Methods of Analysis in Musicology

Author(s): Nicolas Ruwet and Mark Everist

Source: Music Analysis , Mar. - Jul., 1987, Vol. 6, No. 1/2 (Mar. - Jul., 1987), pp. 3-9+11-
36
Published by: Wiley

Stable URL: [Link]

REFERENCES
Linked references are available on JSTOR for this article:
[Link]
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@[Link].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
[Link]

Wiley is collaborating with JSTOR to digitize, preserve and extend access to Music Analysis

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
NICOLAS RUWET

METHODS OF ANALYSIS IN MUSICOLOGY

translated and introduced by Mark Everist

INTRODUCTION

Nicolas Ruwet's 'Methods' deserve a broader reception than the

been afforded. When they appeared in 1966, they represented the f
attempt to articulate a music-analytical system which drew on the d
and taxonomic procedures of anthropology, linguistics and ethn
they also form a large part of a system which has generated m
comment, especially in the French and French-Canadian musi
worlds,'* in its twenty years' existence; furthermore, they constitu
few sets of analytical methodologies which initially address repert
than those of the 'common practice' era. The concentration in
monodies from the twelfth to the fourteenth centuries makes their exclusion
from contemporary discussions of the 'analysis of early music', with its all-too-
often duplicative obsession with voice-leading procedures,2 all the more
regrettable.
Ruwet himself admits to two types of influence in the evolution of his
methods: the mostly oral teaching of Andre Souris and Pierre Froidebise,3 and
the less intangible, better known, work of Gilbert Rouget, Roman Jakobson and
Claude Levi-Strauss.4 The relationship between the work of these authors and
Ruwet's was discussed in detail by Jean-Jacques Nattiez in 1975 as a preamble
to his systematic overhaul of Ruwet's 'Methods' in Fondements d'une simiologie
de la musique. 5
The most significant response to the original (1966) publication of Ruwet's
'Methods' was, however, from the ethnomusicologist Simha Arom, whose
19696 article concentrated initially on Ruwet's notation but went on to discuss
alternative views of the segmentation of the first piece discussed by Ruwet, the
fourteenth-century Geisslerlied. Whether Arom's rectilinear analyses are clearer
or more productive than Ruwet's paradigmatic presentations still seems to
be an area for further inquiry, although there seems little doubt that such
explanations as those offered, for example, by Ruwet's Ex. Id are handled with
* For notes to the introduction, see page 7 below.

MUSIC ANALYSIS 6:1-2, 1987 3

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

much greater finesse in rectilinear analysis. Arom's c

the view of Jean Molino, so influential on Nattiez's F
procedures do not supply a key, but a bunch of ke
analytical possibilities...',7 and played an important ro
of Ruwet's hierarchic structures.
At the end of the first paragraph of his Geisslerlied analysis, Ruwet states that
'clearly, it would be very difficult to apply the procedure to the presentation of
polyphonic structures'.8 This provoked one of the few examples of a strict
attempt to apply Ruwet's methods to a repertory which he had already
acknowledged as problematic: Jean-Michel Vaccaro, leaning on both Ruwet's
'Methods' and his contribution to the function of text in vocal music,9 produced
in 1975 an analysis of the polyphonic chanson by Guillaume Costeley to a text by
Ronsard: Mignonne allon voir.10 This is not the occasion to discuss the
relationship of Vaccaro's 'Proposition' to Ruwet's 'Methods', except to note the
silence which has greeted these endeavours from those who seek to explain the
musical processes of the sixteenth-century chanson and of Renaissance music in
general.
In the same year as the appearance of Vaccaro's 'Proposition', David Lidov
contributed two studies to the theoretical tradition created by 'Methods', one
which reviewed and revised the analytical system, particularly with regard to
Ruwet's interpretation of Kalenda maya and Guiot de Provins' Molt me
mervoilt," and another which furthered the discussion with another medieval
monody, by Li Tresoriers de Lille (Pieros li Borgnes): Haut honor d'un
commandement.12 Here, perhaps even more than in Ruwet's analyses, the
technical problems posed by the original notation take on greater significance;
they are particularly pressing, for example, in Lidov's account of rhythmic
figures.13 This, along with the relevant Ruwet analyses, points to the urgent
need for a review of the texts of these analyses as much as their methodology.14
It would be wrong to suggest that music before 1600 has had a monopoly
of subsequent discussions of 'Methods'. Gilles Naud's studies of Xenakis'
Nomos Alpha15 not only parallel Nattiez's study of Varese's DensitJ 21.5 but also
take Xenakis' own comments on musical segmentation as a starting point for the
analysis. Whilst Naud eschews Ruwet's paradigmatic display of the musical
data, the analysis of Nomos Alpha opens up fascinating lines of inquiry which
depend directly on 'Methods'.
Almost contemporary with the work of Lidov and Vaccaro was a series of
analyses of secular music by Machaut, designed to demonstrate 'pitch
patterning of a quasi-ostinato character', which were produced by Lawrence
Gushee in September 1974.16 The sceptical response to these analyses prompted
a further contribution from Gushee in 197517 which included a series of
'paradigmatic diagrams' of some monophonic chansons of Adam de la Halle and
monophonic dance music. These were subsequently published in 1982.18 Given
the obvious distributional nature of Gushee's analyses, it is curious that the
work of Ruwet, and the large body of literature which had been thus engendered
by 1982, was consigned to a single sentence:

4 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

One should refer to the recent (sic) application of the

Ruwet to a 14th-century Geisslerlied, ultimately
author's ignorance of the documentary evidence (emp

Gushee does not explain Ruwet's 'ignorance', but

to Ruwet's statement: 'Since I am not interested here in problems of
transcription, I take the transcriptions as data, without prejudging their
validity'.20 This seems strange in view of Gushee's self-confessed original use of
Wilkins' edition of Adam de la Halle's chansons.21 Unlike Gushee, however,
Ruwet is not primarily concerned with the elucidation of a single song, be it by
a troubadour, trouvere or flagellant, but with analytical method (hence his title).
Nevertheless, there are indeed problems with the texts that Ruwet selects for his
analyses, problems which centre on the question of the rhythmic evaluation of
thirteenth- and early fourteenth-century notation. The influence on Ruwet of
Friedrich Gennrich and Gustave Reese reflects the latters' view of the rhythmic
structure of medieval song; in all cases except the Geisslerlied,22 this involves the
assumption that the application of the principles of modal rhythm is appropriate
to this repertory. There have been many challenges to this assumption,23
although there is still a great reluctance to overthrow the modal orthodoxy. The
result of this schismatic view of the rhythm of trouv re and troubadour song is
that analyses of this music which use editions employing modal rhythm are
likely to be condemned for inaccuracy. In the specific cases discussed in
'Methods', the issue is further complicated by the fact that, in Ruwet's stage
(b)24 of his analytical procedure, the total duration of each segment is taken as
an index for the further segmentation of level I units. However, it is
unconstructive simply to dismiss an analytical methodology out of hand on
these grounds alone. The assessment of the degree of variation between, on the
one hand, an analysis based on a 'free declamatory' edition of the song and, on
the other hand, Ruwet's original analysis would make a valuable contribution to
the development of the analysis of medieval song.
More significantly, perhaps, Gushee exaggerates his response to Ruwet's
dependence on the philosophies of Popper and Hayek. Gushee writes:

A strictly applied paradigmatic technique rests on concepts of identity or

similarity with respect to pitch letter name patterns .... It is a technique for
segmentation, replacing other criteria - cadences, proportions, text
structure - with that of pitch recurrences, and pretends to isolate minimal
formal elements.25

The idea that Ruwet's paradigmatic technique replaces other criteria with
'pitch letter name patterns' seems extreme, and adjusting 'replacing' to
'complementing' might be more accurate. More important, however, is the
concentration on the apparent 'strictness' of this 'paradigmatic technique'.
Ruwet himself says (in 1966), referring to and quoting Zellig Harris in extenso,
that 'the procedure is much more one of verification, meant to ensure that the

MUSIC ANALYSIS 6:1-2, 1987 5

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

analysis is coherent, than a discovery procedure in the

Nattiez's discussion of Ruwet's 'Methods' and their accretions, in
Fondements, was the most comprehensive to date.27 Synthesising the
distributional ideas of Ruwet and Arom with Jean Molino's concept of
tripartition, Nattiez gave a revised form of Ruwet's 'Methods' pride of place in
his 'neutral level' and generated analytical examples which have served as
models for subsequent inquiry.28 The relationship between Ruwet's 'Methods'
and Nattiez's Fondements is obscured by the former's recantation in Musique en
jeu.29 If the work of Nattiez, Lidov and others provided the stimulus for his
retraction, a more deep-seated explanation was given by Ruwet himself:

For myself, I have serious doubts concerning the validity and interest of
inductive procedures; I would adopt a more rationalist and more
'theoretical' procedure; I believe in the possibility and the necessity of the
research of universals. I distrust relativism and behaviourism, etc. There is
certainly no question here of offering another critique of empiricism and of
positivism. Others have done it, better than I could ever do. I shall refer only
to the classic texts on the question . .30

Ruwet's 'classic texts' are Noam Chomsky's theoretical monographs on

generative grammar31 and the more popular works of Karl Popper and
Friedrich Hayek.32 More specifically, for Ruwet:

A more serious reading of Popper has convinced me that the ideal of the
tabula rasa is illusory and that, no matter what we do, as soon as we reflect
on any subject, we always approach it with preconceived theories.33

The question of the significance of positivism and empiricism in Ruwet's

'Methods' and in Nattiez's use of them strikes a chord with Gushee's apparent
criticism of Ruwet's 'Methods' as a 'strictly applied paradigmatic technique'.
Ruwet's critique of his original (1966) ideas is puzzling. It astonished Nattiez,
who asked if 'Ruwet has not substituted for a reading of [an earlier version of
this part of Fondements] the sometimes rapid criticisms aimed by Chomsky and
his disciples at the taxonomic perspective'.34 Nattiez continued, paradoxically
making the same point as had Ruwet in 'Methods':

It is absolutely wrong to pretend that the classificatory procedure [la dicarche

classificatoire] . . . appeals neither to intuitions nor to hypotheses (Nattiez's
emphasis).35

Whatever the confusions surrounding Ruwet's retraction, there is no doubt

that his 'Methods' of 1966, albeit in the guise of Nattiez's neutral level, and not
Ruwet's 'Thdorie et mrthodes' of 1975 have had the more lasting influence.
Both are discussed with relative impartiality in more detail in Reinhard
Schneider's review in Semiotik der Musik36 of Ruwet's contribution to music

6 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

semiotics, although the difference between the

'Thdorie et methodes' is perhaps underplayed.
There will be many objections to the viewpoin
Certainly the emphasis on modal categorisation
equally true that many of the suggestions explic
investigation of parametric/non-parametric e
application of this approach to other repertories), as
far only been hinted at (comparative studies and
empiricism and pragmatism in the context of this an
be taken up.37
Devotees of voice-leading procedures have held
analysis of 'early' music for too long at the expen
productive, lines of analytical inquiry. Ruwet's 'M
familiar to any semiotician of music; they should al
exposure in the analysis of music composed before 1

NOTES TO THE INTRODUCTION

1. Ian Bent offered the most sympathetic treatment of Ruwet's 'M

'Analysis, III, 7: Distribution', The New Grove Dictionary of Music and
ed. Stanley Sadie (London: Macmillan, 1980), Vol. 1, pp.377-8
summarised Ruwet's analysis of Guiot de Provins' Molt me mervoil.
2. Perhaps the best articulation of this point of view was given by Saul
Analysis of Pre-Baroque Music', Aspects of Schenkerian Theory, ed
(New Haven: Yale University, 1983), p.133: 'We have no other rec
understanding the music of the past but to rely upon what Schenker h
Its validity is unquestionable; its limitations, none'. Whilst the contex
article would seem to preclude any discussion of techniques not ass
voice leading, such unquestioning acknowledgement of the prim
procedures should not go unchallenged.
3. Both Souris and Froidebise held positions respectively at the cons
Brussels and Liege. With the exception of the encyclopaedia arti
Ruwet in the footnotes of 'Methods', none of Souris' voluminous
assist in tracing the influences on Ruwet's thinking.
4. See p.11 and note 4.
5. Jean-Jacques Nattiez, Fondements d'une simiologie de la musique, S&ri
(Paris: Union G6nerale d'Editions, 10/18, 1975), pp.240-4.
6. Simha Arom, 'Essai d'une notation des monodies " e fins d'analys
musicologie, Vol. 55, 1969, pp.172-216.
7. Nattiez, Fondements, p.254.
8. See below, p.20.
9. Nicolas Ruwet, 'Fonction de la parole dans la musique vocale', Rev
musicologie, Vol. 15, 1961, pp.8-28; reprinted in Langage, musique, poesi
po6tique (Paris: Seuil, 1972), pp.41-69.

MUSIC ANALYSIS 6:1-2, 1987 7

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

10. Jean-Michel Vaccaro, 'Proposition d'un analyse po

xvie siecle', Revue de musicologie, Vol. 61, 1975, pp.35-5
11. David Lidov, 'Musical Phrase Structure in the Theo
Meyer, and Ruwet', On Musical Phrase (Montreal: U
[Internal Publication], 1975), pp.35-77.
12. Lidov, 'Syntactical Strata in Music', ibid., pp.79-85
13. As is the case with Ruwet's Geisslerlied analysis, Lidov
in Gustave Reese, Music in the Middle Ages (New Yo
in his analysis of Haut honor d'un commandement. The
there is only one of a number of possible versions of th
Bibliotheque de l'Arsenal 5198, p.232.
14. See below, pp.4-5.
15. Gilles Naud, 'Pour une methode d'analyse du n
musicale', 2 vols (These de Maitrise, University of
d'une analyse semiologique de Nomos Alpha', Musiq
pp.63-72.
16. Lawrence Gushee, 'Two Central Places: Paris and the French Court in the Early
Fourteenth Century', paper read at Symposium '"Peripherie und Zentrum" in der
Geschichte der ein- und mehrstimmigen Musik des 12. bis 14. Jahrhunderts',
Berlin, September 1974; published in Gesellschaft fiir Musikforschung: Bericht iiber
den internationalen musikwissenschaftlichen Kongress Berlin 1974, ed. Hellmut Kfihn
and Peter Nitsche (Kassel: Birenreiter, 1980), pp. 147-8; see also the discussion on
pp. 153-7.
17. Lawrence Gushee, 'Analytical Method and Compositional Process in some
Thirteenth and Fourteenth-Century Music', paper read at symposium 'Aktuelle
Fragen der musikbezogenen Mittelalterforschung', Basle, 1975; published in
Aktuelle Fragen der musikbezogenen Mittelalterforschung: Text zu einem Basler
Kolloquium des Jahres 1975, Forum musicologicum: Basler Beitrige zur
Musikgeschichte 3 (Basle: Amadeus, 1982), pp. 165-91.
18. See above, note 17.
19. Gushee, 'Analytical Method', p.171.
20. See below, p.20 and note 23.
21. Nigel Wilkins, ed., The Lyric Works of Adam de la Halle, Corpus mensurabilis
musicae 44 (n.p.: American Institute of Musicology, 1967); see also Gushee,
'Analytical Method', p. 169.
22. Runge's edition of this piece (see below, note 23) is a compromise between
diplomatic facsimile and edition with equivocal indications of rhythm; the
rhythmic transcription used by Ruwet is the responsibility of Reese.
23. The best review of this question and summary of the challenges to the assumptions
of modal rhythm is in Hendrik van der Werf, review of Hans Tischler and Samuel
Rosenberg, eds, Chanter m'estuet: Songs of the Trouveres (London: Faber and Faber,
1981), in Journal of the American Musicological Society, Vol. 35, 1982, pp.539-54.
24. See below, p.18.
25. Gushee, 'Analytical Method', p.172.
26. See below, p.20.

8 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

27. Nattiez, Fondements, pp.239-78.

28. Ibid., pp.297-356.
29. Nicolas Ruwet, 'Thdorie et methodes dans les etu
remarques retrospectives et prdliminaires', Musique enjeu,
30. Ibid., p.12.
31. Noam Chomsky, Syntactic Structures, Janua linguarum
van Wijk dedicata 4 (The Hague: Mouton, 1957); Aspects of
(Cambridge, Mass.: MIT Press, 1965); Language and
Harcourt and Brace, 1968); Studies on Semantics and Gener
Linguarum: studia memorie Nicolai van Wijk dedicata -
Hague: Mouton, 1972).
32. The works cited by Ruwet are Karl Raimond Popper, Conj
The Growth ofScientific Knowledge, 3rd ed. rev. (Lond
Paul, 1969); The Logic of Scientific Discovery, 6th ed. re
1972); Objective Knowledge: An Evolutionary Approac
Clarendon Press, 1979); Friedrich A. Hayek, The Sensory Or
Foundations of Theoretical Psychology (London: Routledge
Studies in Philosophy, Politics and Economics (London: Rou
1967).
33. Ruwet, 'Th6orie et Methodes', p.13.
34. Nattiez, Fondements, p.256.
35. Ibid.
36. Reinhard Schneider, Semiotik der Musik: Darstellung und Kritik, Kritische
Information (Munich: Wilhelm Fink Verlag, 1980), pp. 179-228.
37. Furthermore, it is regrettable that such a study as David Halperin, 'Distributional
Structure in Troubadour Music', Orbis musicae 7, 1979/80, pp. 15-26, which leans
heavily on Harris, for example, apparently refuses to acknowledge the existence
either of Ruwet's own work or its derivative literature.
38. I would like to acknowledge the assistance of John Taylor (King's College London),
who kindly read the translation and offered much advice.

MUSIC ANALYSIS 6:1-2, 1987 9

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

METHODS OF ANALYSIS IN MUSICOLOGY

For Andre Souris [ 1966]

In every semiotic system,* the relationship between code and message can be
described from two different points of view, depending on whether one
proceeds from the message to the code or from the code to the message.1
In the first case, the procedure is analytic; in principle, it is indispensable
whenever, as in the case of an unknown language, of exotic music or myths, etc.,
the message alone is given. The work of the analyst then consists of
deconstructing and manipulating the corpus (all given messages) in various
ways in order to derive the units, classes of units, and rules of their combination
which together constitute the code. The crucial problem here is that of
discovery procedures, in other words, analytical criteria. For twenty years,
Structural Linguistics - at least in Denmark and in the United States - was
preoccupied almost exclusively by these problems and elaborated various
analytical models based on such explicitly defined criteria as the principle of
commutation in the glossematic school, or that of contextual substitution in
American distributional analysis.2 An outline of the discovery procedure may be
found applied to myth in Levi-Strauss;3 more recently, researchers have tackled
the problem in semantics as well as in stylistics.4
Once the code has been deciphered, a reverse procedure allows the generation
of messages from this code according to rules of derivation which can
themselves be rigorously clarified.5 Thus, in contrast to an analytical model, a
synthetic model is available which proceeds from the most abstract and general
elements and results in specific messages. From this point of view, the grammar
of a language, when formulated synthetically, appears as a sort of machine
capable of generating all - and nothing but - admissable, 'well-formed' or
'grammatical' sentences in that language. At first sight, the synthetic model

* Nicolas Ruwet, 'M6thodes d'analyse en musicologie', Revue beige de musicologie, Vol. 20, 1%6, pp.65-90; reprinted in
Langage, musique, poisie, Collection po6tique (Paris: Seuil, 1972), pp. 100-34.

MUSIC ANALYSIS 6:1-2, 1987 11

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

offers nothing new; it is simply the mirror image of t

use is to prove the validity of the analytical model. It
the latter's faithful representation of the facts and,
productivity; if the analytical model is correct, its syn
generate messages which did not appear in the or
definition) but which will be recognised by subjects a
In fact, this conception of the relationship betw
approximately that of Hjelmslev - is oversimplified
more fundamental purposes. As theoreticians of g
shown, it seems very difficult to formalise discovery
and the rigorous application of such tests as commuta
leaves remainders; these can be reduced only by
considerations of a quite different order - such as th
applied to the whole of the system (the code). On the
to hold that the initial data of the analysis amount to
Hjelmslev's 'yet unanalysed text', to the American
statements) which would constitute the only mean
soon as a rather limited objectivity is abandoned, one r
more varied data at his disposal, for example, all
judgments made by the subjects upon the code which
furnish a complete series of indices to the structure o
Thus, although the establishment of the code con
existence of analytical procedures, these will be nec
multiple, and it will be only at the level of the formu
that the code can be described uniformly with maxim
simplicity. Furthermore, as experience has shown, th
models have always had a static character hardly suited
of fundamental problems, that of creativity, of prod
semiotic systems, and that of the universal laws whic
code consists essentially of two parts: inventories of el
combination and operation. Now, analytical mo
inventory, whilst neglecting the question of rules. He
hence also their lack of universality: it is at the level o
that languages (or musical systems) diverge most,
direct these elements present a much more general ch
It was necessary briefly to indicate the limitations of
undertakes to establish discovery procedures in music
over apparently very difficult methodological pr
ultimately limited: at the level of the synthetic mode
longer relevant. Thus, a question - which can i
preoccupied linguists for a long time: must th
proceeding 'from top to bottom' or conversely 'f
Hjelmslev, the given data comprises the unanalysed te
can be very large, as great as the sum of phrases utter
the analysis takes the form of a progressive separatio

12 MUSIC ANALYSIS 5:2-3, 1986

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

smaller parts defined by their mutual relationships -

elements which are ultimately irreducible. Harris, o
from relatively brief statements and first derives m
which he groups progressively into greater clas
syntagms, sentences) in order finally to attempt di
of these two types of analysis, when applied to the s
but, since, in any case, a single procedure can never
the two procedures ceases to be crucial: in practi
mixed. Of course, it is very useful to have envisage
of the use of such and such a particular procedure.
Notwithstanding these reservations, it rema
discovery procedures, even if partially insufficient,
guarantee that the synthetic model will not change i
history of linguistics bears witness to their nece
reached the present stage of generative grammar if
analytical research had not liberated it conclusiv
normative systems of traditional grammar.

II

Let us now consider the present state of musicology

distinction between the two models. It may be no
problem of this distinction has never been raised; b)
ever been explicitly elaborated; c) that musical an
example the one given by Pierre Boulez of The Rite o
the discovery criteria on which they depend. In
harmony, fugue, etc., are analogous to tradition
synthetic, only partially explicit, and tainted wit
known. More strikingly, the least contestable suc
field of western tonality - Gevaerts - as well as in th
or rhythms - Brailoiu9 - have received a synthetic f
always presented starting from the most abstract e
circle of fifths, for example, according to G
reconstructing from it the whole diversity of ac
synthetic formulation presupposes numerous prelim
- correct to judge by the value of these exercises
almost never made explicit.10
I shall illustrate the need for resorting to discovery
two types of problems - closely related moreover: t
on the one hand, and, on the other, that of the segm
of different hierarchic levels.
Gevaert and Brailoiu both give tables, one of diatonic modes, the other of
prediatonic systems, developed deductively. They illustrate them with
examples but do not pose the crucial question: given any corpus of modal

MUSIC ANALYSIS 6:1-2, 1987 13

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

melodies how does one recognise that one of these me

mode? Or again: how does one recognise that a melody
one mode to another, or that it presents a hierarchic
modes? The earliest specialists of Gregorian chant
appeared as an embyronic discovery procedure, that a
mode whose last note was the 'tonic'; ' but this proced
In fact, despite the insistence of various musicolog
formulae, for example, in the determination of mode
gap between, on the one hand, specific messages (the
for example) and, on the other hand, the system of m
the most abstract part of the code. All the rules allowin
former to the latter (and vice versa) are missing.
Another problem: for everyone, it seems self-evi
with a minimum of complexity is subject to a hier
divides into parts on different levels. Thus, according
melodies divide into periods, these in turn into phrases
and the latter into incices. Such analyses raise a multi
least of which is the validity of the taxonomic concep
that they seem to imply (see below). One could also as
of period, phrase, etc., are susceptible to general or u
conversely, they must be seen only as ad hoc notions, v
But the crucial question, first and foremost, is the foll
which, in such and such a case, have presided over the
takes the trouble to reply to this question, as if the o
were manifest.14
This question involves a series of others. Here are a
section A into two segments a and b, are these div
difference of timbre, opposition of register, melodic
similarity or contrast of rhythms, equal or unequal d
Or is there a combination of these elements in play
similarities of or differences between segments? Can
by others? Do I obtain the same results, for example,
on rests and then on cadences - a correspondence that
chorale - or, on the contrary, does recourse to di
different segmentations which introduce ambiguities
case is certainly very common, and its study wou
accounting for, among other things, variants of inte
impose a hierarchy on the various criteria, the one onl
allows ambiguities to remain? Can one establish pr
validation of a chosen criterion?' Do universal criteria
to distinguish essentially syntagmatic criteria (the
criteria (based on elements' internal and/or extern
depending essentially on the substance (rests yet a
depending on essentially formal criteria (repetition, v
These questions may well appear futile to music

14 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

However, their importance in the study of non-Eu

contestable. On the other hand, the application of expl
to more familiar musical systems can well lead only
already intuitively recognised. But that itself is far fr
is indeed very useful to be able to verify the elaborati
step by recourse to intuition; once this procedu
determined, it can then be applied to the study of less
Moreover, even in fields as well known as that of fugue
of segmentation lead to the revision of traditional analy
intuition of the best musicians).16
I was saying above that the question of the division (
and the establishment of scales and modes were re
structure can serve as a pointer in the process of divis
is therefore a danger of introducing a vicious circle in
consider the principal criteria adopted by Bruno Nettl
a given piece. These are: a) greater frequency and du
comparison with others; b) final position of this note
phrases; c) its terminal position in the song. What inte
It presupposes that one already has at one's disposal crit
the piece into 'individual phrases and sections' and that
circle, these criteria exclude all reference to the sc
(Nettl does not furnish explicit criteria for divisi
however that the only possible discovery procedure wi
to scales. One can, on the contrary, imagine the
procedures, the one proceeding from division to scales
division, with the second confirming the validity of the
at the beginning of this article, that there certainly d
satisfactory discovery procedure: the more one m
procedures, the better that will be for the final const
is essential that one of the procedures does not pres
other.

III

In this article, I shall deal especially with the procedures of division, even if it
entails, in a given example, indicating their influence on modal analysis. Tw
methods are available to derive a procedure. One can either begin from already
completed analyses, and try to reconstruct the criteria, not necessaril
homogeneous, which have dominated there, or choose one given principle
perfectly explicit, even if it means accepting that it may prove inadequate
require improvement, and even be rejected. It is this last path that I shall try to
follow.
First of all, let us leave aside reference to rests - certainly inadequate if the
segmentation is taken far enough- as well as recourse to the linguistic structur

MUSIC ANALYSIS 6:1-2, 1987 15

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

of the words,'" in the case of vocal music. Let us supp

subsidiary recourse to these types of data, either t
provided in other ways, or as indices to possible segm
purely musical analysis runs into great initial difficult
criterion of repetition is not immediately applicable).
all of formulating procedures based on specifically m
Furthermore, it is useful to introduce a theoretical
types of musical element which I shall call respecti
parametric.19 A parametric element can take two form
which is constant throughout the whole duration of a
the tempo in certain Bach allegros, or the monodic cha
vocal melody. Here, clearly, this element is of no help
piece. Secondly, the element takes the form of a binar
the piece into sections characterised by the presence of
other term of the opposition; cf. the opposition so
singing, the opposition piano (= 'solo')/forte (= 'ens
choral music, that between the 'original' and the 'echo
'complete'/'incomplete') in certain musics of the b
minor opposition in the slow movement of Beethoven
opposition of high/low register (it is also an oppositio
march of Stravinsky's Renard, etc.. As these exam
oppositions are composite and combine several dim
can be very easily dissociated, in which case the segm
be different according to whether one considers the
each case, the principle of segmentation will be the sa
defined in terms of contrasts, and these depend on th
in the section of one of the two terms of the binary o
On the other hand, a non-parametric element canno
opposition; rather, it is characterised by a fairly l
distinctions of the same dimension (cf. the nume
produced by the diatonic or chromatic scale or, again,
intensities, modes of attack, in serial compositions
shown - in another language - it is not possible to det
and such a musical dimension has a parametric or n
Parametric dimensions in one culture, or at one pe
parametric in another.
In this article, I shall take no account of paramet
therefore be considered constant throughout the
analysed. I shall confine myself to non-parametric
repetition as my principal criterion of division. I shal
appreciation of the enormous role played in music, at
and I shall try to develop an idea proposed by Gilbert

.. Certain fragments are repeated, others are not; it is

absence of repetition - that our segmentation is based.

16 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

of notes appears two or more times, with or wit

considered a unit. As a corollary, a sequence of not
once is also considered a unit, whatever its length and
of its articulations (especially silences) ... 21

Before going further, it is necessary to specify

repetition, and to define the assumptions on wh
criterion of repetition is based. Repetition signifies
spaced at different places in the syntagmatic chai
raises the question: identity from which point of v
physical point of view, two actual events are never
degree of abstraction is therefore inevitable, and th
be asked. We will not ask this question here, and will consider certain
elementary identities as given. On the other hand, we must decide which
dimensions - pitch, duration, intensity, timbre, etc. - will be the basis on which
two different segments will be considered as repetitions of one another. Here,
given that the examples will be borrowed from the western literate tradition,
and will be monophonic, only pitch and duration will be considered. But it must
be remembered that segments, variable as to pitch and duration, can be
considered as repetitions as long as they are identical in other respects.
One therefore treats as data minimal elementary identities of pitch and
duration. More precisely, one can express the situation by saying that one
initially possesses some mechanism which is capable of recognising a pitch, an
interval, or a definite unit of duration (for this mechanism, a c' is always a c', a
minor third is always a minor third, and a minim a minim). Moreover, in the
particular procedure chosen here, identities of pitch and duration are treated
together, at least at the beginning: only segments which are at once
simultaneously identical from the perspectives of pitch and duration are
considered identical. At a later stage of the procedure, the two dimensions may
possibly be disassociated to provide units which, as repetitions of one another
from one only of the two points of view, will be considered as transformations
of one another (or variations). This procedure seems to have suited the type of
material used (it saves time: a procedure which separated the two dimensions
would have ended up with the same results, but by a longer route), but it is not
obligatory; there are musics - fourteenth-century isorhythmic motets especially
- which require one to separate the two dimensions from the outset.

IV

This much said, here, in its broadest outlines, is the description of a procedure
of division, based on the principle of repetition, and applied to monodies.
(a) Our 'machine for identifying elementary identities' passes along the
syntagmatic chain and identifies similar fragments. One considers as level I

MUSIC ANALYSIS 6:1-2, 1987 17

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

units sequences - the longest possible - which are r

either immediately after their first statement or
segments. This first operation yields such structu
X,A + X + A + Y + A,A + A + B + B + X,A + B + A + X + B + Y, etc.
(repeated sections, level I units, are represented by the first letters, the
'remainders', by the last letters of the alphabet).
(b) The remainder or remainders are considered provisionally as units of the
same level I (cf. the quotation from G. Rouget); this analysis is strengthened or
weakened by recourse to other criteria. The total duration of the segments can
yield an initial index: if, by the operation (a), one has derived a structure A +
A + X, X will in principle be considered as a unit of the same level as A if its total
duration is approximately the same as that of A (in this case, to show that (b) has
taken place, one may, in the notation, replace X, Y, Z, etc., by B, C, D, etc.,
and A + A + X is written A + A + B). It should be noted that, in having
recourse to the equivalence of the segments' duration, we are only applying the
principle of repetition on a more abstract level: X is, from the point of view of
its absolute duration, all other things being equal, a repetition of A.
(bl) The results of (b) can then be consolidated by recourse to indices
provided by the rests, or by linguistic analysis of the words in the case of vocal
music.
(c) If operations (b) and (bl) have failed to result in, and if the remainders are
not admissable as, level I units, two alternatives emerge: (1) X, Y, etc., are
much shorter than A, B, etc.; these remainders are put off to a later stage in the
analysis, awaiting the results of following operations (d); (2) the remainder is
much longer than A, B, etc.; in this case, either, thanks to the operations of (b),
(bl), (d) it can be segmented22 into level I units, which will be transformations
of A, B, etc. - and then, for example, A + A + X will be described as A + A +
B + C - or else it will reduce later - after a new application of (a) to units derived
at level I - into units of level II, or, finally, it must be considered as an
unanalysable unit of level O (see below, (e)).
(d) Often, one will be led to consider various units - both among A, B, etc.
and X, Y, etc. - as being transformations (rhythmic and/or melodic variants) of
one another. Thus, for example, A + A + X will be rewritten A + A + A', or
again A + B + A + B will be rewritten A + A' + A + A'. It would be essential
to draw up a list of types of possible transformations and to describe the
procedures which allow their derivation. I shall limit myself to a few remarks
(leaving aside the question of transpositions, particular transformations which
scarcely pose any problems).
(dl) A first class of transformation will be derived if one applies, as has already
been suggested above, the principle of repetition separately to pitches and
durations. One then obtains rhythmic transformations of the same melodic
structures, and vice versa.
(d2) Other transformations will introduce such more complicated operations
as permutations, additions or subtractions of certain elements. I shall not enter
here into the detail of these operations, except to note a few in the course of the

18 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

analysis of the examples.

(d3) It is important to note that, in order for a section
a transformation of another section A, it is often neces
application of the operation (a) at a lower level; this
units, such that, for example, A = a + b, and X =
appears as a strict repetition of A and, from other
duration, rests, text structure, etc. - as long as X is eq
considered a transformation of A: X + A'. We see
necessity, in the course of the procedure, to shunt, tha
from top to bottom, now from bottom to top. Another
at the very beginning, since, having started from 'the
units of duration and pitch - we then, with the operat
'the top'.
(d4) Often, in deriving transformations - especially by the operation (d1) - one
is led to revise an initial segmentation, established by (a) and (b). Let us suppose
that these two operations have resulted in a structure A + x + A + y (with very
short remainders). If (dl) shows that A + x is identical to A + y from the point
of view of durations, for example, and if other factors intervene as well - such
as the absence of a rest between A and x, A and y, but the presence of a rest
between x and A - it may be stated that A + x is a single unit, of which A + y
is a transformation, and one will rewrite the structure as A + A'.
(e) We can now tackle a problem of which (d4) is only a particular case. Let
us suppose that the operation (a) has produced such structures as
1)A+X+A+Y...
or

2)X+A+Y+A...
A question arises which we had first of all ignored: ca
in (1), A + X and A + Y, and in (2), X + A and Y + A, con
higher than level I (let us call this level O)? The operati
of replying to this question, and one is obliged to resor
Here are the two most important; both appear to me to
describe (1) as (A + X) + (A + Y), and (2) as (X + A
(el) The ending of X and Y in (1), that of A in (2) - in
(1), those of X and of Y in (2) - are marked in a special
the elongation of the final note (compared with the
elongation in the other units).
(e2) Later analysis - that is to say, essentially, the ope
(d) - shows that Y is a transformation of X.
It remains to be said that once the units of level I
procedure must be applied again, beginning with o
derive the units of level II, and so on, until one arrives a
the elementary units from which one began.

MUSIC ANALYSIS 6:1-2, 1987 19

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

Let us now illustrate the procedure by examples, beginning with the most
simple. Any difficulties encountered, and problems raised, will appear
progressively. First, a fourteenth-century German Geisslerlied: Maria muoter
reinif maft (cf. Ex. 1).23 A few words first of all on the graphic presentation of the
examples. I have found it illuminating, in the study of monodies, to follow a
procedure applied to the analysis of myths by Claude Levi-Strauss - the latter
who was himself influenced by the musical notation of orchestral scores.24
Equivalent sequences are, as far as possible, written one below another in a
single column , and the text is to be read, ignoring the spaces, from left to right
and from top to bottom. Thus, certain traits of structure become immediately
apparent, as are certain ambiguities. Clearly, it would be very difficult to apply
the same procedure to the presentation of polyphonic structures.
It must be stressed that, in the actual functioning of the analysis, various
stages of the procedure do not necessarily follow in the given order. The
procedure is much more one of verification, meant to ensure that the analysis is
coherent, than a discovery procedure in the strict sense of the term. Doubtless,
it would always be possible to apply it rigorously in the given order, and one
would obtain the same results, but it is much faster and more economical to
make use of it in order to verify the results of an analysis obtained purely
intuitively and sometimes very rapidly. This is a situation well-known to
linguists. Therefore, in the analysis of the examples, and so as not to prolong
the demonstration inordinately, I shall often allow myself to be quite elliptical,
confident that the reader will be able to reconstitute for himself the series of
operations which have been carried out.
Let us consider our Geisslerlied. A first application of the procedure derives,
at level I, the structure A + A' + B + B, without any remainder (A' to take
account of slight variants, b against b', bb against a, and then the crochet a
divided once into two quavers).26
The explicit series of operations would have in fact given:
(a)X + B + B;
(b) a negative result: no equivalence of absolute duration between X and B (no
more than between A and B);
(c), (d): X = A + A'; A' is a melodic transformation (without change in
duration) of A ([Link]); it is certain that, intuitively, one would have already
derived the level II units, and that it is in terms of b and of b'- rather than of A
and A' - that one would have identified the transformations.
If, despite the negative result of (b), A, A' and B are considered units on the
same level (I), this is particularly because of the results of (d), and because at the
later stage, A, A' and B will appear to be made up in part of identical elements
(cf. d3).
A second application of the procedure gives level II units, obtained for
example in the following manner:

20 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

Ex. l a: Geisslerlied
b

b A

b w.
Al
II I

if I

d di

Ex. Ib Ex. Ic Ex. Id

T T I

c I

MUSIC ANALYSIS 6:1-2, 1987 21

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

(a) A'= x + b + y + b;
(b) A' = a + b + c + b (durations of a, b, c are identical);
(d) A = a + b + c + b' (b' = Tm - abbreviation for melodic transformation -
of b);
(a) B = z + b' (b', identified in A, is found here);
(b)B = d + b'.
A third application of the procedure gives the units of level III (designated by
the means of subscript numbers to the designations of the level II units, e.g. al):
(a) d = dl + dl;
(a) c = c1 + dl;
(b) a = al + a2, b = bl + b2, b'= b'l] + b2 (all these units being equivalent in
duration to dl); in addition:
(dl) (cf. Exs lb and Ic): al, b'1, bl, cl, a2 are all melodic transformations of the
same rhythmic structure (four crochets); similarly, b2 is a Tm of dl.
Finally, a fourth application of the procedure allows the derivation of a
certain number of units which are either repetitions or various types of
transformations (transpositions, inversions, recurrences, Tm) (cf. Ex. Id).
What prevents us from talking of units of level IV, besides the fact that they are
of very unequal length (some are as long as the level III units), is that these units
encroach on one another in various ways. The discontinuous character of units
and levels - which appears essential to a taxonomic conception of the musical
structure - thus appears obscured there. If, on the other hand, one pushes the
segmentation further, one ends up with the minimal units postulated at the
beginning, and the procedure has exhausted its results.
Let us single out as one of the essential results of this analysis the asymmetry
that it uncovers, at all levels: asymmetry between A (varied to A', and composed
of three sub-units) and B (not varied on repetition, and composed of two sub-
units), asymmetry between a, b, c (composed of two different segments) and d
(composed of two identical segments), a more subtle asymmetry between a
(whose two sections are only Tm of each other) and b, c (whose two segments are
varied at once melodically and rhythmically), and finally, asymmetry as the
result of the encroachments of the 'level IV units'.

V.1

This analysis has not had recourse, at any of its stages, to data relative to scale
or mode. On the other hand, it is possible to use its results to derive the modal
structure of the piece. A clear-cut hierarchy of the different notes used results
from the analysis into units of different levels, and this hierarchy does not entail
the introduction, at least not directly, of quantitative criteria. The principal
criterion singled out is that of the initial, final or intermediary position which
the notes occupy in various units. Initial and final positions are considered
as taking priority,27 and it is accepted that the initial and/or final positions in the
units of a higher level carry more weight than the same position in units on a

22 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

lower level.
1. fis the first and last note of A, A'and B; fis also the first note of a2 and dl;
2. c is the final note of a;
3. a is the first note of b, final note of c, final note of al, bl, b', cl, dl;
4. d is the first note of c, and absent everywhere else;
5. bb is the first note of b2; everywhere else, it is in an intermediary position
(note however that b is the most 'welded' [soudge] of the level II units, both
because bb moves from a to a, and as a result of the encroachment of b2 and of
the unit of 'level IV', which is a retrograde transformation of dl; thus, even here,
the position of bb is very close to a position of transition);
6. g only appears in a position of transition. A secondary criterion - derived in
practice from the first - takes into account the role of different notes as passing
notes, ornaments, etc., their place in conjunct vs disjunct motions, and the fact
that they are or are not immediately repeated;
7. f, c, a, d are the only notes to be linked - inside a unit, or at the boundary
between two units - by disjunct motion;
8. bb and g only ever appear in conjunct motion, ascending-descending or
descending for bb , ascending or descending for g;
9. f (as the final note of A and of B), c (as the final note of a) and a (at the
boundary between c or d and b or b') are the only notes to be repeated
immediately;
10. the variant b against b' accentuates bb 's character as a passing note and the
stronger position of a.
All these traits allow the derivation of a very clear modal hierarchy that one
could characterise as f major, with an oscillation towards the relative minor in
c, and some traces of pentatonicism. But it must be noted that these aspects -f
major, pentatonicism - only have significance if one places this piece within a
larger context. If one limits oneself to a particular system of which our
Geisslerlied is the sole message, to speak of major (without the leading note, e)
or of pentatonicism (f- g- a - c - d, when nothing authorises us - on the contrary
- to lend more weight to g than to bb ) is equivalent to a distortion of the facts.
Only if one replaces this piece in a much larger corpus does it appear as a case
of major or of pentatonicism, and its underlying system appear as a sub-code of
the tonal system or a sub-code of the pentatonic system.

VI

Let us take another example, a chanson by the trouvere Guiot de Provins: Molt
me mervoil (cf. Ex.2).28
1. An analysis based on the metre and the rests would immediately give eight
distinct level I units; our procedure ends in the same result by the following
route:29
(a)X + B + Y + B;
(b) X = A (same duration as B);

MUSIC ANALYSIS 6: 1-2, 1987 23

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

Ex. 2a: Guiot de Provins

A en

I/. -I m . I-

do ~ Fmrr -~--
( "l ' I';

(c) and (d): Y resolves into units of equal length to A or B (I sho

procedure, which implies shunting - reference to level II units - and re
to transformations: C = C'), whence the structure:

A+ B +C + D+ E + F +C'+ B

2. The application of (a) - and if necessary of (d) - to a lo

series of level II units, as in the following table:

A=x+a B=y+b
C=z+a D=w+b
E=v+a" F =?
C'= z'+ a' B= y + b

This table suggests two things. Firs

resolution; this fact introduces an asy
only unit to end on g (as opposed to c
one should ask if it is necessary to con
second-level units of the same status as
duration between these remainders an
the remainders do not represent simp
furthermore, there exist transforma
remainders - all have the same durati
simple. It thus seems better to acco
different notation.
3. One can now consider whether t
themselves in units of a higher level
without effect, but another criterion
is visible if one recognises that (A + B), (C + A), (C' + B) are so many
manifestations of a single abstract structure defined in terms of the relationship
between units of level II; this structure is described by the following formula
(where the brackets indicate that a choice must be made between the units

24 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

Ex. 2b

Ex. 2c

Ex. 2d

F[-3
S,x - . z, z-
o I I _

Ex. 2e

13

- _

Pa

MUSIC ANALYSIS 6:1-2, 1987 25

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

which they enclose):30

x y

z a b
The case of F raises a p
corresponds to the first p
from an external (distrib
the formula, but that it d
4. If one wishes to take th
to posit an intermediary
From the perspective chos
segments which do not en
are repeated in different
b, and if neither a nor b
neither a + c nor d + b - i
be units; (a + b) must be
Guiot de Provins, one fi
level II but only, occasio
each other. I leave open th
the positing of an autono
be found in Ex.2, especi
realise that, insofar as
immediately successive se
different level I units, bu
transformational correspo
of level I some sort of
comparison with the relat
In Ex.2 are indicated a few of these transformations which have the effect of
blurring the boundaries between the units of level II. Particularly noteworthy
are those: (i) which link B and D to C ([Link].2e), introducing an oblique
relationship in comparison with the contrast between A, C and B, D, and
(ii) which link F respectively to B, C' and E ([Link].2d), thus connecting F
to the rest of the piece, from which, in terms of level II units, F appeared to
be detached.

VII

Here is another example, the famous estampida: Kalenda maya by Raimbau

de Vaqueiras ([Link].3).31
1. By the application of (a), a unit A is derived without difficulty an
immediately repeated; then, by (a) and (c), two sequences B + x and B + y a
derived which, by virtue of (d4), are rewritten B and B'- in a transformationa

26 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

Ex: 3a: Raimbaut de Vaqueiras

'00

_ _ _-u _ ? ? . .L " L . . -'_ , ,_ _50 \ ! - - , I I t /d i

relationship: B = in suspense, B = conclusive; then a problem arises: must one

consider the sequences represented in the table respectively by c and by D (or
D'), as distinct level I units or, on the contrary, starting from the principle that
one tries first of al to derive the longest pos ible units, should one group them
as a single unit (c + c + D, then c + c + D') that would be rewrit en C (C'at its
second appearance)? In fact, it is not pos ible to give an unequivocal response
to this question, and this is certainly a case of ambiguity writ en into the
structure itself. Without doubt, by its length (criterion (b) , c can be as imilated
into units of a lower level, al the more since c appears effectively as a Tr(=
rhythmic transformation) of the final 'motif' of A and B'. On the other hand, c
shares with the level I units, A and B, the privilege of being immediately
repeated after its first appearance, which, in a sense, puts it on the same footing
as these units. Final y, D begins and ends in the same way as A and B, and we
wil se that these three units stand in a transformational relationship; to fuse D
with c in the single unit C would make it lose this characteristic. That is why I
think that one can represent a first segmentation of the piece by the fol owing
formula, which preserves the ambiguity:

A + A + B + BI+ c + c + D + c + c + D'

2. We have ignored another ambiguity. In terms of absolute dura

MUSIC ANALYSIS 6:1-2, 1987 27

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

Ex. 3b32
[Ai ] [A2]

A P]. i

Ex. 3c

A 4 T" [partially T']

a 00 m . l
N --- .

twice as long as B. This fact, t

derived not to end on c (hence
in a sense, A is equivalent to (
A2. But ([Link].3b) another tr
reduction 'through the midd
into B and then into D, A is
'motifs', a and b, whose au
melodic) transformations, on
elements (especially, c is a Tr
of segments of intermediary

28 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

To sum up, if the analysis quite clearly produc

higher and lower - units which satisfy crite
encroachment and of autonomy, it also reveals mu
not only at the intermediary levels, but even in th
units of the two extreme levels (cf. the case of c). T
of a unit keeps its value, that of the level (i.e., tha
become blurred.

VIII

A last example, which is also a troubadour chanson: Be m'anperdut by Bernard d

Ventadour ([Link].4).33
An initial application of (a) readily produces the structure A + A + X.
However, the difficulties begin immediately one tries either to reduce th
remainder X or to segment A into smaller units. In this last case, anoth
application of the procedure at a lower level only yields very short units, reduc
to one, two or three notes (Ex.4c); at intermediary levels, one only find
segments with quite badly defined contours, linked only by relationships
transformation and not of simple identity, and whose autonomy is weak.
As for the remainder X, the application of (a) and of (d) produces a long fin
segment (A1.2) which is a repetition, with slight transformations, of a large pa
of A. But this segment is itself preceded by a shorter segment, of five minim
(A.,1) in a double transformational relationship to the beginning of A ([Link].4
on the one hand, it is equivalent, in absolute length, to the whole of th
beginning (everything which precedes the equivalent of A1.2), and, on the oth
hand, it represents a melodic transformation by 'diatonic filling-in' of the mot
of minims a - c - d -f. The lack of autonomy of these two segments A1.1 and A1
as well as the great length of A1.2 in comparison with A, prevents their being
considered as constitutive units of an intermediary level. One can group them
in one segment A1, which, from certain perspectives, could be placed on t
same level as A - absolute duration, similarity of termination (A1.2) - but which
from other viewpoints, particularly absence of the initial motif, seems only to
equivalent to a fraction of A.
Having extracted A1 from X, there now remains, at the beginning of Ex.4b
a segment (A1.2) which, with transformations a little more complex34 than in t
case of A1.2, corresponds to a shorter part of the end of A. The relative autonom
of this segment could also lead one to postulate a division of A into two parts
nearly equal length (A = Y + A1.2), but this would be to posit a unit (Y) which,
partially encroaching on A1.2, does not have, any more than the latter, a well
defined existence.
Finally, it is Ex.4c which gives the clearest picture (although not exhaustive,
it does not indicate the relationships between small units, which could easily be
detected) of the structure of the piece. The segments in brackets correspond to
'large units', partially fitting in to one another; the short segments that are

MUSIC ANALYSIS 6:1-2, 1987 29

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

Ex. 4a: Bernard de Ventadour

Ex. 4b

Vs - "-duratio

I 1'r f
Vs [durationl

found in the same vertical columns correspon

segments represented on the same stave, in th
intermediary units. This table also quite c
procedure which consists of inserting - or
inside larger segments.

SUMMARY AND CONCLUSION

Starting from the distinction between analytical model and synthet

have stressed in this article the necessity of developing the analytic
view in musicology, or in other words, the urgent need to elabor
procedures for deciphering the code from messages. At the same tim
stressed the limitations of this type of procedure. I have principally
a procedure of segmentation, based on the criteria of repetition
formation, and I have tried to apply it to a few medieval monodies.
It is too early to draw conclusions from a study which represents
beginnings of research. Many things stated here will need to be tak
elaborated: that is the case with the distinction between paramet
parametric elements, as well as the very notion of transformation. In
it will be necessary to invent discovery procedures for recognising,
transformational relationships between elements. These relations
in very simple cases, have here been considered as self-evident.

30 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

Ex. 4c

twice

I F t I.-

_______ A. -'-....._t- f.I_.

J t >x
rn. L

Kf
t F ,-_-
--

L Jlj7.

MUSIC ANALYSIS 6:1-2, 1987 31

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

I would however like to draw attention to two of the

1. It seems to be the case that, at least in certain
segmentation into units of distinct levels is unproble
paragraph), a procedure of segmentation of the typ
important consequences for scalar and modal an
alternative to or as a control over more traditional
space, I have hardly stressed this point, but, for
application of the procedure sketched in section V.
from the modal point of view by Edmond Costere35
error of these analyses. Moreover, one can begin to
discovery applicable to the question of pyens, a proce
terms of distribution of intervals, would complete
contribution.
2. Precisely because it attempts to define rigorously levels and units, the
procedure we have followed has led us - by different paths from those followed
by Andre Souris - to challenge a purely taxonomic conception of musical
structure. Even in such a simple case as that of Ex.1 (cf. section V), it is
impossible to represent the structure completely in the form of a series of
interlocking structures, with level I units wholly resolving into discontinuous
level II units, and these in their turn into discontinuous level III units, etc. The
main reason for this state of affairs stems evidently from the fact that musical
syntax is a syntax of equivalences: the various units have between them all sorts
of relationships of equivalence - relationships which can unite, for example,
segments of unequal length - one segment will seem to be an expansion, or a
contraction, of another - and also segments encroaching on one another. The
consequence of all this is, as has been seen, that it is impossible to represent the
structure of a piece of music by a single arrangement.

NOTES

1. In this article, I shall treat music as a semiotic system, sharing a certain n

common traits - such as the existence of a syntax - with language and other
of signs. I shall leave completely aside the genuinely aesthetic aspect and es
the question of whether aesthetics can be reduced to semiotics. Moreover, i
field of terminology, because of the reference to notation necessarily implie
use of the word 'text' in music, I shall use, in preference to Hjelmslev's dic
of system and text, Jakobson's, derived from the theory of communication,
and message.
2. Louis Hjelmslev, Prolegomena to a Theory of Language, trans. from the Danish by
Francis J. Whitfield, rev. ed. (Madison, Wisc.: University of Wisconsin, 1970);
Knud Togeby, Structure immanente de la langue frangaise, Travaux du cercle
linguistique de Copenhague 6 (Copenhagen: Nordisk Sprog- og Kulturforlag,
1951); Zellig Sabbettai Harris, Methods in Structural Linguistics (Chicago: Chicago
University, 1951); Discourse Analysis Reprints, Papers on Formal Linguistics 2

32 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

(The Hague: Mouton, 1963); Paul Lucian Garvin, On

Papers, Janua linguarum: studia memoriae Nicolai v
(The Hague: Mouton, 1964).
3. Claude Levi-Strauss, Anthropologie structurale, 2 vols
trans. Claire Jacobsen and Brooke Grundfest Schoepf
(New York: Basic Books, 1963; 2/London: Allen Lane
Chapter 11, 'The Structure of Myths'.
4. Cf. especially Algirdas-Julien Greimas, Semantique stru
Langue et langage (Paris: Librairie Larousse, 1966); M
for Style Analysis', Word, Vol. 15, 1959, pp. 154-74; 'V
du style', Word, Vol. 17, 1961, pp.318-44.
5. Cf. the work of the school of generative-transformat
States and particularly Noam Chomsky, Syntactic S
studia memoriae Nicolai van Wijk dedicata 4 (The Ha
Issues in Linguistic Theory, Janua linguarum: studi
dedicata - series minor 38 (The Hague: Mouton, 1964) a
Introduction to Transformational Grammars (New
Winston, 1964).
6. Noam Chomsky, 'Some Methodological Remarks on
Vol. 17, 1961, pp.219-39. In a way, Chomsky takes up t
Claude Levi-Strauss that the analysis of myths is only
types of data: the texts and their ethnographic context;
is necessary, but it is insufficient (cf. Claude Levi-
forme: reflexions sur un ouvrage de Vladimir Prop
Philosophiques et Economiques, 19 vols, ed. Jean Lacro
Science Economique Appliqu&e M:1-19 (Paris: Insti
Appliqu&e, 1960), Vol. 7, pp.3-36. In music, one can see
of these additional data can correspond: the analys
disposal a corpus of recorded pieces but also the de
information on how they are played, data on the condit
commentaries - be they only the titles of works -
indirect indices to the structure of the code.
7. Pierre Boulez, 'Strawinsky demeure', Musique
consecutively, ed. Pierre Souvtchinsky, Biblio
musicologie (Paris: Presses universitaires de France,
8. Franqois Auguste Gevaert, Traite d'harmonie theor
Henry Lemoine, 1908). I mention in passing my gra
drew my attention to this fundamental work, now alm
who was one of the few to have appreciated all its valu
9. Constantin Brailoiu, 'Le giusto syllabique bichron
pp.26-57; 'Sur une melodie russe', Musique russe, 2 v
ed. Pierre Souvtchinsky, Bibliotheque internation
Presses universitaires de France, 1953), Vol. 2, pp.32
notions liminaires', Cercle international d'etudes ethno-m
26 Septembre 1954, ed. Paul Collaer, Les Colloques

MUSIC ANALYSIS 6:1-2, 1987 33

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

Elsevier, 1956), pp.64-96.

10. One certainly finds in the works of Gevaert or Brailoi
allow, in certain cases, the reconstruction of an analytic
Gevaert, Traitr d'harmonie, Vol. 2, pp.64-5. Brailoiu
certain number of indices by which pyens are analytica
not group them in a systematically ordered procedure -
errors of interpretation, cf. Ex. 10, p.343. Let us nevert
of discovery procedures is beginning to preoccupy some
least implicitly, by Andre Souris (cf. for example '
musique, 3 vols (Paris: Fasquelle, 1958-61), Vol. 3, p.437)
who are led there inevitably from the question of trans
progress of Gilbert Rouget, as well as Nicholas M. Engla
Transcription and Analysis: A Hawke Song with Musi
Vol. 8, 1964, pp.223-77).
11. Cf. especially Gustave Reese, Music in the Middle Ages
pp.161-2.
12. Cf. for pentatonicism, Brailoiu, 'Mlodie russe', pp.351-2. It is clear that formulae
- different types of incipit, etc. - are used only for illustrative purposes. The
problem arises moreover as to how the autonomy of these formulae is established,
which again poses the question of discovery procedures, in another domain (see
below, problems of segmentation).
13. Reese, Music in the Middle Ages, p.169, and the articles by Andre Souris in
Encyclopidie de la musique, 3 vols (Paris: Fasquelle, 1958-61).
14. In my own analyses of Debussy - cf. Nicholas Ruwet, 'Note sur les duplications
dans l'oeuvre de Claude Debussy', Revue belge de musicologie, Vol. 16, 1962, pp. 57-
70; reprinted in Langage, musique, poesie, Collection poetique (Paris: Seuil, 1972),
pp.70-99 - the segmentation of analysed fragments is taken as acquired data: see,
for example, the analysis of the passage of Fetes into sections and subsections (p.75).
Nevertheless, I give a rudimentary rough sketch of the discovery procedure a propos
the Prelude of Pellgas (p.90) [page numbers refer to those of the reprint].
15. For example: let us suppose that, in using the rests, I divided the given section into
two segments A and B, this segmentation has a good chance of being confirmed if,
searching for equivalences in the internal structure of A and of B, I discover that A
and B have the same absolute duration (in terms of time, of bars, etc.) and/or that
A = a + b, and that B = a + c (so that A and B are equivalent from a certain point
of view).
16. The application of a procedure broadly analogous to that based on the principle of
repetition, which is exposed below, to the fugues in the Well-Tempered Clavier,
allows the derivation of units of various levels, often corresponding to units defined
by traditional theory (exposition, episode, subject, etc.). One can however
determine in a purely formal manner that it is impossible, in the case of a particular
fugue, to speak, for example, of counter-subject or primary subject, etc.
17. Bruno Nettl, Music in Primitive Culture (Cambridge, Mass.: Harvard University,
1956), p.46. I do not necessarily wish to take responsibility for the idea that it is
possible to determine a tonic in non-tonal music - cf. Brailoiu, 'Mlodie russe',

34 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 METHODS OF ANALYSIS IN MUSICOLOGY

pp.346-7, on the instability of the tonic in pentatonic

is purely of method. It is to be noted that Nettl m
qualitative (b) criteria. It might have been appropr
clearly, and to try to define scales in purely struct
relative frequency of such and such a note does not ap
18. It is obvious that the linguistic structure of the text m
of vocal music. But this obvious point does not allow t
as is current in the study of the 'forms' of the mediev
Middle Ages, p.224, analyses Por conforter ma pesan
follows:
(1) a + b + a + b + c + d + E.
If one provisionally leaves aside the contrast between the chorus (E) and the soloist,
a musical analysis based on the criteria of repetition gives a primary segmentation
into A + A + X; at a second level, A resolves into a + b, and X into a' + c (a' = a
transposed). Hence:
(2) a + b + a + b + a' + c (c then is divided on the basis of a totally different
principle, the opposition chorus/soloist). The linguistic analysis gives a different
structure; from the point of view of rhymes (which is only one of its aspects), one
has (3) which can be superimposed on (2) in the following manner:
(3)m + n + m+ n + n + m + p
(2) a + b + a + b + a' + c.
This analysis has only a demonstrative value; it seeks above all to react against a
tendency to level out linguistic and musical realities by projecting multidimensional
structures onto a single plane. Let us add that, in the purely linguistic field, the
distinction into several levels is essential: the segmentations obtained in terms of
syntax do not necessarily correspond to those obtained in terms of phonology or
metrics. Cf., for illustrations, Claude Levi-Strauss and Roman Jakobson, 'Les
Chats de Charles Baudelaire', L'homme: revue franfaise d'anthropologie, Vol. 2, No.
1, January-April 1962, pp.5-21; trans. Fernande M. De George in The
Structuralists: From Marx to Levi-Strauss, ed. Richard T. and Fernande M. De
George (Garden City, N.Y.: Anchor Books, Doubleday, 1972), pp. 124-46, and my
article 'Sur un vers de Charles Baudelaire', Linguistics, Vol. 17, 1965, pp.69-77;
reprinted in Langage, musique, podsie, Collection poetique (Paris: Seuil, 1972),
pp.200-9. Finally, it is hardly necessary to observe that many linguistic elements
may play a part in a musical analysis in so far as they are musical elements: thus rhyme
can act as timbre or syllabic/melismatic opposition as staccato/legato opposition,
etc.

19. This distinction is not absolute and a whole range of possibilities exists.
stress that the word parametric is used here in a slightly different way than in
writings of serial musicians. To designate what they understand by parameter
shall speak simply of dimensions of the musical substance.
20. Cf. especially Andre Souris, 'Forme', Encyclopidie de la musique, 3 vols (P
Fasquelle, 1958-61), Vol. 2, pp.119-23.
21. Gilbert Rouget, 'Un chromatisme africain', L'homme: revue franf
d'anthropologie, Vol. 1, No. 3, September-December 1961, p.41.

MUSIC ANALYSIS 6:1-2, 1987 35

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]
 NICOLAS RUWET

22. With, possibly, new short remainders to transmit to a la

example, A + A + X = A + A + B + C + y (level I units
case letters and those of level II by lower-case letters).
23. Cited after Reese, Music in the Middle Ages, p.239, who
Die Lieder und Melodien der Geissler des Jahr 1349 nach d
Reutlingen (Leipzig: Breitkopf und Hirtel, 1900), p.9.
here in problems of transcription, I take the transc
prejudging their validity.
24. Cf. Levi-Strauss, Anthropologie structurelle, pp.234-5.
25. Cf. Harris, Methods in Structural Linguistics, p.1: 'T
constitute a necessary laboratory schedule in the se
completed before the next is entered upon. In practice, l
short cuts and intuitive or heuristic guesses, and kee
particular language before them at the same time .... Th
procedures listed below is therefore as a reminder in
research, and as a form for checking or presenting the
26. To simplify, in the course of the analysis, I shall tak
variation, since the second is well explained as a combina
of the syllabic and metric structure of the words (t
unaccented respectively, against a single accented syllabl
level II units, b and b'.
27. This criterion, sufficient in this form for the piece in
refined. It would be necessary at least to take account of
(anacruses, etc.).
28. From Friedrich Gennrich, ed., Troubadours, Trouveres,
Das Musikwerk 2 (Cologne: Arno Volk Verlag, 1951), p
29. That this route is longer does not necessarily mean that
only more rigorous but it brings forth information which
for example, it indicates that, from a certain point of v
enclosed within the two statements of B comprises a sin
30. This notation is borrowed from transformational lingu
Transformational Grammars. I disregard here the differ
and z'.
31. Gennrich, Troubadours, Trouveres, p. 16.
32. In this table, as in the following one, certain segments
within parentheses - in the same row but in different co
occurrences are equivalent to a single element and signif
equally well appear in two units but that it has - mor
assigned to one.
33. Gennrich, Troubadours, Trouveres, p. 13.
34. These, with the insistence on the a which they introdu
giving a suspensive aspect to this section, which contrasts
section in A.
35. Edmond Costere, 'M6lodie', Encyclopidie de la musique, 3 vols (Paris: Fasquelle,
1958-61), Vol. 3, pp.178-81.

36 MUSIC ANALYSIS 6:1-2, 1987

This content downloaded from

[Link] on Tue, 17 Oct 2023 [Link] +00:00
All use subject to [Link]