Segmentational Approaches of Atonal Musi PDF
Segmentational Approaches of Atonal Musi PDF
Segmentational Approaches of Atonal Musi PDF
Stefanie Acevedo
A Thesis
December, 2010
Committee:
Gene Trantham
© 2010
Stefanie Acevedo
ABSTRACT
The complexity of atonal musical structures has led theorists to offer varying analyses of
atonal works. This ambiguity stems from the intricacies of human perception: Is it
segmentation, the analyst must provide supporting evidence in the music. Due to the
wide range of perception, this evidence yields analyses that are more or less
persuasive, but neither correct nor incorrect. David S. Lefkowitz and Kristin Taavola,
James Tenney and Larry Polansky’s perception-based theory. Tenney and Polansky’s
theory is rooted in visual Gestalt perception and provides the foundation for Dora A.
identify segmentational boundaries that support published analyses of two atonal works:
the fourth of Anton Webern’s Fünf Sätze, Op. 5 and an excerpt from Arnold
segmentational criteria: the sonic, which refers to acoustical properties, and the
Lefkowitz and Taavola note that Tenney and Polansky’s theory cannot be applied
to polyphony. Although Tenney and Polansky concede this point, Hanninen encourages
the use of her theory for polyphonic segmentation. She does not, however, provide a
method for addressing polyphony. Thus, I combine aspects from Lefkowitz and
iv
I find that sonic and contextual criteria in the music strongly support the analyses
by George Perle, Allen Forte, Gary Wittlich, and Charles Burkhart. Due to the emphasis
criteria; however, sonic criteria also reinforce their segmentations and sometimes may
even support their contextual criteria in places lacking local sonic criteria. Thus, the
analyses and suggesting that atonal music can legitimately be heard in different ways.
v
To my father
It's only when you grow up, and step back from him, or leave him for your own career
and your own home—it's only then that you can measure his greatness and fully
appreciate it.
-Margaret Truman
vi
ACKNOWLEDGMENTS
First and foremost, I would like to thank my family. Their love, encouragement, and
support are, without a doubt, the most important part of my life. I thank my father, to
whom I dedicate this thesis, for his struggle and never-ceasing quest to see me
succeed. ¡Te quiero, papi! Jacqueline, you are not only like a mother but a best friend;
thanks for your love, your support, and your zest. My aunts, uncles, and cousins—you
provide never-ending love, merriment, and inspiration; thank you. And to the elders of
the family: Nona and Abuelito—both of you have gone through this life giving it all
you’ve got, never giving up, and seeing your families grow into a diverse group of
individuals that love and care for each other no matter what. I can only aspire to achieve
I want to thank my friends for their inspiration and endless support. I would
especially like to thank Kristopher, Nick, Aleks, Dennis, Heather, and Zoey for not only
thinking with me, but also enduring endless weeks of stress and harsh treatment!
Ultimately, I would not have gotten where I am without the aid of great teachers
and mentors. Dr. Per Broman has been a wonderful advisor, encouraging me and
devoting endless amounts of time toward helping me with my endeavors. I thank Dr.
Gene Trantham for putting time into this project despite the hardships in his life; he has
great dedication and has been a great influence. Dr. Robert Fallon, thanks for all of your
help throughout my masters career, especially with that abstract magic! Both Dr. Nora
Engebretsen and Dr. William Lake have not only taught me how to be a better scholar,
but have guided me through difficult times and shown me how to persevere through it
vii
all. I love you both very much. Dr. Christopher Alan Williams and Dr. Smith Alexander
Reed—two great friends, teachers, and role models—thanks for all the late-night chat
sessions, coffee breaks, office hours, and unending support and encouragement. You
There are many other people in my life that I could not include here. However,
you know who you are and I thank you from the bottom of my heart for being a part of
my quest.
viii
TABLE OF CONTENTS
Page
Measures 1 – 3 ............................................................................................. 19
Measures 4 – 8 ............................................................................................. 26
Overview ................................................................................................. 33
Discussion ................................................................................................. 54
APPENDIX: SCORES.............................................................................................. 61
x
LIST OF TABLES/FIGURES
Table Page
Figure Page
B. Phenosegment Support for E/F-sharp Dyad in the Viola, mm. 1-4 .......... 40
Allen Forte, in his preface to The Structure of Atonal Music, describes atonal music as
“complicated music [that] has not been well understood.”1 In the forty years since the
first publication of Forte’s groundbreaking book, much work has been done in the
development of atonal theory; no longer can it be said that atonal music is too complex
or completely misunderstood. The theoretical field has developed widely and many
offshoots of atonal theory have led to the development of models and methodologies to
describe not only how this type of music is written and structured, but also how it is
musical structures into groups or segments, which Dora Hanninen defines as “the basis
for subsequent musical organization and interpretation.”2 The different theoretical tools
that have emerged have led to multiple segmentations, or interpretations, of the same
musical work. As Christopher Hasty states in his article “Segmentation and Process in
Different analyses, then, do not detract from the credibility of the music, but instead add
This thesis examines different analyses of two atonal works, Fünf Sätze, Op. 5,
No. 4 by Anton Webern and an excerpt from Arnold Schoenberg’s Klavierstücke, Op.
11, No. 1, and defines the most musically apparent boundaries that may have impacted
1
Allen Forte, The Structure of Atonal Music (New Haven, CT: Yale University Press, 1973), ix.
2
Dora Hanninen, “Orientations, Criteria, Segments: A General Theory of Segmentation for Music
Analysis,” Journal of Music Theory 45, no. 2 (Autumn, 2001): 345.
3
Christopher Hasty, “Segmentation and Process in Post-Tonal Music,” Music Theory Spectrum 3
(Spring, 1981): 59.
2
the analysts’ segmentations. The multiple segmentations serve to not only authenticate
Hasty’s ideas about ambiguity, but also lead to a more complete and musically evident
What Is Segmentation?
Many theories of segmentation have been developed, some involving human perception
studies) is to decipher how the brain perceives music, and subsequently, segments the
information. Two main theories are discussed here: David S. Lefkowitz and Kristin
of musical perception, and James Tenney and Larry Polansky’s temporal Gestalt
perception, which has a basis in visual Gestalt perception and serves as the foundation
Approach,” David S. Lefkowitz and Kristin Taavola state that musical analysis relies on
segmentation:
Typing the search string “music segmentation” into any library database brings up
hundreds of articles, books, and other resources dwelling on what should appear to be
4
David S. Lefkowitz and Kristin Taavola, “Segmentation in Music: Generalizing a Piece-Sensitive
Approach,” Journal of Music Theory 44, no. 1 (Spring, 2000): 171-229.
5
Employed for analysis in this thesis and discussed in a later section.
6
Lefkowitz, “Segmentation in Music,” 171.
3
the simple concept of dissecting a musical passage. And it might, indeed, be a simple
concept were it not for the intricate workings of human perception and its implications
visual, auditory, or other, must be condensed into smaller units. For example, a ten-
digit phone number is a string of numbers too long for one to easily parse visually and
thus, is usually divided into three separate groups, making it easier to discern (Figure
1). This principle of chunking is also true for music, or any type of aural perception.
chunks the stream of sound into separate entities. These entities can either be defined
Hanninen, segments are groupings of musical events that are bounded by musical
borders based on changes in musical criteria;8 James Tenney and Larry Polansky
7
Hanninen’s definition depends on whether the group in question involves a phenosegment or
not. Others’ vary. David S. Lefkowitz and Kristin Taavola say, “There is a roughly ideal number of [three
to five] notes that combine to form a segmentational group;” The number is drawn from human perception
studies. These ideas vary due to the theorists’ respective hierarchical definitions of segments and their
sub-components. Lefkowitz, “Segmentation in Music,” 181.
8
Hanninen, “Orientations, Criteria, Segments,” 426.
9
James Tenney and Larry Polansky, “Temporal Gestalt Perception in Music,” Journal of Music
Theory 24, no. 2 (Autumn 1980): 208.
4
differences between musical events that create differences or “disjunctions.” So, for
example, if pitch were the parameter, a method of measurement may be the intervallic
distance between pitches. Thus, if the musical example consisted of a string of major
seconds and then a perfect fourth, the disjunction would occur when that different
Due to the many similarities between visual and aural phenomena, and the
perception of these, some music theorists have drawn parallels between visual Gestalt
theory and music perception theory.10 James Tenney and Larry Polansky, in their article
algorithm for music segmentation. They define three musical differences that can be
perceived between temporal gestalt units:11 state (the mean value of a unit’s measured
parameter), shape (the contour of a parameter across time), and structure (the relation
of the unit to others across hierarchical levels). At the lowest hierarchical level, or what
Tenney and Polansky call the “element-level,” it is hard to discern differences between
the temporal gestalt units’ shape and structure (due to the lack of hierarchy and
extended time). Thus, they argue that state is the only quality of a unit that pervades
through all hierarchical levels, and is the essential difference that drives segmentation.12
Tenney and Polansky’s algorithm for segmentation, derived from their theories
10
For more information on Gestalt Theory, please visit the website for the Society of Gestalt
Theory and Its Applications. The International Society for Gestalt Theory and its Applications, Gestalt
Theory and its Applications, http://www.gestalttheory.net.
11
A temporal gestalt-unit is defined as “time-spans” perceived by the listener. These can include
anything from a motive to a movement of a piece, depending on hierarchical structures. In either case,
these are largely determined by perception and the smallest temporal gestalt-unit can be equated to
Lefkowitz’s and Taavola’s event. Tenney, “Gestalt Perception,” 217.
12
Tenney, “Gestalt Perception,” 214-216.
5
A new [temporal gestalt-unit] at the next higher level will be initiated in perception
whenever a [temporal gestalt-unit] occurs whose disjunction (with respect to the
previous [temporal gestalt-unit] at the same hierarchical level) is greater than
those immediately preceding and following it.13
Events in the same hierarchical stream of musical information will only be segmented if
the measurement of a musical parameter is greater from the one before and after it.
Starting with the first note, consider each pair of adjacent notes in turn. For the
magnitudes of any three adjacent notes, x, y, and z with respect to the sonic
dimension in question, determine the (non-directed) intervals |y - x| and |z – y|. If
|z – y| < |y - x|, then place a segment boundary between x and y. Note that the
algorithm always considers and compares the size of the following interval before
placing any boundary…14
Thus, looking at the melody in figure 2a, a segmentation, according to the pitch
segmentations may occur when two separate parameters create boundaries at different
points. For example, the pitch parameter, as stated above, could remain the same as in
figure 2a, but a rhythm parameter may imply a different segmentation. Also, Tenney and
13
Ibid., 217.
14
Hanninen, “Orientations, Criteria, Segments,” 426.
15
Lefkowitz, “Segmentation in Music,” 173.
6
Polansky’s theory is designed for the segmentation of monophonic music, an issue that
*Tenney/Polansky: Compares the notes, thus focusing on the top row of differences (Is the interval
between A and B larger than the distance between B and C? If so, there is no disjunction.)
*Lefkowitz/Taavola: Compares the intervals, thus focusing on the bottom row of differences (Is the
difference between intervals the same? If yes, there is no disjunction).
Results will sometimes give the same segmentation, but the NOMENCLATURE is different.
“change in the rate of change.” They argue that this change in definition allows for
and Taavola, who focus on the proximity between parameters. Thus, given a parameter
of pitch, Lefkowitz and Taavola’s algorithm performs operations on the similarity of the
intervals between the pitches, while Tenney and Polansky’s method compares the
them, how are they measured, how do different musical styles lead to their
16
Lefkowitz, “Segmentation in Music,” 177.
7
interpretation? All of these and many other questions naturally influence the “method”
chosen for segmentation and each method’s inherent problems and implications.
As mentioned before, segmentation theory has been developed for many years and
there have been many postulated methods. There is, however, a general lack of
codification of segmentation procedures and theories. Two theories have been touched
upon thus far, leaving out the fundamental theory employed in this thesis: Dora
Hanninen’s theory of segmentation.17 James Tenney & Larry Polansky’s theory is the
on the possibilities of not only multiple analyses, but, in contrast to Lefkowitz and
Taavola, also emphasizes the importance of the music by stating that segmentations
must be audible.19 The theory focuses on the interaction of three main criteria (sonic,
contextual, and structural) and how these affect groupings that enable the emergence of
Musical segments are, in essence, groups of musical material that are isolated
from other groups according to boundaries. These groups, according to Hanninen, are
17
Hanninen, “Orientations, Criteria, Segments,” 345-433.
18
This section includes a very concise overview of Dora Hanninen’s theory of segmentation,
which will be employed for analysis. Please refer to the article for more detail and specific analytical
illustrations. Ibid., 345-433.
19
Hanninen, “Orientations, Criteria, Segments,” 413.
20
There are three criteria involved in Hanninen’s theory, as opposed to the four used by
Lefkowitz and Taavola. Hanninen points this out as another flaw in their theory: the four domains do not
amply represent the “fourteen dimensions listed.” Ibid., 423.
8
single criterion that delineates a musical group (that is, a “one-to-one mapping”) is
called instantiation; two or more criteria that identify the same musical group (two-to-one
The sonic criteria are acoustical properties such as pitch, timbre, rhythm, attack,
note, with magnitude, and exists within one dimension. There are two subtypes of sonic
criteria (S1 and S2); S1 defines criteria that are adjacent in sequential time and S2
defines criteria that may or may not be temporally adjacent.22 Contextual criteria rely on
interactions between musical events, and thus are also perceptually grounded. These
criteria define musical aspects such as pitch contour, set-classes, interval classes, and
structural criterion would describe aspects such as how a V-I motion may relate to the
Serialism.
21
Ibid., 357-358.
22
“Music is represented as a string of temporally adjacent events; the analyst then identifies and
compares intervals in pitch, attack-points, dynamics, or some other sonic dimension formed by pairs of
temporally adjacent tones…The resulting segmentation is a series of disjunct time-slices. I group all such
segmentation criteria based on temporal adjacency under the heading subtype 1.”; “Sonic criteria can
also be predicated on adjacency in linear dimensions other than time, such as pitch, duration and
dynamics. I call these sonic subtype 2.” Ibid., 360-361.
23
Hanninen, “Orientations, Criteria, Segments,” 364.
24
Ibid., 375.
9
and descriptor. The letters S, C, and T are used to denote sonic, contextual, and
structural criteria respectively. Each label, in turn, also contains a subscript with extra
information. In the case of sonic criteria, the subscript contains not only subtype one or
two, but also a descriptor such as adjacency, attack, pitch, rest, dynamics, etc. (i.e. S1-
pitch or S2-dynamics). Similarly, the contextual criteria will denote a subtype (such as set
class) as well as a descriptor of said subtype (i.e. Cpc R<9A10> where PC stands for pitch-
class set, R for retrograde).26 Structural criteria are comparably shown, however, due to
the atonal focus of this thesis, will not be discussed in further detail.27
limited number of which will be employed for analysis (Table 1). Following her theory,
For example, using the pitch sonic criteria (S1-pitch), a musical line will be analyzed for
boundaries based on the size of intervallic distances between the pitches (Figure 3a).
25
Criterion not described by Hanninen.
26
Hanninen, “Orientations, Criteria, Segments,” 359-387.
27
Since there is no definitive orienting theory used to describe the musical works analyzed in this
thesis (i.e. serialism, reductive analysis, etc.), structural criteria will not be used for analysis.
10
Using the formula |z – y| < |y - x| (where each set of three notes is labeled respectively;
criterion (in this case, S1-pitch). A phenosegment (denoted by a caesura), in turn, arises
*Since there are no common boundaries between the genosegments, except the beginning and end, the
phenosegment arising from all three genosegments includes the entire musical example.
It could be argued that Dora Hanninen’s segmentation theory emphasizes the analyst’s
individuality and perception. Each theorist hears different hierarchies and groupings,
suggesting different analyses about a work and, thus, leading to segmentations that
may differ from each other. Hanninen’s theory is a tool that can lead to a better
It is important to note that the purpose of employing the theory is not to define
what segmentations are correct and which are wrong, nor to explicitly define a routine
11
for musical segmentation.28 If used for analysis after the fact, as in this thesis,
Hanninen’s theory can aid the understanding of why certain segmentations were
chosen since individual perception defines how one chunks musical segments. Thus,
the criteria and processes described are only suggestions of how one might go about
segmenting a work.
Lefkowitz and Taavola’s theory, which weighs the strength of musical segments.29 This
Hearing is subjective and perception even more so. As stated by Fred Lerdahl
and Ray Jackendoff in their influential work A Generative Theory of Tonal Music,
“Rarely do two people hear a given piece in precisely the same way or with the same
segmentations based on atonal sets than, say, a college music student specializing in
18th-century music. Yet, as musicians, these two very different individuals can agree on
28
“The theory is not a methodology (although it might be used as one), nor is it prescriptive.
Rather, it is a flexible, neutral, conceptual framework and language—both terminology and notation—that
analysts can use to identify and grasp sub rosa aspects of [segmentation.]” Hanninen, “Orientations,
Criteria, Segments,” 346.
29
“The overall aim of the methodology is a musically-responsive weighting of different
parameters’ effect on perceptual grouping.” This, inherently, leads to the weighting of segments produced
by the selection of stronger parameters. Lefkowitz, “Segmentation in Music,” 172.
30
Fred Lerdahl and Ray Jackendoff, A Generative Theory of Tonal Music (Cambridge, MA: The
MIT Press, 1983), 3.
31
“The ‘experienced listener’ is meant as an idealization…Occasionally we will refer to the
intuitions of a less sophisticated listener, who uses the same principles as the experienced listener in
organizing his hearing of music, but in a more limited way. In dealing with especially complex artistic
issues, we will sometimes elevate the experienced listener to the status of the ‘perfect’ listener—that
privileged being whom the great composers and theorists presumably aspire to address.” Lerdahl,
Generative Theory, 3.
12
Dora Hanninen’s theory, Lefkowitz and Taavola’s work ensures optimal chunks for
perfect ratio of discontinuities and events within different specified domains or musical
However sound, those mathematical models do not guarantee that the human brain will
determined that [the] ideal number [of events within a chunked entity] is between three
to five.”33 The mathematical model takes into account this formula and derives the
strongest possible boundaries within each domain (or a combination of domains), thus
according to statistical testing and mathematical formulas, but serve only as that, a
model for the perfect segmentation of the musical passage. They cannot accurately
portray how one (or many) would hear these musical ideas.
musical theory should describe multiple ways of organization to appease the possibility
of different hearings.34 Taavola and Lefkowitz, in turn, while proposing that different
32
The domains defined are pitch, articulation, timbre, and rhythm. Lefkowitz, “Segmentation in
Music,” 175-182. Hanninen has issues with these as they are limiting “fourteen dimensions” into four
domains. Hanninen, “Orientations, Criteria, Segments,” 423.
33
Lefkowitz, “Segmentation in Music,” 181.
34
“A theory of a musical idiom should be concerned above all with those musical judgments for
which there is substantial interpersonal agreement. But it also should characterize situations in which
there are alternative interpretations, and it should have the scope to permit discussion of the relative
merits of variant readings.” Lerdahl, Generative Theory, 3.
13
methods of segmentation are viable, contradict the idea through the use of an exact
Thus, one could employ this theory when attempting to understand segmentational
processes but not necessarily to concretely segment a piece of music. The theory
certainly should not be used to comprehend the meaning of said segments within a
musical structure.
Aside from the philosophical and perceptual issues with Lefkowitz and Taavola’s
fundamental issue leads to contradictions within the algorithm, which may or may not be
the point where the rate of change changes. Lefkowitz and Taavola employ multiple
Taking into account that the distance between each subsequent note is shorter by one
eighth-note each time (thus having the same rate-of-change), a discontinuity occurs
after the fourth note, at which point the rate of change becomes three eighths. Now,
consider a similar scenario employing the pitch domain: figure 4b features the exact
35
Lefkowitz, “Segmentation in Music,” 220.
36
The depth and strength of the flaw has not been examined within the algorithm or mathematical
computations, only within the definition of the discontinuities.
14
same rate-of-change as figure 4a, yet the parameter is pitch. Following the similar
schematic for segmentation, a segment would arise after the fourth note (between the A
and B-flat). However, looking at the melodic line, it seems logical that there should be
no segmentation between the B-flat and A due to their intervallic proximity, especially
The complex issues described above limit the use of Lefkowitz and Taavola’s
theory. Yet polyphonic music still poses a problem for the use of Dora Hanninen’s
theory. Unlike Lefkowitz and Taavola’s method, Tenney and Polansky admit that their
theory should only be used on monophonic music.37 Hanninen, however, implies that
the method can be used for other textures; she mentions drawbacks to theories limited
to homophonic music38 and labels criteria that allow simultaneities.39 Still, there is no
analysis of polyphonic music in Hanninen’s article and no mention of the flaw with
37
Tenney, “Gestalt Perception,” 212.
38
Hanninen, “Orientations, Criteria, Segments,” 352.
39
See the definition of S1-attack. Ibid., 362.
15
Tenney and Polansky’s method, which is the fundamental algorithm used for
Hanninen’s theory.
A wise word of advice from Lefkowitz and Taavola now applies: “It is important to
understand that we need not be slaves to any theory or system.”40 Their theory provides
a method for dealing with polyphonic music, which can be employed using Hanninen’s
criteria. A work can be heard as a single line or multiple lines at once (and everything in
isolating different domains/criteria into single lines (what Lefkowitz calls the SINGLE-line
Polyphonic Analysis
It is impossible to predict the multiple ways in which one may hear polyphony. However,
polyphonic analysis, like that of Lefkowitz and Taavola’s theory, using sonic and
contextual criteria.42 They perform a simultaneous analysis, which they call a SIMUL
analysis, by isolating a specific parameter (for example, timbre) and analyzing the
interaction of the multiple lines as a whole.43 This concept can be used in combination
with Hanninen’s theory. Since a genosegment is defined as the grouping based on one
40
Lefkowitz, “Segmentation in Music,” 219.
41
Ibid., 208.
42
It must be clear that this is only a slight approximation. The multitudes of ways in which musical
lines can interact to form a polyphonic texture are innumerable and cannot be fully understood with this
method.
43
Lefkowitz, “Segmentation in Music,” 208.
16
specific criterion, it could be stipulated that the grouping of multiple genosegments for
Figure 5a). In turn, these SIMUL genosegments can be combined across multiple
However, it is imperative that some sense of polyphonic analysis be used. Most music
is polyphonic, and thus a monophonic analysis is extremely limited in order to define the
perception of segmentation. The methods employed here for polyphonic study are very
limited; this is only a demonstration of the potential arising from the fusion of two diverse
The following section describes analyses for sections of two works. The analyses
A survey of main sonic criteria precedes the discussions of three analyses, which, in
Overview
As Allen Forte states in his article “The Magical Kaleidoscope: Schoenberg’s First
Atonal Masterwork, Opus 11, Number 1,” the piece has been analyzed from multiple
perspectives and has given theorists many hardships, contradicting Schoenberg’s belief
that the piece was one that was easily understood.44 Various approaches have led to a
multitude of different analyses, relying on tonality and atonality alike.45 However, the
focus will be on atonality in this thesis. Some of the theorists who have endeavored to
analyze the work include Allen Forte, George Perle, and Gary Wittlich, three analyses
Opus 11, No. 1 begins with a passage that introduces the fundamental material
of the piece. Perle describes the first three measures by placing emphasis on the
trichords,46 while Forte focuses on hexachordal divisions.47 Wittlich brings both ideas
into play and describes the trichords as the building materials of the hexachords.48
44
Allen Forte, “The Magical Kaleidoscope: Schoenberg’s First Atonal Masterwork, Opus 11,
Number 1,” Journal of the Arnold Schoenberg Institute 5, no. 2 (November 1981): 127-129.
45
“There is a long history of published analyses of this work, beginning soon after the score was
published, in October, 1910…Many of them attempt to place the music in some kind of tonal framework.”
Ibid., 129.
46
George Perle, Serial Composition (Berkeley, CA: University of California Press, 1991): 10-15.
47
And as such, this thesis will focus on segmentations of hexachordal divisions, and will only
touch on divisions of dyads, trichords, tetrachords, and pentachords as mentioned in Forte’s article. Forte,
“Magical Kaleidoscope,” 127-129.
48
Gary Wittlich, “Interval Set Structure in Schoenberg’s Op. 11 No. 1,” Perspectives of New
Music 13, no. 1 (Autumn-Winter, 1974): 42.
19
Measures 1 – 3
A rest on the downbeat of the fourth measure sectionalizes the beginning of the piece.
This sudden repose brings attention to the importance of the opening statement in the
first three measures. This opening material is clearly divided into two main parts: a
melodic segment in the right-hand and a harmonic segment comprised of two chords on
the second beats of mm. 2-3. Gary Wittlich describes each of the two parts as
autonomous hexachords, each playing a fundamental part in the fabric of the work.49
Similarly, Forte also describes these two hexachords, along with a multitude of others,
Focusing on a monophonic (or SINGLE) analysis of the sonic criteria for mm. 1-
leading structure in which the soprano is an accompanied melody. The main sonic
genosegments appearing in this passage are S1-pitch, S1-duration, and S1-rest (Figure 6).
Dynamics do not exert segmenting power, and due to the short length of the passage,
S2 criteria do not seem to play a large part either (S2-duration criteria will be discussed in a
polyphonic analysis).
The soprano line is fairly conjunct and thus can be described by the connection
measure two, after which the segment continues far beyond the reach of measure three
(due to the sameness of the intervals between measures 4-8). The S1-duration and S1-rest
criteria, however, end at the rest (m. 4) due to their emphasis on musical time. A strong
49
Wittlich, “Set Structure,” 42.
50
Due to the focus on atonality, interval size is evaluated by interval-class 1 increments. The first
pitch is technically its own genosegment due to the size of the following interval, thus creating three total
genosegments. However, the majority of the melody is included in the two genosegments and very limited
in segmentation.
20
phenosegment appears from the melodic line due to the interaction between all three of
measures 1 and 4.
In a similar fashion to the melodic line in the excerpt, the subservient harmony
lines feature genosegments described by the subtypes for rest, duration, and pitch.
the harmony (chordal texture) and the repetition of said harmony (mm. 2 and 3), there is
(Figure 6).
each criteria subtype while combining the individual voice parts. As a whole, it is
bass; Figure 7). This is in part due to the strong attack of the left-hand chords and the
Unfortunately, however, the SIMUL segmentation does not support the right-hand
which he calls cells, in the first three measures of Op. 11, No. 1 (Figure 8).51 These
trichords can be by the contextual criterion CSC 3-3. Gary Wittlich, in his own analysis,
denotes trichords as being “the most common structures of the piece,” specially the 3-3
trichord.52
The only sonic criteria that appears to support Perle’s segmentations are SIMUL
S2-duration and S1-rest for the harmonic trichord appearing in measure 3. The other
trichords (in measures 1 and 2), however, are not readily apparent from the sonic
genosegments of the section. The first trichord segmentation (appearing in the right-
hand melody mm. 1-2) is supported by a contour contextual criterion. A split in the
melodic line is not apparent in the single sonic criterion for said line, yet there appears
delineates a contextual split in the top voice, which leads to the emergence of the
Perle’s trichord in mm. 1-2 (if the repeated B in the alto line is omitted). While other
supporting contextual criteria are limited, a strong case is made for the trichord
51
Perle, Serial Composition, 11.
52
Wittlich, “Set Structure,” 42.
53
Perle, Serial Composition, 10-15; Wittlich, “Set Structure,” 41-42.
23
In contrast to Perle, Allen Forte segments the first three measures into four
separate hexachords: CSC 6-Z10, CSC 6-Z44, CSC 6-21, and CSC 6-16 (Figure 10a).54 Sonically,
the most readily apparent hexachords are CSC 6-16 and CSC 6-Z10 due to the strong
stratification of the left and right hands into melody and accompaniment chords. He
describes these two hexachords as the thematic components of the work. The two
segmentations are supported by the SINGLE-line analyses of sonic criteria (Figure 6),
while CSC 6-16 emerges through the phenosegment established by the SIMUL sonic
criteria S1-pitch and S2-duration (Figure 7). Two contextual criteria help to establish the
CSC 6-Z44 hexachord: motion by Cip <+4> in the bass staff and Ccseg <210> in the melody
(Figure 10b). The chordal nature of the alto, tenor, and bass lines discourage any
attempt to divide the three into separate groups, yet the similar movement that exists
only between the tenor and bass voices supports a decision to segment the voices
54
Forte, “Magical Kaleidoscope,” 139.
24
separately. The phenosegment that would justify the segmentation with CSC 6-21 is only
supported by the S1-pitch criteria. CSC 6-21 contains the strong melodic line of the right
hand and the alto and tenor voices, an odd segmentation on the basis of sonic criteria.
Thus, the SINGLE criteria of S1-rest and S1-duration for the bottom voices lead to a
phenosegment, but exclude the bass voice; the strong correlations between the three
Anachronistically speaking, one could say that Gary Wittlich’s analysis in “Interval
Set Structure in Schoenberg’s Op. 11, No. 1” builds on the analyses by Allen Forte and
George Perle. Wittlich discusses the prominent segmentations from both theorists’
analyses, including the importance of trichords and two main hexachordal figures in
25
mm. 1-3. Wittlich’s segmentations into trichords are the same as Perle’s except for an
extra segmentation of a CSC 3-5, which creates a second trichord segmentation of the
aforementioned Ccseg<210> criterion that supports the first trichord division of the line
stipulated by Perle.
emerge from criteria CSC 4-2, CSC 4-5, and CSC 4-19 (Figure 11b). The melodic S1-pitch criteria
supports one of the melodic line tetrachords (CSC 4-2) while no apparent sonic criteria
segmentations in measure 2 and measure 3; these include the chord structures and the
downbeat pitch of the melody (apparent from CSC 4-5 and CSC 4-19). Because of the attack
offset between melody and harmony, a SIMUL phenosegment does not emerge from
S1-attack. However, due to the simultaneity, harmony is a strong parameter for the
contextual parameters. This quality, on the other hand, hurts the emergence of Wittlich’s
last tetrachord segment (m. 3) since the bass voice is omitted from the segment,
despite the support from the SINGLE S1-pitch criteria. Either way, Wittlich argues for the
importance of these tetrachords and trichords due to their relation to his hexachordal
segments,55 narrowing the “supersets of practically all the smaller sets of the piece” into
two of Forte’s same hexachordal segmentations (delineated by CSC 6-Z10 and CSC 6-16).
As previously mentioned, these two hexachords are the most apparent due to strong
55
See Table 2. Inclusion of the Primary Tetrachords and Hexachords in Wittlich, “Set Structure,”
44.
26
sonic support from criteria and the inherent homophonic texture, as well as the strong
Measures 4 – 8
After a short three-measure introduction, the texture of the movement becomes more
four-part texture as in the prior three measures. As before, the four distinct parts (S-A-T-
B) can be analyzed for individual segments, and combined for a SIMUL approach.
The structure shows that measures four and five are reiterated in measures five
and six yet the attacks of each voice are offset by an eighth-rest. Despite this change,
the segmentations apparent in the first appearance of the musical passage still hold for
the repetition, the only difference being a change in genosegment support. Each
individual line is a repeats its own respective motive. The soprano line is a repetition of
the E and G, which leads to a S1-pitch genosegment that pervades throughout most of the
four measures, except for a disjunction in measure seven due to the following material
(in measure 9). Thus, the more influential genosegments of the soprano line (that is,
those which combine to create a phenosegment) are the S1-rest and S1-duration that break
the line into three separate segments (Figure 12a). The alto line, which begins with an
attack concurrent with the soprano line, is a string of seconds that essentially
embellishes the top line.56 Except for an eighth-note neighbor, the alto line would be
equal in rhythm to the soprano. The neighbor resolution (Bb to B) is highlighted in the
disjunction apparent from the S1-dur and S1-pitch criteria that brings forth a genosegment
56
However, the alto line will diverge from the soprano by an eighth rest, as previously mentioned.
27
beginning in the second note of the motive. Thus, the phenosegment that appears is
*Due to the pitch proximity of the pitches in m. 7, these are considered part of the tenor line despite stem
directions.
Like the soprano/alto pairing, the tenor and bass voices in these measures also
D Major chord in which the fifth is further elaborated through a string of minor seconds
in order to resolve to a B (a case which can be made due to the alto sonority held during
the cadence). The bass merely serves as a pedal tone to the figure. The SINGLE
analysis of the tenor is very straight forward in that a phenosegment arises from S1-pitch,
the tenor motive. This is further amplified by an S1-rest genosegment in the bass (Figure
12b). Since the bass line is only one pitch with the same rhythmic duration, no
segments are supported by S1-duration or S1-pitch. A slight departure from the motivic line
dissolves so that by measure 7, the top voice and bottom voice motives are no longer
juxtaposed, allowing for the addition of the embellishment to the bottom voice.
This figure shows the genosegments for the four individual lines (top=soprano, bottom=bass). Notice the
lack of bracket-endpoint overlap, which prevents the emergence of four-part genosegments.
Figure 13b: S1-rest Phenosegments from SIMUL Analysis of Left Hand, mm. 4-8
*Text in parentheses denotes which voices support the genosegment (S = Soprano, A = Alto, T = Tenor,
B = Bass).
A SIMUL analysis of each individual subcriterion for all four parts does not lead to
prominent genosegments (Figure 13a). This is probably due to the strong pairing of the
29
right and left hand voices and variation of attacks in the reiterations. Thus, the analysis
must also support the structure of the musical passage, leading to a SIMUL analysis of
the two top voices and the two bottom voices individually along with a combination of
the four voices. This analysis shows support for an S1-rest phenosegment in the tenor
and bass voice (Figure 13b). Also, due to the consequent attack points, the S1-attack
arising from pairing of voices (Figure 13c). Despite the lack of S1-attack between left and
right hands, though, there could be a case made for an S1-rest phenosegment arising for
all four voices because of the rests in the lower voices and the ambiguity for the
beginning of the S1-rest criterion (see dotted phenosegment lines in Figure 13a). This
repetition of the material) and measures 7 to 8. There is a division between the top
voices, which contains an iteration of the criterion CSC 5-Z38, and the lower parts, which
introduce the hexachord CSC 6-Z39 (Figure 14). This division into top and bottom voices is
not only supported by the aforementioned sonic criteria, but also by further contextual
criteria as explained by Forte.57 The rhythmic offset in the repetitions does not alter the
support for the individual contextual criteria, mainly due to the support from other
contextual criteria. For example, the prominence of CSC 5-Z38 is not only supported by the
same hexachord heard as the attack on the third beat of measure 4 (and thus S1-attack),
but also by the relationship to another structural hexachord, CSC 6-Z42, which is
introduced in measure 7. Also, Forte describes the bottom voices and their contextual
criteria as inclusive: a pentachord appearing in the tenor voice (CSC 5-Z37), which is
57
Forte, “Magical Kaleidoscope,” 140-141.
30
included in a hexachord with the bass pitch (CSC 6-Z39).58 These contextual criteria are
phenosegment, respectively.
George Perle’s analysis describes the importance of the new passage for the
introduction of a second theme. The tenor voice’s prominent line returns throughout the
work and is derived from Perle’s basic cell CSC 3-3 and CSC 3-12. He states that these
ideas return throughout in the development and even later in the introduction as
(arising from the ambiguity of S1-rest) could show enough support for the importance of
Perle further describes that the predominant factors of the musical structure (of
said second theme) are the contextual criteria Cic 3 and Cic 4 (Figure 15). These interval
58
Forte, “Magical Kaleidoscope,” 140-141.
59
Perle, Serial Composition, 14.
31
classes, which occur as cadential harmonies, are derived from the relations of the
pitches in the basic cell CSC 3-3.60 Likewise, Allen Forte also emphasizes the importance
of the CSC 3-3 as a cadential component of measure 5.61 These contextual correlations
are visually obvious, especially when naming the trichord as a component of CSC 5-Z38.
Sonically, the harmonic implications are supported at measure 4 and 6 through S1-attack
in the right hand. The rhythmic disparity, however, later causes an interaction between
the sonic criterion and the contextual; the criterion S1-attack supports CSC 3-3 at measure 4
and 6, a criterion that later heralds the sustained CSC 3-3 at measure 8 without the need
Gary Wittlich compares two methods of analysis for the four-measure excerpt.
He recognizes the basic cell emphasis as described by Perle and gives his own
interpretations, which give prominence to the criteria CSC 3-3, CSC 3-1, and CSC 3-8. Wittlich
CSC 3-3 segmentation on the third beat of measure four, but also emphasizes a CSC 3-3
prior to that (Figure 16). To support said segmentations, Wittlich argues that Cp [78E] is
60
Perle, Serial Composition, 13.
61
Forte, “Magical Kaleidoscope,”141.
32
emphasized due to the prominence at the onset of the piece, a similar argument to
Perle’s.
and hexachordal inclusions not mentioned in the text of his article, are themselves
contextual criteria that support the segmentations.62 The sets are derived from similar
trichords employed by Perle, as mentioned before, and are also subsets to Forte’s
hexachords. Therefore, some of the similar sonic criteria support Wittlich’s analysis. The
division of top and lower voice aid in the division of segments by trichords that include
only soprano/alto or tenor/bass voices. The S1-rest for the soprano and alto helps to
differentiate the trichord on the second beat of measure 4 from the trichord including the
bass and tenor of the bottom voice. S1-rest for the alto is also support for the trichord
emphasizing the lone alto line. Due to the hierarchical nature of the phenosegments, it
was possible to state that S1-pitch supported a segmentation of the tenor line as a whole
melodic swoop, but the actual genosegments of said criterion provide support for the
analysis given by Wittlich. This same tenor line is described to be part of the hexachord
CSC 6-Z39, which is related to both CSC 6-16 and CSC 6-Z10.63 The segmentation of the line
into two hexachords is the same as Forte’s analysis and thus similarly supported.
62
Wittlich, “Set Structure,” 44-45.
63
Ibid.
33
Overview
fourth movement of Op. 5 has been a favorite for analysis.64 Analyses by three theorists
are discussed in depth here: George Perle Charles Burkhart, and Joseph Straus. All
analyses agree that the piece consists of a three-part form. However, there is a bit of
contention as to where the exact boundaries lie. Burkhart believes that the sections are
demarcated by a seven-note motive (at mm. 6, 10, 13).65 Perle, however, states that the
second section occurs at measures seven through nine.66 Straus is a bit more
ambiguous, claiming that measures seven through nine are clearly a contrasting
section, yet later stating that the seven-note figure at measure ten ends the second
section.67 Taking into account the expression markings of the piece and the seemingly
punctual purpose of the seven-note motive, three sections emerge: section A (mm. 1-6),
section B (mm. 7-10) and section C (mm. 11-13). The following discussion will focus on
The orchestration of the movement allows for a clearer SINGLE analysis. Yet,
there are some areas that feature double stops and other such non-monophonic playing
64
Charles Burkhart, “The Symmetrical Source of Webern’s Opus 5, No. 4,” The Music Forum 5
(1980): 317.
65
“To me this little piece has long been a tantalizing puzzle, particularly so the three transposition
of what I call the ‘7-note figure’—that is, the rising unaccompanied figure…that demarcates the end of
each of the piece’s three sections.” Charles Burkhart, “Symmetrical Source,” 318-319.
66
Perle, Serial Composition, 16.
67
Joseph N. Straus, Introduction to Post-Tonal Theory (Upper Saddle River, NJ: Prentice Hall,
2005), 119.
34
abilities. Such figures will be analyzed to portray the aural qualities of the piece, more
than the pitches or similar attributes which appear on the written paper.
The abilities of the instruments also allow for a variety of timbral interchanges.
Dora Hanninen does not define a quality for timbre or articulation due to their
will be, for the purposes of this thesis, a one-dimensional approach to timbre. As
Section A, Measures 1 – 6
The first two measures of the piece serve as an introduction to the movement, in which
the violins introduce the fundamental pitch-sets of the musical material employed
overlapping of the established sets.71 A slight pairing of voices begins the piece, in
which the violin one and two enter with tremolos and the viola and cello play an
accompanying role. This is quickly dissolved at measure three, during which the voices
contrapuntally share motives. The change of texture supports the division of a two-
measure quasi-introduction.
68
Hanninen, “Orientations, Criteria, Segments,” 362.
69
“One interim solution is to treat timbre and articulation as nominal spaces. Musicians have long
done this in practice, registering simply the presence or absence of change: change indicates disjunction
and marks a segment boundary, whether in S1-timbre, S2-timbre, S1-articulation,
S2-articulation.” Ibid, 363.
70
Burkhart, “Symmetrical Source,” 319; Straus, Post-Tonal Theory, 120.
71
Perle, Serial Composition, 16.
35
tremolos of the violins. The analyses put forth by Perle, Straus, and Burkhart pay close
attention to the pitch material of the figures. However, it is difficult to aurally discern the
exact pitch material of these ornamentations, and thus, it is a prerogative to label these
as non-pitched figures. The pizzicato in the first violin also poses an issue as to what
combination of pitch material is heard, and thus is left out of the S1-pitch analysis (as will
A SINGLE analysis of the violin one part supports a division of the section into
three phenosegments (Figure 17). Two are strongly supported by S1-pitch, S1-duration, and
S1-rest (in mm. 3 and 4), while S1-tone also supports the first (mm. 3). A third
phenosegment arises at the end of measure four, but continues through to the next
section of the piece. However, the material after the section break does not contain
anything aside from rests, so it could be said that the phenosegment indeed ends at the
end of the section. Thus, the phenosegment contributes to the delineation of formal
structure. The violin two parts also supports this delineation, mainly due to the seven-
note motive.
36
supported by S1-pitch, S1-duration, S1-rest, and S1-tone. This new phenosegment, however,
ends immediately before the seven-note figure in measure five as supported by S1-pitch,
S1-duration, and S1-rest.The figure could be called its own phenosegment, as a boundary
this boundary before section B is only supported by S1-rest and S1-tone (Figure 18).
The viola part contains two very strong phenosegments, supported by all
evaluated criteria (Figure 19). Once again, the second phenosegment lingers into
section B, but only containing rest material before the onset of viola ostinato in measure
7. The musical material of the viola is not as varied as that of the violins, as the
37
instrument only plays for pitches in the span of six measures. Thus, the main breaking
segmentation points are characterized by its steep change in register and its change to
The cello part contains perhaps the most ambiguous material of the section. The
strongest phenosegments would only arise from support by three criteria at any one
time (Figure 20a). The double stop at measure four adds to this ambiguity: the addition
of the extra note juxtaposes boundary notes, in essence abolishing said disjunction.
Thus, if the G were placed linearly next to the C-sharp, a genosegment boundary would
38
occur. However, this slight temporal modification liquefies the segmentation, creating a
SIMUL analysis of the four instrumental parts does not show a large amount of
consensus. Generally, the S1-tone criterion is highly inconsequential in the scope of the
SIMUL analysis. A phenosegment does seem to have a bit of support at the end of
measure four, where the violin one and viola parts interact to form a disjunction in the
S1-pitch, S1-duration, and S1-rest criteria at measure four, along with some participation from
tremolo pitch material. His ideas stem from the transposition of an intervallic cell. The
CSC 4-8 and CSC 4-9. Perle pays close attention to the iterations of these sets as they
appear at measure three (Figure 22a)—CCS 4-8[E045] and CCS 4-9[E056]. These sets contain
39
three common tones, excluding pitch classes E and F-sharp, a dyad that appears as the
highest pitches of the opening tremolos.72 This segmentation is not supported by the
aforementioned analysis due to the aural qualities of the tremolo. However, a later
melodic iteration of the dyad in the viola (measures two and three) is highly supported
by genosegments arising from the SINGLE S1-pitch/rest/duration criteria (Figure 22b). The
dyad also appears at the end of the section (mm. 5-6) in the violin one and cello parts,
two iterations that are minimally supported by the respective S1-pitch criterion.
Perle alludes to segmentations of CSC 4-8 and CSC 4-9 throughout measures three
to six, yet only clearly describes five iterations (Figure 22a).73 The SINGLE analysis
supports iterations occurring in violin one (Violin one S1-pitch/duration/rest, m. 3) and violin
two (Violin two S1-pitch/duration/rest, m. 4). The cello figure (mm. 4-5) is not strongly
supported, however, as the phenosegment arising from any SINGLE criteria extends to
measure six. The CSC 4-8 occurring in measure three is also not supported by any
SINGLE criteria, but the S1-attack criterion supports the inclusion of the viola pitch on the
upbeat of beat two. The more interesting support, however, occurs at the downbeat of
measure four where the all of the parts sound simultaneously. The SINGLE criteria for
the second violin, which are not included in the segmentation, are actually the support
as the melodic line genosegments detract from the inclusion. Once again, like in the
cello part in measure two, a quasi-elision has occurred and thus, aids the support for
72
Perle, Serial Composition, 16.
73
Ibid.
40
importance of the CSC 4-9 and CSC 4-8 criteria, but also connects them to larger over-
arching concepts, including a CSC 8-9 criterion and the multiple ways in which the eight-
note set is realized in order to create symmetry. Burkhart describes the opening tremolo
measures as segments including a CSC 4-9, CSC 4-8 and CSC 4-16 (Figure 23a).74 The
genosegments created by S1-attack, S1-tone, S1-duration, and S1-rest show great support for the
74
Burkhart, “Symmetrical Source,” 319-324.
41
harmonic iterations of CSC 4-9 and CSC 4-8 in the violins. The CSC 4-16, occurring in the violin
Just like Perle, Burkhart describes the importance of the dyad occurring in the
top register of violin one, as well the polyphonic interplay of CSC 4-9 and CSC 4-8 in
measures three through six. Burkhart, however, delineates the tetrachords, seven in
total, four of which are identical to Perle’s segmentations (Figure 23b).75 The remaining
segmentations are a little difficult to show support for using sonic criteria. Some of the
segmentation can be supported in individual parts but not through the four parts. For
example, the violin and cello parts in the CSC 4-9 segment of measure 5 are supported by
S1-pitch, yet the viola part has no support for said segmentation. In this context, it is easier
to explain the support through the inherent contextual criteria of the segmentation.
It is logical that Joseph Straus’s analysis would build on previous analyses. While
the overarching theories digress (Perle’s basic cell versus Burkhart’s symmetry), the
three segmentations studied increasingly build on each other and share many ideas.
Straus’s analysis is a bit more pedagogical (due to its inclusion in an atonal theory
Straus, like the other analysts, emphasizes the sets that underlie the first
measures of the movement. However, he also makes mention of a CSC 5-19 appearing in
measure two, a figure which includes the cello’s E-flat pedal (Figure 24a). Straus
defines this set due to its complementary nature against CSC 7-19, the fundamental set of
the seven-note motive.76 The S1-duration for the cello voice is the only sonic supporter of
said segmentation, but the relationship between the aforementioned contextual criteria
is strong.
75
Burkhart, “Symmetrical Source,” 319-324.
76
Later discussed in detail. Straus, Post-Tonal Theory, 122.
43
Straus also makes mention of trichord segmentations, which have generally been
ignored by Perle and Burkhart. CSC 3-4 and CSC 3-5 are both subsets of CSC 4-8 and CSC 4-9
and play an important role in contrapuntal interplay of measures four and five, or the
canon section, as Straus nicknames it (Figure 24b).77 Melodically, Straus identifies CSC
3-4 in the violin one part, which is strongly supported by S1-pitch, and the cello part, which
contains limited support by the same. Harmonically, the segmentations are also
77
Straus, Post-Tonal Theory, 124-125.
44
apparent on every beat of measure five, interchanging between CSC 3-4 and CSC 3-5.78
However, only two of these harmonic segmentations are truly supported by any of the
sonic criteria analyzed: the pickup to measure five, the upbeat of the first beat, and the
Section B, Measures 7 – 11
The texture of the piece once again changes in measure seven as the three lower parts
take subservient roles and the first violin takes on the melody. The section ends with a
seven-note punctuation in measure ten, which also heralds a tempo change on the
*The genosegments are arranged by part (Violin I = top; Cello = bottom). Notice how the section is
marked by rests on both sides, thus creating a segment despite how the genosegments line up.
Due to the shorter length of this section, it is not necessary to divide it into parts
like the first section, and no strong phenosegments appear that would indicate
otherwise. However, it is imperative to point out that a strong S1-rest offsets section B
78
Straus, Post-Tonal Theory, 124.
45
from the rest of the movement (Figure 25). The viola sets up an ostinato that is strongly
supported by the SINGLE analysis criteria S1-rest, S1-tone, and S1-duration. To create
harmonic interest, the cello and second violin add pedals, individual lines which are
supported by the cello’s S1-rest/pitch/tone and the violin’s S1-rest. No strong SIMUL
phenosegments exist, however, most likely due to the offsetting of the parts’ attacks.
The violin one contains the melodic interest, and contains a highly sectionalized line
according to S1-pitch/rest/duration.
George Perle focuses on the ostinato and the relationships between the voices.
He denotes a segmentation that emphasizes the criterion CSC 4-16 emerging from the
combination of the pitches for the three accompaniment voices (Figure 26).79 It could be
said that the delay of the first violin causes the accompaniment to be more audible and
therefore a segmentation would be supported. In this case, the most prominent sonic
criteria would be the S1-rest, despite no support for a true phenosegment. Perle also
points out a prominent dyad (E/F-sharp), which appears in the cello and viola (as a G-
79
Perle, Serial Composition, 18.
46
flat).80 Due to the high harmonic range of the cello, these two pitches sound next to
each other in range. A segmentation is supported by the S1-attack criterion due to the
simultaneity of the cello’s attack and the first G-flat of the ostinato. As the viola ostinato
dies out (in m. 9), the violins and cello come to rest on CSC 3-4, establishing the
foundation for a CSC 4-8 that arises with the last pitch played by the viola.81 There are no
strong sonic criteria to support this segment, and thus the support must rely on the
Figure 27: Straus’s Segmentation of the Viola Ostinato, mm. 7-9 and Support
(Segmentation Source: Straus, Post-Tonal Theory, 121.)
behind symmetrical analysis. Thus, his segmentations are limited to the CSC 4-16 as
proposed by Perle and a discussion of their relation to the original CSC 4-16 appearing in
the first violin in measures one and two.82 Similarly, Straus focuses on the
accompaniment’s emergent CSC 4-16 set.83 However, he also mentions the subset
produced by the viola line alone, a CSC 3-12. This new, completely distinct subset is used
for contrast.84 The viola’s S1-pitch criterion shows many divisions during this ostinato,
each of which supports the subset as it matches the three-note division pattern
80
Perle, Serial Composition, 16.
81
Ibid., 18.
82
Burkhart, “Symmetrical Source,” 325-327.
83
Straus, Post-Tonal Theory, 121.
84
Ibid., 119.
47
The third section of this movement does not contain material that is musically similar to
that of the first section. As Straus mentions, though, the final section still sounds very
similar to the beginning.85 This is due to, as Straus states, a similar pizzicato figure to
that of measure two and the stark contrast of sets employed in section B. Thus, a return
The last three measures of the movement contain genosegments that support a
stark segmentation into two phenosegments, a disjunction that occurs before the
pizzicato at measure twelve (Figure 28). The first violin line does not contain the
pizzicato or any material after the second beat of measure twelve, and thus is not
included in the segmentation. The SINGLE analysis shows a strong support for the
85
Straus, Post-Tonal Theory, 122.
48
phenosegment in the other lines. The viola and cello lines contain no material after the
pizzicato, yet they both show S1-tone/rest support for a phenosegment break before it (the
viola also includes a S1-pitch criterion). The only voice that contains music afterward, the
violin two splits the pizzicato into a phenosegment that is supported through
S1-pitch/tone/rest. Due to the strong, and similar, phenosegments arising from the SINGLE
analysis, the SIMUL analysis shows the same support stemming from S1-rest/tone/pitch.
George Perle does discuss the final section in depth. He mainly focuses on the
seven-note motive and its relation to the material in section A.86 Burkhart, on the other
hand, brings the piece full-circle by discussing the return of CSC 4-9 and the transposition
of this set that contributes to his symmetrical theory.87 As he states, the pizzicato chord
nearing the end (at measure twelve) is one of the most structurally important moments
of the movement, a statement that mirrors Straus’ comments about its formal
86
Perle, Serial Composition, 17-18.
87
Burkhart, “Symmetrical Source,” 327-330.
49
implications. As Straus segments the chord, it harmonically forms a CSC 4-9 criterion.88
Burkhart segments this harmonic criterion, but also combines it with other pitch classes
from the sonorities that appear earlier in the measure (Figure 29).89 By itself, the
pizzicato chord is most strongly supported through S1-attack. However, the segmentation
including the other pitches is a bit more ambiguous and the only strong support appears
to be the contextual criterion. Burkhart also mentions a CSC 4-28 set that is more relevant
The main emphasis on the fourth movement of Webern’s Op. 5 is placed on a seven-
note motive that appears three times during the work (Figure 30). The analysis of the
motive has led to the discussion of its influence on the form of the movement and the
sets employed.
The figure itself is comprised of CSC 7-19 and it is generally permutated through
motive during all three appearances. The first, in measure six, is supported in the
supported by S1-duration/rest/tone. The last iteration, however, is the only one that is fully
88
Straus, Post-Tonal Theory, 122.
89
Burkhart, “Symmetrical Source,” 329.
90
Ibid., 330.
50
relationship stemming from the basic cell described above. Thus the first iteration of the
motive includes the exact initial four-note CSC 4-16 in the first violin part.91 Not only that,
but the two notes of semitone dyad (E/F-sharp) are also adjacent (Figure 31a). No
91
Perle, Serial Composition, 16.
51
sonic criterion supports this segmentation of the motive into a tetrachord and a trichord,
yet the dyad segmentation is supported by S1-pitch. The contextual relations also serve
as support for both the segmentation of the motive and the segmentation occurring in
measure one. The second iteration of the motive contains the CSC 4-16 set in the middle
(the third through sixth notes; Figure 31b).92 Again, there is no strong support for a
segmentation that would divide the motive into a dyad, a tetrachord, and a dyad. Yet,
once again the appearance of the dyad is supported by a genosegment in the S1-pitch
criterion of the viola. No direct CSC 4-16 from the beginning of the movement appears in
the last appearance of the motive and thus a segmentation is mostly dependent on the
Figure 31a: Perle’s Segmentation of the First Figure 31b: Perle’s Segmentation of the Second
Seven-Note Motive, m. 6 Seven-Note Motive, m. 10
(Source: Perle, Serial Composition, 16.) (Source: Perle, Serial Composition, 18).
also dependent on the appearance of incomplete CSC 4-28 criteria that relates to an
inventory of pitch classes for the movement. Thus, he argues that the seven-note
motives are really based on eight-note groups, an explanation he uses to prove his
points about symmetry.93 The contextual criteria, CSC 4-28, however, is not present in the
92
Perle, Serial Composition, 16.
93
Burkhart, “Symmetrical Source,” 322.
52
iteration of the motives as Burkhart segments them, so it does not work to support said
segmentation (Figure 32a). On the other hand, a CSC 3-10 is apparent, and thus can be
Figure 32a: Burkhart’s Segmentation of Figure 32b: Burkhart’s Segmentation of the Second
the Seven-Note Motive, m. 6 Seven-Note Motive, mm. 10-11
(Source: Burkhart, “Symmetrical (Source: Burkhart, “Symmetrical Sources,” 325-327.)
Sources,” 322.)
Like Perle’s analysis, Burkhart’s segmentation of the first iteration of the seven-
note motive focuses on a tetrachord (CSC 4-16) followed by a trichord (CSC 4-28). The
groupings of the last two appearances of the motive, however, vary from Perle’s:
Burkhart follows the pattern of tetrachord and trichord to segment the motives into a CSC
4-16 and CSC 4-28 once again (or CSC 3-10 for the purposes of this analysis).94 The sonic
criteria still, however, do not support these segmentations. There is a stronger case
made for the contextual criteria and its repetitive nature in the three motives, however,
along with the aforementioned segmentation that lingers into measure eleven (Figure
32b). The beginning of each part (except the second violin) re-states pitch-classes from
94
Burkhart, “Symmetrical Source,” 326.
53
the end of the seven-note motive. Despite no sonic-criteria support, the contextual
also makes two very important observations, which may add answers to the questions
motive’s CCS 7-19.96 The segmentation is mainly supported by S1-rest criteria. There is also
a complementary set to CSC 4-9 in the last two measures of the movement (CSC 8-9;
Figure 33).97 This segmentation contains the pizzicato chord and the seven-note motive,
so as mentioned before, the sonic criteria support is very strong (Figure 28).
95
Straus, Post-tonal Theory, 119-124.
96
Ibid., 122-123.
97
Ibid.
54
IV. CONCLUSION
Discussion
This study served two main purposes: to describe the contextual and sonic criteria
apparent in the musical examples, and to describe how the criteria may have supported
the segmentations by several theorists. The analyses were, indeed, limited and
sampled a very small part of the large body of work that has been implemented on
segmentation analysis. No single thesis would be able to discern all of the intricate inner
workings of each excerpt. However, it is interesting to note that, however cursory and
contextually based the analyses were, each theorist’s analysis was supported by a
dependent on contextual criteria, for example, Burkhart’s and Straus’s analyses of the
canonic section in measures four through six of Webern’s Op. 5, No. 4. It is possible
that the highly trained theorist did indeed hear these segmentations. Other explanations
are also probable: these types of segmentations rely on the existence of a similar
contextual criterion for support. In that case, prior contextual criteria were introduced in
the first measures of the piece and were supported by sonic criteria. Thus, in essence,
the aural cues that influenced the initial segment may indirectly influence the perception
of the latter contextual criteria; the contextual criteria that seem to have no sonic
support are subconsciously recalled in the mind through recurring aural cues.
55
Dora Hanninen believes that there should be a minimum of sonic support for
segmentations since “we must be able to hear them as musical units.”98 It is arguable
that the amount of sonic support needed to justify an analysis depends on the purpose
of said analysis. However, whether the theorist decides to employ an aural perspective
contextual criteria in order to define groups of musical material, this thesis shows that
sonic criteria serve as the majority of support for the segmentations. Also, the
contextual criteria employed could also be aural cues for certain trained listeners as
that the theorists are, in fact, processing their segmentations musically, however
subconscious their musical judgment and whatever their analytical aims may be.
expounded by Hanninen,100 and supported by the analyses of the two pieces in this
thesis. In more than one instance, sonic criteria alone did not describe the support for
the contextual criteria. For example, in the analysis for measures four through six of
Schoenberg’s Op. 11, No. 1, the segmentation CSC 3-3 was originally supported by sonic
criteria. However, as the motives were rhythmically rearranged, the sonic criteria failed
to support later segmentations of said contextual criterion. The earlier sonic support,
however, solidified that segment in the mind, and thus when it reappeared in a different
manner, the contextual criterion was the one that supported the segmentation.
98
Hanninen, “Orientations, Criteria, Segments,” 413.
99
“Sonic and contextual criteria identify features of a sound-surface (sonic disjunctions and
contextual associations) that may be grasped by basic perceptual and cognitive faculties.” Ibid., 357.
100
Ibid., 413.
56
sonic criteria. Theories such as Allen Forte’s Structure of Atonal Music rely on
contextual criteria; thus, an exclusive sonic-criteria-based analysis would only hinder the
contextual-criteria based analysis would detract from the musical context. Sure, an
experienced listener may be able to hear sets and retrogrades of those sets, yet the
average listener may not. Just like when analyzing tonal or any other type of music, it is
important to employ both sonic and contextual criteria, as indirectly done by Straus,
Wittlich, Forte, Perle, and Burkhart, in order to better understand the depth of the
musical structure.
a precise and common language to communicate their musical findings. All of them
were able to relay their findings, yet some of the writing described the segmentations as
if they were self-evident; take, for example, a passage from Perle’s discussion of
Webern’s Op. 5, No. 4: “At the same time the figure as a whole is a linearization of
elements comprised within the initial transpositions of x and y, from the conclusion of
bar 3 through the beginning of bar 4.”102 Of course, Perle wrote his analyses before the
to discuss their contextual criteria findings, it seemed difficult to communicate the ways
101
Hanninen, “Orientations, Criteria, Segments,” 413.
102
Perle, Serial Composition, 16.
103
For example, Allen Forte’s The Structure of Atonal Music.
57
Further Research
It is without question that much work still remains to be done in segmentation theory.
This thesis works to describe the criteria that may have influenced segmentations of
Schoenberg’s Op. 11, No. 1 and Webern’s Op. 5, No. 4. Although only hinted, this study
proposes the development of a polyphonic method that would work with Dora
Hanninen’s theory. However, this method is still far from perfect and much still needs to
As discussed in the introduction, there are many doubts about how humans
perceive polyphony. It is difficult even for a computer program to describe the exact
segmentations and combinations of sounds that are extracted from a musical stream.104
Even so, who is to say that the rules of perception for monophonic music are fully
understood? This thesis has combined two theories in order to arrive at some
Lefkowitz and Taavola’s imperfect SINGLE versus SIMUL methodologies. The result is
not precise, but it is an important step in a very intricate and winding path toward the
Dora Hanninen defines subcriteria that give insights to the potential for analysis
polyphonic attributes is required for further work. For example, throughout the analysis,
there was a need for describing simultaneity of notes, despite a lack of concurrent
104
Lefkowitz and Taavola attempt to do this with a SIMUL analysis, yet the matching of voices
and mathematical operations to decipher differences in pitch height and other such criteria do not break
the code of human polyphonic perception.
58
attacks (for example, to define harmonies). This lack of subcriteria is, thus, detrimental
Closing Thoughts
musical human mind. Thus, the methods of analysis should include cognitive roots in
order to truly grasp the musical understanding. As Dora Hanninen states in her writing,
it is imperative for theorists to be aware of the underpinnings that may influence their
analyses in order to achieve a more fruitful musical experience as it extends not only to
listening, but also performance, music instruction, and other musical endeavors.105
105
Hanninen, “Orientations, Criteria, Segments,” 347.
59
BIBLIOGRAPHY
Baars, Bernard J. and Nicole M. Gage, eds. Cognition, Brain, and Consciousness.
London: Academic Press, 2007.
Beach, David W. “Pitch Structure and the Analytic Process in Atonal Music: An
Interpretation of the Theory of Sets.” Music Theory Spectrum 1 (Spring 1979): 7-
22.
Burkhart, Charles. “The Symmetrical Source of Webern’s Opus 5, No. 4.” The Music
Forum 5 (1980): 317-334.
Forte, Allen. “The Magical Kaleidoscope: Schoenberg’s First Atonal Masterwork, Opus
11, Number 1.” Journal of the Arnold Schoenberg Institute 5, no. 2 (November
1981): 127-168.
Forte, Allen. The Structure of Atonal Music. New Haven, CT: Yale University Press,
1973.
The International Society for Gestalt Theory and its Applications. Gestalt Theory and its
Applications. http://www.gestalttheory.net.
Lerdahl, Fred and Ray Jackendoff. A Generative Theory of Tonal Music. Cambridge,
MA: The MIT Press, 1983.
Straus, Joseph N. Introduction to Post-Tonal Theory. Upper Saddle River, NJ: Prentice
Hall, 2005.
Tenney, James and Larry Polansky. “Temporal Gestalt Perception in Music.” Journal of
Music Theory 24, no. 2 (Autumn 1980): 205-241.
Webern, Anton. Fünf Sätze für Streichquartett. Vienna: Universal Edition, 1922.
60
Wittlich, Gary. “Interval Set Structure in Schoenberg’s Op. 11 No. 1.” Perspectives of
New Music 13, no. 1 (Autumn-Winter 1974): 41-55.
61
APPENDIX A: SCORES
NOTE: Only measures 1-8 are studied in this thesis. Measures 9-12 are given for
contextual reference.
62
Source: Anton Webern, Fünf Sätze für Streichquartett (Vienna: Universal Edition, 1922).