1. Problems of Mahler Performance Practice
 Although Gustav Mahler conducted about 70 performances of his own works during his lifetime,
some of which are fairly well documented, it is not easy to grasp the performance style he developed for his own music in a coherent manner. What we do know is that many performances (conducted by himself or others) prompted Mahler to continually revise his scores with considerable scrutiny in an attempt to “elucidate” or “bring out” the meaning of the score-script, resulting in a number of detailed performance instructions in his scores that is unusually high for this period. In respect to tempo, Mahler most of the time refused to indicate metronomic tempo marks, as he insisted that the “correct” tempo – as well as dynamics and orchestral balance – had to be adapted to the interplay between composed structure, performance space, performing musicians, and audience: […] [in my scores] everything is spelled out in detail by means of the note-values and rests. Of course, I am referring to the things that can be written down. All the most important things – the tempo, the total conception and structuring of a work – are almost impossible to pin down. For here we are concerned with something living and flowing that can never be the same even twice in succession. That is why metronome markings are inadequate and almost worthless; for unless the work is vulgarly ground out in barrel-organ style, the tempo will already have changed by the end of the second bar. Therefore the right inter-relationships of all the sections of the piece are much more important than the initial tempo. Whether the overall tempo is a degree faster or slower often depends on the mood of the conductor; it may well vary slightly without detriment to the work. What matters is that the whole should be alive, and, within the bounds of this freedom, be built up with irrevocable inevitability.
Many sources testify that Mahler in his orchestral concerts indeed changed tempi from bar to bar
in the tempo rubato tradition (labelled “Phrasierungsrubato” by Jürg Stenzl ) oriented towards melody ( Melodie) , cantability ( Gesanglichkeit), and clarity ( Deutlichkeit), as set most prominently by Richard Wagnerʼs essays Über das Dirigieren (1869) and Beethoven (1870), a tradition which – mediated by Hans von Bülow – was of considerable influence on Mahlerʼs concept of orchestral performance, although Wagnerʼs, Bülowʼs, and Mahlerʼs performance aesthetics surely cannot simply be equated. In any case, the skepticism towards a “literal,” one-dimensionally text-oriented performance, also mirrored in Mahlerʼs much-discussed revisions of the established symphonic repertoire (encompassing most prominently Beethovenʼs Ninth Symphony), which can be traced back to the Romantic aesthetics of the ineffable essence of the musical art work, forms a strong link between Mahler and the Wagnerian tradition and allows one to distinguish Mahlerʼs performance aesthetics from the more “classicist” approaches of his contemporaries Karl Muck and Richard Strauss, although at the same time his relentless insistence on “clarifying” musical structure testifies to a specific variant form of modernity-driven text-oriented conducting.
Despite this skepticism towards a regulation of tempo, some metronome marks for individual movements from Mahlerʼs symphonies have been transmitted. Metronome marks in the conducting scores of Willem Mengelberg – who stood in Mahlerʼs highest esteem – were in most cases made in close communication with Mahler, documenting Mahlerʼs own tempi as witnessed by Mengelberg in many rehearsals and performances, as for example in the case of the Fourth Symphony
– though in this case the metronome marks were clearly not observed in Mengelbergʼs recording of this symphony. Anton Webern also claimed to have received metronome marks for the first three movements of the Fifth Symphony from Mahler personally. In some works metronome marks in the manuscript score were removed by Mahler for the printed edition, but in several cases metronome marks found their way into the final score, as in several sections of the First Symphony. Finally, the durations of some of Mahlerʼs performances of his works have been measured (by the composer himself and others), allowing for tentative conclusions about the chosen tempi. “Historically informed performances” of 19th- and early 20th-century music are still rare, and in Mahlerʼs case it is mainly Roger Norringtonʼs recordings of the Symphonies 1, 2, 4, 5 and 9 that have been associated with an explicit attempt at taking performances of Mahlerʼs own time as a (flexible) point of orientation, with such inherited reference points playing a considerable role in the conductor’s concept of performance.
Hermann Danuser pointed out already in 1992, however, that Mahler was very well aware of the limitations and discrepancies of such an “auctorial performance tradition.”
After an obviously unsuccessful performance of his Second Symphony in an arrangement for two pianos, featuring among others his friend and assistant Bruno Walter as one of the performers, he lamented that this event showed “the truth about every socalled [ sic] ‘tradition’: there is no such thing! Everything is left to the whim of the individual, and unless a genius awakens them to life, works of art are lost.” The elusiveness of tempo indications in particular frustrated Mahler repeatedly: “‘One would almost be tempted to write in no tempi and no expression marks,’ he said in exasperation, ‘and leave it to the performer to understand and articulate the music in his own way.’”
The objectives of the
Mahler-Bund, founded during the Mahler Festival in Amsterdam in 1920 (with Arnold Schoenberg acting as its first president) and dedicated to firmly establishing a performance practice of Mahlerʼs music committed to the composerʼs ideals, could not obscure the fact that early on even those conductors who were declared (or declared themselves) Mahler’s “heirs” or “soulmates,” particularly Oskar Fried, Willem Mengelberg, Bruno Walter, and Otto Klemperer, not only pursued radically differing concepts of Mahler interpretation in general – in simplified terms: biographical and motivic-thematic (Mengelberg), classicist (Walter), and objectivist (Klemperer) – but in many cases introduced readings which were clearly at odds with Mahlerʼs own performances. This situation is further complicated by the observation that performances of the same work by the same conductor may reveal profound deviations. Klempererʼs 1971 recording of the Second Symphony lasts 28 minutes longer than his 1951 recording.
 Mahlerʼs piano performances of his own works, documented on a 1905 Welte-Mignon piano roll,
testify to a highly flexible conception of tempo with bar-to-bar tempo changes as described above. Mahler stated that the “correct” tempo was arrived at “when everything can still be heard” – a statement by which he obviously implied that the tempo “space” required by singers should be respected: “A tempo is correct when everything can still be heard. When a figure can no longer be perceived because the notes begin to overlap, the tempo is too fast. In a Presto the limit of distinctness is the right tempo: beyond that, the effect is lost,” to which he added that “when the audience seemed unmoved by an Adagio, he slowed down the tempo instead of increasing it, as is usually the case.”
In order to “bring out the intention,” Mahler even encouraged others to make changes to his scores “if required by the performance space and the quality of the orchestra.”
It is documented that Bruno Walter, for example, indeed followed this practice by adding retouchings in the orchestration of some sections of Mahlerʼs works, including the beginning of the recapitulation in the first movement of Das Lied von der Erde.
Whereas many sources testify to a considerably high tempo of Mahlerʼs performances,
the intention to “bring out” details of the musical structure may to some extent require, in contrast, a relatively slow pace. Theodor W. Adorno in particular seems to have defended the latter principle, even against one of the apparently most “authentic” Mahler conductors Bruno Walter, whose 1952 recording of Das Lied von der Erde met with Adornoʼs criticism that in the first movement “just to make it go quickly, the pesante element of the basic character is missed completely.” A rehearsal in which Jascha Horenstein had the Vienna Symphonic Orchestra play a passage from the first movement of the Ninth Symphony “in slow motion, in order to technically control the entangled, dissociated string voices,” in contrast, was highly praised by Adorno und Hans Wollschläger. The two listeners agreed that “it should sound like that, if it could still sound like that.” In addition, the two Mahler advocates shared a deep skepticism towards the increasingly fast tempi of the “star conductors” during the 1960s, who, according to Adorno, tended to conceal the fragility of Mahlerʼs works by the “urgency of ‘I know that already,’” thus making unheard the ruptures in the musical structure, the signs of modernity in Mahler’s scores.
In sum, it seems exceptionally difficult to draw sound criteria for tempo design and performance aesthetics from these historical sources. Hans Wollschlägerʼs argument that every Mahler movement has a “basic tempo of which all apparent changes only form organic deviations”
in particular is not supported by convincing evidence. The performance history – as will be amply demonstrated in the present essay – is a strong argument against such a normative concept of interpretation; it rather constitutes Mahlerʼs work as an inexhaustible source of potential, new, or re-invented performance strategies. Still – to cite three categories of performance introduced by Hermann Danuser – in the case of Mahlerʼs music, the idea of a “historical-reconstructive mode” of performance, oriented mainly (though not exclusively) towards the performance style favoured by the composer-conductor himself, arguably cannot neatly be distinguished from “traditionalist” and “contemporary” styles of performance. The imagination of a performance “faithful” to Mahlerʼs principles – even though it might be impossible to pin them down to authoritative “rules” – has occupied an important place for many performers and commentators from the beginning, considerably shaping the way musicians and audiences have conceived of Mahlerʼs music. 2. The Limitations of Structuralist Performance Styles vs. Performance as Analysis in Real Time
Mahlerʼs proposition that the key criterion for the disposition of tempo is that “everything can still be heard” may be regarded as a main source for a specific type of structuralist performance style that came to be associated with the Schoenberg School. The intentions of the
Verein für musikalische Privataufführungen were to pursue “the greatest possible clarity” and the “fulfilment of all intentions of the author as derived from the work.” Rudolf Kolisch explained that for Arnold Schoenberg, “all technical means are to be subordinated to the musical idea without utilising associations from the spheres of emotion.” Similarly to the Schoenberg School’s aesthetics of orchestration, the function of which, according to Adorno, was to provide an “X-ray photograph” ( Röntgenphotographie) of the work, performance, for Schoenberg and his followers, had the foremost function of bringing out musical lines and communicating the structural design. This tendency was considerably grounded by the adoration Mahler enjoyed in the Schoenberg school.
When such a “structuralist” performance style is subjected to critical scrutiny, it soon appears that it involves many paradoxes and dead ends. This is conceded even in studies which are highly sympathetic of the basic project of harmonizing structural analysis and performance. In a recent study of the 1964 recording of Beethovenʼs Violin Concerto by Rudolf Kolisch and René Leibowitz – two performers deeply shaped by Schoenbergʼs performance ideals – which involves a thorough discussion of the conceptual background as well as a
close listening of the musical result – Thomas Glaser outlines that Kolisch and Leibowitzʼs “authoritative” performance model, which considers analysis as precondition for any kind of viable (hermeneutic or practical) interpretation, is necessarily confronted with a “difference between theoretical aspiration and the area of practice.” The mandatory consequences drawn by both performers from a detailed analysis of the musical text do not materialize without compromise and leave many open questions.
It is for such reasons that recent musical performance studies have articulated considerable skepticism towards “overexposed” structural features in performance – a skepticism that dates back at least to Heinrich Schenkerʼs theory of performance.
And yet it is evident that the options for a performer to “mark” (or “unmark”) formal-spatial boundaries or macroformal functionality also beyond an “orthodox” realization of “phrase arching,” that is, the conventional way of slowing down the tempo at the beginning and end of musical phrases, are very rich and multifaceted and that as yet they have hardly been systematized. More importantly, the impact of performance decisions for macroformal perception is not at all limited to a simple “projection” of score-based analysis. The kind of “formal analysis in real time” described by fortepianist Robert Hill rather comprises a large set of strategies pertaining to tempo/timing/rubato/agogics, synchronicity/ asychronicity, caesuras/fragmentation/continuation, dynamics/accentuation, timbre, registration, etc. – aiming at a “connection between the moment and the whole” – that has the “potential to expand our listening habits fundamentally.”
 It seems promising, therefore, to explore exemplary situations and strategies of the performers’ impact on the “creation” of large-scale form. However, it should be remembered – keeping in mind Adornoʼs and Wollschlägerʼs criticism of a performance style which “undoes” key features of Mahlerʼs modernity mentioned above – that “macroformal analysis” here cannot tautologically presuppose a large-scale formal coherence but must prominently integrate the “death-wish”
of formal rupture and disintegration. 3. Performative Strategies in Recordings of the First Movement of Das Lied von der Erde
To what extent, then, do recorded performances of Mahlerʼs
Lied von der Erde reveal a concept of “formal analysis in real time”? When we think of the structural use of tempo contrasts and deviations – along with the organization of dynamics and the balancing of musical layers or voices – as the most obvious means of marking or “creating” large-scale form in performance, we first have to take into account that – as mentioned above – Mahlerʼs scores usually tend to point out such deviations with considerable accuracy. Nevertheless, the subsequent discussion will provide ample evidence for the fact that performance concepts even of such a detailed musical text may vary drastically.
The somewhat contested place of Mahlerʼs
Lied von der Erde in the history of genres and the ambiguity between strophic lied form and developmental sonata form may help us to frame the question about an “analysis in real time” during a performance of this work. We might ask, for example, whether it employs developmental strategies which let a whole movement, or ultimately even the entire cycle of six movements, appear as one overarching process, or whether it is possible to observe individual solutions for the different movements or sections of the music that let the Lied appear more as a series of “strophes” or even fragments, each of which is somehow contained within itself. Of course this question seems highly theoretical at first and requires differentiation and refinement during the analytical process.
Following a method described as “augmented listening” by Nicholas Cook,
I aim to confront a quantitative analysis of recordings (“distant listening”) with a qualitative analysis of specific key sections or processes (“close listening”), aiming at a mutual improvement of the methodological constraints of both strategies. I begin by approaching 23 recordings of the first movement, selected from a recording history of about one hundred titles, including the entire period spanned by recordings of Das Lied (1936–2017) and covering many renowned Mahler conductors (table 1). Table 1: Recordings of Mahlerʼs Das Lied von der Erde taken into consideration for the analysis of the first movement
The first movement lends itself particularly well to a tempo analysis, firstly because of its stable 3/4 meter and its clear whole-bar pulse, and secondly for its various indications of tempo change and adjustment in the score. Moreover, this opening movement plays an important role in claiming the
Lied for the genre of the symphony as it clearly betrays the contours of sonata form, with the principal compositional challenge being to translate the regular stanza-refrain structure imposed on Li Boʼs poem by Hans Bethge into the dynamic developmental principle of a symphonic sonata form.  Mahler achieves this mainly by discarding the third refrain after the third stanza, thus creating a close dramatic link between the development (third stanza) and the strongly curtailed recapitulation (fourth stanza), resulting in a basic formal structure of three continuously expanding musical strophes or rotations. Table 2 provides a synoptic overview of the movementʼs formal design, with special emphasis on the formal function of the tempo changes indicated by Mahler. The sections rendered in grey colour mark important formal turning points and key moments that deviate from the main tempo Allegro pesante (Ganze Takte, nicht schnell) more or less obviously. This applies especially to the three refrains which are marked by the indications “Ruhig” [calm], “Sehr ruhig” [very clam], and “Gehalten” [sustained], all three of them being clearly separated from the preceding and subsequent sections by ritardandi. In addition, both expositions (stanzas 1 + 2) distinguish the three verse groups from one another by ritardandi or a modification of the main tempo to “sostenuto” (mm. 29–30; 121–124) and “etwas gehalten” (mm. 45–52; 137–152).  In comparison to these detailed tempo instructions, it is striking that the entire development section (123 measures) eschews tempo changes, with the sole exception of the indication “Leidenschaftlich” [passionatly] at measure 291, provoked by the question occurring in the text “Du aber Mensch, wie lang lebst denn du?” – while it is not clear whether this indication also implies a tempo change, and if so whether it suggests a faster or a slower tempo. In any case, this “passionate” character marks the beginning of a long intensification on all musical levels which culminates in measures 353 to 368 (“Hört ihr, wie sein Heulen / Hinausgellt in den süßen Duft des Lebens!”). This intensification creates a very close entanglement between development and recapitulation, despite the clear motivic parallelism of measure 326 with the beginning of the movement. Table 2: Formal design of the first movement Das Trinklied vom Jammer der Erde from Mahlerʼs Das Lied von der Erde
The results of the “distant listening” sessions of the 23 recordings are summarized in diagram 1 and table 3 (all tempo values throughout this article are rendered in beats per minute = bpm). It is hardly surprising that the tempo stability globally tends to increase in the more recent recordings. The degree of tempo stability/flexibility can be detected from the proportional values of the standard deviation (second and fourth line of table 3), indicated here for the “Tempo I sections” exclusively (i.e., without calculating the sections where a modified tempo is indicated in the score; table 3, first line) and for the entire movement (table 3, third line).
The maximum of tempo instability, however, is not found in the earliest recordings of Walter 1936 (14.8/20.3%) or Schuricht 1939 (14.2/21.2%) but in Leonard Bernsteinʼs 1966 recording: Both within Tempo I (15.4%) and for the entire movement (22.3%), Bernstein reaches maximum values; thus, his interpretation indicates a very variable Tempo I as well as a strong difference between Tempo I sections and those with modified tempo. Diagram 1: Mahler, Das Lied von der Erde, i; tempo graphs of 23 recordings; tempo values in bpm referring to a dotted half note/whole bar beat (linear scaling)
Table 3: Mahler, Das Lied von der Erde, i; average values and standard deviations of the tempo graphs shown in diagram 3. The four columns of the table (from left to right) refer to (1) the main tempo for Tempo I sections (mean value for all sections without tempo modifications in the score), (2) the proportional standard deviation of Tempo I sections (indicates the degree of tempo stability/flexibility), (3) the mean tempo for the entire movement, (4) the proportional standard deviation of the mean tempo.
 Diagram 2 enlarges the tempo graph of Bernsteinʼs 1966 in comparison to his 1972 performance.
The 1972 (live) recording basically pursues the same dramaturgy, though mostly in a less pronounced manner. Particularly remarkable in the 1966 performance is that Bernstein reduces the tempo in the ritardandi before the development and recapitulation to a virtual standstill to less than a quarter of the subsequent fluent tempi (m. 202 = 17.3 bpm vs. m. 205 = 62.9 bpm; m. 392 = 16.4 bpm vs. m. 394 = 75.2 bpm / m. 398 = 78.8 bpm; diagram 2). Equally obvious is the continuous acceleration of tempo in the second half of the development until the abruptly slanting curve before the beginning of the third refrain (mm. 368–369; video example 1). The intensity which features James Kingʼs brilliant performance of this scene of the ape on the graves is substantially indebted to this tempo dramaturgy. The key words “ [Du aber Mensch,] wie lang lebst denn du?” (m. 298/299), “an all dem morschen Tande [dieser Erde!]” (m. 317/318), “Seht dort hinab!” (m. 328–330), “Ein Aff’ ist’s” (m. 348–350) are highlighted by accelerating tempo levels (62.4–67.6–71.9–79.9 bpm), the climax at the words “… hinausgellt in den süßen Duft des Lebens” (mm. 357–365) is intensified by a continuous fast tempo up to measures 358/359 (68.3/66.4 bpm) and an abrupt deceleration immediately before the climax point in measure 360 (43.8 bpm), further decreased to 29.2 bpm at the beginning of the third refrain (m. 367). This large-scale tempo dramaturgy is enhanced by a considerable reduction of tempo in the subordinate theme section in the development (mm. 281–290, main tempo: 45.5 bpm/48.1 bpm in 1972), a tempo change which – just like all those previously mentioned – is not indicated in Mahlerʼs score, as well as the sustained tempo in the third and final refrain following the climax ( main tempo: 35.8 bpm/37.2 bpm in 1972). Diagram 2: Mahler, Das Lied von der Erde, i: tempo graph Bernstein 1966/1972 (logarithmic scaling)
Video Example 1: Mahler, Das Lied von der Erde, i, mm. 289–381: Bernstein 1966 (James King; Wiener Philharmoniker; Leonard Bernstein)
The minimum of tempo deviations in diagram 1/table 3 is
not found – as might have been expected – in recordings of the recent two to three decades but is marked by recordings from the 1960s and 70s, including the famous rendition of Otto Klemperer with soloists Fritz Wunderlich and Christa Ludwig from 1966 (10.2% for Tempo I sections), as well as those of Bernhard Haitink from 1975 (10.0% for Tempo I sections) and Jascha Horenstein from 1972 (11.2% for Tempo I sections; this value is close to that of the seven recordings since the 1990s, all ranging between 11.0 and 12.0% for Tempo I sections). (All three recordings also have relatively low though less significant values for the deviations for the entire movement, indicating that they do not tend to pronounce the difference between Tempo I and modified tempo areas too rigorously.) Horensteinʼs extremely slow reading of the movement (46.0 bpm), with only Klaus Tennstedtʼs 1984 interpretation (45.5 bpm) taken at a slower pace, renders an impression of the conductorʼs “slow motion” approach testified by Adorno and Wollschläger, cited above. Klemperer, in contrast, who had recorded the movement in a frantic tempo in 1948 and 1951 (67.9 and 65.6 bpm respectively ), tends towards a very moderate approach in 1966 (55.5 bpm), close to that of Haitink (53 bpm). Klempererʼs, Haitinkʼs, and Horensteinʼs interpretations of the climactic area show no stronger or lesser deviations of tempo than during the rest of the movement (diagram 3). Klemperer, for example, provides ample space for Fritz Wunderlichʼs restrained “Heulen,” in sharp competition with the insisting counterpoint-ostinato on the E5 of the glockenspiel (which is particularly present in this recording as if to materialize Anton Webernʼs enthusiastic report on its role in this movement ; video example 2). Thus, in contrast to Bernstein neither Klemperer, Haitink, nor Horenstein appear at first glance to employ tempo design as a means of sharpening the form of the movement. Relying on a text-oriented performance, they seem to refrain from a clarification of the composed formal structure via tempo organization.
Such a qualification, however, requires a still “closer” listening mode. Similarities and differences between these three “controlled” interpretations might be illustrated by a short listening exercise towards their grasp at the instrumental subordinate theme in the development (mm. 281–293) and the pre-climactic zone marked by “Leidenschaftlich” (mm. 292ff.) up to the recapitulation (diagram 4). All three conductors interpret the first phrase of the “Leidenschaftlich” zone (mm. 292–301) with a slight increase in the basic tempo (Klemperer 1966: 53.1 60.1 bpm; Horenstein 1972: 44.2 47.8 bpm; Haitink 1975: 53.1 54.5 bpm). The basic dramaturgical concepts, however, are clearly discernible: After slowing down the instrumental subordinate theme to 53.1 bpm, Klemperer clearly marks off the individual phrases of the “Leidenschaftlich” section (mm. 292–304; 305–315; 316–325 – as marked by the verses of the text: Du aber, Mensch, wie lang lebst denn du? / Nicht hundert Jahre darfst du dich ergötzen / An all dem morschen Tande dieser Erde!) by differing tempo nuances (60.1–56.7–54.3 bpm), stabilizing his basic tempo 55.5 bpm at the beginning of the recapitulation. The interpretation is very much guided by the “over-pronounced” text-articulation of Wunderlich, apparently in tune with Mahlerʼs ideals of a tempo guided by vocal melody. Horensteinʼs and Haitinkʼs tempo changes are more linear, as can be seen from the normalized graph in diagram 4: Both continuously increase tempo until the first phrase of “Leidenschaftlich,” starting off from a slower tempo before the subordinate theme. In the further process, Horenstein in particular marks the entry of the recapitulation with a
Luftpause prolonging measure 325, the glockenspiel octave sharply interrupting the vocal line (video example 3), whereas Haitink enhances the entanglement of development and recapitulation by merging the phrases before and after measure 326 into a larger phrase structure (video example 4). Diagram 3: Mahler, Das Lied von der Erde, i: tempo graphs Klemperer 1966, Horenstein 1972, Haitink 1975 compared (logarithmic scaling)
Video Example 2: Mahler, Das Lied von der Erde, i, mm. 289–381: Klemperer 1966 (Fritz Wunderlich; New Philharmonia Orchestra; Otto Klemperer)
Diagram 4: Mahler, Das Lied von der Erde, i, mm. 203–344, normalized tempo graphs Klemperer 1966, Horenstein 1972, Haitink 1975 compared (logarithmic scaling)
Video Example 3: Mahler, Das Lied von der Erde, i, mm. 289–381: Horenstein 1972 (John Mitchinson; BBC Northern Symphony Orchestra; Jascha Horenstein)
Video Example 4: Mahler, Das Lied von der Erde, i, mm. 289–381: Haitink 1975 (James King; Koninklijk Concertgebouworkest, Amsterdam; Bernard Haitink)
 The interpretations of the three refrains can be considered as a key to understanding the different performance concepts of the entire movement. The recurring maxim “Dunkel ist das Leben, ist der Tod” at the (end of) each refrain is set by Mahler – in contrast to earlier sketches of the music
– in three different keys: G minor, Ab minor and A minor, implying a rising tessitura of the vocal line and thus an intensification of its bodily expressive content. The cadential closure in the first and third refrain is evaded or “deferred” by an open diminished seventh chord on the tonic bass (mm. 90; 393), letting the word “Tod” fall into the abyss as it were, which after the third refrain and the preceding surrealist ape scene rings associations with the absurdity of death and the rebellion against it (further enhanced by the flatter tongue sounds in the high woodwinds of the postlude ritornello that already in the beginning set the atmosphere of the movement together with the glockenspiel).
An obvious tempo dramaturgy in the interpretation of the three refrains set by Bruno Walterʼs 1936 recording (table 4) was to decelerate the tempo of the refrain stepwise, making its contrast to Tempo I increasingly more pronounced – an approach which seems to be justified by the difference in Mahler’s indications: “Ruhig”/“Sehr ruhig” [mm. 77–89], “Sehr ruhig” [mm. 179–202], “Sehr ruhig”/“Gehalten” [mm. 369–392]. However, from the 1970s onwards a clear tendency to increase the tempo of the third refrain can be observed, possibly in an attempt to retain the tension aroused by the climactic area and to create the impression of incisiveness towards the end (further enhanced by the short and dry final chord). More recent recordings, namely Jonathan Nottʼs 2017 reading, even invert the tempo dramaturgy, rendering the first refrain in the slowest and the third refrain in the fastest tempo.
Table 4: Mahler, Das Lied von der Erde, i, main tempo of the three refrains (mm. 77–89; 179–202; 369–392) compared to Tempo I in the 23 analyzed recordings
Independent of the question of “accuracy to the score” (at least the indication “Gehalten” in the third refrain may be said to be ambivalent, as it is not clear whether it suggests a slower or a faster tempo than the preceding “Sehr ruhig”), both dramaturgies might turn out to be convincing. Whereas an increased refrain-tempo in the final refrain might be heard as a consequence of the preceding climax and its grotesque atmosphere, a continuous deceleration of the refrain-tempi might symbolize the inescapability of the maxim’s insight of transitory life. This might be further supported by the important addition of the third refrain, which is, contrary to refrains 1 and 2, expanded by an additional pair of verses (“Jetzt nehmt den Wein! Jetzt ist es Zeit, Genossen! / Leert eure goldʼnen Becher zu Grund!”), often marked out by a strong ritardando in measure 381 which ostensibly symbolizes the time the singer-protagonist takes to empty his cup, a further formal marker making the final occurrence of the maxim’s line even more effective and chilling. Thus, a criterion for assessing the plausibility of the different tempo dramaturgies might be whether the tempi chosen for the three refrains principally highlight their architectonic function or whether the tempo dramaturgy communicates the refrains as part of a transformational process spanning the movement as a whole. Approaching this delicate question of course involves taking into consideration parameters beyond tempo. It can be relevant, for example, whether the evaded cadence after the first and third refrain and the “hands-on” character of the ensuing tutti ritornello increase in intensity towards the end of the movement (which would support a process-oriented reading of the movementʼs form). A short comparison of three readings of refrain 3 may illustrate the broad choice of solutions available (video examples 5–7).
In his 1939 recording, Carl Schuricht (Concertgebouw Orchestra; tenor: Carl-Martin Öhmann) offers an extreme ritardando in measure 381 (the emptying of the cup). The main tempo of the third refrain – already taken considerably slowly (34.6 bpm) – is here reduced to 12.4 bpm (video example 5). A pronounced contrast to this interpretation is heard in the most recent recording conducted by Jonathan Nott (Vienna Philharmonic, tenor: Jonas Kaufmann) which almost seems to “rush” towards the “Tod” in m. 393 (video example 6). Nottʼs tempo dramaturgy is at times almost imperceptible (proven by the small difference between his deviations within Tempo I and within the entire movement, shown in table 3: 11.2/13.7%). In contrast, Schurichtʼs interpretation offers a clear “escalation” of the formal-dramatic situation in refrain 3 when compared to his rendering of refrain 1. However, both performances miss an important detail in the final key moment of the movement: The “shocking” effect of the evaded cadence in measure 393, compositionally enhanced by the ritardando in m. 392, is mitigated if the soloist and/or orchestra intensify the dynamics of the penultimate note of this cadence – contrary to the score, which does not indicate a crescendo except for the quaver-upbeat in the orchestra immediately before measure 393. Again it is Bernstein who (in both his recordings) stages this contrast in the most pronounced manner, but it is also very clearly audible in the live recording of Josef Krisp and Fritz Wunderlich from 1964 (video example 7).
Video Example 5: Mahler, Das Lied von der Erde, i, mm. 367–405: Schuricht 1939 (Carl-Martin Öhmann; Koninklijk Concertgebouworkest, Amsterdam; Carl Schuricht)
Video Example 6: Mahler, Das Lied von der Erde, i, mm. 367–405: Nott 2017 (Jonas Kaufmann; Wiener Philharmoniker; Jonathan Nott)
Video Example 7: Mahler, Das Lied von der Erde, i, mm. 367–405: Krisp 1966 (Fritz Wunderlich; Wiener Symphoniker; Josef Krips)
If the final “Tod” occurs in such a somewhat “expected-unexpected” manner, to some degree unlikely after the protagonistʼs harmonious invitation to empty the cup, being further drawn into the codaʼs surrealist and short-cut recapitulation of the beginningʼs atmosphere and its tendency towards panic, frenzy, and catastrophe, it seems clear that the symphonic character of the movement – and thus of the
Lied von der Erde as a whole – can be brought out very convincingly. While a macroformal tempo dramaturgy may facilitate the process of grasping the important entanglement of development and recapitulation, emphasizing the movementʼs end-oriented formal dramaturgy, the clear articulation of such “ruptures” in the musical narrative might arguably be found even more important for performing “analysis in real time,” not least because these ruptures tend to fulfil important text-related as well as formal functions. This also suggests that the singerʼs lied interpretation – the articulation and pronunciation of the text, the characterization of the “tenor persona,” and its development in the course of the movement – and the symphonic dramaturgy may illuminate and consolidate one another. 4. Perspectives on Questions of a Performance-sensitive Analysis of Cyclic Form in Das Lied von der Erde
 A thorough discussion of the performative strategies applied to the cycle of
Das Liedʼs six movements as a whole is beyond the scope of a single article and would require highly time-consuming “augmented listening” research into the entire recording history. What can be offered here instead is a reflection on which research questions such a study might pose and a short exemplification of distant listening (based on highly selective quantitative data), as well as a short concluding example of how “close listening” might contribute to this research area.
Basic questions to be tackled by research into the performance of the whole cycle are:
(1) How are the individual movements related to one another in terms of tempo/duration as well as dramaturgy (tempo, sound, and timbre changes, articulation of both orchestral and vocal structures/text, etc.)?
(2) Is the relationship between the two large “Abteilungen”
of the Lied (movements 1–5/6) balanced, or are there clear indications that the emphasis is placed on the finale?
(3) Does a performance show clear attempts at strengthening the cyclic integration by analogous tempi, by a highlighting of motivic or timbral connections, or by similar means, or is there rather a tendency to “isolate” the six songs into fragmentary pieces? The latter tendency might become particularly dominant during the last movement, the fragmentary, loose-knit character of which has often been discussed. Placing the emphasis on this fragmentary character (which to a certain, though limited, degree may also be observable in the other movements), could imply an interpretation of the
Lied not as a symphonic form in an empathetic sense but as a more contemplative reflection of the “world” which Mahlerʼs symphonies aim at creating.
A basic material (to be expanded further) for a discussion of these questions is provided in diagram 5, which lists the durations of all six movements in 92 different recordings.
The fact that there is a difference of more than 18 minutes between the fastest and the slowest recording (50:22/68:41, see table 5) is hardly surprising given the evidence of diverging Mahler performance traditions cited up to this point. Nor does it come unexpectedly that tempo variability in the slow movements 2 and 6 turns out to be somewhat stronger than in the other four movements. Diagram 5: Mahler, Das Lied von der Erde, durations of all six movements in 92 selected recordings
Table 5: Mahler, Das Lied von der Erde: maximum, minimum, and mean durations as well as proportional standard deviation in 92 selected recordings
It is evident that these absolute values are of limited significance. Diagram 6 therefore aims to complement these data by indicating the proportional values of the individual movements within the overall duration of a cycle (100%). The weight attributed to the
Abschied finale (movement 6) fluctuates between 50.9% (Colin Davis/Jessye Norman 1981) and a mere 43.4% in Klempererʼs 1951 recording with Elsa Cavelti. Colin Davis pursues a particularly interesting dramaturgy in his recording: First, both slow movements are extremely stretched in time – Davisʼ Abschied is the slowest among all recordings with a duration of almost 35 minutes, mostly due to the prolonged recitatives; his second movement is also among the slowest versions (11:11/11:16 in Davisʼs second recording 1988), with only Bernsteinʼs (11:23/1966) and Wolfgang Grohsʼs (11:44/2002) coming in at a slower tempo. Davisʼ concept could not least be due to a tribute to the prominent soloist, providing her with the necessary “time-space” to unfold the unique qualities of her voice. In contrast, movements 1 and 3 are considerably “compressed,” with the 11% proportion of the first movement reaching one of the minimum proportional values and the duration of 2:42 of the third movement the shortest absolute duration among all selected recordings. Klempererʼs 1951 recording, in turn, reaches a very high value for the proportion of its second movement: its absolute duration of 9:00 minutes, though below the average value of 9:33, results in a proportion of 17.2%, the maximum among all selected recordings.
A comparison of these two sets of data with the remaining recordings seems to suggest three different basic dramaturgies:
(1) Davisʼ recording is an example of a strategy in which
Der Abschied is taken as the telos and centre of the work, suggesting a tendency to reduce the other five movements to extended “introductions” to this massive statement; in this concept the two Abteilungen last about half of the entire duration each.
(2) Klempererʼs dramaturgy, in contrast, tries to balance out the weight of movements 2 and 6 by reducing the finaleʼs weight, increasing that of
Der Einsame im Herbst, and further reducing the proportion of the remaining four movements. It is clear that this dramaturgy is not at all incompatible with the first one, since both may focus on the architectonic link between the two slow movements.
(3) In contrast, many recordings testify to a more traditionally “symphonic” dramaturgy in which the durations of movements 1 and 2 are almost equal in length, which implies that the first movement gains prominence and can thus function (ultimately in conjunction with the second movement as a form of twofold introduction to the cycle) as a palpable balance to the
Abschied. This dramaturgy seems obvious in Karajan 1972, Walter 1953, Walter 1960, or Jochum 1963, among others. Diagram 6: Mahler, Das Lied von der Erde, proportions of the six movements in 92 selected recordings
To conclude, let us turn again to the dramaturgies of Davis 1981 and Klemperer 1951 and hear how their contrasting approach materializes towards the “heart” of the final movement as compared to Bruno Walterʼs much-discussed recording with Kathleen Ferrier from 1952.
Adorno implied that a difference in character and expression between the “main tempo” sections (indicating the funeral march tempo of the central section, mm. 288–373) and the recitatives was key to the performed form of this movement, a criterion which Bruno Walterʼs 1952 recording obviously did not meet: “The recitative[s] in the final movement as expressive as the main themes – thus the entire form wrong.”
 Video examples 8 to 10 document the beginning of the recitative (mm. 374–380) at the outset of the second part of the movement, in which the poem by Meng Haoran has been set, soon punctuated by reminiscences of the preceding C minor funeral march (mm. 381–389). The alto recites “erzählend und ohne Espressivo” [narrating and without expression] over a low
C ostinato and accompanied by regular tam-tam strokes. The highly differing articulation of all three soloists, closely though not exclusively emerging from the different tempo conceptions, makes it abundantly clear that we find ourselves in entirely different “narratives” at this point (diagram 7): The funeral march is taken in an extremely hurried tempo by Klemperer (77.8 bpm, fast even when compared to the 68–70 bpm of his recording in Budapest three years earlier), making the recitative (60.7 bpm, 78% of the preceding tempo) appear as a clear formal point of orientation, with the considerable rubato (16.6% standard deviation) only slightly masking the hurried character which this tempo still implies for the singer (video example 8). The only real independence from the implicit urgency is gained in the doubled key word “warum” (“Er fragte ihn, wohin er führe / Und auch warum, warum es müsste sein“), on which Elsa Cavelti performs a considerably “expressive” tenuto and portamento to expand the cadence before the march enters in an unchanged hurried tempo, leading to almost grotesquely virtuosic woodwind figures in mm. 382–386.
In terms of form, Davis and Norman (video example 9) follow a similar basic principle, their recitative establishing a tempo of 40.4 bpm, about 87.4% of the preceding main tempo of the funeral march (46.2 bpm). Although their recitative has a more stable tempo than Klempererʼs and Caveltiʼs, the massive tenuto causes the impression of a “beat” to vanish from the music entirely (repeatedly falling to beats below 30 bpm). The exact opposite impression is rendered by Walter and Ferrier (video example 10): They even increase the tempo of the recitative (60.0 bpm) compared to the preceding march tempo (57.7 bpm), making the recitative appear as a sung intermezzo rather than as the beginning of a new formal and narrative area. Though their readings have different consequences on the overall form, we might interpret Walterʼs and Klempererʼs restraint here as an attempt to save Mahlerʼs music – and particularly the final movement of the
Lied–from the “sentimental” implications it had been associated with early on. Video Example 8: Mahler, Das Lied von der Erde, vi, mm. 374–389: Klemperer 1951 (Elsa Cavelti; Wiener Symphoniker; Otto Klemperer)
Video Example 9: Mahler, Das Lied von der Erde, vi, mm. 374–389: Davis 1981 (Jessye Norman; London Symphony Orchestra; Colin Davis)
Video Example 10: Mahler, Das Lied von der Erde, vi, mm. 374–389: Walter 1952 (Kathleen Ferrier; Wiener Philharmoniker; Bruno Walter)
Diagram 7: Mahler, Das Lied von der Erde, vi, mm. 323–389, tempo graphs Klemperer 1951, Walter 1952, Davis 1981 (logarithmic scaling) 5. Conclusion
The intention here was not to sort out “authentic” from “inauthentic” Mahler performance styles. Rather, our “augmented listening,” particularly when read against the historical evidence introduced in the first two sections, seems to suggest that it is ultimately impossible to define an “adequate” performative interpretation for Mahlerʼs work in a narrow sense. Mahlerʼs objective to “bring out” important details and nuances in performance seems at first glance to be matched particularly well by a slow performance such as Horensteinʼs (for the moment ignoring deficiencies in recording technique which can be observed in many recordings far into the 1990s
). But this surely does not render Bernsteinʼs tempo-based dramaturgy or Klempererʼs “hurried” early recordings “inauthentic.” And neither does it imply that ultimately there are no criteria for distinguishing a “good” from a “bad” interpretation. In terms of the model pursued here, namely performance as “formal analysis in real time,” it seems obvious that the potential for using tempo as a means of marking formal turning points or intensifying formal processes has not only been a part of Mahlerʼs explicit aesthetics of performance as documented in his highly detailed tempo markings and his habit of changing tempo “from bar to bar” but may be traced also as implicit in the musical work and its internally “symphonic” dramaturgy, which might be rendered explicit by conductor (as seen most vividly in Bernsteinʼs example), singer and orchestra. In such a case, the performers indeed seem to act as co-creators who – while doubtlessly believing to a certain degree in an “autonomous” dimension of the work – have gone beyond a simplistic model of “authoritarian” performance practice indebted to a composerʼs intentions. References