Monday, August 2, 2021

Some varieties of improvisation, from New Orleans, through bebop, to Cage, and to K-Pop

Nahre Sol recently posted a video about improvisation, a subject she’s been posting about for several years: Why is Improvisation SO DIFFICULT for Classical Musicians?

Independently of what she said, there are two obvious answers to her question: 1) it isn’t part of their training and so is utterly foreign, and 2) it is to some extent surrounded by a mystique that magnifies the difficulties imposed by lack of training. It’s as though improvisation is this special thing where you have to learn the secret handshake and go through the initiation ritual before you’re allowed to even begin doing it.

I don’t know where the mystique came from – it was there when I was young in the 1960s – but I suspect it may be a reflex of the process whereby improvisation disappeared from the classical tradition. For, as Sol points out, improvisation was important in the Baroque era and into the 19th century. Thus Mozart was renowned for his improvisation and I recall reading somewhere that someone had even suggested that some of Beethoven’s piano sonatas are derived from improvisations. But by the late 19th century improvisation had all but disappeared – though as Sol notes, it continues among organists (many of whom have to improvise interludes and such for church services).

Why did it disappear? I’m not sure we really know. One thing, as Sol points out, is that it is pretty hard to imagine how one could create a piece of music where all the musicians in a symphony orchestra (60 to a 100 or more musicians) are busy making stuff up. How do they coordinate what they’re going so that the resulting performance is a coherent musical experience?

Roles in an ensemble

But the problem exists for small ensembles as well. A classical string quartet is very different from a small jazz group, quartet, quintet, or more or two more, or, for that matter, a rock group or a bluegrass group. In a jazz group there is a clear distinction between the rhythm section – generally piano (or sometimes guitar), bass, and drums – and the front line – generally one or two horns, trumpet and/or saxophone, and/or trombone. The rhythm section is responsible for maintaining the basic form of the piece, harmonic structure and rhythmic pulse, while the front line musicians are free to improvise melodic lines. The rhythm section players are thus more constrained than the front line soloists.

This distinction between front line and rhythm section doesn’t exist for a string quartet (two violins, viola, and cello) or for other chamber ensembles. While the players will take different roles in the course of a piece, sometimes more foreground, sometimes more background, there is no one player who can outline and anchor the harmonic structure the way a pianist in a jazz group can. While strings can play double and even triple and quadruple stops, I suspect you’d have some very annoyed cellists if they were forced to play triple and quadruple stops all the way through a piece, nor would it sound very good. Triple stops and even quadruple stops can be used here and there, but employing them on a continuous basis is something reserved to Paganini-style stunt pieces. Consequently all the players in a string quartet must play a role in maintaining the harmonic structure of the piece. And while one player will be assigned the melodic line (often, but not always the first violin), that is quite different from the relatively free improvisation of a horn soloist in a small group jazz.

Returning to jazz, we need to recognize that the scope of improvisation itself has changed from one style to the next. In the earliest form, New Orleans style traditional jazz (later revived as Dixieland), you didn’t have solos. While the group did recognize the distinction between front line (typically clarinet, trumpet, and trombone) and rhythm section (generally tuba rather than string bass, banjo and/or piano, and drums), the front line musicians did not play solos. One instrument, often the trumpet, would play the melody at the beginning and end while the other two instruments would play stylized counter melodies. Between the beginning and the end the front liners engaged in group improvisation (with the rhythm section), highly stylized and constrained counterpoint lines in relation to one another. The solo developed from something called a ‘break’: the rhythm section would stop playing for a short time, say four bars, and one of the front liners would improvise a line to fill the space. By extending the break and bringing the rhythm section in for background you have a solo. Sidney Bechet and Louis Armstrong appear to have been the prime movers here.

Starting in the mid-to late 1920s, traditional jazz gave way to the big bands of the swing era. These bands consisted of a rhythm section (piano, and sometimes guitar, bass, and drums) and trumpet, saxophone, and trombone sections, each with three to four members. The big band was a composer’s and arranger’s medium. Star soloists would be given a chorus, maybe two, in arrangements, but they weren’t expected to carry a performance. Nor is it clear to me just how much improvisation they did, as opposed to un-notated composition – something I’ll take up in the next section.

It really isn’t until we come to the bebop era, starting the early to mid 1940s, that the improvised solo comes to dominate jazz. Bebop was a small group form, typically quartets and quintets. The arrangements were simple:

1. a brief introduction from the rhythm section,
2. the front line plays the head (i.e. melody), often in unison, but sometimes in harmony,
3. solos from the front line, often the pianist would take a solo, sometimes the bassist, and often the solo section would end with the front-line trading four and eight-bar phrases with the drummer, and finally,
4. repeat the head.

The soloists would each have the same number of choruses, though the slow tempo of ballads often required different arrangements, and each was free to do as they pleased within the general constraints of the form. Later soloists were not required to use material provided by an earlier soloist, nor was there any requirement to somehow create a form that somehow went beyond the theme-and-variations format.

In the late 1950s and early 1960s various musicians experimented with freer formats, either simplifying to simple modal forms or loosening harmonic constraints. In time avant-garde groups would all but eliminate the difference between front line and the rhythm section. At that point any player was free to take any role in the evolving performance.

Let’s move on.

Un-notated composition

Back in 2018 Ethan Iverson interviewed Carla Bley, the (stone cold genius) jazz composer. She talked of having met Louis Armstrong:

I actually met Louis Armstrong! That was really amazing. It was in the subway and he was with his wife, they were going home to Queens. I just practically fell over. I couldn’t believe it. I thought that he lived on a cloud somewhere, in a palace, not in Queens. He was a very nice person, nice to me, anyway. I, of course, revere him for the same reason as I revere Count Basie. Everything is perfect. Every note, all the phrasing, is perfect. He worked on it. He told me he didn’t have that ethic that the rest of the musicians have, always coming up with something new, never repeating themselves. He just worked on a solo until he had it, and that was really freeing for a person like me, used to writing everything down. I couldn’t really enjoy improvising. When it turns out great, I am thrilled. But on the way to sounding even good, I’m in a terror. I just cannot relax.

Let’s be clear about that. When you listen to a Louis Armstrong solo, whether on a sound recording or a YouTube clip, most likely you are not hearing something he improvised on the spot. Rather, you are hearing something he’d worked over until he was satisfied. He then played that on all further occasions. What you are hearing is a composition which has not been notated. It will vary in minor ways from performance to performance, but it is always recognizably the same melodic line.

Consider this 1928 recording of “West End Blues,” in particular, listen to the opening cadenza:

Here’s a live performance from 1955 (with a helpful transcription):

After almost three decades he’s still playing the same cadenza, and his solo chorus at the end is the same as well.

Was he improvising or not? It depends. I’d be very surprised if he’d lost the ability to make up interesting melodies on the spot. He just was not interested in doing that once he’d arrived at an approach he liked. A saxophonist I worked with in a rhythm and blues band some years ago was the same way. Once he’d arrived at a solo he liked, he stuck with it.

In Armstrong’s case that solo has become part of the jazz canon, and, as such, part of the apparatus of jazz education that has been build up over the last five or six decades. Thus, last year something called the “West End Blues Challenge” swept through jazz trumpet circles on YouTube. Trumpeter after trumpeter played that cadenza. I don’t know how many versions there are; I’ve only listened to a half-dozen or so. All of them are technically competent, and musical as well. And they’re different in small ways. How could they not be? They’re different musicians, no?

I would imagine most of them were working from one of the transcriptions floating around. That is, like a classical pianist learning a Beethoven sonata, they worked from a score. And like the pianist, they also listened to recordings. I don’t know when this cadenza was first transcribed, but the first transcription I am aware of is the one David Baker published in Downbeat magazine a month or two after Armstrong’s death in the summer of 1971. That’s the version I’ve been practicing ever since. I recorded an a capella version of “West End Blues” on YouTube back in the early days, 2006, and opened with that cadenza (excuse the lousy sound quality). By that time I’d been practicing from Armstrong transcriptions since the mid-1960s. Judging from the copyright dates on those transcriptions, people were transcribing and publishing transcriptions within weeks after Armstrong’s records were released (these are from the so-called Hot Five and Hot Seven recordings of the late 1920s).

Jazz musicians operate within a large and well-developed culture of improvisation that no longer exists for classical musicians. You listen to recordings – every jazz biography I’ve read talks of listening and playing from recordings – learn solos from recordings, make your own transcriptions, learn some “theory” here and there, and practice patterns, the cycle of fifths, ii V I, etc. You do this year after year and eventually you get good, sometimes really good.

But this is a digression, let’s get back to un-notated composition. How common is it in jazz?

I don’t think we know. Back in 1973 Joachim Berendt suggested, on the basis of recordings, that un-noted composition was common in the big band era (The Jazz Book, pp. 193, 132-33) and Guntar Schuller made similar observations (Swing, pp. 165, 173-176, 445, 481). Thus if you listen to various versions of “Take the A Train” that Duke Ellington recorded, you’ll hear Cootie Williams playing pretty much the same trumpet solo. Was he doing what Armstrong had done, playing the solo he liked, or was he simply playing what the fans obligated him to play because they wanted to hear what was on the record? As far as I know, the problem hasn't been investigated, though I haven’t looked into it in awhile.

It is clear, however, that bebop musicians began insisting on making up something new each time you play a solo on a given tune. And that practice continued in subsequent styles. It is bebop common practice that has given us the idea of improvisation as coming up with something new. Note, however, that the major burden of newness falls on the soloist. The rhythm section players work in more constrained ways, though good ones feed off of and are responsive to the soloist and vice versa.

What happens when you toss out all the rules?

William Parker and John Cage

It may sound like chaos, but it isn’t. Not when it’s well done. Playing without any rules, though, is in some ways more difficult than playing with and so within them.

Some years ago I attended a workshop conducted by William Parker, an avant-garde jazz bassist and composer. My notes indicate there were five players, including Parker, but I seem to remember more. No matter.

We started with something Parker called “Number 14” — which he concocted on the spot. As he said, it starts with four fours. “What’s that?” you ask. You play four notes at a rapid clip, and do that four times in a row. “What four notes?” you ask. “He didn’t tell us, and none of us asked.” We all picked the notes we wanted when it came time to play the piece. But there’s more to “Number 14” than four fours. After four fours there’s a pause, that’s so, just like that, long. After the pause everyone picks a high note and does a long descending gliss(ando) to some low note. “What high note to what low note?” you ask. “Do you really think there’s a specific answer to that question?” says I. And then we play a long trill. “On notes of your choice,” you remark. “Yes, that’s it.” After that, guitar one plays a simple one two figure and repeats it four times. Then the ensemble does another high to low gliss to trill. That’s the “head” to “Number 14.” After the head, it was up to us and the music to negotiate the flow.

So we went through the head a couple of times and then played it down. It must have gone for twenty or thirty minutes. It started out pretty raggedy, but then things started to settle in—though “settle “ is not a particularly good word to use here. There’s no easy way to describe the music that evolved. Sometimes there was a pulse, sometimes there wasn’t. Even when there was a pulse, there where times when some people didn’t follow it. Sometimes everyone was playing, sometimes only one or two were. Sometimes loud, sometimes soft, sometimes fast, sometimes slow, always shifting. Sometimes full-tilt bozo, sometimes approaching serene.

This is a very demanding way to make music. Oh, if sound is all you’re after, it’s easy. Music is something else. You have to listen to everyone else – at least as much as possible – and figure out how you can fit it. Yes, I know, you should always be attending to the other musicians. But it’s one thing to do that in a situation where you’re playing from a fixed score, or within well-understood rules, it’s something else entirely when anything can happen.

While I have, shall we say, a temperamental inclination to go for the solo spotlight, I keep that firmly in check in situations like this. Yes, I will step out, but I’ll also play long tones to fill in the background and nudge things around a bit (by changing volume), or play bass lines – to be sure, in a treble register, but, functionally, I’m a bass player.

This sort of thing is not foreign to the world of classical music, at least not the classical avant-garde. When the score looks more like a work of abstract art or a circuit diagram than like a traditional notes-on-a-staff score, you are improvising whether or not you call it that. Why? Because there is no standard way to translate those marks on the page into musical sounds. You have to make it up.

Thus, back when I was researching my book on music (Beethoven’s Anvil: Music in Mind and Culture, 2001) I took a trip to Wesleyan University to visit a family friend, Jon Barlow. He was a musicologist, with a particular interest in Carnatic music, and pianist. By the time I visited him in the late 1990s his Parkinson’s had advanced to the point where he could no longer play from the standard classical repertoire. But he could play the piano music of John Cage. He asked me to perform one of Cage’s Etudes Australes with him. The pieces were written for solo piano, but Jon, who had known and worked with Cage, didn’t think it would violate the spirit of the work to have me play with him.

Cage derived the scores these pieces from star charts. Here’s how Wikipedia describes his composition process for these piece:

First, Cage put a transparent strip of about three-quarter inch over the maps. The width of the strip limited the number of stars used. Within this width Cage was able to discern the twelve tones of the octave. Then through chance operations using the I Ching, he transferred these tones to the available octaves for the left and right hands. The resulting notes reflect only the horizontal positions of the stars, and not all stars are used, because the maps used a variety of colors, and Cage's chance operations limited the choices every time to specific colors. In the end Cage would have a string of notes and ask the I Ching which of them are to remain single tones and which are to become parts of aggregates. In the first etude this question is answered by a single number, in the second by two numbers, etc. So as the etudes progress, there are more and more aggregates: in the first, most sounds are single tones, in the final, thirty-second etude, roughly half of the sounds are aggregates. The aggregates themselves were selected from the list of available aggregates, described above.

Though the pieces are written on score paper, there are no bar lines, dynamics, or specified tempos.

While Jon played from the score, the nature of the score obviously gave him a great deal of latitude. I had no score at all, nor did I look over his shoulder. I simply listened to him and fit in where I could. As far as I can tell, things went fine.

Were we improvising? Well, what do you mean? Apparently Cage himself disapproved of improvising, or at least the concept. But when you discard the standard harmonic and melodic structures of most kinds of Western music, and you produce such idiosyncratic scores, what’s left is certainly not Bach or Beethoven, nor Charles Wuorinen either. If performing such scores isn’t improvisation, then what is it?

Which is to say, we don’t have this one thing called improvisation. Improvisation, like fully-notated composition, takes many forms.

Miles Davis controlled the “long-line” of a performance

While jazz critics tend to late-period Miles Davis, when he included pop tunes in his repertoire, I heard a performance of “Time After Time” in 1987 which is one of the finest musical performances of any kind I’ve heard. I’ve listed to at least a dozen live versions of that tune on YouTube. Here’s a post where I compare two live versions (Chicago 1989 and Munich 1988) with the original studio version. The studio versions resemble on another more than either of them resembles the studio version. They’re longer (almost 10 minutes vs. a little over 5 minutes) and have a different overall shape (long-line), two high points vs. one.

A couple days later, after more listening and more thinking, I made a post where I said:

Over the last week I’ve listened to, I don’t know, say a dozen live versions of Miles Davis playing “Time After Time.” Some clock in at 9 minutes or so, others run 11 or perhaps more minutes. But they all have the same form, which I discussed earlier, a form not on the recorded version. Just how and when this form emerged – in rehearsal for the road, on the road, spontaneously or deliberately plotted – I don’t know, but once arrived at, it seems to have stuck. I’m guessing that if I listed to live versions of other tunes Miles was playing at the time that I’d hear the same thing, a consistent overall flow in the performance.

Two things are involved. There’s the overall flow of musical and emotional energy. But also occurrence of specific musical features, motifs, climaxes. In the case of “Time After Time” I’ve laid some of these out in my post.

My question: How common is this? First of all, with Miles, but then more generally in jazz performance.

I don’t know the answer to that question. In the case of big-band music the arranger – who may be the composer in some cases – controls the overall shape of the performance. But that is generally not the case with small group performances where, beyond the opening and closing performances of the head and the standardized ordering of solos in between, there is no overall compositional control of how the performance flows.

At the moment my impression is that Davis was doing something new. Perhaps not radically new, but a bit new. I’m pretty sure it’s not notated – no one is performing from a score – nor is it (by which I mean the overall shape) freely improvised on the spot. It is a routine the band has arrived at over time. It is un-notated composition, but operating on a different aspect of the performance from the un-noted standard solos that Louis Armstrong played.

Henry Lau’s free-styling

Finally, I want to look at something rather different. I want to look at some videos that a K-Pop star, Henry Lau, has done with young musicians. Lau was born in Canada about three decades ago and trained to become a classical musician when he responded to a talent search and went to Korea where, in time, he became a star. Starting about two years ago he began posting videos he made with musical prodigies. I’ve already written about these videos in previous posts. This one is in particular of particular interest: Kids and Music 6: Get Aboard Henry Lau’s Freestyle Express.

In the fourth and fifth videos he’s working with a young pianist, Seo Yul Shin. In the fourth one he introduces her to free-styling (his term for improvising), which was new to her. She gets the idea immediately. In the fifth video they perform a song they made up together. How many young classical musician in Korea, or elsewhere, are going to see videos like this and take them as a cue to do their own experimenting?

Now watch this video, where he performs with a young violinist, Sohyun Ko:

You’re welcome to watch the whole video, but pay attention from 5:20 on. They start playing a pop song, “Dynamite” by BTS. At 6:32 he calls for free-styling (presumably they’ve worked on this a bit earlier in the day) and she gives it a try. At about 7:48 she hesitantly offers a song that she had written and could Henry free-style with it? Who could he say ‘no’? It goes very well. Notice that Henry begins and ends with the melody she played. At the end he encourages her to spend a half hour every day free-styling.

Will she do it? I hope so.

This is something that Lau has done with a number of musicians and likely with do with others, as the second season in this series has only begun. Will there be a third season? If there is it will certainly have free-styling.

Will free-styling catch on with young Korean musicians? I have no idea. But Henry Lau is a pop star with a greater reach than any classical musician. But what will teachers and parents think?

A friend tells me that Korea has 20% of the world’s market for classical recordings. It is not unusual to have classical-style passages in K-Pop recordings; Lau does it all the time. Will young Koreans play a significant role in re-introducing improvisation into classical music?

Who knows.


  1. I would think that in the classical music world -- at least in part -- that the rules of give and take in social discourse enacted roles in the performance realm. The composer became the presiding voice, and the music was the dominant expression of that voice. The musicians and conductor keeping aligned with the hierarchies of their respective relationships with the music. Something like that. Along with which music somehow had become something "out there" to be realized as if divinely rediscovered for earthly measure.

    1. Yes, there certainly is a Platonic idealization of the composition so that, yes, it is treated as something out there to be imperfectly approximated in performance. I suspect this goes hand-in-hand with the Romantic cult of the genius, where the composer becomes that divinely inspired creature.

    2. That may shed some light on the fact that liturgical organ music, especially in France, is an exception. You don’t want to have the secular composer-god competing with God in the church. So you allow the organist to improvising, thereby consigning composers to a secondary role.

      Composers after Bach certainly wrote religious music, but none were so given over to it as he was. And there was room for improvisation in the accompaniment roles and in ornamenting melodies.

    3. (It's me, Larry. Not sure how I lost the tag with my name . . . anyway.)Richard Armstrong, in teaching voice, was categorized at Banff as "Extended Voice Specialist". He said that his joke was that the extended voice -- i.e. capable of tuvan overtone singing, polyphony, and several octave ranges -- is actually the normal voice, and that in western culture we've lose that connection of voice and body, noting that infants can vocalize for hours without any strain, sore throat, etc. Richard thought it interesting that as we've become industrialized, we've lost that connection with the body to make motor sounds, peep tones, etc. (Richard can do all this.) Classical music seems to have been caught in that same zeitgeist of disconnect of sound and the body. You can also see a similar trail of disconnect in the changes wrought in poetry in the place of sound and the body; poetry becoming more and more of the mind when it segues its way further from rhyme and actively composing voice. (Read an issue of American P0etry Review; the poems usually all sound the same.) The complex reasons for this hand over to the Platonic idealization of imagination in composition are multilayered in culture.

    4. Yes, the organist was allowed to be divinely inspired.

    5. In thinking about jazz we need to recall the role of the church in Black music. The relationship is obviously close in gospel, which IS church music while jazz is secular. But many jazz musicians grew up in the Black church. They heard performances that were divinely inspired. As such, there was no need for a composer.