Yale University Department of Music

An Instance of Parapraxis in the Gavotte of Schoenberg's Opus 25

Author(s): Henry Klumpenhouwer
Source: Journal of Music Theory, Vol. 38, No. 2 (Autumn, 1994), pp. 217-248
Published by: Duke University Press on behalf of the Yale University Department of Music
Stable URL: http://www.jstor.org/stable/843773
Accessed: 09/03/2010 21:59

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=duke.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

Duke University Press and Yale University Department of Music are collaborating with JSTOR to digitize,
preserve and extend access to Journal of Music Theory.

http://www.jstor.org
 AN INSTANCE OF PARAPRAXIS IN

THE GAVO'I'E OF SCHOENBERG'S

OPUS 25

Henry Klumpenhouwer

In the second of his IntroductoryLectures to Psychoanalysis, Sig-

mund Freud introduces a class of psychological phenomena which he
calls "Fehlleistungen," customarily translated as "parapraxes" but
more generally known as "slips."' In Freud's discussion, parapraxes
are events in which "a person who intends to say something may use
another word instead (a slip of the tongue [Versprechen]),or he may
intend to do the same thing in writing, and may or may not notice what
he has done. Or a person may read something, whether in print or
manuscript,different from what is actually before his eyes (a misread-
ing [Verlesen]),or he may hear wrongly something that has been said
to him (a mishearing[Verhoren])"(Freud 1966, 25). Freud asserts that
such mistakes, although they usually attractlittle attention or are sum-
marilydismissed, do indeed have a sense of their own, a sense that can
reflect upon the wishes or purposes of the person who makes the
error.
Edward Cone has catalogued a number of "mistakes" in Schoen-
berg's music, and suggested corrections (Cone, 1972). One such error
217
occurs in measure 5 of the Gavotte of Schoenberg'sSuite, opus 25: as
Cone shows, according to the row structureestablished in the music
preceding it (including the entire preceding movement, the Praelu-
dium, which is based on the same row) and confirmedby the 23 mea-
sures of music that follow, Gb4 there should be Gt4.
Cone's case is particularlywell-argued and his proposed correction
is convincing. Yet if we take seriously Freud's assertions about para-
praxes, it will be worthwhile examining the slip more closely: such an
investigation has the potential to suggest, within a limited scope, a
psychological context for Schoenberg's writing of the Gavotte.
Example 1 gives the Gavotte's first six measures. We begin by con-
firmingGb in measure 5 as an errorwith respect to the serial structure
of the work. It will be necessary to establish some local context of the
flawed row statement, the third presentation of the row in the Ga-
votte. As we shall see the firstfour row statements are best considered
together as a unit.
Example 2 depicts the opening two row statements in measures 1
through 3. The row in measures 1 and 2, labelled J for reference, is
partitioned into three tetrachords, presented melodically in alternate
hands. The diagram displays the pitch-classes of J in two horizontal
lines to show their assignment to the relevant hand. The first three
pitch-classes in the second tetrachord, the "BACH tetrachord," are
repeated in reverse order as an echo accompanyingthe end of the last
tetrachord.2
The partitioningof the row into tetrachordsis carriedthrough into
the second statement (measure 3), labelled K for reference. As in the
first row statement, the tetrachords are presented melodically, in al-
ternating hands. The treatment of the row elements further parallels
the first statement by echoing (in reverse order) the first three pitches
of the second tetrachord as accompanimentto the end of the last tet-
rachord. The pitch-classcontent of row K is derived from row J under
IE, an operation that leaves the dyad G/Dl invariantin the first tet-
rachord, and the dyad Gb/Ab invariant in the last.3
Before examining the row statement with the alleged "error," we
shall examine the row statement that immediately follows it. Example
3 depicts the fourth row statement (measures 5 and 6), labelled M for
reference. Unlike rows J and K earlier, row M presents the three tet-
rachords more or less simultaneously: the first and third tetrachords
(Et, Eb, Db, Gb and D, F, C, Ft) in the right hand; the second tet-
rachord (At, Ab, Cb, Bb), in the left. The relative registral displace-
ment between the hands is not strictly observed: the second
tetrachord, in the left hand, is the highest of the three tetrachords at
A~, the second highest at Ab, and the lowest at Cb and Bb.

218
Etwas langsam( J = ca72) nicht hastig

Example 1. Etwas langsam (J = ca 72) nicht hastig

J K
m. 1-2 m.3

r: Etl Ftl Gl Db Gb E6 Ab D i l B Ah Gtl Db Ab C6 Gb C

I: BslCl Al Bb (Al Cl Bt) Eb Db Ft El (Fl Da Eb)

Example 2
M

Etl Eb Db Gil
t
( Dll Fl Cll F
l: Ahl Ab , Cb Bb

Example 3

Example 4 depicts the flawed third row statement, labelled L for

reference. The relevant music is displayed in three lines, to show tet-
rachordmembership. Unlike example 3, vertical page position in ex-
ample 4 correspondsroughly to relative registralposition in the music.
There is one exception: Et, the "last" pitch in the diagram, is placed
higher (vertically) in the example than its relative registral position
requires; the visual placement is to show that the pitch belongs to the

219
 L
Bb B4 Db Gb
F4 F[ Eb El
Cl A4 D4 Ab

Example 4

"middle tetrachord," F, F#, Eb, E". But this is the only departure
from the putative association between relative registral position and
tetrachordalstructure.
Examiningthe diagramfurther, we see that each pitch-classbut Gl
is represented, and the pitch-class Ft/Gb is represented twice. Ac-
cordingly, L is a flawed row statement, flawed with respect to the man-
agement of the aggregate. No other such departure from strict serial
patterningappears in the rows J, K, or M, or indeed in any other row
statement of the Gavotte. L can easily be corrected to conform to a
comprehensiblerow statement in the present serial context by viewing
the accidentalin Gb as a "slip of the pen," and assumingthat Schoen-
berg's intent was to write G~. Changing the accidental in front of the
letter G from b to 4 yields the row form PBb; each tetrachord in ex-
ample 4 can then be derived from the correspondingtetrachordin row
J, a statement of the form PE, under T6.4 Example 5 displays in a
network format the twelve-tone operations that relate the four row
statements discussed so far. Arrows labelled with the appropriateop-
erations extend among the nodes that represent the four row state-
ments. The row statements themselves are represented by both their
reference labels, J, K, L, and M, and their row form labels, PE, IBb,
PBb, IE.5
The diagram suggests the pairings J/K and L/M. Members of each
pair are contiguous in the music, and relate under IEB.Each member
of the first pair maps to the correspondingmember of the second pair
under T6. And looking back to examples 2 and 3, we see that the row
statements within each pair are presented in the same way: the tetra-
chords of J and K are presented melodically, one after the other, and
in alternating hands. Moreover, elements of their respective BACH
tetrachordsare repeated in the same way, as an echo, accompanying
the last tetrachord. Both L and M present their constituent tetra-
chords simultaneously, to form chords.
The Gb in measure 5 can also be understood as a flaw without at-
tending to aggregate completion. We read in Schoenberg's essay
"Compositionwith Twelve-Tones (I)" that the invarianceof the dyad
Gt/Db among the four row forms is an animating idea of the piece
(Schoenberg, 233-34). The dyad closes the first tetrachord of each
row statement represented in example 5: in all four statements, the

220
 lBk
IBO B

(PE? IBTQ,
>IRBB B)
> PB
) QiI
T6 T6

Example 5

first tetrachordis a musically prominent element, appearingeither as

the opening melodic unit, or as the registrally highest melodic unit,
the "harmonized" melody (with the partial exception of M as ex-
plained above). Consideration of the initial tetrachords, E-F-G-Db
in row J and Bb-A-G-Db in row K suggests that the Gb in the "mel-
ody" of row L, Bb-B-Db-Gb, is an error on melodic/thematic
grounds, and that Gb should be G~. That conclusion is strengthened
by the appearanceof Et -Eb -Db -Gt, the initial tetrachordin the row
statement that follows.
I see a number of ways of dismissingthe mistake. The most serious
of these is the argument that the Gb in measure 5 replaces the Gl re-
quired by row structurefor non-serial musical reasons, and that there
is no slip at all. Such a view will be driven by the belief that in this
piece (or perhaps in general), serial structurehas no independent aes-
thetic status; it is merely a precompositionalprocess necessary to gen-
erate a stockpile of possibilities which must be then edited to conform
to less definable musical and aesthetic laws. I am sure that there were
and are composers who approachserialismin this way; but everything
Schoenberg says about "composing with twelve tones" indicates he
was not one of them.
We might, for instance, observe how Gb relates to its immediate
context, as a preservation of the Db5/Gb,4dyad in measure 3; or as a
"contamination"of the sequence Gb-Eb-Ab at the end of the first
row statement in measure 1. But neither observation amounts to much
of an explanation. It will be necessary to discuss why Schoenberg con-
ceptually connects the music in measure 5 to the music in measure 1
or the music in measure 3. And none of this will get us very far. From
a serialist's perspective the slip seems egregiously overt, yet it has
survived through Schoenberg's proofreading passes-and there must
have been several over time. We must come up with an explanation
strong enough to account for the original parapraxisand Schoenberg's
subsequent resistance to correcting it.
TakingFreud's work as a methodological startingpoint, we need to
discover in the parapraxisthe two mutually interfering purposes: the
disturbedfunction and the disturbingfunction (Freud 1966, 46ff). The

221
 disturbedfunction (or sense) is simple to establish. Knowledge of it is
the basis of an ability to recognize the relevant mistake in the first
place. Accordingly, in this case the disturbed function is the serial
structureof measures 4 and 5.
Discovering the disturbingpurpose, however, is more difficult. On
Freud's view, it is indicated by the result or outcome of the slip. Ac-
cordingly, the investigation must begin by recognizing the validity of
the parapraxis, and then go on to consider the parapraxisin its own
account (Freud 1966, 49). We shall do this by carryingout an analysis
of the music relevant to the flawed row statement in measure 5. We
shall use as an analytical method the construct of Klumpenhouwer
Networks.
Example 6 abstracts five trichords from measures 4 and 5 of the
Gavotte. The chords are relevant to row L and the beginning of row
M, and are labelled appropriatelyfor reference. The chords are dis-
played so that vertical page position indicates the relative registralpo-
sition of the pitches that constitute each chord. Example 7 interprets
as KlumpenhouwerNetworks the chords L1, L2, and L3 of example
6; the networks are labelled 11, 12, and 13for reference.6 In each net-
work the T2-arrow indicates transposition by 2. Thus in network 11
C~ is Bb transposed by 2; in network 12B~ is Ah transposed by 2. An
IDB-arrowindicates inversion about Dl and Dt. Thus in 11Bb inverts
to Ft about Db and Dt; in 12 Bt inverts to F# about Db and D~. An
ID -arrow indicates inversion about D and Eb. Thus in 11F inverts
to C~ about D~ and Eb; in 12F$ inverts to B~ about D~ and Eb. The
relationships also hold, with the appropriatechanges, for 13.
The three networks shown in example 7 are related in a special
way: all contain as corresponding elements of structure an arrow la-
belled T2, an arrow labelled ID", and an arrow labeled ID1. We call
networks that contain identical graphs "strongly isographic" (Lewin
1990, 84; Klumpenhouwer 1991, 329).
Example 8 displays the chords once again, but visually rearranges
them to stress the correspondences under strong isography: the dia-
gram does not use vertical page position to reflect the relative registral
position of pitches within each chord. Instead the diagram arranges
the pitches vertically to conform to the referential graph at the left,
the graph common to all three networks. Each pitch-classon the right
side of example 8 can function as content of the node to its left in the
graph. So the dyads Bb/F in 11, A/F# in 12, and Db/D in 13may all be
inserted as contents of the nodes p and q, respectively, in the graph:
the pitch-classes in each pair relate under ID. The pairs F/C in 11,
F#/B in 12 and Dt/Eb, in 13 may all be inserted as contents of the nodes
q and r: the pitch-classesin each pair relate under ID. The pairs BbI/C
in 11, A/B in 12 and Db/El in 13may all be inserted as contents of the

222
 Bb Bit D{ Gb Eb
F4 FP Eb A{ Dt
Cl At DOI E4 E4

L1 L2 L3 L4 Ml

[0271 [025] [012] [024] [012]

Example 6

Xa~ ID~
T2 (T2
I(
(R IDE
(13
11
11 12 13

Example 7

-1 +4
O p Bb ----------- >A -------------- >Db

D< +1 -4
T2 O q Ft --------------
. F -- ---------- --->D

-1 +4
o r CI ------------- >Bh -------------- -Eb
11 12 13

Example 8

nodes p and r: the firstpitch-classrelates under T2 to the second pitch-

class in each pair.
Example 8 further investigates the strong isographies among 11, 12,
and 13, by providing arrows between corresponding pitch-classes in
adjacent chords; the arrows are labelled with "voice-leading inter-
vals," which measure the appropriatemagnitude (in semitones) and
direction through pitch-class space. The plus sign represents "up" or
"clockwise" in pitch-class space; the minus sign, "down" or "coun-
terclockwise." The format helps us to conceptualize those elements
of structure that persist through the three chord-interpretations.
"Voices" that carry elements related under T2 move the same mag-
nitude and the same direction from chord to chord: "-1" from 11
to 12; "+ 4" from 12 to 13; and "+ 3" from 11 to 13. The interval + 3
is derived by following the relevant arrows and combining under

223
addition the intervals that label the arrows. "Voices" that carry ele-
ments related under an inversion operation-either ID or ID1 move
the same magnitude but in opposite directions from chord to chord:
"-1/+ 1" from 11 to 12; "+ 4/-4" from 12 to 13; and-following the
relevant arrows and doing the appropriate arithmetic-" + 3/-3"
from 11 to 13.
Example 9 displaysthe three chords L1, L2, and L3 once again; the
chords are displayed to reflect the relative registral positions of their
constituent pitches, following the visual layout of examples 6 and 7
earlier. Arrows connect pitches related under the strong isographies
investigated in example 8: the arrows are derived by maintainingthe
voice-leading connections of example 8, but removing the interval
arrow-labels. Three registral positions are represented on the left by
numerals: numeral 1 represents the highest pitch within each chord;
numeral2, the second highest pitch within each chord; numeral 3, the
third highest pitch within each chord. Underneath each arrow ar-
rangementthe example provides the relevant permutationof registral
positions 1, 2, and 3 brought about the strong isographic correspon-
dences. The registralarrangementof the pitches in L1 permutes under
"(13)" to the registral arrangement of L2. The expression (13) de-
notes that the pitch in registralposition "1" of L1 relates via the cor-
respondences detailed in example 8 to the pitch in registral position
"3" of L2, while the pitch in registralposition "3" of L1 relates to the
pitch in registral position "1" of L2. The symbol (13) concisely rep-
resents the exchange between registral positions 1 and 3 from L1 to
L2. The pitch in registralposition "2" of L1 relates to the pitch in reg-
istral position "2" of L2. Since the pitch in the second registral posi-
tion of L1 relates to the pitch in the second registralposition of L2, the
notation (13) does not specify any change that registral position. Ex-
ample 9 shows how the registral arrangement of pitches in L2 per-
mutes under (123) to the registral arrangement of L3. Under the
correspondencesdetailed in example 8 the pitch in registralposition 1
of L2 relates to the pitch in registralposition 2 of L3, the pitch in reg-
istral position 2 of L2 relates to the pitch in registralposition 3 of L3,
and the pitch in registral position 3 of L2 relates to the pitch in reg-
istral position 1 of L3.
The arrow format and the permutations in example 9 help us to
integrate the investigations carried out in example 8 with the regis-
trally ordered display of example 7. Looking back to example 7, and
bearing in mind the investigationsof examples 8 and 9, we can see the
graph of 11 "registrallyinverted" in the graph of 12; the T2-arrowsin
each graphrun between the highest sounding and the lowest sounding
pitches, but in opposite directions. And the relative registralpositions
of the I- and ID -arrows are exchanged: in 11the IDt-arrow relates

224
 1 Bb \ B Db
2 F4 F||EF
3 Cl At Dt
(13) (123)
Ll L2 L3

Example 9

the top two pitches; the IE^-arrow,the bottom two; vice versa in 12.
Comparingthe registrallyordered graphs 12 and 13 of example 7, still
bearing in mind the relevant voice-leading arrows in example 8 and
the (123) permutation in example 9, we see that the graph of 12 "reg-
istrally rotates" into the graph of 13;visually tracing each arrow from
12to 13in turn-first, the T2-arrow,then the IDt-arrow,and finally the
IDa-arrow-makes the registral rotation from 12 to 13 clear.
Example 10 shows ml, a network interpretation of M1, the first
chord of row M, and includes for comparison 11, the network inter-
pretation of L1 studied earlier. Like network 11, network ml has a
T2-arrow. But where 11 has an IDb-arrow,ml has an ID -arrow; and
where 11 has an ID-arrow, ml has an Ia-arrow. Example lla inves-
tigates the relationship further. It shows the two networks 11and ml;
the constituent pitch-classes and the relevant graphs are visually
aligned to stress corresponding elements of structure. The relevant
graph of 11 appears to the left of its pitch-class node content; the rel-
evant graph of ml, to the right of its pitch-class node content. The
structures in the example are also arranged to parallel the previous
display in example 8; in all four networks given by examples 8 and 1la
the T2-arrow extends from the top node to the bottom node.
Comparing horizontally aligned arrow labels in the two graphs,
we see that T2 of 11correspondsto T2 of ml; IDDof 11correspondsto
ID of ml; ID! of 1 corresponds to IEI of ml. Accordingly, the op-
eration that maps 11 to ml involves maintainingthe T-labelled arrow
and "transposing"by two semitones the indices of the inversions that
label the other arrows of 11 in order to yield the arrow labels of ml:
applyingT2 to ID6 of 11yields IEDof ml; applyingT2 to ID: of 11yields
IElof ml. We can represent the relationshipbetween the arrowlabels
of 11 and ml as <T2>; the notation reflects a connection between
<T2>-a one-to-one mapping of the group of pitch-class operations
onto itself in the manner just described-and plain T2, a one-to-one
mapping of the collection of twelve pitch-classes onto itself.7
Example lla also investigates the "voice-leading"relationship be-
tween the correspondingT2-related pitch-classesin 11and ml; the ex-
ample provides voice-leading arrows labelled with the appropriate

225
 T2 : EE
If
E (c~ K__~T2
11 ml

Example 10

voice-leading intervals. Bb and C, the T2-related pitch-classes in 11,

both move up 4 semitones (in pitch-class space) to reach D and E,
respectively, the correspondingT2-related pitch-classes in ml.
We saw earlier in example 8 that I-related pitches of a given net-
work move by voice-leading intervals of the same magnitude but in
opposite directions to reach corresponding pitches in a second net-
work, if the two networks are strongly isographic. But, as we have
seen, 11and ml are not strongly isographic;they are isographicunder
the network operation <T2>. Examples lib and lic investigate the
voice-leading intervals from I-related pitches in 11 to the correspond-
ing I-related pitches in ml. Example llb studies the voice-leading
structurefrom the ID -related pitch-classesin 11, B and F, to the ID-
related pitches in ml, D and El . The arrow labelled " + 4" extending
from Bb in 11 to D in ml is copied over from example lla; the new
arrow, extending from F in 11 to Eb in ml, is labelled with a voice-
leading interval expressed in two different ways. Above the arrow the
diagram gives the interval -2. Below the arrow the interval is ex-
pressed as a combination under addition of two intervals: down four
semitones (-4) plus up two semitones (+2). The interval -4 is the
inverse of the interval that labels the arrow extending from Bb to D;
+ 2 is the interval necessary to yield (in combination with -4) the in-
terval given above the arrow, -2. Representing the interval from F to
Eb as the combinationof -4 and + 2 shows the effect of <T2> on the
voice-leading structurefrom pitches related under ID6 in 11to pitches
related under IDI in ml.
Example llc investigates voice-leading from the ID -related pitch-
classes in 11, F and C, to the correspondingIE-related pitch-classesin
ml, Eb and El. The arrow extending from C in 11 to E~ in ml and
labelled +4 is copied over from example lla. The arrow extending
from F in 11to Eb in ml is copied over from example llb; once again
the correspondinginterval is expressed in two ways, as -2, and as a
combination (under addition) of -4 and +2, to display the effect of
<T2> on voice-leading structure.
The idea of "transpositionby two semitones" suggested in the re-
lationship <T2> is principallyreflected in two fields: first, in the dif-
ference between the inversion arrowlabels of 11(ID6 and IDt ) and the

226
 +4
B..................------------------- Dt o
IDq DtIEb
-2
T2 o Ft-- ----------------- -> Eb T2 o

Igt~^ I Eb
+4
Cl ------------------->
El

11 ml

Example 11a

+4
0
Bb ---------.-------- ->D 0

IDb
ID4

-2
0 - o0
F4 ----- -+E
(-4 +2)

11 ml

Example llb

+4
o Bb----------------- >D

ID
-2 Eo
T2 o Fh4------- ---------. Eb T2 0
(-4 +2)
)E +4
)
PEi

V0 Cl4 ----------------- >-E

11-----......---.. .-- ml
<T2>

Example llc

correspondingarrow labels of ml (ID and IE6, respectively); and sec-

ond, in the voice-leading organization from 11 to ml, by transposing
two semitones one face of the symmetricalvoice-leading wedge that
extends from an I-related dyad in one network to the corresponding
I-related dyad in the next network.

227
 +1
A ID~ ob----------------- > i
/ I
E
ID
+1
T2 o Dtl----- ------ Eb T2 0
(-1 +2)

+1
O Eb -----------------> El O

13-----------------> ml
<T2>

Example 12

Example 12 explores the voice-leading structure from 13 to ml.

Since the network 13has the same graph as the network 11, <T2> will
map the arrowlabels of 13to the arrowlabels of ml, just as it maps the
arrow labels of 11to the arrow labels of ml. The investigation will be
interesting, because we can study the effect of <T2> on voice-leading
structurein a different context, and because L3 maps to Ml under the
operation T1. We will be interested in how relationshipTl, which ex-
tends from the pitch-class collection L3 to the pitch-class collection
M1, interacts in the present context with <T2>, which extends from
the graph of the network 13 to the graph of the network ml.
The elements in example 12 parallel those in example 11. The
pitch-classes related under T2 in 13--D and Eb -move the same in-
terval (up one semitone) to achieve their correspondingpitches in ml,
Dt and Et, respectively. Below the arrow extending from D~ in 13to
Eb in ml, the relevant voice-leading interval is expressed as the com-
bination of one semitone down (in symmetricalcontrarymotion to the
two arrows labelled one semitone up) plus two semitones up. The re-
sult, given above the arrow extending from D~ in 13 to El in ml, is
+ 1, the voice-leading interval that labels the other two arrowsextend-
ing from pitches in 13. So replacingthe voice-leading intervalswith the
corresponding transposition operation, T1, we can yield a diagram
that explains the two structures L3 and M1 as Tl-related pitch-class
collections.
Example 13 summarizesthe relationship between the networks 11,
12, 13 and the network ml, by showing just their graphs-that is, the
networks emptied of their node content-and the <T2> operation
that maps the arrow labels on the common graph of networks 11, 12,
13 to the arrow labels on the graph of network ml.
Example 14 reexamines 13 and compares it to 14, a network inter-
pretationof L4, the chord with the flaw, Glb.Network 13is copied over
228
 IDb I
D|D4

T2 ) T2

11 = 12 = 13--------------------------- ml
<T2>

Example 13

T2
A -^T2

13 14

Example 14

from example 7 earlier. L4 is interpreted by a network that has an

arrow labelled T2, and two arrowslabelled with inversion operations,
IF and ISG. Accordingly, network 14 is isographic to network 13.
Examples 15a and 15b show the two networks 13and 14again. They
are visually rearranged, to align correspondingelements of structure,
and to relate the investigation to similar investigations earlier. Exam-
ple 15a looks at the voice-leading structurefor the T2-related pitches
in 13, Db and El, to the corresponding T2-related pitches in 14, re-
spectively, Gl and Al. Both voice-leading arrows are labelled with
voice-leading interval + 5.
In example 15b arrows extend from all three pitches in 13 to cor-
respondingpitches in 14. The arrowsfrom the two T-related pitches in
13 to the corresponding T-related pitches in 14 are copied over from
example 15a. The third arrow, from D~ in 13 to Et in 14, allows us to
study the voice-leading from I-related pitches in 13 to their relevant
counterpartsin 14. The interval that labels the arrow is displayed, as
before, in two formats: above the arrow the example shows the mag-
nitude and direction of the interval from Dt to E simply as + 2;
below the arrow the example expresses the interval as a combination
(under addition) of down five semitones (-5) and up seven semitones
(+7). The interval -5 is the inverse of the interval that labels the

229
 +5
O Db ----------------- ->Gb 0

T2 o D1l El T2 o

+5 i
o Eb---------- -----Ab o

13 14

Example 15a

+5
o D ------------------ >Gb o

ID4 +2 F
o9 D .------ ----- E o

IE I Gb
+5
o Eb------------------>Ab o

13 -----------------. 14
<T7>

Example 15b

arrows connecting the T-related pitches; +7 is the interval necessary

to produce in combination with -5 the voice-leading interval shown
above the arrow, + 2. The voice-leading structureof 13and 14suggests
that the arrow labels of 13 map under <T7> to the corresponding
arrowlabels of 14. CombiningT7 and ID, the operation that labels the
arrow between Db and Db in 13, yields IF, the corresponding arrow
label in 14. Likewise, combining T7 and IE, the operation that labels
the arrow between D and Eb in 13, yields IGbin 14.
Example 16 is a "network-of-graphs"that displays the relation-
ships among 13, 14, and ml. The arrow extending from 13 to 14, and
labelled <T7>, is based on the work just done in example 15; the
arrow extending from 13 to ml, and labelled <T2>, is derived from
example 11 earlier. Since 13is strongly isographicto 11 and 12, the ex-
ample can represent the relationshipsamong the all the networks that
interpret the chords in measures 4 and 5 (given originally in example
6). The arrow extending from 14 to ml is labelled with <T7>; that
interval is derived from applying the procedure done in connection
with 13and ml (and with 13and 14)to the networks 14and ml. But the

230
 <T2>

o 0o >o
11 =12=13 <T7> 14 <T7> ml

Example 16

group structureof angle-bracketrelationships allows us simply to cal-

culate <T7> as the appropriatelabel for the arrow that extends from
14to ml: <T7> is the operation that in combination with <T7> will
yield <T2>. Example 17 investigates, along the lines of example 15
earlier, the <T7> relationship from 14 to ml with respect to voice-
leading structure.
Example 18 displays the three chords L3, L4, and M1; here the
chords are presented to reflect the relative registralpositions of their
constituent pitch-classes; arrows join pitch-classes in adjacent chords
that are associated under the network relationships studied in exam-
ples 15 and 17. The pitch-classes of L3 correspond (under <T7>, the
relevant network relationship) to pitch-classes of L4 in the same reg-
istralposition: the highest pitch in L3 relates to the highest pitch in L4;
the second highest pitch in L3 relates to the second highest pitch in
L4; and so on. The example notates this permutation, in which related
elements maintain their registral position, as ( ). Comparing L4 and
M1, we see that (under <T7>, the relevant network relationship) the
pitch in the highest registral position of L4 relates to the pitch in the
second-highestregistralposition of M1; the pitch in the second highest
position of L4 relates to the pitch in the third highest (i.e., lowest)
registral position; the pitch in the lowest registral position of L4 re-
lates to the pitch in the highest registralposition in Ml. The registral
rearrangement is represented on the example by the permutation
(123), the permutation relevant to the registral relationship between
L2 and L3 studied earlier in example 9.
Example 19 integrates the study in example 18 with the voice-
leading studies undertaken in examples 15 and 17. Example 19 dis-
plays the graphs of the relevant networks, 13, 14, and ml; page
position reflects the relative registral position of the constituent node
contents in each structure. Below the three graphs, the example pro-
vides the pertinent angle-bracket network relationships. Below that,
the example copies over from example 18 the pitch-class content of
each chord. The registral permutation arrows of example 18 are re-
placed with voice-leading arrows labelled with the appropriatevoice-
leading intervalscopied over from the pertinent earlier examples. The
intervalsthat label arrowsinvolved with pairs of I-related pitch-classes
are expressed, as before, in both simple and compound formats to

231
 +8
o Gb----------------- > D o

-1
T2 T
E.....----------------- E T2
(-8 +7) .

Ain +8 *1E
oo\ Ab------------------> El o

14 ------------ ------ ml
<T7>

Example 17

1 Db > GL Eb
2 Eb > Ab ^ D
3 D4 > Etl E4
( ) (123)
L3 L4 Ml

Example 18

emphasize the effect of the network relationships. By following the

arrowsextending from Db and Eb in 13to Gb and Ab in 14to Dt and
Et in ml, we can trace the T2-relationshipthat persists through the
three networks. By following the arrowsextending from Dt and Eb in
13to Ab and Et in 14to El and Eb in ml, we can trace the ID arrow
label of 13, transformed under <T7> to IG in 14, and transformed
again under <T7> to IEEin ml. By following the arrows extending
from D~ and Db in 13 to E~ and Gb in 14 to Eb and D in ml, we can
trace the ID' arrow label of 13, transformed under <T7> to IF in 14,
and transformedagain under <T7> to IE in ml. The procedurehelps
us to hear the identical registral arrangement of 13 and 14, and the
"registralrotation" from 14 to ml.
Example 20 takes advantage of certain recursive properties that
hold between pitch-class operations and the angle-bracket network
operations. The example copies over the network-of-graphsfrom ex-
ample 16, but replaces each angle-bracketoperation <Tn> with the
pitch-class operation Tn. Accordingly, the node content must also
change, from pitch-class operations to pitch-classes; example 20 re-
places the graphof 13(and of 11and 12as well) with the pitch-classBb;
the remainingnode content is calculated from there, to create a well-
formed network. I chose Bb carefully: the node content of the net-

232
 ( T2 T2 ()
IFr I?

o0 0

<T7> <T7>

+5
Db > Gb +8 Eb

+5
Eb >Ab +8D

+2
D4 > El (-8 +7) El
(-5 +7)

Example 19

T2

BI
- T7 T7

Example 20

work in example 20 duplicates the pitch-classcontent of L1, the initial

chord of row L. Furthermore, comparing the network of example 20
with the chord L1 in example 6, we see a correspondence between
"registrallyhigher" in chord L1 of example 6 and "chronologically
prior" in the network of example 20; the association is particularlyin-
teresting because it is already suggested by the registrallydirected ar-
peggio of L1 in the score.
Yet the allusion by the network-of-graphsin example 16 (via ex-
ample 20) to the content of L1 is ironic: the angle bracket operations
in example 16 are premised on one particular interpretation of L1
(and the four chords that follow L1), an interpretation (11) that in-
volves one T2-labelled arrow and two I-labeled arrows; but the net-
work of relationshipsthat pertain to 11and the networks isographicto
11 suggests an altogether different interpretation of L1, one that in-
volves three T-labelled arrows and no I-labelled arrows.
Returningonce again to the chord L4 in example 6, we see that the
network interpretation 14 presented is not the only network we can
construe from the collection L4 that is isographic to the networks of

233
11, 12, 13 and ml. There is another possible network interpretation
of L4 that has, as elements of structure, a T2-arrow and two arrows
labelled with inversions.
Example 21 investigates the other network interpretation.It copies
over from example 19 the networks 13 and ml: the alternative inter-
pretation of L4 is labelled 14', for reference. We recall that 14, the ear-
lier interpretation of L4, labels an arrow extending from Gb to Ab
with T2, an arrowextending between E and Gl with IF, and an arrow
extending between Ab and E~ with IG. The new interpretationof L4,
14', labels an arw e extending fm G to
g from t A with II- an arrow
tending from E~ to Gb with T2, and an arrow extending between Ab
and E with IG8. Example 21 visually presents the networks 13, 14',
and ml to reflect the relative registral positions of their constituent
pitch-classes. Example 22, however following similarly designed ear-
lier examples, displays the 13, 14', and ml to align correspondingel-
ements of structure. The graphs of 13, 14', and ml are shown above
the relevant pitch-class collections. The visual alignment of corre-
sponding arrow labels helps us to calculate the appropriate network
relationships. Comparingthe arrowlabels of 13and 14', we see that T2
of 13corresponds to T2 of 114' I I of to I of 14', and
13corresponds
ID of 13correspondsto ID; of 14'. Since the transpositionarrowlabel,
T2, persistsfrom 13to 14', and since applyingT9 to the inversion arrow
labels of 13 yields the inversion arrow labels of 14', <T9> is the rel-
evant operation to map 13to 14'. Comparingthe arrowlabels of 14'and
ml, we see that T2 of 14' corresponds to T2 of ml, IG6 of 14' corre-
sponds to ID of ml, and ID^ of 14'correspondsto IE6of ml. Since the
transpositionarrow label T2 persists from 14' to ml, and since apply-
ing T5 to the inversion arrow labels of 14' yields the inversion arrow
labels of ml, <T5> is the relevant operation to map 14' to ml.
The angle-bracket operations are given in network format under
the three graphs. Under the three graphs, example 22 shows the con-
stituent node contents. Voice-leading arrows, labelled with the perti-
nent intervals, help to integrate the network operations and voice-
leading structure;the two notational formatsof the intervalsthat label
the middle arrow between 13 and 14', and between 14' and ml, clarify
the connection.
Example 23 is a registrally ordered display of the information in
example 22; all the structuresof the previous example are maintained
here in example 23. The display allows us to also investigate the per-
mutation of registral arrangementsunder the given network relation-
ships. Examiningthe voice-leading arrowsshows the relevant registral
permutationof the elements in one chord to the correlate structureof
the next chord. The pertinent permutational cycles are noted under
the voice-leading arrows. The registral permutation represented by

234
 1i Et i,) D0
I DEb
I'
T2 (AE b i4 (D~T2

13 14' ml

Example 21

o0~ ~ o~0 o
T2
t

% 31QT2
o% T2 (
i Db ( ) IDD
Gb IET

'<T9> <T5> >

?Q

+3 +10
D ----------------- ----- Ell------------------- -- D

+6 -5
DtI---------------------- Ab ----------------------> Eb
(-3 +9) (-10 +5)

+3 +10
Eb ----------------- ---- Gb ---------------------> EE

Example 22

the notation (132) relates the registral arrangement of 13 to that of

14', and the registral arrangementof 14' to that of ml; (132) is the in-
verse operation of (123), the permutationobserved in example 9 from
L2 to L3, and from L4 to M1 in example 18. In other words, (132)
represents a "registralrotation" of the elements in 13to the elements
of 14' and the elements of 14' to the elements of ml, but a rotation
undertaken in the opposite direction to that taken by the registral
rotation (123).
Example 24a isolates from examples 22 and 23 just the relation-
ships among the arrow labels of 13, 14', and ml for further study. The
structure displayed in example 24a is a "network-of-graphs,"along
the lines of example 19 earlier. The node on the left contains the graph
common to 11, 12, and 13; the node in the center contains the graph
of 14', the alternative interpretation of "flawed" chord L4; the node
on the right contains the graph of ml. The angle-bracketoperations
<T9> and <T5> that label the arrows from 13 to 14' and from 14' to

235
 ) T2 D) D0)
Eb 0 T2
ID 0 T2 0

GO T2
I 0
Eo? 0o E

u <T9> 3 < T5>

?
a>(^)>(8

(132) (132)

Example 23

<T2>

" " >o > o

11 = 12 = 13 <T9> 14' <TS> ml

Example 24a

T2 T2

T9 T5 T9
E a p T5

Example 24b Example 24c

ml are derived from example 22 and 23. <T2> is the result of com-
bining <T9> and <T5>, but it can also be derived from the earlier
examination of the relationship between 13 and ml in example 12.
Example 24b is a network of pitch-classes and pitch-class opera-
tions derived from the network of graphs in example 24a; the <Tn>
operations, operations on pitch-classoperations, are replaced with Tn
operations, operations on pitch-classes. Example 24b replaces the
graphs 11, 12, 13, 14', and ml contained in the nodes of example 24a
236
with pitch-classes. The node content 11 = 12 = 13in example 24a is re-
placed in example 24b by the pitch-classFt; the pitch-classcontent of
the remainingnodes is calculated from there. Ft was carefullychosen:
the resulting node contents correspond to the pitch-class content of
L2, the second chord of row L. Comparingthe registrallyordered el-
ements of chord L2 in example 6 and the network in example 24b, we
see a connection between "registrallylower" in the chord L2 of ex-
ample 6 and "chronologicallyprior" in the network of example 24b.
Comparingthe network in example 24b and the network in example
24a is also suggestive: F#, the content of the central node of the net-
work in example 24b, corresponds to the graph of 14', in the central
node of example 19a, which interprets the chord that contains the
flaw, Gb; the two networks in example 24 suggest an interesting con-
nection between Ft in L2 and the "wrong note" Gb in L4.
Example 24c investigates a different network of pitch-classes de-
rived from the network of graphs in example 24a. The node content
here-Ct, A~, D-corresponds exactly to the initial three pitch-
classes of the collection CO,AO, DO, Ab, the third tetrachord of the
row statement, which appears as the "bass-line"of the chords L1, L2,
and L3. In light of the allusion to serial structure, we can derive yet
another network of pitch-classes following the same procedure, one
that recapitulatesthe last three pitches in the right hand of measure 1,
Gb, Eb, Ab; the connection is especially suggestive given the present
context.
The preceding analysis of the "mistake" and its immediate musical
context reveals a coherent musical ideation involving all four chords in
row L and the first chord of row M. Moreover, the chords relate in
motivic ways, ways that reflect upon the pitch content of the two
chords that open the excerpt, as one alternates between interpreting
L4 with 14 and interpretingL4 with 14'.
Within the frameworkof Freud's methodology we have, by means
of the network analysis, touched upon the disturbingfunction, the sec-
ond of the mutually interferingpurposes of the parapraxisin measure
5. We may contend that function to be the musical structuringre-
vealed by the network analysis:or put less strongly, we may argue that
Schoenberg's serial conception of measures 5 and 6, the disturbed
function, is altered by a wish to engage in musical ideation something
like that captured by the network analysis. It may be claimed that in
order to be thorough we must carry out a similar network analysis of
the "corrected" form statement, and take the musical ideation re-
vealed in that analysis to represent the disturbed function. However,
someone making such an objection forgets that it is not the ideation
revealed in such an analysis that we sense is being interfered with.
Rather, it is specificallythe serial structure of the pertinent music.

237
 Under our interpretation, the parapraxiscarried out in music here
reproducesthe pattern of Freud's examples of linguisticparapraxes,in
which the semantic direction of a given syntacticalunit is shifted, more
or less drastically, by a very small alteration in phonetic or morpho-
logical structure. Here the syntactic unit, the "musical phrase" in
measures 5 and 6, loses something of its serial structureand takes on
something else very different, by alteringa single accidental, from ~ to
b. And glancing back to L4 in example 6, we see that the serially cor-
rect version of chord, Ab-Gl -E, cannot be interpreted by a network
that is isographicto all five networks 11,12,13and ml, since T2 cannot
label an arrow in any pertinent network interpretation.
However, satisfactionwith the network analysis as a representation
of the disturbed function will diminish when we consider the chords
L1 and L2 more closely. We have seen in our analysis so far how the
alternative networks-of-graphsallude to the content of L1 and L2.
But the two chords themselves make importantreferences, references
that suggest the disturbed function of this parapraxisis more compli-
cated than asserted earlier. Example 25 is an excerpt from the last
of Schoenberg's Six Piano Pieces, op. 19, written about ten years be-
fore the Gavotte. The passage is measures 4 and 5, and concludes the
opening section of the piece. Five chords are labelled for reference, V,
W, X, Y, and Z. ComparingV and X-the firsttwo chords in the right
hand of example 25-to the firsttwo chords of row L in example 1, we
see that L1 and L2 are precisely the chords X and V, at least with re-
spect to pitch-class content and registral arrangementof constituent
pitch-classes: in LI and L2 the chronological and registral relation-
ships of V and X are reversed. Comparing example 1 to example 25
somewhat further, we see more similaritiesbetween the two passages.
The melodic dyad of chords Y and Z, Gt3/F#3, and the relevant bass
note, Et2, in example 25 relate suggestively to content of the flawed
chord L4 in example 1. Indeed we can connect Gb, the mistake itself,
in measure 5 of the Gavotte to both pitches of the melodic dyad in
measures 4 and 5 of opus 19/vi. Gb relates to GOtby the conventions
of musical notation: both are written and read as inflections of G, the
note found on the second lowest line of the treble staff. Gb relates to
Ftl by conventions of serial thinking; that is, via pitch-class member-
ship.
And we can relate other structures in example 25 to our network
analysisof the Gavotte. Example 26 excerpts from example 25 the five
indicated chords; the relevant normal form is listed under each chord.
The example interpretsthe five chords V, W, X, Y, and Z as Klumpen-
houwer Networks, labelled v, w, x, y, and z, respectively. The net-
works 11, 12, 13, 14, and ml in the analysis of the Gavotte along with

238
 w cE.-
. _ Z
Example 25

the normal forms relevant to the collections L1, L2, L3, L4, and M1
are included in example 26 for reference.
Networks v and x, the interpretationsof V and X, are strictlyiden-
tical to 12 and 11, respectively, since the pitch-class contents V and X
are strictlyidentical to L2 and L1, respectively. The remainingchords
in the excerpt from op. 19/vi may be interpreted by networks that are
isographicto v and x; that is, W, Y, and Z may each be interpretedby
a network that labels one of its constituent arrowswith T2, and labels
its other arrows with inversion operations. The rules for combining
pitch-class operations dictates that the indices of the inversions will
differ under T2.
Comparing y in opus 19/vi to 14 in the Gavotte, we see that the
two networks label their arrowswith identical operations: T2, I8G(or
IF), and IF. And both w in opus 19/vi and ml in the Gavotte, label
their arrows with T2, ID6, IEl. Since the graphs of v (and x), y, and
w are identical to the graphs of 11 (and 12 and 13), 14, and ml, respec-
tively, we can extend the angle-bracketoperations that obtain among
the graphs in the Gavotte, shown in example 26 under the last three
networks, to the networks v, w, x, and y. Example 27 copies over the
Gavotte-network from example 25; the op. 19/vi-network is con-
structed on the basis of the similarities just observed in example 26.
So the graphs of five of the networks in the excerpt from op. 19/vi
occur in the analysis of the Gavotte; only the graph of network z does
not. Example 28 investigates that network in more detail. The exam-
ple includes for context the networks x and y, the interpretations of
chords X and Y. All the structuresin the example are rearrangedfrom
the registrallyordered display of example 26, in order to align corre-
sponding elements among the three networks. We can derive the op-
eration <T7> as the pertinent mapping of x to y from example 27.
The intervals that label the voice-leading arrows from the pitch-class

239
Op. 19/vi, mm. 5, 6

Db t _ D IE
ID4
Ea
i F2 If
T2 ( T2 (
ID4
IEN _ )>T2
y
v x w y z
[0251 [0271 [027] [026] [024]

Gavotte, mm. 4, 5
E~)
T ID ) T2 / (P T2
T2 |F T2 () ID @ IF
iF
() (@
( D
iD J ?IG _ _ T2
(EG)

11 12 13 14 ml
[0271 [025] [012] [024] [012]
<T7> j~ ,OT0 < T7>
<T2>>

Example 26

content of x to the pitch-classcontent of y display the effect of <T7>

on the voice-leading structure.
Comparingthe arrow labels of y with the correspondingarrow la-
bels of z, we see that T2 of y maps to T2 of z, IF of x maps to IE of
z, and IF of y maps to IFof z. Since T2 persists from y to z, and since
we can "transpose"by ten semitones the indices of the inversion op-
erations in y to generate the inversions of z, <T10> is the relevant
angle bracket operation from y to z. Under the graphs y and z, the
example displays the pertinent node content of Y and Z and studies
the voice-leading structure under <T10>.
Example 29 is a network-of-graphsthat incorporates the relation-
ships studied in example 28; the display indicates the hand to which
each interpreted chord is assigned. The single node at the top of the
network represents the graph common to both v and x. The left side
of the network, which contains the nodes relevant to w, y, and x along
with the angle bracket operations that obtain among them, is just the

240
 <T7> <T7>
> 0 > 0
v=x y w

<T2>

<T7> <T7>
0.,, -o 0o
11=12=13- 14 ml

<T2>

Example 27

/D4^ / ) F f */ E
0IT2D T2 0 :

<T7> <TIO>

+4 +0
B .....................-
-- D4 ----------------- ---- > Dl

+3 +10
F G .----------------- ----- >F
---------...-.--------
.>
(-4 +7) (-0 +10)

+4 +0
Cl---------------------> El----------------------> EI

Example 28

network displayed in example 27 visually rearranged. The node con-

taining z and the arrow extending from y to z and labelled <T10> is
derived from the previous example; the arrow extending from the
node v = x to the node z, and labelled <T5> is derived by combining
<T7>, which labels the arrow that extends from node v = x to node
y, and <T10>.
Example 30 takes advantage of recursive opportunities between
angle bracket operations and pitch-classoperations. It copies over the
network-of-graphsfrom example 29, but replaces each angle-bracket
operation <Tn> with the pitch-classoperation Tn. The node content
241
 righ.t:

<T2> / \

<T7>\<T5>

v
<T7> <T10>
left. () ,\

Example 29

T2

T7
T7 T10

Example 30

must also change, from pitch-class operations to pitch-classes. Exam-

ple 30 replaces the graph common to v and x with the pitch-classFP;
the remaining node contents are calculated from there, to create a
well-formed network. I chose F carefully:the node content of left half
of the network-F, C, G-duplicates the pitch-classcontent of W, the
initial chord in the left hand. And the node content of the right half of
the network-F, C, B -duplicates the pitch-class content of X, the
second chord in the right hand.
Example 31 is another network of pitch-classes derived from the
network of graphs in example 29; the example copies over the part
of the network in example 29 relevant to the chords in the left hand
only; the angle brackets operations <T7> and <T10> are replaced
with the correspondingpitch-class transpositionsT7 and T10; the op-
eration that labels the arrowextending from the node on the left to the
node on the right is derived by the rules for combining transpositions:
T7 followed by T3 yields T10. The graphs that occupy the nodes in
example 29 are replaced with pitch-classes, to create a well-formed
network. The resultingpitch-classcontent duplicatesthe content of V,
the opening chord in the excerpt.

242
 T3

T7 _ T10

Example 31

Example 32 compares the various network relationships observed

in the analysis of the excerpt from op. 19/vi to the network relation-
ships observed in the analysis of the excerpt from the Gavotte. Ex-
ample 32a gives the network of network-relationships that obtain
among v and x (as content of the same node), w and y; example 32b
gives the network of network-relationshipsthat obtain among v, x, y,
and z. Both networks are simply extracted from the diagram in ex-
ample 29. Since the two networks have the same graph, they are
strongly isographic. They are in turn strongly isographic to the net-
work of network-relationshipsthat obtain among 11,12,13,14, and ml,
in the analysis of the Gavotte, copied over here as example 32c. Ear-
lier examples have shown how a graph of transpositionsderived from
the graph common to examples 32a, 32b, and 32c can suggest similar
chords in each excerpt: L1 in the Gavotte and X in op. 19/vi.
Example 32d gives the network of network-relationshipsthat ob-
tain among w, y, and z of op. 19/vi; example 32e gives the network
relevant to 11,12,13,14' (the alternativeinterpretationof L4), and ml.
Both networks label corresponding arrows with identical operations,
and are thus stronglyisographic. Variousearlier examples have shown
how a graph of transpositionsderived from the graph common to ex-
amples 32a, 32b, and 32c can suggest similar chords in each excerpt:
L2 in the Gavotte and V in op. 19/vi.
By referring to the Network model of chord structure, we have
been able to associate the flawed version of the excerpt from the Ga-
votte with the excerpt from op. 19/vi. Specific graphs are common to
the analyses of the two passages, though they do not entirely overlap.
The graph of 14' cannot interpret any of the five chords in op. 19/vi;
and the graph of z cannot interpret any of the five chords in the ex-
cerpt from the Gavotte. Even so, we are able to construct identical,
strongly isographic networks of graphs for both excerpts, which in
turn suggest the pitch content of the two chords that open both ex-
cerpts. The associations revealed in the network analysis are not
obvious. Some connection may be suggested by the observed relation-
ship between the error, GI, supported by E 2 in the Gavotte and the
melodic dyad Gt-F# supported by E~2 in op. 19/vi. Nevertheless, an-
alyzing the chords in the two passages as pitch-class collections has
very little explanatory power in this case.

243
 <T2>

a) uIZ) T
3- ><T(i <
>O ?
<TT'
<T7> <T7>

<T2>

b) >GD >
) 2 <T7> (E) <T7>

<T2>

c) 11=12=13 >
< T7> ^-
><T7> <T7>

<T2>

d) d) (
<T9> <T5>
(

<T2>

e) 11=12=13 >
<T9> <T5>

Example 32

In our earlier account of the parapraxis,formed before we exam-

ined op. 19/vi, the purpose that involves the serial conception of the
excerpt from the Gavotte was disturbed by a wish to bring about the
musical ideation represented by the network analysis of the flawed
version. In that account the two mutuallyinterferingpurposes are not
related in any obvious way. The structureand relationshipspresented
in the network analysis of the excerpt from op. 19/vi suggest a more
elaborate, and more powerful explanation of the parapraxis, and an
account that can place the disturbedfunction in a series of associations
with the disturbingfunction.
Example 33 visually arranges the various structures discussed so
far to suggest such a series.8 At the top of the diagram the example
represents as letter O the putative correct version of the Gavotte,
244
 O) CorrectVersionof Gavotte,measures4, 5

SerialConceptionof Gavotte, measures4, 5

ChordsLI and L2 as allusionto op. 19/vi, measures5, 6

MusicalIdeationof op. 19/vi, measures5, 6

O(i MusicalIdeationof flawedversionof Gavotte, measures4, 5

(iO FlawedVersionof Gavotte, measures4, 5

Example 33

measures 4 and 5. By this I mean only the arrangementof pitches ap-

propriate to such a version. This version is engendered by Schoen-
berg's serial conception of the relevant passage of the Gavotte,
represented on the diagram by the letter P. The serial conception of
the excerpt in turn yields the chord L1, by gathering together the first
three pitch-classes of each tetrachord in PBb, and the chord L2, by
gathering together the second pitch-class of each tetrachord in PBb.
The chords themselves allude to the music in the right hand of op.
19/vi, measures 6 and 7. The reference by L1 and L2 to op. 19/vi is
represented on the diagram as letter Q.
At letter R example 33 represents the musical ideation of the rel-
evant passage from op. 19/vi, as that is captured by the network anal-
ysis. The ideation involves the two chords, U and W, referredto by L1
and L2. The role of U and W, however, extends beyond functioning
as node contents in u and w: the two chords are reflected in the re-
lationships among all the chord interpretationsin the excerpt. Signif-
icant structures in that analysis are also relevant to letter S, which
represents the network analysis of the flawed version of the Gavotte,
measures 4 and 5. This in turn engenders the flawed version itself; let-
ter T represents the particulararrangementof pitches relevant to the
published version.

245
 The diagram allows us to connect, through the series of observed
associations, the hypothetically correct version at O and the flawed
version at T, which differs from O by a single accidental. The diagram
also helps us to articulate a new account of the two mutually disturb-
ing functions in the parapraxis.As before, we take the disturbedfunc-
tion of the passage to be the serial conception of measures 4 and 5.
The disturbingfunction is the wish to carry through the quotation of
op. 19/vi, past the two chords in the right hand at measure 6.
This account suggests that something like the following occurred
to Schoenbergas he wrote the passage of the Gavotte. He had noticed
at some time-either when he constructed the row, or when he was
working on the Gavotte itself-that the three tetrachords of PBb,
when presented as simultaneous four-note melodies, generate chords
that have the same content as certain important chords in his earlier
op. 19/vi. We might further assume that he was pleased or at least
struck by the connection. The well-circulated claim that Schoenberg
conceived op. 19/vi as a tombeau to Mahler is particularlysuggestive
in this context, especially since in op. 25 Schoenberg is writing a par-
ody of a baroque keyboard suite that involves the cryptogram of
Bach's name as an importantharmonic and melodic device (Stucken-
schmidt 1977, 108; Lewin 1982-83, n.9).9 The desire to continue with
the quotation, out of a compulsion to repeat the composing of opus
19/vi, or out of the dynamicsof his relationshipwith Mahler, suggests
itself to Schoenberg; but he suppresses it. The desire, however, finds
an effective way to manifest itself, as a mistake in writinga flat instead
of a natural before the G in measure 5.
The suggested scenario recalls Freud's concluding remarks about
"slips of the tongue," a phenomenon to which slips of the pen may be
reduced. He writes:
... [T]he disturbingpurpose is recognizedby the speaker; further-
more, . . . that purpose announces itself before the slip. But, . . . it is
forced back. The speaker decides not to put it into words, and after
that the slip of the tongue occurs:after that, that it is to say, the pur-
pose, whichhas forcedbackis put into wordsagainstthe speaker'swill,
eitherby alteringthe expressionof the intentionwhichhe has permit-
ted, or by minglingwith, or by actuallytakingits place. This, then, is
the mechanismof a slip of the tongue (1966, 65)
In our case the chords LI and L2 are viewed as written traces of the
disturbingpurpose; but they can also trigger the intention in Schoen-
berg's subsequent editing passes, or at least distracthim from correct-
ing the error.

246
 NOTES

1. Fehlleistung translates more literally as "flawed action."

2. Schoenberg conceived the BACH tetrachord to be the last in the row, as it is
in the Praeludium. In his essay "Composition with Twelve Tones (I)," Schoen-
berg assumes a listener of the Gavotte who is already familiar with the preced-
ing movement, the Praeludium, and who is able to discern the relationship
between the row forms of the two movements (Schoenberg 1975, 232-33). I
will, however, continue in this paper to reference the tetrachords as they appear
in the Gavotte.
3. I am using Lewin's model for pitch-class inversion: IE signifies inversion about
the pitch-class dyad E/BI,. Applying I, to pitch-class X involves calculating
the interval i from X to one element of the dyad E/Bb -say, E-and then cal-
culating the pitch-class that lies i from the remaining element of the dyad-in
this case, Bb (Lewin 1987, 54).
4. As far as I can tell, Gb appears in all extant versions of the Gavotte, from the
initial complete handwritten copy to the first published text. And it appears in
the Gesamtausgabe. There are no surviving sketches (Brinkmann 1975, 30ff.).
5. Schoenberg discusses these four row forms as the fundamental serial structures
in the Suite, op. 25; PE corresponds to his "Basic Shape" (Schoenberg 1975,
232).
6. Throughout the essay, uninterpreted pitch-class collections are labelled by
upper case letters; the network interpretations of pitch-class collections are la-
belled by corresponding lower case letters.
7. David Lewin defines the angle bracket operations in Lewin 1990, 88-90. My
notation differs somewhat from Lewin's; my <Tn> corresponds to his <l,n>.
8. The diagram itself invokes the procedure Freud used to explain his own mo-
mentary forgetting of the name of Signorelli as the artist responsible for the
frescoes in the dome of Orvieto (1960, 38).
9. Lewin's discussion includes some speculation about how the various musical
structures in op. 19/vi can represent (among other things) funeral bells at Mah-
ler's funeral (a common reading of the long held chords) and reminiscences
about Mahler's conducting.

LIST OF WORKS CITED

Brinkmann, Rheinhold. 1975. Arnold Schonberg: Abt. II, Klavierwerke und Orgel-
musik. Riehe B, Band 4. Kritischer Bericht. Mainz: Schotts S6hne/Vienna: Uni-
versal.
Cone, Edward. 1972. "Editorial Responsibility and Schoenberg's Troublesome
'Misprints'." Perspectives of New Music. 11.1: 65-76.
Freud, Sigmund. 1960. Psychopathology of Everyday Life. Translated by Alan
Tyson, and edited by James Strachey. New York: W.W. Norton.
. 1966. Introductory Lectures on Psychoanalysis. Translated and edited by
James Strachey. New York: W.W. Norton.
Klumpenhouwer, Henry. 1991. "Aspects of Row Structure and Harmony in Mar-
tino's Impromptu No. 6." Perspectives of New Music. 29.2: 318-354.

247
Lewin, David. 1982-83. "Transformational Techniques in Atonal and Other
Music." Perspectives of New Music 21.1-2: 312-71.
. 1987. Generalized Musical Intervals and Transformations. New Haven:
Yale University Press.
. 1990. "Klumpenhouwer Networks and some Isographies that Involve
Them." Music Theory Spectrum 12.1: 83-120.
Schoenberg, Arnold. 1975. Style and Idea. Translated by Leo Black, and edited by
Leonard Stein. London: Faber & Faber.
Stuckenschmidt, H. H. 1977. Schoenberg: His Life, World, and Work. Translated
by Humphrey Searle. London: Calder.

, /' 2-~"I
_ III

oe--
~~11
II ~I ~ q, ~

248