Volume 12, Number 1, February 2006
Copyright � 2006 Society for Music Theory

Avior Byron*

The Test Pressings of Schoenberg Conducting Pierrot lunaire: Sprechstimme Reconsidered

Get the free RealPlayer to hear the audio examples.

Get the free DjVu browser plug-in to best view the examples.
Get DjVu Plug-in

Comment on this article

KEYWORDS: Sprechstimme, Schoenberg, performance, notation, recording, test pressings, Pierrot lunaire, analysis, singing, score, recitation, pitch contour, Stiedry-Wagner, Eine blasse Wäscherin, interaction, reproduction, real-time, deviation

ABSTRACT: Newly discovered recordings of Schoenberg conducting Pierrot lunaire open a window into the workshop of Arnold Schoenberg (the conductor) and Erika Stiedry-Wagner (who performed the Sprechstimme). These recordings reveal that in a period of not more than three days, Schoenberg accepted relatively great freedom in the Sprechstimme pitch contour; as well as a contradictory tendency towards consistency and a certain systematic approach towards pitch, which does not always adhere to the score. Before examining the recordings it was not possible to know whether the relation between the performed Sprechstimme and the score was controlled, systematic, or simply a matter of chance. The recordings shed new light on what has been described by Boulez, Stadlen and others as the "Sprechstimme enigma": namely, how Schoenberg expected the Sprechstimme to be performed. The history of Schoenberg's writings on Sprechstimme demonstrates that his perception of it changed along with the development of his performance aesthetics in general. Based on evidence from the recordings as well as on recent performance studies theory, I will claim that the Sprechstimme enigma is greatly clarified when one understands that there are simultaneously two types of notation in Pierrot lunaire: one for the instruments that tends towards a reproduction of a sound object, and another for the Sprechstimme which involves a process of greater real-time interaction between performer and score. Although the Sprechstimme from the workshop of Schoenberg and Stiedry-Wagner may be regarded as an extreme case study, it magnifies in a way what also happens in performances of other types of music.

REFERENCE: MTO 13.2: Byron and Pasdzierny, “Sprechstimme Reconsidered Once Again: ‘... though Mrs. Stiedry is never in pitch’”

Received January 2006


[1.1] In his 2005 article "Schoenberg as performer of his own works" Ronald Jackson speculated that 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 condemned or favored."(1) Indeed the well documented web-site of Wayne Shoaf does not mention more than one recording of the very few that exist with Schoenberg conducting; the Arnold Schönberg Center web site has some further information but here too one can find only a single recording of each composition. Recently, I discovered several further recordings of Schoenberg conducting Pierrot lunaire. The first is a recording of a broadcast from 17 November 1940 in New York. The second is the test pressings of the 1940 commercial recording. The broadcast does not appear in any catalogue. The test pressings appear in a printed catalogue from 1986, yet they lay dormant for 65 years in Schoenberg's Nachlaß since only now they were transferred in a very delicate and expensive process from LPs to CDs.(2) In this article I will focus on the test pressings.(3)

[1.2] Pierrot lunaire is considered by many as Schoenberg's most famous masterpiece. For a long time the 1940 recording of Schoenberg conducting the piece was the only commercial recording available of his mature music.(4) Referring to it in a letter from 30 September 1940 to Moses Smith, Schoenberg claimed to be "very happy about the records." The test pressings for the recording were made between 24 and 26 September 1940 in Los Angeles, in which the following musicians participated: Erika Stiedry-Wagner, recitation; Rudolf Kolisch, violin and viola; Stefan Auber, cello; Edward Steuermann, piano; Leonard Posella, flute and piccolo; and Kalman Bloch, clarinet and bass clarinet.(5)

[1.3] After the different test pressings were recorded they were given to Schoenberg to choose the best ones for the published recording. Dika Newlin, a student of Schoenberg, wrote on 2 October 1940 that Schoenberg had two or three test pressings of each side of the records. He wanted some "outsiders who were not quite as familiar with the music" to help him choose the best test pressing of each song. She reported that "Estep, Stein [other students of Schoenberg] and I were the only outside people there� Otherwise, there were performers and family: Kolisch with wife, Auber, Posella, Khuner, and Mrs. Seligmann." Newlin claimed: "I vigorously participated in selecting the test pressings, a rather trickish job, since in many cases the differences were slight yet important."(6)

[1.4] The test pressings include 16 records recorded on one side at 78 speed. These are not simply different performances. If up to now we had only a single picture of a historical event of Schoenberg conducting Schoenberg, now we have several pictures, or as it were--a short film of the very same occasion. This grants a rare opportunity to enter the workshop of the artist and observe not only the final product, but also the process of creation. It also provides a new perspective on the degree of stability and change of this historical interpretation..

[1.5] There are 34 test pressings (see Table 1). The songs "Eine blasse Wäscherin," "Valse de Chopin," and "Madonna" were recorded four times; "Gebet an Pierrot," "Raub," "Rote Messe" and "Galgenlied" were recorded three times. "Mondestrunken," "Colombine," "Der Dandy," "Enthauptung," and "Die Kreuze" were recorded twice; all the remaining nine songs were recorded once (not included in Table 1).

[1.6] One of the curiosities of Op. 21 is the Sprechstimme; and it is especially interesting when examining the singing of Erika Stiedry-Wagner in the recording with Schoenberg conducting. In this study I will focus mainly on Sprechstimme in the song "Eine blasse Wäscherin."(7) Many commentators have noted that Stiedry-Wagner sings with an inaccurate pitch.(8) In Western art music, singing in a pitch that is inaccurate to some extent is not uncommon, yet in the performance of Stiedry-Wagner this is done in an explicit and exceptional manner. I will examine not only how accurately Stiedry-Wagner sings the pitches in "Eine blasse Wäscherin," but also whether or not deviations are consistent in the four test pressings. If consistency may be found, it is interesting to consider its extent and whether it has greater implications on the understanding of how Schoenberg expected the Sprechstimme to be performed, and how Stiedry-Wagner built her performance? Finally, I suggest that a reconsideration of Sprechstimme in light of the test pressings has larger implications for understanding the role of the performer in relation to the score of many other musical works that were influenced by Pierrot lunaire or which influenced its creation, as well as of other works in Western art music.


[2.1] David Hamilton has claimed that Schoenberg's Sprechstimme "has been enormously influential in breaking the virtual stranglehold the bel-canto model of vocal production had maintained on Western Music� the vocal writing of Boulez, Crumb, Berio, and Maxwell Davis, for example, is unthinkable without Pierrot's rupture of that fruitful but necessarily limiting monopoly."(9) The Sprechstimme technique used by the speaker in Pierrot lunaire, as well as in other works by Schoenberg, has been described as posing "an enduring and perhaps insoluble interpretive enigma for the performer."(10) Both Darius Milhaud and Pierre Boulez, who conducted the piece, described it as creating "insoluble problems."(11) Boulez claimed that "although we possess authentic documentation on the subject, it is still difficult to form any precise idea of Sprechgesang."(12) Lorraine Gorrell, who has performed the speaker's role in Pierrot lunaire, reported the puzzling situation she found herself in when she approached the work. When listening to different recordings she discovered that "one performer sang the voice part, duplicating the pitches indicated in the score, while several other singers spoke the part but only vaguely approximated the indicated pitches."(13) The recording of Pierrot lunaire made by Schoenberg also did not solve this problem since Stiedry-Wagner "did not match pitches or observe the indicated intervallic relationships. In fact, she sometimes did not even follow the direction of the indicated pitches. I began to wonder why Schoenberg had given pitches at all."(14) A sense of the difficulty of performing Sprechstimme can be gained from the following report. Stiedry-Wagner said that Schoenberg and other members of the Society for Private Musical Performance

wanted first Gutheil-Schoder to [sing Pierrot lunaire]� But it is a thing you have to talk, to speak, not to sing. And so she found it � too difficult, and then Alban said, "Why don't you ask Erika Wagner? She's an actress and she's a singer," because I also sang. I had Liederabende and I sang operettas. And so it happened that I did Pierrot. I studied it with Erwin Stein� I was working very hard, oh very hard. Because you know it's very difficult to speak in rhythm--strong rhythm. And then the Sprechmelodie, you know� He wants a certain line to speak, with low tones and with high tones, and it is very difficult to keep the tone if you have to speak a word through one whole measure--a long time. It's very difficult to keep it without singing� you need a very, very big Skala--deep and high and you have to be a Sprecher--you have to know how to speak, not how to sing, and that's the main thing. And it's very wrong--Schoenberg always told me it's wrong--to sing� And this is really very difficult. So it was a Spezialität--a specialty for me.(15)

In several recordings--for example, one by Pierre Boulez and Yvonne Minton from 1977--one can hear a precise production of the Sprechstimme pitches indicated in the score, which Minton sings.(16) A performance decision which closely observes the pitch is not without sense since many of the songs contain significant pitch relationships between the Sprechstimme and the instruments. For example, in "Parodie" there are canons between voice and viola and between voice and piccolo. Yet "failing" to observe the indicated pitch is not only the practice of Stiedry-Wagner but also that of many other speakers.(17)

[2.2] Further problems were mentioned by Boulez when he wrote that there are "people the tessitura of whose singing voice is wider and higher than that of their speaking voice," while with others the opposite may occur. He concludes that Pierrot lunaire "is thus both too high and too low."(18) Also he pointed to the fact the when one speaks, the duration of the sound is usually short.(19)

[2.3] There have been several attempts to solve the Sprechstimme enigma. Peter Stadlen's article "Schoenberg's Speech-Song" from 1981 is a classic in the literature on Sprechstimme.(20) Yet this article has several weaknesses. First, it contains an implicit assumption that the performance of Sprechstimme should be identical or at least similar in all of Schoenberg's compositions. Yet Schoenberg imagined different kinds of Sprechstimme in different periods and for different compositions; for example, in Pierrot lunaire he wanted the adherence to pitch to be greater than that in the Gurrelieder.(21) Stadlen's article reviews what it sees as Schoenberg's contradictory attitudes to Sprechstimme, with no relation to the contexts of his different performance aesthetics. Demonstrating how Schoenberg's writings on Sprechstimme can be more fully understood in the light of his changing performance aesthetics is beyond the scope of this article, yet I will point to significant moments of change. Most importantly, I claim that this so-called contradiction appears only if one understands the role of the singer as that of reproducing the score or a sound object. My study will reveal that it is not that Schoenberg simply tolerated Sprechstimme performances that were not faithful to the score, but that he did not expect an exact reproduction of a sound object (at least with regard of the pitch parameter) from Stiedry-Wagner. We will see that Schoenberg's acceptance of Stiedry-Wagner's "out-of-tune" Sprechstimme was not due to limited musicality (she was after all more an actress than a singer), but something that was part of the conception of the piece. At the end of this article I will offer an alternative view to the role of the singer in Pierrot lunaire which much clarifies this so-called contradiction. In order to explain the contradiction that Stadlen wrote about I will further review the history of Schoenberg's Sprechstimme.

[2.4] The origin of this technique has been traced to Engelbert Humperdinck in his 1897 melodrama Königskinder, as well as to the "old" Austrian theater speaking, yet Schoenberg's Sprechstimme is a fresh and new conception.(22) Richard Kurth argues that "Schoenberg's Sprechstimme is a representation of speech and a substitute for both speech and song. It emphasizes its own peculiarity and disorients the listener's customary response to words' sound and meaning."(23) Schoenberg was influenced by Albertine Zehme's original conception of Pierrot lunaire, and it is not unlikely that he was also influenced by her conception of recitation.(24) Zehme, who was deeply interested in the artistic and mystical sides of recitation, was the author of a treatise on the subject named Die Grundlagen des künstlerischen Sprechens und Singens from 1920.(25) In a note entitled "Why I must speak these songs" which she attached to one of her March 1911 Berlin performances, she wrote:

        The words that we speak should not solely lead to mental concepts, but instead their sound should allow us to partake of their inner experience. To make this possible we must have an unconstrained freedom of tone. None of the thousand vibrations should be denied to the expression of feeling. I demand tonal freedom, not thoughts!
        The singing voice, that supernatural, chastely controlled instrument, ideally beautiful precisely in its ascetic lack of freedom, is not suited to strong eruptions of feeling--since even one strong breath of air can spoil its incomparable beauty.    
        Life cannot be exhausted by the beautiful sound alone. The deepest final happiness, the deepest final sorrow dies away unheard, as a silent scream within our breast, which threatens to fly apart or to erupt like a stream of molten lava from our lips. For the expression of these final things it seems to me almost cruel to expect the singing voice to do such a labor, from which it must go fourth frayed, splintered, and tattered.
        For our poets and composers to communicate, we need both the tones of song as well as those of speech. My unceasing striving in search of the ultimate expressive capabilities for the "artistic experience of tone" has taught me this fact.(26)

This "unconstrained" expressive conception of the role of the voice, which expresses something that is more than "the beautiful sound alone," will be one of the keys to understanding Schoenberg's puzzling concept of Sprechstimme.

[2.5] On a page from a working autograph of Die glückliche Hand, Op. 18 from September 1910 Schoenberg explained that the notes with crossed stems "must be spoken at exactly the prescribed time and sustained as indicated; the pitch is to be realized approximately through speech."(27) The first (complete) manuscript of Pierrot lunaire had no preface, yet on the page of "Gebet an Pierrot" (which was the first song to be composed in the cycle) from about March 1912 Schoenberg wrote: "The recitation should hint at the pitch."(28) In a letter to Berg from 14 January 1913 Schoenberg wrote: "Regarding the melodramas in the Gurrelieder: the pitch notation is certainly not to be taken as seriously as in the Pierrot melodramas. The result here should by no means be such a songlike Sprechmelodie as in the latter� [There is] no [need to keep the] � interval proportions!."(29) The idea of keeping the interval proportions (and not the exact notated pitch) in Pierrot lunaire is manifested in an early version of the famous preface to the piece. It can be found in a collection of manuscripts which were written between March 1912 and January 1914.(30) Here we find a different version of the text: "it is the duty of the performer to perform the rhythm absolutely precisely, and to transform the notated melody into a Sprechmelodie by always keeping the relationship(31) between the pitches."(32) The same concept is also expressed several years later, on 13 January 1921, in a letter Erwin Stein wrote to Schoenberg: 'It is really unbelievable, how clear the expression, and also its intensity, is fixed with the [notated] speech intervals. When one speaks what is [notated] there, one understands the expression, even if one did not feel it at first. Although the register differences [Lagenunterschiede] and the proportions of the intervals are very important'.(33) This letter reveals that even members of Schoenberg�s circle were not completely sure about the performative status of the notated pitch. If it were clear that the 1914 preface demands an exact reproduction of notated pitch, then Stein would not seek the confirmation of 'the master' concerning the superiority of interval proportions in relation to exact pitch. On 8 July 1923 Schoenberg himself confirmed this when he wrote to Josef Rufer:

The pitches in Pierrot depend on the range of the voice. You have to consider them 'well' but not 'strictly'. You can divide the range of the voice in as many parts as half tones are used; perhaps then every distance is just a 3/4 tone. But you don't have to carry this out in a pedantic way, because the pitches do not realize harmonical proportions. Of course the range of the speaking voice is not enough. Well, the lady has to learn to speak with 'head voice'; every voice can do that� The most important thing is to get the 'Sprechmelodie'.(34)

The following suggestion by Erwin Stein, which appeared in an article in the journal Pult und Taktstock in 1927, seems to correspond to the indication in the early preface (and not the one that appeared in the published score) and these letters:

Though shown in absolute pitch notation, the intervals are only meant to be relative. The initial note is so short that it is of no harmonic consequence. The reciter is therefore free not only to transpose his part according to the type of his speaking voice and regardless of the other instruments, but also to narrow down the compass and tessitura� What is essential is that the proportions of the melodic line be retained: a high note has to be relatively high, a low note relatively low; a fourth must be a wider leap than a third, and a minor second a smaller step than a major second.(35)

On 13 May 1927 Schoenberg wrote to Stein from Berlin that this article was 'an excellent article, full of clarity, intelligence and understanding.'(36)

[2.6] Yet a very different conception of Sprechstimme can be found in the 1914 preface that Schoenberg wrote to the first printed edition of Pierrot lunaire:

The melody indicated by notes in the part of the speaker (with certain specially indicated exceptions) is not intended to be sung. The performer has the task of transforming it into a speech melody by taking well into consideration the indicated pitches. He can do this by
        I. keeping to the rhythm just as precisely as he would when singing, i.e., with no more freedom than he would take in a sung melody;
        II. being quite conscious of the difference between a sung tone and a spoken tone: the sung tone maintains its pitch without change, the spoken tone touches upon it but then leaves it immediately by descending or ascending. The performer must always be on guard against falling into a "singing" manner of speech. That is absolutely not intended. But neither should he aim for a realistic-natural speech. Quite the opposite, there should always be a clear difference between customary speech and speech that contributes to a musical effect. But this should never remind one of song.(37)

The demand for "taking well into consideration the indicated pitches" seems to contradict the mentioned above conceptions of approximation and suggestiveness of the Sprechstimme in relation to the notated pitch. Stadlen suggests there was "a conflict, from the very beginning, in Schoenberg's mind between a desire for speech character and another, seemingly incompatible desire for an exact rendering of the notes."(38) However, this is only a restatement of the problem and not its solution.

[2.7] There is something strange in Schoenberg's instructions in this version of the preface: how can the "indicated pitch" be communicated if one must immediately deviate from it? Schoenberg's demand for the speaker to perform her part in a manner which is neither singing nor normal speaking leaves the performer in a state where clear performance instructions are not to be found. This places considerable responsibility on the performer--a responsibility that may lead to unpredictable results. Indeed, the many recordings of this piece demonstrate the diversity of the interpretations of the Sprechstimme part that have been made up to now.(39) Richard Kurth writes: "At once both speech and song, Sprechstimme is also neither� Its exact rhythms, its sometimes purposefully uncoordinated placement of stress, and its exaggerated contours lend Sprechstimme an unnatural air which is modern in its artificiality, and which recalls the strange poses of the miming Pierrot� A stimulated, exaggerated form of recitation, it distorts the sounds of normal spoken language and alienates them; but at the same time, it also augments the mimetic potential of words by (re)presenting their sounds and rhythms in an altered (and heightened) form."(40) Perhaps this is also one of the reasons that Sprechstimme has invited not only a large variety of performative interpretations, but also very different responses by listeners.(41)

[2.8] On 16 August 1922, Schoenberg wrote to the singer Marya Freund concerning Pierrot lunaire: ��I am anxious to explain to you why I cannot allow any will but mine to prevail in realizing [dargestellten] the musical thoughts [musikalischen Gedanken] that I have recorded on paper, and why realizing them must be done in such deadly earnest, with such inexorable severity, because the composing was done just that way."(42) The reason for this severity, as well as for the relatively great silence by Schoenberg during the 1920s on the issue of Sprechstimme, might be that he was too influenced by and aware of the anti-interpretation movements of the 1920s, such as the Neue Sachlichkeit, to correspond on a matter which grants the performer such great freedom from the score. Several of Schoenberg's writings contain complaints about performers in the 1920s who interpreted the score too freely;(43) the concept of Sprechstimme as it appears in Pierrot lunaire demands respect for the score on the one hand, while also requiring a freedom of interpretation on the part of the speaker, something which to Stadlen seemed contradictory. Yet what exactly needs to be realized here? What exactly are these musical thoughts? I believe that it is clear neither from this letter nor from other evidence that reproduction of pitch is the main issue (note that Schoenberg had put emphasis here on the severity of the process of composing).

[2.9] Christian Schmidt quotes the following passage from Schoenberg's instructions in Moses und Aron: "The musician often cannot refrain also from melodically notating these mere speech figures. But also these are not to be sung. Evidence: they are beyond the 12 tones! But perhaps the singer should extract out of the [melodic] line what expression to conceive."(44) Also Edward Steuermann, Schoenberg's disciple and the pianist who performed Pierrot lunaire since its beginning, claimed: "It seems to me one has to find the expression of each sentence that will cover an entire line."(45) Schmidt asks whether the Sprechstimme notated is "music-for-the-eyes" (Augen-musik) and if it was notated only because the composer could not release himself from the responsibility of relating to harmonic, counterpoint, and motivic-thematic ideas?(46) Schmidt does not answer his question but rather claims: "Doubt is advisable, and will arguably also remain."(47)

[2.10] A different attitude than in the 1920s can be found in the last decade or so of Schoenberg's life. On 31 August 1940 he wrote to the Stiedrys: "We must ... refresh the Sprechstimme thoroughly--at least, because I aim this time to try to bring out this light, ironical-satiric tone ... in which this composition was originally conceived. Furthermore, times and aesthetic concepts are changing, so what at the time seemed to us Wagnerian, or in the worst case Tchaikovskian, today is, for example, Puccini, Lehar and so on."(48) Notwithstanding the ironic tone of the second sentence, one should not discount Schoenberg's intention to create a new kind of Sprechstimme.

[2.11] In 1949 Schoenberg wrote in "This is my Fault" as follows:

In the preface to Pierrot lunaire I had demanded that performers ought not to add illustrations and moods of their own derived from the text. In the epoch after the First World War, it was customary for composers to surpass me radically, even if they did not like my music. Thus when I had asked not to add external expression and illustration, they understood that expression and illustration were out, and that there should be no relation whatsoever to the text. There were now composed songs, ballets, operas and oratorios in which the achievement of the composer consisted in a strict aversion against all that his text presented.(49)

Indeed we will see that in the recording from 1940 Schoenberg allowed Stiedry-Wagner to make unnotated changes in her Sprechstimme that have a direct correlation to the mood and character of the text. In other words, the indication from the preface of Pierrot lunaire should be understood in its historical context and not necessarily as the definitive, or most significant, wish of the composer.

[2.12] In a letter to the conductor Hans Rosbaud from 15 February 1949 Schoenberg claimed that the speaker in Pierrot lunaire "never sings the theme, but, at most, speaks against it, while the themes (and everything else of musical importance) happen in the instruments."(50) Likewise, in a letter to Daniel Ruyneman from 29 July 1949 concerning Pierrot lunaire, Schoenberg wrote that "none of these poems is determined to be sung, but rather they must be spoken without fixed pitch."(51) In these two cases Schoenberg emphasized the speaking side of Sprechstimme. Such an emphasis may have resulted from performers he had heard who simply sang the part. However negative performance experiences may not have been the only issue here at hand. Schoenberg's performance aesthetics of that time did not advocate interpretations that negate the performer's capacity to express him- or herself in ways that deviate from the strict indications of the score.(52) Sprechstimme in Pierrot lunaire has an in-built demand for interpretation by the performer; and when this is denied by performers, Schoenberg saw it as a misinterpretation of his music.

[2.13] On 2 January 1951 Schoenberg wrote to the Stiedrys that in contrast with Pierrot lunaire, the Sprechstimme in the melodrama of the Gurrelieder should relate "in no manner [to] pitches." Schoenberg stressed: "I believe that I have written them only in order to represent my phrasing of the notes, the accentuation and the recitation more urgently. Therefore please no speech-melodies."(53) This demonstrates that Schoenberg used the conventional notation of pitches in a non-conventional manner. Although he expected greater fidelity to pitch in the Sprechstimme of Pierrot lunaire than in that of the Gurrelieder, this does not mean that in Pierrot lunaire the performer should simply reproduce pitch. The fact that one can find structures in the pitch contour of the speaker's part which are identical to those of the instruments and which correspond to similar compositional techniques, does not necessarily mean that these should be "brought out" in performance. Perhaps also in this case (as in Moses und Aron) the relations in the score are more part of the compositional process, and not necessarily to be communicated by performers and perceived by listeners.(54)

[2.14] The change in conception of the Sprechstimme in Schoenberg's last decade or so of his life is also revealed in his new way of notating it in works such as Ode to Napoleon (1942), A Survivor from Warsaw (1947) and Modern Psalm (1950), where he used a single line (instead of five) for notating the approximate pitch of the speaker (see Examples 1-4). In these works the speaker (or reciter, or narrator) is arguably discouraged from singing (compared to Pierrot lunaire). The fact that Schoenberg also used accidentals in this late type of notation may seem contradictory, yet this paradoxical situation seems to support his call for performers not to read his notation of Sprechstimme at face value. Similarly, Stadlen writes about the fact that in Pierrot lunaire one can find Sprechstimme notes with and without note heads (see Example 5). He concludes from this that the notes with heads should be sung at the given pitch. I would like to suggest that it is also possible that Schoenberg intended to grant the performers different levels of freedom where the notes without heads should be sung in an even freer manner than the notes with note heads. In other words, the fact that there are different types of notation here does not mean that notes with note heads must be sung in exact pitch. It is possible, again, that Schoenberg intended here three levels of interpretation: 1) notation of the instruments which should be precise with relation to pitch; 2) notation of Sprechstimme with note heads which may be less precise; and 3) notation of Sprechstimme without heads which grants the speaker even more freedom in determining the pitch.


[3.1] I mentioned above Lorraine Gorrell's observation that Stiedry-Wagner's Sprechstimme is off-pitch. This is something that one notices immediately when analyzing Sprechstimme pitch in the different test pressings. However, the test pressings reveal further interesting information. Figure 1, which covers mm. 5-17 of "Eine blasse Wäscherin," has the Sprechstimme at the upper stave and the pitches of the four test pressings on the four staves below it.

[3.2] I used the computer program "gram" to detect Sprechstimme pitch. This program creates a spectrogram as in Figures 2 and 3. The user of the program can position the cursor on the spectrogram and the pitch is given in Hertz values. In Pierrot lunaire in general, and in our recording in particular, there is a special problem in deciding where to position the cursor; since Stiedry-Wagner often glissandos (see Figure 2) after and/or before the notated pitch (or its equivalent in her singing). In several cases I determined the pitch according to the dynamics--a place with higher dynamics (marked with darker black in the spectrogram--see for example Figure 2) was most likely the pitch which she tried to reach; or according to an average pitch, for example, when there was vibrato. Although the pitches of the test pressings in Figure 1 should be understood as an approximation, the possible degree of mistake is not larger than a quarter-tone and my transcription in Figure 1 is, therefore, accurate. In very few cases was there doubt about the exact pitch (for example, because of the voice being overridden by the instruments); these places appear in Figure 1 with a question mark. The duration, shown in Figure 3, was measured in relation to the commencement of each word of the text.

[3.3] It seems that Stiedry-Wagner allowed herself to transpose the pitch, usually, a third or fourth lower. Yet the transposition is not done consistently within the different test pressings: if one compares the four test pressings in mm. 5-6, one will notice that except for test pressings 1 and 3 in m. 5, which are almost identical, all the rest start at different pitches and move within the phrases in a manner which is very free compared with one another (hear Sound Examples 1-4). A further comparison of these measures and others suggest that she did not have a strict idea of the intervallic content and the degree of transposition; and that much of this was improvised at that moment and changed from test pressing to test pressing. For example, the starts of the phrase in m. 5 was transposed a minor third to a fifth lower (depending on the test pressing), while the phrase beginning at the middle of m. 7 starts in test pressings 1 and 3 at the very same note as indicated in the score (and in the other test pressings--transposed not more than a minor second away)!

[3.4] Sometimes the pitch content of the notated melody seems to be almost completely ignored: in the phrase starting at the middle of m. 9 one can see that except for the movement in spaces of seconds and of a prima, there is little resemblance between the test pressings. In test pressings 2 and 4 there is a repetition of a single pitch (in each case a different one), something which emphasizes the speech character, yet which was not indicated in the notation (hear Sound Examples 5-8). Another example can be found in the middle of the phrase which crosses mm. 13-14: the last four notes of m. 13 are different than those indicated in the score: in that Stiedry-Wagner does not reach the low point that Schoenberg notated. Yet here also, one can find a high degree of consistency among the test pressings, despite the discrepancy with the score.

[3.5] The examples mentioned above demonstrate the variety and liveliness of the singing of Stiedry-Wagner. The following examples will show that this improvisatory character was accompanied with a contradictory tendency towards stability. The start and end of the first phrase (mm. 5-6) are very similar: the first three notes of test pressings 1 and 3 as well as the last three notes of test pressings 1, 2 and 3 accordingly have almost identical pitches. In contrast, the start of the phrase is more similar to the score than the end of it: there is consistency among the different test pressings and a very close resemblance to the score (hear Sound Examples 1-4). A similarity between the pitch of the test pressings and the score can be found also at the end of phrases (see the first note in m. 9) or at the start of phrases in many other places in the song: for example, the last two notes in m. 14, which are indicated to be sung, contain pitches very similar (indeed almost identical in this context) to those indicated in the score. When the notated pitch is systematically observed by Stiedry-Wagner in several test pressings it proves that it was not done by chance. Many times the exact pitch is not kept but the intervallic content is observed: the last phrase in m. 10 and the first one in mm. 11-12 observe (by large) the direction and the intervallic relations of the notated melody. Note that the transposed starting tones as well as the ending ones in mm. 11-12 are again similar if not identical among the different test pressings.

[3.6] In several cases one is obliged to notice a simultaneous tendency of change and stability in single phrases. The last two notes of the first phrase of the song (first two notes in m. 7) start a gesture with a pitch that is close to that of the score (first note) and ends much lower than indicated; the singer ignored the last pitch of this phrase in favor of a heightened expression of the prolonged gesture. A similar phenomenon occurs in m. 15 where the last two notes are supposed to be performed as a step upwards of a diminished fourth, yet in practice all test pressings contain a step upwards which is greater than an octave. A greater "exaggeration" can be found in the last two notes of the next phrase in m. 16.

[3.7] The phrase starting at the middle of m. 8 and ending at the beginning of m. 9 approximately observes the pitch of the peak of the phrase and that which ends it, while the pitch starting this phrase is usually between a fourth and a fifth higher than indicated (hear Sound Examples 9-12). The tendency to observe the pitch contour of phrase endings (in spite of variety in the body of the phrase) can also be found at the start of m. 10, and the end of mm. 15 and 16. A tendency to observe phrase peaks can be found in m. 11 (see the high but especially the low point in the melody); and in m. 14 where she attempts (and usually succeeds in all test pressings) to reach the c'' twice (see Figure 2, Figure 1 m. 14 and hear Sound Examples 13-16). In m. 10 the c'' which is the peak of that phrase is transposed equally in all test pressings to g'. In m. 11 the beginning of phrase is transposed in test pressings 2-4 about a major third lower. In short, this consistency is not by chance and is probably a result of the singer paying more attention to the starts and ends of phrases.

[3.8] Most interesting are the places where there is consistency among the test pressings which is contradictory to the direction of the melody in the score. It is possible to see such a tendency in the last four notes of m. 6: at the word "Nachtzeit" she is consistent in singing the first pitch higher than in the previous word "Zur" and than going down; this is opposite to what is written in the score (see Figure 3, Figure 1 m. 6 and hear Sound Examples 9-12). See also the 4th and 5th notes of m. 12; the relation between the 3rd and 4th notes of mm. 14 and 16. This suggests that at times Stiedry-Wagner was consistently performing a contour which was different from that in the score. The consistency of her interpretation reveals that it was not pure improvisation, and that this additional conscious or unconscious "structure" was probably defined before performing the test pressings in the studio. The performer's "structure," which is sometimes in contrast with the pitch in the score, possibly fits the words better (at least from the point of view of the singer). We saw that many years before the recording took place she studied this composition with Erwin Stein, and that she remembered, even after so many years had passed, that she "was working very hard, oh very hard"; and that Schoenberg wrote to her just before the recording took place that they "must ... refresh the Sprechstimme thoroughly." The hard work in preparing the piece as well as the many performances done under Schoenberg's baton created this constant character which at times collides with the score and at other times corresponds with it.

[3.9] Another possible reason why Stiedry-Wagner "was working very hard" when preparing her interpretation was because "it's very difficult to speak in rhythm--strong rhythm," as she herself testified. I mentioned Schoenberg's early preface written some time between March 1912 and January 1914: "it is the duty of the performer to perform the rhythm absolutely precisely"; as well as his 1914 preface arguing that the performer should keep "the rhythm just as precisely as he would when singing, i.e., with no more freedom than he would take in a sung melody." The issue of rhythm is prominent in Figure 3. This figure must be read with caution, since what on paper may seem to be sounds with different durations may be perceived in listening as identical. Above Figure 3 one can find the rhythmic values of the first phrase. It seems that the words affect not only the pitch contour, as described above, but also the rhythm of the phrase. For example, in test pressing 3 the rhythm indicated in the score of the word "Nachtzeit" is distorted when the "Nacht-" turns out to be shorter than notated. One perceives the syllables of this word as of equal length. We can see that both rhythm and pitch contour are in contradiction with the score indication in this place. However, rhythmic deviations are usually not significant or systematic. The great deviations in pitch are compensated for by a rather strict adherence to notated rhythm. By fixing one parameter (rhythm) and giving much more freedom to another (pitch), Schoenberg created a situation where there is a mutual creation of musical meaning on the parts of composer and performer.


[4.1] The contradiction that is mentioned by Boulez, Stadlen and many other authors results from the 1914 preface that demands "taking well into consideration the indicated pitches" on the one hand, and the practice of Stiedry-Wagner (under Schoenberg's conducting) not to do so on the other hand. Other evidence mentioned above, the claim that "pitch notation" should be taken seriously (1913), the early preface asking to keep "the relationship between the pitches," Schoenberg's 1922 letter to Maria Freund, and finally, his very act of notating exact pitches, all contribute to the sense that Sprechstimme must involve some serious relation to the notated pitch.

[4.2] Yet, Schoenberg also crossed the stems of the notes and wrote as early as 1912 that the "recitation should hint at the pitch."(55) In Moses und Aron he points to the fact that Sprechstimme is beyond the twelve tones, and that the singer should extract from the notation the expression (as opposed to singing the exact pitches). If Boulez pointed out that themes in the voice part have relations to those in the instruments, Schoenberg, as if predicting this, claimed in 1949 that "the themes (and everything else of musical importance) happen in the instruments." During the same year he even went so far as suggesting that Pierrot lunaire must be spoken "without fixed pitch"; and he developed in several late works a Sprechstimme with no conventional pitch notation. All this contributes to the feeling that the notated pitch is not to be observed strictly.

[4.3] It was well known that Schoenberg's Sprechstimme invited diverging interpretations by various performers, that Stiedry-Wagner's interpretation did not strictly observe the indicated pitch and that Schoenberg accepted such interpretations. What was not known was the degree of freedom that Schoenberg granted Stiedry-Wagner: it was not known whether the "correct" pitches that she sang were done so by chance or on purpose; it was further not known whether the "wrong" pitches were completely experimental, or had some relation (even if remote) to the score. This study of the test pressings reveals that there is a high degree of consistency in Stiedry-Wagner's singing in several cases: some keep the exact pitch indicated in the score, some follow only the intervallic relations indicated in the score, and some do not observe the pitch melody of the score at all. The tendency to keep the pitch of the intervallic relations was especially prominent at melody peaks and phrase boundaries. The consistency among the test pressings was kept even when it was sometimes against the intervallic direction of the notated melody. One could never know whether Stiedry-Wagner's off-pitch Sprechstimme was based on a new score that she created or whether it was purely a real-time experimentation; the test pressings, however, show a degree of freedom as well as of systematic behavior, and reveal that while Stiedry-Wagner prepared her Sprechstimme in advance (hence some of the systematic features), she left many aspects of her performance to real-time interaction (hence the variety among the test pressings).

[4.4] Stiedry-Wagner did not have a strict approach to the observance of notated pitch. In some of her performances the pitch was carefully observed, while in others merely "hinted" at. This, and the fact that places that collide with the score are consistent, suggests that she was working consciously or unconsciously with a performative contour that does not correspond to or deviate entirely from that of the score. In other words, it seems that Sprechstimme is meant to be constructed and highly influenced by the process of building an interpretation by the performer.

[4.5] Albertine Zehme and Stiedry-Wagner were actors and not professional singers; although, Stiedry-Wagner could control pitch since she sang in Liederabende and operettas. Except for praising her publicly for her performance in the recording, Schoenberg constantly chose her again and again (under his baton, that of Erwin Stein and others) for more than twenty years. This would not have occurred if her off-pitch singing would have been seen as problematic in his view. Furthermore, there are two famous stories of Schoenberg's inability to recognize when a player used an instrument in a wrong transposition.(56) These stories are far from being simple descriptions of reality of Schoenberg's ability to decipher pitch, since there is a heavy suspicion that these players were playing tricks on the composer; in addition, he may have been concentrating on other issues when conducting this extremely new music.(57) In other words, the assumption of limited ability on the parts of Stiedry-Wagner and Schoenberg to control or decipher pitch does not explain the Sprechstimmee phenomenon.

[4.6] Schoenberg's Sprechstimme may be seen as a concept which resists the view of music as solely the composer's sound which needs to be reproduced by passive performers. I mentioned above Albertine Zehme's 1911 statement that words are not merely concepts to be reproduced, but rather they "allow us to partake of their inner experience" which is achieved by an unrestrained tone. She concluded that "Life cannot be exhausted by the beautiful sound alone." And indeed, the revealed nature of the test pressings seem to suggest that Sprechstimme notation can be seen also "as a stimulus to the performer to respond in a musically meaningful way," and not only as a tool for reproducing a sound object.(58) As mentioned above, the test pressings include both elements which stay relatively stable among different test pressings, as well as changing elements which support the view that performance "consists of an interpretive engagement with the notation � done in real-time, that has to be enacted afresh on each occasion."(59) Although these words were used by Nicholas Cook with reference to a more rhythmically complex composition by Bryn Harrison, Pierrot lunaire can be seen as a predecessor of this type of approach to notation. The delicate balance between the stable and dynamic elements should not be seen as a contradiction, pace Stadlen, Boulez and others, but as a source of strength. However illogical it may seem, Sprechstimme in Pierrot lunaire engages the performer in action which connects both composer and performer in a fresh, mutual act of creation. The singer Jane Manning had recently written: "The very essence of this masterpiece seems to stem from the extent to which it allows for constant renewal and refreshment in the very act of performing it� This is surely why it remains a consistently fascinating and satisfying task for the vocal performer."(60)

[4.7] In a conversation between Adorno and Boulez about Pierrot lunaire, Boulez said:

I heard beforehand a very curious statement: Leonard Stein from Los Angeles told me once that, for example, when they were first rehearsing the Ode to Napoleon in Los Angeles, Schoenberg demonstrated a few passages himself, and it was completely different than notated, because for him the expression is in the end more important than the notation. For the author that is of course possible. But if one stands before the score as an interpreter, one has to initially have respect for the text; for if one distances oneself too far from the text, then it is no longer necessary to have a score, and perhaps I am therefore stricter than Schönberg. (Boulez laughs)(61)

Boulez's preconception of respect to score that an interpreter must have (a "respect" that the composer himself did not seem to have) conveys a strong belief in the interpreter as a servant of the score that the creator (composer) communicates to the listener. If one takes seriously Schoenberg's interpretations of his own music, it may be concluded that a different approach should be conducted by the performer. Indeed, Boulez and Adorno understand the absurdity of this view in light of Schoenberg's practice and after Boulez laughs (at the end of the quotation above), Adorno answers: 'You are in this respect truly more papal than the pope'. Boulez fails to see a middle way between adhering strictly to the score and not using it at all.

[4.8] If one understands the role of the performer as one of reproducing a sound object then Boulez and Stadlen are right in detecting a contradiction. However, the test pressings of Pierrot lunaire confirm that a perfect reproduction was not Schoenberg's intention. Indeed, some aspects of pitch stay stable, and the rhythm is reproduced quite closely; however, the test pressings also reveal something new: the many changing elements and in some cases, their systematic nature, prove that great real-time interaction was expected from the speaker. The contradiction disappears if one understands the role of pitch not as one of a perfect reproduction (neither of pitches, nor of intervals) but as one that involves interpretative interaction in real-time. The 1914 preface indication that the "performer has the task of transforming it into a speech melody by taking well into consideration the indicated pitches" should be understood, I suggest, as a process of translation of pitch as if into a different language. Put bluntly, this means that the resulting pitches might be very different from those in the score. We saw above that Richard Kurth argues that Schoenberg's Sprechstimme is a "substitute for both speech and song." Also the other preface indication that "the spoken tone touches upon it but then leaves it immediately by descending or ascending" imply that an exact reproduction of the notated pitch (or intervals, or contour direction) is indeed not the main issue. The improvisatory yet systematic nature of the test pressings suggests that the "taking well into consideration" of pitch meant something different from reproducing exact pitch. What counts is not reproduction but a dramatic interpretive engagement with pitch, and more important with the text. In another place Stiedry-Wagner wrote that in performance with other singers sometimes people laugh, yet she argued that one must speak the part in a dramatic manner and that when she did so--no laughter was heard. Schoenberg wanted the speaker to carefully interact with the notated melody in a manner that will transform it; the aim of the hard work, mentioned above, that Stiedry-Wagner had done was not to reproduce the pitch but to transform it in a dramatic and improvisatory manner.

[4.9] The test pressings reveal what was previously almost unimaginable: Schoenberg accepted very different performances (although not completely different) of the Sprechstimme notation by Stiedry-Wagner in a period of not more than three days. It is not clear why Schoenberg decided to choose test pressing number 4 of the song for the commercial recording; the analysis of test pressings shows that faithfulness to the score was not a consideration since none of the test pressings is clearly preferable to others in this respect.

[4.10] It seems to me that in Pierrot lunaire there are actually two types of notations: one for the instruments which demands a relatively precise rendition of pitches; and one for the Sprechstimme which demands real-time interaction of the singer with the notation which creates a much higher degree of unexpected results. The fact that there are pitch structures in the Sprechstimme melody that have relation to the melodies in the instruments does not necessarily mean that these should always be "brought out" in performance. After all, compositional constructions that helped or fascinated composers in the process of composing--features that appear in the score yet are not necessarily to be perceived by listeners, is not an uncommon phenomenon in the history of music.

[4.11] Sprechstimme is indeed a bizarre phenomenon in traditional classical singing. Nevertheless, in many ways it magnifies what happens also in more conventional singing. Deviation from notated pitch, as well as a certain creative and systematic approach which refuses to be reduced to score indications, is frequent in many types of singing. In this sense it is possible to speak of a singer's contour as a source which may have not less authority than what is traditionally known in music analysis as 'the structure' or 'the melody' notated in the score. This is of course valid also for much music of the twentieth century which includes singing techniques influenced directly or indirectly by Schoenberg's Sprechstimme.

[4.12] The history of Schoenberg's conception of Sprechstimme proves that he understood it differently in different periods. Schoenberg/Stiedry-Wagner's 1940 workshop is after all only one historical occasion. Schoenberg was dealing with a particular performer, in a particular setting, and these constraints are perhaps somewhat contingent. Nevertheless, if one examines Sprechstimme history as well as the test pressings of the 1940 recordings, it is hard to avoid the notion that in spite the changes in conception, Schoenberg did expect that the singer will be always on a continuum (to use a metaphor coined by Cook)--more on the side of interacting with the score, and that the instruments will be more on the side of reproducing a sound object.(62) In this sense, the shifting and contextualized picture of Sprechstimme which I presented above does not contradict the larger context, namely that of an interaction-reproduction continuum. One can interpret this score while staying within the framework of Schoenberg's general intentions about Sprechstimme (tending more towards the interaction side of the continuum in relation to the instruments), as well as what one may reconstruct as Schoenberg's more local intentions (in 1940 or at any other period). Perhaps the greatness of this composition is that at different times, Schoenberg placed emphasis on different, and seemingly contradicting aspects, while keeping the larger spirit of the Sprechstimme in relation to the whole ensemble.(63) In this regard, Pierrot lunaire offers an endless variety of possibilities and musical meanings.

Comment on this articlee

Avior Byron
Royal Holloway, University of London and Bar-Ilan University
Yoni Netanyahu 20, apartment 17
Givat Shmuel 54424

Copyright Statement

Copyright � 2006 by the Society for Music Theory. All rights reserved.

[1] Copyrights for individual items published in Music Theory Online (MTO) are held by their authors. Items appearing in MTO may be saved and stored in electronic or paper form, and may be shared among individuals for purposes of scholarly research or discussion, but may not be republished in any form, electronic or print, without prior, written permission from the author(s), and advance notification of the editors of MTO.

[2] Any redistributed form of items published in MTO must include the following information in a form appropriate to the medium in which the items are to appear:

This item appeared in Music Theory Online in [VOLUME #, ISSUE #] on [DAY/MONTH/YEAR]. It was authored by [FULL NAME, EMAIL ADDRESS], with whose written permission it is reprinted here.

[3] Libraries may archive issues of MTO in electronic or paper form for public access so long as each issue is stored in its entirety, and no access fee is charged. Exceptions to these requirements must be approved in writing by the editors of MTO, who will act in accordance with the decisions of the Society for Music Theory.

This document and all portions thereof are protected by U.S. and international copyright laws. Material contained herein may be copied and/or distributed for research purposes only.

Prepared by
Brent Yorgason, Managing Editor
Updated 15 May 2008