Thoughts on Improvisation: A Comparative Approach

Author(s): Bruno Nettl

Source: The Musical Quarterly, Vol. 60, No. 1 (Jan., 1974), pp. 1-19
Published by: Oxford University Press
Stable URL: http://www.jstor.org/stable/741663
Accessed: 06-06-2018 18:42 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/741663?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

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

Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access
to The Musical Quarterly

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
 bOR
IL

Vol. LX, No. 1 JANUARY, 19

HIR

THE MUSICAL
QUARTERLY
THOUGHTS ON IMPROVISATION:
A COMPARATIVE APPROACH

By BRUNO NETTL

IN 1938 Ernst Ferand published what continues to be the only

large, scholarly book on the subject of musical improvisation.'
A rather thorough search reveals that this book was but little re-
viewed and was received with only mild enthusiasm.2 Although it
develops many concepts with considerable sophistication, and gives
much data as to the nature of improvisation in various periods of
European music history, it seems not to have had great impact on
its own time, despite the fact that, through his later encyclopedia
articles and anthologies, Ferand has obviously remained the single
outstanding authority in international musicology on this subject.
Perhaps his work was taken lightly simply because the subject of
improvisation was taken lightly by musicologists; it seems to have
been regarded as something not having cardinal importance3 - as
not truly art, but craft, which results in such "microcosmic" altera-
tions or elaborations of composed music as ornamentation or the

1 Ernst Ferand, Die Improvisation in der Musik (Zurich, 1938).

2 Reviews appeared in few journals, among them Music and Letters, XX/3 (July,
1939), 337-39; and Rivista musicale italiana, XVII (1939), 425-26.
3 For an illustration of this attitude, see the article "Extemporization," in the fifth
edition of Grove's Dictionary of Music and Musicians.

Copyright? 1974 by G. Schirmer, Inc.

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
2 The Musical Quarterly

realization of figured bass, or in such music

ability of a few organists to extemporize fugu
gested to them on the spur of the moment.
The concept of improvisation has become m
inent in musicology since that time, largely bec
attention given to musics which appear to depen
ily on improvisation than does European art m
many studies of jazz and of Indian, Indonesian,
African music that deal explicitly or, more fr
with improvisation. Courses, both practical an
voted to this subject, as are sessions at scholarl
the area of scholarly conceptualization Ferand i
and while he made admirable use of the information available to
him on non-Western music, it is probably time to begin to reth
the idea of improvisation, to see whether it merits consideration as
single process, whether it has integrity as an idea separate fro
other, related ideas about creating of music, and whether all t
things that we now call improvisation are indeed the same thing
In studying the definitions found in music dictionaries and
cyclopedias, we encounter two apparently conflicting views of
provisation. Some sources, such as MGG (in an article by Feran
indicate the relevance of the concept to non-Western, particula
tribal, musics, and state that, given the absence of notation, th
are basically improvised. Others6 confine the idea of improvisa
only to music for which there is basically a notation system fr
which the improviser departs; in this view non-Western music

4 Seminars on comparative approaches to improvisation have been held by

author, for example, at the University of Illinois. Meetings and symposia range
e. g., the large, six-lecture series, organized by Leonard Meyer and Ella Zonis at
University of Chicago in 1968-69, to shorter sessions at meetings of societies, e. g.,
Society for Ethnomusicology annual meeting in 1973, at Urbana, Illinois.
5 Ferand, in an article in Die Musik in Geschichte und Gegenwart, which par
his thinking in Die Improvisation in der Musik, says: "The division - taken
granted in Western musical life of today - that splits the original unity and sim
taneity of creation and reproduction was and is foreign to the musical usage of
primitive and many other non-European cultures; the inventor and executor
composition, the producing and reproducing musician, were originally in most
one and the same person" (Vol. VI, col. 1095).
6For 'example, Riemanns Musiklexikon, Sachteil (Mainz, 1967), in an unsig
article says: "Strictly speaking, only in the West, and even there only beginning
rather late historical stage, can one speak of improvisation, since non-Western
older European music stands outside the distinction between composition and
formance which is essential to the concept of improvisation" (p. 390).

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
 Thoughts on Improvisation 3

which one cannot really distinguish between impr

composition, cannot be said to represent either c
Harvard Dictionary7 implies this by entirely omitting
music from its article. Obviously the relationship of i
to composition and notation is a complex one, on wh
no general agreement.
Specifically or implicitly accepted in all the genera
is the suddenness of the creative impulse. The impro
unpremeditated, spur-of-the-moment decisions, and be
not thought out, their individual importance, if not o
tive significance, is sometimes denied. The above-men
clopedias are proportioned so as to spend far less spac
visation and the kinds of music that result from it than on various

kinds of composition; and, consonant with the general West

view of the process, they treat it as a minor art or craft of musici
ship.
The general literature on improvisation deals largely with the
phenomenon in Western music and attempts to gain a historical
perspective on the role of the process in the performance practice
of the past. Only briefly mentioned are the large number and
variety of non-European musical systems which make sharp dis-
tinctions between improvisation and the performance of standard-
ized pieces, or those in which the improvisatory element is a major
component of all performance. Ferand was well aware of the
existence of these systems and refers to them frequently, without
making them a major substantive component of his work.8 A new
edition of his study, or a new, truly comprehensive book on the
subject, would have to examine the various types of improvisation
that are actually recognized, though perhaps not designated by a
single specific term, in India. It would have to come to grips with
the special nature of the maqam and dastgah systems (the modal
structures of the Arabic world and Iran) and comment on the inter-
relationships of various performances of an epic, by one performer,
in the Yugoslav heroic song tradition. It should deal with the kinds
of group improvisation found in African drum ensembles and in the
music of the Indonesian gamelans. It should, moreover, try to

7 Willi Apel, Harvard Dictionary of Music, 2nd ed. (Cambridge, Mass., 1969),
pp. 404-5.
8 About forty-five pages in Die Improvisation in der Musik are devoted to
"primitive" and oriental musics.

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
4 The Musical Quarterly

explore the notion that the improvisatory p

varying degrees in tribal music such as that of
and it should explain the improvisatory side o
tion in European folk music.
The customary picture of improvisation, then
be greatly expanded by an understanding of no
Unfortunately, the recent ethnomusicological l
to this end in only a piecemeal fashion, providi
of systems and subsystems without giving mu
nature of the concept, and certainly making l
comparative. (Indeed, when it comes to prov
view, Ferand is still the most insightful schola
pages will be to present, very briefly, some th
tive sort on the nature of improvisation as
process, using material from a number of cult
if not to clarify, then to pose some questions a
this elusive idea. If the concept of improvisat
at all viable, it should be considered one of the few universals of
music in which all cultures share in one way or another.

Improvisation and Composition

Improvisation and composition are opposed concepts, we are
told9 - the one spontaneous, the other calculated; the one primi-
tive, the other sophisticated; the one natural, the other artificial.
But, on the other hand, we are also given to believe that improvisa-
tion is a type of composition, the type that characterizes those
cultures that have no notation, a type that releases the sudden
impulse to music through the direct production of sound. We hear
that improvisation ends where notation begins, yet at the same time
we are told that certain non-Western cultures which do not use
notation distinguish between the two processes, if not explic
then by the way they internally classify their musics. Thus,
we feel that we know intuitively what improvisation is, we find t
there is confusion regarding its essence.
For example, let us take the idea of unprepared and sud
creation as the major criterion and examine it in the light of
we know about American Indian music, one of the best described
non-Western musical cultures. It is known that the Plains Indians

9 This statement is based on Ferand and on other previously cited articles.

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
 Thoughts on Improvisation 5

seek visions in order to learn songs, and that these s

them during periods of ecstasy sometimes brought o
and self-torture. The sudden creation of a song is
statements from Plains Indians that songs can be and
learned in a single hearing."' Are these songs improvis
are first sung? It would appear so. But we are told
visionary, upon learning a song, sings it to himself an
"works it out" when he walks back to his band or tribe. Once the

song exists, it takes on the trappings of establishment, like a

position in the Western sense. The composer is known, re
bered, and named, and the circumstances of composition are
times recounted before singing. Repeated recordings show th
song remains reasonably stable. Is this composition or im
visation?
For the Pima Indians of the southwestern United States, on the
other hand, songs existed (in the supernatural world) but had to
be "unraveled" by humans in order to be realized." Here we find
an idea which may approximate that of Western-style composition
in its resort to calculation. But there is no notation, and the style of
Pima songs, generally speaking, is actually not very different from
that of Plains music.

A third North American example comes from the Eskimo, wh

recognize two ways of making songs, a conventional one and
improvised one - the latter being the type of song represented b
the famous song-duels wherein quarreling men settle disputes b
mocking each other.12 The use of a repertory of standard formu
is suspected,13 but this is something shared with at least some oth
cultures which improvise. Again, the improvised songs show
basic dissimilarity of style when compared with the composed o
traditional songs.

10 Bruno Nettl, "Studies in Blackfoot Indian Musical Culture, Part II," Ethno-
musicology, XI/3 (September, 1967), 301.
11 George Herzog, "A Comparison of Pueblo and Pima Musical Styles," Journal
of American Folklore, XLIX (1936), 333; and his "Music in the Thinking of the
American Indian," Peabody Bulletin, May, 1938, p. 2.
12 A number of sources on Eskimo composition are discussed and summarized in
Alan P. Merriam, The Anthropology of Music (Evanston, Ill., 1964), pp. 175-77.
13 Several Eskimo song texts dealing with composition are given by Laura Boulton
in the booklet accompanying the record album, The Eskimos of Hudson Bay and
Alaska (New York, Folkways Record FE4444, 1954), pp. 4-5. Among the most inter-
esting is this one: "All songs have been exhausted. / He picks up some of all / And
adds his own / And makes a new song."

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
6 The Musical Quarterly

Is there, then, an essential difference between

the improvised American Indian materials? Styl
negligible, notation is not present, and it is on
of inspiration in certain cases that allows us to ma
But this kind of distinction can also be made in Western art music,
by comparing, for example, the painstaking and often protracted
method of Beethoven with Schubert's quick, spontaneous creation
of lieder.14 Again, broadly viewed, the styles are not substantially
different. Should we not then speak perhaps of rapid and slow
composition rather than of composition juxtaposed to improvisa-
tion? And would we not do well to think of composition and
improvisation as opposite ends of a continuum, with a Schubert and
a Beethoven at the extremes, likewise with the improvising Eskimo
at one extreme, the Pima at the other, and the Plains Indians
somewhere in the middle?

Having ventured the hypothesis that the juxtaposing of com

position and improvisation as fundamentally different processes
false, and that the two are instead part of the same idea, we can now
turn to the concept of improvisation in various cultures. It obviously
exists in Western music of the present, and, as Ferand shows ove
and over again, it has existed (occasionally perhaps without e
plicit formulation) in the past. Its conscious acceptance among no
Western cultures is less certain. Alan Merriam,15 in discussing com
position in a number of tribal societies, does not bring out that
these cultures single out the improvisatory nature of any of the
musical products; and this applies even to those among them th
articulate ideas about composition. Similarly, an explicit term fo
improvisation seems not to be used in various Asian cultures. The
is some evidence, however, that the thinking of South and Wes
Asian musicians does differentiate the kind of music which we
would regard as improvised from other kinds of music. In Nor
Indian music such terms as badat (growth) and vistar (expansio
are used. In Persian music one type of improvisation, in which
musician modulates from one dastgah (mode) to another - som
thing actually rather unusual - is known as morakkab-kha
(i. e., in the style of the composer-singer).16 The term taqsim, whic
14 Ferand is, of course, aware of the supposedly improvisatory nature of Schube
compositional process, at least in certain instances. See Die Improvisation in der M
pp. 3-4, and the references therein.
15 The Anthropology of Music, chap. 9.
16 Theodore Solis, "The Sarod: Its Gat-Tora Tradition" (M.A. thesis, Uni-

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
 Thoughts on Improvisation 7

denotes a normally nonmetric improvised solo form im

Arabic music, is derived from the term for "division
implying variation and perhaps, with it, improvisation
of types or styles of improvised music in the South India
have specific names, and the performers are evaluated ac
the quality with which they bring out these styles; here
facts indicate that the concept of improvisation is at le
below the surface. The statements by Mantle Hood that
of a Javanese gamelan get paid in accordance with how
their instruments are in providing improvisatory relationsh
main fixed melody17 suggest the same conclusion. Even
examples are enough to point to the possibility that in ce
cultures improvisation is recognized as one concept of m
tion. Yet it is interesting to see that the composition-im
dichotomy is not evidenced; rather, a series of differ
along our continuum are struck. Thus, how much t
performers are required to depart from the fixed melody in
music is a matter of graduation. Thus also in South Ind
where some kinds of improvisation (such as niraval) t
from a composed piece are distinguished from other
alapana and tanam) based only on the characteristics of
the distinction is understood to signify different degrees o
In Persian music the players may (1) perform the r
material learned as a basis for improvisation) with relat
change, (2) perform a free fantasia upon this material wh
it essentially intact, or, again, (3) modulate from one m
other (the kind of performance referred to as morakk
and these three procedures indicate varying degrees of in
for the performer in working from a model, that is, dif
grees of compositional activity on the part of the perfo
short, the lines that different cultures might draw betw
composition and improvisation will appear at different
a continuum.
How do the musicians who improvise regard the differen
among their performances? A few detailed studies conclude th

versity of Hawaii, 1970), p. 55; Nelly Caron and Dariouche Safvate, Iran: Les T
tions musicales (Paris, 1966), p. 128.
17 Mantle Hood and Hardja Susilo, Music of the Venerable Dark Cloud, Introd
tion, Commentary, Analyses (Los Angeles, Institute of Ethnomusicology, UC
1967), p. 14.
18 Caron and Safvate, Iran: Les Traditions musicales, pp. 127-32.

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
8 The Musical Quarterly

musician who repeatedly improvises upon the

does so quite predictably. But of course he hard
the same thing twice. On the other hand, impr
certain cultures tend to maintain that they do
thing each time and to denigrate the idea th
performance to performance.
An example of this attitude can be found in t
a Persian musician who was asked to comment on the fact that two
of his performances of the same dastgah were rather different: h
denied that there was a difference.'9 When confronted with the
concrete evidence of the recordings, he admitted the existenc
the differences, but not their significance, and implied that th
sence of what he performed in a dastgah is always the same.
statements show that in Persian music it is possible to disting
between what is essential and what is not; the former remains con-
stant and is learned, while the latter is simply a result of the per-
former's mood of the moment. This view is not unlike the typical
Western view of improvisation in the seventeenth- and eighteenth-
century music, that is, that the composed piece is the essence and
the improviser's filling in of ornaments and passages is something
of a much lower order of importance.
But Middle Eastern musicians do not always hold such views.
Another Persian musician stressed the importance of never repeat-
ing himself, of never playing or singing a piece the same way. As an
example he cited the nightingale, which is respected by Persian
musicians because it is thought never to repeat itself in its song.
Similarly, an Arabic musician, in commenting on the style of
taqsim, felt that his own performance was basically unstructured
and that he never played anything twice (which is correct in a
narrowly technical sense), so he was surprised at how much his
various performances fell into patterns and types, how predictable
his work was. On the other hand, as though emphasizing the per-
sonal character of a performer's interpretation, some Indian musi-
cians speak of a relatively set, albeit very individual way of im-
provising by saying such things as "I perform Raga X in this way."
19 Most of the remarks on Persian music in this paper stem from my field work
in Iran during 1967, 1968-69, and later. It is impossible for me to acknowledge all the
help which I received from many Persian musicians and others. I should like to thank
the U. S. Fulbright Commission in Iran for supporting me, to the University of Illinois
Research Board and the Center for International Comparative Studies for further sup-
port, and to Nour-Ali Boroumand, Hormoz Farhat, William Archer, and Mohammad
Massoudieh for many kinds of help in Iran.

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
 Thoughts on Improvisation 9

Thus, the attitude of Middle Eastern and South Asian musicians

towards their own improvisation varies greatly, which only adds
confusion to a delineation of the concept. A thorough study that
compares performance and conceptualization on the basis of de-
tailed field research is badly needed. Yet already we who recognize
the notion that there is a difference between improvised and com-
posed material are aware that the improviser, when he performs a
variety of versions of one mode (raga, dastgah, maqam, etc.) is
really doing precisely that - performing a version of something,
not improvising upon something. In other words, he is giving a
rendition of something that already exists, be it a song or a theoreti-
cal musical entity. And its basic "table of contents" is set.
We may again hark back to the American Indians, of whom it
is said that they do not improvise, but simply perform their songs
in numerous variants arrived at by oral tradition. It is sometimes
difficult to see why two rather different performances (without even
the guidance of words) are regarded as variants of the same song,
and why two others that sound practically alike are taken to be
separate musical items. We can only conjecture that the Indians'
idea of musical entities is different from ours. Perhaps the way to
approach the "improvised" music of South and West Asia is like-
wise to say that performers sing or play the same piece, but that
their idea of what is a "piece," a musical unit with its own integrity,
is simply different from ours. In some cultures the concept is nar-
row, in others broad - here the unit is a specified sonata with opus
number, there it is a maqam or a raga. We may then find the
concept of improvisation altogether unnecessary. Instead, we can
say that each musical culture has its set of musical macro-units, e. g.,
songs or pieces, or modes (ragas, maqamat, dastgahs) as performed
by a particular musician, and that the degree to which the sound
realizations of the unit are similar varies with the culture, compris-
ing the system of musical conceptualization, the question of freedom
for the performer, etc. This approach is novel only insofar as it
allows us to think of all musics as having basic musical entities
which exist and are performed, rather than dividing music into
"fixed" and "improvised" types.
Let us look briefly at another aspect of the composition-impro-
visation relationship. In those musics which are said to be impro-
vised a number of compositional techniques and devices at the
microcompositional level appear to be characteristic. Among them
are repetition, simple variation of short phrases, melodic sequence,

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
 10 The Musical Quarterly

the tendency to start two successive sections with

the tendency to increase the length of sections as
progresses, and perhaps others. Now all of these t
present in the "set" or "fixed" compositions of c
their distribution significant? The "fixed" compo
of the cultures that have systematic improvisation,
Iran, exhibit features found in improvisation.
metric devotional songs that are central to the v
mental concert music of the South Indian tradit
have repetition and minor variation of short lin
improvisations upon musical themes taken from
esting, therefore, to find a given culture using es
compositional techniques, whether the material is,
composed or improvised. This fact may eventually
on the ultimate genesis of musical systems. Is it p
cultures which used certain compositional techni
by the very nature of these techniques to develo
provisation? Or did some cultures, perhaps mi
their musical traditions by allowing performers
their renditions of a model, and then permit mu
techniques developed in the "free rendition" perio
history, to devise more stably fixed melodies? T
which, even more than elsewhere in this paper,
other than speculate.
A final observation concerns the use in Western music of the
special techniques whose frequent occurrence in improvisatory
styles is characteristic. It appears that some of these techniques are
especially common in certain periods of music history, particularly
the Baroque, and likewise in the works of certain composers (such
as Mozart, Chopin, and Schubert) who were known to work rapidly
and spontaneously, and whose output was great. It is thus con-
ceivable that, mutatis mutandis, the musical thought processes of a
Schubert are much more closely related to the improvisations of
Indian and Middle Eastern musicians than are those of a Beethoven.
The fact that Schubert wrote down certain of his works rapidly and,
if we are to believe some descriptions,20 without working and re-
working them very much, could lead us to regard his musical think-

20 See, for example, Alfred Einstein, Schubert: A Musical Portrait (New York,
1951), p. 92. Ferand, Die Improvisation in der Musik, p. 4, refers to the possibility that
certain works by Liszt, Chopin, and Schubert may actually be "Fixierungen von Im-
provisationen."

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
 Thoughts on Improvisation 11

ing as basically improvisatory. Thus, rather than making

ence or absence of notation the major criterion for imp
and composition, and dividing music into two spheres, t
provisation-oral tradition and that of composition-nota
could speculate upon the division of the world's musical
and of their subsystems - genres, periods, composers -
groups. One of these would be the music which is carefu
out, perhaps even worked over with a conscious view to i
innovation from piece to piece and even from phrase
the other, that which is spontaneous but model-boun
created, and simply conceived. The first gives up spont
deliberation, while the second eschews a search for innov
favor of giving way to sudden impulse. Neither need be
improvisation, and neither is restricted to any particul
musical or cultural complexity.

Improvisation and the Model

Having apparently done what we could to demolish t
improvisation as a concept separate from composition
now reinstate it for the purpose of examining certain
performance wherein the musician is free to contribut
of his own spontaneous making. The improviser, let us h
always has something given to work from - certain thin
at the base of the performance, that he uses as the groun
he builds. We may call it his model. In some cultur
theoretical terms are used to designate the model: ra
(the basic concepts of melodic and rhythmic organizatio
and other, basically modal configurations - patet in Jav
Balinese gamelan music, dastgah in Iran, maqam in A
Turkish music. In these cultures or others, names of m
formance-forms are used to designate the model: the blu
of chords in jazz, the nonmetric taqsim in Arabic musi
other hand, the model may be a specific composition, an
may in effect designate the model by giving the names
jazz, of kritis in South India, of specific songs in the Yu
tradition. Or, again, styles indicating some specific pitch or
content (figured bass in Western music, or abadja and k
rhythmically distinctive styles of West African drumm
reveal the existence of an improvisatory model and its r
as such by its culture. The same is true of such styles

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
12 The Musical Quarterly

gatt (nonmetric or metric) in North India, an

tanam, niraval, and svara kalpana (three types
distinguished mainly by the treatment of text a
each style here being defined by a certain type
relatively rare instances of polyphonic improvisat
be a tune sung by one voice (against which the
provise) and a set of allowable harmonic interval
characteristic sequences. There are, then, very
models used in the world of improvisation. Thes
like to examine comparatively, in accordance wit
in the two continua of density and audibility.
There would appear to be a wide gulf betwee
which is basically a temporal event with beginni
end, and a raga, maqam, or dastgah, which is at f
to be a theoretical construct, something a perform
as a unit, keeping all its elements more or less in
But to these "theoretical" models of Asian music it turns out that
musicians take a time-oriented approach, considering them object
that have beginnings and endings, and that consist somehow
sequences of thought. Thus we may take it that each model, be
a tune, a theoretical construct, or a mode with typical melod
turns, consists of a series of obligatory musical events which mu
be observed, either absolutely or with some sort of frequency,
order that the model remain intact. For example, in a jazz im
provisation based on a series of eight chords, these chords must
themselves appear in some acceptable form, and in order. In a pe
formance of Perisian avaz, the nonmetric improvisation that is ce
tral in a complete performance of Persian classical music, certai
signposts or points of reference must appear;21 these are certai
central tones, opening and closing motifs, melodic indications sig
nalling the coming of closing sections, etc. It is possible to make
list of these points of reference - a kind of table of contents of t
mode. All of them need not appear in a performance, but som
must be evident, in order for the performance to warrant cred
bility. Similar points of reference appear in the taqasim of Arabi
music. These include not only beginning and cadential motifs bu

21 The best explanation of these signposts appears, passim, in Hormoz Farha

"The Dastgah Concept in Persian Classical Music" (Ph.D. diss., UCLA, 1966). See a
Bruno Nettl, with Bela Foltin, Jr. Daramad of Chahargah: A Study in the Performan
Practice of Persian Music (Detroit, 1972), pp. 25-28, and the just published work b
Ella Zonis, Classical Persian Music: An Introduction (Cambridge, Mass., 1973).

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
 Thoughts on Improvisation 13

also the lengths of sections, signposts of a different kin

which, are at least to a degree obligatory; thus, a long se
come at the end of a taqsim, and possibly about the mi
whole piece as well. The improviser and the listener use
section as a way of gauging the progress of the piece. On
hand, in the model of Indian alap (the nonmetric int
exposition of a raga) as in Persian avaz and Arabic ta
gradual introduction of higher tones functions as point
erence.

Diverse though these points of reference may

close together or far apart they are may be meas
roughly. We can refer to this measurement a
paring various types of models, we find that tho
tively dense, those of Persian music, of medium
of an Arabic taqsim or an Indian alap, relatively
Figured bass, and Baroque music in which a s
ornamentation, are perhaps the densest mode
density of a model have an effect on a perform
seems likely that a performer of improvisation u
tends to vary less from performance to performanc
model lacks density, and that the kinds of impro
that goes on in various musical cultures, differen
in style and content, can therefore be compared.
A musical repertory, composed or improvised
as the embodiment of a system, and one way of
system is to divide it theoretically into its comp
units are, as it were, the building blocks which
lates, and which musicians within the tradition make use of,
choosing from among them, combining, recombining, and re-
arranging them. These building blocks are, even within a single
repertory, of many different orders. They are the tones selected
from a tone system; they are melodic motifs; they are harmonic
intervals and interval sequences in improvised polyphony; they
are types of sections (e. g., the exposition of sonata forms). These
few examples show how greatly varied they are in extent and size.
Studying the building-block components of improvisatory styles
is similar to studying density, since the "points of reference" that
govern density are in some cases identical with the blocks, or with
demarcations between them. Among the various improvisatory
styles already mentioned, building blocks may be illustrated by
the following: (1) In some West African drum ensembles each of

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
14 The Musical Quarterly

several percussionists play a single short motif

posing it to the others in a highly complex poly
The master drummer uses these motifs, mixing
them, drawing on them, one by one; they there
ing blocks, and they are of a relatively short
size. (2) In the Yugoslav epic tradition short mu
accompany the individual ten-syllable lines are
A singer sings one line again and again, though
tions, until for a number of reasons a new but
melody is taken up. These themes are among th
of the Yugoslav tradition. (3) In North Indian d
role is filled by the theka, the specific audible
with a tala, the theoretical metric pattern.
In the Arabic taqsim, as in some other styles,
building blocks can be observed.22 We have, fir
of the maqam, from which the performer draws m
in any order as long as the melodic movement is
a higher level there are motifs of three to five
sociated with each maqam; these evidently mu
occasionally. Beyond this the taqsim is composed
of sections, most easily identified by length; the
even their length, are building blocks, for a mu
them in certain kinds of order and carry out in eac
musical function--the long sections contain mo
sidiary maqams, the shorter ones serve to establish
and the shortest ones provide a kind of dramatic
The point to be made is that in both folk and
tories composers use similar kinds of building bl
of Czech folk songs, for instance, provided evid
number of motifs appears over and over again, a
the songs; in fact, the entire repertory can be r
set of short phrases from which composers see
And, on a large scale, composers of Western art m
to have used similar procedures.
However, one certainly receives the impression
of building blocks in an improvised style is mu
than in "composed" music, and that perhaps

22 This information is based in part on a study of sixteen

Nahawand as performed by Jihad Racy, carried out by the au
with sponsorship of the University of Illinois Research Board

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
 Thoughts on Improvisation 15

repertory can be developed only if the options are

terestingly enough, the building blocks sometimes op
cosmically in the manner of the entire model in macro
a Persian musician does in a single gusheh (a section of
with specific range and thematic content) is a good de
takes place in the entire performance; the relationship
ponents of a gusheh very much resembles the relation
gushehs in the whole piece. In the Arabic taqsim and t
alapana the way in which a tone acts as the basis for em
and thereby acquires emphasis is similar to the way in
levels and ranges such as tetrachords are explored and
as focal points in the entire model. Large building bloc
performer considerable freedom for internal variatio
section type of the taqsim is subject to much more var
say, the ornamented note in a seventeenth-century pi
gree to which the improviser actually makes decisions
sudden inspiration that some regard as the hallmark of
tion, is partially dependent on the size of the building
larger the blocks, the greater the internal variability.
Points of reference and building blocks are definitive
istics of the model on which improvisation may be bas
gard to both, improvisatory music show itself to be d
degree, but not in essence, from the traditions on w
posed" music is based. The same is true where "audibil
model is involved. We are dealing with a wide rang
types: notes, cadential figures, section types identified
melodic phrases, or lines, rhythmic lines or formulas, e
chord sequences, and modal concepts to which are atta
group of traits - scales, motifs, and typical sequences of
in range and tonality, as well as rhythmic tendencies.
these a question arises: to what extent does the mod
the material that is actually heard by the student or p
This question directs us of course to the role of t
each of the improvisatory styles that we have consider
dent must in some way learn the model before he can
upon it, though the formality of his instruction vari
and Persian classical music it is intense; in at least some Arabic
music and in jazz it is mainly a matter of listening and trial-and-
error participation; in Yugoslav epic-singing it is intermediate to
these methods, for songs are learned from a specific teacher whose

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
16 The Musical Quarterly

role is critical.23 But more important for us here

the model itself. In some systems it is actual mu
be performed without improvisation, in some
material that the musician learns but does not execute in a true
performance, and in still others it is largely theoretical su
matter, consisting of verbal instructions and exercises.
Models exhibiting more or less comprehensive audibility
exemplified by the show tune that becomes the basis for
improvisation and by the Yugoslav epic song that is learned in
before it becomes subject to variation by the performer. Slig
less "audible," in the sense that it is rarely regarded as somet
to be actually performed, is the radif of Persian classical mu
long series of compositions that are in style not different from
improvisations, that are developed by each teacher from the c
mon tradition and taught to students who must memorize th
before going on to improvise. Performances of Persian music
easily be classified according to how much they depart from
radif, or teaching version (and some performances depart very
from it).
Of medium audibility are the kinds of models that consist
partly of theoretical concepts that are taught, and partly of short
building blocks such as motifs. The improviser does not perform
in his own way a piece that already exists, but uses materials that
have been learned to some extent in isolation, or at least in a form
which makes it impossible simply to reiterate them. Two examples
of these practices are (1) some aspects of the classical music of
India (whose models are incredibly complex, numerous, and multi-
faceted) in which exercises that provide insight into the character-
istics of a raga are taught, and (2) certains facets of Arabic music,
in which the practitioners may learn the theory of the maqamat,
but focus, in their practical study, on musical motifs and other
matters that are not by themselves performed as music.
"Inaudible" or "minimally audible" models exist in a number
of cultures, but particularly in Western music. For example, while
the pianist for a silent movie typically had at his disposal certain
chords, chord progressions, and melodic materials, he did not use
any traditional model of his '"score" (his performance as a whole)
or even of sections of any considerable length. Similarly, the or-
ganist who performs a fugue extempore on a theme given to him

23 Albert B. Lord, The Singer of Tales (Cambridge, Mass., 1960), pp. 13-29.

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
 Thoughts on Improvisation 17

by the audience has, aside from the theme itself (and t

him), no model more specific than whatever is generic
Some improvisers in avant-garde music of the 1960s wis
the use of any model, and if, parodoxically, if the inno
is to be the model, it can hardly be considered an au
Thus, while a model of some sort is a necessary condit
improvisation, the audibility of the model, like its den
from culture to culture and from repertory to reperto

Improvisation and the Performer

It is axiomatic that improvised music requires a great
effort on the part of the performer than does composed m
indeed, one way we may perhaps define improvisation is
ing the degree to which the performer is creatively in
us consider what is known about the way in which per
repertory differ from each other in the performance
based on one model, and to what extent the performan
individual differ from each other.

A number of studies have been made investigating this problem

Among the most significant or relevant are those of: (1) Katz,24
who compares two generations' performance practice in one genre
with particular regard to ornamentation; (2) Nettl and Riddle,25
in who explore the work of one performer and his range of pra
tice in one Arabic maqam; (3) Nettl and Foltin,26 who investigate
the range of a number of performers of one Persian mode, i
cluding only the nonmetric improvisations; (4) Massoudieh,27 who
task is similar but involves a different, smaller sample; (5) Wi
kens,28 who compares one amateur and one professional musician
in the Persian tradition; (6) Touma,29 who tries to comprehend th
scope of performance of one Arabic maqam, Bayati, as used in sev
eral forms and by several musicians; (7) Zorana Ercegovac,30 whos
24 Ruth Katz, "The Singing of Baqqashot by Aleppo Jews: A Study in Music
Acculturation," Acta Musicologica, XL (1968), 65-85.
25 Bruno Nettl and Ronald Riddle, "Taqsim Nahawand, a Study of Sixteen Per
formances by Jihad Racy," unpublished paper.
26 Bruno Nettl and Bela Foltin, Jr., Daramad of Chahargah.
27Mohammad Taghi Massoudieh, Awaz-e-Sur (Regensburg, 1968).
28 Eckart Wilkens, Kiinstler und Amateure im persischen Santurspiel (Regens-
burg, 1967).
29 Habib Touma, Der Maqam Bayati im arabischen Taqsim (Berlin, 1968).
soZorana Ercegovac, "The Song of Baghdad: A Comparison of Nine Perform-
ances" (M.A. thesis, University of Illinois, Urbana, 1973).

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
 18 The Musical Quarterly

unpublished master's thesis deals with excerpts of

one song by three singers, on nine occasions, in
Yugoslav epic poetry; (8) Reichow,"1 who tries to
notion of a maqam on the basis of both compositi
ance; (9) Gerson-Kiwi,32 who explores the Persian
on the basis of one musician's work; and (10) Rycr
the work of one East African guitarist. This is by
plete list, but it represents much of the work d
focuses on the degree of variation in performanc
or one tightly knit group, or on the basis of on
ority of such studies have been based on Middle E
curiously few dealing with Indian music or jaz
duced. But perhaps the nature of model performa
retical system, or the soloistic quality of the Mid
is the reason for this imbalance.

In any event, clearly ethnomusicologists are only at the begin-

nings of their study of improvisational technique. More inform
tion on individual performers and on performances based on on
model, and then comparisons of these, are needed before muc
insight can be gained. But what we already have at hand indicate
some trends for which future scholars may watch. First, perform
ances by a given musician working with one model vary far less
those of different musicians. This is obviously what a reasonable
guess would have been; it is apparent that an individual musician
establishes a personal practice from which he does not depar
very much. Cursory examinations of Indian music34 show this.
Second, most cultures have a rather specific set of expectations of
the performer, including the requirement of sticking reasonabl
close to the model. The musician who is highly creative and trie
to avoid using the points of reference and the building blocks o
the model is chastised for his ignorance of the model; an identica

31 Jan Reichow, Die Entfaltung eines Melodiemodells im Genus Sikah (Regens

burg, 1971).
32 Edith Gerson-Kiwi, The Persian Doctrine of Dastga Composition (Tel-Aviv,
1963).
33 David Rycroft, "The Guitar Improvisations of Mwenda Jean Bosco," African
Music, III/1 (1962), 86-102.
34 I1 am indebted to Daniel Neuman for information and discussion of the con-
cept of improvisation in Northern Indian music. See, however, N. A. Jairazbhoy, Th
Rags of North Indian Music (Middletown, Conn., 1971), p. 31, which implies agre
ment with the point made here, but which also indicates that "when the musician is
performing beyond his normal capacity . . . the music becomes alive."

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms
 Thoughts on Improvisation 19

performance each time is almost equally unacceptabl

haps not quite as culpable, and this point is of great
cause it strengthens our view that the improviser is re
the performer of a traditional piece, establishing his o
rendering it. Third - and here the difficulty of generalizin
perhaps most evident - performers tend to feel that they a
an extant piece, not improvising with mere reference t
extant. An exception seems to occur in certain improvisa
of Western music, which is music that has all along sin
improvisation as a separate, minor art, the sphere of th
tively few musicians competent to engage in it.
We might close with the observation that the recent
non-Western music have indeed added enormously to o
standing of improvisation. But most of these studies - t
above are the exceptions - concern themselves with the
tion and explication of models, not with the ways in w
cians deal with a model and depart from it. Even th
that are concerned with the latter topics take a prescr
tude, indicating what may be done; few examine, thro
parison of field records, what actually is, or has been, d
But the conclusion which recurs again and again in ou
is that perhaps we must abandon the idea of improv
process separate from composition and adopt the vi
performers improvise to some extent. What the pian
Bach and Beethoven does with his models - the scores and the
accumulated tradition of performance practice - is only in d
not in nature, different from what the Indian playing an a
Rag Yaman and the Persian singing the Dastgah of Shur do
theirs.35

35 This study was carried out with the help of the University of Illinois R
Board which provided research assistance. I am grateful to Doris Dyen Root for
compiling bibliography and abstracting literature which was in part used here. The
paper was written while the author was an Associate of the University of Illinois
Center for Advanced Study.

This content downloaded from 213.22.1.232 on Wed, 06 Jun 2018 18:42:38 UTC
All use subject to http://about.jstor.org/terms