Once the Notes Are Correct, How Do You Practise Melody, Balance, and Direction?

Geoffrey Chen

Building a Piano Piece · Article Nine

The pitches and rhythms are substantially correct, yet the performance resembles an unlayered inventory of notes. The melody exists on the page but does not become a continuous line above the accompaniment. Dynamic marks are observed, but phrases have no destination. Every chord is clean and the whole remains flat.

Advice at this stage often says “play with more feeling”, “bring out the melody”, or “do not be mechanical”. These phrases name an outcome, not a practice task. Expression is not emotion added outside accurate playing. On the piano it must be realised through audible relations of timing, dynamics, articulation, note overlap, balance and pedal.

The first step from accuracy towards expression is to translate abstraction into relationships that can be heard, recorded and compared:

Do not ask only how to play with more feeling. Ask which voice must remain perceptually continuous, where a phrase accumulates and releases, and which relations of key velocity and timing can make that structure audible.

Piano tone cannot be continuously reshaped after the attack

Unlike voice, strings or wind, the piano gives the player little direct means of adding energy after the hammer has struck. Control lies in the preparation and attack that determine hammer speed, in timing among notes, key release, relative intensity of subsequent events, and pedalling that changes duration and resonance.

A notated long note cannot literally crescendo after its attack. Pianists make a continuing line perceptible through the dynamics of later voices, harmonic density, timing and directed attention. Expression is therefore relational rather than an isolated property of one tone.

A modelling study of piano dynamics found that performed intensity could be predicted to some degree from pitch, metrical position, timing and phrasing. The research cannot tell an individual learner exactly how loudly to play one note. It supports the broader conclusion that dynamics are systematically connected with musical structure rather than being arbitrary emotional fluctuation.

Make a voice map

Mark the current perceptual foreground, supporting bass and controlled inner voices. Do not assume that the highest pitch is always melody or the left hand always accompaniment. Counterpoint, crossings and inner lines change roles.

Use three levels:

foreground: the line that must be followed continuously now
middle ground: harmony, response or important inner voice
background: texture that must exist without masking the others

Play foreground and bass separately and sing their direction. Block or lightly play the middle ground. Then restore all notes. The aim is not a fixed decibel difference among layers, but audibly intelligible changes of role.

Practising melody does not mean accenting every melody note

Equal emphasis on each melodic note produces a chain of bumps. A line has syntax: departure, motion, focus, delay and release. Sing or play it without accompaniment and mark a long direction rather than placing a small crescendo–diminuendo over every two beats.

When accompaniment returns, reduce it to a level that can be controlled while preserving melodic continuity. If the line can be heard only through a sudden bodily push on every important finger, examine whether the accompaniment is too loud, chords are struck as undifferentiated blocks, fingering breaks the line, or dense texture acoustically masks the melody.

Voicing a note within a chord requires differentiated key descent and timing, not a universal instruction to “put wrist weight on the fifth finger”. Exact movement depends on the hand and chord; no single posture should be imposed. Begin with the foreground note, add one background note, and gradually reconstruct the chord while recording the result.

Direction exists across several notes

A dynamic marking indicates range and relation, not an exact decibel. A crescendo should answer: where does it begin, towards which structural event does it accumulate, and what follows the arrival? An equal increase on every note without harmonic, rhythmic or phrasal purpose produces only a mechanical slope.

Research often finds expressive timing and dynamics varying together with phrase structure. A kinematic and acoustic study of Romantic piano excerpts observed systematic changes across expressive conditions and score locations. The study involved ten pianists and cannot make a visible body gesture the cause of expression. It illustrates that expression coordinates multiple parameters around structure rather than operating one volume control.

Draw a tension route through a phrase: stable, increasing, focal, departing. Realise it once through dynamics, once through timing and articulation, then combine them. The recording should decide which version makes the structure audible, not which felt most intense to play.

Use recording without confusing self-perceived and external change

The bodily experience at the keyboard may differ from balance heard at a distance. Recording temporarily changes the player into a listener and is especially useful for melody continuity, bass weight, phrase-end timing and pedal mixture.

In a study of twenty-five advanced piano students, groups that listened to Disklavier playback rated aspects of their second performances as improved, but external expert ratings found no significant intervention effect. The study does not show that recording is useless. It shows that one playback and one repeat do not guarantee an externally perceptible improvement. Feedback needs an explicit target, repeated testing and sometimes another listener.

Write the question before recording: “Does the melody remain continuous across the left-hand harmony change in phrase two?” On playback, answer only that question and choose one modification. Simultaneously judging tone, rhythm, pedal, dynamics and emotion usually leaves only “it seems better”.

Example: Mendelssohn’s Song without Words, Op.19b No.1

Mendelssohn’s Song without Words, Op.19b No.1 places a vocal line above flowing accompaniment. The score is available through IMSLP. If every accompaniment note is clear and equally weighted, the melody may lack line even when nominally louder, because the ear is continually captured by background attacks.

Play melody and bass first to establish the long direction and harmonic supports. Block the inner accompaniment into chords to hear its function. Restore the figuration, making the background light but directed rather than so faint that pulse disappears.

Record three deliberate versions: accompaniment clearly too light, clearly too heavy, and apparently balanced. Extreme contrast trains the ear more effectively than a stream of adjustments too small to identify. Listen from outside the immediate keyboard perspective and decide whether the melody can be followed continuously.

A twenty-minute sound-practice session

This is an adaptable synthesis, not a validated prescription:

  1. Two minutes: mark foreground, bass and changes of role.
  2. Three minutes: sing or play the principal line and decide one phrase direction.
  3. Three minutes: play bass and harmonic skeleton.
  4. Five minutes: restore texture one layer at a time.
  5. Three minutes: record one complete phrase without interruption.
  6. Two minutes: leave the keyboard and answer one prewritten listening question.
  7. Two minutes: modify and record again.

At the end, name the changed mechanism: lower accompaniment key velocity, smoother melodic substitution, less phrase-end delay, clearer bass, or an inner voice becoming foreground. “More feeling” alone cannot reliably reconstruct the result tomorrow.

When should expression begin?

Not after every note is perfect. From first reading, a learner should know which line matters, where harmony travels and what character the rhythm possesses. Otherwise, technical repetition establishes an unlayered default that must later be overwritten.

Expression is also not permission to hide reading or timing problems. Correctness and expression are not alternatives; they are accumulating constraints. Pitch and rhythm make the text recognisable. Voice, time, dynamics and articulation turn it into musical speech.

Once the notes are correct, the next stage is not to add emotion. It is to decide which relations must be audible, then train those relations to survive different instruments, distances and conditions of attention.

Primary sources and further reading

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.