Vous êtes sur la page 1sur 52

A Polychordal

Approach to
Serial Harmony
Part I

Barnaby Hollington

CONTENTS

Introduction

Chapter 1

Interval-Class Set Taxonomy

Chapter 2

Tonal Consonance and Dissonance

Chapter 3

Four Problematic Terms: Atonality, Post-Tonality,


Pantonality, Polytonality

12

Chapter 4

The Polychordal Approach

16

Chapter 5

The Art of Thinking Clearly: A Polychordal and Serial


Analysis

21

Chapter 6

Serialism, Polychordally Conceived

44

Bibliography

46

INTRODUCTION

My research aims to test the effectiveness of a consciously polychordal approach in


clarifying so-called atonal/post-tonal1 harmonic trajectories to the listener. My term for the
system I am developing is the Polychordal Approach. I anticipate that my research will
demonstrate that this method allows a composer to minimise the cognitive opacity 2 that,
according to Lerdahl (1988), Hicks (1991), Meyer (1967), Taruskin (1994, 2008) and other
commentators, renders much atonally/post-tonally-conceived harmony particularly serial
harmony - inaccessible to most listeners.
The Polychordal Approach is founded on the premise that, within the bounds of equal
temperament, all pitch-class sets have potential tonal implications as commentators such as
Rti (1958), Parncutt (2009) and Ads (2012) maintain. Most pitch-class sets have multiple
latent tonal centres i.e., they are inherently polychordal. A few pitch-class sets possess only
one possible tonal centre i.e. they are inherently tonal. Any pitch-class set can therefore be
spaced, horizontally and/or vertically, in such a way as to maximise the audibility of its tonal
connections i.e., in all but the simplest cases, spaced as a polychord. In the process, one or
more tonal centres within the pitch-class set will be perceived as the focal point(s) or tonal
anchor(s) of the pitch-class set. This is the key to meaningful cognition of any conceivable
succession of pitch-class sets by the listener; the key to consistently avoiding the cognitive
opacity that will otherwise continue to prevent atonally/post-tonally-conceived music from
ever communicating meaningfully with a wider audience.

1 Post-tonal and atonal are the generally accepted terms for the type of
harmonic territory that my music tends to cover. I hold that neither term
satisfactorily describes my music. Moreover, both terms pose certain more
general problems: for example, certain composers conceive of their music in
atonal or post-tonal terms, but whether listeners always perceive
atonally/post-tonally-conceived harmony in the same terms in practice is another
matter. The question will be explored further in Chapter 3.
2 Lerdahl, F.: Cognitive Constraints on Compositional Systems in Generative
Processes in Music, ed. John A. Sloboda (Oxford: Clarendon Press, 1988), p.231.

The primary purpose of this paper is to demonstrate the workings of the Polychordal
Approach, through harmonic analysis of sections of The Art of Thinking Clearly, for solo
piano. This analysis, in Chapter 5, forms the main body of this paper.
But before any cogent discussion of the Polychordal Approach can ensue, three sets of
terminological and conceptual hurdles must first be overcome:
1. Some well-documented difficulties arising from Fortes pitch-class set system. I
have devised a new, more transparent system which addresses those difficulties.
2. Conceptual problems arising from varying definitions of consonance and dissonance.
3. Contradictions arising from the terms atonal and post-tonal, and limitations
imposed by the accepted definitions of the terms pantonal and polytonal.
Chapters 1-3 deal with each of the three questions outlined above, preparing the ground
for an explanation of the rationale informing the Polychordal Approach in Chapter 4. Chapter
5 will analyse the workings of this approach in practice, dissect some of the harmonic
phenomena created through employing the Polychordal Approach in conjunction with
serialism, and in the process - refute Lerdahls claim (1988) that serialism can only result in
harmonic opacity, from the listeners perspective.3 Chapter 6 concludes by considering some
of the broader implications of this Polychordal Approach to serial harmony.

3 I do not dispute Lerdahls assertion (1988) that the harmony in serial and other
atonally/post-tonally-conceived works such as Boulezs Le Marteau Sans Matre
(1956) is cognitively opaque. I do contend, however, that serialism can be
rendered tonally clear. Schoenberg and Berg both attempted to achieve this, with
mixed results. Webern (1933) also held that his own serial harmony could be
interpreted in tonal terms. After 1945, the generation of composers who took
over serial techniques chose to write as though tonal implications can and should
be eradicated from new music altogether, typically resulting in cognitive
opacity. But if previous generations of serial composers have failed to achieve
harmonic clarity, from the listeners perspective, it does not follow that current or
future generations of serial composers cannot attain the desired coherence.
Indeed, the analysis in Chapter 5 will demonstrate that certain serial techniques
can also be used to increase harmonic intelligibility, in themselves, if skilfully
handled.

CHAPTER 1
INTERVAL-CLASS SET TAXONOMY

For the purposes of my research, Interval-Class Set Taxonomy is intended as a substitute


for the classification system devised by Allen Forte in The Structure of Atonal Music (1973)
ostensibly a pitch-class set taxonomy - which is currently in widespread use. In seeking to
demonstrate the effectiveness of the Polychordal Approach over all conceivable harmonic
territory within equal temperament, I intend to use not only all conceivable interval-class
sets, but exploit each portion of this expressive territory roughly equally, over my
composition portfolio as a whole. Individual pieces will concentrate on different portions of
this territory, to different expressive ends.
Fortes convoluted classification system precludes sufficiently concise and accurate
presentation of the full range of tonal and expressive implications possible within equal
temperament for my purposes. Several commentators have identified flaws in Fortes
classification system and analytical methods, including Taruskin (1987), Perle (1990) and
Parncutt (2009). Those points most pertinent to my research are summarised below.

1A. Pitfalls in Fortes Pitch-Class Set Taxonomy


Parncutt (2009) establishes that Fortes pitch-class sets are not, in fact, sets of pitchclasses, but sets of interval-classes, defined as though they were sets of pitch-classes. Under
Fortes system, the listed elements of a pitch-class set are not pitch-classes, but rather a
group of numbers which would correspond to pitch-classes only if the first element listed
(always 0) happened to be C, and only then if the set were uninvertible. All we can ascertain
from this list of numbers and then only indirectly, through arithmetic - are the intervalclasses between adjacent pitch-classes (by substraction). An additional interval-class vector
is appended to each Forteian pitch-class set, listing all the interval-classes within the set, but
this is not enough to distinguish them from one another: there are instances of separate
Forteian pitch-class sets sharing the same interval-class vectors.

Furthermore, Fortes system treats inversionally-related sets e.g. the minor triad and
the major triad - as identical. Evidently, the minor and major triads do not possess the same
expressive qualities, and the same can be said of other numerous other inversionally-related
sets in both Western and non-Western music4. Nor can the tonal implications of inversionallyrelated sets be considered identical. The Polychordal Approach to post-tonal/pantonal
harmony is concerned with these tonal implications. Therefore, any system that does not
distinguish between major and minor, nor between any other inversionally-related intervalclass or pitch-class sets, would hinder my research.
As a consequence of listing the elements of each set as a numbers indicating neither the
pitch-classes, nor (directly) the interval-classes, nor (if the set is invertible) which way up the
interval-classes are, Forteian pitch-class set analysis is an unnecessarily labyrinthine,
tortuous affair. Too often, the system effectively does little more than muddy the water. Perle
(1990) has persuasively demonstrated how this has adversely affected Fortes own analytical
work.
I have therefore devised a simpler, more elegant and more consistent system, which
avoids each of the pitfalls described above: Interval-Class Set Taxonomy.

1B. My proposed solution Interval-Class Set Taxonomy


Under Interval-Class Set Taxonomy, the listed elements of each set are interval-classes. I
list only the intervals between adjacent pitch-classes, rather than the full interval-class vector.
For example, the pitch-class set [B, C, D, F], is a subset of the interval-class set [1, 2, 3, 6]. In
other words, the interval-class between B and C is 1 (a semitone), the interval-class between
C and D is 2 (2 semitones), etc. Of course, I could have listed the elements as [2,3,6,1],
[3,6,1,2] or [6,1,2,3]. To avoid confusion, the smallest number(s), i.e. interval(s) are always
4 For example, Hindustani Thaats have clearly-defined, distinct expressive
qualities. Two pairs of Thaats happen to be inversionally-related Khamaj
(flirtatiousness, sensuality) and Asavari (renunciation, sacrifice, pathos); and
Bilawal (devotion and repose) and Bhairavi (devotion and repose, but also
compassion and sadness). The interval-classes between the degrees of the
Khamaj Thaat are, in ascending order: 2,2,1,2,2,1,2 (= the Mixolydian mode).
The corresponding interval-classes for Asavari are 2,1,2,2,1,2,2 (= the Aeolian
mode). For Bilawal, 2,2,1,2,2,2,1 (= the Ionian mode); for Bhairavi, 1,2,2,2,1,2,2
(= the Phrygian mode).

listed first therefore [1,2,3,6]. Under Interval-Class Set Taxonomy, the name of the intervalclass set [1,2,3,6] is simply 1236.
Under this new system, inversions of interval-class sets are listed as separate sets. All
that is necessary to discover the inversion of an interval-class set is to reverse its numbers:
1236 becomes 6321. But the 1 is then placed first: so 6321 becomes 1632 (listed as a distinct
set from 1236). In other words, to invert an interval-class set under my system, one keeps the
first number in place, and flips the rest around: 1236, flipped, becomes 1632, 1227 becomes
1722, 1245 becomes 1542, and so on. The major triad, 354, flipped, becomes the minor triad,
345. 1272, flipped, remains 1272, i.e. 1272 is uninvertible. Thus, unlike Fortes system,
Interval-Class Set Taxonomy allows both inversions and (self-evidently) transpositions of a
set to be spotted instantaneously.
Interval-Class Set Taxonomy also allows the analyst to list sets of pitch-classes, e.g. [B,
C, D, F], without risking ambiguity. I shall henceforth use the phrase sets of pitch-classes to
denote this kind of list, so as to avoid confusion with Fortes pitch-class sets, which are not
sets of pitch-classes.

CHAPTER 2
TONAL CONSONANCE AND DISSONANCE

The next terminological and conceptual minefield requiring clearance, before any
meaningful discussion of the Polychordal Approach can ensue, is the question of consonance
and dissonance. There are several conflicting definitions. These fall under two broad
categories tonal (Plomp and Levelt, 1965)5 or sensory (Lerdahl, 1988), versus
contextual (Hill, 1986) or musical (Lerdahl, 1988).6 These describe two very distinct
phenomena, which happen to coincide for most of the history of Western art music. Contrary
to assertions from Rosen (1975) and other authors, both tonal and contextual dissonance can
remain meaningful, and can be coherently and consistently described with reference to
atonally/post-tonally-conceived music, in which the two do not coincide - where tonal
dissonances have been emancipated, i.e. no longer necessarily serve as contextual
dissonances.
Plomp and Levelt (1965) and other commentators employ the terms tonal dissonance
and tonal consonance to describe the sensory difference between intervals, and therefore
also between chords. (Lerdahl and others prefer sensory consonance/dissonance). According
to Plomp and Levelt, a tonally dissonant interval produces audible beats between the
partials of the two tones. These beats occur when the difference between the two frequencies
falls within a critical bandwidth. Plomp and Levelt thus identify the most tonally dissonant
interval as the minor 2nd, followed by the major 2nd and the tritone. Through the centuries,
there have been many divergent rationalisations of tonal consonance and dissonance. The
resultant hierarchies are generally quite similar. I consider Plomp and Levelts hierarchy and
definition the most persuasive, for the purposes of my research.

5 Plomp, R. and Levelt, W.J.M.: Tonal Consonance and Critical Bandwidth, in


Journal of the Acoustical Society of America 38 (1965), pp.548-560.
6 This is a broad generalisation. Hill (1986) identifies three types of definition.
Cazden (1980) identifies fourteen separate definitions, but was unaware of
Plomp and Levelts work at the time of writing.

Another group of theorists and composers, including Babbitt (1965), Rosen (1975),
Cazden (1980), Kamien (2010/1976), and Anderson, have argued that consonance and
dissonance can only properly be defined within the context of a specific piece of music, or, at
best, a given set of harmonic rules as defined in a particular era. Rosen (1975)7 writes:
[A dissonance is] any musical sound that must be resolved, i.e.
followed by a consonance: a consonance is a musical sound that needs
no resolution, can act as the final note, that rounds off a cadence.

Tonal or sensory consonance/dissonance is crucial to the Polychordal Approach.


Contextual or musical consonance/dissonance is less central to my concerns, although not
insignificant.
One may measure the relative levels of tonal dissonance between
interval-class sets simply by counting the instances of the more tonally
dissonant intervals within each set, according to Plomp and Levelts
hierarchy. On that account, borrowing my terminology from Krenek (1940),
I am listing interval-class 1 (the semitone) as a strong dissonance, and
interval-classes 2 (the tone) and 6 (the tritone) as mild dissonances. The
most basic hierarchical strata can be distinguished simply by examining
the number of semitones in the interval vector. Under Interval-Class Set
Taxonomy, the names of the sets themselves give this information: it is
simply a case of counting the number of 1s in the set name. This
produces, for tetrachords, the following hierarchy:

Strong
Dissonance
s
3 semitones

Sets

2 semitones

1128, 1137, 1146, 1155, 1164, 1173, 1182, 1218, 1317, 1416, 1515

1 semitone

1227, 1236, 1245, 1254, 1263, 1272, 1326, 1335, 1344, 1353, 1362,
1425, 1434, 1443, 1452, 1524, 1533, 1542, 1623, 1632, 1722

0 semitones

2226, 2235, 2244, 2253, 2325, 2334, 2343, 2424, 2433, 3333

1119

7 Rosen, C.: Arnold Schoenberg, (New York: Viking Press, 1975), p.33.

We can then add the second and sixth numbers of Fortes intervalclass vectors for each set (i.e. the rest of the dissonance vector), to
determine the number of mild dissonances i.e. tones (2) or tritones (6).
For tetrachords, 11 hierarchical strata ensue:

Strong
Dissonances
3 semitones

Mild
Dissonance
s
2
tones/tritone
s
2
tones/tritone
s
1
tone/tritone
0
tones/tritone
s
2

1
0
4
3
2
1

Sets

1119

1128, 1146, 1155, 1164, 1182, 1515

1137, 1173, 1218, 1416


1317

1227,
1362,
1722
1254,
1344,
2226,
2244
2253,
2343

1236, 1245, 1263, 1272, 1326,


1425, 1524, 1542, 1623, 1632,
1335, 1452, 1533
1353, 1434, 1443
2424
2325, 2334, 2433, 3333

The following chart shows similar hierarchies for all Interval-Class Sets
containing five elements or fewer. Interval-Class Sets (IC Sets) of the same
colour contain the same number of strong dissonances. Within each such
category, higher rows contain more mild dissonances; lower rows contain
fewer of these. Inversionally-related sets will always contain the same
levels of tonal dissonance, and are now shown side by side. This chart will
suffice for the analytical purposes of this research paper (Chapter 5); later
in my research, I will expand this classification to include all 350 IC Sets.

10

Of the 350 possible IC Sets (136 of which are listed above), only 8 are
tonal consonances: silence, 0 (unison), 39 (minor 3rd/major 6th), 48 (major
3rd/minor 6th), 57 (perfect 4th/5th), 345 (the minor triad), 354 (the major

11

triad) and 444 (the augmented triad). Of all the IC Sets serving as
contextual consonances throughout the history of Western art music until
the late 19th century, there are only 7 types (if one includes silence) 7 of
the 8 possible tonally consonant IC Sets listed above. The exception is 444
(the augmented triad). This obvious shared inherent trait none of these
7 IC Sets contains tonally dissonant interval-classes; whereas, of the other
343 possible IC Sets, all but one contain tonally dissonant interval-classes
casts doubt over Charles Rosens claim that a consonance or dissonance
should only be defined contextually:
It is not the human ear or nervous system that decides what is
a dissonance, unless we are to assume a physiological change
between the thirteenth [century, when only 4ths, 5ths and octaves
could serve as contextual consonances] and fifteenth century
[when 3rds and 6ths could also serve as contextual consonances,
but 4ths could no longer do so].8

That all contextual consonances in the history of conventional Western


tonal grammar are also tonal consonances, by Plomp and Levelts criteria
and that, 444 aside, no other tonal consonances are possible - would
suggest that the human ear does, in fact, decide what is a dissonance.
Whilst IC Sets possess measurably variable inherent levels of tonal
dissonance, in practice these differences do not correspond directly to
perceptible levels of tonal dissonance. Rather, IC Sets constitute only the
first

and

most

straightforwardly

analysable

of

several

factors

in

determining perceptible tonal dissonance levels. The other relevant


factors are spacing, register, dynamics and timbre. My research is not
8 Rosen, C.: ibid., p.33. Moreover, Rosens implication that perfect 4 ths never
serve as contextual dissonances between the Renaissance and the 20 th century
is not strictly accurate. Throughout this period of Western musical history,
although a 4th could not serve as a cadential resting-point in itself, 4 ths frequently
do appear at cadential resting-points, in the middle or upper voices of larger
chords. Considering that in conventional Western tonal music, chords are
[considered] dissonant if they include even a single dissonant interval [Hill, C.:
Consonance and Dissonance in The New Harvard Dictionary of Music
(Cambridge, Mass. and London, 1986), p.197], if 4 ths had truly been considered
dissonant between the Renaissance and the 20 th century, 4ths would never have
appeared at cadential resting-points over that period but they do.

12

primarily aimed at finding failsafe methods of measuring tonal dissonance


levels although later in my period of research, I will be measuring the
effect of Polychordal spacing on perceived levels of tonal dissonance, and
considering the effect of register. However, the classification of IC Sets by
relative tonal dissonance levels enables me to define different expressive
territory in broad terms from piece to piece, or from passage to passage 9.
Consequently, I can test the effectiveness of the Polychordal Approach
across the entire available so-called post-tonal harmonic spectrum.

9 For example, The Art of Thinking Clearly, analysed in Chapter 5, uses very
tonally dissonant IC Set material - yet in practice, the polychordal spacing of
those sets results in a much lower level of perceptible tonal dissonance than the
IC content would suggest. I will demonstrate this fully at a later point in my
research.

13

CHAPTER 3
FOUR PROBLEMATIC TERMS: ATONALITY, POST-TONALITY, PANTONALITY,
POLYTONALITY

Several theorists and composers have asserted that all sets of pitch-classes have tonal
implications, including Reti (1958), Webern (1933)10, Auhagen (1994), Ads (2012) and,
most significantly for the purposes of my research, Parncutt (2009). 11 Parncutt argues that it is
impossible for a composer to eliminate tonal implications from their music, unless they also
eliminate pitch altogether, as happens in Sciarrinos da un Divertimento (1970), Reichs
Clapping Music (1972), some of Xenakiss Plades (1978) and most of Varses Ionisation
(1931). The term atonal, therefore, ought only to apply to music which excludes pitch:
Since every interval, sonority and melodic fragment has tonal implications,
even the so-called atonal music of composers such as Ferneyhough,
Ligeti and Nono is full of fleeting tonal references.12
All Tn-types of cardinality 3 [i.e. trichords] have tonal implications
For example, [IC Set 192, e.g. C, D, Eb] may be heard as either the 1st, 2nd
and 3rd degrees of a minor scale or the 6th, 7th and 8th degrees of a major
scale, suggesting that either the [C] or the [Eb] may be heard as a point
of reference. The major-third (4-semitone) interval[-class] embedded within
[IC Set 138, e.g. C, Db, E] suggests that its reference pitch is [C],
regardless of whether the pattern is heard as Neapolitan, Arabic or Flamenco 13
...it is surprising that many pc-set [pitch-class set] theorists tacitly consider
all pc-sets a priori to be equivalent or value-free, as if they had no tonal
implications or as if tonal implications did not exist. Can the tonal
10 One can certainly take the view that even with us [Schoenberg, Berg,
Webern] there is still a tonic present I certainly think so. Webern, A.: The Path
to the New Music, W.Reich, ed., L.Black, transl., (Bryn Mawr, Pennsylvania:
Theodore Presser, in association with Universal Edition, 1963(1933)), p.39.
11 Parncutt, R.: Tonal Implications of Harmonic and Melodic T n-Types in
Mathematics and Computing in Music, T.Klouche and T.Noll, eds. (Berlin:
Springer, 2009), pp.124-139.
12 Ibid., p.124.
13 Ibid., p.126.

14
implications that we learn from music simply disappear (which is
psychologically implausible), or are they arbitrary (which is psychoacoustically and ethnomusicologically implausible)? 14
...in real music heard by real human beings, pc-sets will always have
tonal implications.15

In other words, if a composer conceives of their harmony in so-called atonal or posttonal terms that is, denies or ignores the question of tonal implications in their harmony
altogether, or considers that tonal associations only appear within certain harmonic territory,
and that other territory remains free of such associations - this will not prevent most or
perhaps any listeners from hearing, feeling and attempting to contextualise those frequent
unintended or unacknowledged fleeting tonal references that inevitably become audible. (Of
course, experiences vary according to each listeners prior musical experience, and
psychological disposition.) It would be unreasonable and unworkable for a composer to
demand that listeners suppress this part of their perception and cognition, particularly
considering that identification of tonal implications tends to induce the impression of
understanding the harmony on some level.
If the term atonal, as commonly understood, is unsatisfactory, the same can be said of
post-tonal. The premise behind post-tonal is that only certain elements of the kind of
harmony in question have tonal implications and that these can only ever represent isolated
instances of decontextualized, post-modern references to a bygone system within a wider
musical context in which tonal implications are otherwise supposedly either absent or
irrelevant, as in atonality.
Parncutts article (2009) is restricted to the tonal implications of trichords, and does not
address the question of multiple roots or tonics within a single sonority i.e. polychords.
Once these are considered (the Polychordal Approach), it is possible to identify all the
possible hypothetical tonal implications of any given set of pitch-classes. It might have been
tempting to conclude that all so-called atonal or post-tonal vocabulary is therefore in fact
theoretically polytonal, albeit to greatly variable levels of intelligibility in practice,
depending on how clearly any given vocabulary is presented, in polychordal terms. However,
the term polytonal, as generally understood, implies that each of the multiple tonalities
14 Ibid., p.126.
15 Ibid., p.126-7.

15

present in any one sonority is clearly sustained over a reasonably extended period, as in
Milhaud or Stravinsky e.g. one strand of the texture might proceed in G major over 16 bars,
with another strand proceeding in C# minor. By contrast, in atonally/post-tonally conceived
music, although any given sonority is theoretically analysable in polychordal terms,
successive sonorities will normally outline very different tonalities. There might, at best, be
an element of, say, G major, somewhere to be found within two or three adjacent sonorities,
but the sense of any textural strand remaining in G major will generally vanish almost as
soon as it has appeared (and provided that the spacing of the IC Sets has allowed it to become
relatively unambiguously discernible in the first place, whether by accident or by design). All
we are left with, then, are the fleeting tonal references that Parncutt refers to.
Schoenberg, of course, preferred the term pantonality to signify this type of harmony,
borrowing the term from Rti, and either ignoring or misreading Rtis original definition.
Schoenbergs personal definition of pantonality signified the relation of all tones to one
another.16 Certainly, since the Polychordal Approach aims to maximise the audibility of
multiple latent tonal centres in any conceivable pitch-class context, the term pantonal,
which could in theory etymologically be interpreted as simply suggesting the hypothetical
presence of any tonalities in any possible harmonic context, might ostensibly seem a viable
upgrade on problematic terms such as atonal or post-tonal. In that sense, one might be
tempted to apply the term pantonal to any music composed using any conceivable IC Set
vocabulary particularly, perhaps, my own music, which is explicitly designed to maximise
the audibility of tonal implications of all IC Sets. But this would match neither Rtis original
definition nor Schoenbergs attempted re-definition. There are points in common with Rti,
but Rti also asserts that pantonal music requires tonal dissonances to resolve contextually,
whereas so-called atonal music (which in Rtis terms would include music written using
the Polychordal Approach) does not:
in the specific atonal sphere the [tonal] dissonances appear without
being identified as [contextual] dissonances, as though there were no
tension, no longing to be resolved inherent in them. But in the pantonal
utterances of our time, which at their face value may appear just as full of
[tonal] dissonance we see quite a different tendency [Pantonality] is
based on constant tensions, i.e. [tonal] dissonances which, however,
are constantly resolved yet whenever a tension seems to abate, a new
dissonance is again at work.17
16 Schoenberg, A: Theory of Harmony, transl. Carter, R., (Berkeley and Los
Angeles: University of California Press, 1978 (1911)), pp.432-3.

16

Of course, in Schoenbergs so-called atonal music, for which he preferred the term
pantonal, tonal dissonances do not need to resolve. But Schoenbergs re-definition of
pantonality also implies that although all tones are related to one another, tonal implications
no longer exist:
Even a slight reminiscence of the former tonal harmony would be
disturbing, because it would create false expectations of consequences
and continuations. The use of a tonic is deceiving if it is not based on all
the relationship of tonality.18

Any attempted appropriation of the term pantonal to describe the kind of harmony in
question would therefore require a third separate definition, cherry-picking certain elements
from both Rti and Schoenberg, and discarding other elements. Further confusion would
ensue, regarding the very different harmonic practices of composers such as Debussy and
Wagner, for whose music Rtis original definition was tailored.
In short, the types of harmony created through the Polychordal Approach cannot be
accurately described by any of the terms atonal, post-tonal, pantonal or polytonal.
Indeed, it could be argued that none of the four terms accurately describe the types of
harmony that they purport to describe. (Just as pitch-class sets are not sets of pitch-classes,
and dissonances are not necessarily dissonances, in tonal and contextual terms respectively.)
Much musical terminology is based on assumptions that appear to hold universally, but in
truth only prove useful for certain styles or traditions. From time to time, existing
assumptions are questioned, rules are updated or broken, and new styles evolve but in
certain cases, the contradictions, inconsistencies and fallacies behind the original definitions
of certain terms remain. These can present formidable obstacles to a musician attempting to
break new musical ground, and then accurately explain what they are doing, using language
that other musicians can understand. Perhaps any musician whose practice happens to
genuinely redefine what is musically possible is in a position to suggest that, besides
introducing new terms, certain existing terms perhaps ought to be redefined, as and where
necessary. In that respect, I have every sympathy for, say, Schoenbergs attempt to hijack the
term pantonal for his own ground-breaking purposes, although I cannot agree with his
attempted re-definition.
17 Rti, R.: Tonality, Atonality, Pantonality (London: Rockliff, 1958), pp.72-73.
18 Schoenberg, A: Style and Idea (London: Williams and Norgate, 1951), p.108.

17

18

CHAPTER 4
The Polychordal Approach

The Polychordal Approach is founded on the following hypothesis:


1. Within equal temperament, each of the 350 possible Interval-Class Sets, and each of
the 4,096 (or 212) possible sets of pitch-classes, without exception, is not only
analysable as either polychordal (in most cases) or straightforwardly modal, but also,
crucially, perceptible as such. In other words, any set of pitch-classes contains one or
more latent tonal centres.
2. Therefore, if a composer presents the tonal centre(s) of a given set of pitch-classes
with sufficient clarity, most if not all listeners will sense one or more harmonic
anchors for that sonority.19
3. The same holds true for any conceivable succession of sets of pitch-classes. I.e.
provided that a composer consistently maximises the audibility of polychordal
connections, any conceivable succession of sets of pitch-classes can be rendered
harmonically intelligible, and therefore harmonically meaningful to most listeners.
To illustrate point 1, let us examine a relatively tonally ambiguous or so-called atonal
set of pitch-classes - [E, F, F#, G, C, Db], belonging to IC Set 111513. This set of pitchclasses contains numerous clearly diatonic subsets, including:

Gb 7th [Gb, Db, F]


C major
[C, E, G]
th
F# 7
[F#, C#, E]
th
F9
[F, C, (E), G]

19 There has been considerable debate among musicologists and composers


concerning whether one can, in fact, perceive more than one simultaneous tonal
centre. Van den Toorn (1983), Babbitt (1949), Hindemith (1942), Baker (1983)
and Forte (1955) have argued that only one tonal centre can be perceived at any
one time. Tymoczko (2002), Taruskin (1990), Milhaud (1923), Stravinsky, and
Thompson and Mor (1992) all claim the contrary. Perhaps the most persuasive
cases have been made by Wolpert (2000) and Hamamoto, Botelho and Munger
(2010), whose research suggests that some listeners can perceive multiple tonal
centres, and others cannot. As a rule, listeners familiar with Western Art Music
are more likely to perceive several tonal centres simultaneously.

19

Db major 7th
E minor 9th

[Db, F, C]
[E, G, F#]

It is easy enough to present the set of pitch-classes [E, F, F#, G, C, Db] vertically as a
polychord i.e. as a chord in which two or more of the possible diatonic subsets are clearly
audible. There are many possible solutions. The first five of the six solutions below involve
clear registral separation of the two or three diatonic units. The sixth is perhaps a little more
ambiguous, as the registers are interwoven. Nevertheless, the sixth chord is also clearly
audible as a polychord. The first two chords occur in The Art of Thinking Clearly, to be
analysed in Chapter 5.

Where two or three diatonic subsets are emphasised in this way through spacing, the
tonal centres of each of these subsets are then clearly perceptible as rival, simultaneouslysounding tonal centres. Furthermore, of these tonal centres, one often dominates. In the first
of the six chords above, it is reasonable to expect that most, if not all listeners will perceive C
major more strongly than Gb7 and while some listeners may only perceive C as the tonic,
others may also perceive Gb as a rival, if weaker tonic. 20 In the second chord, due to both the
greater registral proximity of the two subsets, and the C major subset shifting to the first
inversion, C major will still normally be perceived as the stronger tonal centre, but rather less
emphatically than in the first chord. In all cases above, the lowest tonal centre happens to be
the strongest F#7 for the third and fourth chords, F9 for the fifth, C major for the sixth.
I maintain, furthermore, that where one pitch is clearly perceived as tonally stronger, all
of the other pitches in the sonority can also be heard, to some extent, modally in relation to it.
20 See Hamamoto, Botelho and Munger (2010).

20

For example, the F in the second chord above can be heard by an astute listener not only as
the 7th of Gb major or minor, but also simultaneously as the 4 th of C major. The top Db can be
heard not only as the 5th of Gb major or minor, but also as the flattened supertonic of C, and
the Gb as a blue note in C, as well as a rival tonic in its own right.
Indeed, if any one of the constituent elements of any conceivable non-diatonic set of
pitch-classes were arbitrarily assigned as a hypothetical tonic, each of the other pitch-classes
could be defined in relation to that tonic according to either a Western or non-Western mode,
or else a hybrid of two existing modes. In the case of [E, F, F#, G, C, Db], the presence of
three consecutive semitones prescribes hybrids in all cases - but whichever arbitrary tonic one
assigns, there are numerous hypothetical solutions. Here are solutions for each of the 6
possible tonics within the set. Each solution is a hybrid of a Western scale and a Hindustani
Thaat:

21

To give a more extreme example, the set of pitch-classes [C, C#, D, D#, E, F, F#, G]
could, for example, be construed as a blend of part of the C major scale [C, D, E, F, G] and C
Todi [C, Db, Eb, F#, G]. If a composer were to desire to present that particular set of pitchclasses in those specific terms, and were sufficiently skilled, all pitch-classes within the set
could then be rendered audible in terms of their relation to the tonic C.
Since all possible IC Sets - and so theoretically all sonorities within equal temperament possess latent tonal centricity, it should therefore be theoretically possible to present
sequences including any pitch-class vocabulary in the form of readily identifiable harmonic
patterns. Metaphorically, if one clearly perceives a sequence of harmonic locations, one can
then infer the direction of movement between those locations, and so follow a harmonic
journey. Any such harmony then effectively becomes functional. I hold that the term

22

functional can potentially can become applicable to other existing or future types of
harmony, besides conventional Western tonal grammar provided that the new types of
harmony also appear to function, i.e. possess the following:
1. Clearly perceptible harmonic centricity throughout a piece of music
metaphorically, as a series of harmonic locations.
2. The presence of harmonic movement between various locations.
3. Discernible patterns governing this movement.
The harmonic conventions we hear in operation in the Western tonal tradition create the
illusion of purpose, simply because the patterns and the inherent internal logic are so
transparent. Shermer (2011) has coined the term agenticity to describe the tendency to
infuse patterns with meaning, intention and agency. 21 If we perceive patterns clearly, they
appear to function, to have a purpose, but this is, strictly speaking, illusory. Traditional
Western tonal harmony satisfies the three above criteria more thoroughly than any other
harmonic system yet devised, and is therefore generally regarded as the only possible
embodiment of functionalism in harmony. However, it does not follow that other types of
harmonic locations and trajectories cannot be made to seem equally purposeful, and therefore
function in a distinct, yet equally compelling way.
In Chapter 5, I shall analyse both my attempts to clarify harmonic locations via the
Polychordal Approach, and my use of serialism to create patterns. Of course, legions of
critics from Hindemith (1952) to George Benjamin (quoted in Nieminen and Machart, 1997)
have maintained that the kinds of patterns created by serialism will always defy aural
comprehension a straightforward position, which in almost all cases seems to hold true.
However, the Polychordal Approach allows serialism to operate in a completely different, far
more audible way, as will be demonstrated in the following pages. The resultant music is both
centric, and in its own way, functional.

21 Shermer, Michael: The Believing Brain (New York: Times Books, 2011), p.87.

23

CHAPTER 5
THE ART OF THINKING CLEARLY: A POLYCHORDAL AND SERIAL ANALYSIS

Of my recent compositions, The Art of Thinking Clearly (2013) serves the most usefully
as an introduction to the Polychordal Approach. Each of my recent works covers distinct
harmonic ground, and employs one or more devices not found elsewhere in my work. But the
combination of territory and techniques covered in The Art of Thinking Clearly enables a
clearer exposition more of the key concepts than would be possible via discussion of Partita
(2013), Prosthesis (2014), Two Sketches (2014) or Velvet Revolution (2014).

Example 1: The tone-row used in The Art of Thinking Clearly


All of the pitch-class material in The Art of Thinking Clearly derives from a single 12tone row. An analysis of any rows constituent IC Sets will reveal which portions of the
spectrum between consonance and dissonance it employs. The following example shows each
of the dyads, trichords, tetrachords and pentachords formed by adjacent elements of row for
The Art of Thinking Clearly:

24

Of course, one could draw up similar charts for all possible hexachords, heptachords,
etc. But from the smaller IC Sets identified above, we can already identify the general trend,
in terms of levels of tonal consonance or dissonance, with reference to the chart produced in
Chapter 2. The following chart demonstrates that the row for The Art of Thinking Clearly
uses solely the most tonally dissonant portion of the available spectrum. The larger the IC
Sets in question, the greater the concentration of tonal dissonance. This is readily apparent,
visually: in the tables of IC Sets comprising more than 2 elements, the relatively tonally
consonant ends of the spectrum simply do not feature. Of the possible trichordal territory, the
7 least tonally dissonant IC Sets - 37% of the available spectrum never occur. In
tetrachordal terms, this grows to 42%. In pentachordal terms, the figure is 58%.

25

Naturally, the inversion of this row frequently used throughout the piece - would
produce a similar-looking chart, because any two inversionally-related IC Sets will always lie
on the same rung of the tonal consonance-dissonance spectrum. Throughout The Art of
Thinking Clearly, therefore, most IC Sets are tonally dissonant with occasional, notable
exceptions deriving from certain conscious choices concerning the development of this
central material. Of course, as mentioned in Chapter 2, the level of tonal dissonance of an IC
Set is only one of several parameters contributing to the perceptible tonal dissonance of an
actual sonority. At a later point in my research, I anticipate that I will be able to demonstrate
that the Polychordal Approach tends to result in vertical spacings which tend to generally
reduce levels of audible tonal dissonance.
But at this stage, since the pitch-class content of The Art of Thinking Clearly is, for the
most part, very dissonant, it already follows that besides certain obvious general expressive
consequences - this material will have posed me greater challenges, on the whole, than more
consonant material would have done, given that my aim is to test whether the Polychordal
Approach can successfully facilitate perception of any conceivable harmonic vocabulary, and
trajectory, within equal temperament. If the harmony audibly functions in a cognitively

26

transparent manner, consistently throughout The Art of Thinking Clearly, this would suggest
that the Polychordal Approach is also likely to succeed in its aims, when applied to more
consonant material given that such material tends to be more tonally explicit.22
We can now analyse the workings of the Polychordal Approach more closely.

Example 2A: Pentachordal Cycle, bars 1-7, polychordally conceived


The Art of Thinking Clearly opens with a 36-bar harmonic arch. In quasi-Schenkerian
terms, there is a clearly perceptible tonal shift from B (firmly established in bars 1 to 7) to F
(a quasi-cadential resting-point in bars 33-36). This shift is achieved through a combination
of two factors: polychordal vertical spacing (the Polychordal Approach), and a cyclical serial
process, moulding the horizontal harmonic movement.
To begin with, let us examine the polychordal spacing in bars 1-7, itself one of several
harmonic mini-cycles within the wider, 36-bar harmonic arch. In the example below, I have
divided each chord into two or more sub-chords, to illustrate the presence of multiple tonal
centres in each sonority. I have shown what I perceive as the strongest tonal centres within
each chord in bold, but I am more than happy to acknowledge the probability that some
22 There are, however, two significant exceptions. Of the more consonant IC Sets, 444
(the augmented triad) and 336 (the diminished triad) are problematic in this regard:
whilst they are relatively tonally consonant (according to Plomp and Levelts criteria),
they are also inherently tonally unstable (in contextual terms). As noted in Chapter 2, of
the 8 tonally consonant IC Sets, 444 is the only one never to have served as a contextual
consonance in the history of Western tonal grammar. The inherent instability of 444 and
336 would appear to explain why, in the course of composing Partita, in which the
available 19 trichordal IC Sets were spread evenly over two tone-rows which generated
most of the pitch-class material, 444 and 336 became notably under-represented in
practice: subconsciously, I was avoiding this harmonic vocabulary to some extent. The
underlying harmonic aim of my next work - Prosthesis - was, therefore, to explore the
tonal implications of 444 and 336 in order to test the effectiveness of the Polychordal
Approach on this more tonally ambiguous material.
Elsewhere, it is remarkable that
none of Babbitts ostensibly all-trichordal rows actually contain either of these trichords
the other 17 trichords (in Fortes classification, only 10) are always present, but 444
and 336 never feature at all. These so-called all-trichordal rows are employed in many
of the works from Babbitts middle period, including Paraphrases (1979), Dual (1980), Ars
Combinatoria (1981), the Piano Concerto (1985), Lagniappe (1985), The Crowded Air
(1988) and Consortini (1989). Mead (1994) claims that the omission was made because,
unlike other trichordal IC Sets, 444 and 336 cannot be ordered to represent the four
classical transformations [original, inversion, retrograde, retrograde inversion]
unambiguously, and the latter [336] is the single trichord type that cannot generate an
aggregate [of all 12 pitch-classes]. (p.155-6). Perhaps, consciously or otherwise, the
inherent tonal (rather than serial) ambiguity of 444 and 336 may have played a greater
part than Babbitt would ever have admitted, or that Mead would ever admit.

27

listeners will in some instances perceive other sub-chords more strongly than I do, including
sub-chords not shown below. I do not specify 1 st inversions, 2nd inversions, etc, partly because
these will be self-evident, and partly to avoid unnecessary clutter.

The polychordal content of this 8-chord extract may be summarised, chord by chord, as
follows:
1. The tonal centre of the first chord is emphatically B. Take away the top Eb, and we
have a straightforward B minor 9th chord. Respell the top Eb as a D#, remove the D,
and this time we have a B major 9th both sub-chords are anchored by the same
pitch-class, B.
2. I perceive the strongest tonal centre in the second chord as an E major 7 th, with the 4th
(A) included, but there is an additional, rival tonal centre: the lowest 3 pitches form a
Bb major 7th.

28

3. The C major triad included in the third chord establishes C as the prevailing tonal
centre, with the F and F# perceptible as the 4 th and sharpened 4th within a C major
tonality. A faintly perceptible secondary tonal centre is an F# flattened 9th, given that
F# is the lowest note of the chord.
4. In the 4th chord, B major is again the overriding tonal centre, except this time in the
2nd inversion (the first chord was in the root position). The C# is perceptible as a 9 th;
the C natural as a flattened 9th. I am tempted to hear a secondary sonority, with an Eb,
Gb and C over a non-existent Ab possibly because I have internalised the harmonic
cycle enough to anticipate the Ab in the following sonority. Whether any other
listeners would hear this possible secondary focus is another matter. Parncutt (2009)
and other commentators certainly acknowledge the cognitive possibility of a listener
hearing a tonal centre that does not actually sound in the chord.
5. The fifth chord is the most tonally ambiguous, with at least five tonal sub-chords,
and at least four plausible tonal centres. The highest four pitches could be heard in
terms of either a Bb major 7th (+ 4th) or an Eb flattened 7th (+4th). The low A could
also be cognised as a sharpened 4th in Eb, suggesting that Eb is perhaps the strongest
tonal centre but this is far from clear-cut, to my ears. Ab and D are other plausible
tonal centres.
6. The sixth chord might be heard in terms of an overriding F minor 9 th but to some
listeners, perhaps the same pitches might appear to be anchored around Bb. One
could also sense the lowest four pitches as the first four notes of the Todi Thaat,
starting on E.
7. C# major would seem the prevailing tonal centre in the seventh chord, with the B and
C (enharmonically B#) perceptible as the 7th and sharpened 7th respectively.
8. The eighth chord is the same as the first. It is audible as a quasi-cadential resting
point a return to a tonic established in the first chord.
Another notable feature of this chord sequence is that each chord contains one or two
pitch-classes in common with its adjacent chords. These common notes are in most cases
registrally fixed, thus acting as pivots between successive sonorities. This voice-leading
decision was intended to further clarify the harmonic movement. But besides presenting each
successive chord as explicitly as possible, in polychordal terms, and clarifying the voiceleading as much as possible, there is another contributing factor to the harmonic cohesion of
this opening chord sequence one that helps to explain why the return to the first chord in
bars 5-7 feels like a perfect cadence.

29

Example 2B: Pentachordal Cycle, bars 1-7, serially conceived


That factor is the serial mechanism behind the pitch-class content. This claim might
seem contentious and improbable, certainly to advocates of Lerdahls dictum that serialism
causes cognitively opaque harmony (and it frequently does exactly that). Furthermore, most
serial devices are not directly intelligible to listeners. But this does not render all serialism
esthesically meaningless, if the composer is sufficiently skilled.
The serial mechanism in operation here governs all harmonic movement from bars 1 to
36. The following table summarises the content of each successive chord until bar 6, in serial
terms:

Element
Pitch-Class
Chord 1
Chord 2
Chord 3
Chord 4
Chord 5
Chord 6
Chord 7
Chord 8

1
C#
x
x
x
x

2
B
x
x
x
x

3
Eb
x
x
x
x

4
D
x
x
x
x

5
A
x
x
x

6
Bb

7
Ab

8
E

9
G

10
F

11
F#

12
C

x
x

x
x

x
x

x
x

x
x

x
x

Each successive pentachord shifts along the series, but retains alternatively either one or
two pitch-classes from the previous chord. With each new group of pitch-classes comes a
corresponding shift in tonal centres; given sufficiently clear polychordal spacing and good
voice-leading (including, in this case, pivot notes), these tonal shifts will appear audibly
coherent. But besides hearing new harmonic ground with each new chord, one can sense,
from Chord 4 onwards, that the sequence of tonal centres heard in Chords 1-3 is now being
developed approached again, but with some changes - as shown by the blue lines and red
crosses. The second blue line, through Chords 5, 6 and 7, traces through similar harmonic
territory, in a similar sequence to Chords 2, 3 and 4: in pitch-class terms (and therefore tonal
terms), Chord 5 is quite similar to Chord 2, Chord 6 is quite similar to Chord 3, and Chord 7
is quite similar to Chord 4. Consequently, following Chord 7, we expect something like
Chord 8. The satisfaction of hearing Chord 8 is enhanced further, because:

30

1. Chord 8 is identical to the Chord 1; one senses that a cycle has been completed.
2. The tonal centre is unambiguously B.
3. B seems to be recurring at fairly regular intervals: in Chord 1 (root position), then
Chord 4 (2nd inversion), and now returning decisively in Chord 8.
4. The lowest two pitches in Chord 8, B and C#, are carried on from Chord 7,
boosting their tonal weight.
Of course, the success of this type of mechanism depends entirely on whether the tonal
centres of each chord can be rendered sufficiently audible in the first place. The following
example shows the tonal and serial movement, side by side:

31

Example 2C: Full Pentachordal Cycle, bars 1-36


In theory, the serial cycle begun in bars 1-7 would have lasted for 24 chords, consisting
of three sub-cycles of the kind demonstrated above, in which the first chord is identical to the
eighth. The entire sequence, in terms of elements 1-12 of the tone row, would have proceeded
as follows:

Sub-cycle 1
Sub-cycle 2
Sub-cycle 3

Chord 1
1-5
5-9
9-1

Chord 2
4-8
8-12
12-4

Chord 3
8-12
12-4
4-8

Chord 4
11-3
3-7
7-11

Chord 5
3-7
7-11
11-3

Chord 6
6-10
10-2
2-6

Chord 7
10-2
2-6
6-10

Chord 8
1-5
5-9
9-1

In practice, I chose to set up the chord consisting of elements 9-1 of the row as a quasidominant (F being the unambiguous tonal centre in this case), with the first chord elements
1-5 serving as a quasi-tonic (anchored on B). I then appended the quasi-dominant to the end
of sub-cycle 2, as the next chord in the sequence, repeated sub-cycles 1 and 2 before finally
presenting sub-cycle 3, and completing the cycle. In quasi-Schenkerian, quasi-tonal terms,
the movement from B to F is thus delayed, with the chord comprising elements 10-2
functioning as a transitional sonority, comprising two tonal centres B and F. There are brief

32

flourishes, in the form of chord-multiplications; otherwise the harmonic sequence simply


spins round:

33

34

35

36

Example 3A: Chord Multiplication in Appalachian Spring


Several serial techniques are inherently polychordal. That is: by their very nature, these
serial devices can present multiple tonal centres, simultaneously and/or successively, and
these tonal centres are readily discernible to listeners if skilfully handled by the composer.
One such technique is chord multiplication, developed by Boulez and others. Chord
multiplication involves the superimposition of multiple transpositions of an initial chord. If
the initial chord to be multiplied is spaced so as to maximise tonal intelligibility, and then
transposed at intervals conducive to clear discernment of the multiple tonal centres that result,
chord multiplication can often yield vast, highly complex sonorities whilst retaining harmonic
intelligibility.
The spacing of the transpositions of the original cell is crucial, if the resultant
polychordal structures are to remain aurally coherent. The best way of achieving this is to
ensure that the transpositions of the initial chordal cell do not overlap registrally allowing
the ear to pick out the multiple presentations of that cell (and tonal entity) with relative ease.
The following is a particularly transparent example of chord multiplication, albeit in a tonal,
non-serial piece:

37

The transpositions of the initial Ab major cell could scarcely have been rendered more
aurally transparent. If one were to present these pitches as a block chord in the manner shown
above (Harmonic Aggregate), the highest nine pitches would certainly still articulate the
three relevant tonal centres in an aurally coherent and satisfying manner. If the low Ab were
then moved down an octave, the same could be said of the entire chord.
If we consider a hypothetical alternative, in which Copland had allowed the
transpositions of the original Ab major cell to overlap registrally, the results would have
remained tonally intelligible to the listener, so long as the pitches were presented melodically
and not as a chord. The following hypothetical example demonstrates that, with the registral
overlap and a block chord presentation, any sense of tonal coherence would disappear,
despite the harmonic clarity of the original cell:

38

In order to render the block chord shown above tonally intelligible to the listener, the
composer would need to separate out the various transpositions of the Ab major cell
registrally.

Example 3B: Chord Multiplications in The Art of Thinking Clearly, bars 378-408
If a composer chooses to apply chord multiplication to the broadest possible range of IC
Sets, the original chordal cell to be multiplied will normally be rather less tonally
straightforward than the simple major triad multiplied in Example 3A. In most such cases,
besides ensuring a clear polychordal spacing of the original cell, registral separation of the
cells transpositions is even more important, if the results are to be harmonically intelligible.
In The Art of Thinking Clearly, and in other works, I typically have the highest note of a
cell as the lowest note of an adjacent transposition, and vice versa. If the initial cell is
judiciously spaced, in polychordal terms, this is often enough to ensure that the multiple tonal
centres are presented with sufficient clarity to the listener. Over a series of seven chord
multiplications in bars 129-135 of The Art of Thinking Clearly, this device is applied
consistently, with no exceptions or changes to spacing: no further work was necessary in this
case, to ensure the requisite harmonic cohesion.
However, in the sequence of six chord multiplications selected from bars 378-408,
shown below, each instance required a certain amount of tinkering, to ensure harmonically
satisfactory results. For the purposes of this discussion, bars 378-408 are therefore somewhat
more interesting. In the example shown below, besides a breakdown of tonal centres and
intervals for each actual harmonic aggregate, I have shown the hypothetical aggregates that
would have resulted, had I stuck dogmatically to serial principles throughout, and ignored
empirical harmonic reality. Pitches changed from the hypothetical model are shown as
diamond shapes, and pitches hypothetically present but omitted in practice are shown as
crosses.

39

40

In the 6 chord multiplications shown above, the tonal centres are easy to analyse, given
that one original cell-type is then replicated elsewhere and this is aurally evident. E.g. in the
first chord multiplication shown, the lowest Ab 9 + 4 cell (i.e. Ab, Bb, Db, Eb) is replicated
as further 9 + 4 cells higher up. Likewise, the minor 6/5 cells in the second multiplication,
the major 9th cells in the third multiplication, the 9ths in the fourth multiplication, and so on.
I made three types of change to the hypothetical models. Each of the three kinds of
modification originated from the same concern: to maximise the audibility of the multiple
tonal centres.
The reasoning behind the first type of change is straightforward: low registers normally
require wider intervals between pitches, to ensure aural clarity. Therefore, in each case shown
above, one of the lowest pitches is moved down an octave. In the fourth and fifth instances,
this the second-lowest pitch. In all other cases, the lowest pitch is moved down. I do not
always make this type of adjustment: e.g. in the sequence of chord multiplications in bars
129-135 this was never harmonically necessary.

41

The second type of change resulted from a desire to avoid octaves, in this instance.
Again, this is not always necessary: in those instances, I keep the octaves. For example, over
bars 129-135, there are eleven instances of octaves between right and left hand:

In bars 378-408, however, the presence of octaves in the second and fourth chord
multiplications would have interfered with clear perception of the multiple tonal centers. In
general, where the root notes of these tonal centres are doubled at the octave, as is the case in
bars 129-135, the octaves prove aurally satisfying: they reinforce the tonal anchors. In the
second and fourth examples shown on pages 36-7, however, the octaves reinforce pitches
which otherwise would not have been perceived as tonal centres consequently, the
polychordal spacing would have been rendered more opaque and less aurally convincing, had
I allowed the hypothetical models to stand in this particular texture. In serial terms, therefore,
I cheated, and moved certain pitches in the right hand up or down a semitone, as indicated
by the diamond-headed notes. Of course, this type of tinkering further increases the tonal
complexity of the sonorities in question, but such increases serve merely to add harmonic
spice to the chord, rather than upset the balance of tonal centres, as the octaves would have
done.
The third and final amendment to the hypothetical serial models occurs in the fifth chord
multiplication (page 37). In this case, had I kept the highest note of the original cell as the
lowest note of its upper transposition, and vice versa, the highest five pitch-classes of the
resultant chord multiplication would have matched the lowest five pitch-classes
(Hypothetical (A). This is due to the interval-class distance between the lowest and highest
pitch-classes of the original chordal cell, G and Db respectively evidently, two tritones

42

make up an octave. Again, the resultant octaves are not necessarily a problem: in the third
chord multiplication (p.34) the two F major 9ths sound perfectly well either side of a B major
9th. In the fifth chord multiplication, the addition of a minor 3 rd between the various
transpositions of the original cell simply created a richer harmonic effect.
On one hand, these sonorities could only have been conceived through serial thinking: in
all cases, the similarities between the actual sonorities and their hypothetical serial models far
outweigh the small divergences. However, as soon as the results are not as harmonically
convincing as they might be, I opt to break serial exactitude. The overriding concern is with
setting up multiple tonal centres, and presenting these audibly and euphoniously. I
consistently steer clear of those serial techniques that actively hinder harmonic intelligibility
by their very nature. But rather more serial devices can prove helpful in this respect than is
commonly thought. Certainly, basic retrogrades, inversions and transpositions can be
employed to aid tonal coherence. The same is true of certain types of serial cycle (as shown
in Example 2) and chord multiplication (Example 3). Likewise, Messaiens chords of
transposed inversion device is essentially serial, and can enhance harmonic clarity, if
intelligently used. I have used this technique frequently elsewhere, but it only appears once,
fleetingly, in The Art of Thinking Clearly, so discussion of this technique will have to wait
until a later point in my research.

Example 4: Krenek Rotations in bars 255-265, as perceived tonally


The final serial technique used in The Art of Thinking Clearly to be discussed here,
likewise, facilitates tonal coherence by its very nature. The term Krenek rotation is my own:
it was first employed by Krenek in Lamentatio Jeremiae Prophetae (1944). The technique
has been widely discussed, particularly with reference to Stravinsky and Boulez, most
notably by Wuorinen (1979). One starts with a small melodic series of pitch-classes (often
registrally fixed as specific pitches), and rotates the intervals of the series, starting from the
same pitch-class on each occasion. Whilst the mechanism itself is well-known, discussions of
its tonal implications are uncommon. Julian Andersons Harmonic Practices in Oliver
Knussens Music since 1988: Part I (2002) is an exception. Anderson maintains that

43

Knussens use of Krenek rotations in Flourish with Fireworks (1988) establishes one pitchclass A as a tonic:23
Given the importance of the pitch A in the linear rotations [Krenek
rotations] outlined above, this pitch starts to assume the function of a focal
point to the harmony, an easily recognizable modal tonic which guides the
ear through the many simultaneous complexities of the musics textures .

This explanation arguably holds true in Flourish with Fireworks, but does not apply to
my use of Krenek rotations in The Art of Thinking Clearly and elsewhere, nor would it apply
to other instances of this device in the work of certain other composers, including Stravinsky
and Boulez. The mechanism depends on a pivot note, around which the series of intervallic
rotations occur. Naturally, since this pitch repeats, it tends to assume a more important
harmonic role than most, or all other pitches. But this is not normally enough to establish it as
a tonic. The following example shows three Krenek rotations of portions of the retrograde
row in bars 255-265 of The Art of Thinking Clearly:

23 Anderson, J: Harmonic Practices in Oliver Knussens Music since 1988: Part I


in Tempo, New Series, no221 (July 2002), p.4. Anderson also holds that Knussens
use of Messiaens Chords of Transposed Inversion technique, likewise,
contributes to establishing A as a recognizable modal tonic. Whilst I agree with
Anderson with reference to the use of both of these techniques in Flourish with
Fireworks, neither mechanism necessarily always establishes one pivotal pitchclass as a tonic. Far from it! Rather, both devices more commonly establish
multiple, rival tonics that is, they are both inherently polychordal devices. I
shall demonstrate this with reference to Chords of Transposed Inversion at a later
date. In this example from The Art of Thinking Clearly, my focus is solely on
Krenek rotations.

44

45

Focusing, for the time being, on the three Krenek rotations out of context (Serial
Mechanisms (2)), it is easy enough to see how one might succeed in establishing the pivot
notes (F#, A and F respectively) as tonal centres, if one so desired. Simply by playing through
each of the matrices, spaced as they are under Serial Mechanisms (2), and giving a heavy
sforzando on each iteration of the pivot note, the pivot note becomes readily discernible as
the tonic in all three cases. However, in practice, none of the three pivots in this example
function as tonics (except briefly, at the start of Rotation C of the third matrix). Instead, the
pivots take on a variety of other clearly identifiable harmonic functions, in three very
different ways for each of the three matrices. For this, I am thankful! Were the harmonic
implications of Krenek rotations so predictably crude and static in practice, the ear would
quickly tire of the device.

46

In the first Krenek rotation box in the example above, C functions as a tonic throughout,
rather than the pivot note (F#). This happens, despite F# repeating four times, and the C
occurring only once. The repeated Gs and Es/Ebs function audibly as the dominant and
major/minor mediant respectively; all other pitches, including the F# pivot, are perceived in
terms of their relation to C. This is partly due to the structural placing of C within a larger
context, partly due to the sforzando, and partly because it lasts longer than the other notes.
In the second Krenek rotation box, the perceptible tonal centres shift quickly, with the
pivot note, A, assuming a new function each time. Initially, it is heard as the 7 th of B
major/minor. In Rotation B, it is heard as the mediant of an F# minor 7 th. In Rotations C and
D, it is heard as the mediant of an F major 7 th. Thus, the major/minor 7th sonority of the
original melodic cell proves more important than the repeated pivot note in establishing tonal
centres. This is a by-product of an intentional spacing decision: were the outline of the initial
7th not so strongly felt, perhaps the repeated As might have begun to sound like tonics an
option I chose not to take in this instance.
In the third Krenek rotation box, the upper voices in the right hand radically alter the
harmonic perspective of the left hand line. I have only shown some of the more powerful
tonal centres, each of which coincides with an occurrence of the pivot-note, F. At the start of
Rotation B, the F is heard as the dominant of a Bb major 9 th (with a flattened supertonic). At
the start of Rotation C, the F functions briefly as a tonic, as mentioned above. At the end of
the example, immediately after Rotation D, the F is heard enharmonically as an E#, operating
as the mediant of C# major.
This example illustrates that where Krenek rotations succeed in facilitating harmonic
clarity, they do so mainly because they audibly transform a distinctive initial harmonic cell
e.g., in the first Krenek rotation box, a 4-note chromatic cluster, and in the second Krenek
rotation box, a major/minor 7th. Of course, this can only work if the initial harmonic cell is
presented sufficiently clearly. The fact that a pivot note also happens to repeat is normally of
secondary importance. To suggest that Krenek rotations establish harmonic cohesion mainly
through the banal repetition of a single pitch24 would be to sell the technique short.

24 Anderson does not quite suggest this, but one might easily misinterpret the
statement.

47

CHAPTER 6
SERIALISM, POLYCHORDALLY CONCEIVED

Serialism was originally conceived as a means of creating order from chaos. In so-called
atonal/post-tonal music, in theory any sequence of pitches is permissible. The original
concept behind serialism was to create patterns, particularly patterns of pitch-classes which
hypothetically ought to give the listener a cognitive foothold of some kind, in a context
where, otherwise, anything goes. In Schoenbergs words:
Composition with twelve tones has no other aim than comprehensibility. 25

But in practice, as Lerdahl (1988) and numerous other commentators have argued, serial
patterns are in most cases insufficiently discernible to fulfil this intended function. According
to Lerdahl et al., serialism can never achieve the comprehensibility that it was originally
devised to achieve, particularly in harmonic terms; serialism will always remain cognitively
opaque, as will most, if not all atonally/post-tonally-conceived harmony in general. Meyer
(1967) argues much the same thing, with one crucial exception - Meyer acknowledges the
possibility of a breakthrough:

the human mind searching for control, through prediction, will discover
implicative relationships wherever and whenever a modicum of redundancy
exists if I am correct in contending that it is of the very essence of human
behaviour to predict, then it will be no easy matter to invent a series of sounds,
no matter how disjunct and disjointed, which cannot be made to serve as the
basis for some sort of functionally interpreted inferences. 26
Although the rules and constraints of the twelve-tone method may help to
25 Schoenberg, A: Style and Idea (London: Williams and Norgate, 1951), p.103.
26 Meyer, L.B.: Music, the Arts and Ideas: Patterns and Predictions in TwentiethCentury Culture (Chicago and London: University of Chicago press, 1967),
pp.298-299. [And see my brief discussion of Shermers agenticity in creating
the illusion of functionality, pp.18-19: if we discern any mobile harmonic pattern
sufficiently clearly, our brains are wired to interpret the pattern as imbued with
function and purpose.]

48
keep the composer from inadvertently writing passages that sound tonal [I
partly disagree] they do not preclude the possibility of functionalism [my
italics].27
the fact is that the compositional constraints imposed by the rules of
serialism are much less comprehensive than those imposed by the traditional
constraints involved in the composition of tonal music.28
The [serial] system is patently incomplete. That is, it provides no rules for
, , for choosing pitch register [my italics], , , and so on. 29

I hold that the Polychordal Approach allows serially-conceived harmony to be heard


functionally. The Polychordal Approach imposes an additional set of rules onto existing
serial rules. This new grammar consistently aims at one thing: to clarify whatever harmonic
vocabulary might be thrown up by the serial mechanisms. In doing so, the Polychordal
Approach allows far greater cognitive transparency than was hitherto possible serially. It
arguably completes the rules of serialism or, more accurately, possibly represents one of
several potential means of completion, some of which have yet to be devised. In so doing, the
Polychordal Approach is intended to enable serialism to finally achieve Schoenbergs original
aim: comprehensibility.
Evidently, one could also apply the same Polychordal principles to any so-called atonal
harmony, without using serialism. But in my own experience, serialism and Polychordality
complement one another beautifully. If Polychordality seems to complete the rules of
serialism, serialism also seems to complete the rules of Polychordality. I am loath to break the
symbiotic relationship. Although I have occasionally strayed from serialism in a few passages
from works such as Partita (2013), Dectet (2008) and Bagatelle (2007), I generally prefer to
keep using both, side by side.

27 Ibid., p.242.
28 Ibid., p.240.
29 Ibid., p.303.

49

SELECTED BIBLIOGRAPHY

Ads, Thomas, and Service, Tom: Thomas Ads: Full of Noises (London: Faber and Faber,
2012).
Anderson, Julian: Harmonic Practices in Oliver Knussens Music Since 1988: Part I in
Tempo, New Series, no221 (July 2002), pp.2-13.
Anderson, Julian: Harmonic Practices in Oliver Knussens Music Since 1988: Part II in
Tempo, New Series, no223 (Jan. 2003), pp.16-41.
Auhagen, Wolfgang: Experimentelle Untersuchungen zur Auditiven Tonalittsbestimmung
in Melodien, in Klner Beitrge zur Musikforschung, vol.180 (Kassel: Bosse, 1994).
Babbitt, Milton: The String Quartets of Bartk, in The Musical Quarterly, vol.35, no3 (July
1949), p.380.
Babbitt, Milton: The Structure and Function of Musical Theory in College Music
Symposium, vol.5 (1965), pp.10-21, reprinted in Perspectives on Contemporary Music
Theory, B.Boretz and E.T.Cone, eds. (New York: Norton, 1972).
Baker, James M.: Schenkerian Analysis and Post-Tonal Music in Aspects of Schenkerian
Theory, ed. David Beach (New Haven and London: Yale University Press, 1983), pp.153186.
Cazden, Norman: The Definiton of Consonance and Dissonance, in International Review of
the Aesthetics and Sociology of Music, vol.2 (1980), pp.123-168.
Copland, Aaron: Appalachian Spring - Suite (score), (New York: Boosey & Hawkes, 1944).
Croft, John: Musical Memory, Complexity, and Lerdahls Cognitive Constraints, unpublished
Masters dissertation, University of Sheffield, 1999.
Edwards Allen: Flawed Words and Stubborn Sounds: A Conversation with Elliott Carter
(New York: W.W.Norton & Co., 1971).
Forte, Allen: Contemporary Tone Structures (New York: Teachers College, Columbia
University, 1995.
Forte, Allen: The Structure of Atonal Music (New Haven and London: Yale University Press,
1973).
Hamamoto, Mayumi, Botelho, Mauro, and Munger, Margaret P.: Non-Musicians and
Musicians Perception of Bitonality in Psychology of Music, vol.38, no4 (October 2010),
pp.423-445.
Hicks, Michael: Serialism and Comprehensibility: A Guide to the Teacher, in Journal of
Aesthetic Appreciation, vol.25, no4 (Winter 1991), pp.77-85.

50

Hill, Christopher C.: Consonance and Dissonance in The New Harvard Dictionary of Music
(Cambridge, Mass. And London: Harvard University Press, 1986), pp.197-199.
Hindemith, Paul: A Composers World, Horizons and Limitations (Cambridge, Mass.:
Harvard University Press, 1952), reprinted in Contemporary Composers on Contemporary
Music, Expanded Edition, ed. Elliott Schwartz and Barney Childs (New York: Da Capo Press,
1998), pp.80-98.
Hindemith, Paul: The Craft of Musical Composition, Book 1 Theoretical Part, transl. Arthur
Mendel (London: Schott & Co, 1942).
Hogan, Clare: Threni: Stravinskys Debt to Krenek in Tempo, vol.141 (June 1982), pp.2229.
Kamien, Roger: Music: An Appreciation, 10th edition, (New York: McGraw-Hill, 2010
(1976)).
Kostka, Stefan: Materials and Techniques of Twentieth-Century Music (Upper Saddle River,
New Jersey: Pearson Prentice Hall, 2006).
Krenek, Ernst: Studies in Counterpoint (New York: Schirmer, 1940).
Lerdahl, Fred: Cognitive Constraints on Compositional Systems, in Generative Processes
in Music, ed. John A. Sloboda (Oxford: Clarendon Press, 1988), pp.231-259.
Lerdahl, Fred, and Jackendoff, Ray: A Generative Theory of Tonal Music (Cambridge, Mass:
MIT Press, 1983).
Mead, Andrew: An Introduction to the Music of Milton Babbitt (Princeton: Princeton
University Press, 1994).
Meyer, Leonard B.: Music, the Arts, and Ideas: Patterns and Predictions in TwentiethCentury Culture (Chicago and London: University of Chicago Press, 1967).
Milhaud, Darius: Polytonalit et Atonalit, in Revue Musicale, vol.4 (1923), pp.29-44.
Nieminen, Risto and Machart, Renaud, eds.: George Benjamin (London: Faber & Faber,
1997).
Parncutt, Richard: Harmony: a Psychoacoustical Approach (Berlin: Springer-Verlag, 1989).
Parncutt, Richard: Tonal Implications of Harmonic and Melodic Tn-Types in Mathematics
and Computing in Music, T.Klouche and T.Noll, eds. (Berlin: Springer, 2009), pp.124-139.
Perle, George: Pitch-Class Set Analysis: an Evaluation in The Journal of Musicology, vol.8,
no2 (Spring, 1990), pp.151-172.
Perle, George: Serialism and Atonality: An Introduction to the Music of Schoenberg, Berg
and Webern, rev. edn. (Berkeley and Los Angeles: University of California Press, 1977).
Plomp, Reiner, and Levelt, Willen J.M.: Tonal Consonance and Critical Bandwidth, in
Journal of the Acoustical Society of America, vol.38 (1965), pp.548-560.
Rti, Rudolph: Tonality, Atonality, Pantonality: A Study of Some Trends in Twentieth-Century
Music (Westport: Greenwood Press, 1958).

51

Rosen, Charles: Arnold Schoenberg (New York: Viking Press, 1975).


Schiff, David: The Music of Elliott Carter (London: Faber & Faber, 1998).
Schoenberg, Arnold: Style and Idea (London: Williams and Norgate, 1951).
Schoenberg, Arnold: Theory of Harmony, transl. Carter, Roy, (Berkeley and Los Angeles:
University of California Press, 1978 (1911)).
Shermer, Michael: The Believing Brain (New York: Times Books, 2011).
Taruskin, Richard: Chez Petrouchka: Harmony and Tonality chez Stravinsky in Music at the
Turn of the Century (Berkeley and Los Angeles: University of California Press, 1990), pp.7192.
Taruskin, Richard: Does Nature Call the Tune?, first publ. New York Times, 18 September
1994, reprinted with a new postscript (2008) in The Danger of Music and Other Anti-Utopian
Essays (Berkeley and Los Angeles: University of California Press, 2009), pp.46-50.
Taruskin, Richard: Reply to Van den Toorn, in In Theory Only, vol.10, no3 (October 1987).
Thompson, William F. and Mor, Shulamit: A Perceptual Investigation of Polytonality, in
Psychological Research, vol.54, no2 (1992), pp.60-71.
Tymoczko, Dmitri: Stravinsky and the Octatonic: A Reconsideration, in Music Theory
Spectrum, vol.24, no1 (2002), pp.68-102.
Van den Toorn, Pieter: The Music of Igor Stravinsky (New Haven and London: Yale
University Press, 1983).
Webern, Anton: The Path to the New Music, W.Reich ed., L.Black transl. (Bryn Mawr,
Pennsylvania: Theodore Presser, in association with Universal Edition, 1963 (1933)).
Whittall, Arnold: Serialism (Cambridge: Cambridge University Press, 2008).
Wolpert, Rita S.: Attention to Key in a Nondirected Music Listening Task in Music
Perception, vol.18 (2000), pp.225-250.
Wuorinen, Charles: Simple Composition (New York: Longman, 1979).

Vous aimerez peut-être aussi