrument, anxiety and an unexpected movement error.
Auditory memory
The player hears the next melody, bass or harmony before it arrives and detects deviation. Remembering only the general impression of a recording is insufficient; internal voices and exact transitions must become audible.
Visual and spatial memory
This includes score layout, notation, keyboard geography and hand positions. Page memory changes with editions, so “top left of page three” should connect with bar, harmony or structure rather than stand alone.
Conceptual and structural memory
The player knows keys, harmonic routes, thematic entries, sections, differences among similar passages and exceptional bars. Description can range from “the opening material returns in the bass” to detailed analysis. What matters is an articulable map appropriate to the learner’s knowledge.
Expressive and intentional memory
The player knows where a phrase travels, where colour changes, where time needs monitoring and which voice comes forward. This memory answers not only “what note is next?” but “what am I doing now?”
These categories are practical distinctions, not isolated neural boxes. Every category need not be equally detailed, but one serial action chain should not bear the entire load.
Begin memorising on the first day
Memorisation is often treated as a final phase: learn with the score, then remove it. A more informative practice is early, small retrieval. Read a short unit, cover or close the score, state the structure, sing the sound or play once, then immediately verify against the notation.
In a case study of an expert pianist learning Debussy’s “Clair de lune”, the performer attempted memory early, organised practice through structure and worked on retrieval cues. The published study is another expert case, not a mandatory schedule. It demonstrates that memory can be active retrieval during learning rather than an accidental product of fluency.
Early short tests expose distortions while the score is available for immediate correction. If similar passages are allowed to merge for several weeks, the erroneous motor sequence can become highly stable.
Establish landmarks that can be entered randomly
Choose major section openings, thematic returns, modulations, texture changes, points after difficulties and the coda entry. Give each landmark at least three pieces of information: opening harmony or bass, starting fingering or position, and musical intention.
Put landmarks on cards or a randomised list. Before playing, hear the opening internally, locate the hand positions, then continue to the next landmark. If the point can be reached only by rolling in from the previous section, it is not yet an independent entry.
Also start at nonstructural places such as the middle of a phrase or on a weak beat. Structural landmarks create a large map; random-bar tests examine local knowledge. They serve different functions.
Four away-from-the-keyboard tests
Verbal map
State section order, key or harmonic movement, principal difficulties and returns. Without specialist vocabulary, descriptions such as “the melody moves to the bass” or “the same material rises and returns” remain useful.
Audiation
Hear a passage without moving and follow it in the score. If internal sound is only a vague mood, isolate melody, bass or rhythm.
Silent fingering
Simulate critical fingering and locations on a table, closed lid or without fully depressing keys. This cannot reproduce real key feedback; it tests whether movement order is cued only by sound.
Written skeleton
Write bass notes, chord labels, initial pitches or diagrams at landmarks. Recopying the whole score is unnecessary. Externalise structure and the differences among similar places.
Mental-practice research asked professional pianists to imagine sound, keyboard and movement, and found effects on motor anticipation. The experiment had a small expert sample and technical material, so it does not make mental rehearsal equivalent to playing. It supports its use as a supplementary test of memory gaps.
Example: similar material in Debussy’s “Clair de lune”
Debussy’s “Clair de lune” from the Suite bergamasque contains return, transformation and changes of texture. The score is available through IMSLP. A similar sound world can encourage memory of atmosphere while exact harmony, register and entry conditions remain weak.
Mark the principal sections and returns, then attach bass, hand position and a first expressive intention to every entry. A cue might combine “this bass location—this right-hand fingering—sound emerging from a distance”. Draw entries randomly without playing the preceding music.
Compare similar passages side by side and mark the first genuinely different pitch, harmony or voice. Many lapses are not total forgetting; the mind selects the wrong member of a similar family. Contrast can encode the difference more effectively than repeating each version separately.
Simulate interruption and recovery
At a stable stage, create safe interruptions. When a timer sounds, stop, name the current section, harmony and next landmark, then resume from the next beat or landmark. A teacher or colleague can call out starting places.
Do not create startle in unsafe physical tasks. The interruption is an attentional test. Its aim is not to prove that forgetting never occurs, but to establish recovery: preserve pulse, simplify texture, jump to a landmark or re-enter rather than return automatically to bar one.
Is memorisation mandatory?
Not every piano performance must be memorised. Solo traditions, exam rules, accompaniment, chamber music, contemporary repertoire and personal goals differ. Playing from a score still requires structure, anticipation and recovery; memorisation is not automatic evidence of understanding.
If memory is chosen, its purpose should be reliability and musical freedom rather than the absence of a music desk as the only sign of maturity. Where memorisation demands substantially aggravate anxiety, decisions should respond to performance rules, teaching advice and health.
Evidence that memory is not only in the hands
The player can begin from many locations; sing or state what follows; identify differences between similar sections; continue after a pause; preserve structure when the instrument, tempo or hand condition changes; and recover after a minor slip.
Finger memory should not be eliminated. It is necessary for fluency. It simply should not carry all the risk alone.
Reliable memory is not a rope from the first note to the last. It is a map with roads, landmarks and alternative routes. Automaticity allows rapid travel; structure and intention reveal where the player is.
Primary sources and further reading
- Review of performance anxiety, retrieval practice and expert memory
- Piano performance as expert memory
- Learning “Clair de lune”: retrieval practice and expert memorisation
- Mental practice and motor anticipation in skilled pianists
Continue reading: Building a Piano Piece series page
If you would like to bring these ideas about listening, understanding, and practice to the keyboard, you might try ScoreFlow, an app I developed to make score reading and daily practice flow more naturally together.
Discover more from Geoffrey Chen
Subscribe to get the latest posts sent to your email.