• 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

6. Changing Performance Styles: Piano Playing

¶1 ‘What if’ games can be silly but revealing. What if recording had been invented fifteen years earlier? We might well know how Liszt played. Fifty years earlier and we’d have Chopin and Robert Schumann; seventy-five years, Beethoven. Just five years and we might have Clara Schumann. We do have Brahms, almost inaudibly. ¹ We do have Rachmaninov and Bartok on disc, even Mahler on piano rolls. But how much difference does it make? Curiously, the styles of unknown pianists seem much more intriguing and important than those we know. Suppose that we didn’t have Rachmaninov nor Bartok. It seems a fair bet that we’d be fascinated, certain that their playing would reveal their music to us in a wholly new light, and that it is a tragedy that no recording was made. I think it’s fair to say that Bartok’s playing does, but only in the general sense that all recordings of modernist music from before c. 1950 are much more flexible and expressive than the ideology of modernism would lead one to expect. And perhaps to an extent Rachmaninov’s too, but only in the sense that recordings of late romantic music from before c. 1950 tend to be less ‘romantic’ than we might expect. ² Either way, wonderful and fascinating as their playing is, it’s not so revealing that were all records to be lost in an instant we should, having known them before, feel the sense of loss we feel now for Beethoven or Chopin. The uncomfortable truth is that we like to imagine the past we don’t know as far more extraordinary than the past we do. So whereas Rachmaninov and Bartok are people we can understand as something like ourselves, just much more gifted, Beethoven, Chopin and Liszt remain gods, or heroes from mythology (as you prefer).

¶2 It’s worth bearing this in mind when we look at modern attempts to recover past styles. For nineteenth-century music it’s really in piano playing that this has seemed to matter most to modern musicians, perhaps because almost all the composers we most want to understand were pianists. Yet attempts at finding out how they played have argued forwards from the 18^th century rather than backwards from the 20^th. Of course there are no recordings from the 18^th century, and so what people have argued from are styles already (re)created for the music of Bach and Mozart—styles based, if on anything other than modern ‘early music’ traditions, on readings of treatises interacting with (modern) personal taste. But if you do believe that Mozart and Beethoven played in something like the style we enjoy from modern fortepianists, you have then to explain how pianists got from there to Reinecke, Grieg, Pachmann, and Paderewski (born 1824, 1843, 1848, 1860) whose recorded styles of playing are utterly unlike Historically Informed Performance, both in details and in their ethical approach to the composer’s score.

¶3 Of course, all those players were recording in the first decade of the twentieth century, and their styles could have changed radically during their lifetimes. But it would be unusual. Most performers documented across the twentieth century have not changed that much, and since an ability or willingness to change one’s manner of being musical is subject (presumably) to fairly deep psycho-physiological states—the way one thinks about musical sounds and the way one has linked that thinking up with what one’s body can do with a musical instrument—it’s quite hard to argue that different stylistic or social conditions would produce a vastly different rate of change in people’s personal performance styles. Some do change significantly—Artur Rubinstein springs most readily to mind, and we’ve seen change (perhaps to a lesser extent) in the young Kreisler—but it’s rare; and in a moment we’ll look at a pianist close to Brahms who certainly didn’t.

¶4 So it seems on balance reasonable to assume that, there being so many elderly pianists recording at the beginning of the twentieth century, a lot more than violinists or (of course) singers, we do have quite a good body of evidence for piano playing in, let’s say, the third quarter of the nineteenth century and perhaps a little earlier. It doesn’t sound at all like modern fortepiano playing. If we’ve got anywhere near eighteenth-century manners of playing in the HIP movement, then someone needs to explain what it was that transformed piano playing between about 1820 and 1840. The alternative is that HIP has significantly mistaken late eighteenth-century style. There’s some evidence for that, and we’ll look at it in a moment. But first of all, because they give us a coherent body of evidence, albeit fairly small and rather late, I’d like to look at the pupils of Clara Schumann, conveniently collected together in a landmark LP and later CD issue from Pearl. Before we begin, though, a word about schools of piano playing.

¶5 Early recorded pianists, like violinists, tend to be grouped into ‘schools’ according to their descent from teacher to teacher. The International Piano Archives at Maryland offers a convenient schematisation, from which the following very partial family tree (Figure 10) is extracted. ³

¶6 What does it amount to? One could argue, without too much stretching of the imagination, that Curzon and Rosen have things in common, especially a precise attention to detail, but it’s perfectly obvious that they owe that to their own personalities and not to Liszt, who is their latest shared ancestor in the descent from teacher to pupil. One has only to listen to Leschetizky’s pupils to hear the contradictory range of approaches that could be attributed to Liszt (or indeed to Beethoven) if this kind of descent meant anything at all by itself. We badly need a detailed study of schools that brings together memoires, teaching books, annotated editions and recordings. Though the evidence from any one player is likely to be confusing, given enough data groupings may well emerge. But until then we need to be wary of the whole notion of schools. Whether there is anything common to a teacher’s pupils can be determined only by listening and, supposing the evidence is available, by looking at their technique in action. ⁴

¶7 Let’s suppose that some pupils do have a lot in common. How much can be attributed to their teacher, and how much deduced about their teacher’s playing? (Those are two questions, not one, because what one teaches and what one does are not always perfectly synchronised.) Listening alone would be an imperfect route to an answer. Typically the kinds of things piano pupils learn from a teacher involve ways of holding the hand, fingerings, solutions to specific technical problems more easily seen than heard, single features that for the listener are hard to separate out and that may be masked or complicated by other features in a mature style. Often they are recognised instantly by a fellow pupil because of their vivid memory of similar lessons, so testimony may be valuable evidence. In the absence of visual evidence, or reliable and clear testimony, reconstructing teachers’ styles from the common features of pupils’ needs very analytical listening, separating out details, and comparisons with pupils of others (who should contrast), very wide and detailed knowledge—and a performer’s experience—of the instrument and its performing tradition; and even then conclusions are going to be speculative. Alternatively one needs meticulous scientific and statistical analysis. Perhaps it could be done, in which case common features emerging that weren’t common to most players and weren’t a by-product of other features might be features of the teacher’s playing. But would it be worth the effort? The kinds of things one was left with, if one were rigorous about it, would be small and open to having been modified by all the rest, so how a teacher sounded in practice still couldn’t be known. Pupils play differently, and that’s part of the way style changes. The history of performance on record shows that above all else. The genetic analogy is obvious: family trees tell us no more about the behaviour and tastes of our ancestors than they do about piano teachers.

Pupils of Clara Schumann

¶8 All that said, I very much doubt that readers will now cease wondering how Beethoven played, and nor can I. And that’s why, despite all my warnings about the hopelessness of the task if one approaches it reasonably scientifically, I still want to talk about the pupils of Clara Schumann. Because they do have important things in common, and those things do make sense in relation to testimony; and while one cannot of course come anywhere near the sound of her playing, let alone her father’s, Friedrich Wieck (also Robert Schumann’s teacher), nevertheless there are still things one can reasonably suggest were features of earlier nineteenth-century German style, and they’re not easily compatible with modern assumptions.

¶9 Wieck’s emphasis was on legato line and singing tone, in other words on touch, ⁵ and it was above all with the subtlety of her touch that Clara impressed critics. Shaw wrote, in 1877, ‘If any words could do justice to the poetic expression and beauty of touch which distinguishes Madame Schumann’s art, they are such as would appear overstrained and out of place in these columns’ (his emphasis). ⁶ Her pupil Franklin Taylor mentioned that ‘the peculiarly beautiful quality of tone she produced ... was obtained by pressure with the fingers rather than by percussion ... [T]he fingers were kept close to the keys and squeezed instead of striking them’. ⁷ Eugenie Schumann (Clara and Robert’s youngest daughter), in a chapter of her autobiography entitled ‘An Attempt’ to write about Clara’s style of playing, mentions how the Brahms Opp. 116 and 117 pieces, once Clara had lived with them for some time, became ‘plastic creations, glowing with life and tenderness’. And she quotes a Brahms inscription, ‘To Frau Klara Schumann, the greatest singer’, apt, so Eugenie felt, because of the way her touch ‘conveyed like a beautiful human voice every shade of emotion.’ Eugenie remembered, in the accompaniment of ‘Schöne Freunde’ from Liederkreis, ‘How the legato of the melody hovered above that of the bass’, a remark that may make more sense when we look at the separation of bass and melody in Fanny Davies’s playing in a moment. ⁸

¶10 None of this can tell us anything precise about her sound, but, while bearing in mind all the caveats outlined above, it is something to place alongside the sounds made by her pupils. For what little it’s worth uncorroborated by sounding evidence, Clara gave public concerts from 1830 until 1891, and it’s not unreasonable to assume that her mature performance style was in place by the late 1830s (around the age of 20, since she’d been performing at the highest level for some years already). As we’ve seen, that is not to say that her style remained unchanged, but testimony in reviews and memoirs to significant change in her performances has yet to be reported in the modern literature, which (even the best of it) tends to recycle the same quotations. Further research on this would be useful.

¶11 What of her pupils? Natalie Janotha studied with Schumann in the 1870s, was recommended by her as a deputy, and was said to play ‘exactly like Clara Schumann’. ⁹ Bernard Shaw reviewed her in 1889 and found her playing ‘beautiful, suggestive, poetic’, repeating also his previous opinion that ‘here at last was someone who could replace Madame Schumann.’ He continued, prophetically, ‘Though Miss Janotha occasionally breaks out in waywardness and displays of strength, suggestive of possession by a fitful musical demon, yet I know no pianist of her generation whose playing is more sustainedly and nobly beautiful than hers.’ ¹⁰ Her four short recordings made in 1904, however, are disappointing, ¹¹ despite a whirlwind Song Without Words (Op. 67 no. 4, MM=c. 130, the same speed as Raoul Pugno recorded the previous year, considerably faster than the 112 at which it’s commonly played today), and presumably reflect the condition in which Shaw found her in 1892, ‘idly displaying her rare dexterity of hand and her capricious individuality of style without a ray of thought or feeling’. ¹²

Fanny Davies

¶12 Much more evidence survives in recordings by Fanny Davies, who studied with Schumann from 1883, Ilona Eibenschütz, and Adelina de Lara (both from 1886). (Leonard Borwick, reported to be her favourite pupil, left no recordings.) Of these three, Fanny Davies (born 1861) now seems, from her recordings, the most able and it is easiest to imagine, listening to her, the use to which Schumann might have put her legendary touch. ¹³ Davies’s recording of Robert Schumann’s Davidsbundlertänze, book 2 no. 5, is a case in point. (Sound File 20 (wav file)) ¹⁴ The three layers, melody, ascending arpeggio/scale figures, bass, are differentiated by loudness with unusual regularity. Using a spectrum analyser one can see how precisely she is able to match the loudness and the balance of upper partials between notes within each layer. (We’ll discuss this further in chapter 8.) Franklin Taylor’s description of Schumann’s tone and ‘squeezing’ of the keys is easy to relate to Davies’s playing, as is Amy Fay’s account of the consistency of her pianissimo throughout a Mendelssohn Song Without Words, ¹⁵ and Eugenie’s memory of her mother playing chords of equal strength yet exquisitely mellow tone. ¹⁶ Just as interesting, though, in the wider picture of turn-of-the-century style, is the emphasis on legato—inculcated by Wieck and which incidentally Clara Schumann particularly admired in Mendelssohn’s playing. ¹⁷ Early recorded pianists achieve extreme legato by overlapping notes, which can happen in Davies even where staccato dots are notated: these tend to be read as ‘less rubato’ rather than ‘shortened’. In the context of rubato playing, that momentary strictness of time is perceived as relatively marked, an equivalent effect to staccato in more regular playing. What happens, in other words, is that the player moves the whole gamut of touch a step towards the legato side, just as in post-Second World War pianism it tended to be moved a step towards staccato (a large step, in some cases, especially in Bach).

¶13 One aspect of layering and of rubato that is absolutely fundamental to early recorded piano style is the playing of bass notes somewhat before the notes they support. ¹⁸ Davies again offers particularly carefully-calculated examples, and it’s worth looking into them in more detail before we try to understand the more extensive use of this technique elsewhere. Sound File 21 (wav file) is her recording of the last of the Davidsbundlertänze book 2 no. 9. ¹⁹ Figure 11 shows a score annotated with the timings of each beat and (beneath the staff) the length of time by which the bass anticipates the beat to which it belongs. ²⁰ How do we know that the bass is early and the other notes not late? By looking at the consistency of note lengths between passages with bass anticipation and without, by considering whether the bass or melody sounds alone, and by listening: it’s generally fairly obvious which way one perceives it.

¶14 The first thing to say is that the differences between these numbers, though they look large, are small in time: they count hundredths of a second, so the difference in the bass anticipations between 12, 13 and 14 is inaudible, 17 can just be perceived as fractionally noticeable, 24 definitely so. 09 is almost inaudible and may be accidental. In the beat durations a difference between 63 and 37 is very noticeable, between 45 and 49 not. Overall, then, there is a very consistent rhythm of beats that is way off strict 3/4, a long–short–fairly long that gives shape to the notationally monotonous sequence of crotchet beats. Harmonically the second beat simply fills in the harmony belonging to the first, so that it’s almost an appendage in structural terms rather than an independent event; the third beats lead into the first so are slightly shorter; and it’s these facts about the composition that Davies’s performance values and communicates. Working within this framework we can now begin to understand some of the details. Early bass notes give weight to a chord and especially to its melody note that comes after. ²¹ The relative weights again reflect and communicate Davies’s perception of the composition.

¶15 In bars 3 to 6 the most noticeable anticipation is of the C-sharp in bar 5, marking the point at which the bass begins to move; it’s reduced in bar 6, a point of local arrival but insignificant tonality (although this anticipation is so small that it may be accidental), and increases again as the bass moves on, with the longest anticipation before the first dominant (of the dominant) in bar 9, reflecting a quasi-crescendo of harmonic intensification towards the repetition of this phrase (bars 10 into 3). The dominant itself is left unaccented, perhaps in order not to emphasise the dominant cadence too strongly at the first time ending: it gets a normal anticipation the second time around. The second statement of the phrase is treated regularly until the much larger anticipation that emphasises the structural juncture with the next section, starting in bar 11. Having made a large anticipation there Davies continues them for another three bars, perhaps for coherent continuity (it’s become a motif, in John Rink’s terms) ²² before dropping it entirely as the harmonic intensification goes into reverse and the music winds down towards the principal dominant pause at bars 19-20. Again the new section (bar 21) is marked with bass emphasis. Here the harmonic sequence repeats the opening, and so does Davies’s pattern of bass anticipations (showing that the absence at bar 4 was deliberate, incidentally), before it is again dropped from bar 27, where the second beats need more emphasis. It returns to make a structural harmonic point at bars 30-1, and momentarily at bar 47 for hypermetrical emphasis, and finally at bar 53 where Davies separates out the minim sixth, emphasising its role as a drone above the bass melody.

¶16 That was boring to read, but it makes the point that it’s not hard to see good structural reasons for almost all Davies’s early bass notes. Naturally, her treatment of beat lengths overall is just as logically related to the score. (See the numbers above the staves in Figure 11.) The first note of bar 3 is much shorter the second time round because we are now in the middle of a larger musical unit, and the third beats are longer than the first as Davies emphasises the slurred melodic figure rather than the basic structure the second time through. The first beat of bar 12 is longer than usual in order to make a point about the expressive size of the interval leading up to a much higher note. The ritardando is absolutely strictly observed, as one would expect given Clara Schumann’s insistence to Fanny Davies that she should ‘Play what is written; play it as it is written. It all stands there.’ ²³ And Davies changes from unequal beats to equally spaced notes for the written out arpeggio in bars 19-20, emphasising its stillness (and that the purpose of unequal beats is to give vitality to the score). Speeding up is gradual through bars 21-23; regularity recurs for the sequence in bars 27-30 (the lengths of beats 2 and 3 recur; see bars 37-40 for a comparable situation). In bar 32 a longer first beat emphasises the first occurrence of the dotted motif (and see bar 42, second beat for the next variant of it). At the end the ritardando is again begun absolutely precisely. Once again, all this makes sense. What might in later times be dismissed as wayward rubato is in fact meticulously deployed in order to communicate facts about the composition. ²⁴

Ilona Eibenschütz

¶17 Rubato is more (or, depending on your point of view, less) than structural in the playing of Ilona Eibenschütz. She was born in 1873, which makes her twelve years younger than Davies, and was considered to be somewhat wild by Clara Schumann and by her fellow pupils. ²⁵ Nevertheless, Brahms seems to have taken to her, he gave her the first private performance of his Klavierstücke Opp. 118 and 119 in 1892, and it was she who gave the first performance of Op. 119 in London in 1894. That makes her recording of no. 2 rather interesting historically: musically it is remarkable. ²⁶

¶18 Figure 12 compares Eibenschütz’s tempo in the Intermezzo in E minor, Op. 119 no. 2, with Wilhelm Kempff’s. ²⁷ Kempff was born in 1895, so he’s a full generation younger, and his performance, at least in terms of tempo (articulation is distinctly choppy and would have appalled Clara Schumann), represents something more like a typical modern performance. At any rate it gives us a rough measure of modern normality against which to place Eibenschütz. (In contrast to most published tempo graphs, slower speeds are towards the top of the Y axis, faster towards the bottom, since intuitively one is inclined to read uphill as slowing down, downhill as speeding up. This way round the graph is easier to apply to an imaginary performance, or to follow against Eibenschütz’s.) Only the first 17 bars are analysed here, in the interests of space, but they give an adequate picture. It’s immediately obvious that compared to Kempff’s more modern performance, Eibenschütz’s rubato is hair-raising, mapping out not just structural points in the composition but also momentary inflections of pitch and harmony which seem to her more pathetic (emphasised by slowing and getting quieter) or vigorous (faster and louder). The triplet episode, bars 13-17, in which so many linger, she finds increasingly animated till it is rounded off with a huge ritardando emphasising the chromatically descending inner voice at bar 17, before racing off again with the semiquaver off-beats in the following section. And so on. That this is what Brahms enjoyed, and the first audience heard, gives us a lot to think about. It might be tempting to try to argue that Eibenschütz’s playing in 1952, when this recording was made (she was 79), might not have had much to do with her playing in 1894. But in fact her 1962 recording of Brahms’s Waltz, Op. 39 no. 15 is remarkably like her 1903 recording of the same piece, ²⁸ confirming again the broad stability of most players’ style through their lives; and so we simply have to accept that this is approximately how Op. 119 no. 2 sounded when it was young.

¶19 Eibenschütz seems always to have been an impulsive player, and one would not expect everyone to have played it like this. But then so was Brahms. Eugenie Schumann wrote in her autobiography:

It was not always perfectly enjoyable to hear Brahms play his own compositions, but it was always highly interesting. .. He played the themes with great emphasis and curiously free rhythms ... so that one had the impression of strong light and shade. When he came to passionate parts, it was as though a tempest were tossing clouds, scattering them in magnificent fury. ²⁹

¶20 I said of Fanny Davies that it was ‘easiest to imagine, listening to her, the use to which [Clara] Schumann might have put her legendary touch’, and one might be tempted to say something similar of Eibenschütz, Brahms and wide rubato. But of course its being easy to imagine doesn’t indicate that something existed. Davies’s ability to use that touch to produce multiple layers isn’t shared to quite the same extent by Eibenschütz or de Lara, and Eibenschütz’s rubato isn’t like Davies’s or de Lara’s, so we must be cautious about including either in the equation when trying to say anything reliable about Clara Schumann’s or Brahms’s own styles of playing (though at least for Brahms we have that one semi-audible recording). What they do always have in common, though, is this expressive rubato, varying the length of beats from moment to moment in line with the shifting character of the composition and the changing relationship between melodic, harmonic and metrical direction and emphasis; also their habitual use of arpeggiated chords; the slight but ever-present dislocation between melody and accompaniment.

¶21 Does that suggest that these were characteristic of Clara Schumann’s playing also? Well, it may, and one could easily read Eugenie Schumann’s testimony in that light, but equally these features may have changed somewhat through changes in general style since Schumann’s youth. We can’t be sure that they go back to c. 1830. What we can say is that they are not inconsistent with anything said about her playing, or indeed the playing of other pianists from early in the nineteenth century. We must be prepared to consider the possibility that early nineteenth-century playing was much more expressive and less metronomic and neat, and certainly less highly articulated (remember all the emphasis on legato and melodic line) than modern HIP approaches would wish to suggest.

¶22 What this case study shows, it seems to me, is that we can’t know anything much for certain about unrecorded playing styles. But if we are careful, and try to stick to the evidence and avoid too much hopeful guesswork, we can use traditional humanities research methods—assembling a selection of the evidence to argue a point—to sort through incomplete data with a view to renewing our sense of the possibilities for what might have happened in the past. And a fresh look at the possibilities can help to free us from some of the dogmas that constitute current approaches to performing music in the light of historical information. In the end, in other words, these kinds of studies of recordings, arguing backwards, don’t permit reliable historical discoveries; they do help us rethink our own view of the music. And they do that because, however hard we try, it’s extraordinarily difficult to free ourselves from the desire to know how people played in the past. We need to admit, though, that it suits us rather well not to know too exactly. When we know, as we do for early twentieth-century playing, there’s remarkably little appetite for reproducing the style ourselves. When we don’t, we try to reproduce it as hard as we possibly can. In other words, we deceive ourselves: really it’s ideas to direct our own approach that we want, not models to copy.

¶23 And ideas are exactly what this kind of historical research filled out with imagination, in which musicology specialises, gives us so well. At its very best, generating performances that do all they can to adopt what's known from the documentary remains, however unpalatable to modern musical taste, practice-led research into the performance styles of eras closely pre-dating recording can have real historical value. Whether the results will ever reach commercial performance, where musicians' convenience (what can be achieved in limited rehearsal) and public taste (what people will pay to hear) run hand in hand, remains to be seen.

The oldest recorded pianists

¶24 The pupils of Clara Schumann make a useful case study also because their playing is by no means the most different among early recorded pianists from what we’re used to today. They’re not the oldest pianists on record, nor, with the possible exception of Eibenschütz’s Op. 119 no. 2, the most surprising. If we want to reach as far back as possible then we have to look to reproducing piano rolls. The manifold problems with these as evidence were outlined in chapter 3: they are a very questionable source, but when there is nothing else they can be hard to resist, particularly so in the case of one of the very oldest recorded pianists because his playing is so extraordinary. Carl Reinecke was born in 1824, was a friend of Mendelssohn and the Schumanns, and later taught at the Leipzig conservatory where his pupils included Grieg, Sullivan, and the conductors Muck and Weingartner. He was recorded onto Welte Mignon rolls in 1905. In his performance of the slow movement of Mozart’s Piano Concerto K. 537 there are interesting decorations, some of them Mozartian, others Chopinesque; but much more striking is the timing of the written notes. Most chords are arpeggiated upwards, so consistently that one wonders whether this is harpsichord technique surviving into nineteenth-century playing, or nineteenth-century pianism applied to Mozart. The notes of the melody are almost always delayed, sometimes by as much as 1/5 of a second, which may not sound very much but, when one is used to notes placed vertically in a score being played exactly together, seems a very long time as one listens. If one isn’t familiar with this kind of playing it can seem as if the recording has somehow mis-synchronised the hands, but in fact the sequence of notes through the rest of the texture makes clear that that’s not so (and indeed technically it would be a very unlikely accident). The fact is that a lot of the oldest pianists we can hear did play with far more disjunction of melody and bass than Clara Schumann’s pupils. In Reinecke’s case, the regularity of the arpeggiation and of the accompaniment beat lead us strongly to the sense that the melody is late rather than the bass early, and that too tends to increase the sense of alienation for a modern listener, who is at least familiar with the idea of a low bass note sounding early when an impossible stretch forces it. Reinecke’s playing can’t be explained away like that, and the fact that he is playing Mozart, for us such a regular composer, simply makes the strangeness of his playing more acute. The thought that there might be a historical cause in his youth is almost frightening because of the wholesale rethink it would force about everything we imagine as Classical.

¶25 On the other hand, we can also find among early recordings a considerable number of pianists who play in a manner much more like our own. Grieg (b. 1843, studied in Germany) we have already looked at in another context in chapter 3. Raoul Pugno (b. 1852, studied in France) trained under Georges Mathias, a pupil of Chopin whom Cortot remembered as giving the vivid impression of passing on Chopin’s ideas: ³⁰ he plays with the rubato that almost all share but without nearly so much dislocation of melody and bass (in fact it’s rather rare in his playing). ³¹ Aleksander Michalowski (b. 1851, studied in Germany but also sought out a Polish Chopin tradition), also plays more regularly than the Clara Schumann pupils (let alone Reinecke), though still with much rubato and with extremes of speed that are very commonly found at this time. ³² This puts into some perspective the more wayward playing of the much more prolific Vladimir de Pachmann (b. 1848), who studied in Vienna with a pupil of Czerny (which just shows how little can be assumed about teachers and their grandpupils). ³³ Examples include adding material, sometimes several bars, to the beginnings or ends of pieces, virtuoso cadenzas in the middle, passing notes to make a passage more pathetic, as well as the more common octave extensions in the bass that one finds in many pianists of his and later generations. His rubato goes much further than average, though often to strikingly poetic effect. And Pachmann does not shrink from thoroughly changing a composer’s instructions in order to make an effect that he finds interesting. A case in point is one of his recordings of Chopin’s ‘Minute’ Waltz, Op. 64 no. 1, which, as he explains to the audience, he starts playing straight, then with pedal, a slower middle section, staccato ‘a la Pagannini’, legato ‘a la Chopin’, finishing with a little turn. ³⁴ The whole thing takes 2’41”. ³⁵

¶26 Pachmann was an extreme case, exacerbated by his endearingly high opinion of himself which he tended to express in approving remarks at the ends of items in his concerts and occasionally on disc, ³⁶ but the fact is that changing a composer’s text was absolutely normal among many nineteenth- and early twentieth-century pianists. Our literal, often pious attitude to the notation developed only gradually during the first half of the last century. Performers before then took the much more pragmatic view that if a change would make a more moving effect in the context of the way they’d shaped their performance then it was appropriate. It was their responsibility to move the audience as profoundly as they could: the composer’s text was just one element that contributed to the result. As we saw in chapter 2, it’s a perfectly reasonable view, for which the justification comes in the sound and the way that sound makes listeners feel. If we understand music as what it feels like then there can be no watertight objection to the means used to generate good feeling, provided that they do no serious harm to anyone; and whether one sees the dead composer as harmed by changes to his text is a matter only of belief.

¶27 Changes along these lines (albeit less far-reaching) are to be found all over early recorded pianism: octave doublings, chords in place of notes, glissandi in place of scales, modulating improvisations between items, elaborations of all sorts; the line between interpretation and arrangement is impossible to place. At the same time there are pianists (Leopold Godowski a particularly striking example on account of his rhythmic as well as textual strictness) who play it straight, as straight as a modern pianist, reminding us that elaboration is a matter of temperament until the practice ceases altogether after the Second World War.

¶28 Another thing made possible by a more sceptical view of notation was a variety of types of triple meter. When we see 3/4 today we tend to read it as more or less three equal length notes, with some kind of emphasis (which may be provided by the composition rather than the performance) on the first. A hundred years ago there were many different types of three beats in a bar. We’ve seen Davies playing long-short-fairly long; and I mentioned in passing in chapter 2 above Ignacy Friedman (b. 1882, studied in Poland, Leipzig and Vienna) playing Chopin’s Mazurka, Op. 24 no. 4 with the motif short-long-fairly long, as much as 2:5:4 on occasion. ³⁷ An earlier example (one of a large number from which one can choose almost at random) would be Alfred Grünfeld (b. 1852, studied in Prague) recorded in 1905. ³⁸ His Mazurka, Op. 33 no. 4 uses a typical two-bar rhythmic motif, medium-long-fairly long / very short-long-medium (roughly 5:7:6 / 4:8:6). Paderewski plays Chopin’s Waltz, Op. 31 no. 1 (especially on disc, slightly less so on roll) relatively evenly, but with a tendency towards medium-longer-short. By contrast, the last section of his Mazurka, Op. 24 no. 4 tends more towards slightly longer-medium-shorter (e.g. bars 116-20: 6:7:4 / 4:4:3 / 5:4:3 / 4:3:2 / 4:4:3). ³⁹ Paderewski was born in 1860 and studied in Poland; Friedman too was Polish, but it’s a moot point as to whether Polish origin gives pianists more insight into Chopin playing.

¶29 One could continue over many pages to cite examples of comparable patterns of beat lengths. Most of Chopin’s output is in 3/4, and there’s clearly room for a study of early recorded performances that looks both for consistency within genres and (more likely to produce interesting results) in the relationship between rhythm and structure. It will need to be statistically based, however, because it’s as easy as you may now expect to single out pianists who place emphases in different places in the bar. The real point was not to be formulaic but to be interesting, to make the music live or speak or sing, in other words to make it vital, like a living human with responses and feelings that change from moment to moment in response to incoming data (which in this case includes the score). And that meant constantly changing, and in constantly varying gradations, any parameter that one’s instrument was able to change: for pianists loudness and timing, and colour when humanly possible.

Benno Moiseiwitsch

¶30 The emphasis placed on legato by so many nineteenth-century teachers makes the legato playing of early recorded pianists particularly worth studying, and this is another area ripe for research. Benno Moiseiwitsch (b. 1890, studied in Russia, then with Leschetitzky in London) offers a particularly clear example in his 1940 recording of Chopin’s Nocturne, Op. 9. no. 2 (Sound File 22) (wav file), ⁴⁰ though many others would do just as well. There are several aspects to his legato technique, some are relatively straightforward, others less so. In the first category come the continuation of notes through overlapping them—in other words holding down a note until after the next has been begun to sound—and half pedalling, pressing down the sustaining pedal just enough to allow strings that were hit loudest to continue to vibrate without leaving quieter accompanying notes freely vibrating as well. Using these techniques it’s quite normal for Moiseiwitsch to keep notes sounding long after the end of their written duration without having to resort to full pedal, so that the texture remains clear and yet sustained.

¶31 More interesting, because it’s so unexpected, is his ability to make notes get louder well after they’ve been struck. One might think this is impossible on a piano, and for notes sounding alone it is. Once the key has struck the string the note increases in amplitude during the attack phase very rapidly, far too rapidly for one to be aware of anything but the striking note, and then it immediately starts to decay, quickly at first and then gradually more slowly. Clearly, after the attack phase, an individual note cannot get louder. But some of Moiseiwitsch’s long melody notes in this Nocturne do: in bars 2 and 6 the b”-flat, in bars 4 and 8 the e”-flat, in bar 9 the c”, and so on. Spectrogram analysis (of which more in chapter 8) shows at once how this works. ⁴¹ It depends on several factors: 1) the composer’s score providing subsequent notes in the accompaniment whose partials have the same frequency as the fundamental of the continuing melody note—for example an accompaniment note an octave (preferably), a twelfth or two octaves below the melody (or all of these); 2) the accompaniment notes being struck loud enough for their partials to audibly reinforce the melody’s decaying fundamental but soft enough not to overwhelm it with the loudness of their own; 3) the melody note having been struck loud enough to continue clearly above the changing accompaniment but soft enough for the partial of the new accompaniment note to audibly increase it; 4) the resonances of the particular piano that’s being played. Pedalling 5) may also play a role. So it’s very difficult to do, requiring minute control and very careful listening, and it’s only possible when the harmonic context allows it. But in a piece like the E-flat Nocturne, with long melody notes reinforced by consonant arpeggiated accompaniment, the basic conditions are in place: all that’s needed is a superlative pianist to make long melody notes come twice. In theory there’s nothing period-specific about this, it needn’t be a product of a particular performance style; but it’s a technique that flourishes within a fundamentally legato attitude to piano playing in which beautiful connections from note to note are valued more, and contribute more to the shaping of longer continuities, than their articulation. Consequently one hears it more before the Second World War than since; but even so, as with most features discussed thus far, not all pianists practise it by any means.

¶32 If we look at early twentieth-century piano playing as a whole, then, and glance across at violin playing and singing, we can see a variety of practice that may surprise us if we imagine music at that time reflecting a socially constrained world. Far from it; playing was more expressive than it was to become in more modern, ‘liberated’ times, and it was much less bound by convention. Far more was acceptable than half a century (or indeed a whole century) later. Like singers and violinists, pianists valued melodic continuity and beauty of line more than articulation or rhythmic exactness. (One can understand, perhaps, why Stravinsky’s insistence on rhythm and literal representation of the score had to be repeated so insistently, and how that whole attitude seemed modern.) ⁴² The flexibility of music in performance was understood as key to its expressivity. It’s not impossible that there was a process of expressive inflation during the latter part of the nineteenth century that led to the sorts of performances we’ve been examining. That would make the reaction of the modernists easier to comprehend. Certainly, pianists active in the 20s and 30s are, like singers of the Lotte Lehmann generation, possible to see as coming at the end of a tradition; the last and in some ways the most intense of the musical poets, soon to be superceded by the young actors of the post-War years.

Alfred Cortot

¶33 Alfred Cortot (b. 1877) can very well stand as representative of this last piano-poetic generation. Like Lehmann he was one of the most intensely expressive performers on record, and like her he wrote in detail about the interpretation of specific pieces as an extension of his lessons and masterclasses. He studied in Paris with Emile Descombes, a pupil of Chopin, ⁴³ which may or may not explain his reverence for Chopin’s intentions and his repeated emphasis on the need to reproduce the ‘aristocratic reserve’ he believed to be characteristic of Chopin’s playing, even to the extent of advising, in the Prelude Op. 28 no. 12, that one should hold back in order not to exceed the limitations of Chopin’s technique. ⁴⁴ The modern listener may sometimes wonder what Cortot’s Chopin playing would have been like if it were unrestrained; and it’s hard for us to understand how one can square such profound respect for a historical style with rewriting passages of Chopin’s scores to produce a better effect. ⁴⁵ For Cortot, though, it was Chopin’s effect that he was producing, just doing it better. Examples include extending cadenzas, replacing single notes with chords, and enlarging chords (all these examples from Cortot’s fascinating annotated edition of the Ballades). ⁴⁶ And in performance they are undoubtedly thrilling, not to be wished away but rather to be appreciated as absolutely right in context, the context of a Cortot performance. Perhaps the most evocative example comes in the Prelude, no. 15, bars 71-5, five rumbles of distant thunder in the bass leading into the final return of the opening material which it is hard to wish were not there. ⁴⁷ (Sound File 23 (wav file) gives the 1926 recording.) ⁴⁸ Hardly less effective is the unnotated crescendo that begins his performances of the twenty-second Prelude and gives it a force it lacks in any more literal representation. (Sound File 24) (wav file) ⁴⁹

¶34 Notoriously Cortot was unbothered by wrong notes if the spirit of the performance was right (Op. 28 no. 22 is by no means the worst example, incidentally), and in so many of his recordings one has to be extremely unsympathetic to the beauties in order to object fundamentally to the mistakes. In his book on Chopin Cortot wishes that more performers would ‘leave the problems of technique where they belong - in a place of secondary importance - and allow... them to place their imagination rather than their fingers at the service of the inner significance of the music.’ ⁵⁰ The deliberate changes and the accidents all belong to a consistent approach to making music from raw compositions. Cortot had an exceptionally acute aural imagination that, applied to his perception of the shifting meanings dormant in a composition, and tied to a fine control of timbre through touch, timing and pedalling, led him to shape and colour every passing moment of a score. He also had a Lehmann-like ability to conjure up effects in words, and an intellect that enabled him technically to describe what he was doing and why. He makes a very good subject for study, therefore, if one wants to understand how the most expressive piano playing works.

¶35 His 1928 recording (the second of five) ⁵¹ of Chopin’s Prelude in E minor, Op. 28 no. 4, makes a nice example (Sound File 25 (wav file)), ⁵² not least because it has never been published before (it turned up unexpectedly in a copy of the 1926 set in the King’s Sound Archive), but also because performances of the piece have been studied many times, and comparisons can be made and the replicability of earlier findings tested. ⁵³ Eric Clarke, in a thought-provoking study (Clarke 1995), used modern performances captured via a MIDI piano to examine the relationship between tempo and dynamic changes and the representation of musical structure. Among other things, Clarke showed how the two performances of Op. 28 no. 4 by Robin Bowman offer contrasting readings of the musical structure by using similarly expressive tempo and dynamic changes in different musical contexts, and also a similar reading of a musical structure using different patterns of change. Change in tempo and/or loudness, it seems, is more significant than the kind of change (louder or softer, faster or slower). It follows that musical structures do not require for successful performance particular patterns of note lengths and loudnesses: they only require that changes be made in patterns that relate to structure. Within that very broad requirement, the performer has an inconceivably large number of possibly combinations of timing, loudness, and (in vibrato, portamento, and expressive intonation) pitch to combine in ways that feel good. The next interesting area for study, then, concerns that ‘feeling good’: why do particular combinations seem to work in particular performance contexts? Cortot may help us begin to find some answers because his moment-by-moment combinations are so striking and so easy to observe.

¶36 Figure 13 offers an edition annotated with the lengths (in hundredths of a second) and loudnesses (in dB) of each horizontal event in this 1928 recording. (It would be easy to turn the numbers into a graph, but actually a graph tells us nothing that they do not; though, turning them into coordinated graphs of loudness and duration could make a nice classroom exercise.) There are a number of points where Cortot plays the melody note late, ⁵⁴ and they’re worth looking at more closely. In most cases he doesn’t just delay the melody note, he also extends the accompaniment chord so that the melody note has at least (usually more than) the full length it would have had if played on time. Note that, unlike in the Davies example above, it’s not a question of playing the bass early. Cortot actually inserts a lot of extra time into a sequence of events that is otherwise fairly regular. One might think this would produce a halting rhythm, the music proceeding in fits and starts, but actually it doesn’t sound like that. Why not?

¶37 For Cortot the falling semitone crotchet-dotted minim figure, c’’-b’, c’’-b’, c’’-b’, b’-flat-a’, b’-a’ etc., is the focus of attention, and the way he plays it forms a kind of motto shape that applies to many aspects of the piece. To understand his performance we need to understand this shape. The first note is a peak from which the second falls away—generally the first is a step up from the preceding note so that in context the shape resembles so many one finds in music and in life: rising and then falling. Like peaks in other domains it takes some effort to reach up to it, but one descends from it more easily. Even if one didn’t know what Cortot had said about the piece (and we’ll look at that in a minute) it would be evident that he imagines it as like a sigh: a longer and louder breath in, expelled in a shorter and quietening breath out. It would be an imaginative or contrary pianist who read it much differently, given the chromatically descending accompaniment sequence of 7ths to 6ths.

¶38 The difference comes rather in the way pianists represent this figure in sound. For Cortot the figure is the focus and goal of every bar; but, as the discussion of Clarke suggested, there are various ways of representing it even within one coherent performance: what counts is that there be a change in at least one dimension of the sound in order to mark it. Which direction the change takes, and in which dimension, is open to the performer to decide according to the local context. It would be possible to do nothing at all, and simply leave it to Chopin, representing the notated score literally and relying on the changing harmony and melodic line to do all the expressive work. In more recent decades, say between 1950 and 1980, pianists have tended more in that direction. But for Cortot the task was to make that shape as expressive as it could possibly be. And that means making large changes in more than one dimension while maintaining a sense of continuity and general consistency of mood (albeit with changes of intensity along the way). It’s a question of how far he can go without losing that coherence.

¶39 Cortot creates several very noticeable gestures (or dynamic shapes) ⁵⁵ that recur, matched to recurrences of compositional motifs. We’ll look more closely at expressive gestures in chapter 8. My aim here is to show how expressive gestures can be used not (or, if we take Cortot’s programme for this performance seriously, not only) to shape particular moments in ways that model sounds or feelings from life (which is one of the things we’ll be doing in chapter 8) but rather to show how tempo/loudness envelopes can be used repeatedly throughout a performance to do music-structural as well as emotionally expressive work.

¶40 Here again, just as in Fanny Davies’s performance of the Davidsbundlertanz, it’s helpful to invoke John Rink’s notion of ‘performance motif’. What, using the more neutral terminology of acoustics, I’ve begun by calling tempo/loudness envelopes, are easily conceptualised as motifs, especially because they are tied throughout the performance to the motifs so easily seen in the score. As with composition motifs, the recurrences of these envelopes generate consistency, yet—and this gives them an additional and essential dimension—each statement contains within itself change that does expressive work, and does it again and again as the envelopes repeat. Outstanding, as I’ve just suggested, is the treatment of the ‘sigh’ figure. As Figure 13 shows (with perhaps unnecessary precision, but unquestionably shows), the associated performance gesture consists of delayed melodic upbeat, anticipated by the left-hand accompaniment, falling into a delayed downbeat. The first and usually the second note are louder than their surroundings, the second note less so than the first, assisting the modelling of a sigh. The delayed onsets separate the melody from the accompaniment, adding to the motif’s prominence and (paradoxically) to the sense of melodic legato. The motif is elongated when the composition provides extra notes, as at the end of bar 7 but also bars 11 and 19 where Cortot gives the grace note the greatest weight (the score in Figure 13 gives not Chopin’s notation but a notation closer to Cortot’s performance). And it’s elongated also where the extra notes are concealed or differently placed, right at the end of bar 10 and in the middle of bar 18, at the start of bar 12 and at its end in the elongated c’, as well as at the related end of bar 20.

¶41 A second motif that recurs all through serves to shape the left-hand accompaniment. The first quaver is lengthened, starting an acceleration pulled back through the fifth to the seventh quavers, with the seventh as the goal, underpinning and slightly anticipating the melodic crotchet. The acceleration/deceleration in each bar is hard not to sense as inhalation/exhalation. The pattern is applied to the melody in the solo bar 12 and to the turn in bar 16. But see too how it is changed after the climax, so as to follow the harmony changes, so that the louder melodic line that Cortot has been making out of the top notes of the left hand right from the start now becomes the main source of melodic motion. The continuity this provides holds together the compositional liquidation or dissolution through these bars, completed by Cortot’s very individual decision (made consistently in all his performances of the piece) to play the penultimate bar’s chords somewhat staccato.

¶42 Cortot has, then, a very clear sense of the composition’s structure but his aim seems not to be to represent the structure—not simply to give an account of it—but rather to bring it to life by making it breathe more deeply, obviously and evocatively than would later be considered well-mannered.

¶43 Now let’s look at how Cortot uses these motifs at key moments in unexpected ways. The c’–b figure from bars 3-4 is much quieter, the b in fact no louder than its surroundings. Compared to the pattern established till now the marked lengthening but sudden quietening of the ‘sigh’ figure completely confounds expectations, yet it remains part of a coherent process of speeding up and slowing down through the bar and of gradual decrescendo. Its effect demonstrates vividly how powerfully expectations are involved in our response to moments in performances. We have only three bars behind us, but we have configured our expectations sufficiently to the pattern already set up to be easily effected by this change in just one dimension of it. The sigh is long and quiet, and evokes a deeper sense of despair than one might think possible from a few microseconds more and a few decibels less. It emphasises just how responsive we can be to minute changes in sound (for excellent evolutionary reasons), and how much the smallest details of a performance matter. The boldness, precision, and emotional effect of these sorts of gestures are key to Cortot’s personal style.

¶44 Something similar happens, perhaps slightly less successfully because of the repetition, at bars 10-11. Then, after the melodic cadenza in bar 12, bar 13 and what follows comes a much stronger contrast than one would expect from the score. The whole texture is much louder, and the lengthenings around the fourth and first beats become smaller, creating an acceleration. The weight of the climax is greater than one might expect, supplemented by Cortot playing the left-hand Bs at the start of bar 17 an octave lower than notated. (And incidentally he also adds a central B in the final left-hand chord of the piece.)

¶45 Cortot is going beyond what’s normal for us, and indeed beyond what was normal for most pianists of his or the preceding generation, because he’s not just marking up the formal properties of the piece in a manner consistent with period style. He’s giving moments expressive properties well beyond those required by the representation of compositional form. The piece’s larger AA’ form is marked, certainly, but the A and A’ are contrasted far more than even Cortot’s contemporaries might normally have thought was required. Clearly Cortot, like Lehmann, is working within an expressive world in which nothing is too extreme as a representation of deeply felt emotion provided only that it makes sense within the stylistic world of the whole performance.

¶46 There are many more points that could be made about this recording and about the data derived from it in Figure 13. It’s easy to see similar patterns recurring throughout the performance and operating simultaneously on different levels. Readers familiar with classical music analysis will not have failed to notice that many of the things I’ve said about this performance are the sorts of things that people used to say about scores when analysis was less unfashionable than now. I can’t help that, though. These are the things that this performance does. It’s not impossible that analysts in search of motivic working and unity were led along, without realising it, by the extent to which performance contributed to the sense that those things mattered. Perhaps one is really more aware of them at work on one in the performance dimension than in the score; in other words, things that performers were doing we tended to ascribe to scores alone. I argued along these lines in the opening chapter, and this is a good example of why. It’s also possible that current fashion is failing to do justice to the strength of classical analytical views of how music works, though I’m happy to leave that to analysts to argue more fully.

¶47 I don’t think that there can be any reasonable doubt, though, that convincing performances are in many respects coherent, however unexpected they may be: one couldn’t take a phrase from a performance of this piece by Cziffra or Godowsky or Gould and drop it into a Cortot recording and get any kind of sense (though there are any number of examples of editors making less obvious, more plausible insertions once tape made it possible). Cortot’s expressive gestures here have no place on this sort of scale in later or in much contemporary playing, then, and for a listener accustomed only to later playing they will not make immediate sense. We see this again and again in unsympathetic comments on early recorded performance. And this is the case precisely because, just as within Cortot’s performance itself, the rules of expressivity can be radically reset by performers, both over years on the very large scale, and over seconds on the small. Over years, these kinds of gestures become normal and then increasingly abnormal; over seconds they are formed in one configuration and then changed into another: the principle is the same. Music works, as analysts know so well, on our expectations, variously soothing them and upsetting them according to what composers and performers need to do in order to communicate expressively with us.

¶48 With all these points in mind we can now see what Cortot imagined when he came to represent the piece through metaphor. Here are his comments as noted down by Jeanne Thieffry, probably sometime in the early 1930s, soon after this recording.

This piece must be played in the mood of a mourner, with the face dimly veiled and the eyes heavy with tears. The beginning should be so quiet as to be hardly perceptible: it should be a mere murmur. In the left hand the finger that establishes the change of harmony should alone emerge a little from the pianissimo. In the right hand there should be nothing but a lament, the lament of a mourner who can no longer lift up his voice.

At the twelfth bar pass from the feeling of sadness to sudden terror. Here it is no longer a question of correctness, of strict time, but of superhuman sorrow. Emphasize the chords of the left hand by thrusting the fingers well into the keyboard, and let the clamour of the right hand reach a kind of frenzy. Then the lament dies away. There is a pause—a long one—and then in the last chords a note of doom. The highest note alone of these chords will be held for its full value. The others will be abandoned previously, at the same time as the pedal. ⁵⁶

¶49 You’ll see at once how like this is to Lehmann’s notes for students: the same mixture of acting and technical instruction. The performer is told how to feel as well as what to do, and for very good reasons. There is no more efficient way of communicating to a student performer how to shape a piece in a particular way than to tell her how to feel or what to imagine the music is evoking. It’s painfully unfashionable, but none the less true, that representing music through metaphor is a direct way of communicating precisely what one hears or wants to hear. It’s a far more direct route than specifying note lengths and loudnesses, as comparing my text and Cortot’s, as representations of his performance, makes abundantly clear. The ease with which we map metaphor onto music, and vice versa, tells us something very important about the way music works—not (or not only) as a system of abstract patterns giving us mathematical pleasure but as a system of triggers in which patterns are mapped by the brain from sound onto memories with related properties, and vice versa. There’ll be more to say about this in chapter 8. For now the important point is that however ‘extreme’ or ‘fantastic’ or ‘indulgent’ or ‘romantic’ Cortot’s and Lehmann’s approaches to performance and teaching may seem, they do make very good use of the way in which our brains construct musical meaning. There is much more to this manner of performance than just 1930s fashions in sentiment, and that is why, for anyone who has become acclimatised to it, it can still be so powerful today, and why there is so much for modern performers to learn from it.

Artur Rubinstein

¶50 We’ve seen how singing style changed quite suddenly after the Second World War, and the same is true of piano playing. An interesting barometer is Artur Rubinstein, one of those relatively unusual musicians whose style changed markedly during a long life. Rubinstein was born in 1887 in Poland, but studied mainly in Berlin where he made his debut in 1900 aged 13. His musical development was under the watchful eye of Joachim, whose style of violin playing Rubinstein realised was relatively plain (‘ascetic restraint and nobility’) only once he’d heard Ysaye, whose much more intense expressivity had an immediate impact on him. ⁵⁷ It was this newer style that Rubinstein developed in himself, and to that extent he began very much as a child of his time. He rethought his approach to practice and technique in 1934, ⁵⁸ but while much has been made of the resulting improvement in his playing (not least by himself), ⁵⁹ it was in fact much later that his way of being musical became modernised. When and how this happened has yet to be investigated. But because Rubinstein recorded substantial parts of his repertoire twice, once in the 1930s and again in the 1960s, we can hear that his performance style did change substantially. While Cortot played in much the same way throughout his recording career—compare his conception of Schumann’s ‘Der Dichter spricht’ in its representations on disc in 1935 and on film in 1953 ⁶⁰ —Rubinstein fitted much more easily into the general stylistic context of his time, even in his 70s and 80s. His recordings of Chopin’s Mazurka, Op. 68 no. 3 illustrate this clearly. The first was made in 1938 for HMV, ⁶¹ the last in 1966, ⁶² and even a casual hearing suggests that the latter recording is rhythmically more regular, the three beats of each bar less widely varied in length than in 1938. If we compare it to a selection of modern recordings we can expect to hear that this trend towards regular 3/4 continued through the later 20^th century, as the performance practice for Chopin’s 3/4 music became increasingly homogenised. Where once the different composed languages of the nocturne, Polonaise, mazurka and valse were expressed through correspondingly different habits of rhythmic emphasis, increasingly the three beats of the bar became more even, the composed language less noticed by the performance style. ⁶³ Figure 14 confirms this in a small way, charting Rubinstein’s beat-lengths in 1938 (the most jagged line) against those of his 1966 recording and those of four younger players all producing smoother lines. It’s easy to see how Rubinstein in 1966 sits between the extremes, reflecting a significant change in his style, matching changes in the playing of the first post-War generation. That said, there are interesting exceptions among the younger players. Luisada is considerably freer than the others. And like Luisada, Indjic, though playing evenly in the A-section material, in the middle section, based around an ostinato fifth, is much more free. From this we might propose that modern players, faced with a crotchet ostinato in Chopin, are more likely to switch on rubato than early recorded pianists, though much less likely to use it in more varied textures. More examples need to be analysed to test this, of course. ⁶⁴

¶51 A more sophisticated analysis of the data, by Craig Sapp, using hierarchical correlation plots, ⁶⁵ compares the curves of these graphs with one another to represent visually how alike they are. The bottoms of the triangles represent similarities between pairs of performances at the level of the beat; as one moves up, larger and larger groups of beats are compared until at the top one sees only the complete piece; so the more uniform the colour patterns further down the triangle the more alike are the performances. (Plate 8) This shows particularly clearly, in a way that one can read from the tempo graphs but is less immediately apparent there, the similarities between Rubinstein’s recordings, for which the colour patterns are very alike. Although an important respect in which they belong to different generations is by showing different attitudes to rubato, the 1966 performance is in some ways an evened-out, more regular version of the 1938 one. The only other performance like it is Chiu’s, leading to the possibility that Chiu had Rubinstein 1966 in mind as a model. Similarly, there are a few but quite striking similarities between Rubinstein 1938 and Luisada, though these are certainly clearer from a straight comparison of the tempo graphs (and, of course, from listening) than from the correlation plots. And there are structural but not surface (and so not stylistic) features in common between Luisada and Indjic.

¶52 Figure 15 charts data for a selection of performances of the first 16-bar phrase of Chopin’s Nocturne in E-flat, Op. 9 no. 2, from Josef Hofmann in 1911, through Rubinstein in 1936, to Maria Pires in 1996. ⁶⁶ The contours of the graphs tell the story plainly. The three recorded after the Second War are noticeably less jagged than the four before, except that Leopold Godowski plays with a regularity that, while uncharacteristic of his time, is a very noticeable feature of his personal style in all his recordings. In general, though, the beats are less varied in length (there is less rubato) from the 1960s and 90s than in the 10s, 20s and 30s. Of course these are not the only features of piano playing that changed. There is less anticipation of the melody by the bass in the second half of the century—although Kathryn Stott is a fascinating exception here, again a feature of personal style out of keeping with the norm around about. ⁶⁷

¶53 How variation in colour and in dynamics are handled is harder to say from early recordings, and therefore generalisations about change in those dimensions over the course of the century are more difficult to make: much closer study will be required before we can put all these elements of piano playing together into a more accurate story. But it does seem reasonable to suggest that details at the note-to-note level varied across a much broader scale early in the 20^th century than later on. As players after the War were less inclined to allow large contrasts from note to note, so their attention shifted to higher levels, delineating formal structure expressively rather than microstructure. This had a variety of consequences—and this is true for all kinds of music-making, not just for piano playing.

¶54 In moving their focus of attention from the detail to the whole, from the surface of the composition to its underlying structure, from descriptive to analytical playing, performers responded to much broader changes of focus in society. We’ve seen a related phenomenon in post-War singing and its search for deeper meaning. The baroque-instrument and Urtext movements reflected similar shifts in attitude, as did the contemporaneous growth of musical analysis. ⁶⁸ In many ways music-making after the second War was a more serious, now ethically-based activity. While the previous focus on the musical surface, on the shaping of tunes and local gestures, enhanced classical music’s entertainment value, allowing audiences to follow its shifting moods from moment to moment, continually moved by something new, the more austere analytical view of post-War performance (and commentary, of course) increasingly downplayed music’s entertainment value and insisted instead on its music-structural integrity: it was (as Hanslick had insisted long before) to be a language of its own, not representative of anything else. This is, of course, exactly the attitude underlying the post-War avant-garde, another manifestation of the same widespread change in attitudes on the part of music professionals. Composers ceased to be entertainers and became experimenters, scientists even. ⁶⁹ One could argue that recording made this possible, or at least alleviated the worst economic effects by allowing music to be heard repeatedly and conveniently, compensating for the greater difficulty of making sense of it at first hearing, and also by turning performers who might, for the older generation at least, seem austere into (musical) household names. It seems possible that without the LP and the consequent reinvigoration of the recorded music market, the less approachable style of post-War players would have had considerable economic disadvantages. ⁷⁰ But recording is not in itself sufficient explanation for the continuing popularity of classical music performance through the 50s, 60s and 70s. Listeners surely became better at hearing details, so that as performers worked with much smaller changes from note to note listeners adapted with them. How else can one explain the deep and universal admiration for players who worked with a relatively subtle palette of expressive gestures like Curzon, Brendel, or Perahia?

¶55 Not everything was treated austerely, either. The complexity of post-War performance practices is evident when one adds to this melting-pot of possible causes and effects those pieces (mostly Chopin) which continued to be performed with wide rubato through the whole of the 20^th century. The Prelude, Op. 28 no. 4, whose 1942 performance by Cortot we examined earlier, is a case in point; the Berceuse another. Looking across a century’s performances of the e minor Prelude it’s actually those of the 1940s (Cortot excepted—his as we’ve seen was fixed already by 1926) that seem the plainest, Rubinstein’s of 1946 especially so. ⁷¹ More recently pianists have struggled to bring back intense expressivity to it, not always very convincingly since this kind of rubato-led expressivity no longer comes naturally. Pogorelich’s 1990 recording seems to this listener particularly awkward, with notes stretched in different places from bar to bar, not always with any very obvious connection to the voice-leading, as if the stretching itself were a guarantee of expressivity, regardless of how it’s done. ⁷² And to an extent it probably is: for reasons that will become clearer in chapter 8 our brains probably do interpret rubato as expressive in itself; but of course it’s much more so when it happens in places where expressivity makes sense to us for other reasons too. ⁷³

¶56 A more interesting example, because it is still so often done so well, is Chopin’s Berceuse. Here the repetition throughout the piece of a left-hand ostinato (tonic-dominant seventh until almost the end) allows no other harmonic guide for the use of rubato. If expressivity is to be signalled through rubato then it’s going to have to be led by the melodic content of the right hand. Several of the early recorded pianists—Moriz Rosenthal (b. 1862, rec. 1930), Josef Hofmann (b. 1876, rec. 1918), Ignacy Friedman (b. 1882, rec. 1928)—play it relatively straight, and very fast (Hofmann takes three minutes, Friedman 3’20” and Rosenthal a little over 4’, as opposed to almost 6’ for Maria-João Pires in 1998). ⁷⁴ Paderewski (b. 1860, rec. 1922) uses more rubato, but pays most expressive attention to differentiating the characters of the right-hand sections using rubato and variation in tone. Friedman modifies the left-hand quite extensively (presumably to make it more interesting). Cortot is typically free with rubato, producing a kaleidoscopically nuanced performance that responds to every twist and turn in the right-hand line. Backhaus (b. 1884, rec. 1927) takes a similar but more restrained approach, using accelerando and ritardando over longer stretches rather than moment-driven rubato. But all these pre-War performances belong to a similar stylistic world in which rubato is placed so as to point up details of voice-leading that seem to have expressive potential.

¶57 Jumping to the 1960s and to Perlemuter (b. 1904, rec.1960) we find hardly less rubato, indeed possibly more than Hofmann or Friedman, with Perlemuter almost Cortot-like in his stretching and contracting of beat-lengths. But there’s an important difference: Perlemuter uses this rubato to shape whole phrases rather than momentary gestures. And that tends to be the way it is used by most players in the second half of the century. This is an important change, reflecting a more structural, less rhetorical approach to shaping music in performance. Another is the much greater length of most post-War performances of the Berceuse. If anything, one might expect these two factors—rubato and length—to encourage each other, greater rubato making longer performances, and indeed that is what happens in Cortot’s, which is the longest of the pre-War recordings, but in general that is not the case. Nor is it a matter of phrase-end rubato taking longer (or involving less compensation) than gesture-led. The post-War performances are just much slower, which fits with general trends in performances of all sorts. ⁷⁵

¶58 We’ll look at a more typical example of post-War pianism in chapter 8 (Schnabel playing Schumann), in discussing the expression of structure, pursued as a common goal by most pianists from the late 1940s. Where the Berceuse is different is in its continuing to attract wide rubato throughout the rest of the century. As I’ve suggested, this is tied up with the nature of the ostinato which would bring death to any performance without flexibility at some other level. But what is most striking about the rubato in the latest 20^th-century recordings is, just as in the e minor Prelude, its waywardness. It’s as if performers no longer knew how to manage it. Both Vásáry (b. 1933, rec. 1967) and Ashkenazy (b. 1937, rec. 1976) produce wonderfully meditative performances, in tune, one might say, with their times, but both grind almost to a halt on the final page. Theirs are much less lively performances than those on the earlier recordings, requiring more concentration from an audience. Howard Shelley (b. 1950, rec. 1991) turns attention towards the left hand, shaping it in two-unit (four-bar) phrases, and allowing it to play a role in shaping the performance while the right hand becomes more decorative. Again, this is a structural approach to performing that pays less attention to the musical surface (a slightly curious decision in this piece, perhaps), and is in that sense modern. Maria-João Pires (b. 1944, rec. 1998) plays the piece extremely slowly, with big dynamic surges to generate expressive phrasing so that much more work is done by dynamics, and less by rubato, than in the earlier recordings. This, too, is a modern tendency, switching attention from timing towards loudness as a dominant agent of expressivity. Demidenko (b. 1955, rec.1992) makes good use of an accelerating and then decelerating ostinato, which breathes well, but gets sidetracked towards the end by an interest in bringing out inner voices. Nakamatsu (b. 1968, rec. 1998) plays the simpler passages straight, but uses rubato (well) in the more complicated phrases, so that the beginning and end of the piece seem to belong to a different expressive world from the more complex middle. Kissin (b. 1971, rec. 1998) uses half-pedalling throughout, aiming perhaps to produce by blurring the sound the dreamy impression once made more subtly by other means, and his rubato is wayward. All of these are wonderful pianists, albeit in very different ways, yet they seem to have little shared sense of how a piece like this can be made to work. The means to it are simply not part of current piano technique, so that each pianist has to invent a solution for themselves as best they can. While it can be very hard, close up, to see how style is developing around one, the Berceuse does offer a particularly revealing measure of the way in which available technique has changed.

¶59 It must be evident from this brief discussion of post-War piano playing that there is everything still to do if we want to sort out how playing developed through periods that are still relatively close to us. It takes a lot of perspective to see the general features of a period’s performance style and to understand how they relate to their cultural context, and in the end only a lot of time is going to give us that. But we could make a good start by assembling detailed studies of a range of performers and repertories, and I hope that readers may want to contribute to that. There is a huge amount of fascinating research waiting to be done.