This article was downloaded by: [University of Western Ontario]

On: 12 November 2014, At: 16:56

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,
UK

Journal of Musicological
Research
Publication details, including instructions for
authors and subscription information:
http://www.tandfonline.com/loi/gmur20

Schoenberg as Performer of His

Own Music
a
Roland Jackson
a
Professor emeritus, Claremont Graduate University
Published online: 31 Jan 2007.

To cite this article: Roland Jackson (2005) Schoenberg as Performer of His Own Music,
Journal of Musicological Research, 24:1, 49-69, DOI: 10.1080/01411890590915494

To link to this article: http://dx.doi.org/10.1080/01411890590915494

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the
information (the “Content”) contained in the publications on our platform.
However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness,
or suitability for any purpose of the Content. Any opinions and views
expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the
Content should not be relied upon and should be independently verified with
primary sources of information. Taylor and Francis shall not be liable for any
losses, actions, claims, proceedings, demands, costs, expenses, damages,
and other liabilities whatsoever or howsoever caused arising directly or
indirectly in connection with, in relation to or arising out of the use of the
Content.

This article may be used for research, teaching, and private study purposes.
Any substantial or systematic reproduction, redistribution, reselling, loan,
 sub-licensing, systematic supply, or distribution in any form to anyone is
expressly forbidden. Terms & Conditions of access and use can be found at
http://www.tandfonline.com/page/terms-and-conditions
Downloaded by [University of Western Ontario] at 16:56 12 November 2014
 Journal of Musicological Research, 24: 49–69, 2005
Copyright © Taylor & Francis, Inc.
ISSN 0141-1896 print / 1547-7304 online
DOI: 10.1080/01411890590915494

Journal
241Taylor
Taylor
GMUR
3
1
2005
10.1080/01411890590915494
49515
Roland
Schoenberg
2 &Jackson
ofFrancis,
Musicological
&asFrancisTaylor
Performer
Inc. Research
of and
His Francis
Own Music
325 Chestnut StreetPhiladelphiaPA191060141-18961547-7304

SCHOENBERG AS PERFORMER OF HIS OWN MUSIC

Roland Jackson
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

Professor emeritus, Claremont Graduate University

Five recordings of Schoenberg conducting his own music have been

preserved: Verklärte Nacht, “Lied der Waldtaube,” Pierrot Lunaire,
Suite Op. 29, and Von Heute auf Morgen. These offer a valuable
resource, showing many aspects of Schoenberg’s own approach to inter-
pretation. Although produced between 1927 and 1940, the recordings
reveal that Schoenberg adhered closely to the late–Romantic tradition of
performing that was typical at the turn of the century. Most striking is
Schoenberg’s pervasive use of rubato as well as his frequent deviations
from his own metronomic markings. The recordings—in conjunction with
his published and rehearsal scores—also give evidence of Schoenberg’s
detailed attention to dynamic and articulative nuances, and provide a
model for Schoenberg’s use of Sprechstimme. In his writings, Schoenberg
expressed firm convictions concerning a composer’s rights, which were,
as he said, confined to an adherence to the notes and rhythms. To the
performer, on the other hand, he accorded a number of freedoms (as was
typical of the time around 1900), especially those of dynamics, tempo,
and timbre.

Recordings by composers of their own works offer invaluable insights

into how their music might be performed. A few composers (for instance,
Rachmaninoff or Bartók) made a considerable number of such record-
ings; others did so only rarely. Among the latter was Arnold Schoenberg
(1874–1951), who, despite his enormous importance in his own time, left
only a few recorded examples of performances of his works. These few
examples show, perhaps surprisingly, that—despite his significant fore-
shadowing of future developments in twentieth-century music—his inter-
pretations owe more to late–nineteenth-century traditions of performance
than they do to those of the unfolding twentieth century.
 50 Roland Jackson

Five recordings of Schoenberg conducting his own music have come

down to us.1

1. Verklärte Nacht (1899), mm. 1–200; (recorded in Berlin, 1928,

orchestra unknown);
2. “Lied der Waldtaube,” Gurrelieder (1900–1901); (recorded in
New York, April 7, 1934, with Rose Bampton, soprano, and the
Cadillac [i.e., the NBC] Symphony Orchestra);
3. Pierrot Lunaire (1912); (recorded in Los Angeles, September
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

24–26, 1940, with Erika Stiedry-Wagner, reciter, and Kolisch,

Auber, Steuermann, Posella, and Bloch, instrumentalists);
4. Suite, Op. 29 (1925–1926); (recorded in Paris, December 15,
1927, world premiere);
5. Von Heute auf Morgen (1928–1929); (recorded in Berlin, 1930,
with Margot Hinnenberg-Lefèbre, soprano, and Gerhard
Pechner, baritone, orchestra unknown).

The recordings of Verklärte Nacht, “Lied der Waldtaube” (Gurrelieder),

the Suite Op. 29,2 and Von Heute auf Morgen are in the possession of the
Arnold Schoenberg Center, and Pierrot Lunaire has been released commer-
cially by Columbia Records.3 A sixth recording, involving Schoenberg’s
Kol Nidre (also available at the Center), includes only the brass instru-
ments playing two measures (199–200), but it is nonetheless remarkable
for preserving Schoenberg’s own comments to his orchestra during a
rehearsal prior to the world premiere (in Los Angeles, October 4, 1938).4

1
I would like to express my gratitude to Dr. R. Wayne Shoap, former archivist in the
Arnold Schoenberg Institute at the University of Southern California, for aiding in the
accessing of the institute’s Schoenberg recordings and related materials and for unstintingly
offering of his time and advice with regard to a whole range of questions. The Arnold Schoenberg
Institute has since been relocated to Vienna, where it is known as the Arnold Schoenberg
Center in Vienna (scanned copies of their holdings are currently available at http://
www.schoenberg.at).
2
The record cover bears the inscription “Première 1927.” Another early recording—not (as
one might conjecture) a mere copy of that of 1927—appeared in France in 1947(?) on a Contre-
point label: Contrepoint CO 10/13 (Paris, 1947?), 78.
3
Pierrot Lunaire initially was recorded in Los Angeles, September 24, 1940, on Columbia
M461 (71157-D/160-D), 78, and was subsequently redubbed and reissued on the following:
Columbia MM 461 (71161-D/64-D) (XH23-XH30) (1948), 78; Columbia ML 4471 mono
(1951), LP; Philips L 01 515 L mono (1961), LP; CBS 61 442 mono (1974), LP; Odyssey Y
33791 mono (1975), LP; CBS Sony 20 AC 1887 mono (1984), LP; CBS MPK 45659 digital
mono (1989), CD. See R. Wayne Shoaf, The Schoenberg Discography, 2nd ed., revised and
expanded, with foreword by Leonard Stein (Berkeley, CA: Fallen Leaf Press, 1994).
4
The orchestra remains unidentified, although according to R. Wayne Shoap, it might very
well have been the WPA Orchestra of Los Angeles. Other possibilities include the San Diego
Symphony or the Los Angeles Philharmonic.
 Schoenberg as Performer of His Own Music 51

Although Schoenberg made his recordings in the second quarter of the

twentieth century—between 1927 and 1940—his approach to perfor-
mance in many ways reflects the period shortly before and after 1900, a
more intensely Romantic time, when individuality reigned and performers
were given more freedom to make expressive additions of their own to a
musical score. Such performances tended to be more exaggerated than
those to which we have grown accustomed today, conditioned as we are
to a more literal interpretation. Inserting tempo rubato, adopting different
tempi in accordance with the nature of a theme, and an abundance of
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

expressive nuances were typical of the late–Romantic style. Each of these

types of musical alteration contributed to the highlighting of themes or
motives, something that had become a central feature of musical style
around 1900. And although it is certainly true that composers during
the latter part of the nineteenth century sought increasingly to notate these
features into their written scores, such features also were often added
improvisatorially in performances of the time.
Recordings from shortly before and after 1900 afford an important
means for discovering just what elements tended to be added in perfor-
mances. Especially valuable are the renditions of the composers them-
selves: Mahler, Scriabin, Grieg, and Debussy, among others, left recorded
versions of their own works on player pianos such as the Welte-Mignon,
allowing us to observe how they at times went beyond what was present
in their notated music. The same is true of Schoenberg’s recordings,
which provide (along with his rehearsal scores and preserved personal
comments and writings) a rich legacy, inviting the researcher to look
deeply into various aspects of his performance. What seems most striking
is the extent to which the recordings digress from what is present in
Schoenberg’s published (or even in his rehearsal) scores.5 As such they
bring to light a side of the composer hitherto unexamined, revealing
aspects of his musical style he presumably deemed too obvious or too
intrinsic to require being written down. An examination of Schoenberg’s
recorded performances reveals his use of tempo rubato, his digressions from
his indicated tempi, his added articulative or dynamic nuances, and his
manner of realizing Sprechstimme. They also shed light onto Schoenberg’s
feelings about the performer, as expressed in certain of his personal
writings.

5
For the works considered here, Schoenberg’s most frequent publisher was Universal Edi-
tion (U), the others being Birnbach (B), Boelke-Bomart (BB), Belmont (Bel), Eulenburg (E),
Kalmus (K), International (I), and Schotts Söhne (S). Thus we have Verklärte Nacht (B, E, I, K,
U), Gurrelieder (U, Bel), Pierrot Lunaire (U), Suite, Op. 29 (U), Von Heute auf Morgen (S), and
Kol Nidre (BB). The Complete Edition at present includes Von Heute auf Morgen and Pierrot
Lunaire.
 52 Roland Jackson

SCHOENBERG AND TEMPO RUBATO

Schoenberg rarely used the word rubato, yet his performances abound in
rubato effects.6 For Schoenberg, tempo rubato—the momentary hastening
or slowing down of the music—seems almost to have been used instinc-
tively and, as evidenced in his recordings, was frequently present despite
the absence of specific indications in the score. This is apparent with
regard not only to his own music, but to that of other composers as well.
An example may be found in his recorded performance of the second
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

movement of Mahler’s Symphony no. 2 by the Cadillac (or NBC) Sym-

phony in 1934.7 In it Schoenberg invariably interrupts the normal rate of
motion at the ends of important phrases or sections as a means of punctu-
ating them. Example 1 provides an illustration; here the melody in the
first violins is gradually slowed on the last four sixteenth notes of the
measure, leading gracefully into the cadence on A
. Such momentary
ritards were nowhere indicated by Mahler, nor are they marked into
Schoenberg’s (conducting?) copy of Mahler’s Symphony, now in the pos-
session of the Center.
While Schoenberg’s procedure here informs us specifically about his
approach to Mahler, at the same time it casts considerable light on the
performance of his own music, the recordings of which contain numerous
instances of such unmarked ritards. Schoenberg, to be sure, frequently
marked into his scores ritardando, molto ritardando, or their equivalents
(such as sehr zurückhaltend or pesante), usually followed shortly thereaf-
ter by a tempo. But what is unusual about these markings, as evidenced in
the recordings, is Schoenberg’s propensity to extend the ritards back
from the place where they are actually indicated (often by several mea-
sures), thereby making them far more emphatic. A few instances may be
cited:

Example 1. Gustav Mahler, Symphony no. 2, second movement, first

violins, mm. 30–31.

6
Richard Hudson points to Schoenberg’s extensive rubato effects in Stolen Time: The
History of Tempo Rubato (Oxford; Clarendon Press, 1994), 356–358, singling out for discus-
sion Erwartung and the Piano Concerto.
7
The recording is available in the Arnold Schoenberg Institute.
 Schoenberg as Performer of His Own Music 53

“Lied der Waldtaube,” mm. 1, 6, 15, 69, 106;

Pierrot Lunaire:
No. 1, mm. 36–37—prior to a poco rit. in m. 38;
No. 5, mm. 35–42—prior to a rit. in m. 43;
No. 19, mm. 43–49—the conductor’s score has a molto rit. in
mm. 50–51;
No. 20, mm. 28–30.

Other verbal expressions in German or Italian can constitute a means of

Downloaded by [University of Western Ontario] at 16:56 12 November 2014

signaling momentary changes of pace; some examples follow:

espressivo (e.g., Verklärte Nacht, m. 18), a slight slowing;

ruhiger (e.g., Von Heute auf Morgen, m. 851), slower;
wesentlich ruhiger (e.g., Von Heute auf Morgen, m. 289), consider-
ably slower;
steigernd (e.g., Verklärte Nacht, m. 24, 31), faster.

Another marking that affects the pace is the crescendo, which often is
associated with a slight hastening. These various indications occur often
in Schoenberg’s music and can now be interpreted more precisely in light
of his recorded realizations.
Turning to Schoenberg’s performance of his own works, Verklärte
Nacht—of which the first 200 measures are preserved in his recording—
offers a useful starting point, providing a window into his unique manner
of introducing rubato. To be sure, rubato is more markedly present in this
early work (it was completed in 1899) than in his later compositions. At
the time of the recording in 1928, Schoenberg would have have been
working from his orchestral version of 1917. Many of the rubato changes
introduced spontaneously in the 1928 recording are not included in the
1917 score, and were subsequently added by the composer in Italian into
the 1943 revised edition. Only the calando (=diminuendo and perhaps
ritardando) in m. 41 runs contrary to Schoenberg’s earlier indication
accelerando, although the metronomic indications provided in the two
scores are both surprisingly faster than those in the recordings (see Table 1).
Schoenberg commences at MM. = 52 in his recording, but already in
o
m. 3, the last part of the measure is slightly slowed (52 to 48) due to a
miniscule lingering on beat 4, probably calculated to set apart the repeat
of the motive on the last sixteenth note. A similar effect recurs in m. 13,
only somewhat more intensified (56 to 46). The espressivo in m. 18 is
accompanied by a slight lessening (to 44), the steigernd in m. 24 by a slight
increase (to 54), and the steigernd (i.e., crescendo) in m. 31 and continuing
crescendo in m. 32 by an even greater increase (60 to 69). This culminates in
 54 Roland Jackson

Table 1. Verklärte Nacht: Tempos and Indications, Published and Recorded

Versions (mm. 1–50)
m. Tempo (1917 ed.) Actual tempo (1928 rec.) Additions (1943 rev. ed.)

1 sehr langsam MM. = 52 o

3 52/48 [changes within]
13 56/46 [changes within]
17 48
18 espress 44
21 rit.
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

24 steigernd 54
25 [rit.] poco a poco accelerando
27 rit.
28 molto rit.
29 etwas bewegter 56 poco più mosso
31 steigernd, cresc. 60 [and accel.]
32 cresc. 69 poco accelerando
34 [ = 96, S’s conducting score]
o moderato [ = 84]
o
37 72 [top of mel. contour]
39 [rit.] rit…
40 [a tempo] a tempo
41 [rising scale, accel.] calando
45 63
46 [unmarked rit.]
47 rit. [rit.] rit.

an even greater acceleration (to 72) in mm. 36–37, as the melodic line reaches
its crest (see Example 2).8 Also noteworthy is Schoenberg’s extending back-
ward by a full measure the ritard marked in m. 47, making it more pro-
nounced (a Schoenbergian characteristic mentioned above).
What this performance of Verklärte Nacht displays most conspicu-
ously is its continual fluctuation of the pace. Schoenberg seems through-
out his career to have remained averse to the idea of maintaining strict
metrical exactness; for example, as late as 1948 he said, “almost every-
where in Europe music is played in a stiff, inflexible metre—not in a
tempo, i.e., according to a yardstick of freely measured quantities.”9 The
expression “freely measured quantities” is indicative for Schoenberg, and
he made this a central aspect of his approach to performance. Consider,
for example, what Marcel Dick, one of Schoenberg’s performers, had to
say about his rehearsals:
8
The crescendo with a rising melodic line (and decrescendo with a falling line) had already
been advocated by Pierre Baillot in his L’art du violon, nouvelle méthode (Paris, 1834).
9
“Today’s Manner of Performing Classical Music of the So-Called ‘Romantic’ Type”
(1948), as reproduced in Style and Idea: Selected Writings of Arnold Schoenberg, ed. Leonard
Stein (New York: St. Martin’s Press, 1975), 320.
 Schoenberg as Performer of His Own Music 55
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

Example 2. Schoenberg, Verklärte Nacht, mm. 3–37; metronomic

fluctuations.

He was terribly meticulous about rhythms… And when you finally got
it … he says, “Yes, but it sounds stiff and … it has to be free … and
not in a strait jacket.”10

It seems significant in this regard that later recordings of Verklärte

Nacht by other conductors, such as Robert Craft and David Atherton,
tend to be more consistent in their pace than Schoenberg’s had been (see

10
Cited in Joan Allen Smith, Schoenberg and His Circle: A Viennese Portrait (New York:
Schirmer Books, 1975), 112.
 56 Roland Jackson

Table 2. Verklärte Nacht: Tempi in Performances by Craft and Atherton

(mm. 1–50)
m. Indication Craft recording Atherton recording

1 sehr langsam MM. = 63

o MM. = 44
o
17 cresc. [rit.] 48–46
18 espressivo 63 44
21 rit. [ignores rit.]
24 steigernd 54
26 [slight accel.]
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

28 63
29 etwas bewegter 88
34 [1943: moderato = 84] 69–72
37 96
40 80 [slower]

Table 2).11 Both Craft and Atherton, of course, were able to profit from
Schoenberg’s 1943 edition, with its added directives. But with respect to
Schoenberg’s initial tempo of 52, Craft is rather more rapid (at 63),
whereas Atherton is slower (at 44). Both adhere more strictly to their
original tempo, however, accelerating only at particular moments—in
contrast to Schoenberg’s more pervasively fluctuating realization.
Schoenberg’s use of rubato, his ever-present ritards and accelerandos,
both marked and unmarked, place him squarely within the late–Romantic
tradition. As Mathis Lussy had indicated in 1874, some performers digressed
in this regard in nearly every phrase.12 But it was the composer Franz Liszt
who earlier played a key role in establishing shifts of pace as a routine proce-
dure. He very much opposed, for instance, the mechanical up-and-down
beating of many conductors of his time, and he emphasized the impor-
tance of shaping an entire phrase rather than being concerned about the
beats of a measure.13 He also wrote that the introduction of rubato was some-
thing that could be left to the taste and momentary feeling of gifted players.14

SCHOENBERG AND THE METRONOME

The latter part of the nineteenth century saw an increasing distrust of the
metronome. Liszt, for example, wrote that a metronomical performance

11
Robert Craft, Canadian Broadcasting Symphony Orchestra, Columbia Masterworks
recording M25 694 (ms 6531/32), stereo (1963), LP; David Atherton, London Sinfonietta,
Decca recordings SXLK 6660/64, stereo (1974) LP, and SDD 519, stereo (1977), LP.
12
Traité de l’expression musicale (Paris, 1874; 8/1904; Eng. 1885), 163.
13
[“Preface”], 12 Symphonische Dichtungen (Weimar, 1856).
14
Gesammelte Schriften (Leipzig, 1882), v, 231.
 Schoenberg as Performer of His Own Music 57

was not only tiresome but nonsensical.15 Wagner gave up the metronome
entirely after Tannhäuser in favor of verbal indications. Even the more
conservative Brahms indicated that the “metronome is of no value,” and
“as far at least as my experience goes, everybody has sooner or later with-
drawn his metronome marks”; he himself did so in his Ein deutsches
Requiem.16
This was the background against which Schoenberg formulated his
ideas concerning the metronome and, although he provided markings for
many of his works, he seems to have done so only reluctantly, warning in
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

the prefaces to many of them that “the metronomic indications are not to
be taken literally, but only as a suggestion.”17
Schoenberg’s recordings reveal that he frequently digressed from his
own “suggested” markings, sometimes to a considerable extent. Perhaps
in recognition of such disparities, he once wrote:

As an interpreter of his own works the composer sets a standard more

through the way his performance alters than for example through the
tempi he is supposed to have taken.18

A finalization of tempo appears to have given Schoenberg some diffi-

culty, and he frequently changed his mind as he proceeded from the
sketches19 to the published scores. In the end, the recordings show still
other tempi.
We can take Pierrot Lunaire as a case in point. The sketches most
often display a tempo designation (in German), followed by a note that
corresponds with the basic beat. The published score adds metronomic
numbers, which are sometimes adjusted in the rehearsal score; the record-
ings exhibit still other tempi (see Table 3). For number 1, published with
the indication = ca. 66, the rehearsal score shows “76” in red pencil,
o
whereas the recording proceeds at only 60. In his sketch for number 14,
Schoenberg initially stipulated nicht langsam, whereas the publication
reverts to langsame = ca. 56 and the recorded version is at 46. For num-
o
ber 16, the sketch shows sehr rasche , the published score ziemlich rasch
o
oca. 126, and the recording is taken at 104. Number 20 has leicht bewegte , c
15
Ibid.
16
Bernard D. Sherman, “Tempos and Proportions in Brahms: Period Evidence,” Early
Music 25 (1997), 463.
17
Point 7 in the prefaces to works published by Universal Edition (for example to the Suite,
op. 29, 1927). Subsequently this recommendation was reproduced by Schirmer when this com-
pany took over Schoenberg publications after World War II.
18
“Mechanical Musical Instruments,” in Style and Idea, 328.
19
Schoenberg’s sketches for the works considered in this paper are now in the possession of
the Arnold Schoenberg Center in Vienna.
 58 Roland Jackson

Table 3. Successive Tempi in Pierrot Lunaire (Sketches, Scores, Recordings)

no. Sketch Published score Rehearsal score Recording

1 bewegt ( = ca. 66)

oo –76 oo = 60
4 mässige o mässige ( = ca. 66) fliessend aber abwechslungs = 76
real ( = 60–92)
o
8 sehr langsams o gehende ( = ca. 88)
oo oo = 88
14 nicht langsam langsame ( = ca. 56) = 46
15 mässig bewegte o o mässig bewegte in abwechslungs-realen o = 56–60
( = 56–70) Bewegung
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

16 sehr rasche o o ziemlich rasch o = 104

( = ca. 126)
17 ( = ca. 120) 132 = 104–108
20 leicht bewegte c cc leicht bewegte
( = ca.126–132)
cc = 88–92 (100)

Table 4. Pierrot lunaire Published and Recorded Tempi

no. Tempo in the score Tempo in the recording

3. oo 152 oo 108
12. 120 88
13. oo 120 88
16.
17. cc 126
126 (132 in reh. score)
oco 104
104–108
18. 144 co 104
19. cc 120–132 94–114
20. 126–132 co 88–92 (100)
21. o 120 92

whereas the published score adds ca. 126–132, and the recording is real-
c
ized at = 88–92. In numbers 4 and 15, we find the expression “in quite
c
variable movement” (abwechslungsreal and in abwechslungsrealen
Bewegung, respectively) added to the rehearsal score, in recognition that
the piece calls for changing tempi throughout.
What is perhaps most unexpected about the recordings is how fre-
quently their tempi are slower than those indicated in the scores, often
markedly so. In Pierrot Lunaire, for example, several movements are
taken at 75 percent or less than the tempo called for in the published
score (see Table 4). Many of the tempi in Von Heute auf Morgen are even
more startling (see Table 5). Here the sections at mm. 254, 283, 483, 522,
938, and 991 are at half (or almost half) the published marking. The number
of incidences seems to preclude the possibility of misprints in the score
(e.g.,  instead of at m. 254).
o
 Schoenberg as Performer of His Own Music 59

Table 5. Von Heute auf Morgen, Published and Recorded Tempi

m. Tempo in the score Tempo in the recording

254  96 oo 96
283 o 126 69–72
289 wesentlich ruhiger co 80
294 o 126 112
301 Tempo 1 cc ca. 96–108
483 o 126 (wie Takt 283) 112 (slower than 283)
493 (theme = that of 294) 112 (about = to 294)
oo
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

522 Ruhig  100 100

834 Breit . 72
o . 88
oo
851 Nicht langsamer bloss ruhiger 88
o 70–75
940  84 84
oo
991 . 60
oo . 33
1041 152 100
o
Also to be noted in Table 5 are the following: mm. 289, where
wesentlich ruhiger (essentially calmer) evokes a tempo at half the speed
of the preceding; 301, where “Tempo I” is realized at about half the speed
of the preceding “Tempo I” (in m. 294); 834, where (exceptionally)
Schoenberg performs more rapidly than the published marking; 851,
which is slower than the preceding measure, despite the admonition “not
slower” (nicht langsamer) in the score; and 1041, performed at about
two-thirds the published marking.
Were Schoenberg’s conspicuously slower tempi a result of uncertain-
ties surrounding an initial performance? He seems to imply as much by
the following:

Nowadays I take everything in my works a basic degree quicker than

at the earliest performances, when, partly for technical reasons (diffi-
cult and inadequate dynamics), partly to obtain flexibility, I con-
sciously and unconsciously took everything much too slowly.20

This, however, was written in 1926, prior to any of his recorded perfor-
mances (which date from between 1927 and 1940). “At the earliest per-
formances” might be pertinent to the Suite, Op. 29, which presumably
was recorded at the time of the premiere in Paris, December 15, 1927. In
it, Schoenberg’s tempi are certainly slower than his markings. Most
remarkable is the second movement and variation 1 in the third move-
ment, both of which are realized at half (or less) of the designated
tempo. Were the metronomic values in the printed score meant to be an
20
“Mechanical Musical Instruments,” in Style and Idea, 326.
 60 Roland Jackson

Table 6. Suite, Op. 29, Recorded Tempi in Movements 1, 2, and 3

mvt. Tempo in the score Tempo in the recording

1. . = 72 . = 66

2.
(m. 68) coo= 132
= 80
cco = 92
= 72
3. o= c. 126 oo= 84–88
(var. 1)  = 104 = 104
(var. 2) = ca. 80 = 66–72
(var. 3) cco= 100 cco = 80
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

(var. 4) = 144 = 138

(return of th.) [ = 126]
o o= 76, then 96

eighth note (second movement) and a quarter note (third movement,

variation 1)? Table 6 shows the tempi of the first three movements as
marked in the printed score and as realized in Schoenberg’s recorded
performance.
Schoenberg’s quote is less pertinent to Pierrot Lunaire, however,
which was recorded considerably later (1940) than his earliest renditions
(between 1912 and 1924). In spite of its lateness, many of the movements
are unusually slow with respect to his own markings.
Among the most striking of Schoenberg’s changes is his according of a
fresh tempo to subsidiary themes within a movement, even though such
digressions are rarely indicated through metronome markings.21 The
fourth movement (Gigue) of the Suite, Op. 29, affords a notable example
(see Table 7). Here Schoenberg commences at 92 (a lessening from his
indication in the score), but with the appearance of differing thematic
ideas. In mm. 23 and 48, he diverges considerably from this opening
speed, in each instance adopting a pace he apparently deemed more
appropriate to the themes in question.
Comparing David Atherton’s recording of the Gigue, we are struck by
his quicker pace (one more in accord with Schoenberg’s marking of
“about 100, but preferably quicker”). At the same time Atherton, like
Schoenberg, also slows his tempo correspondingly at mm. 23 and 48 (see
Table 7).
Schoenberg, in the slowing of his tempi for subsidiary themes, once
again adheres to a late–Romantic tradition, one that seems to have

21
Concerning such unnotated changes, Schoenberg wrote in “Today’s Manner of Perform-
ing Classical Music” (Style and Idea, 320) that “suppressing all emotional qualities and all
unnotated changes of tempo and expression … came to Europe by way of America, where no
old culture regulated presentation, but where a certain frigidity of feeling reduced all musical
expression.“
 Schoenberg as Performer of His Own Music 61

Table 7. Suite, Op. 29 Tempi in the Fourth Movement (Gigue)

m. Tempo (score) Schoenberg recording Atherton recording

1 o . = ca. 100 (aber eher etwas rascher) oo . = 92 oo . = 112–116

23 . = 63 . = 84
38 oo . = 80–84
48 . = 48 oo. = 60
51 . = 72
64 Tempo I oo . = 80 . = 112–116
oo
84 (theme = m. 23) . = 58 . = 84
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

stemmed primarily from Wagner, who, according to Henry Smart in his

London concerts of 1855, “[reduced] the speed of an allegro—say in an
overture or the first movement—fully one third on the entrance of the
cantabile phrases.”22 In the Gigue, Schoenberg’s theme at m. 23 was simi-
larly lessened in speed by about a third, while that at m. 48 was even slower,
reduced by almost half the original tempo. The convention of performing
secondary themes more slowly—although not to this extreme—goes back to
the earlier nineteenth century. Hummel (in 1828) wrote, concerning the
“singing [“cantante”] passages in an allegro,” that they “should be delivered
with some yielding, in order to give them the necessary feeling.”23

SCHOENBERG AND EXPRESSIVE MARKINGS

The late nineteenth century saw a steady increase in the use of expressive
markings: articulative, accentual, or dynamic, of which the frequent inser-
tion of swells and diminuendos was especially typical.24 Such markings
went hand in hand with the rising importance of motivic ideas, as they
made these more individualized and more readily distinguishable from
one another.
Schoenberg continued in this propensity and seems to have added a
number of markings of his own, such as the combinative nuances shown

22
Henry Smart, Review, London Times (June 17, 1855).
23
Johann Nepomuk Hummel, Ausführliche theoretish-praktische Anweisung zum Piano-Forte-
Spiel (Vienna, 1828); A Complete Theoretical and Practical Course of Instruction on the Art of
Playing the Piano Forte (London, 1829), 417–418.
24
The frequency of markings seems to have reached a culmination in a composer such as
Reger, whose mature works show nearly every note or chord in a musical continuity to be
nuanced or dynamically gradated in some fashion. A further late-century manifestation can be
observed in “instructive” musical editions of earlier music (e.g., by Siegmund Lebert or Hans
von Bülow), which impose an excess of markings on earlier music, including even that of
Bach.
 62 Roland Jackson

in Figures 1b and c, which require considerable discrimination on the part

of a performer.25
The study of Schoenberg’s recordings allows us to observe more fully
his manner of interpreting his own special markings, as well as the various
other markings he drew upon. For instance, the straight line appears in
conjunction with the repeated notes in the second viola and second cello
parts in the beginning measures of Verklärte Nacht (see Example 3),
forming an ostinato background that allows these instruments to stand out
more pointedly within the polyphonic texture. Here, the recording has a
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

particular value in that it shows that Schoenberg, aside from following his
stated meaning—the lengthening of a note—also introduces a slight begin-
ning accent followed by a rapid decrescendo toward the end of each of
the notes.
Another marking Schoenberg sometimes drew upon was the short swell
and diminuendo (< >) on individual notes, as in the English horn solo at
the beginning of “Lied der Waldtaube” (Gurrelieder), where four of the
notes are so marked (see Example 4). During the nineteenth century, the

Figure 1. Some detailed markings used by Schoenberg.

a. – the note should be lengthened (tenuto and portato)
b. '– the note should be accented and lengthened
c. –. the note should be well held-out (ist gut aushalten), but nonetheless
separated from the note following by a small pause or interruption.

Example 3. Schoenberg, Verklärte Nacht, second viola and second cello,

m. 1.

25
Among a list of nuances Schoenberg provided for his works published by Universal and
Schirmer (aside from the three shown here) are the following, each with its own particular sym-
bol: “accented like a strong beat,” “unaccented like a weak beat,” “hard, heavy, martelé (for
short notes),” “light, elastic, thrown,” and “not to be weakened, and often even to be brought
out (mainly on upbeats).”
 Schoenberg as Performer of His Own Music 63

Example 4. Schoenberg, “Lied der Waldtaube” (Gurrelieder), mm. 1–3.

short swell and diminuendo had become a common nuance; inherited

from the earlier messa di voce, it acquired an accentual significance, espe-
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

cially in string playing, where it also sometimes had the meaning of a

momentary vibrato. In Schoenberg’s recorded interpretation, the sign was
realized simply as a quick decrescendo. Here again the ritard (sehr
zurückhaltend) is extended back into the preceding measure, as was typi-
cal in Schoenberg’s performances.
In his rehearsal copy of this work, Schoenberg also displays a concern
for a clear separation between successive musical ideas by marking in
with red pencil a curving line following the English horn’s melody, thereby
setting it apart from the onset of the new motive beginning in triplet
thirty-second notes in the piccolo and clarinet in E
.
Rather unusual is Schoenberg’s introduction of the sign < > in connection
with a single chord played on the piano in m. 34 of the “Valse de Chopin” of
Pierrot Lunaire (see Example 5), because in this instance, the sign could not
have been associated with any swelling or diminuendo. On the recording,
Schoenberg appears to have interpreted it simply as a slight accentuation.
This example, limited to the piano,26 brings to the fore Schoenberg’s pre-
dilection for highly detailed markings. Each successive chord in the piano
part is differentiated both dynamically (moving through pp, swell, sf, and
pp) and articulatively (proceeding through sustained, staccato, staccato,

Example 5. Schoenberg, Pierrot lunaire, “Valse de Chopin,” mm. 33–34

(piano part only).

26
In the movement “Valse de Chopin,” the voice’s Sprechstimme is also accompanied by the
flute and bass clarinet (or clarinet).
 64 Roland Jackson

elongated, and presumably accented chords). On the basis of the recording,

it seems evident that the composer wanted each of these nuances to be heard,
and for the parts to be balanced against one another. An equality between the
parts and the formation of a contrapuntal whole was Schoenberg’s apparent
ideal, something he expressed in his writings, two of which follow:

The highest principle for all reproduction of music would have to be …

that every note is really heard, and that all sounds … stand out clearly
from one another.27
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

[F]or an interpretation to be in keeping with our technique [that of

contemporary rather than of baroque music] one must expect to hear
all the parts with equal prominence.28

As Smith points out, Schoenberg’s “total concentration and involvement

with even the smallest details caused him difficulties in rehearsing.”29 This is
borne out in the recording that preserves Schoenberg’s conductorial direc-
tions during part of a rehearsal of Kol Nidre in Los Angeles in 1938.30

Oh yes. This I would like to hear, trombones and bass tub … and the
tuba. Please, in 99, the first … the second and third beats are legato, and
should be [inaudible] … let us say quasi dolce. Yes? Not too loud. But
then comes two very short staccatos. Don’t make them too long please.
Yes? You accompany this sung [inaudible] melody. Yes? And it’s too
quick to the melody. And the staccato does not mean characteristic. See
… Listen please! It serves only so that one hears better the singer. You
know? And it needs very short notes, but not accented: bump, bump.
Yes? So then please now play 99 and 100.

“Legato … quasi dolce … very short staccatos … so that one hears better
the singers”—such references reveal a musician intent on achieving the
most minute of differences and on attaining just the right balance between
the instruments and singers, while at the same time keeping the lower
brass distinctive through their articulations.
Schoenberg’s conducting scores, or Handexemplare, also available in
the Center, form a valuable adjunct to the study of his performances. They

27
“Essay on Performance,” reproduced in Style and Idea, 319.
28
“Mechanical Musical Instruments,” reproduced in Style and Idea, 327.
29
Smith, Schoenberg and His Circle, 115.
30
Cited in The Schoenberg Era: Schoenberg the Man, an unpublished transcript from a radio
series, originally broadcast April 20, 1981, and distributed by the Public Broadcasting Associa-
tion, p. 36.
 Schoenberg as Performer of His Own Music 65

Pierrot Lunaire
11. (m. 1) cresc. is crossed out
19. (m.22) p < is added in the cello
(m. 40-41) ff is added in the cello (between hairpins)
Verklärte Nacht
(m. 138) fpp is substituted for fp
(m. 188) ff is substituted for f
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

Suite, Op. 29
2nd Mvt. (m. 90) p is added in the viola
(m. 99) f is added in the viola and cello
Figure 2. Some dynamic changes in Schoenberg’s rehearsal scores.

contain various additions (in colored pencil) entered during or prior to

rehearsals, offering a rare glimpse into the last-minute concerns of a com-
poser preparing for a performance. In them his most typical consideration
was the attaining of just the right dynamic balance between the parts, a
few instances of which are cited in Figure 2.31

SCHOENBERG AND SPRECHSTIMME

Schoenberg’s detailed description of Sprechstimme in Pierrot Lunaire

apparently goes beyond that of any previous user of the device.32 In the
Vorwort to the published version (Universal Edition), he sets forth the fol-
lowing key points:

The melody given in the speaking voice through notes … is not meant
to be sung… The sung tone maintains unchangeably the pitch; the
speaking tone indeed approximates it, but then abandons it immedi-
ately by falling or climbing. 33

31
For further information concerning these scores, see Jeremy McBride, “Schoenberg’s
Annotated Handexemplare,” Journal of the Arnold Schoenberg Institute 5 (1981), 183–202.
32
The earliest examples of the device, most notably in Humperdinck, are described in some
detail by Sharon Mabry in Vocal Problems in the Performance of Pierrot Lunaire, Op. 21 by
Arnold Schoenberg (DMA paper, Peabody College, 1977).
33
“Die in der Sprechstimme durch Noten angegebene Melodie ist . . . nicht zum Singen
bestimmt ... der Sprechton gibt [die Tonhöhe] zwar an, verläβt sie aber durch Fallen oder
Steigen sofort wieder.”
 66 Roland Jackson

The principal idea here is that the “spoken part” should approxi-
mate speaking rather than singing. In this light, the story of Alma
Mahler takes on added significance. She recounted that Schoenberg
once heard Pierrot Lunaire “sung” by Marya Freund (directed by
Milhaud) and scarcely recognized the result as his own work.34 Her
singing was obviously contrary to his directives, which call for a form
of heighted speech, wherein the notated pitches are to be approximated,
then immediately slid away from, upward or downward, as happens in
actual speech. In Schoenberg’s recording, the Sprechstimme is presented
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

by Erika Stiedry-Wagner, who follows the contour of the notes without

matching the pitches. Only on the lengthier syllables, however, is the slid-
ing at all apparent, and it is almost invariably downward. Most of the notes
are clipped short, lending the recitation its supernormal effect, in which the
voice part is markedly set apart from the more normal sounds of the
accompanying instruments.35
Schoenberg’s approach may be elucidated by contrasting it with a pri-
marily “sung” performance, as heard in a recent recording by Phyliss
Bryn-Julson made in Frankfurt in 1991.36 In an excerpt from No. 2 (“Col-
umbine”), mm. 17–26 (see Example 6), we observe Stiedry-Wagner fol-
lowing the general shape of the notated melody, with some exceptions (on
“zu lindern,” for instance, she ascends rather than descends), although
none of the given pitches are replicated. Downward slides give an
uncanny emphasis to the lengthier syllables—Stro-me, wei-βen, and wun-
der-ro-sen—while the remaining syllables are too brief to permit any
noticeable sliding.
Bryn-Julson’s approach could not be more different. Hers is a “sung”
realization, apparently more akin to Marya Freund’s (mentioned above).
Bryn-Julson reproduces the given notes throughout and almost entirely
avoids any slides away from the pitches. Only on one syllable, wei-(βen,
is a downward slide introduced, and this is more in the nature of a

34
As reported in Smith, Schoenberg and His Circle, 88.
35
The gliding between pitches in Sprechstimme is suggestive of the portamento, a vocal
technique prominent during the late nineteenth and early twentieth centuries, but the two are
essentially different. In portamento, the gliding points up the pitch arrived at, whereas in
Sprechstimme it emphasizes the pitch (or approximated pitch) being left and moves to no par-
ticular point of termination. Schoenberg seems to have had a general antipathy toward the por-
tamento. In a marginal note added to the Genesis Suite (1945), he made it clear that—at least in
this work—portamento was to be avoided altogether: “always without Hollywood style of
vibrato and portamento, even large intervals must not be connected by gliding, but if necessary
by stretching. This gliding is of a detestable sentimentality.” It is of note that Schoenberg’s own
recordings contain no evidences of portamento.
36
On RCA Victor with Peter Eötvös conducting the Ensemble Modern, Opera House,
Frankfurt, December 9–15, 1991. Recorded on RCA Victor Red Seal 09026 61179-2, digital
stereo, 1993, CD.
 Schoenberg as Performer of His Own Music 67
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

Example 6. Schoenberg, Pierrot lunaire, "Columbine," mm. 17–26, real-

izations of Sprechstimme by Stiedry-Wagner (indicated above the staff) and
Bryn-Julson (indicated below the staff)

pronounced portamento proceeding from a’ directly down to c#’ (neither

singer pays heed to the f#). Bryn-Julson suggests a speech-like quality by
adopting semi-breathless, nonsustaining tones.
The emotional impact of these two performances is greatly affected by
the singers’ interpretation of Sprechstimme. Aside from Stiedry-Wagner’s
closer adherence to Schoenberg’s idea of the technique, the singer also
conveys—moreso than does Bryn-Julson—a quality of the macabre or
grotesque well suited to the sense of Albert Giraud’s poetry, on which this
song cycle is based.

SCHOENBERG AND THE PERFORMER

Schoenberg felt that a performer should respect the composer’s idea about
a musical work in preparing a realization of it: “Interpreters rights, are
there not also author’s rights? Does not the author, too, have a claim to
 68 Roland Jackson

make clear his opinion about the realization of his work[?]”37 At the same
time, he also embraced the late–Romantic attitude of according consider-
able leeway to the performer. Going beyond other writers of the early
twentieth century, Schoenberg spelled out and made quite explicit the
performer’s—as opposed to the composer’s—contribution:

For the true product of the mind—the musical idea, the unalterable—
is established in the relationship between pitches and time divisions.
But all other things—dynamics, tempo, timbre, and the character,
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

clarity, effect, etc., which they produce—are really no more than the
performer’s resources, serving to make the idea comprehensible and
admitting of variations.38

This indicates what in Schoenberg’s mind constituted the true divi-

sion between the composer and the performer—that is, a distinction
between the musical substance as formulated by the composer (the
pitches and time divisions) and the performer’s freedom to vary this sub-
stance in his or her presentation of the work. To the composer belongs
“the true product of the mind,” to the performer a wide range of possibil-
ities for the enhancing of this product of mind, including the dynamics,
tempo, and timbre, as well as (more abstractly) the character, clarity, and
effect. Such a difference quite possibly explains why Schoenberg indi-
cated that in his own performances—undoubtedly including those of
the present recordings—there could be no finalized version, that his or
any other composer’s performance “can by no means remain the finally
valid one.”

insofar as the mechanization of music [through recordings] … states as

its main aim the establishment by composers of a definitive interpreta-
tion, I should see so advantage in it, but rather, loss, since the com-
poser’s interpretation can by no means remain the finally valid one.39

It is unfortunate that we have only one version of each of Schoenberg’s

recorded works. Had he left more than one, how much more we would
know about the kinds of freedoms he would have condoned or favored.
What seems most likely, though, is that he would not have advocated that
other performers slavishly duplicate his own interpretations.
Schoenberg’s recordings remain of inestimable value. They do not
require imitation of their particulars, but they do provide guidelines of a

37
“About Metronome Markings (1926),” cited in Style and Idea, 342.
38
Mechanical Musical Instruments,” in Style and Idea, 326.
39
“Mechanical Musical Instruments,” in Style and Idea, 328.
 Schoenberg as Performer of His Own Music 69

general sort as to what was most meaningful to the composer: the momentary
slowings and hastenings, the taking of themes at differing tempi accord-
ing to their character, and the adding of nuances to individual notes,
thereby individualizing more markedly the musical ideas. Although
Schoenberg’s approach is rather more exaggerated or “Romantic” than
modern taste might readily accept (witness the more staid or regular
recordings by Craft, Atherton, and others), it seems nonetheless emi-
nently worthwhile to resurrect his early-century manner in the interest of
remaining faithful to his vision. The recordings provide us an opportunity
Downloaded by [University of Western Ontario] at 16:56 12 November 2014

to enter into Schoenberg’s time, an era more intense and more extreme in
many ways than the one to which we have grown accustomed. More
importantly, they offer us a model that can be followed, if not in the letter
at least in the spirit.
Downloaded by [University of Western Ontario] at 16:56 12 November 2014