Greg Segal with Hyam Sosnow

 revised 3/03

This is an overview of the ILCS, or Improvisation Level Classification System. The purpose of this system is to quickly show the amount or type of improvisation in a given piece of music, indicated by two letters and a number (IL1-IL6).


I developed this system while working on the liner notes for the recent Jugalbandi recordings. I found myself becoming frustrated with how difficult it was to explain the facts of improvisation to the general public. Improvisation by its nature remains an amorphous thing in most people's minds. Some, with little knowledge of the art form, often get confused and think you're pulling their leg about having "made it all up on the spot" if they discover that any aspect was pre-planned; others will think it's normal to pre-plan a basic structure for an improvised piece and never consider that you might literally have pulled every drop of it out of thin air. It became obvious that some simple means of explanation was needed for how much of the music was improvised, the method involved, etc. I thought letters and numbers, related back to simple definitions, would serve as a quickly identifiable rating which could then be listed along with title, author information, recording dates, etc.


This system is the product of many years of experience with the full range of musical improvisation- from total to none- and is meant to cover those extremes and as many shades in between as I could hit without complicating things too much. Hyam Sosnow and I refined my original draft in an attempt to take things down to the essentials; the categories here are as basic as we could make them. Making those divisions was difficult and these choices will no doubt cause some controversy.


Hyam's input was responsible for very definite, specific changes in the system from how it was originally conceived, and we collaborated on all changes after the initial proposal.


Though created for the Jugalbandi project, the ILCS was intended from the first for general use. It is my hope that others will find this system handy either for classifying their own work, or as a tool to explain the various levels of improvisation to curious parties, students, etc.- GS


Note, 3/03: The ILCS (Improvisation Level Classification System) has undergone a change: for practical use, we have decided that the area between levels 3 and 4 was a bit too broad, and needed a further classification level in order to be properly descriptive while still encompassing both the qualitative and quantitative aspects* of the system. So there are now 6 levels.




IL1: totally improvised- no pre-communication whatsoever

 Comment: This is pretty much just what you'd think it is. Of course there is communication as the piece is happening- non-verbal, but listening to each other and responding. There might be gesturing at times with facial expressions or body language. The music has most likely erupted spontaneously, with one person playing and the other player(s) joining in.


IL2: a) a few words or song title mentioned prior to starting ("let's play some funk", or "let's do something that fits 'Moving Towards Kyoto'"), etc.; b) one player begins with an idea the others haven't heard

Comment: The very lowest limit of prestructuring, based around the scantest of ideas. A style of music or some basic element is suggested; or a title is given as something to base musical impressions on; or one player has a preconception that the other player(s) will improvise to without having any information prior to the start of playing.


IL3: Piece based around a newly composed riff, idea or chord structure- riff/idea/chords barely formed (and may remain so), arrangement and any solos improvised

Comment: The next step: an identifiable, repeatable riff or chord structure, usually of short duration and maleable form (i.e., identifiable but not completely fixed, a flexible motif). Here we are concerned specifically with a motif freshly invented and no actual surrounding structure beyond that. Usually such a piece will be worked into a song- changing it to IL4, IL5 or even an IL6- but in the early stages here referred to, it remains little more than a sketch whose form is still being defined. Once the motif is defined enough for exact replication, with set borders between it and improvised sections, it will be classed as IL4 or IL5 (with certain rare exceptions).


IL4: Mostly improvised piece based on a known structure/riff.

Comment: The structural element- what little there may be of it- is set and repeated with each version, but the overwhelming majority of the piece is improvised. A good example of this would be "Uncle Sun": the head, all 90 seconds or so, is very clearly defined, identifiable, and repeated in all versions- composed, in other words! There usually follows 20 to 30 minutes of totally improvised music. Conceivably though, this could be applied to any song: take a few readily identifiable notes- not the full structure- and have that as your only touchstone with the rest being improvised, and you have an IL4.


IL5: Composed piece w/ improvised section (s)- usually, head/improv/head, but variations are endless. Most jazz combos throughout the past 75 years or so improvise at this level. At one extreme, only the solos are improvised; at another, the arrangement goes completely fluid.

Comment: This is the improvisation that most people are familiar with, not just from jazz but from blues and psychedelic music. In practical use, this refers to a piece that's somewhat/slightly improvised and is based on a known structure/riff.


IL6: Fully composed piece

Comment: Why? You get it.


In practical application, pieces change classifications very easily. Something that begins as an IL1 can end up becoming anything up the scale right up to IL6 (to that extreme only if the musicians are really, really patient). Something that starts off an IL5 or 6 can become an IL4 if the mood strikes the players to take some small kernel of something and expand it into the stratosphere.

Obviously this system doesn't cover all the bases; it can't. But with enough examination even difficult pieces can usually be classified. 

For example, how would you classify a piece where there is a definite rhythm and definite chord shape, but no definite key or progression? Hyam and I were faced with that when attempting to classify "Reciprocal Demonology", off Jugalbandi 2000's "The View Is Better From The Top Of The Food Chain". After much discussion, we were able to classify it as IL3- but only after having to redefine the classification.

 Or take the case of "Ivy Mike". This was a piece done with Dog Neutral, Jugalbandi's predecessor. (A version was done by Jugalbandi for the 2000 sessions but was ultimately left out to avoid a 4th disc.) The structure consisted of a 5 note riff, repeated at will throughout an improvised arrangement. The riff had fixed intervals and rhythm but no fixed key; and of course, during the improvising, variations on its intervals and rhythm were fair game. The keyboard and guitar started off from a 4-count, playing the riff in unison, but the starting note was purposely not discussed beforehand. After 2 repetitions the keyboard and guitar would change keys, destination also not discussed. The idea was for us to be as surprised as the audience at the interval. Sometimes it was pleasant and other times fantastically atonal (which could be pleasant in its own way). After this, each player slid away from the meager structure provided by the riff, and keyboard, guitar and bass could go anywhere only to bring the riff to the surface again at any point. Would it be answered? If so, how? Or would it be ignored? We never knew where it was going and neither did the audience. And yet because of the simplicity of the motif this was an easily identifiable piece. Now then- how would you classify that? We thought it would make a good test for the system. It was a tough one. Ultimately, Hyam and I decided that it had to be IL3 as well.

Another type of problem was posed by an older Jugalbandi piece, "Uncle Sun". The structure: very short head/lengthy improv/?. Sometimes it resolves to the head again (or a variation on it), sometimes it doesn't. The head is composed, but it is followed by a sea of improv. Hyam initially wanted to class this as IL1, because of the sheer weight of IL1 material in each version; in my opinion, according to the old classification this was clearly IL4, despite how short the composed section is. He felt that classing it as IL4 would be misleading to the audience, and just poorly descriptive; I felt that the whole story would have to be gleaned from liner notes, but that shouldn't change the rules for classification. We finally arrrived at a compromise by adding another level to the system. At times like this, you want to abandon all systems and let people figure it out on their own! But the will to provide a useful tool was stronger than the flight response, so the system was revised.  

Lastly, there is the problem of applying this system to solo improvisation. This is less of a problem than it seems; certain aspects of the system (references to communication between players) simply become inapplicable. IL1 becomes anything improvised without a preconception; IL2 becomes anything improvised with only a basic feel/style in mind, or based on impressions of a title; IL3 is an improvisation based on a very fresh and new riff the soloist has just come up with; IL4, 5 and 6 are basically unchanged.


Qualitative & Quantitative: When starting to classify the pieces for J99, Hyam and I realized we were viewing the system from 2 very different perspectives. I was thinking of it as primarily qualitative, he as quantitative. Translation: to me it was more about the stages of creative development, and for him it was about the amount of time actually spent improvising during a given piece. After discussing this, it was decided that in order to encompass both we needed to add a level, which is how you now see it. We hope this will not frustrate people into indifference, and that they will find the change useful.