While it might be thought that musical improvisation is the specialty of American jazz, it has long been a common – indeed, perhaps basic – feature of music throughout the world. Arab, Indian, Iranian, and African musicians have all long been familiar with it. From the Middle Ages through the Renaissance in European music, it was standard practice to improvise a line in counterpoint over a cantus firmus. In the classical era, keyboardists often competed with each other in improvisational contests – Mozart, for instance, against Clementi, or Beethoven against rivals such as Hummel. Performances extempore are still standards features of organ recitals.
Improvisation in concert music declined in the nineteenth and twentieth centuries. However, by the late twentieth century, composers and performers began to revive improvisational practice. It has attracted the attention of musicians such as Lucas Foss, Pierre Boulez, and Terry Riley, for example, and groups such as the multi-faceted New York organization “Bang on a Can.” Currently, soloists in classical concerti exhibit a trend toward replacing stock cadenzas with novel impromptu efforts of their own.
For many listeners, the paradigm example of improvisation is jazz. In mainstream jazz, a “head” – usually based on a 32-bar jazz “standard,” such as “Body and Soul,” or 12-bar blues pattern – is played over once, or perhaps twice, framing improvised solos. The improvised melodies are played on the harmonic and rhythmic foundation provided by the head. Alternative chords are often allowed, depending on style. After a sequence of solos, the performance will normally end with a reprise of the head. There are many variations to the basic pattern. Several musicians may trade off with each other. Or, as in classic New Orleans jazz, many musicians can improvise collectively. The basic pattern was challenged by the rise of so-called “modal” jazz in which, instead of improvising on melodies that fit a set of chords, soloists would create wide-ranging variations within a single scale.
Neither of two extreme points of view about improvisation can be sustained. One of these might be termed the romantic perspective, according to which improvisation is utterly rule-free music-making – music created “without previous preparation,” as one work on piano instruction puts it (Palmer 1992: 109). Too often, though, the ex nihilo view is based on an equation of the improvised with the primitive or unschooled. Such a view, as applied to jazz, was popular in the mid-twentieth century among certain French journalists – for example, Robert Goffin, who often extolled the most untrained, most “frenzied,” versions of jazz as the most authentic (Goffin 1944: 124).
This sentimental perspective reflects naïveté about the basic resources that improvisational performances inevitably presume. Experts on Iranian instrumental music, for instance, explain that improvisers in that tradition must learn several hundred elements that make up the repertoires of what is called the radif (Nettl 1992). Analogous considerations apply to jazz improvisation, as demonstrated by one massive study of the topic (Berliner 1994). Jazz musicians also internalize a cache of musical forms – for example, meters, bar lengths, chord progressions, and even phrase patterns – as frameworks and as material for improvised solos. Whatever Coleman Hawkins was creating in his famous 1939 recording of “Body and Soul,” it was not the harmonic motion instanced by that song. He simply accepted it as a pattern for his solo. Even Keith Jarrett’s famously “free” piano improvisations were typically built upon a vamp of familiar chords.
The freedom of the improviser is also limited by what she must not do. In Ghanaian drum music, for instance, only certain instruments are allowed to improvise and they can do so only within prescribed limits (Chernoff 1982). Unwitting musicians who beat out novel pulses without regard to customary practice could easily confuse the dancers and other musicians. In jazz, too, the most daring soloist realizes that there are any number of things she is not supposed to do. Even in a “free jazz” context, a keyboardist is not normally allowed to interpolate Chopin’s Ballade in G minor or to beat the piano with a baseball bat. Further, there are contextual stylistic constraints. It might seem that while playing with Charlie Mingus, Eric Dolphy had as much freedom as could be imagined. In fact, Mingus encouraged those qualities in Dolphy’s playing that fit the conception of the music he wanted to realize.
Part of the explanation for the mystificationist perspective on improvisation is that most of us nowadays are mere auditors of the activity rather than participants. A partial antidote is the useful analogy some have suggested between musical improvisation and linguistic activity, in which we all participate. For instance, the highly interactive playing of jazz musicians has been framed as a musical conversation (Hagberg 1998: 480–1; Kraut 2007: 57–65, 177–82).
Equally extreme, of course, is the view that, once the materials that go into the process are understood, improvisation can, in effect, be explained away. According to this perspective, what we call “improvising” – unless it be mere noodling – always follows a preconceived plan. However, even if a performance were to consist of a dreary pastiche of learned material, there is no reason not to regard it as genuinely improvisational – unless the sequencing itself had been worked out in advance.
Judgments about improvisation quality have been made from both an extra- and an intramural point of view. Winthrop Sargeant, whose study of the musical elements of jazz betrays a peculiar love–hate relationship with its subject, lodges the complaint that in American jazz in general, “a sturdy repetition” of the music’s basic harmonic elements always underlies “the apparent freedom of improvised” jazz (Sargeant 1976: 247). Elaborating the thought, he gives jazz a generally lower place in a scale of musical values, when compared with that of opera and concert music (Sargeant 1976: 253–78). Sargeant’s provocative and detailed discussion of the matter merits more attention than we can give here. But he could be faulted for his tacit assumption that forms of music that privilege harmonic variety, at the expense of other values, are superior.
Nevertheless, even when viewed from a properly intramural perspective, issues of improvisational excellence are still complex. Generally, good improvisers will exhibit technical facility and display a resourceful and imaginative reach. But beyond these platitudes it is hard to generalize. Gambits appropriate to one musical genre or style would be inept or meaningless in another. However, the phenomenology of the knowledgeable listener’s experience does suggest one additional but fairly constant norm of artistic quality in improvised music.
With any kind of unfamiliar music, one can be interested in how it will go. Where a work is familiar, a listener can take an interest in the interpretive choices of the performer. However, with improvised music a knowledgeable listener’s focus of interest is complex from the outset. One will be interested in how the musical line itself unfolds and whether it hangs together. At the same time one will be interested in aspects of the activity itself (Alperson 1984: 23; Brown 2000: 121). And this is where a peculiarly salient norm surfaces. Even when a performance is going well, a knowledgeable listener will be alert to the musician’s willingness to take risks, at the peril of the quality of the musical line. If a performer’s choices get her bogged down, or if she runs out of ideas, one worries about how she will deal with the problem. If she pulls the fat out of the fire, we will applaud. This is not only true of jazz. In Iranian instrumental music, again, experts tell us that the unpredicted phrases are most prized (Nettl 1992: 191–2).
However, even here, we find a spectrum of degrees such that knowledgeable judgments will be highly contextualized. Given the style of music he played, we do not mind that Louis Armstrong worked out aspects of his performances in advance. By contrast, listeners have different expectations for music played by Charlie Parker. As alternative takes of his recording of “Embraceable You” for Dial records show, Parker would go in strikingly different directions with a given song, from one performance to an immediately successive one. Further, even more local circumstances make a difference. For instance, a solo in an Ellington concert would be expected to follow a prescribed melody more closely than a jazz jam using the same song – “Take the A Train,” for instance.
Consider the worry of the jazz journalist who complained about hearing pianist Ray Bryant play “After Hours” in what sounded like a note-for-note copy of his famous recording of it on Verve Records (Gioia 1992: 52–3). In a commentary on the example, Andy Hamilton is perplexed to explain a relevant difference between the two, given that the subsequent performance was, like the original, “fine blues piano” (Hamilton 2000: 177–8). In fact, Hamilton has stumbled onto a perfect illustration of the fact that we tacitly appeal to a musician’s intentions in order to mark an improvisation as such. Hamilton goes on to grant that there is an “improvised feel” in improvised music. But the observation fails to do any work. (What if Bryant’s original performance had not, in spite of its “improvisational feel,” actually been improvised – that it had been written out, for instance, or was a copy of a previous improvisation?) We may not be able to say with certainty what Bryant was doing on either occasion – but whatever it was depends partly upon his intentions at the time.
It is striking that a principled analysis of the concept of the improvisational has been so elusive. Some have approached the concept by relating it to another supposedly less daunting one – composition, for instance. Here, two opposite strategies open up. One is to illuminate improvisation by contrasting it with composition. The other is to try to demonstrate affinities between improvisation and composition.
Borrowing words from the jazz pianist Bill Evans, Ted Gioia states that improvisational jazz differs from many other artistic practices, including musical composition, by its dependence upon a “retrospective” rather than a “blueprint,” or “prospective,” model. In the prospective model, artists make decisions about what is to come next in light of an overall conception. With the retrospective model, “the artist can start his work with an almost random maneuver – a brush stroke on canvas, an opening line, a musical motif – and then adapt his later moves to this gambit.” The jazz improviser may proceed from his opening move in any number of directions (Gioia 1992: 60). However, there is no reason a composer, too, might not begin with a random maneuver and adapt later moves to the initial one. Furthermore, there is no reason why an improvising musician needs to play in the absence of an overall conception of what she is doing.
A novel way of contrasting improvisation and composition was articulated by the composer Ferruccio Busoni. Taking a stand against the common modern platitude voiced by Arnold Schoenberg and others that the performer of a composed work is only the servant of it, Busoni claimed that improvisation is historically, and perhaps logically, more fundamental than composition. Compositional notation, he states, “is to improvisation as the portrait is to the living model.” It is only “an ingenious expedient for catching an inspiration, with the purpose of exploiting it later.” An interpreter of a notated work thus has the obligation to do his best to “restore” what “the composer’s inspiration necessarily loses through notation” (Busoni 1962: 84).
Of course, it is difficult to sustain the thesis. First, Busoni appears to assume mistakenly that all musical works for performance are tied to scores. Second, as Stephen Davies has explained, works for performance in general have some degree of thinness – that is, some degree to which the work’s instructions, whether through a score or otherwise, leave some performance decisions to be determined by the performer (Davies 2001: 3, 20). To add that these decisions should be guided by some more fundamental model lying, so to say, behind the scored work is hardly helpful. (What would the criterion possibly be of a successful restoration?) Third, even if the concept of a musical work has only developed in relatively recent music history, it does not follow that fully fledged musical works are awkward attempts at catching something more original.
An opposed approach is to stress affinities between improvisation and composition. Gunther Schuller, in one of his exhaustive historical studies of jazz, recommends that we should see a jazz soloist’s recorded performance as a “work in progress” (Schuller 1968: x). If the similarity of Charlie Parker’s recorded solos to compositions seems less than obvious, consider that when he recorded his music, the final product issued to the public would typically be picked as the best of several recording “takes.” (And Parker’s case is not unique.) So, there may be some correspondence between this practice and the kind of trial-and-error methods of composers.
An example not limited to the territory of recorded music comes from the life of J. S. Bach. While at Potsdam, it is said, Bach improvised a three-part ricercare for Frederick the Great, and wrote the music down only later when he returned to Leipzig. According to Peter Kivy, “the composing was already done” when Bach improvised the piece (Kivy 1983: 124–5). Can we generalize from this kind of case?
In what might have been the first extended philosophical treatment of improvisation in English, Philip Alperson attempts to make the connection between improvisation and composition by means of a rather complex argument. He first establishes a reciprocal relationship between composition and work-performance (Alperson 1984: 19–20). In narrower senses of these concepts, it is customary to distinguish the two. However, in a broad sense, Alperson urges, composing always involves performance – for example, running over music in one’s mind if not actually playing it aloud. In a broad sense, too, the converse holds, given that, as already noted, there is always some degree to which the instructions for a work leave some decisions to the performer.
Now, when Alperson turns to improvisation, he says that we have an activity in which the improviser “practices simultaneously the interdependent functions of composition and performance in both the broad and narrow senses of the term” (Alperson 1984: 20). By these moves, the gap between improvisation and composition is gradually closed so as to yield the wanted analysis: improvisation is the composition of a musical work as it is being performed (Alperson 1984: 20).
Alperson was challenged on the grounds that he makes his case only by using the concepts of both composition and performance too loosely (Spade 1991). When arguing for the necessity of (improvisational) performance to composition, he sticks pretty closely to our standard concept of composition. However, when he turns to the converse point, Alperson is using “composition” in a much looser sense, where it now means something like “determining the sonic properties of a performance.” Analogously, part of the time Alperson uses “performance” in a standard sense – roughly, the tokening of a pre-existing work-type. However, when arguing that composition requires performance, he shifts to a loose sense of “performance,” where the mere generation of some musical sounds qualifies. The grain of truth in Alperson’s view might simply be that both improvisation and composition are creative activities.
If we compare improvisation and composition as practices, we can discern general reasons why the one cannot be assimilated to the other (Brown 2000: 114). Let us profile them.
The French existentialists were fascinated by the idea of forced choice, according to which every moment in life is latent with an anxiety-charged choice among alternatives. This may be an exaggerated picture of human life in general, but the thought might have some application to improvisation. By contrast with the improviser, the composer can take time out in her project – indeed, set it aside for years. The improviser must plunge ahead and do something. Stretches of silence can be musically functional in all music, whether composed or not. However, a pause in the process of composing a work does not become a potentially unfortunate feature of the work. With improvisation, time-outs resulting from fatigue or a lack of inspiration carry costs.
Indeed Alperson, whose overall theory seems to neglect it, notes such a difference between improvisation and composition (Alperson 1984: 23). At any point, the composer can alter what has so far been laid down. Not only can compositional projects be revised up to the point of publication but they can also perhaps be revised beyond that point, as examples by Stravinsky and others show. A subsequent effort by an improviser might be superior to a previous one, but it cannot count as a revision of an earlier one.
Finally, the improviser’s choices ramify, in the sense that she must produce onthe- spot responses to something already laid down. An extended improvisation is a continuous feedback loop, such that later phrases are responses to previous ones.
Now, none of the foregoing rules out that Bach, on the occasion cited earlier, was composing as he improvised. However, to generalize from that case to composition as a general practice is implausible. It is part of the practice of composing that composers do avail themselves of the conventions that allow the sorts of revisions and time-outs that are not allowed in genuine improvisation. (Imagine the riskiness of composition were it otherwise.)
Another way to explicate improvisation is as part of the very concept of work-performance. Perhaps, as some have maintained, improvisation is not a curiously separate and distinctive form of performance, but an inevitable dimension of any music-making whatsoever (Gould and Keaton 2000; Benson 2003).
For instance, Carol S. Gould and Kenneth Keaton argue that “all musical performance, no matter how meticulously interpreted and no matter how specific the inscribed score, requires improvisation” (2000: 143). Basic to the argument is the now familiar view – which the theory shares with Alperson’s – that musical works underdetermine their performances. The authors go on to claim that such work-performances count as improvisation, for improvisation is “a relation between the score and the performance event” (Gould and Keaton 2000: 145). (Throughout, it should be noted, the authors, like Busoni, assume what might be challenged – that all works are scored.)
In order to support this broadening of the concept of improvisation, the authors must interpret improvisation in such a way that an improvisation need not be spontaneous (Gould and Keaton 2000: 144–5). So, a specific thickening of the instructions for Beethoven’s Op. 135 that the Guarneri String Quartet might work out in advance would, in this theory, qualify as improvisatory. However, this would surely be stretching the concept of improvisation to the breaking point. At the very least, a necessary condition of an improvisation is that it involves spontaneity.
In a later essay, Alperson wrote, to “improvise is to do or produce something on the spur of the moment” (Alperson 1998: 478). There must be something to the idea. But can the matter be that simple?
To improvise, let us say, is to make decisions about the music one is playing as one plays. Note that we must of course avoid equating “music” here with “a musical composition.” But the formula faces other more serious difficulties of clarification. First, what should we make of the implicit temporal marker in the phrase, “as one plays”? Surely an improviser’s decision to go one way rather than another must have been made at least a nanosecond before following through. In fact, though, we may not know enough about the mechanics of mental activity to decide the issue one way or another, so let us leave the matter open: an improvisational move is one made at the time of or slightly before the move itself – where we shall assume that either formulation would make the addition of “spontaneity” in the formulation unnecessary (Kania forthcoming). However, we cannot avoid fuzzy cases here. If Sonny Rollins lays out a second chorus while playing the first, should we regard the second as improvised?
Further, what is it to make “decisions about the music”? Given what we saw about the inevitable resources that are drawn upon in improvisational performance, it is not clear what this concept means, or how it applies. As already suggested, even very free improvisations have some structural guidance. The genuine keyboard improviser, for instance, is not simply noodling.
At one point, I tacitly answered the foregoing question when I wrote, “an improviser makes substantive decisions” about what to play “while playing it” (Brown 1996: 354). But what kinds of decisions are substantive? Elsewhere it has been suggested that, in jazz, “an improvised performance is one in which the structural properties of a performance are not completely determined by decisions made prior to the time of performance,” where “structural properties” include melody, harmony, and length as opposed to “expressive properties” such as “tempo, the use of vibrato, dynamic, and so on” (Young and Matheson 2000: 127). But, first, the concept of a structural property remains unclear. (By what criterion would we distinguish between structural properties and others?) More generally, it is difficult to see why the musical properties that can be improvised should be restricted at all – except to those over which the improviser has control.
Let us grant then that an element of spontaneity is involved in any performance we term “improvisational.” With that qualification, can we then say that improvisation and work-performance “differ more in degree than in kind” (Gould and Keaton 2000: 143)? One might try to illustrate the view with a thought experiment: imagine a stretch of music consisting of, say, a hundred notes, such that some are specified by a score, with the others to be filled in spontaneously by the performer. Now, imagine many such sets in a spectrum, such that in some of them very few notes are to be filled in, while in others a great many are. The array might be thought to illustrate how the supposed difference between the two kinds of performance is only a matter of degree.
However, the thought-experiment at best illustrates the banality that in such a situation we have potential vagueness, since we cannot indicate a precise point at which a performance is no longer a work-performance but an improvisation. To conclude from that fact that there is no difference between the two kinds would involve a version of the so-called “slippery slope” fallacy. Further, the thought experiment has left out of consideration what the performer intends – that is, what she thinks of herself as doing. Does she think of herself as spontaneously fleshing out a work while remaining faithful to its composer’s style? Or does she think of herself as exploiting a given musical structure as a point of departure for music of her own?
The difference – in kind – between the case where a performer thickens a relatively thin work while performing and the case where she improvises ought surely to go something like this: in the former kind of case, the performer fleshes out a pre-existing structure rather than using it as a springboard for what Stephen Davies terms a “gravity defying” departure from such a structure (Davies 2001: 17).
But there are two grains of truth in the Gould–Keaton view. First, we can envision cases on the boundary between the two types of performance. In jazz pianist Uri Caine’s recently recorded performance of Mozart’s Sonata in C major, the “wrong” notes throughout can be assumed to have improvisational intention. But the performance does on the whole follow the general structure of the written music. Second, even within the class of uncontentiously improvisational performances, some may be more so than others. A typical solo by Louis Armstrong is less improvisational, for instance, than one by Charlie Parker. However, comparisons across musical genres will be difficult – if possible at all – for it is not clear how to enumerate the available options in one context by a measure that would apply in the other. How could we determine whether a bop solo by Charlie Parker is more or less improvisational than a classical Iranian performance on the ‘ūd?
The inclusion of stretches of improvisation in a performance does not rule out that such a performance may still count, ontologically, as being of a work. (Consider a piano concerto containing an improvised cadenza.) So should we simply borrow our ontology for improvisational performances from the best available view about musical works? Such a view would be hard to generalize because it would leave certain cases homeless – certain free jazz performances, for instance, which are not of any antecedent work. Is it possible that the concept of work-hood simply does not apply to such cases? Upon what does the question turn?
First, let us assume that by “art work” we mean something that can be re-identified – revisited, as it were, on multiple occasions. Obviously, a Keith Jarrett improvisation cannot be revisited in the way that we can revisit the Las Meninas of Velasquez, which can be found on a wall in the Prado, where we can go see it anytime we can afford to do so. But how could we possibly revisit an improvisation that is, so to say, entirely in-the-moment? As it happens, Jarrett’s Cologne improvisations were transcribed and published. However, a performance of one of them from the sheet music, or indeed, a copy of it by any means, whether by Jarrett or by anyone else, would surely lack an essential feature of the original, namely that – with the necessary qualifications – the music was created as Jarrett performed it. Given its once-only character, must we conclude that a Jarrett improvisation is not an art work? But now consider a visual work of performance art – such as those organized by Alan Kaprow, which, given their presumed spontaneity, could not be copied without loss of authenticity. In spite of this, such once-only events in visual art are documented and discussed just like art works in general.
Perhaps a musical improvisation is not an art work because an art work is something worked on over time (Kania 2008: 6–7). True, we can cite examples of art works that were in fact not worked on over time – Coleridge’s poem, Kubla Khan, for instance, if we accept the poet’s story about its spontaneous genesis. The reply, however, is that Coleridge could have worked on it over time.
Another reasonable criterion of workhood is that an art work is the focus of critical attention. By this criterion, Jarrett’s performances presumably would be works in their own right – if we are untroubled by the thought that it seems conceptually impossible for these musical works to have more than a single instance.
So with different criteria of workhood we get various problematic results. And sooner or later, we will find ourselves asking whether it is relevant that ECM recorded Jarrett’s performances for us to listen to as often as we wish. And would this be relevant because recordings do magically allow us to revisit an ephemeral event even though it has slipped into the past? Or is it because the Jarrett recording itself takes on the status of an art work? An ontology for improvised performances remains unfinished business.
See also Jazz (Chapter 39), Ontology (Chapter 4), and Performances and recordings (Chapter 8).
Alperson, P. (1984) “On Musical Improvisation,” Journal of Aesthetics and Art Criticism 43: 17–29.
—— (1998) “Improvisation – An Overview,” in M. Kelly (ed.) Encyclopedia of Aesthetics, vol. 2, New York: Oxford University Press, pp. 478–9.
Benson, B. (2003) The Improvisation of Musical Dialogue: A Phenomenology of Music, Cambridge: Cambridge University Press.
Berliner, P. (1994) Thinking in Jazz, Chicago: University of Chicago Press.
Brown, L.B. (1996) “Musical Works, Improvisation, and the Principle of Continuity,” Journal of Aesthetics and Art Criticism, 54: 353–69.
—— (2000) “‘Feeling My Way’: Jazz Improvisation and its Vicissitudes – A Plea for Imperfection,” Journal of Aesthetics and Art Criticism 58: 112–23.
Busoni, F. (1962) “Sketch of a New Aesthetic of Music,” in Three Classics in the Aesthetics of Music, New York: Dover.
Chernoff, J. (1982) African Music, Chicago: University of Chicago Press.
Davies, S. (2001) Musical Works and Performances: A Philosophical Exploration, Oxford: Clarendon Press.
Gioia, T. (1992) The Imperfect Art, Oxford: Oxford University Press.
Goffin, R. (1944) Jazz from the Congo to the Metropolitan, New York: Doubleday.
Gould, C. and Keaton, K. (2000) “The Essential Role of Improvisation in Musical Performance,” Journal of Aesthetics and Art Criticism 58: 143–8.
Hagberg, G. (1998) “Improvisation: Jazz Improvisation,” in M. Kelly (ed.) Encyclopedia of Aesthetics, vol. 2, New York: Oxford University Press, pp. 479–82.
Hamilton, A. (2000) “The Art of Improvisation and the Aesthetics of Imperfection,” British Journal of Aesthetics 40: 168–85.
Kania, A. (2008) “Works, Recordings, Performances: Classical, Rock, Jazz,” in M.D. Dack (ed.) Recorded Music: Philosophical and Critical Reflections, Middlesex: Middlesex University Press, pp. 3–21.
—— (forthcoming) “All Play and No Work: The Ontology of Jazz,” Journal of Aesthetics and Art Criticism.
Kivy, P. (1983) “Platonism in Music: A Kind of Defense,” Grazer Philosophische Studien 19: 109–29.
Kraut, R. (2007) Artworld Metaphysics, Oxford: Oxford University Press.
Nettl, B. (1992) The Radif of Persian Music: Studies of Structure and Cultural Context, rev. edn, Champaign: University of Illinois Press.
Palmer, K. (1992) The Piano, Chicago: NTC Publishing Group.
Sargeant, W. (1976) Jazz, Hot and Hybrid, rev. edn, New York: Da Capo Press.
Schuller, G. (1968) Early Jazz: Its Roots and Musical Development, New York: Oxford University Press.
Spade, P. (1991) “Do Composers Have To Be Performers Too?” Journal of Aesthetics and Art Criticism 49: 365–9.
Young, J. and Matheson, C. (2000) “The Metaphysics of Jazz,” Journal of Aesthetics and Art Criticism 58: 125–34.