Thoughts on Improvisation.

Improvisation – making things up on the spot. Making something out of what is to hand. Creating without prior preparation. Well, they are the dictionary definitions of “to improvise”, but to what degree is improvisation preparation free? Using what is to hand requires a prior understanding of the properties of what we have to hand. Making something up on the spot depends on our ability to think quickly, instantaneously, even. These are not abilities that we have without proper prior preparation.

For example, let's look at our ability to hold a conversation. Before we can utter a word, we have to have an internalised practical grammar generator, as well as a vocabulary of words and phrases. Then, to be able to relate to the subject of the conversation we need some knowledge of the subject. Also, we need to know where our knowledge is lacking, so we can ask questions. What at first seems to be spontaneous interaction has a great deal of learning and knowledge behind it.

It is the same in music. Making up music that is coherent, and has an internal logic, depends on our having considerable prior knowledge and skills. We need to have knowledge of the components of music – the relationships between pitches, time and loudness, as well as knowledge of how to create sets of musical events that are relevant to each other.

We can think about several categories of improvisation. First, solo playing. Here, what we are doing is constructing a piece of music that has only one creator – if I am playing, that is me. Our musical creation is therefore really only relating to itself. One section relates to other sections through time. The player has to maintain a memory of what he is doing, so that he can make the phrases within the line of music relate back to what has gone before, and project expectations as to what will happen next.

In his book, “The Analysis and Cognition of Basic Melodic Structures: The Implication-Realization Model,” (1992) Eugene Narmour proposes two universal hypotheses, that repetition implies that repetition will continue, and that change implies that change will continue. Thus, when we hear something repeated, we expect it to keep repeating; if it changes, we expect it to continue to change. How those expectations are fulfilled is what makes the music interesting. Do exactly what is expected, all the time, and the music becomes boring for a listener. Always do the unexpected, and it becomes too difficult to relate to. There has to be a balance between doing what is expected and doing what is not.

There are two other levels of expectation fulfilment: doing something only slightly different from the expectation, or doing something quite different, but related to the expectation. This gives us four levels of expectation fulfilment, which apply to composed music as much as to improvisation:-

Table 1. Four level hierarchy of Expectation Fulfilment

1) Fulfilment

Playing exactly what is expected.

2) Semi-fulfilment.

Playing something similar to, or slightly different from what is expected.

3) Substitution.

Playing something that is different from, but formally related to what is expected.

4) Violation.

Playing something without formal relation to what is expected.

The difference between the last two levels of fulfilment may depend on the listener's level of musical experience. Note that the player is a listener, too. The player uses the same hierarchy of expectancies and fulfilments as the audience to maintain his own interest in what he is playing.

Then we come to playing with other musicians. First let's look at a duo. This seems to have more in common with our earlier illustration of a conversation. Two parties both contribute to a musical creative process resulting in an interactive work where whatever one party does may or may not influence or be influenced by what the other party does. Usually, we expect both parties to a conversation to talk about a single subject at any given time. There are other unspoken rules of conversation that H. P. Grice discussed under the general title of “The Co-operative principle.” (“Logic and Conversation” (1975))

Grice drew up these “maxims of conversation”:-

Table 2. Grice's “Maxims of Conversation”


    Make your contribution as informative as required. (Don’t say too much or too little.)

    Make the strongest statement you can.


    Do not say what you believe to be false.

    Do not say that for which you lack adequate evidence.


    Be relevant. (Stay on topic.)


    Avoid obscurity of expression.

    Avoid ambiguity.

    Be brief (avoid unnecessary prolixity).

    Be orderly.

Grice also suggests that these maxims may be flouted deliberately for conversational effect. We might exaggerate for the sake of emphasis, or say the opposite of what we mean, to be sarcastic. Such flouting creates what Grice calls “conversational implicature”, where we infer more than what has been said, and which depends on our knowing what cultural conventions of conversation are in play.

I suggest that a parallel set of maxims are applicable to musical improvisation, and more work needs to be done to work them out. Because music does not have a set of specific denotive meanings, some conversational maxims cannot simply be applied across to music. For example, the two maxims of quality depend on a concept of truth that is irrelevant to music. There are no statements in music that are “true” or “false”. However, others, such as the maxims of relation or manner can be seen to be relevant with just a little tweaking.

One aspect of how these maxims of musical co-operation could work would be the level of freedom involved in the improvisation. If our musical duo is playing jazz, and the improvisation is on the harmonic framework of a standard, the maxim of “relevance” means that both players make their musical statements relate directly to that melodic, chordal and rhythmic structure. There are several levels of relevance. First, there is the melody itself, and the harmonic framework that constitutes the chord sequence. This is the tune, as written. We can play this “as written” (little or no freedom), or we can alter phrasing and note timings, or add decorative turns. We can add extensions to chords, or even substitute other chords, but, at this first, quite tight, level of freedom, we need to state the melody and chords more or less plainly, according to the genre.

Then we come to the improvisation on the structure. Here, the maxim of relevance would require us to stick to the chords, and refer to the melody. This means keeping our musical statements within the tonality of the piece, changing where it changes, and staying the same where it does not. It also means basing our phrasing on the time signature of the tune, maybe even relating our musical statements to the contour of the original melody. It means sticking to the number of bars of the tune's structure. A thirty two bar sequence stays a thirty two bar sequence.

More complex improvisations would make changes to these applications of “relevance”, as we applied the four levels of expectation fulfilment as outlined in the first table above. The idea of “flouting” would apply to playing “outside”, where a player deliberately uses non-harmonic tones to create a different kind of tonal effect. Similarly, playing in a different time signature would create unexpected rhythms.

Not only are we looking here at the relevance to the original structure, but also to the relevance of what one player is doing to what the other player is doing. Each player needs a fundamental assumption that the other player will play material that is relevant to what they are doing themselves. This is a basic level of trust between the players. Now, there will be things played that do not seem quite so relevant. Sometimes it will be a mistake, or misjudgement. Other times it will be a deliberate flouting of the maxim of relevance that will provide interest in the music.

When the improvisations are based on composed frameworks, such a standard or a blues, it is quite clear to the experienced jazzer how to maintain levels of relevance. In our duo playing a jazz standard, both musicians are relating directly to that framework as well as to each other's playing. Relating back to the overall structure means that there is already common ground between the players. Where there is no pre-determined structure, that is, in “free” improvisation, the common ground has to be discovered and determined in the process of playing. Making relevant musical statements in this context requires a different approach. We have to pull each other up by the bootstraps. We have to commit ourselves to making a musical statement, without being totally certain that what we do will be relevant to what the other is doing.

Of course, when the musicians are experienced players, they draw on their internal stores of knowledge, as both a source of material, and a store of methods of playing. If the players know each other well, those internal stores will also contain knowledge about the other's playing approach. We all have favourite phrases and ideas. Some are technique based, others theoretical. These can provide some common ground between the players. The beginnings of one player's familiar musical units will give the other player cues to imply what might happen next. There will be more that each player can predict about the other's playing. Of course, the player may or may not follow through with those implications.

There is another level still to this two player interaction, and that is where players avoid direct reference to what the other is doing, and wait to see how their separately generated material interacts without deliberately making it relevant. This can inspire all manner of new ideas. The relevance comes from the perceptual process of trying to make sense of the combinations of sounds presented to the listener.

When we start to think about groups with more numbers, the set of possible interactions between musicians grows. In our duo, we have Player A and Player B. We can have A solo, B solo, A taking lead and B supporting, B leading and A supporting, both leading, or both playing less intensely. Without considering these weightings, we can trim this to three groupings. A, B and AB.

If we add a third party, we have A, B and C all solo, AB, AC, BC all duos, and A+B+C as a trio. The weightings amongst the members can also vary as with the duo. That is seven possible musical combinations, even without the weightings. With four members, this set of sub-groups rises to fifteen possible units, and with a fifth member, you get thirty-one. Again these groups are subject to a variety of weightings. Thus the complexity of relationships between the musicians rises exponentially as you add members. This may explain why improvising groups tend to be small – it is easier to keep tabs on fewer interactions, both for the performers and the listeners. There are limits to how much information we can process at a time.

Another way that interactions within a larger group can be organised is by the players choosing different roles for themselves. In our duo example, if one musician plays a single note instrument, or frontline, and the other a chord instrument, it is typical that the chord player will act as an accompanist to the frontline player. They will provide harmony and bass to the single line's melodies. With more instruments, this division of labour gets more distributed, so a typical group would have frontline, chords, bass and drums, with each working within a specified musical dimension, be it melody, harmony, root movement or rhythms. This is not to say that each instrument is entirely restricted to their most obvious role or function, but that such self-regulation does happen, and helps organise the music.

The concept of organisation is the key to understanding what we are doing in making music. We are organising sound. A composer creates a program, or recipe for how other people should organise sound in a given piece. The classical musician brings that recipe into existence, and actually organises the sound using instruments. The jazzer takes a piece that is already written, and improvises an organised commentary on the structure of that piece, or even reconstructs it into another form, and the free improviser creates spontaneously a musical recipe, organises the sound, makes commentaries on that organised sound, and takes it to another place, all in the moment.