How Do You Compare Interpretations Without Merely Copying a Favourite Pianist?

Building a Piano Piece · Season Three: From Score to Interpretation: Building a Performance of Your Own · Article Eleven

Listening to different performances readily becomes one of two activities: ranking famous pianists, or collecting effects from a favourite recording. Ranking produces preference without necessarily producing learning. Copying can transform one’s sound quickly, but it cuts another performer’s conclusion away from their tempo, instrument, body and whole design.

Useful comparison does not begin with “Who is best?” It asks: what decisions did these performers make about the same textual relationship? How do their decisions work together? If one is transferred into my performance, does it still work?

Compare parameter combinations, not names

A performance can be described in at least six dimensions: baseline tempo and sectional proportion; local timing; dynamics and levels of climax; articulation and pedal; voice focus; repeats, ornaments and textual choice. Record date, instrument, edition and media conditions as well, because not every sonic difference is interpretive.

Avoid labels such as “poetic”, “objective”, “Romantic” and “faithful” until they have been unpacked. Replace “very free” with observations: phrase endings broaden substantially, melody often follows accompaniment, and the sectional baseline changes. Replace “clear” with: limited pedal, distinct voice onsets and stable pulse make contrapuntal entries traceable. Precise description reduces the temptation to imitate a personality impression.

Performance research also finds that timing, dynamics, articulation, pedal and movement form individual combinations; performers do not evaluate their own and others’ work identically. Studies of self- and other-evaluation and individuality in pianistic performance establish preference and difference as real objects. They do not turn liking into correctness.

Make your own baseline first

Before listening extensively, build a provisional version from a reliable score. Record a tempo range, principal phrases, harmonic arrivals, voice plan and unresolved questions, then make a complete recording. The baseline need not be mature. It makes later change visible.

If one familiar recording loops from the first day, its movement, timing and dynamics can be absorbed without judgement. Once the model is removed, the student remembers effects but cannot explain functions. With a baseline, external versions become evidence and alternative hypotheses rather than a default answer.

This does not require weeks of isolation. Even a short, honestly provisional baseline is enough. Its purpose is to reveal the assumptions you already carry before prestigious names add their authority.

A three-pass comparison

Pass one: blind-listen to the whole

Choose three recordings and label them A, B and C without performer names. Listen to a complete section at roughly comparable playback levels. Do not read reviews simultaneously. Record broad structural impressions: where does direction form or disappear, how are sections proportioned, and which voice can be followed?

Blind listening cannot eliminate prior knowledge, and a style may reveal a player. It does reduce the immediate substitution of name, prize and reputation for hearing.

Pass two: compare one question at a time

Select three to five short locations. Compare only one parameter on each pass. First record duration and rate change from section opening to the initial arrival. Next listen only to bass–melody balance. Then compare release and pedal at the cadence. Do not try to evaluate emotion, structure, colour and historical truth within five seconds.

Use three columns: observation—effect—interpretation. Observation contains an audible or measurable fact. Effect states its consequence for direction or clarity. Interpretation proposes a reason and remains explicitly inferential. For example: “B enlarges time before the arrival; the cadential level becomes more prominent; the performer may understand it as the boundary of the section.”

Pass three: restore the whole performance

A local effect must survive whole-work proportion. One pianist’s first climax may be large because a still larger dynamic field follows. Copying only that point leaves one’s own later music nowhere to go. Restore at least a full section before and after, asking how decisions support one another.

This pass prevents a common error: assembling an interpretation from isolated favourite moments. A convincing performance is a network of proportions, not a playlist of impressive details.

Do not confuse recording loudness with pianistic dynamics

Mastering, microphone distance, compression, streaming normalisation and playback equipment alter loudness and colour. Be cautious when comparing absolute dynamic range. Relative levels, voice relations and timing within one recording are more dependable. A version that is louder in headphones does not prove that the pianist produced greater live dynamic range.

Tempo comparison also requires separating total duration from local proportion. Omitted repeats change duration; editions, ornaments and pauses may differ. Establish comparable text before measuring comparable sections. A time figure without textual equivalence can create false precision.

Turn a heard choice into an experiment

Extract no more than one testable proposition from each recording. For example: “Reducing overall loudness while clarifying the inner voice at the second thematic entry may make it function as response.” Do not copy exact milliseconds or output level. From the baseline, make three versions: original plan, low-amplitude introduction and conspicuous introduction.

Listen first for whether the original problem is solved. Then inspect costs. Has pulse been damaged? Is technique less secure? Has later dynamic space been consumed? Does the solution work only on this piano? A borrowed choice that cannot be explained from textual relation or reproduced in changing conditions should not enter the final design.

Use an interpretation-comparison table

The last row is essential. Every solution may sacrifice something. A very slow pace expands detail but can weaken long line. Strong rubato can clarify syntax while reducing dance impulse. Sharp voice separation can diminish integration. Comparison does not discover a performance without cost. It reveals what each pianist prioritises.

Keep observation and value judgement separate long enough for disagreement to teach something. A performance you dislike may expose a textual possibility; a beloved one may rely on a proportion that does not fit your own account.

Let the work generate the question, not the size of the collection

Owning twenty versions is not necessarily more instructive than comparing three closely. Begin from a question raised by score and baseline: is this cadence a local breath or a sectional boundary; which voice gives the second thematic entry a changed meaning? Select recordings that create a genuine contrast, not only the most famous or mutually similar versions.

After comparison, write a provisional judgement and state what evidence would change it. If A’s treatment works only at a slower baseline and damages the line at your tempo, record that condition rather than declaring A wrong. Comparison then investigates conditions within the work instead of ranking taste.

A listener can also perform a descriptive task: mark perceived arrivals, changes of leading voice and places where direction is lost, without being given performer names. Listener response is not a final court, but it tests whether your language corresponds to audible result.

Know when to stop listening

If only one recording remains in the inner ear, if each practice session begins by checking how another pianist plays, or if every version converges on one model, suspend external listening for a week or two. Use the score, analytic notes and your recordings. Later, cold-listen and ask whether borrowed treatments have become decisions you can independently explain and reconstruct.

Complete refusal of recordings is not independence. A performer still inherits teachers, editions and current convention without awareness. Comparing conflicting versions can make those habits visible.

The goal is not to “combine the best of everyone”, a phrase that often produces collage. Differences reveal a problem space: tempo, voice and time can be organised through these competing relationships. Keep only what your own chain of evidence can support.

The value of comparing recordings is not finding an answer to copy. It is making previously invisible performance decisions describable, testable and rejectable.

Primary sources and further reading

• Self- and other-evaluation in piano performance
• Timing, articulation, dynamics, pedal and individuality in piano performance
• Structural and narrative imagery in expressive piano performance
• Personal taste and modelling of expressive piano dynamics
• CHARM research on recorded-performance analysis

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.