• Home
• About CHARM
• Discography
• Sound Files
• Research Projects
• Outputs
  • Publications by Project
  • Publications by Year
  • Publications by Researcher
  • Presentations
  • Changing Sound of Music
    • 1. Introduction
    • 2. Performances
    • 3. Recordings
    • 4. Singing
    • 5. Violin playing
    • 6. Piano playing
    • 7. Style change
    • 8. Expressivity
    • 9. Conclusion
    • Updates
    • Bibliography
    • Captions
    • Indexes
• Analysing Recordings
• History of Recordings
• Download Area

4. Changing Performance Styles: Singing

¶1 Sound File 13 (wav file) is a transfer by Roger Beardsley from a 1911 recording of Elena Gerhardt singing Schubert’s ‘An die Musik’ (D 547), accompanied by Arthur Nikisch. ¹ It’s worth hearing now because it helps us to focus on one of the key questions raised by early recordings. Why did people sing and play the notes on the page in ways that to us seem so strange? How can ‘An die Musik’, to take just one example, ever have made sense to people sung like this? Yet it did. Gerhardt was one of the most celebrated singers of her time, and this is now one of her most valued and valuable recordings (as I discovered when trying to acquire a copy to use for this book). For Eddy Sackville-West and Desmond Shawe-Taylor in 1951, Gerhardt as an interpreter of German song ‘developed a mastery of phrasing, enunciation and tone-colour which have set a standard difficult to approach’, ² and in this recording she achieved ‘superb results’. ³ Yet for John Austin (of Kangaroo Ground, Australia), writing an Amazon review of a modern reissue on October 9, 2000,

Conductor Arthur Nikisch and soprano Elena Gerhardt were musicians of great renown in their day, but their 1911 performance of ‘An die Musik’ is frankly appalling. Nikisch plays the opening accompaniment quickly, then slows to half speed when Gerhardt enters. What follows, for nearly four minutes, is not so much a tribute to music as a travesty of it. Altogether there is little offered by the early 20th Century generation of singers that I should like young singers of today to hear. ⁴

¶2 What has changed? And what does John Austin fear might happen if young singers heard Gerhardt? Was her performance always appalling, and was everybody who so admired her wrong? Or is it just that taste has changed? If the latter, then is ‘taste’ a strong enough word to describe the reshaping of responses to such an extent that a modern writer can fear for the musical health of the young? And if people have changed that much, why might hearing Gerhardt have any appeal for young singers today? What would be the danger in their hearing her? Clearly powerful emotions are involved in responses to performances like this, and one of the things we’re going to need to do in this chapter is to explain how such differences are possible. In subsequent chapters we’ll begin to see why people feel so strongly about them.

¶3 First, though, we need some context. The best way to provide this, I think, is to take a selective tour of twentieth-century performance practices, working chronologically from the earliest recordings, and taking instruments—voice, violin, piano—that have somewhat self-contained traditions before we try to look either at what they have in common or at how performance works on a much more detailed scale. ⁵ With those overviews to hand, we’ll be in a better position to think about the mechanisms that cause these changes in style and about the relationship between styles of music-making and writing about music. Then we can come back to the more fundamental problem of the changing nature of musicianship.

¶4 I ought not to need to justify beginning with singing, where there is text that directs emotional responses more narrowly, and only then moving on to instrumental playing, where there is not. But the prejudice of musical academia in favour of instrumental music (‘absolute’ music), precisely because its meaning is not reified by text, is so long and was until recently so strongly established that I probably should say something about it. Aaron Ridley has dealt with this at length from a philosophical point of view, ⁶ arguing that music particularises text as much as vice versa. That is so; and we shall see as we go along, especially in chapter 8 below, how multiple cues combine to particularise our sense of what any kind of music means to us: ⁷ text is only one source of cues, a very significant one when it’s in a language we understand, but not (as it were) the whole story.

¶5 But my reason for starting with song is simpler than that. I want us to be able to see musical sounds in relatively precise relation to sounds from life before we move on to look at musical sounds whose relationship with anything else is much less definable. What I’m really examining in this book are the things performers do with sound that make music (emotionally) moving. ⁸ My argument is that this happens from moment to moment as performers ‘shape’ some aspect of the musical sound—its frequency, loudness or pitch (or a combination of those)—in relation to the musical score on the one hand and on the other in relation to associations we bring as listeners (and the performers bring similarly) by knowing what sounds like those mean in the environment and in our lives. Understanding the relationship with scores is easy; or at any rate, musicology has been doing it very well for generations and I have nothing to add. But understanding how musical sounds draw on sounds from life is much harder, because the associations are rarely crude; if they were, music would be so commonplace in its expressivity that it would bring no added value: one might as well hear the environmental sounds themselves and have done with it. If we want to make any progress towards understanding what a momentary adjustment of loudness in a Bach partita is doing for us, then we need to gain a lot of understanding by working first of all in musical situations that are much simpler. And this is why I begin with song. Words help us to understand the things singers do with the music that sets them. If we can see some coherent procedures at work there, ways in which musical sounds refer to other kinds of experiences of change over time, then we can make progress towards understanding situations in which there is no text, and no intention to emulate text, simply change in sound over time. For my purposes, aiming to find out more about how music is expressive and of what, song is a very good place to start.

¶6 The other thing I want to do―because it’s a necessary preliminary to understanding how music moves us―is to propose and illustrate a variety of ways of studying musical performances, using recordings as my sources. And so as we go through the next three chapters, as well as showing how style changed I’ll be offering various approaches to finding out what performers are doing from moment to moment, before in chapter 8 we find out how to get down to the smallest details.

¶7 I mentioned when we were looking at the limitations of early cylinders and discs that singers recorded particularly well, especially sopranos and tenors. This was due to the limited frequency response of the equipment coinciding with the most important part of the spectrum for vocal sounds—between about 150 and 2000Hz—and the relatively high energy that a voice could direct down the recording horn compared with other kinds of instruments. So listening to an early recording one could get a more complete sense of the sound of a voice than, say, a piano, whose sound couldn’t be focussed, or a violin whose sound wasn’t strong enough. The instrument most recorded, therefore, in the early years was the voice. Producers, as we now call them—‘experts’ in the Gramophone Company’s terminology—quite literally went out of their way to find and record as many singers as possible in order to build up a catalogue of saleable discs as fast as they could. We’ve already seen Fred Gaisberg taking a year-long trip to India and the Far East in 1902-3, and bringing close on 600 recordings back. He had made similar trips to Russia in 1900 and 1901, and before that to Italy, France and Spain in 1899, and to Italy again in 1902, focusing his search for suitable artists always on singers, albeit with excursions for popular instrumental items. ⁹ Consequently, most of the oldest musicians we have on record are singers.

¶8 On the face of it, that might not seem a very promising statistic, for the oldest voices might not always be the ones we most want to hear. More interesting is that these are the musicians trained earliest in the 19^th century, and for anyone trying, as many do, to extend the evidence of recordings back into a time before recording was invented, this is particularly tantalising evidence. We have to be very careful, though. Is an old voice fairly representative? Do musicians change their performance styles with the times? (We’ll come back to this later, especially in chapter 6 on piano playing: ¹⁰ broadly, some do and some, perhaps most, don’t, or not very much. But of course this conclusion is based on 20^th-century evidence, and patterns could have changed.) At any rate, it seems fair to assume that we know more about singing than anything else from the later 19^th century , and much more than we do about, say, orchestral playing. The earliest recorded singers tantalise us, therefore, and it’s very tempting to suppose that the earliest sounds we hear from them represent a tradition going back towards composers—Brahms, Wagner, Schumann, Schubert, Beethoven—with whom the western classical tradition remains obsessed but about whose performance we otherwise know with certainty little (Brahms and Wagner) or next to nothing (anyone earlier).

¶9 The other tantalising factor, though it may be misleading, is that vocal performance styles changed early in the recorded era; indeed, they probably had already changed for many singers before recording began, so that only the oldest singers seem to belong to the older style. It’s very tempting to suppose, therefore, that those few represent a tradition going back a long way. Several very interesting attempts have been made to relate their styles to earlier 19^th-century writings on singing, ¹¹ and one can certainly make a plausible argument along those lines. But arguments are one thing, and recorded sounds quite another. The former, in the end, are wishful thinking (though possibly correct wishful thinking), the latter are evidence. So if we’re to give proper weight to the evidence we do need to be cautious in arguing back. ¹² That said, what we hear from these oldest recorded singers is very fascinating.

¶10 Let’s begin with perhaps the most famous. Adelina Patti was born in 1843, and since she was already performing professionally in her teens it’s likely that her performance style was fairly well-formed by 1860. Her recordings, though, were made in 1905 and 1906 after she’d stopped giving public performances. ¹³ It’s obvious that her voice is not what it was; she finds it hard to sustain long phrases without taking a breath, high notes can screech, chest notes can be rasping, and from time to time a note breaks altogether, but these flaws due to age are less striking than the qualities that remain. ¹⁴ Outstanding among them is the ease with which she gets from note to note, even in the fastest decorative passages. The impression of fluidity this produces results from very short onsets for each note she sings; her voice hits each pitch precisely, its full tone and volume reached very quickly, so that the sequence of fully-voiced pitches is hardly interrupted at all by the formation of each new note. Partly this is a matter of forming consonants very briefly; and sometimes they can be so brief as to be inaudible, in which case diction suffers and we seem to be hearing vocalise rather than sung text. Partly it is due simply to the fast reaction time of her vocal chords. In addition, in melismas she joins notes together with rapid portamento, even on repeated pitches which she typically manages by scooping down very fast at the end of the first note and up rather less fast at the start of the second.

¶11 Together with this goes a clear and even tone, caused by strong first and second harmonics (the fundamental and octave), a slightly less strong third harmonic (twelfth), and strong partials in the region between 2500 and 3500Hz. How bright it was we can no longer know for sure, since the recordings transmit no frequencies above 4000Hz. Hanslick, Wolf, and Emmanuele Muzio (writing to Verdi) all described it as ‘silvery’, ¹⁵ a metaphor usually associated with brightness caused by a few strong upper partials, so we may well have lost something significant (Muzio thought it already lost in 1886), but it may also be meant to convey an element of fluidity, so that we can’t know the balance between their sense of her voice as limpid and as bright. But even so, we can easily agree with 19^th-century critics who found it ‘melting’, ‘liquid’ (Emma Eames), ¹⁶ ‘flexible’ (Charles Dickens and John Cox), ¹⁷ and above all ‘pure’ (Eames, Camille Belaigue). ¹⁸ Her timing is relatively exact, at least from one bar to the next, although there is flexibility from note to note. ¹⁹ It is a voice that emphasises line, continuity and melody much more than drama or rapid changes of emotion. Beautiful singing seems to matter more than emphasising remarkable moments in the text. Generally, most of the expressive work is done by changes in amplitude, which are wide and surprisingly well recorded for this date, and by much more subtle variations in timing from beat to beat. Less commonly she scoops up to notes—the technique long known in the teaching tradition as cercar la nota—in order to emphasise specific points in text or musical structure. ²⁰ Portamento across pitch intervals she uses quite a lot, but in some songs much more than others; her vibrato is fast and by subsequent standards shallow, so much so as on occasion to be barely noticeable. In its general character her singing is unaffected and simple, and much more like modern early music singing than was the singing of the generations immediately following her.

¶12 How much of this is really about Patti’s voice and how much about the sound of CD reissues is a question that’s complicated by the impossibility of knowing the speed at which the original discs were recorded. As I write I have six transfers to hand, and a vinyl pressing from the original metal, of matrix 683c (G & T 03084), a recording of ‘Ah, non credea mirarti’ from Bellini’s La Sonnambula made at her home in Wales in June 1906. Three transfers agree on a pitch for the key-note of 414Hz (modern a’-flat); ²¹ one is pitched at 407 (flat a’-flat); ²² two (including a CD transfer of the metal) at 392Hz (modern g’); ²³ and one at 430 (a flat modern a’). ²⁴ There is almost a tone between the highest and the lowest, and none is necessarily correct, for reasons we saw in chapter 3. Patti’s voice sounds very significantly different at the highest and lowest of those pitches. It’s always tempting to accept the compromise (which is what I’ve done for the description of her voice above, based on the Symposium transfer), and as you might expect it does indeed sound less girlish than the highest transfer, and less matronly than the lowest, but there is no way of knowing which was Patti. We simply have to accept that we can’t know for sure how she sounded or whether the comparisons we make with other singers are quite accurate. ²⁵

¶13 Patti was taught, like many singers of her generation, by Manuel Garcia, son of the Manuel Garcia who was one of Rossini’s singers. Garcia the younger wrote up their teaching in Traité complet de l'Art du Chant, Paris, 1841 and 1847. ²⁶ Both here and in the later summary of his teaching, as learned at first hand by Hermann Klein, all the emphasis is on steadiness and purity of tone, managed through even breath pressure, open vowels, voice production forward in the mouth, and above all perfect legato producing sequences of equally powered and coloured notes strung seamlessly together, enhanced by lightly-sung portamento. ²⁷

¶14 We can get another angle on this tradition of fluid melodic singing by listening to one of the oldest recorded musicians, Sir Charles Santley, born in 1834, so aged almost 70 when he was recorded in 1903. Though English, Santley studied in Italy in the 1850s and then with Garcia in London, and so it is not surprising to find him singing in this same legato style. A fine example is his recording of ‘Non più andrai’ from Mozart’s Le Nozze di Figaro. According to Hanslick his voice was already aging in 1886, ²⁸ and by 1903 we are clearly listening to an old man; but vocal colour is not by any means the only factor determining our sense of a singer’s skill: among others that shine though Santley’s performance are steadiness and consistency of tone, subtlety of rubato, and this same ability as Patti has to sing legato without being shapeless. ²⁹

¶15 The steadiness comes from an even vibrato, slightly slower and considerably wider than Patti’s, as is characteristic of male compared to female singers around 1900; consistency of tone is modified only by the darkened vowels of his pronunciation (lower and further back in the throat than one would expect today); the rubato is very slight, and corresponds to stresses in Italian rather than to meanings of words; the legato is achieved by breath control, of course, and by the ease with which notes sound, just as in Patti, but unlike Patti Santley manages what one might think an almost impossible task of articulating legato, done by decreasing the loudness of articulated notes at their ends but without stopping them entirely, and on occasion by simultaneously sliding below the next note in order to crescendo and scoop into it: the scoops are fast enough that one perceives them not as pitch change but as emphasis. When he wants something more continuous he has graded options to hand, at the first level continuing the tone without diminuendo, at the second joining notes with portamento (though he does this rarely). ³⁰ The most expressive words are sung out, when there is time, and with more vibrato (for example, ‘molto onor’) and, when there isn’t time, with a small fast scoop.

¶16 The other fascinating detail, more discussed by the critics, is his ornamentation, which is extremely fast, flexible and integrated into the legato by widening two vibrato cycles far enough to reach an adjacent pitch. ³¹ Surprising only because the character of the performance is brisk and bracing (as the text suggests) is the cadenza at the end of the B-section, which uses Mozart’s pause sign to leap up a fifth, hold it, run down a scale through an octave and a half to bottom G, and then slide back up to the G from which he began, before taking a breath and launching into the pseudo da capo. By articulating the descending scale more than usual Santley manages all this without a trace of sentimentality; which is a useful lesson in the integration of ornamentation into a performance: done in the right way it has none of the sense of indulgence that is too often attributed to early 20^th-century singing by writers unattuned to the stylistic context.

¶17 Compared to our own expectations this performance seems informal, light-hearted, fluent, easily achieved (though at 69 years old it cannot possibly have been), and in a way natural. This last word is a tricky one, of course, and it can only be used as a comparative—natural compared to what we’re used to. Untrained would be another way of putting it. But in fact there’s nothing untrained about these singers. They simply make sounds that are less unlike amateur singing than trained singing became in the generation that followed them. And the reason for this change, above all, was orchestrally conceived opera. As singers from Verdi onwards had to make themselves heard over a growing orchestra—as Wagner, Puccini, and then Strauss required—vocal training had to develop a sound that would produce strong harmonics relatively high up the spectrum. The so-called ‘singer’s formant’, developed in Italian tenors and spread rapidly, carried effectively over much larger accompaniments, but at the cost of lost flexibility: the energy required from each note could not be generated so quickly; notes take longer to reach their full amplitude and they require wider vibrato to distinguish themselves from the instrumental sounds around them. The marked style change that occurred between this generation, therefore, and the next—singers who grew up with verismo and Wagner—was caused more than anything by the demands of composers. We shall see that this is far from always being the case.

¶18 I want to look at one other elderly singer from the first decades of recordings in order to focus for a moment on the question of how texts were understood and how the meaning found in them was conveyed in the early 20^th century. Texts, and especially song texts―whose poetry was often more subtle and more capable than opera libretti of bearing a range of interpretations―did not always mean the same things to singers then as they mean now. And it’s possible to see how meanings have changed in line with changes in performance style. Indeed, the two are inextricably bound up together.

¶19 Sir George Henschel, born in 1850, was a singer, pianist (he studied with Moscheles), composer and conductor, in fact the first conductor of the Boston Symphony Orchestra. ³² As a singer he accompanied himself, which gives his recordings of songs a particularly interesting quality, with coherence between voice and piano both in timing and interpretation. In his teaching Henschel laid particular emphasis on the expressive potential of vowels. ‘..the five vowels ... are what we may call the primary colours of the voice. By skilful and judicious mixing of these colours, a singer should be able to produce as many shades of (let us say) the vowel A as a painter can produce of (let us say) the colour red.’ And he used to make his pupils sing whole songs on vowels alone ‘with the object of expressing the character of the music by mere vocalisation.’ ³³

¶20 Later on in this study we’re going to look at expressivity in Schubert songs and so it’s two of Henschel’s Schubert recordings I want to focus on here, ‘Das Wandern’ (the first song from the cycle Die schöne Müllerin) and ‘Der Leiermann’ (the last from Winterreise). The first was recorded for HMV in 1914, the year of his retirement, and again for Columbia in 1928 (we shall look at their very close similarity in chapter 8); the second was recorded as a coupling for the Columbia release. They were singled out for particular praise in the first book-length study of Schubert’s songs, by Richard Capell, first published in 1928. ³⁴ In a list of approved recordings, without much comment, Henschel’s Columbia discs get Capell’s most enthusiastic endorsement: ‘These records show, no doubt, how Schubert intended his songs to be sung. There is no show and no self-consciousness about this singing. The performance strikes the right balance between voice and piano’, as well it might with Henschel accompanying himself. ³⁵

¶21 Henschel’s seems to us now, as it did to Capell, a straightforward approach. In ‘Das Wandern’ (Sound File 14 (wav file)) ³⁶ the miller’s boy feels precisely what he sings: nothing could be more delightful than to go travelling. The jauntiness in Henschel’s sound is produced by making the notes short, attacking them hard and cutting them off early, and by raising and lowering the pitch. On a spectrogram each note is an inverted U, with a rapid crescendo to the top and equally fast decrescendo down. The sense a listener gets is of top-of-the-world good humour: there are no hidden meanings, no self-deception; the boy is looking forward to a delightful adventure. For Capell the boy at the start of Die schöne Müllerin is ‘a lovesick lad in a green valley.’ ‘Schubert’, he says, ‘simply did not know what to do with the bold and the bad of the earth. But he lent his luckless young miller tones that he could not have bettered if he had wanted them for himself. And surely just such a one would he himself have been if he had fallen to such a milleress’s charms; timid and rapturous, flower-plucking and star-gazing, a fount of tenderness, a gulf of despair.’ ³⁷ It’s easy for us to laugh at this, just as we sometimes laugh at the singing of 90 years ago. But this is how it seemed. A poet’s love was pure, generous and honest, and it was in the same spirit that one should sing. And so at the far end of Müller’s story, in ‘Der Leiermann’ (Sound File 15 (wav file)), ³⁸ Henschel responds by narrowing his vocal cavity, holding it almost immobile, pulling all the vowels towards the [ᵄ] (‘arr’) position, maintaining a monotonous colour and level to suggest the stasis of profound depression. Nothing could be more obviously contrasted with his performance of ‘Das Wandern’; ‘Der Leiermann’ becomes, as Capell recommends, ‘An almost toneless song’ and Henschel ‘an unforgettable interpreter’. ³⁹

¶22 In relating Capell and Henschel it’s worth remembering not only that Capell’s was the first study that considered Schubert’s songs worth examining in depth—earlier writers on Schubert tended to dismiss most of them as relatively trivial—but also that Capell’s much more sympathetic view of them comes after some thirty years of recordings which had made at least some of them better known and widely loved. ⁴⁰ It’s an interesting early example of how recordings changed the way music was understood. ⁴¹ What had been a reflection of Schubert’s superficiality was now a tribute to his integrity, thanks in large part to the manner in which his songs were being sung. It’s not hard to see how these two attitudes fed on one another. The conviction of singers increased that of writers, and theirs in turn fed back into an increasing intensity in performances.

¶23 So during the 1920s and 30s we tend to see a gradual inflation in the emotional intensity of Schubert song performance on record. Of course this is not an isolated style change. It happens, too, because of a general trend towards more emotionally expressive performances in all kinds of music during the first four decades of the 20^th century. The change in vocal production, brought about by verismo and Wagner opera, is another factor, and together they all play a part in the manner of singing we find in the next generation. All these factors―singing, playing, writing, training―were shaped by the general communicative norms of their time. Modernism was, to a considerable extent, a reaction against precisely this sort of emotionalism. ⁴²

¶24 We can take Lotte Lehmann as a first representative of this more intense style. ⁴³ Born in 1888, Lehmann was in her early career best known as an opera singer, especially in Strauss and Wagner. Her discography runs to well over 400 items, not including interviews and master-classes, reflecting her long life (she died in 1976) and enthusiastic engagement with recording. ⁴⁴ Lehmann was appreciated not so much for the quality of her technique, which anyone could hear was flawed, as for her exceptional ability to communicate feelings through her voice. And in a way, the imperfections in her singing contribute to her effectiveness as a communicator. One of the variables is the colour of her voice, which runs from harsh at the bottom (strong in the fundamental and in dissonant upper partials) to childlike at the top (strong lower harmonics only, giving a pure tone almost like a treble). ⁴⁵ In the normal course of events training evens this out, but Lehmann’s training, at least by her account, seems to have been unsatisfactory in the extreme. Together with changing tone goes a vibrato that is even in wavelength but uneven in pitch.

¶25 To a listener the sound seems to suggest a curious mixture of stability and instability: we sense warmth (the strong consonant lower harmonics), dependability (the regularly beating vibrato combined with warmth of tone) and yet vulnerability. The pitch instability is there not just within the vibrato but also in a tendency to drift upwards very slightly during a note, too little to be perceived as sharpening, but enough to sound yearning, as if her voice is reaching towards the listener. A sense of eagerness is increased by her tendency to swoop up to notes at the start of a phrase, and to slide up from note to note in a rising melodic line, sometimes right through a note as she passes up the scale. Her approach to linear continuity comes not so much from very fast note onsets, as in Patti, as from portamento and these other kinds of pitch slides. Indeed, because her breathing was never well controlled, she breaks up lines into shorter-breathed units more than most. ⁴⁶ Yet she was uncommonly valued and loved as a singer who seemed to involve her audience in every nuance of a song.

¶26 It may be that listeners felt happier then with humanly imperfect voices than we do now after a long period in which evenness and clarity have been prized most highly. But to understand Lehmann’s effectiveness as a communicator we need to look also at the advice she gave to students on interpretation. Teaching books for singers from the 18^th, 19^th and early 20^th centuries were concerned almost exclusively with technique, and based around exercises. Lehmann’s, and those of a handful of other singers from the 1920s through the 40s, were much more inclined to advise on interpretation, and especially on what feelings a singer needed to convey. ⁴⁷ Here she is on Schumann’s ‘Ich grolle nicht’: ⁴⁸

Change the quality of your voice which has been dark and flowing, at ‘Wie du auch strahlst in Diamantenpracht’. Sing with a bright tone, disparagingly and ironically, as if you were saying: “But don’t think I don’t see through you!” ... Sing broadly, with sorrowful accentuation ‘das weiss ich längst’. ... Turning away from the beloved, still trembling from your outbreak of bitterness, you now speak more to yourself. ... completely absorbed with yourself you repeat, trembling, ‘Ich grolle nicht’. Beginning this verse with a restrained piano will also give a stronger effect in building up the dramatic climax of the song.

¶27 What we get here is a mixing of advice on vocal colour and dynamics―more or less technical points―with description of how the character portrayed feels within the story, but expressed as advice on how the singer himself should feel: the singer, in other words, becomes the protagonist; the more successfully, the stronger the musical result. It’s method acting for musicians. And for Lehmann it clearly works. A recording of her reading this text (Sound File 16 (wav file)) ⁴⁹ corresponds exactly to her advice to singers and makes interesting comparison with her recording of the song from June 1930 (Sound File 17 (wav file)), ⁵⁰ for the similarities between them emphasise the extent to which this highly expressive, deeply felt singing, calls on styles of acting for its communicative power.

¶28 Into this same stylistic world comes Elena Gerhardt’s recording of ‘An die Musik’ with Arthur Nikisch, the recording with which this chapter began. Here again, the performance decisions are not simply musical; in this case they are not even led very directly by Schubert’s notation, whose repeated quavers look so even. In Gerhardt’s recording, quavers in the first two lines of the song vary from 0.34s to 1.04s in length, with most distributed fairly evenly between 0.4s and 0.8s. In other words a quaver can vary in length by a factor of 1:3, and quavers frequently vary across as little as a beat by a factor of 1:2. Whatever drives this performance it’s certainly not the notation. But if we understand it as Lotte Lehmann would then it makes much better sense. ‘An die Musik’ is about music itself, its solace, its power to change one’s mood from despair to ecstasy; it’s about the most profound topic of which a singer can sing. Gerhardt isn’t singing the score, she’s singing her feelings for music—or at least, acting them out. In a performance-stylistic world in which intensity of feeling is everything, her rubato makes perfect sense. (We’ll look at this example more closely in chapter 8.)

¶29 Heinrich Schlusnus was an exact contemporary of Lotte Lehmann, born in 1888, ⁵¹ and it’s useful to compare his singing with hers to see whether men manage this intensely expressive style differently. His voice is much more controlled, both naturally and artistically; I suppose one could say, adopting terminology of the period, more manly (strong, direct, but not insensitive). ⁵² The intensity is there all right, but less in tempo flexibility or pitch scoops (generally less used in male singing) than in amplitude and colour. His 1928 recording of ‘An die Musik’ (Sound File 52) (wav file) makes a useful comparison with Gerhardt’s; ⁵³ he’s really quite strict with tempo, only lengthening notes markedly right at the end, ‘Ich danke Dir’: and there, as he thanks Music for all it’s given him, he does adopt almost as overt an emotionalism, fervently stressing those words, as does she. But for the rest, the intensity comes less from tempo or sudden change than from increasing amplitude to the high-point of a phrase and then quickly releasing it, and from an interesting ability to change the depth of his vibrato (increasing it for expressive passages) and the balance of harmonics, bringing in harmonics not required by the vowel in order to make a note glow. The resulting impression is of stability—the manliness—coupled with strength of feeling, with the emphasis on the strength; where Gerhardt or Lehmann are willing to let pitch or duration off the leash, Schlusnus simply adds to what he already does without any parameter ever cutting loose. Nevertheless, the effect is of an intensity that clearly belongs within the same emotionally expressive world: it could never be mistaken for a younger Henschel or an earlier Fischer-Dieskau. (Though if he’d made a more beautiful sound the younger Peter Pears might have come close.)

¶30 If we look back at Patti, now, we can easily see that there has been a very considerable change in performance style since the beginning of the century. Many features are shared, including features we no longer hear today—portamento and wide scoops especially, and because those are so obviously strange to us now we tend on first hearing to group all these singers together as quaint, or even tasteless. But the ways Patti and Lehmann, or Santley and Schlusnus, use their voices are actually very different. For the oldest recorded singers linear continuity is everything; beautiful singing, expressive of the general sense of a text, is much more important than following and communicating moment-by-moment changes in the emotional state of the character they represent, which for Lehmann and Gerhardt, and in a more restrained manner, for Schlusnus, was what singers were there to do. Different performance styles, and different conventions of emotional communication, go hand in hand, and music accordingly takes on different functions and to a certain extent even comes to mean different things.

¶31 Why the Second World War marked a watershed in musical performance style remains to be investigated; but it did. After the War that intensely expressive style seemed hopelessly old-fashioned, exaggerated and unrealistic. Perhaps it was simply that naivety was impossible after the discovery of the concentration camps, so that singers of the generation who’d come to adulthood during the Nazi period found they could no longer represent 19^th-century love poetry without a degree of irony. It’s not coincidental that Freud finally became a significant influence in popular German thought during the post-War decades. Although psychoanalytic research had continued through the state-approved Göring Institute during the Nazi period, a new generation of German psychoanalysts considered themselves to be starting afresh in 1945. Institutes of psychoanalysis were founded between the late 40s and the mid 60s, and psychotherapy gradually established itself within the German healthcare system between the mid-50s and mid-70s. ⁵⁴ Young Germans after the War were in effect the first generation for whom psychoanalysis offered an obvious way of understanding human behaviour. It’s not hard to see how in this context, especially given the weight of guilt and insecurity about the recent past, a new generation of singers would tend to read opera libretti and song poetry less literally and less innocently than their predecessors.

¶32 This proves to be less true of Elisabeth Schwarzkopf than Dietrich Fischer-Dieskau, whom I shall look at in more depth in a moment. Schwarzkopf (born 1915) grew up in the early 30s and prospered during the Nazi period, singing with Deutsche Oper in Berlin during the War and afterwards at the Theater an der Wien. Apart from the qualities of her sound, which is unusually even across the spectrum, her singing is characterised by strong expressive gestures, often using indrawn breaths and drawn-out final consonants for emphasis; and in that general sense she can be (and has been) seen as a direct successor to Lotte Lehmann. ⁵⁵ On the other hand—and it’s this that really marks her out as a singer from a different world—there is an emotional distance that would have been incomprehensible to Lehmann. She is particularly good at being skittish and sly—a side to her art that always appealed to audiences—by using rapid swoops during notes.

¶33 Deep emotion, on the other hand, she tends to signal rather than to live in the way that Lehmann believed was essential; and perhaps this is one response to the inhumanity of the age in which she matured. Characteristic signals include changing vibrato width from narrow when singing softly (sounding timid) to wide when loud (commanding); pitch variation within notes, especially sharpening through a note to suggest increasing intensity of feeling; gaps between notes, covered over by starting the new note on the old pitch and then very quickly gliding to the new, producing in a song a sense of an unconfident protagonist feeling her way along; and, above all, these loud and long noises for final consonants and breaths, used to suggest alarm. Her singing at these moments comes closest to speech and, because one has the strong sense that this is stage work with the voice, closest to acting. We never make the mistake of thinking that she as the singer feels any of this herself.

¶34 At the same time, none of these effects is so spontaneous or heartfelt that it intrudes on the beauty of her sound: it’s as if she’s happy to play a little, but never with such abandon as to risk getting muddy. And however speech-like the starts or ends of notes may be, the rest is musical through and through: there may be a lot of pitch movement within notes, and scoops between them, as in Lehmann, but the unwritten pitches used along the way tend to be scale notes, not random frequencies that happen to be nearby, so that one gets a sense of musically constrained expressivity, not like real speech; and this too may contribute to the sense that this is formalised—actorly—expressivity. ⁵⁶ In other words, however lovely the sound and however expert she may be at representing texts, Schwarzkopf sings with a detachment that would have been disappointing to a listener brought up on Gerhardt or Lehmann. Eddy Sackville-West and Desmond Shawe-Taylor, in their 1951 survey of music available on record, and still more in their 1955 revision, when more of her discs were available, give very much this impression: that she sings beautifully but has some way to go before she acquires all the abilities of Lehmann or Elisabeth Schumann. ⁵⁷ Of course, they were not to know—one never does while style is changing around one—that Schwarzkopf’s deficiencies were in fact central features of a new style.

¶35 Dietrich Fischer-Dieskau was ten years younger, born in 1925, and so his early development as a singer came during and immediately after the War. ⁵⁸ For a singer, these are the years spent finding ways of using one’s developing voice and technique to communicate narratives and feelings in a manner that seems right for one’s time. Normally, what seems right is likely to be fairly similar to what one hears from singers a decade or two older than oneself. But for that to be true one needs a stable context, both for study and early professional development, and also for a shared sense of how music can best represent the feelings of people around one. It’s hard to imagine how anyone with the slightest ability to empathise—and one can hardly be an engaging singer without it—could grow up as a performer in Germany during the last years of the War and the first of the Allied Occupation, and sing as if nothing had happened. For a thoughtful young singer the naive expression of poetic emotion, in which Lehmann, Schumann, Schlusnus excelled, would have been almost impossible. The context surely demanded a new understanding of music’s ability to evoke unease (at least).

¶36 If we want to understand Fischer-Dieskau’s very characteristic performance style, that’s one factor to consider. Another, of course, is the voice itself. Because a voice develops not just on its own, but in relation to the sounds a singer needs to make, one can’t entirely separate it out from everything else. But even so it seems safe to say that Fischer-Dieskau’s voice favoured drama and dark emotions because of its exceptional dynamic range from quiet and dull (upper partials cut right down) to very loud and bright, in either case with evenly attenuated harmonics from bottom to top, which makes it sound rich and smooth, but also capable of sudden change within a wide range of colour and dynamic. Much of this would have been developed through training and preference, and preference is certainly a prime factor in his very well-known taste for what Walter Legge called ‘his Aussprache beim singen [pronunciation in singing], the Prussian exaggeration of consonants’, which, Legge said, slightly irritated him, as it has many others. ⁵⁹ Their variety is fascinating: long drawn-out consonants, both at the starts and ends of words, allow him to linger over the ideas those words represent, sometimes eerily, sometimes longingly, but always suggesting that more is to be read into them than one might think; explosive consonants give words the force of irresistible energy or a ruthless command. At the same time, the colour and character of his voice can vary from seductively female (lower harmonics only, slides into and out of significant words, consonants almost silent) to overwhelmingly alpha male (every harmonic so strongly present that the upper partials are a blizzard of noise, acoustically barely distinguishable from consonants). All these effects, it’s easy to see, borrow from the sounds of speech that carry these connotations. To extend Legge’s phrase, Fischer-Dieskau declaims through his singing, at the same time bringing deeper and often more disturbing meanings to the words and music than singers of an earlier generation. Consequently we begin to perceive subtexts in songs and arias that would have been thought unseemly, and unhealthily imaginary, before him.

¶37 Like Schwarzkopf, then, he uses speech sounds as expressive gestures, but uses them with less innocent intent. And it cannot possibly be coincidence that over the next twenty years, as Fischer-Dieskau becomes accepted as the gold-standard of Lieder singing, writers begin to find deeper meanings in song texts than were dreamed of by Capell, or Henschel, or Lehmann. Let’s take an example. For Lotte Lehmann, writing in 1945, Schubert’s ‘Am Feierabend’ should end with passionate impatience, ‘the impatience’, she says, ‘of one in love, which causes the bystander some quiet amusement’, giving way in the final bars to ‘dreamy yearning’. ⁶⁰ The listener, in other words, sees the story for what it appears to be, a touching instance of youthful ardour. For Fischer-Dieskau, in his own book in 1971, writing for a culture in which Freud was now ingrained, the energetic music ‘expresses the lad’s fanatical desire for work’, the return of that music at the end gives the piece ‘psychological depth which clearly goes far beyond the poet’s intentions’, and the final bars express ‘not only weariness, but also a deep yearning.’ ⁶¹ See how Lehmann’s ‘dreamy yearning’ has become ‘deep’ here: everything about this music has been ratcheted up a notch, a wide notch.

¶38 And that’s exactly what we hear in Fischer-Dieskau’s recordings of ‘Am Feierabend’: already in 1951 (Sound File 18 (wav file)) featuring powerful contrasts (the descriptions of female/male sounds above is based on this recording); ⁶² by 1971 (again with Moore, and there had been another with him in the interim), the sounds are still more varied, evocative of a wider range of shifting emotions, emphasised by skilful use of the microphone which can bring moments of intimacy and of distance in unpredictably shifting patterns. ⁶³ And it’s from the early 70s on, once Fischer-Dieskau’s view of Schubert, sounding on record and described in his book, had come to seem normal, a model even, that we begin to find commentators speaking more often about Schubert songs in terms of drama and psychological disturbance. For Graham Johnson ‘Am Feierabend’ shows the miller’s boy sealing himself off ‘into his own world of fantasy and longing’, his mood shifting from ‘moonstruck fantasy’ to ‘frustration’, the song as a whole characterised by ‘healthy physical activity combined with unhealthily suppressed feeling’ (this in the liner notes to a recording in which the retired Fischer-Dieskau appears as a speaker, photographed giving Johnson the benefit of his experience). ⁶⁴ It’s conceptually still some way from here to Lawrence Kramer’s recent view of the miller boy as a masochist and ‘Am Feierabend’ as wish-fulfilling fantasy, the miller’s daughter, later in the song, ‘abolishing her father by reenacting his music in an emotionalized, dephallicized, ambiguous form’, as Kramer says. ⁶⁵ But could we have got there without a previous performance tradition in which psychologised readings seemed essential?

¶39 We see especially clearly here something that may well be common, perhaps the norm: changes in musical performance style cause changes in the way listeners and then writers understand the music. For a popular, widely-performed repertory (the situation with new or rediscovered music may be different), conceptions of what composers’ work meant can hardly be kept separate from how the music sounds in concert and especially on record. And in general the influence is going to be from performers to critics, not the other way round, both because it’s easier for performance style to change unnoticed—we’ll see why in chapter 7—so that critics don’t even realise they’re being influenced, and also because on the whole critics listen to more performances than performers read academic books; and this is especially true now that we have recordings.

¶40 Fischer-Dieskau represents a particularly strong instance of this. His influence has been extraordinary not just because he had a remarkable voice, but also because he thought about texts and their musical settings in a way that reflected with unusual sensitivity currents in thinking about the emotions, and especially thinking influenced by psychoanalysis. Given his background and his intelligence that’s hardly surprising; but one needs to see it working together with general trends in performance style away from expressivity achieved through pitch (portamento) and timing (rubato) towards expressivity through declamation and dynamics. A singer with a wide range of power and colour, a period-style inviting him to explore just those features for their expressive potential, a disposition to see the dramatic and dark side of the pieces he sang, and an interest in psychological understanding of human behaviour (which a thoughtful German of his generation could hardly avoid), would be someone with the power and motivation to change radically the way singing developed.

¶41 Fischer-Dieskau is exceptional, of course, but fascinating for the clarity with which he enables us to see the complex interactions of factors that bring about performance style change: voice, personality, historical and intellectual context with all that it implies for the ways people tend to think about communication and the expression of feeling, the performance style of his immediate predecessors (the Schwarzkopf rather than the Lehmann generation in this case), musical performance depends on the interactions of all these, and it’s hard to say that any of them plays a less important role than another.

¶42 There was another factor that affected performance style after the Second War, and it’s been much discussed in recent years with good reason, ⁶⁶ and I’ve already hinted at it above. It’s recording itself. The general thrust of recent discussion of the influence of recording on performance has been that recording tended to flatten out the variation that once existed between styles of performance practised in different countries. When recording began, singing by Germans was very different from singing by the French or Italians; orchestras, usually because of different traditions of wind instrument building and playing, also sounded considerably different; and so those national traditions could be heard also in solo playing. There was diversity, in other words, and as its influence spread recording began to even all this out, so that by the latter part of the 20^th century orchestras sounded almost identical, save for a few local traditions deliberately maintained, and singers sounded much more alike than they used to: ‘international opera’ was now so much the norm that even Germans in Verdi or Italians in Wagner raised few eyebrows.

¶43 It’s become the tendency in recent years, among writers on the history of performance, to regret this homogenisation, which seems very fair. ⁶⁷ But I’d like to suggest that what’s really happened is not quite so straightforward, nor so regrettable, sad as is the loss of a diversity of national traditions. We’ll examine this more closely in chapter 7, when we draw out of the discussions of performances in this and the next two chapters some conclusions about style change. For now it’s enough to suggest that the continuing development of style is itself a form of evolution that introduces fresh approaches to interpretation and fresh meanings to compositions as it renews performance from generation to generation. It’s not at all clear that we should regret that. It inevitably involves losing other approaches along the way, but it’s going to take a lot more than 100 years before we can confidently say that in the medium to longer term anything is declining.

¶44 One thing that recording has undoubtedly caused—and again we can hear it happening—is a trend towards the literal performance of scores. Sometimes this is described as greater accuracy, but whether it’s any more accurate to be literal is a question that takes us unhelpfully back to ontology and the composer’s intentions. So let’s just say more literal. For singers and string players that means less portamento, less rubato, less ornamentation; for pianists it means synchronising the hands so as to play all the notes of a chord together, playing in stricter time, and forgoing doubling notes and the elaboration of scales and arpeggios—removing, in other words, all the things that musicians used to do as a matter of course in order to intensify the expressivity of a performance. As in so many other respects, the post-War generation marks a watershed in these habits too.

¶45 So if we look at the generation of singers born ten years or so after Fischer-Dieskau—for example Janet Baker and Elly Ameling, both born in 1933; Peter Schreier and Nigel Rogers, born 1935; Arleen Auger, born 1939—we can easily see how the pursuit of perfection led to performances that were increasingly regular in all their dimensions. And this is probably why the speech-influenced expressivity of Schwarzkopf and Fischer-Dieskau was not much developed by their successors. Although Peter Schreier was a striking exception, using whispers and semi-spoken syllables at moments of particular intensity, on the whole the generation born in the 1930s cut down on extra-musical sounds, relying instead on vibrato and intensity to do most of their expressive work. For particularly evocative words and moments in a score Janet Baker still uses the occasional portamento (especially in opera) and shapes dynamics and timbre; but evenness and consistency of sound and manner are the norm, avoiding anything that might stand out from its surroundings and—the horror of every modern record producer—draw attention to itself on repeated hearing. Beauty of sound and line—those characteristics of very early recorded singing—again become the main focus for attention, albeit for quite different reasons now.

¶46 Arleen Auger is especially interesting because she changed style in the 1980s by cutting down her vibrato and narrowing her expressive range in order to conform to the ideals of the ‘historically informed performance’ (HIP) movement. The difference is encapsulated in her two recordings of Schubert’s ‘Gretchen am Spinnrade’, in 1978 with Walter Olbertz, and in 1990 with Lambert Orkis playing a fortepiano, ⁶⁸ and the comparison shows particularly clearly that HIP was, for singers at any rate, no more than a continuation of trends evident for decades: more literal performance of the notation, a tendency to regard anything, even vibrato, as an ‘imposition’ (a much favoured criticism at the time) on the score, and the score to be identified (save for documented conventions of ornamentation or notes inégales) with the composer’s intentions. Auger had to make relatively few modifications to her way of singing in order to work with a fortepiano according to (modern) period norms, because as a child of her time she was already singing in a much plainer fashion than her predecessors. ⁶⁹ In a sense, the taste for imitation boy sopranos which began in England with Emma Kirkby in the mid-70s is only a further intensification of the same trend, narrowing the sound still further, maintaining constant speeds and dynamics and the most even tone possible: modernism in sound, you might say, ⁷⁰ and very late in the day.

¶47 Among the many fascinating aspects of the HIP phenomenon was that it was probably the first occasion in the history of music when a change of style was intentionally manufactured by performers and widely adopted. Indeed it was so successful that it came to dominate several generations of musicians in an increasingly wide repertory. ⁷¹ It was a change that could never have happened without recording and the dissemination of style that it encouraged, but even so it conforms to the general principles of style change that we’ve been seeing, away from expression through modifying pitch and timing, towards expression through dynamics and articulation. We’ll look at it in more detail in a moment when we examine string playing. Among singers—and Auger illustrates this very well—the characteristics of HIP were much less marked. Learning to play an instrument that behaves significantly differently from its modern descendant is not easy, although it is at least possible to continue the old way in order to make a living while learning the new. Only a very few singers are adaptable enough to do that—you can’t just pick up the other voice and take it to the appropriate concert—so that on the whole singers had no choice but to be ‘early music’ or ‘mainstream’. At the time Auger was an exception, but one who showed that mainstream and early music styles were not as far apart in their general attitude to sound and score as their advocates tended to imagine.

¶48 The last few years of the 20^th century saw these distinctions blurred, first because of the mainstream’s increasing tendency to sound HIP, and second because within HIP there began to appear the first signs of a re-emergence of expressive singing, which continues at the time of writing (2005). This, without much doubt, is also happening partly under the influence of recordings, but this time recordings from the past, the flood of reissues that have appeared from record companies struggling to cut costs by using their back catalogue instead of new issues, creating and developing a taste for pre-War performance. What used to be tasteless, self-indulgent and over-the-top now begins to seem ideally expressive once again.

¶49 There’s a nice (and early) example in a 1996 recording of Messiah conducted by Paul McCreesh, using period instruments articulating sighing note-pairs in the neatest neo-baroque fashion, in which Bernarda Fink sings ‘He was despised’ with elaborate Handelian ornamentation and yet with a sound and with an intensity of expression, shaped by dynamics and rubato, that could have been made in the heyday of the massed-choir performances. Compare it to Jennifer Vivyan under Sir Thomas Beecham in 1949—the recording with cymbals and trombones in the Halleluja chorus—and you’ll see where in terms of vocal expressivity it belongs. And it works. ⁷² As we saw in chapter 2, music is extraordinarily flexible: styles that we imagine to be polar opposites can mix. Given enough time, musicians ensure that they do.

¶50 What have we learnt so far about performance style? It’s intimately bound up with expressivity. Expressivity, in turn, is achieved by changes in sound from moment to moment, over and above those demanded by the composer, chosen by the performer. What makes a style is that the options performers choose for being expressive are relatively consistent within a performance, within the work of a performer, and within a geographical locale (though this is less the case now, thanks to recordings) and within a period of time (this may be more true now than before, it’s hard to tell). So on the one hand, expressivity in the western art tradition is fundamental to music-making; it’s what makes musical performance musical. Yet on the other, over time it changes. To make sense of this paradox we’ll have to bring more factors into the equation. Although I’ve provided many hints in the discussion of singing styles as to how expressivity works, and why it changes, we’re not going to start to understand the mechanisms through which it works on us as listeners until we’ve looked in more detail at the analysis of expressive gestures in chapter 8. But before we go there it will be helpful to see how expressivity differs in other kinds of instruments.

¶51 In some ways instruments are easier to deal with. They tend to work expressively in fewer musical dimensions; for example, strings can vary pitch, intensity and time, like voices, but have somewhat less control over timbre and, of course, no text. A piano can vary intensity and time, timbre only to a very limited extent, and frequency not at all. A harpsichord can change time, but not much else, which makes it an extremely good instrument for studying rubato and articulation. A flute can vary intensity and time, timbre somewhat, and pitch slightly. And so on. By making careful choices, it will be possible for researchers to make progress in understanding musical performance under reasonably controlled conditions simply by focusing on contrasting instruments. And in that case it may seem perverse to start with singing, which is so variable and so much shaped by singers’ responses to a text. But in fact the text is invaluable because it gives us a way of understanding what expressive gestures are there to signal. We’ll make much more use of this in chapter 8. For now, though, let’s take a closer look at violin playing and at piano playing, to see how their styles changed during the past 100 years, and how they relate to and differ from singing styles. That will give us a broad enough view of musical expressivity for us to move on, in subsequent chapters, to look for ways of understanding how it works.