Durham Research Online

Deposited in DRO:
18 February 2021

Version of attached le:

Accepted Version

Peer-review status of attached le:

Peer-reviewed

Citation for published item:

Dickson, Ian (2012) 'Towards a grammatical analysis of Scelsi's late music.', Music analysis., 31 (2).

Further information on publisher's website:

https://doi.org/10.1111/j.1468-2249.2012.00336.x

Publisher's copyright statement:

This is the pre-peer reviewed version of the following article: Dickson, I. (2012), Towards a Grammatical Analysis of
Scelsi's Late Music. Music Analysis, 31: 216-241, which has been published in nal form at
https://doi.org/10.1111/j.1468-2249.2012.00336.x. This article may be used for non-commercial purposes in accordance
with Wiley Terms and Conditions for Use of Self-Archived Versions.

Additional information:

Use policy
The full-text may be used and/or reproduced, and given to third parties in any format or medium, without prior permission or charge, for
personal research or study, educational, or not-for-prot purposes provided that:
• a full bibliographic reference is made to the original source
• a link is made to the metadata record in DRO
• the full-text is not changed in any way
The full-text must not be sold in any format or medium without the formal permission of the copyright holders.
Please consult the full DRO policy for further details.

Durham University Library, Stockton Road, Durham DH1 3LY, United Kingdom
Tel : +44 (0)191 334 3042 | Fax : +44 (0)191 334 2971
https://dro.dur.ac.uk
 1

TEXT EDITED BY WMD AND AGREED WITH AUTHOR, 23 MARCH 2012

IAN DICKSON
TOWARDS A GRAMMATICAL ANALYSIS OF SCELSI’S LATE MUSIC

Introduction
The late works of the Italian composer Giacinto Scelsi (1905-1988) have often been
treated as if they were incompatible with the concept of musical grammar. First of all,
they seem to lack unambiguous morphological unities, even discrete notes. Their
syntax is also mysterious, because they are based on improvisations, which Scelsi
executed in a meditative trance, believing that he could thus avoid imposing any
rational technique or system on his sound material. Almost any improvisation,
however, presupposes a potentially inferable system of rules, a ‘model’1 or ‘referent’,2
even if the improviser is only vaguely conscious of this model, or denies its existence.
The syntax of Scelsi’s music is thus determined by (although not identical to) the
model on which he relied to create his most successful improvisations. In this article I
suggest some ideas towards a description of this model. Although such a description
can never be proven accurate, this seems to me to be a promising direction to
illuminate this apparently inscrutable music.
Scelsi lived and worked mainly in Rome. His aristocratic origins have often
been associated with the eccentricity of his attitudes and working methods, and thus
with the originality of his late work.3 For most of his career he was known mainly on
the Rome contemporary music scene, especially through his involvement with the
Nuova Consonanza group. He became internationally renowned only in the 1980s,
when his works were featured in major festivals, most significantly the 1987 Cologne
ISCM World Music Days, and taken up by ensembles such as the Arditti Quartet.
In his early music, culminating in La nascità del verbo of 1948, Scelsi was
torn between advanced techniques, including serial writing, and intuitive
improvisation, which he later claimed had been a natural inclination since childhood.4
This period was brought to an end in the late 1940s when he suffered a psychological
crisis, which led him unequivocally to reject the rational orientation of the Western art
music tradition. Convinced that he had made himself ill by ‘thinking too much’, he
 2

resolved to create his music (and poetry) ‘without thinking’.5 In his subsequent music,
therefore, he proceeded through meditation and improvisation, cultivating
‘automaticity’, that is, the release of conscious control of decision-making.6 The most
successful improvisations were then treated as abbozzi, or sketches, and transcribed
by assistants.7
The main characteristic of these works is their focus on pivotal pitch centres.
In the transitional works of the 1950s, which usually originated in piano
improvisations and so were limited to the chromatic scale, the pivotal sounds are
reiterated, blurred with clusters (especially in the piano suites), and embellished with
oscillations and figuration reminiscent of Varèse (as in the wind monodies and
Yamaon of 1954-58). In the mature works, beginning with Elegia per Ty and the
String Trio of 1958, and the more famous Quattro pezzi per orchestra (ciascuno su
una nota sola) of 1959, the oscillations and figuration disappear, leaving a musical
discourse based almost entirely on sustained sounds, which are varied with subtle
inflections of intonation, intensity, timbre and, intermittently, changes of register.
This style was made possible by Scelsi’s adoption of the ondiola, or clavioline,8 an
electronic keyboard instrument allowing precisely these inflections. Usually, several
of these gestures overlap, in such a way as to create the impression of a more complex
but still unitary sound object, which is either fixed to one pitch-class (as in the
Quattro pezzi), or describes a gradual composed-out glissando (as in Xnoybis,
analysed below).
Scelsi’s statements about his own music, which consist of various dictated
texts and interviews, outline his general aesthetic stance. This was influenced by
musical and philosophical traditions of India and Tibet, as filtered through Western
mediators.9 His use of the drone was not merely a superficial imitation of these
traditions, but was motivated by the idea of an inner energy of sound. The ‘true
musician’ or adept was able to find this sound energy within single sounds, and
specifically not by ‘com-posing’10 (that is, by putting sounds together, finding
systematic connections between sounds). Sound had ‘depth’: it was like a sphere, and
the enlightened musician was able to ‘penetrate to the centre’ of the sphere.11 The
Western canon, on the other hand, by relying on abstract systems of notes (points in
relation to the ‘two dimensions’ of pitch and duration), tended to be ‘empty’ of sound
energy.12
 3

Given Scelsi’s esoteric, anti-systematic stance, it is not surprising that his

statements offer no rational explanation of his music. The only technique that he
admitted using was that of meditation: through yoga, he believed, he was able to find
what he called ‘le son juste’, or ‘the right sound’ (not, he specifies, ‘the right note’),13
which allowed him to perceive cosmic sound and, thus, his music. Indeed, like many
improvising musicians,14 he seems to have interpreted the automaticity of his
improvisations, and the resultant feeling that they were creating themselves, as
evidence of his communication with higher powers rather than of practice and skill
acquisition. He viewed his music as a fragment of an autonomous, cosmic sound15
and himself as a passive vessel, a ‘postman’16 delivering the music. Such beliefs have
distracted attention from one of his main innovations, the ‘preservation’ in scored
works of an element of automaticity, comparable to that of the surrealists in the visual
arts.17
The literature on Scelsi has focused more often on his anti-constructivist
philosophy of sound than on his automatist working methods. There were several
reasons for this: the influence of Scelsi’s own statements, embarrassment at the
controversy that arose when the composer Vieri Tosatti revealed his role in the
creation of many of the scores,18 and, above all, the fact that scholars had no access to
the tapes of the original improvisations until 2009, when the Fondazione Isabella
Scelsi made them available in digitised form.19 Several authors have reflected Scelsi’s
own conviction that his music represented a fundamentally novel and exceptional
relationship between sound and musical organisation, that is, that sound was no longer
constrained by considerations of construction. Thus, for instance, Castanet and
Cisternino wrote: ‘an absolutely innovative aspect in the spectral dimension of sound
in Scelsi is the absolutely a-constructive modality with which such thinking takes
form, substance [...]; sound here is not thought of as a material to be treated with
more or less numerical-artisanal techniques and exercises, but rather as a sound-
Klang; a sort of primordial sound [...].’20 In a similar vein, Martin Zenck argued that
‘tone is understood not as a material with determined historical sediments and
compositional implications [...] but rather as a matter whose own dynamism the
composer emphasises.’21 Some writers have seized on Scelsi’s remark that he did not
‘com-pose’, and have suggested that his musical process was instead one of
‘dissolution’ or ‘de-composition’.22
 4

The notion of the irreducible, non-composite character of Scelsi’s music has

long been exaggerated. For some works he used overdubbing, and it seems that he
was able to play two keyboards simultaneously.23 His works are thus undeniably
composite even when they are not polyphonic in the traditional sense. Moreover,
listening to the tapes one can clearly identify ‘boundaries’ at which Scelsi performs
specific mechanical operations at the ondiola (glissandi, changes of timbre, etc.). Such
boundaries are also inferable from the scores and (less clearly) from performances.
Scelsi’s mature style is generally thought to require a new analytical approach,
again because of its supposed seamless quality. Tristan Murail has said that ‘[i]t is
almost impossible to analyse most of Scelsi’s works in formal terms. Time unfolds in
continuous motion without a break.’24 He has also suggested a statistical approach to
Scelsi analysis:

Music always has a model, whether formal or natural. Even the most
abstract art proceeds from models. What is Scelsi’s model – how can one
analyse his music without resorting to a simple and useless description?
The traditional tools of analysis are inappropriate, since there is neither
material, nor combination, nor a clearly articulated form. There remains
the study (perhaps with statistical methods) of shapes, densities, changes
of register and thickenings, of their evolutions and relationships.25

The most thorough recent analyses of Scelsi have adopted the kind of approach
suggested by Murail. Christine Anderson’s work on Anahit attempts to pin down
Scelsi’s notion of ‘sound as energy’ by studying the distribution of loudness, density,
ambit, and various types of rhythmical activity across the piece.26 Johannes Menke
adopts a similar approach to the Tre canti sacri and Konx-Om-Pax, discussing first the
whole work, then each movement according to various criteria (form, dramaturgy,
proportions, and so on), and also giving a thorough classification of types of gesture.27
The underlying problem with Scelsi analyses is that they usually provide a
‘neutral’ analysis of the large-scale proportions, pitch structure, and other features of
the finished works, often focusing on the more polyphonic works, such as Anahit, i.e.,
those with a relatively traditional pitch structure. Given Scelsi’s use of improvisation,
this is an unduly architectonic, teleological approach that neglects the style’s most
unorthodox and perplexing aspect, that is, the redundancy and apparent irrationality of
 5

the musical surface. It leaves us little closer to understanding how these works were
generated, precisely why a particular inflection of intonation might be followed by a
particular change in vibrato, and so on. It seems to me that an analysis of a work by
Scelsi ought to attempt what Jean-Jacques Nattiez calls ‘the inductive move from
analysis of the neutral level to the poietic’,28 that is, it ought to try to characterise the
logic of the improvisations. The answer to Murail’s question (‘what is Scelsi’s
model?’) is ‘the model of improvisation’.
In the following discussion, therefore, I make some preliminary suggestions
regarding this model, in the knowledge, of course, that the improviser is no longer
available to confirm these (not that it would be in character for him to do so). I do not
attempt to reconstruct the precise mechanics of the improvisations (his uses of
keyboards and recording equipment), but rather to scrutinise his most typical
strategies and configurations of gestures. My suggestions are based mainly on the
scores, and assume the ‘accuracy’ of the transcriptions (in fact, the mediation of the
person transcribing the improvisation is another essential poietic element of the
music, although it is not always acknowledged as such).
I focus on Scelsi’s most characteristic style, the ‘one-note’ heterophony of the
Quattro pezzi and the late string chamber music (strings are particularly suited to this
style). My examples are from Xnoybis (1964) for solo violin and the Duo (1965) for
violin and cello. I then relate these to other, less severely constrained works, using
Dharana (1975) for cello and double bass as an example of the way in which Scelsi’s
model interacts with traditional voice-leading patterns.
The theoretical stimulus for my analysis comes partly mainly from writings on
musical grammar and style by Mario Baroni, Rossana Dalmonte, and Carlo Jacoboni,
and Fred Lerdahl. I have also been influenced by the idea of the improvisational
model described, with reference to music of oral traditions, by the ethnomusicologist
Bernard Lortat-Jacob and his colleagues, and by their approach to representing such
models with diagrams. Lortat-Jacob regards the model as ‘a stable reference’, which
can be of various kinds, but which is ‘at least implicitly known by the musician and
perceived by the hearer in proportion to his/her familiarity with the genre, form, or
style of music’;29 each improvisation constitutes a ‘realisation’ of the model, although
the realisations can also influence the model.30 Needless to say, an individual
improvisation model is different in status from communal, traditional ones; on the
 6

other hand, Scelsi’s attitude throughout his writings encourages a quasi-

ethnomusicological interpretation of his work.31

The concept of musical grammar and Scelsi’s morphology

I do not attempt here to formalise a comprehensive ‘Scelsi grammar’ in the way, for
example, that Baroni, Dalmonte, and Jacoboni formalised that used by the Italian
Baroque composer Giovanni Legrenzi in his chamber arias, i.e. testing the rules by
programming a computer to compose in the given style.32 Such a formal grammar
could never be as authoritative in the case of the relatively small, highly individual,
and varied output of a deceased improviser. However, I think the idea of musical
grammar throws light on Scelsi’s work in several ways. Indeed, Scelsi’s own notion
of the intuitive rightness of the son juste seems to invite the comparison.
The concept of musical grammar implies that it is possible to describe
exhaustively the systems of rules by which pieces are generated and because of which
they are recognised as belonging to collectively legitimated styles. This way of
thinking is derived from Saussure and Chomsky, and presupposes a shared historical
langue, preceding the individual parole, and an innate competence on the part of
listeners for learning the langue explicitly or implicitly (as in the case of untrained
listeners who are nonetheless able to detect a mistake in an unfamiliar example of a
familiar style). In general, it is doubly problematic to apply the concept of musical
grammar to avant-garde, scored music, not only because the avant-garde composer
aspires to invent the langue as well as the parole, but also because, in practice (as a
consequence of the ever greater reliance on writing rejected by Scelsi), the new
langues often challenge perceptual processing to such an extent as to make this kind
of ‘learning’ extremely difficult: the avoidance of pitch hierarchy in serial music is
the most obvious example. Fred Lerdahl has argued that music can be ‘cognitively
opaque’; when its ‘compositional grammar’ is not constrained by a ‘listening
grammar’, it becomes divorced from the subsequent listening grammars. Lerdahl
suggested that it was possible for the composer, by imposing certain ‘cognitive
constraints’, to achieve cognitive ‘clarity’ without resorting to existing norms and
techniques of composition.33 Incongruous as it may at first sound, it seems to me that
this is what Scelsi achieved with his austere pitch constraints, although he would
never have expressed the matter in such terms. These constraints, as well as favouring
the perception of smaller variations of pitch than would be perceptible in music with a
 7

richer pitch structure (indeed, Scelsi’s idiom resembles the ideal experimental
conditions for investigating this area), also allow the listener to intuit the music’s
langue, its ‘rules’, and to experience satisfaction at their fulfilment, or suspense at
their delayed fulfilment. For example, one soon learns to expect quartertone dyads to
converge into unisons. This element of expectation and fulfilment might also be
regarded as a ‘traditional’ trait.
While Lerdahl’s attempt to recover a sense of ‘grammatical’ communication is
that of a composer-theorist, Scelsi’s ‘compositional grammar’ was determined by his
way of improvising, and was thus constrained in minute detail by a ‘listening
grammar’, a feedback system consisting of a real-time assessment of his own musical
decisions. Indeed, the gestures used by Scelsi also are also conventions of common-
practice ‘musicianship’, at least for instruments with flexible intonation: strings, voice
(the main media for Scelsi’s late work). Dissonances are often louder than their
resolutions, leading-notes are a little sharp, vibrato is used to shape musical ‘breaths’,
and so on.34 In the Western canon, because these conventions are only occasionally
explicit in the score, they are assumed to be extraneous to the ‘core’ syntax of the
work, that is, to a system of ‘notes’ (although they are understood to be necessary for
a valid realisation of the work). In Scelsi, on the other hand, in the absence of such
systems, these conventions (arguably because they are ‘all that is left’) seem to
constitute, that is, are experienced as the core syntax. Significantly, the fact that these
performance conventions are already familiar to Scelsi’s listeners reinforces the
collectively legitimated sense of expectation and fulfilment that I mentioned above.
Perhaps the brilliance of Scelsi’s overall artistic strategy is that the reduction of his
music discourages both the conservative and the avant-garde listener from perceiving
this underlying familiarity. Only when these performance conventions are prescribed
on paper is one invited to hear them as avant-garde compositional gestures.
At this point, I should confront two possible objections to my use of the term
‘grammar’ with reference to Scelsi. First, one might be sceptical because of the
eccentricity of Scelsi’s working methods, combining automatist improvisation, tape
editing and collaborative transcription, and because of the difficulty of reconstructing
them. However, as Baroni, Dalmonte and Jacoboni point out, it is not only legitimate
but necessary to distinguish ‘structural rules’ from ‘application procedures’:35 the
importance of the distinction is demonstrated by the fact that a human composer and a
 8

computer can adhere to the same (or very similar) structural rules, using completely
different application procedures.
The other objection would be to do with Scelsi’s apparent avoidance of
discrete units. Musical grammar ought to include morphology, defined by Baroni as
‘the identification of different categories of musical structures’, and syntax, ‘rules
connecting morphological unities’.36 As I discussed above, Scelsi is normally
regarded as the composer to whom these terms are least appropriate, because in his
music it is difficult to segment morphological ‘unities’ or distinguish them from their
syntactical relations. For example, would a tremolo in Scelsi count as a morphological
unit or as a syntactical relation between adjacent pitches? When does vibrato end and
tremolo begin? How can one establish rigorous criteria of segmentation in music that
avoids even discrete ‘notes’?
In fact, however, this sort of ambiguity is not unique to Scelsi’s music.
Discussing tonal music, Baroni points out that a familiar phenomenon (for instance, a
triad) may be experienced either as a syntactical relation (between three notes), or, on
a higher level, as a ‘morphological phenomenon’. The same can be said of Scelsi’s
tremoli and glissandi. It would also be a mistake to assume that Scelsi’s avoidance of
discrete units implies the absence of morphological categories. As to the ‘note’,
Baroni continues:

The central morphological concept, common to many musical

cultures, is that of “note”. It can be defined as central because many
musical concepts can be conceived as micro-categories that are
necessary in order to define a note (as traits of a note), while other
concepts can be conceived as macro-units formed by a number of
notes. For example the absolute and relative pitch of a note, its being
a particular degree of a scale, its duration, its metrical accent, are
morphological characteristics, or micro categories, or traits,
necessary to define a particular note, while an appoggiatura is a
morphological unit composed by two notes.37

All of this applies to Scelsi. The peculiarities of his music are that its discourse
focuses on the traits rather than on notes or configurations of notes, and, indeed, that
he makes ‘note abstraction’ as difficult as possible. Even so, the concept of ‘note’
 9

retains its traditional centrality. After all, most listeners confronted with Scelsi’s
music will immediately remark that ‘it’s all one note’, although this is not strictly the
case.

Scelsi’s monotone-based heterophonic style

In this style, Scelsi fuses parts together to create the illusion of a single, complex
sound object; strong individuation of parts occurs but is exceptional. He achieves this
effect of fusion partly by imposing global pitch constraints – ‘harmony’ is based on
unisons, octaves, and quartertones,38 ‘melodic motion’ is mainly by quartertone (or,
less often, semitone) step or glissando – but also by using local disguise strategies.
Pitch, dynamics, vibrato, and register consistently interact in ways that allow Scelsi to
control and often frustrate the listener’s perception of independent parts and of pitch
motion. Without such strategies, it would be difficult not to hear the late pieces simply
as studies in oblique quartertone motion; this would be precisely the kind of
interpretation, privileging pitch relations, of which he disapproved, and which he
sought to discourage.
These strategies involve the manipulation of various categories of perception,
especially salience (generally defined as ‘probability of being noticed’), grouping of
simultaneous sounds, and pitch centrality. Scelsi usually maintains a high degree of
ambiguity and tension with regard to at least one of these categories. I shall discuss
each in turn, beginning with salience.
The music is segmented as follows: a new cell beginning each time a sound
gesture begins or ends, and each time there is any notated fluctuation of pitch, or any
change in the relative loudness, relative extent of vibrato, or timbral relationship of
parts. I assume that the durations of cells are partly contingent on Scelsi’s
psychological process of finding the next ‘right sound’, and partly manipulated for
dramatic purposes, usually with an increase of ‘harmonic rhythm’ in the middle of
each movement.

Salience
In the following analyses, salience is associated with the following: asynchrony (a
single, distinct inflection is more salient than one disguised with another,
simultaneous inflection), increment of texture (a newly introduced part will tend to be
more salient than a continuing part), increment of pitch (the onset of a new gesture at
 10

a distinct pitch, or oblique motion from unison to a distinct pitch – the two types of
event are difficult to tell apart) or pitch motion (from one distinct pitch to another),
loudness, extent of vibrato, and brightness of timbre.39 In other words, almost any
fluctuation in any of the traits characterising Scelsi’s music affects the relative
salience of parts.
With regard to pitch, it is likely that salience will tend to be greater when the
interval between the distinct pitch and the nearest continuing pitch is wider (with the
exception of octaves and fifths), not least because wider intervals are less frequent
than quartertones and semitones; however, I do not attempt to ‘rank’ these intervals
here. We should also note that an increase in extent of vibrato, as well as being a
salient trait in its own right, also constitutes a marginal kind of pitch ‘increment’, but,
at the same time, limits the distinctness of the pitch.
One of Scelsi’s main strategies is to ‘distribute’ salient traits between parts, so
that they compete for attention. For instance, one part may be salient in pitch, another
in loudness, and so on. The balance is, of course, affected by the degree of emphasis
in each trait (a fortissimo may intuitively be felt to ‘outweigh’ a subtle change of
pitch, for instance). However, it would be difficult to quantify these traits or justify
any hierarchy.
Ex. 1 shows the opening of the third movement of Xnoybis. In this work, the
violin is tuned f–g1–b1–d<sharp>2, to allow unisons and quartertones to be played
more easily among the upper strings, and is distorted with a special mute. The score,
characteristically, uses a separate stave for each string. Xnoybis allows us temporarily
to disregard the perception of grouping, as it is rarely in more than two parts. This
particular movement is also without octave doubling.
Fig. 1 highlights distribution of salient traits in the same passage. Each row
corresponds to a cell; each unbroken vertical block is a gesture. Each new gesture
begins a new column, moving from left to right; this is to allow us to compare Scelsi’s
treatment of onsets. The number of the gesture (in square brackets) and the bar and
beat numbers are shown in the left margin, timbre indications in the right margin.
Vibrato is indicated with bold italics, the relative loudness between parts with arrows
(the arrow points towards the louder component; double-headed arrows indicate equal
loudness), pizzicato with +, and glissando with N. All the pitches are in the d2
register.
 11

This schematic form allows various observations. First of all, we can see how
Scelsi approaches increments of pitch. More often than not these coincide with onsets
of gestures. However, he usually limits the salience of new pitches and of onsets by
making them quieter than the continuous sound. In the sixteen bars, only the onsets of
gestures [9], [11], [14], and [16] are louder than the continuous sound, and, in each
case, the new pitch is the lower pitch; this might be defined as another disguising
factor, a factor limiting salience. He also blurs increments of pitch by following a
general ‘rule’ that either the continuous pitch or the new pitch should be vibrato. In
the central part of the movement he breaks this rule, introducing more salient onsets.
The result is a form common in his music: polyphonic perception is blocked, then
temporarily encouraged (especially when the interval between parts stretches to a tone
and a quarter), then blocked again.
Scelsi’s concern with onsets perhaps explains the occurrence of ‘false starts’:
entries [3] and [4] are false starts in relation to [5], likewise [6] and [7] in relation to
[8]. It is impossible to say whether these false starts were ‘genuine’ or rhetorical, or
somewhere in between.
Another tactic is the ‘exchange of traits’ that occurs between parts in bar 2.
Here the traits of relative loudness and vibrato are swapped between parts:

d ¬ c<1 ½ sharp>
d ® c<1 ½ sharp>

The aural result is almost indistinguishable from this:

d ¬c<1 ½ sharp>
c<1 ½ sharp> ¬d

So far I have disregarded the influence of the colour of the strings, which
undoubtedly affects the listening experience significantly. For instance, the opening d2
is assigned to the softer third string, so that the lower auxiliary appears on a higher,
brighter string (the first movement also begins in this way). This may have been
Scelsi’s intention from the outset, or merely a practical solution to the problem that
the alternative distribution would force the performer to sustain the same relatively
 12

uncomfortable hand position for longer (between bars 6 and 12). In either case, I am
supposing that the underlying model, which is my target here, preceded such fine
points of instrumentation and timbre, even if they were conceived before the
particular improvisation that was arranged into this piece. Indeed, many of the
improvisations are significantly different in timbre from the transcriptions, and also
less ‘fine-grained’ in their variety of timbre. Some pieces, such as Maknongan (1976),
exist in versions for more than one instrument or voice. In view of this, is it so
narrow-minded to wonder whether timbre might be a ‘secondary parameter’ in
Scelsi’s late style?40 He undoubtedly subverts the standard relationship between
‘nuance perception’ and ‘categorical perception’, but it is not in his power to abolish
the distinction altogether.41

Grouping
Textures with more than two parts are exponentially more complex and difficult to
represent, since each part interacts with all the others in the ways described above.
Indeed, it may be that these more complex works were generated in a more haphazard
way, that Scelsi could not have processed (‘automatically’ or otherwise) all of these
interactions in such a honed way as he does in Xnoybis. Some dynamic detail may
also have been lost in the transcription process. The analyst therefore needs to be even
more careful not to introduce arbitrary or misguided interpretations.
The works with several parts introduce a new perceptual element, that of the
grouping together of simultaneous sounds. Again, this can be associated with several
competing criteria: synchrony (grouping simultaneous inflections together), register
(grouping clusters together), pitch-class (grouping unisons and octaves together),
uniformity of loudness, or timbral uniformity. In other words, there are several kinds
of similarity or proximity that the ear may privilege from one moment to the next. In
live performance, the spatial separation of instruments will influence perception, but
this does not necessarily reflect the original model.
The complexity of grouping in Scelsi’s textures is illustrated in Ex. 2, from the
first movement of the Duo for violin and cello of 1965.42 The opening two bars can be
represented in the same way as above (Fig. 2). However, in bar 3 the question of
grouping arises with the introduction of a third part, the bass pedal. Two groupings
are plausible: either the two Gs form a unit distinct from the f<1 ½ sharp>1, or the
cluster material in the violin forms a unit distinct from the cello pedal. Which is more
 13

likely to have been Scelsi’s intention? Of course, there can be no clear answer to this
(the ambiguity is intentional), but there are several reasons to favour the latter: first,
the pedal G is the unprecedented element; second, the vibrato of the g1 distinguishes it
from the bass G (it is not a ‘clean’ octave) and fuses it with the auxiliary; third, the
bass is quieter; fourth, the fact that the bass is quieter makes it likely that the perceiver
will notice it fractionally later than the quartertone motion. This interpretation is
reinforced by the fact that the cluster material and the bass pedal are subsequently
inflected independently: the cluster is ‘inverted’ to the quartertone above the pitch
centre (bar 6), and the pedal is doubled at g1 (bar 4), then g (bar 8).
A similarly detailed consideration of these categories is required each time a
new gesture begins. Fig. 3 describes bars 1-13, using brackets to suggest possible
groupings. One can see how Scelsi’s attention and the focus of his decision-making
process shift from one part of the texture to another, usually allowing one group to
continue unchanged, while he inflects another. Changes in overall grouping are less
frequent than adjustments to the relative salience of parts within groups. Most of the
activity between bars 3 and 9 is ‘nested’ within what we might call a ‘middle-ground’
level. From a structural point of view, the passage could be reduced further (Fig. 4);
there is an interesting self-similarity between this and Fig. 2.
Sometimes grouping is influenced directly by salience, in the sense that a
particular gesture is so salient as to seem extraneous. Such events create a sense of
expansion, a sudden move to a higher level of grouping (the previous groups are
suddenly reinterpreted as subgroups); this multiplicity of implied levels may also
correspond to Scelsi’s concept of ‘depth’. In this passage, the most extraneous
element is the vibrato d<1 ½ sharp>2, especially because of its pitch-class (which
suddenly reveals that this is not a ‘one-note’ piece), but also because of its register
and timing (it occurs after a relative relaxation of activity); in performance, the bright
timbre of the open E pizzicato also marks it out. The f<sharp>3 in bar 9, similarly, is
considered extraneous in register, pitch-class (it is not a quartertone sharp and thus not
in octaves with the cello auxiliary), loudness, and timing (it seems to ‘cut off’ the
vibrato); the only other possible grouping would be with the harmonic g1 in the cello,
but the latter is masked. Sometimes, on the other hand, new gestures seem to emerge
not from the ‘outside’ but from within a particular group, e.g. the cello f<1 ½ sharp>1
in bar 9, which grows out of the cluster group.
 14

Grouping is even more elusive in the second movement of the Duo (Ex. 3).
The opening is again representative of the general rhetoric. In the third beat of bar 1
there are two possibilities: to group together the violin g<1 ½ sharp>1 and the
harmonic a1 in the cello on grounds of simultaneity and register, or to group together
the two G<1 ½ sharp>s on grounds of pitch-class and vibrato. This is a more
ambiguous and difficult case (and therefore particularly successful, from Scelsi’s
point of view) and it is only clarified when the a1 ends, inviting the listener
retrospectively to group together the vibrato octave. Indeed, part of the ambiguity of
this case is that the criterion of simultaneity holds only momentarily, so that one may
group the simultaneous onsets on the third beat together and then reinterpret the same
texture, hearing the vibrato octave as a unit.
This passage continues in a similar vein (Fig. 5). The cello a in bar 2, beat 3
can be grouped with the quartertone auxiliary, which now describes a tremolo with
the same a and is thus no longer so closely tied to the sound an octave above.
However, the sudden doubling of the pitch A in bar 3 again shifts the grouping so that
the cello’s g<1 ½ sharp> is suddenly extraneous.
A comparison of Figs. 3 and 5 highlights the contrast between the two
movements. In the first movement, grouping was influenced above all by register (so
that clusters tended to be grouped together); in the second, it is influenced alternately
by register and by pitch-class (so that octaves are grouped together), with the result
that the same component of the texture can be grouped differently from one moment
to the next. It is a virtue of Scelsi’s improvisation model that it allows this kind of
effective contrast in spite of the uniformity of the material. The contrast is nothing to
do with ‘sound itself’, however.

Pitch centrality
In a cluster, the pitch likely to be interpreted as ‘central’ will usually be the one that is
continuous (not newly introduced) and thus stable; if several are continuous, the ear
can be expected to orient itself to a pitch that is doubled at the unison or (more so) at
the octave; possibly, absence of vibrato may also encourage this orientation; other
things being equal, the ear will probably orient itself towards the loudest pitch. The
opening of the third movement of Xnoybis, discussed above, presents an apparently
simple example (Fig. 6). Overall, the c<1 ½ sharp>2 is predominant: is it not only
 15

‘more continuous’ (that is, less frequently interrupted) than the d2, but also louder
(except at the very beginning).
Although pitch centrality is the first notion that springs to mind where Scelsi is
concerned, it seems on reflection even more difficult to assess rigorously than
salience and grouping, not only because there are numerous situations of conflict (for
instance, when one continuous pitch is octave-doubled and another is relatively loud)
but also because it is inherently retrospective. There is no fixed point at which a pitch
‘becomes’ continuous or established. Moreover, frequently recurring pitches are
likely to be retained in the listener’s memory, and will thus be ‘virtually’ if not
literally continuous. Scelsi can therefore only control a sense of pitch centrality
directly when one pitch-class is evidently more stable than any other; as soon as there
is any doubt, matters of salience and grouping will interfere.
This can be seen by again comparing the two movements of the Duo. In the
first movement the pitch centre is never in doubt: it is the almost unbroken g1, often
doubled two octaves below by the cello, and sometimes two or three octaves above as
a harmonic (neither doubling is ever marked vibrato). The pitch-class G is only
interrupted once, at bar 22, revealing an upper auxiliary in the treble register of the
cello, but after two beats it is loudly reasserted. As a result there is a mutually
reinforcing opposition between the centrality of the pitch-class G and the salience of
the incremental pitches, which appear in increasingly complex and dramatic
configurations. This creates the centrifugal rhetoric on the musical surface.
In the calmissimo second movement, on the other hand, the pitch centre is
much less certain. This is because the pitch-class A and its quartertone lower auxiliary
are both used, alternately, as pedals (often doubled, but with no bass ‘anchor’), but are
also both frequently interrupted. Since there is not such a clear opposition between
continuous and incremental pitches, the listener’s perception of pitch centre is likely
to be influenced positively by other salience criteria, especially loudness. The
uncertainty of the pitch centre seems to be the rhetorical focus of the movement, its
source of tension.
To put it another way, the relationship between salience and pitch centrality is
one of negative correlation at moments of pitch increment (since the centrality of the
continuous pitch is defined against the salient element, i.e., the new pitch) and
positive between moments of pitch increment (when no element is continuous and
thus centrality starts to be influenced by intensity).
 16

Tonal and post-tonal allusions in the non-monotone-based works

Not all of Scelsi’s late work is in the idiom that I have analysed above. Some pieces
are in its ‘polyphonic’ equivalent; that is, the overall harmonic constraint is relaxed to
include other sustained intervals (not only unisons and octaves) and chords (often
triads), even though the constraint on melodic motion is usually sustained. This is the
case in the Tre canti sacri (1958) for choir, the violin concerto Anahit (1965), and the
music for large string ensemble. These works consist of simultaneous drones, each of
which is elaborated using the peculiarly Scelsian strategies described above, but
which, when combined, allude to (Western) harmony and counterpoint.
For example, Dharana (1975) (Ex. 4), for cello and double bass, is based on
the major third F–A. Associations of F major are reinforced by the fact that the pitch-
class F is dragged down to its lower auxiliary E (sometimes E<quartertone-sharp>,
which can be heard, at least by the Western listener, as an expressively raised leading-
note), while the A is dragged upwards to A <quartertone-sharp> (heard as a
downward-leaning B<flat>), and that these ‘auxiliaries’ resolve inwards again, like
the E and B<flat> in the dominant seventh in F. It is hard not to suspect that Scelsi
specifically tries to soften this allusion at various points: for example, in bars 9-10,
when the upper auxiliary relaxes onto a, the f in the cello also falls by a quartertone
and increases into loudness.
A different association is prompted when the double bass doubles the E: this
encourages one to reinterpret the tonality as A minor, with a dominant pedal,
especially when the A is then pulled up almost to B<natural> (Ex. 5). Here it is hard
not to interpret the indication ‘quasi <natural>’ (that is, Scelsi’s avoidance of the
compound perfect fifth E–b) as another attempt to attenuate a tonal reference.
Another well-known tonal allusion in Scelsi’s work is the progression from a
G<flat> triad to a 6/4 chord on F that occurs in the first and fourth movements of the
Third String Quartet. Similar progressions abound in Anahit. In Anagamin (1965), he
plays on the tonal implications of the seventh between C and B<flat>. Ohoi (1966)
uses quartal harmony.
These works manifest a delicate interaction between Scelsi’s individual
procedures (regarding nuance) and received syntactical structures (regarding voice-
leading and so on). He thus sets himself an interesting problem, that of preventing the
 17

(Western) listener from relegating all the detail of nuance back to the status of
‘accessory’, back to the realm of performance practice.
The stylistic echoes in Scelsi are not always tonal. There are works that share
traits with other post-war avant-garde composers, such as Xenakis. These include the
pieces for choir and orchestra Uaxactum (1966), Konx-Om-Pax (1969), and Pfhat
(1974), in which drones are opposed with clusters and untuned percussion. Julian
Anderson has noted Scelsi’s debt to Nono in some of his vocal writing.43 The overtly
melodic, modal, archaic style of Antifona and Three Latin Prayers of 1970 constitutes
a third type of stylistic allusion (an eccentric one even by Scelsi’s standards). Many of
the late works, such as Ko-Lho (1966) for flute and clarinet, recall the oscillatory style
of the transitional works of the 1950s.
Other late pieces are exceptional in that the original improvisations were made
on non-keyboard instruments, or were collaborative. Certain performers had such a
significant input in some improvisations that they can be viewed as co-authors of the
resulting works. The outstanding example is the cycle Canti del Capricorno (1962-
1972), for which the soprano Michiko Hirayama effectively turned herself into the
instrument of Scelsi’s improvisation. In such cases, Scelsi’s improvisational model is
fused with that of another individual (even if the latter is attempting to improvise in
the style of Scelsi).

Conclusions
The thrust of my argument is that, in Scelsi’s music, the impression of a magical
emancipation of sound from any syntactical system is achieved by a manipulation of
morphological traits that is so systematic as to constitute a kind of syntax. This is not
to say that his music does not also, on another level, transcend the syntactical; but this
is true of any music.
My observations are based on study of the scores and are intended as
preliminary suggestions, which could be refined by a more detailed comparison with
the tapes of the original improvisations. Such a comparison might also allow a more
rigorous assessment of the mediation and contribution of the transcribing assistants.
Indeed, although the analysis of the tapes is a long overdue step in Scelsi research, it
is not certain that the tapes alone will allow scholars to reconstruct his working
process in as much detail as they might wish, and such a reconstruction would still not
necessarily reveal his music’s structural rules. It could also be fruitful to measure
 18

redundancy in various aspects of the music, using the scores as well as the tapes, and
to conduct listening experiments; perhaps a ‘listening grammar’ could be described.
One of the most interesting and original aspects of Scelsi’s music is that he
was able to realise his model or grammar automatically in the act of improvisation, as
a result of extensive practice. Another is its relation to the grammar or grammars of
musicianship: those semi-explicit rules determining when to widen vibrato, when to
sharpen the leading-note, when to slow down, and so on. In Western concert music,
these nuances, which constitute such an important – if ‘ineffable’ – part of musical
communication, are normally applied to an existing, written-down musical structure;
in Scelsi, the manipulation of this type of nuance is what generates the structure.
 19

Notes

An early version of this article was delivered during the Cardiff Music Analysis
Conference in 2008. I would like to thank the staff of the Fondazione Isabella Scelsi
for their generous assistance.
1. See Bernard Lortat-Jacob, ‘Improvisation: le modèle et ses réalisations’, in Lortat-
Jacob (ed.), L’improvisation dans les musiques de tradition orale (Paris: SELAF,
1987), pp. 45–59. The concept is referred to in several other chapters in this volume.
2. Jeff Pressing, ‘Cognitive processes in improvisation’, in W. R. Crozier and A. J.
Chapman (eds.), Cognitive Processes in the Perception of Art (Amsterdam: North
Holland, 1984), pp. 346–51.
3. See Eric Drott, ‘Class, Ideology, and il caso Scelsi’, Musical Quarterly, 89/i
(2006), pp. 80–120.
4. Franck Mallet (ed.), ‘Conversations avec Giacinto Scelsi’, in Giacinto Scelsi, Les
anges sont ailleurs…, collected writings and interviews, Sharon Kanach (ed.), trans.
Sharon Kanach, Irène Assayag and Fiorella Edel (Arles: Actes Sud, 2006), p. 64. This
interview was originally broadcast on Radio France: France Musique in 1987.
5. Ibid., p. 66. Scelsi also claimed that, while recovering in a Swiss clinic, he had
developed the habit of playing single notes on the piano and meditating on their
acoustic depth. With this story he encouraged the interpretation of his later work as an
extension of this ‘therapeutic’ practice.
6. Pressing 1984, pp. 357–9.
7. This is the most controversial element of his methods. It has often been justified on
the ground that it was consistent with his overall attitude, undermining the centrality
of the written text. In his view the musical ‘work’ was established during the moment
of inspiration, and transcription was a mechanical task that did not – indeed must not
– contribute to the work. Indeed, it is hard to imagine how Scelsi could have
transcribed the music himself without succumbing to the temptation to start editing it
rationally. On the other hand, there is no escaping the fact that he did not credit his
assistants.
8. It appears that Scelsi used two claviolines, confusingly inscribed with the word
‘ondiola’. Elisabetta Piras, Mario Baroni, Gianni Zanarini, ‘Improvvisazioni di
Giacinto Scelsi: il caso problematico dell’ondiola’, i suoni, le onde... Rivista della
 20

Fondazione Isabella Scelsi 19/20 (Roma: Fondazione Isabella Scelsi, 2007-2008), pp.
6-7.
9. Gregory N. Reish, ‘Una nota sola: Giacinto Scelsi and the Genesis of Music on a
Single Note’, Journal of Musicological Research, 25 (2006), pp. 150-60. Reish aptly
describes Scelsi’s late aesthetics as ‘sonorist’.
10. Mallet 2006, p. 83.
11. Ibid., p. 64. Also Giacinto Scelsi, ‘Son et musique’, in Scelsi, Les Anges sont
ailleurs…, p. 126 (originally Rome and Venice: Le Parole Gelate, 1981).
12. Scelsi, ‘Son et musique’, pp. 131–2. Scelsi observed that post-Webernian and
even chance music were, like tonal music, conditioned by numerical systems that
preceded the sound ‘itself’, even when the artistic intention was to give particular
attention to sound quality (this also applies to spectral music).
13. Scelsi, ‘Son et musique’, pp. 128–9.
14. Jeff Pressing, ‘Improvisation: Methods and Models’, in John Sloboda (ed.),
Generative Processes in Music: the Psychology of Performance, Composition and
Improvisation (Oxford: OUP, 1988), pp. 139–40, 142.
15. Scelsi, ‘Force cosmique’, in Scelsi, Les Anges sont ailleurs…, p. 151.
16. Letter to Sharon Kanach, in Scelsi, Les Anges sont ailleurs…, p. 57.
17. Scelsi was well informed in this respect. He was a friend of Michaux and Dalí,
and co-founded an art gallery specialising in abstract expressionism and art informel.
See Alessandro Mastropietro, ‘Action music (1955)... action painting: Su un nodo
della produzione pianistica di Scelsi e su alcune ipotesi definitorie della sua tecnica
compositiva’, in Daniela M. Tortora (ed.), Giacinto Scelsi nel centenario della nascita
(Rome: Aracne, 2008), pp. 119–44.
18. Vieri Tosatti, ‘Scelsi, c’est moi’, Il giornale della musica 5/35 (1989), pp. 1, 10.
19. This development has demystified Scelsi somewhat. Some progress has been
made in identifying and analysing the tapes, above all by the composer and scholar
Friedrich Jaecker. The first results of this research were presented at the conference
‘Scelsi Ritrovato: nuovi percorsi alla luce delle fonti d’archivio’, Rome, November
11–12, 2010. See also Friedrich Jaecker, ‘ “Funziona? O non funziona?” Ein Streifzug
durch das Scelsi-Archiv’, MusikTexte: Zeitschrift für neue Musik 128, 2011, pp. 5-11.
20. Pierre Albert Castanet, Nicola Cisternino, ‘Giacinto Scelsi, quasi una premessa’,
in Castanet and Cisternino (eds.), Giacinto Scelsi: viaggio al centro del suono (La
Spezia: Luna, 1993), p. 11.
 21

21. Martin Zenck, ‘Das Irreduktible als Kriterium der Avantgarde’, in Heinz-Klaus
Metzger and Rainer Riehn (eds.), Giacinto Scelsi. Musik-Konzepte 31 (Munich: text +
kritik, 1983), p. 70.
22. Heinz-Klaus Metzger, ‘Das Unbekannte in der Musik. Versuch über die
Kompositionen von Giacinto Scelsi’, in Metzger and Rainer Riehn (eds.), Giacinto
Scelsi. Musik-Konzepte 31 (Munich: text + kritik, 1983), p. 14. See also Giulio
Castagnoli, ‘Suono e processo nei “Quattro pezzi per orchestra (su una nota sola)
(1959)” di Giacinto Scelsi’, in Castanet and Cisternino (eds.) 1993, pp. 246–7, and
Tristan Murail, ‘Scelsi De-composer’, Contemporary Music Review 24/i-ii (2005), p.
173.
23. Piras, Baroni, and Zanarini, ‘Improvvisazioni di Giacinto Scelsi’, pp. 8–9. Scelsi
would also exploit the possibilities offered by his recording equipment, for example,
reversing the direction of the tapes.
24. Tristan Murail, ‘Scelsi and L’Itinéraire: The Exploration of Sound’, trans. Robert
Hasegawa, Contemporary Music Review 24/ii-iii (2005), pp. 183–4. Murail describes
this kind of time as ‘temps lisse’, smooth or polished time.
25. Murail, ‘Scelsi, De-composer’, p. 179.
26. Christine Anderson, ‘Klang als Energie. Anahit von Giacinto Scelsi’, MusikTexte:
Zeitschrift für neue Musik 81/82 (1999), pp. 72–82.
27. Johannes Menke, Pax. Analyse bei Giacinto Scelsi: Tre canti sacri und Konx-Om-
Pax (Hofheim: Wolke Verlag, 2004).
28. Jean-Jacques Nattiez, Music and Discourse, trans. Carolyn Abbate (Princeton, NJ:
Princeton University Press, 1991), p. 88, n. 15.
29. Lortat-Jacob 1987, p. 46.
30. Ibid, p. 52.
31. See Giovanni Giuriati, ‘Suono, improvvisazione, trascrizione, autorialità, Oriente,
... e Scelsi. Alcune riflessioni di un etnomusicologo’, in Tortora 2008, pp. 263–79.
32. See Mario Baroni, Rossana Dalmonte, and Carlo Jacoboni, A computer-aided
inquiry on music communications: the rules of music (Lewiston, Queenston,
Lampeter: The Edwin Meller Press, 2003). Other important contributions to this field
include Terry Winograd, ‘Linguistics and the computer analysis of tonal harmony’,
Journal of Music Theory 12/i (1968), pp. 2–49, Baroni and Jacoboni, ‘Proposal for a
grammar of melody: the Bach chorales’ (Montréal: Presses de l’Universitè de
 22

Montréal, 1978), and Kemal Ebcioglu, ‘An expert system for harmonizing four-part
chorales’, Computer Music Journal 12/iii (1988), pp. 43–51.
33. Fred Lerdahl, ‘Cognitive constraints on compositional systems’, in Sloboda (ed.),
Generative Processes in Music: the Psychology of Performance, Composition and
Improvisation (Oxford: Oxford University Press, 1988), pp. 231–59. Lerdahl’s
argument has been attacked as conservative. For example, see James Boros, ‘A “New
Totality”?’, Perspectives of New Music 33/i–ii (1995), pp. 538–53. Perhaps I should
specify that I do not consider ‘cognitively opaque’ music inferior. See also Mario
Baroni, ‘GTTM and post-tonal music’, Musicae scientiae. Discussion Forum 5, 2010,
pp. 69–93.
34. See Mieko Kanno, ‘Thoughts on how to play in tune: pitch and intonation’,
Contemporary Music Review, 22/i-ii (2003), pp. 49–51. As Kanno explains, in works
such as those of Scelsi, the performer still has ample responsibility in this regard.
35. Baroni, Dalmonte, and Jacoboni 2003, p. 14, note 18 on pp. 15–16.
36. Mario Baroni, ‘Musical grammar and the study of cognitive processes of
composition’, Musicae scientiae, 3/i (1999), pp. 3–5.
37. Ibid. p. 5.
38. The intervals that most encourage tonal fusion (i.e. are most likely to be
interpreted as ‘comprising partials of a single complex tone’) are the unison, octave,
and fifth – hence their avoidance in polyphonic idioms. See David Huron, ‘Tonal
consonance versus tonal fusion in polyphonic sonorities’, Music Perception, 9/ii
(1991), p. 135.
39. Similar criteria are used in Fred Lerdahl, ‘Atonal prolongational structure’,
Contemporary Music Review, 4/i (1989), pp. 73–4.
40. The familiar term used in Leonard B. Meyer, Style and Music: theory, history and
ideology (Philadelphia: University of Pennsylvania Press, 1989), pp. 14-16. Xnoybis
requires special metallic mutes, but the score comes with an erratum explaining that
one should ‘find a suitable solution by personal experimentation’. Is it, then, the fact
that the sound is altered that counts rather than the particular character of the new
sound?
41. See Diana Raffman, Language, Music and Mind (Cambridge, MA, and London:
MIT Press, 1993), pp. 83–97.
42. This work is also discussed in Ian Dickson, ‘Orality and rhetoric in Scelsi’s
music’, Twentieth-century Music, 6/i (2009), pp. 34–8.
 23

43. Julian Anderson, ‘La note juste’, Musical Times 136/1823 (1995), p. 23.

Bibliography

Anderson, Christine, 1999: ‘Klang als Energie. Anahit von Giacinto Scelsi’,
MusikTexte: Zeitschrift für neue Musik, 81/82, pp. 72-82.
Anderson, Julian, 1995: ‘La note juste’, Musical Times, 136/1823, pp. 22-27.
Baroni, Mario, 1999: ‘Musical Grammar and the Study of Cognitive Processes of
Composition’, Musicae Scientiae, 3/i, pp. 3-21.
Baroni, Mario, 2010: ‘GTTM and post-tonal music’, Musicae scientiae: Discussion
Forum 5, pp. 69-93.
Baroni, Mario, Rossana Dalmonte, and Carlo Jacoboni, 1999/2003: Le regole della
musica: indagine sui meccanismi della communicazione (Turin: EDT).
English translation, 2003: A computer-aided inquiry on music communication:
the rules of music (Lewiston, Queenston, Lampeter: The Edwin Mellen Press).
Castagnoli, Giulio, 1993: ‘Suono e processo nei “Quattro pezzi per orchestra (su una
nota sola) (1959)” di Giacinto Scelsi’, in Pierre-Albert Castanet and Nicola
Cisternino (eds.), Giacinto Scelsi: viaggio al centro del suono (La Spezia:
Luna), pp. 246-259.
Castanet, Pierre-Albert and Nicola Cisternino, 1993: ‘Giacinto Scelsi, quasi una
premessa’, in Castanet and Cisternino (eds.), Giacinto Scelsi: viaggio al centro
del suono (La Spezia: Luna), pp. 10-13.
Castanet, Pierre-Albert and Cisternino, Nicola (eds.), 1993: Giacinto Scelsi: viaggio
al centro del suono (La Spezia: Luna).
Dickson, Ian, 2009: ‘Orality and Rhetoric in Scelsi’s Music’, Twentieth-century
Music, 6/i, pp. 23-41.
Drott, Eric, 2006: ‘Class, Ideology and il caso Scelsi’, Musical Quarterly, 89/i, pp.
80-120.
Giuriati, Giovanni, 2008: ‘Suono, improvvisazione, trascrizione, autorialità, Oriente,
... e Scelsi. Alcune riflessioni di un etnomusicologo’, in Tortora (ed.), Giacinto
Scelsi nel centenario della nascita (Rome: ARACNE), pp. 263-279.
 24

Huron, David, 1991: ‘Tonal consonance versus tonal fusion in polyphonic sonorities’,
Music Perception, 9/ii, pp. 135–54.
Jaecker, Friedrich, 2005: ‘ “Der Dilettant und die Profis”, Giacinto Scelsi, Vieri
Tosatti & co.’, MusikTexte: Zeitschrift für neue Musik , 104, pp. 27–40.
Jaecker, Friedrich, 2011: ‘ “Funziona? O non funziona?” Ein Streifzug durch das
Scelsi-Archiv’, MusikTexte: Zeitschrift für neue Musik, 128, pp. 5-11.
Kanno, Mieko, 2003: ‘Thoughts on how to play in tune: pitch and intonation’,
Contemporary Music Review, 22/i-ii, pp. 35-52.
Lerdahl, Fred, 1988: ‘Cognitive constraints on compositional systems’, in Sloboda,
John (ed.), Generative Processes in Music: the Psychology of Performance,
Composition and Improvisation (Oxford: Oxford University Press, 1988), pp.
231-259.
Lerdahl, Fred, 1989: ‘Atonal prolongational structure’, Contemporary Music Review
4/i, 1989, pp. 65-87.
Lortat-Jacob, Bernard, 1987: ‘Improvisation: le modèle et ses réalisations’, in B.
Lortat-Jacob (ed.), L’improvisation dans les musiques de tradition orale
(Paris: SELAF), pp. 45-59.
Mallet, Franck, 2006: ‘Conversations avec Giacinto Scelsi’, in Giacinto Scelsi, Les
anges sont ailleurs…, ed. Sharon Kanach, trans. Sharon Kanach, Irène
Assayag and Fiorella Edel (Arles: Actes Sud), pp.62-84.
Mastropietro, Alessandro, 2008: ‘Action music (1955)... action painting. Su un nodo
della produzione pianistica di Scelsi e su alcune ipotesi definitorie della sua
tecnica compositiva’, in Daniela M. Tortora (ed.), Giacinto Scelsi nel
centenario della nascità (Rome: ARACNE), pp. 119-144.
Menke, Johannes, 2004: Pax. Analyse bei Giacinto Scelsi: Tre canti sacri und Konx-
Om-Pax (Hofheim: Wolke Verlag).
Metzger, Heinz-Klaus, 1983: ‘Das Unbekannte in der Musik. Versuch über die
Kompositionen von Giacinto Scelsi’, in Heinz-Klaus Metzger and Rainer
Riehn (eds.): Giacinto Scelsi. Musik-Konzepte, 31 (Munich: Edition text +
kritik), pp. 10-23.
Metzger, Heinz-Klaus and Rainer Riehn (eds.), 1983: Giacinto Scelsi. Musik-
Konzepte 31 (Munich: Edition text + kritik).
Meyer, Leonard B., 1989: Style and Music: theory, history and ideology
(Philadelphia: University of Pennsylvania Press)
 25

Murail, Tristan, 2005a: ‘Scelsi and L’Itinéraire: The Exploration of Sound’, trans.
Robert Hasegawa, Contemporary Music Review, 24/ii-iii, pp. 181-185.
Murail, Tristan, 2005b: ‘Scelsi De-composer’, trans. Robert Hasegawa,
Contemporary Music Review, 24/i-ii, pp. 173-180.
Nattiez, Jean-Jacques, 1991: Music and Discourse, trans. Carolyn Abbate (Princeton,
NJ: Princeton University Press).
Piras, Elisabetta, Mario Baroni and Gianni Zanarini, 2007-2008: ‘Improvvisazioni di
Giacinto Scelsi: il caso problematico dell’ondiola’, i suoni, le onde... Rivista
della Fondazione Isabella Scelsi, 19/20, pp. 3-9. (Roma: Fondazione Isabella
Scelsi).
Pressing, Jeff, 1984: ‘Cognitive Processes in Improvisation’, in W R Crozier ad A J
Chapman (eds.), Cognitive Processes in the Perception of Art (Amsterdam:
North Holland), pp. 345-63.
Pressing, Jeff, 1988: ‘Improvisation: methods and models’, in John Sloboda (ed.),
Generative Processes in Music (Oxford: Oxford University Press, 1988), pp.
129-76.
Raffman, Diana, 1993: Language, Music and Mind (Cambridge, MA, and London:
MIT Press).
Reish, Gregory, 2006: ‘Una Nota Sola: Giacinto Scelsi and the genesis of music on a
single note’, Journal of Musicological Research, 25, 149-189.
Scelsi, Giacinto, 1981/2006: ‘Son et musique’, in Scelsi, Les anges sont ailleurs…
(Arles: Actes Sud), pp. 125-139.
Scelsi, Giacinto, 2006: ‘Force cosmique’, in Scelsi, Les anges sont ailleurs… (Arles:
Actes Sud), pp. 151-153.
Scelsi, Giacinto, 2006: Les anges sont ailleurs…, collected writings and interviews,
ed. Sharon Kanach, trans. Sharon Kanach, Irène Assayag and Fiorella Edel
(Arles: Actes Sud).
Tortora, Daniela M. (ed.), 2008: Giacinto Scelsi nel centenario della nascità (Rome:
ARACNE).
Tosatti, Vieri, 1989: ‘Scelsi, c’est moi’, Il giornale della musica, 5/35, pp. 1, 10.
Uitti, Frances-Marie, 1995: ‘Preserving the Scelsi Improvisations’, Tempo 194, pp.
12-14.
Zenck, Martin, 1983: ‘Das Irreduktible als Kriterium der Avantgarde. Überlegungen
zu den vier Streichquartetten Giacinto Scelsis’, in Metzger and Riehn (eds.):
 26

Giacinto Scelsi. Musik-Konzepte, 31 (Munich: Edition text + kritik), pp. 69-

81.

Musical works consulted

Scelsi, Giacinto, 1988: Duo (1965), for violin and cello (Paris: Salabert).
Scelsi, Giacinto, 1985: Xnoybis (1964), for solo violin (Paris: Salabert).
Scelsi, Giacinto, 1986: Dharana (1975), for cello and double bass (Paris: Salabert).
 27

Abstract

Giacinto Scelsi (1905-1988) is a problematic figure for musical analysis, on account

of his extreme anti-rationalism and devaluing of the score. By the 1950s, he was
creating music through improvisation, and delegating the transcription to assistants.
The idiom he evolved was novel not only in its extreme economy of means, usually
consisting of subtle inflections of continuous sounds, but also in its apparent rejection
of any rational organisation.
Analysts of Scelsi’s work have tended to concentrate on large-scale musical
architecture, neglecting the apparently redundant, non-developmental gestures from
which, nonetheless, this architecture must be built up. Many of Scelsi’s advocates
have encouraged this by insisting on the music’s irreducibility and exceptional rapport
with sound – an argument that stems from the composer’s own mysticism: he
attributed the automaticity of his improvisations to the cosmic power of sound, rather
than to the long hours that he spent creating them.
This article explores the idea that Scelsi’s music is conditioned, if not by an
explicit grammar (traditional or avant-garde), then by the ‘model’ of the original
improvisations, and that his manipulation of nuance can be considered as a kind of
syntax. It argues that a grammatical analysis accounts for the persuasiveness and
variety of the improvisations more plausibly than statistical analysis or metaphysical
formulations involving sound ‘itself’.
 28

Biographical note

Ian Dickson studied Music at King’s College London, and received his PhD in
Composition from the University of York, where he was supervised by Professor
Roger Marsh and supported by the Arts and Humanities Research Board.
 29

Captions for music examples and figures

Ex. 1. Giacinto Scelsi, Xnoybis, for solo violin (1964), movement III, bars 1–16.
Copyright © 1985 Éditions Salabert – Paris, France (SLB 2257)
Reproduced by kind permission of MGB Hal Leonard, Italy

Ex. 2. Giacinto Scelsi, Duo, for violin and cello (1965), movement I, bars 1–22.
Copyright © 1988 Éditions Salabert – Paris, France (SLB 2364)
Reproduced by kind permission of MGB Hal Leonard, Italy

Ex. 3. Giacinto Scelsi, Duo, for violin and cello (1965), movement II, bars 1–4.
Copyright © 1988 Éditions Salabert – Paris, France (SLB 2364)
Reproduced by kind permission of MGB Hal Leonard, Italy

Ex. 4. Giacinto Scelsi, Dharana, for cello and double bass (1975), bars 1–14.
Copyright © 1986 Éditions Salabert – Paris, France (SLB 2421)
Reproduced by kind permission of MGB Hal Leonard, Italy

Ex. 5. Giacinto Scelsi, Dharana, for cello and double bass (1975), bars 57–62.
Copyright © 1986 Éditions Salabert – Paris, France (SLB 2421)
Reproduced by kind permission of MGB Hal Leonard, Italy

Fig. 1. Xnoybis, movement 3, bars 1–16, salience relations.

Fig. 2. Duo, movement 1, bars 1–2, grouping.

Fig. 3. Duo, movement 1, bars 1–13, grouping.

Fig. 4. Duo, movement 1, bars 1–13, grouping (reduction).

Fig. 5. Duo, movement 2, bars 1–3, grouping.

Fig. 6. Xnoybis, movement 3, bars 1–11, pitch centrality.

 30

Ex. 1. Scelsi, Xnoybis (1964), movement 3, bars 1–13. Paris: Salabert, 1985.
 31

Ex. 2.
 32

Ex. 3
 33

Ex. 4.
 34

Ex. 5.
 35

[1] (1.1) d
d
[2] (2.1) d ¬c<sharp>
+

d ®c<sharp> +
c<sharp> +
[3] (3.2) c<sharp> +¬d
c<sharp> +
[4] (4.2) c<sharp> +¬d
c<sharp> +
[5] (4.3) c<sharp> +¬c<sharp>+
c<sharp>+
[6] (5.2) c<sharp>+¬c<sharp> +
c<sharp>+
[7] (5.3) c<sharp>+¬c<sharp>++
c<sharp>+
[8] (6.1) c<sharp>+¬d
d
[9] (6.2) d¬c<sharp>+
c<sharp>+
[10] (6.4) c<sharp>+¬d
c<sharp> +¬d
[11] (7.3) d ®c<sharp>
d ®c<sharp>+
c<sharp>+
[12] (9.1) c<sharp>+¬c<sharp>+
[13] (10.1) c<sharp>+¬c<sharp>+
c<sharp> +¬d
c<sharp> +¬d
d+
d+
[14] (11.4) d +®c<sharp>+
c<sharp>+
[15] (12.2) c<sharp>+¬c<sharp>+
c<sharp>+
d+
[16] (12.4) d+ ®c<sharp>+
[17] (12.4) Nd
d+
[18] (13.1) d+ ¬d
[19] (14.3) Nd
d<sharp>
[20] (14.4) d<sharp> ¬ d+
[21] (15.1) d<sharp> ¬ d++
d<sharp> ¬ d+
[22] (16.2) d<sharp>¬d<sharp>
d<sharp>
[23] (16.4) Nd<sharp>¬e-
 36

Fig. 2.

[1] 1.1 g1
[2] 1.3 g1 f<sharp> +1
2.1 g1 g1
[3] 2.2 g1 g1
2.3 g1 g1

Fig. 3.

[1] 1.1 g1
[2] 1.3 g1 f<sharp>+­1
2.1 g1 g1
[3] 2.2 g1 g1
1
2.3 g g1
[3] 3.1 (g1 (f<sharp>+1/g1)) ¬ G
[4] 4.1 (g 1
(f<sharp>+1/g1)) ¬ (G g1)
4.3 (f<sharp> +1 (f<sharp>+1/g1)) ¬ (G g1)
5.4 (g1 g1) ¬ (G g1)
6.1 (ab -1
g1) ¬ (G g1)
6.4 (ab-1 g1) ¬ G
7.1 (ab-1 (g1/g1)) ¬ (G g)
7.4 (ab-1 g1) ¬ (G g)
[5] 8.1 (g1 g1) ¬ (G g)
8.2 (g 1
(f<sharp>+1/g1)) ¬ (G g)
8.4 (g1 (f<sharp>+1/g1)) ¬ G
9.2 ((g1 ¬f<sharp>+1) (f<sharp>+1/g1))) G
1 +1
9.4 ((g f<sharp> ) g1) G
[6] 9.4
1
(((g f<sharp> ) +1 1
g) g) 1
® f<sharp>3
11.1
+1 1
((f<sharp> /g ) g1) ® f<sharp>3
[7] 12.1
+1 1
(((f<sharp> /g ) g1) ® f<sharp>3) ¬
(d<sharp>+2 e+)
12.2 ((f<sharp>+1 g1) ® f<sharp>3) ¬
d<sharp>+2
13.1 (g1 g1) f<sharp>3
13.2 (g1 g1) (f<sharp>3 ¬ g3)
1
13.4 (g g3)

Fig. 4.

[1] 1.1 g1
[2] 1.3 g1 f<sharp> +1
3.1-8.4 (g1 (f<sharp>+1/g1)) ¬ G
[6] 9.4 (((g1 f<sharp>+1) g1) g1) ® f<sharp>3
11.1 ((f<sharp>+1/g1) g1) ® f<sharp>3
[7] 12.1
+1 1
(((f<sharp> /g ) g1) ® f<sharp>3) ¬ (d<sharp>­2
= e+)
12.2 ((f<sharp>+1 g1) ® f<sharp>3) ¬ d<sharp>­2
 37

13.1 (g1 g1) f<sharp>3

13.4 g1 g3
 38

Fig. 5.

[1] 1.1 g<sharp> +

[2] 1.3 g<sharp> + (g<sharp> +1 a1)
(g<sharp> + g<sharp> +1) a1
[3] 2.2 (g<sharp>­/a) g<sharp> +1
[4] 2.3 ((g<sharp>+/a) a) ® g<sharp> +1
[5] 3.1 g<sharp>+ ((a a2) (a+ a1+))
3.2 g<sharp>+ (a1 a2)
3.3 Na (a1 a2)
[6] 3.3 (a1 a2) (a a1)
3.3 a2 ((a a1) a)
3.4 (a a1) Ng<sharp>+

Fig. 6.

[1] (1.1) d
d
[2] (2.1) c<sharp>+ ®d
c<sharp> + ¬ d
c<sharp> +
[3] (3.2) c<sharp> + ¬ d
c<sharp> +
[4] (4.2) c<sharp> + ¬ d
c<sharp> +
[5] (4.3) c<sharp> +
c<sharp>+
[6] (5.2) c<sharp> +
c<sharp>+
[7] (5.3) c<sharp>++
c<sharp>+
[8] (6.1) c<sharp>+ ¬ d
d
[9] (6.2) c<sharp>+ ¬ d
c<sharp>+
[10] (6.4) c<sharp>+ ¬ d
c<sharp> + ¬ d
[11] (7.3) c<sharp> ¬ d
c<sharp>+ ¬ d
c<sharp>+
[12] (9.1) c<sharp>+
[13] (10.1) c<sharp>+
c<sharp> + ¬ d
c<sharp> + ¬ d
d+
d+
[14] (11.4) c<sharp>+ ¬ d+