Article by Louis Nagel
![]()
n considering
this question we are dealing with the heart and soul of music-making.
According to Webster, improvisation is "a course pursued
in accordance with no previous plan." It is far more than
that, in my opinion, and I am not even convinced that there is
no previous plan in the course of a good improvisation. At least
in tonal music, there is a sense of key, a rhythmic organization,
a texture, even a form. These elements are controlled from deep
within a performer's musical being-the musical soul.
It is a circular process. The music comes from the soul of the performer, transmitted through the fingers to the keyboard, heard, and responded to, the response justifying the continuity of sound. Musical logic plays a part in all improvisation whether it is immediately apparent to the listener or not. The improviser is speaking the language of music-the "score" is part of the person even though it is not on a page. Many great composers were famous for their improvisations, none perhaps more noted than Bach who was able to extemporize fugues in as many as four and five voices.
The impulse that allows a performer to play what comes to mind with logic and beauty is not unlike the process we use when speaking. We do not converse from a script, but we do know what we mean when we speak. If we greet friends, the words we use may differ with each greeting but the sentiment is the same. Often a pun or other humorous statement is a form of ad-libbing. We communicate conceptually, not word-by-word. When we perform music by other composers we communicate conceptually, not note for note. We achieve an ideal performance of other composers' music if we can create a sense of "it is happening for the first time." It is no small accomplishment to achieve this sort of improvisatory feeling in a piece of music whose notes and rhythms are indicated carefully by the composer and are immutable.
It is possible to make a fermata in a Haydn sonata or a Mozart cadenza sound as if it is happening for the first time; such is the somewhat freer organization of the musical materials. The fioratura in Liszt's "Sonnet 104" or the bel canto ornamentation in the "Andante Spianato" of Chopin sound improvised if they are played with a certain amount of rhythmic freedom. Such examples are not particularly hard to play or teach. But is it possible to make an entire work sound as if it is being improvised right before the collective ears of the audience? I believe it can be done by understanding the structure of the music, and feeling where liberties might be taken and what the nature of those liberties might be. I would like to explore this idea by looking at the "Nocturne in G Minor, Op. 15, #3," by Chopin. It is an infrequently programmed composition of rare and exotic beauty, and strikes me as a perfect example of a written-in improvisation. It is in three parts, through-composed, and wonderfully pianistic; to me it feels as if the composer improvised the entire piece, recalled it perfectly and wrote it down on the spot!
Improvisation does not usually evolve from lengthy and complex themes. A simple phrase, a rhythmic cell, a particularly pungent harmony that has potential for later development, an unexpected dynamic process-these are the elements used in a successful improvisation, and Chopin uses all of these elements in this piece. The opening phrase (mm. 1-3) is mainly a fall of a fifth and a filled-in rise of an octave. In mm. 4-7, a long note is sustained in the soprano melody while the harmony underneath changes from Gm to Bb. In m. 8 the diminished 7th over the tonic pedal is gently dissonant, a hint of the harmonic intensity later to erupt. The dynamic level is piano, except for a few measures from 16-20, with subtle crescendi and diminuendi. Most telling of all is that Chopin instructs us to play "languido e rubato" a clear order to improvise! (See example below.)

Added together, these elements combine to produce music that is
not exactly simple, and at the same time they invite an improvisatory
mindset for the pianist. For starters, one should not play the
opening measures precisely in time. I would ask a student to feel
the harmonic pull from tonic Gm to relative major Bb and rush
the tempo-creating a sense of expectancy or urgency right at the
very beginning of the piece. This must feel spontaneous to the
performer, not calculated. I would ask the student to play it
one way, and no matter what happens in that performance say "good,
now do it again, but not in the same way." We would talk
about how to emphasize the opening D: is it a percussive attack
or could it also involve a tiny delay before falling to the downbeat?
Again, I would ask the student to try it both ways and feel what
seems appropriate. How much accelerando to the F and its ties
at m. 4? When the music recedes to the tonic again in mm. 12-13,
there must be some acknowledgement by delaying the inevitability
of the resolution. It should sound as if it is being discovered,
not repeated; the student must try different tempo manipulations
and not get locked into only one.
What about the forte in m. 16? This music is louder than the similar spot in measure 4. Is it possible to play this forte non-percussively but with a sense of urgency and pressure into the keys? And when the dynamic recedes from forte (which it must do, although Chopin does not tell us exactly where), can it sound inevitable and yet fresh? The performer must physically FEEL these gestures. They are part of the language of music that can only be alluded to through mere notation. The resolution back to Gm at m. 25, structurally so obvious, seems to me to be a lovely improvisatory gesture that invites much experimentation on the part of the performer.
The repetition of the opening 24 measures does not demand an intentional varying of one's performance plans, but subtle changes are inevitable and one ought not to try to repeat literally the same gestures in mm. 25-50. (See example below.)

At m. 51, the second section of this nocturne begins. The basic
waltz-like rhythm in the left hand continues, but the harmonic
language becomes much more complex. The tiny hint of dissonance
heard previously in m. 8 is going to become very important, as
is the fall of a half step heard in the left hand from Gb to F
natural. Short of offering a detailed analysis of the next thirty-seven
measures, suffice it to say that the music is largely about falling
half steps and rich unsettled harmonies progressing rather quickly.
(See example below.)

At m. 77, there is a powerful series of descending chords which
emphasize the falling half step and culminate on a low C# which
is repeated three times. Then it is spelled enharmonically as
Db and falls yet another half step to C natural. The second section
ends here.
Let me focus on one less obvious detail that contributes to the improvisatory nature of the music. The pitch B natural in mm. 54-61 is used in several totally different harmonic contexts!
I would point out the functions to a student, focusing the ear upon that note. Then I would ask for the B's to be voiced in the left hand as these several measures are played, inviting the student to hear the richness of the harmony from the standpoint of one small component part. This exercise goes far beyond mere chord analysis, and serves two important purposes: It educates the ear to the structure and sonority of the chord, and it invites the ear to become more responsive to color changes. Those B's may feel like the same key being played, but THEY ARE NOT THE SAME TONE! They are very different to a sensitive ear, and there is a delight in hearing a different color with each re-striking of the same key. The bottom line: a good improviser perceives the inherent variables in seemingly-repeated gestures.