Composition Versus Improvisation?

Yale University Department of Music
Composition versus Improvisation?

Author(s): Steve Larson
Source: Journal of Music Theory, Vol. 49, No. 2 (Fall, 2005), pp. 241-275
Published by: Duke University Press on behalf of the Yale University Department of
Music
Stable URL: http://www.jstor.org/stable/27639400
Accessed: 06-11-2016 17:39 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Duke University Press, Yale University Department of Music are collaborating with JSTOR to
digitize, preserve and extend access to Journal of Music Theory
This content downloaded from 117.131.219.47 on Sun, 06 Nov 2016 17:39:55 UTC
All use subject to http://about.jstor.org/terms
COMPOSITION
VERSUS
IMPROVISATION?
Steve Larson
In 1963, Bill Evans recorded the album Conversations with Myself

(Verve 314 521 409-2). To create that album's special sound, Evans first
recorded one piano track. Then, while listening to that track, he recorded
another track to go with it. For many of the selections, once he had two
tracks he liked, he overdubbed a third.
Composed of these improvised lines, Conversations with My self prob
lematizes the traditional distinction between composition and improvisa
tion. The following captures some of that traditional distinction.
Composition is traditionally re
Improvisation is traditionally re
garded as a process in which garded
a
as a process in which per
composer, with pen and paper, formers, with their voices or instru
outside of "real time," uses revi
ments, in "real time," use luck or
sion and hard work to eliminateskill to respond to or incorporate
or avoid mistakes; the composimistakes; the improvisation grows
tion builds on tradition, imposes
out of innovation, exploits freedom,
and relies on talent in an instanta
constraints, and relies on training
Journal of Music Theory, 49:2
DOI 10.1215/00222909-008 ? 2008 by Yale University
241
in a time-consuming process that neous process that involves emo

involves rational reflection and tional invention and intuitive impulse
intellectual calculation to create to create simple, direct expressions.
complex, sophisticated relation
ships.
I suspect that many musicians may also hold this traditional view (and it
is easy to see, given such a view, one reason that compositions are ana
lyzed more often than are improvisations). And I have read books and
articles?some of them by very distinguished writers?that articulate the
distinction this way.
I also suspect that most music theorists would say that, although there
is an element of truth in this distinction, all musical creation really lies
on a continuum between these poles. Yet in this article, I will go further
and claim that, in important ways, the traditional distinction has it back
ward. My evidence is drawn from two selections from Conversations

with Myself: " 'Round Midnight" and "Stella by Starlight."
"'Round Midnight"
The first cut on Conversations with Myself is Thelonious Monk's com
position " 'Round Midnight." Its theme is in the standard thirty-two-bar
AABA song form. Example 1 shows features common to Evans's settings
of the first A section and of the bridge. The first A section begins with a
distinctive motto. This motto is transformed into an ascending-seventh
chord arpeggio in mm. 3 and 5. The bass descends by half-step on two
levels of tonal structure. And the melody echoes that descent in m. 4. The
A section ends with a "linking motive." I call it the "linking motive"

because it participates in a kind of hidden repetition that Schenker called
Kn?pftechnik or "linkage technique": the motive that ends the first A sec
tion also begins the bridge. The changed function hides the repetition so
well that many listeners (including jazz musicians who have performed
the piece hundreds of times) fail to notice this link. After two statements
of the linking motive, its elements are reversed (so that another end
becomes a beginning). The result is that this collection of pitches, Gl?-F
Ek-D (a diminished tetrachord), and its extension Gb-F-Ek-D-Cb-Bl? (all
the descending half-steps in El? minor, both with and without change of
octave for the last two notes) contribute to the coherence and expression
of " 'Round Midnight." (Notice also that conceiving of the linking motive
as two descending half-step pairs, and its extension as three descending
half-step pairs, makes a connection between the linking motive and the
chromatic descent.) I return to all of these features in the analysis below.
" 'Round Midnight" usually begins with a standard introduction (which
has been attributed to Dizzy Gillespie). It features a distinctive melody, a
242
(* = chromatic descents, on two levels)
? ai?7 ? ? linking motive
linking motive
[B]
w linking motive
motive linking
V- v r?K J^P^F^
J
Cm7b5 F7
^te
?
Cm7b5 F7
Bbm7 Abm7
Eb7 Db7
F?m7 Fm7b5
B7 Bb7
mî r^=LJ r j
Example 1. Bill Evans, "'Round Midnig

Myself), first A section and brid
common jazz sequence of descending pairs

to the dominant, and a prolongation of th
dominant. Example 2 offers transcriptions
formances of this introduction.1
Evans's recomposition of this introduct

on the Gb-F-Eb-D diminished tetrachord
he creates the sophisticated motivic and

shown in Example 3.2 He first recorded th
of my transcription (labeled "Right"), w
243
moves down the circle of fifths (see the T5 below the Analysis system in
Example 3). He then added the part given on the top line of my tran
scription (labeled "Left"), which offers successive transpositions of the
diminished tetrachord discussed in Example 1 (see the T7 and T3 above
the middle line of the Analysis system in Example 3). Yet those transpo
sition levels differ from those of the bassline. Finally, he added the rip
pling arpeggios given on the middle line of my transcription (labeled
"Center")?which are transposed according to yet another scheme (see

the T3 and T7 above the top line of the Analysis system in Example 3).
Thus, Evans's version of the introduction exhibits some of the sophisti
cated relationships that we are accustomed to seeing described in analy
ses of composed music. In fact, we should regard the introduction as
"composed"?whether or not he wrote it down, Evans surely could have
reproduced it note for note. So it may be tempting to assume that these
sophisticated transpositional relationships are here because this passage
was composed.
Other elements may be regarded as "composed." The AABA theme
was composed by Monk. Evans had probably settled on the chords, the
Example 2. Three performances of the standard introduction to
"'Round Midnight": (a) Thelonious Monk, 1968.
244
rubato
Ip^nr^g- J j j
r r
"'Round Midnight": (b) Thelonious Monk, 1957.
245
Piano
Wi i,J ^r
Bass
Percussion
? n j r^
ijt^j Jim
Pno. <
Bass
X-J_I
fhffs
^s
v?-<? -*- * -*-/
4?4
-?]-J 7^
Pno.
" 'Round Midnight": (c) Bud Powell.
246
Pno. <
^'iM m
i A
EUE
Example 2(c) (continued)

textures, and the functions of each piano part for these thirty-two bars
before entering the studio. Example 4 offers my transcription of 1 A! and
the first half of 1A2. A transcription of the entire AABA theme may be
found as supplemental material (online only) with this article at jmt

.dukejournals.org. No jazz musician would describe the music of these
sections as an improvisation.
Yet parts of it probably were improvised. Evans surely planned the
basic harmonization and even some of the specific voicings used in his
accompanimental track (shown on the lowest system of Example 4).
Such accompaniments, called "comping" by jazz musicians, are like
Baroque continuo parts. And, like continuo parts, they are usually impro
vised; differences between the first two A sections in Example 4 suggest
that this comping part was also improvised.
The commentary track (middle line of the transcription in Example 4)
was probably improvised, too. Evans may have planned certain licks, or
planned, perhaps in outline, the beautiful gentle registral climb that occurs
in both the comping and commentary tracks over the first two A sections.
Nevertheless, commentary lines like this one are usually improvised. And,
again, differences between these two A sections suggest that this com
mentary line was also improvised; although the first two A sections have
"the same" melody, these statements also appear to have been improvi
sationally varied: notes have been added, subtracted, or embellished, and
rhythms have been varied.
Nevertheless, these A sections also feature the kinds of relationships
that analysts often attribute to composed music. Consider the imitative
polyphony of mm. 5-8 of 1A2 (see Example 5). Imitation at the second
occurs in mm. 5-6. Measures 7-8 present the same melodies at different
times and with different timings. Example 5 shows these relationships by
connecting analogous pitches with lines between the staves.
247
Example 3. "'Round Midnight," introduction
Analysis
Left
Center
Right
In fact, the coordination of the different tracks at the cadence in m. 8
is striking in another way. Notice that each of the tracks settles on a dif
ferent member of the tonic triad when it arrives on the third beat of m. 8.
Each track then moves to another member of the tonic triad on the fourth
beat. And all tracks do this in a way that results in complete triads on both
third and fourth beats. The result is shown in Example 6, with the tran
scription along the bottom, analytical notation in Example 6b, and dura
tional notation in Example 6a. While this may seem like a lucky coin
cidence, a look at the analogous measures of the other phrases in this
performance suggests that something similar happens in most sections.
In other words, sophisticated musical relationships occur not only in
Example 4. " 'Round Midnight," 1 Aj and first half of 1A2
249
(2) (D [1A%)
Example 4 (continued)
the "composed" sections, but also in the sections that involve some impro
visation. In fact, I suspect that some of these musical relationships are
there because they were improvised.
The commentary line of the first bridge (IB) was surely improvised.
In the spaces after each statement of the linking motive, the commentary
line restates the linking motive, but extends and embellishes it. And that
embellishment uses the ascending-seventh arpeggio of the A section. The
fill in mm. 5-8 is analyzed in Example 7. The original melody is elabo
rated in an elegant twisting line (Example 7c) that contains several state
ments of the linking motive (bracketed in Example 7b). The result is

several examples of what Schenker called "hidden repetition." The last
250
Statement of this motive suggests that D^ will follow, but the harmony
requires a different note, Dk Nevertheless, the basic contour, complete
with the following seventh leap, again refers to the linking motive.
These hidden repetitions occurred because these motives and embel
lishments "fell under Evans's fingers." To say this is not to dismiss their
musical significance. In fact, they fell under his fingers because he played
them a lot. And he played them a lot because his ear and his sense of
musical structure and development guided him to practice these patterns
when practicing this piece. Thus, in Evans's case, to say that he played
what fell under his fingers often means that it is well crafted, sophisti
cated, and compelling.3
For the same reasons, it is not surprising to find similarly sophisticated
relationships in Evans's improvised "solos." Consider the first of his

improvised variations on the bridge, a portion of which is analyzed in
Example 8. First, notice that the commentary track (Example 8, Center)
is playing the same licks we saw in the first bridge?but now at a differ
Example 5. " 'Round Midnight," mm. 5-8 of 1A2
251
ent transposition. Now look at the (presumably improvised) "solo" line

(Example 8, Right). This compound melody presents, as its lower strand,
a stepwise descent from Bb (stems down, Example 8b). As that stepwise
descent arrives at Eb, it seems to summarize the preceding material (see
the embedded brackets below Example 8b). Since that second strand has
reached Eb (the same note as the other strand), the compound melody now
takes a new path. But since the Eb is now a seventh with the bass, the line
must continue to D. In m. 4, Evans delays the expected D with a new

embellishment of this same linear progression that also incorporates the
A section's ascending-seventh arpeggio. The resultant hidden repetition,
since it simultaneously completes its path and that of the larger model it
represents (see the embedded brackets below Example 8a), is of the type
I call a "confirmation" (Larson 1987).
Every published Schenkerian analysis of an improvisation by Bill

Evans includes such confirmations. They occur in a live recording of
"'Round Midnight" (Larson 1987). They also occur in his improvisa
Example 6. " 'Round Midnight," cadence in 1 Au m. 8
252
Example 7. " 'Round Midnight," fill in IB, mm. 5-8

tions on "The Touch of Your Lips" (Larson 1998), "Stella by Starlight"
(Larson 1997-98), and "All of You" (Larson 2006).
Such confirmations seem a natural part of Evans's improvisational
style. His improvised lines are often easily viewed as?and heard as?
embellishments of a descending linear progression that threads its way
through the chord changes on which he is improvising. Some of those
lines are what contemporary jazz pedagogy calls "guide tone" lines. But
Evans's playing usually features long stepwise lines at levels closer to

and further from the musical surface, too.4 These lines typically termi
nate in the cadences of the theme on which he is improvising?or at least,
they sound like they are going to (Evans often artfully plays with our
expectations in this regard). Since some of Evans's favorite licks involve
foreground descents, as well, and since the rhetoric of Evans's solos often
exploits gradual expansions and contractions in pitch space, it is not

uncommon for Evans to approach a cadence with something that can be
heard to retrace the steps of the preceding section. The result is that two
versions of the same descending linear progression are completed at the
same time?a hidden repetition of the confirmation type.
Nevertheless, I think that it may strike some readers of this journal as
253
Example 8. "'Round Midnight," first improvised variation on bridge, 2B, mm. 1-4
(D
'
(D
[3:51]
Right
Center
Left
unusual or unexpected that such hidden repetitions would be common in
improvised jazz?even in the improvisations of an artist such as Bill

Evans. Elsewhere (Larson 1998), I have cited and responded to the reluc
tance of scholars to apply Schenkerian analysis to jazz improvisations.
Among these are Greg Smith, who wrote that Schenkerian analysis does
not "take the circumstances of composition in performance fully into
account" (1983, 91). Instead of Schenkerian analysis, Smith suggests

that
an analysis conditioned by a more realistic image of the kinds of relation
ships a player is capable of developing and sustaining in the course of

performance would, presumably, be couched in less ingenious explana
tions of the melodic organization. Such analysis would seek to reveal
devices of structure and expression of a sort a player could conceivably
master in performance. (1983, 126-27)
Smith argues that Bill Evans could not conceive and develop the kinds of
relationships shown in a Schenkerian analysis in the act of improvising,
but rather relied on a set of formulas that fell readily under the hand.
My first response to Smith was that if these patterns are in the impro
visations, it does not matter what we think Evans could or could not
conceive. But I went on to include a transcription of Evans's demonstra
tion of his assertion that "I always have, in anything that I play, an abso
lutely basic structure in mind. Now, I can work around that differently, or
between the strong structural points differently, but I find the most fun
damental structure, and then I work from there." And I noted the striking
similarity between the stages of Evans's description and the levels of a
Schenkerian analysis.
My argument in that article concerned the applicability of Schenker
ian analysis to improvised jazz, but my point here is somewhat different:
I want to explore why our way of thinking about improvisation versus
composition would lead us to embrace hidden repetitions as a significant
artistic feature of composed music and yet be skeptical of the same fea
tures in jazz improvisations.
To dramatize this point, consider received ideas about rhythmic com
plexity in jazz. Notice the rhythmic virtuosity of the sections that follow
the one analyzed in Example 8. Example 9 quotes two passages from

those sections. In the first (2B, mm. 5-6), all parts stay almost entirely on
the offbeats. In the second (2A3, mm. 6-8), the commentary line plays a
long string of dotted eighths, creating a polymetric effect with its implied
3/16 against the prevailing 4/4.1 have described elsewhere (Larson 2006)
how such rhythmic displacement is generated in Evans's music and have
offered other, even more impressive examples (Larson 1997-98).
To make such a journey through the "metric space" of this piece with
such confidence and elegance, Evans must have known that metric space
255
r*
7..J vi.ly
*SW J t
jjps 11
? ?jj ? ? i~? ? c ?P
Lf?Tf Lfjf t-M^
?Jyt?E
? L ? r^Mh ?j ?
Example 9. " 'Round Midnight," rhythmic complexity in two passages: (a) 2B, mm. 5-6.
-?-htp"i y r^ îî ^ jt? i^rn y ^tii^f

^=fe
? ?#?
.^
?
g##
tJ^+ff
^? T^^
J^/f EiP^
?E
Iliis
Example 9. " 'Round Midnight," rhythmic complexity in two passages: (b) 2A2, mm. 6-8.
[4:41]
intimately, must have internalized its possible basic rhythmic paths

securely, and must have developed many ways of traveling those paths
flexibly and fluently. And I suspect that most analysts have no difficulty
believing that he did.
To improvise interesting middlegrounds that include hidden repeti
tions, Evans must have worked in an analogous fashion. Evans must have
known the analogous pitch space intimately, must have internalized its
possible voice-leading paths securely, and must have developed many

ways of ornamenting (in other words, prolonging) those paths flexibly
and fluently. Those familiar with John Rink's excellent essay on "Schen
ker and Improvisation" (1993) will see that such an analogy is consistent
with Schenker's concept of improvisation. Yet I suspect that there are
analysts for whom this analogy is not obvious.
I have encountered a number of musicians who regard jazz as rhythmi
cally complex or difficult, yet tonally simple or easy. Compared to some
twentieth-century concert music, this may be a fair description. Yet I find
it interesting to speculate as to why some people might think of jazz as

rich in rhythmic complexity yet lacking in pitch complexity. Perhaps
rhythmic complexity is more obvious on first listening than pitch com
plexity is. Perhaps typical?and misleading?dichotomies slip into our

thinking unnoticed: jazz versus classical, African versus European, body
versus mind, rhythm versus melody. And perhaps we are misled by our
notions of improvisation versus composition.
Evans was able to improvise in these rhythmically and melodically

compelling ways only because he spent years studying and practicing
patterns: the rhythmic and harmonic patterns upon which jazz standards
are based, the rhythmic and melodic patterns of the specific pieces he
played, the patterns of voice leading that he embellished, and the patterns
of ornamentation that he applied to these other patterns. Although Evans
knew many pieces, he focused his greatest attention on a small subset of
these pieces; he performed and recorded some pieces over and over again
during his decades-long career.
Thus, I feel confident in describing Evans's apparently "instantaneous"
improvisations as the result of years of preparation. In fact, I suspect that
any given mature Bill Evans recorded improvisation took him a great deal
more time to produce than most twentieth-century compositions of com
parable length cost their creators. The real work of producing such impro
visations happens not on stage or in the recording studio, but in the prac
tice room.
According to this argument, then, Evans's intimate familiarity with
"'Round Midnight"?with its motive structure, with its rhythmic and
metric structure, and with its underlying voice-leading structure?helped
to ensure that his improvisations on it would have compelling relation
ships with those structures.
258
However, some readers may object that the distinctive character of

"'Round Midnight" already tends to guarantee both that those relation
ships will be strong and that they will be apparent to the analyst. " 'Round
Midnight" is a slow ballad in an unusual key (Eb minor is not a common
key for jazz standards) whose distinctive melody has many leaps. All these
factors seem to lead jazz musicians to keep their improvised variations
closer to the original melody of" 'Round Midnight" than they might with
other themes that lack these characteristics.
"Stella by Starlight"
"Stella by Starlight" is quite different.5 It has a straightforward mel
ody that is mostly step wise. And Evans performs it at a medium tempo
and in a commonly used key (Bb major). Most jazz musicians would not
expect to hear strong relationships between an improvisation on "Stella"
and its original melody. And, at least on first hearing, Evans's improvised
variations sound more like variations on the harmonic structure of the
theme than like a paraphrase of the melody.
Few would accuse Evans of merely "running the changes," that is, of
mechanically producing melodic lines that, although they fit the chords,
are bereft of global logic (including any relation to the original melody
of the theme). To me, it seems counterintuitive to think (as some writers
seem to have suggested) that Evans's improvised lines could have global
logic but not contain the kinds of middleground structures depicted in
voice-leading graphs. But I will grant that such global logic does not seem
to require reference to the original melody or guarantee the sophisticated
structures (e.g., hidden repetitions) admired in the common-practice
works that theorists typically analyze. Yet, in the analyses that follow, we
will see that Evans's improvised variations on "Stella" not only possess
interesting middlegrounds, but also contain subtle references to the orig
inal melody and a number of hidden repetitions.
Example 10 analyzes the leading melodic line at the beginning of

Evans's first improvised chorus.6 Example 11 analyzes the analogous
material in the next chorus. The discussion that follows refers to both
examples at the same time. In both Examples 10 and 11, level d gives the
right hand of the leading track and the left hand of the comping track.7
Levels a-c analyze this music in stages.
Level c includes every bass note, but rhythmically normalizes them,
and shows some of the compound melodic structures that result in the
bassline.8 All chord roots for what might be called "the basic changes"
are stemmed; most embellishing notes are not. Level c also includes
almost every right-hand note (I assume that the embellishment functions
of the omitted notes are clear). On level c, in the first four measures of
both choruses, asterisks connected with a line show Bb-C-D, an element
259
a <-"~"~~^T 6t?T
& - &- & - %-&
Example 10. Bill Evans, "Stella by Starlight" (Conversations with Myself), first improvis
JîE?LLiEl-__63) (34) (g) (36)
[1:03] (g)
I _ ?I 1._I -^- ?l??J^=^=^
@@@
@^_?_ -???1
of a subtle but elegant compound melodic structure. And throughout, the

use of a limited number of techniques of melodic elaboration gives the
whole an organic unity. On the one hand, it may seem remarkable that?
in a "solo" that was surely improvised?every note shown at this level
makes perfect sense; not only does every melodic gesture clearly express
the chord of the moment, but it also fits into the deeper-level structures
shown on level b. On the other hand, Evans created this recording in the
studio, so he had the opportunity to rerecord any solo that contained such
"mistakes."
In my conversations with musicians and nonmusicians alike, I have

often heard them suggest that improvisers capitalize on their mistakes?
in fact, that mistakes are an important part of improvisation and that
responding to such mistakes and incorporating them into one's improvi
sation is central to the process of improvisation. But Evans's live record
ings also contain very few "mistakes." Despite "folk theories" of impro
visation, "mistakes" simply did not play a meaningful role in Evans's
improvisations. Any theory of improvisation and composition that grants
an important role to "mistakes" completely misunderstands the accom
plishments of musicians like Bill Evans.
Of course, one may reasonably assert that Bill Evans was an unusually
gifted improviser and that his improvisations differed from others' in this
regard. However, in my experience, it is not these kinds of mistakes that
separate Evans from other jazz pianists. There are plenty of competent
improvisers who only rarely make mistakes at this level of musical struc
ture. While Evans was remarkably consistent in avoiding mistakes at this
level, his real gift appears clearly only when we examine the deeper lev
els of his musical structures.
Level b shows how the bass notes group in pairs (with pairs of chords
that are typically in a II-V relation) to project two-bar and four-bar

groups. Level b also shows the basic stepwise skeleton that underlies
Evans's improvised lines. Notice that most of levels a and b may be
described as long passages of descending steps (with some ascending
sevenths or other register transfers). As mentioned above, such lines are
common in Evans's improvisations. Converging lines on level b show

how Evans approaches goal notes in contrary motion, often with a chro
matic lower neighbor or with chromatic passing notes in either or both of
the lines that converge.9
Level a shows how the II-V pairs of level b relate to the more-basic
functional harmonies of the theme. The harmonic reduction shown in this
way agrees with the general outlines of the theory of harmonic substitu
tion described by Steven Strunk (1979), even though it departs from the
particular reading Strunk gives for this song. Level a also shows how
the right hand combines descending steps, ascending sevenths, and the
resolution of unstable notes to create the compound melody that moves
263
1____]___J_ r?3?! _ |________l_?_______l
Example 11. "Stella by Starlight," second improvised chorus
5!î^^--?-?_(?) ? ? ? ? ^?-s
(63) (64) _ _^ ^ _ ^
[1:55]
d <
f f **rr [ir^^r"
ff?f?fr
tfO
1- r?~~~?'?~\
(76)
(77)
d <
largely in parallel sixths. These sixths are expressed both horizontally

and vertically on level a.
It is at this level that I feel we get the clearest picture of what makes
Evans's playing so compelling. Notice that mm. 1-4 of both choruses

refer to the melody (see the top system of Example 2, mm. 1-4) by begin
ning with the descent Bk-A-G-F. But in both cases, the pickup notes in
the preceding two measures elaborate the descending third D-C-Bk (On
level b, a bracket above the pickup notes to the first improvised chorus
points to the embellishment of this pattern with an upper neighbor El?,
creating the pattern D-Eb-D-C-Bb, the subject of hidden repetitions on
many later levels, and an important element of the deeper structure of the
original melody of the theme. While this sort of hidden repetition may
seem remarkable, I believe it contributes less to the elegance of Evans's
improvisation than the features I point out below.) When this D-C-Bl? is
joined to Bb-A-G-F, the result is a descending sixth. That sixth is fol
lowed by a number of "motivic answers," most of which may be described
as an ascending leap of a seventh to an unstable note that resolves down
by step to create another sixth. And these answers are ordered so that they
appear gradually higher and higher. In the next four measures, the origi
nal melody is based on the shape G-F-Bk That shape plays in an impor
tant role in the fifth through eighth measures of both choruses, but it is
also absorbed into the motivic answers mentioned above. The following
four measures begin with El? resolving to D. The first improvised chorus
echoes that resolution, but the second improvised chorus takes an entirely
different path. Yet, again, both continue to echo varied repetitions of the
motivic answers mentioned above. In the following four measures, the
brackets on levels a and b show confirmation-type hidden repetitions (of
El?-D-C-B, a transformation of the D-Et-D-C-Bl? figure noted paren

thetically above).10
Levels a and b show interesting references to the original melody and
a number of hidden repetitions. But mere verbal description misses the
best part; the real elegance of Evans's playing seems best grasped when
we hear it as elaborating the slowly climbing and intrinsically beautiful
motivic answers of balancing steps and leaps shown in level a.
Example 12 summarizes the deeper structure common to both pas

sages?and reveals further hidden repetitions. The lowest two levels of
this example reproduce Examples 10a and 11a. Notice that the sixths
shown horizontally in Examples 10a and 11 a are verticalized in Example
12b?creating a deeper-level expression of the same descending sixth

(= 3prg + 4prg) that began Examples 10a and 11a. Example 12b also ends
with a deeper-level expression of the same confirmation bracketed in
Examples 10a and 11a.
These examples suggest that, despite the differences between " 'Round
267
? ? ? ? ? ? @." ? @
,_^ i ^ , 6Pfg r?-?-?-?i
6prg = 3prg + 4prg_ I-~???~-"?? -? - ? _ "~~~~ " ~~~ 1
Example 12. Common structure of improvised choruse

6prg = 3prg + 4prg ^_^^ j?LjyS] \^~-~~"~^-^
Ex. 10a< 6th 6th 6th 6th
6prg = 3prg + 4prg
Ex. lla< 6th 5th
Midnight" and "Stella by Starlight," Evans's improvisations on these

pieces relate to the melody of the original theme in interesting ways,
contain hidden repetitions on several levels, and owe some of their ele
gance and global logic to middleground motives.
In discussing Evans's recomposition of the standard introduction to
" 'Round Midnight," I also noted interesting transpositional schemes. Are
there such transpositional features to be found in "Stella"? Actually,
despite the noted differences between these pieces, they do share an in
teresting harmonic feature: the harmonic progression of the introduction
to "'Round Midnight" is transpositionally equivalent to Evans's rehar
monization of the final section of "Stella by Starlight." This section is
mm. 25-32 of the theme and is also the basis of mm. 57-64, 89-96,
121-28, 153-58, and 159-64.
The last of these sections is analyzed in Example 13. Just as in Exam
ple 3, the bassline moves down the circle of fifths (see the T5 below
Example 13b). But here, that bassline moves around the circle of fifths
by alternating tritones and descending half-steps (see the T6 and Tj ? below
Example 13c). The result is a pitch-class set that is transposed down a
fifth (see the T5 below the system of Example 13 labeled "Right")?but
now in partial melodic inversion with each transposition. Just as in Exam
ple 3, one part (the part labeled "Right" in Example 13, and the part shown
on the top staff in Example 3) follows a different transpositional scheme?
up a minor third and then down a perfect fourth (see the T3 and T7 between
the staves on the system of Example 13 labeled Right)?but now with

six-note instead of four-note sets. Just as in Example 3, the sequence itself
moves down by major second every two bars (see the T10 above the system
of Example 13 a)?but now with varied melodic realizations (each T10 on
Examples 13b and 13c illustrates one of these varied transpositions). And,
just as in Example 3, additional transpositional relationships emerge. The
piano labeled "Left" (analyzed on the middle system of Example 13c)

presents varied transpositions of three sets: the diatonically filled in de
scending minor third, the descending arpeggiation of a minor triad, and
a five-note subset of the chromatic scale. The T10 and T0 on Examples 13b
and 13c describe the transpositions of the first two of these sets. The T3
and T5 on Example 13c describe the transpositions of the last of these
sets.
My purpose in pointing out these relationships is not to argue that they
ensure the quality of this music. I think that these musical relationships
contribute to our experience of the music. And I think that the complexity
of these relationships adds richness to the music. But I believe that the
quality of music lies partly in its balance of simplicity and complexity.
And I believe that while analysis may illuminate quality, it is a poor way
to "measure" quality. Analysts can always describe simplicity as elegant
and coherent (instead of poor or vacuous) when it appears in something
269
Example 13. "Stella by Starlight," mm. 159-64

they like, and still describe complexity as rich and sophisticated (instead
of inelegant or incoherent) when it appears in something they like.
Instead, my purpose is to examine the implications, for the practice of
analysis, of the distinction between composition and improvisation.

What I find in a comparison of Examples 3 and 13 is that trying to explain
differences in passages as a result of the real-time constraints of impro
visation can be misleading.
In some ways, Example 13 is less complicated than Example 3. One
270
might argue that the real-time pressures of improvisation did not allow
Evans to develop the kind of simultaneous T3 plus T7 against T7 plus T3
that appears in Example 3. While the T3 plus T7 scheme appears in the
accompaniment, so the argument might go, it is just too much to expect
that the improvised line above that accompaniment would express the T7
plus T3 scheme at the same time.
In other ways, Example 13 is more complicated than Example 3. One
might argue that the real-time pressures of improvisation meant that after
271
playing the licks that began this section, it was easier for Evans to play
similar shapes at a less restricted number of transpositional levels than to
do something more consistent.
Neil Sorrell suggests that, with respect to complexity, the traditional
distinction between improvisation and composition has it backward, at
least from the point of view of the history of Western art music: "The
argument that composition enables improvised material to be worked out
in a more complex fashion was perhaps the reverse of the truth. Compo
sition became a means of restraining the excesses of certain kinds of
ostentatious improvisation" (1992, 778-79).

Thus, one could argue that the real-time pressures of improvisation
result in some relationships being simpler and others being more com
plex. While this may be true, it seems more fruitful to regard these rela
tionships as choices (which, at least for a recorded performance, can
always be rejected in favor of another take) and to ask how these relation
ships affect us as listeners.
Conclusion
I began this article by observing that the music on Bill Evans's record
ing Conversations with MyseZ/problematizes the traditional distinction
between composition and improvisation. And we have studied transcrip
tions and analyses of two performances from that recording.
This study suggests a different definition of improvisation. I now under
stand improvisation as the real-time yet preheard?and even practiced?

choice among possible paths that elaborate a preexisting structure, using
familiar patterns and their familiar combinations and embellishments."
And I now understand composition as putting together musical elements
and storing them?whether in memory, notation, or sound-recording
media?in a way that allows, but does not require, revision.
These definitions are not mutually exclusive. Music can be either, nei
ther, or both of these things. Some improvisations are best regarded as
compositions. Other improvisations are not. Some compositions are best
regarded as recorded improvisations. Other compositions are not. Some
aleatoric music, often cited as an example of music that is both composed
and improvised, may be neither. And I suspect that all enduring music is
created by improvisation, whether or not it is recorded in notation.
Conversations with Myself records Evans as both composer and im
proviser. Evans composed many of the pieces on which he improvised,
including one of the pieces on this album. But, in putting together its
musical elements and storing them on this recording, in a studio setting
that allowed but did not require revisions, he was also composing even
when he was improvising. I cannot imagine a different?and still mean
272
ingful?definition of "composition" that would capture what Bach and

Bacharach do, but that would exclude what Evans has done here.
In fact, Evans's improvisations on this album suggest that the tradi
tional distinction between improvisation and composition has it back
ward in some important respects.
To think that improvisation differs from composition because impro
visation is an instantaneous process is misleading. Most of the work of
the improvisations recorded here took place in the practice room, and I
have argued that any given mature Bill Evans recorded improvisation took
him a great deal more time to produce than most twentieth-century com
positions of comparable length cost their creators. For the same reasons,
to suggest that improvisation, as opposed to composition, relies on luck
or skill instead of thought and hard work is similarly misleading.
To suggest that an important part of improvisation lies in how the
improviser responds to or incorporates mistakes is misleading. Seeing so
few mistakes in the work of competent improvisers, I now assume the
reverse: mistakes probably play a more important role in the process
of composition than in improvisation; composition allows for the revi
sion of mistakes, mistakes simply do not play an important role in
improvisation.
To associate tradition, training, and constraint with composition (and
innovation, talent, and freedom with improvisation) also seems mislead
ing. Again, I now assume that the reverse is more appropriate. The real
time pressures of improvisation require more reliance on tradition, train
ing, and clear constraints than do those of composition.
And to associate simplicity with improvisation and complexity with
composition is misleading. Both composition and improvisation must
work to balance simplicity and complexity. In fact, I argue that some of
the sophisticated relationships prized by music analysts appear in the
music of Bill Evans?and in the music of Bach and Brahms?because

both were (according to the definitions offered here) improvised.
273
NOTES
1. The performances are (a) Thelonious Monk, Monk's Greatest Hits (November
19, 1968, Columbia CS 9775 and 32355); (b) Thelonious Monk, Thelonious Him
self (April 5, 1957, Riverside 12-253), reissued on 'Round Midnight (Milestone
M-47067); and (c) Bud Powell, Bud Powell (Quintessence QJ-25381). The timings
[given in square brackets] indicate where to find the transcribed material on the
recording. On the transcriptions of " 'Round Midnight", I have numbered the mm.
(1-8) within each formal section. The measure numbers are circled Arabic numbers
and appear above the bar line that begins that measure. Measure numbers are
circled in the transcriptions and musical examples, but not in the text. I have also
designated the beginning of each formal section. The formal-section designations
appear in square brackets above the bar lines that begin a formal section. The
formal-section designations have three parts: the first describes the (thirty-two
bar) chorus, the second describes the (eight-bar) section, and the third (subscript)
distinguishes the first A section from the second and third. Thus, for example, 2A3
is the third A section of the second chorus.
2. Here, and in Example 13, the notation "Tx" means transposition up x semitones in
the appropriate octave. In other words, x indicates what Joseph Straus calls an
"ordered pitch class interval" (2000, 7).
3. Henry Martin makes this same point in an insightful discussion of Charlie Parker's
use of "pathways" and "licks" (1996, 115-19).

4. Richard Hermann (2004), Henry Martin (1996), and Thomas Owens (1974) have
noted similar lines in the playing of Charlie Parker.
5. "An Analysis Symposium: Alternate Takes?Stella by Starlight" appears in the

1997-98 issue of Annual Review of Jazz Studies 9; see Folio 1997-98, Forte
1997-98, Larson 1997-98, Lindeman 1997-98, Martin 1997-98, and Williams
1997-98.
6. On the transcriptions of "Stella by Starlight," I have numbered measures from the

beginning, not starting anew with each chorus.
7. In m. 41, both parts on the upper two systems play exactly the same notes (the
commentary line, middle system, adds one more note at the end of the bar; I have
shown both parts with divisi notation for beats 3 and 4). I suspect that this hap
pened because Evans had practiced putting certain ideas in certain harmonic con
texts within this piece. But, whatever the reason, the simple fact that it could
happen suggests that we should be reluctant to accept the notion that each new
improvisation is created completely from scratch.
8. On rhythmic normalization, see Rothstein (1990).
9. Larson (1993) has described similar melodic structures in the improvised lines of
Dave McKenna.
10. If we needed further demonstration that structures like those shown in our voice
leading analyses were real for Evans, we might compare the structure shown in
level a with what Evans plays in the commentary line in mm. 109-12.
11. In free jazz, that preexisting structure may be very loosely conceived, but in most
improvisational music, the preexisting structures seem to have very specific pitch
content.
274
WORKS CITED
Folio, Cynthia. 1997-98. "The Great Symphonie Theme': Multiple Takes on 'Stella's'
Scheme." Annual Review of Jazz Studies 9: 3-24 and 103.
Forte, Allen. 1997-98. "The Real 'Stella' and the 'Real' 'Stella': A Response to 'Alter
nate Takes.'" Annual Review of Jazz Studies 9: 93-101.
Hermann, Richard. 2004. "Charlie Parker's Solo to 'Ornithology': Facets of Counter
point, Analysis, and Pedagogy." Perspectives of New Music 42/2: 222-63.
Larson, Steve. 1987. "Schenkerian Analysis of Modern Jazz." Ph.D. diss., University
of Michigan.
-. 1993. "Dave McKenna's Performance of 'Have You Met Miss Jones?'" Amer
ican Music 11: 283-315.
-. 1997-98. "Triple Play: Bill Evans' Three-Piano Performance of Victor Young's

'Stella by Starlight.'" Annual Review of Jazz Studies 9: 45-56 and 105-7.
-. 1998. "Schenkerian Analysis of Modern Jazz: Questions about Method."

Music Theory Spectrum 20: 209-41.
-. 2006. "Rhythmic Displacement in the Music of Bill Evans." Structure and

Meaning in Tonal Music: A Festschrift for Carl Schachter, ed. David Gagn? and
L. Poundie Burstein, 103-22. Hillsdale, NY: Pendragon Press.
Lindeman, Steve. 1997-98. "Miles's 'Stella': A Comparison in the Light of the Two
Quintets." Annual Review of Jazz Studies 9: 57-76.
Martin, Henry. 1996. Charlie Parker and Thematic Improvisation. Lanham, MD:
Scarecrow Press.
-. 1997-98. "The Nature of Recomposition: Miles Davis and 'Stella by Star

light.'" Annual Review of Jazz Studies 9: 77-92 and 109-10.
Owens, Thomas. 1974. "Charlie Parker: Techniques of Improvisation." Ph.D. diss.,
University of California, Los Angeles.
Rink, John. 1993. "Schenker and Improvisation." Journal of Music Theory 37: 1-54.
Rothstein, William. 1990. "Rhythmic Displacement and Rhythmic Normalization." In
Trends in Schenkerian Research, ed. Allen Cadwallader, 87-113. New York:

Schirmer Books.
Smith, Gregory Eugene. 1983. "Homer, Gregory, and Bill Evans? The Theory of For
mulaic Composition in the Context of Jazz Piano Improvisation." Ph.D. diss., Har
vard University.
Sorrell, Neil. 1992. "Improvisation." In Companion to Contemporary Musical Thought,
ed. John Paynter, Tim Howell, Richard Orton, and Peter Seymour. London: Rout
ledge.
Straus, Joseph. 2000. Introduction to Post-tonal Theory, 2d ed. Upper Saddle River,
NJ: Prentice Hall.
Strunk, Steven. 1979. "The Harmony of Early Bop: A Layered Approach." Journal of
Jazz Studies 6: 4-53.
Williams, J. Kent. 1997-98. "Oscar Peterson and the Art of Paraphrase: The 1965
Recording of 'Stella by Starlight.'" Annual Review of Jazz Studies 9: 25-43 and
103-4.
275

Composition Versus Improvisation?

Uploaded by

Copyright:

Available Formats

Composition Versus Improvisation?

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Composition Versus Improvisation?

Uploaded by

Copyright:

Available Formats

Yale University Department of Music

Composition versus Improvisation?

In 1963, Bill Evans recorded the album Conversations with Myself

Journal of Music Theory, 49:2

DOI 10.1215/00222909-008 ? 2008 by Yale University

in a time-consuming process that neous process that involves emo

ward. My evidence is drawn from two selections from Conversations

A section ends with a "linking motive." I call it the "linking motive"

(* = chromatic descents, on two levels)

? ai?7 ? ? linking motive

Example 1. Bill Evans, "'Round Midnig

common jazz sequence of descending pairs

Evans's recomposition of this introduct

he creates the sophisticated motivic and

"Center")?which are transposed according to yet another scheme (see

Other elements may be regarded as "composed." The AABA theme

Example 2. Three performances of the standard introduction to

"'Round Midnight": (a) Thelonious Monk, 1968.

Example 2. Three performances of the standard introduction to

"'Round Midnight": (b) Thelonious Monk, 1957.

v?-<? -*- * -*-/

Example 2. Three performances of the standard introduction to

" 'Round Midnight": (c) Bud Powell.

Example 2(c) (continued)

found as supplemental material (online only) with this article at jmt

Example 3. "'Round Midnight," introduction

In fact, the coordination of the different tracks at the cadence in m. 8

Example 4. " 'Round Midnight," 1 Aj and first half of 1A2

ments of the linking motive (bracketed in Example 7b). The result is

relationships in Evans's improvised "solos." Consider the first of his

Example 5. " 'Round Midnight," mm. 5-8 of 1A2

ent transposition. Now look at the (presumably improvised) "solo" line

must continue to D. In m. 4, Evans delays the expected D with a new

Every published Schenkerian analysis of an improvisation by Bill

Example 6. " 'Round Midnight," cadence in 1 Au m. 8

Example 7. " 'Round Midnight," fill in IB, mm. 5-8

(Larson 1997-98), and "All of You" (Larson 2006).

Such confirmations seem a natural part of Evans's improvisational

Evans's playing usually features long stepwise lines at levels closer to

exploits gradual expansions and contractions in pitch space, it is not

unusual or unexpected that such hidden repetitions would be common in

improvised jazz?even in the improvisations of an artist such as Bill

account" (1983, 91). Instead of Schenkerian analysis, Smith suggests

ships a player is capable of developing and sustaining in the course of

the one analyzed in Example 8. Example 9 quotes two passages from

Lf?Tf Lfjf t-M^

-?-htp"i y r^ ^i^i ^ jt? i^rn y ^tii^f

intimately, must have internalized its possible basic rhythmic paths

possible voice-leading paths securely, and must have developed many

it interesting to speculate as to why some people might think of jazz as

plexity is. Perhaps typical?and misleading?dichotomies slip into our

Evans was able to improvise in these rhythmically and melodically

However, some readers may object that the distinctive character of

Example 10 analyzes the leading melodic line at the beginning of

& - &- & - %-&

J^iE?LLiEl-__63) (34) (g) (36)

I _ ?I 1._I -^- ?l??J^=^=^

of a subtle but elegant compound melodic structure. And throughout, the

In my conversations with musicians and nonmusicians alike, I have

v?-<? -- -*-/

1__]_J_ r?3?! _ |________l_?_______l

6prg = 3prg + 4prg_ I-~???~-"?? -? - ? _ "~~ " ~ 1