• 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

5. Changing Performance Styles: Violin playing

¶1 It’s a surprisingly short step from voice to violin, as writers on violin technique have repeatedly noted. The comparison makes especially clear the similarity, which I’ll be discussing in chapter 8, between expressivity in singing and in instrumental playing. As David Milsom has pointed out, 19^th-century violin teachers habitually referred students to vocal technique for models, and one can hear the similarity of approach in many early recordings. ¹ We saw how from Patti to Lehmann a relatively ‘natural’ sound, in which vibrato, rubato and portamento are all used but are subservient to continuity and flexibility of line, developed into a much more overt emotionalism via heavier and more continuous vibrato, rubato that favoured ritardandi at climatic moments, swoops up to the most significant notes and words, and more dramatic use of dynamics. Similarly in violin playing a relatively plain style that we hear first in recordings of Joachim, using noticeable vibrato only occasionally for emphasis, frequent minor rubato, and light but frequent portamento, seems at first glance to have given way through the 1920s and 30s to a style, typical of Kreisler, Heifetz, Elman and their contemporaries, that depended for its greater emphasis on continuous vibrato, heavier rubato at crucial moments, slower portamento, and emphasis through dynamic accents. How these, and other styles, are tied together over time we’ll consider later on, after we’ve looked at pianists as well. But first let’s look at examples of developments in 20^th-century violin playing and compare them with singers. ²

Joseph Joachim

¶2 Joseph Joachim, born in 1831, was three years older than Santley, and so was in his early 70s when he recorded in 1903. He had studied in Vienna in the late 1830s and early 40s, and then with Mendelssohn, later working with Liszt and most famously with Brahms, whose violin concerto was written for him. How he played is therefore of some historical interest, and for anyone whose notion of Romantic playing involves surging crescendi and heavy vibrato it’s rather surprising. ³ First of all, there is by our standards almost no vibrato; what there is is very light and used most noticeably on longer notes. Portamento is used but mainly in pathetic passages, expressing character, not as a routine means of getting around the instrument. Rubato has two functions; at the level of the beat it is used only when it contributes to characterisation, marking the difference between more forceful (faster) and more pathetic passages (slower); from note to note, however, it is used continuously, stressing (by lengthening) notes of structural importance within a phrase, so that, for example, scales tend to be fast but may linger on melodically or harmonically significant pitches. In these senses Joachim’s playing is not unlike the more expressive end of modern HIP. But Joachim has a much wider range of sounds and styles than is common today. His Bach playing is extremely clean and highly articulated, with very narrow and rather uneven vibrato on the long notes only (suggesting, though, that it might have been common in more reflective music); his Brahms (Joachim’s arrangement of two of the Hungarian Dances nos. 1 & 2) adopts a gipsy-style rubato; in his own Romanze he uses far more portamento to bring out the sentimental character of the composition. In other words, Joachim’s performance style varies in order to emphasise his notion of the composition style, something that happened much less in subsequent generations. To us his Bach playing may sound HIP—and he uses largely gut strings, as was still the norm—but his Brahms uses the Tourte bow to attack chords with passion, and his Joachim is as sentimental as anything from the 1920s. ⁴ Some of these differences must have been evened out by the insensitivity of the recording technology—his relatively consistent loudness throughout may be misleading, therefore—but that so much comes through only emphasises just how powerful and varied his playing must have been. Later 20^th-century playing would have seemed monochrome by comparison. ⁵

¶3 Milsom has made a number of useful comparisons between Joachim and Patti, ⁶ and indeed they have much in common in their limited and targeted use of vibrato, portamento and rubato; but one could equally appropriately compare Joachim to Santley or Henschel, or other early recorded singers who have left more powerful performances, for example Lilli Lehmann (b. 1848). The unbroken melodic continuity managed so well by Patti is there in Joachim’s Romanze but totally foreign to his Bach, and for a vocal analogue to his Brahms one would need to look to the most forceful operatic singing of his time. Equally, his attention to the character of the composition and the range of expressive approaches he takes as a result has no space for the consistency of expressive language that settled on violin playing as vibrato became continuous and universal over the next two decades. Responsiveness to changing characters was crucial, but it’s above all his very limited vibrato that separates him out from violin playing over the next 70 years and that it’s so tempting to suppose was characteristic of his unrecorded predecessors.

¶4 This isn’t the place to try to determine what caused the change in vibrato at the beginning of the 20^th century. There has already been much argument about it and, given its importance and the lack of definitive evidence, it’s a safe bet that musicologists will continue to worry over it for a long time to come. Judging by written evidence from the 19^th century (usefully summarised in Milsom 2003) and by the earliest recordings, among which Joachim’s are crucial, it seems that vibrato changed from being one expressive technique among many, used to different extents by different players and only prominently at moments of greater intensity, to become a fundamental feature of string technique used by everyone all the time—that is to say, on almost every note that was long enough for a finger or wrist to shake and that was not an open string. ⁷ It is very unusual for so significant a change in style to happen so quickly, and it’s hardly surprising that many are looking to understand why.

¶5 While recognising that taste may have played a part, Mark Katz has proposed that the main engine of vibrato change was the introduction of recording. Certainly the coincidence of dates is intriguing. He offers a number of reasons, none of which is in itself decisive. But a single cause seems wholly unlikely: a coincidence of tendencies seems far more probable, and Katz offers a fair number. First, as singers discovered towards the end of the 19^th century, vibrato helps to distinguish a solo line from its accompaniment. Singers, of course, were finding out how to project over increasingly large orchestras in ever larger opera houses, and they modified the colour of their voice as well, developing formants specific to trained singing. Violinists had less occasion for standing out from the crowd, though one could perhaps argue that concerto playing required a more forthright style than it had when orchestras and venues were smaller. But Katz’s argument is that violins registered better in early recordings when vibrato was wider and continuous. Secondly, it covered over bowing noise that became audible once electrical recording was introduced in 1925 (though this is of course an argument about a later period). Thirdly, vibrato disguised poor intonation. How poor soloists’ intonation was around 1900 it’s hard to tell—there aren’t enough recordings to support much generalisation—but while there are polite reasons for not supposing it any worse than now, there really is so much imprecise orchestral playing on early records that it’s tempting to suppose that standards were at least different. So, although it’s not an attractive argument for continuous vibrato, Katz may have a point. Fourthly, (this is actually Katz’s third point, but I’ve separated out his second argument into two) Katz points to the lack of the visual dimension, which to early listeners at home must have seemed so peculiar, and he sees continuous vibrato as a way of indicating an emotional engagement that listeners could no longer gauge from facial expression and bodily gesture. Fifthly, Katz suggests that violinists may have placed more weight on vibrato because it gave them a means of differentiating their own sound from anyone else’s. ⁸ A counter-argument might be that violinists still use continuous vibrato but never tire of arguing that nowadays everyone sounds the same. ⁹ There’s no doubt, though, that in the early decades of continuous vibrato they didn’t; so again Katz may well have a point.

¶6 On the other hand, the dates don’t really work: the change had already happened before recording began. Only in the oldest performers, born before 1850, do we hear the old style. Among younger players it was already disappearing, and the generation born after 1870 used it not at all. That generation would have developed their own personal styles in the 1890s before commercial recording began, and long before recordings became common enough to influence players (which I would think was not until at least the 1920s). The more conventional explanation for the new more demonstrative style, that orchestras had become larger and louder and so ways had to be found for soloists to penetrate through these bigger sound textures, may be good enough, but without recorded evidence covering the change we can’t know for sure.

Fritz Kreisler

¶7 No doubt much more will be said on this subject over the next few years. For us it’s probably enough simply to observe what the recordings seem to tell us, which is that already at the start of the 20^th century, although older players like Joachim, as with singers like Patti, tended not to use much vibrato, younger players were already using it all the time. Fritz Kreisler is always cited as the father of continuous vibrato, ¹⁰ and while his recordings certainly offer very clear examples of its early use, it seems very unlikely that the cause can have been so simple. And indeed when we listen to a wider range of early recorded players it quickly becomes clear that there was an increasing use of vibrato by the generation that followed Joachim’s; ¹¹ the differences, however, are nothing like as striking as some discussions have suggested.

¶8 Table 1 sets out some rough figures for rate (speed) and extent (width) in a chronological sequence of players. ¹² (The numbers were obtained by reading off the timings and frequencies in a spectrum analysis program ¹³ and converting the latter into fractions of a semitone. ¹⁴ Since there’s no practical way of measuring every cycle and averaging them it’s not a very reliable set of numbers, depending on one’s ability to make accurate measurements by hand, and on one’s patience to make enough of them. I don’t make any great claims on either count, but it gives a rough idea, and I hope is accurate enough to allow comparison later on with more modern players.) ¹⁵

¶9 Joachim’s own palette, as I’ve suggested, included a range of sounds and styles, including light and infrequent vibrato in his Brahms playing. Leopold Auer, born in 1845, was recorded in 1920 playing Tchaikovsky’s ‘Melodie’ Op. 42 no. 3 (arranged, like other early recorded favourites, by August Wilhelmj). Of course by 1920 he may have absorbed the manners of younger players, but when he objected strongly to continuous vibrato in his textbook from 1921 I think we can assume that his own, which was continuous but narrow, wasn’t the kind of vibrato he was talking about. ¹⁶ Whatever he found objectionable, then, it’s unlikely to have been the principle of continuous vibrato, but rather a recent practice more noticeable than his own, for example Kreisler’s or even younger players born in the 1880s and 90s. His vibrato shows at least the beginnings of a tendency―perhaps, in fact, the continuation of a tradition. Certainly the next generation was using light but noticeable vibrato much of the time. ¹⁷ Arnold Rosé, born in 1863, is sometimes said to have been one of the last orchestral leaders (of the Vienna Philharmonic) to insist on orchestral playing ‘without vibrato’; ¹⁸ but it’s clear from his recordings that ‘without vibrato’ is a comparative, not an absolute. And players born only a bit later use it all the time, albeit still relatively lightly; examples include the American Maud Powell, born in 1868, and the immensely influential teacher Carl Flesch, born in 1873.

¶10 There is one important difference, though, between the older players’ continuous vibrato and that developed by the younger ones. Joachim, Auer, Viardot, and Rosé, have a vibrato that varies from cycle to cycle, slightly in speed and considerably in depth, without any apparent cause related to the composition. But in Flesch and then Kreisler, followed by Hubermann and Heifetz, we begin to hear players varying vibrato according to changes in emotional temperature within a phrase. At high-points dynamics are louder and vibrato is deeper, at low points, especially phrase-ends, dynamics and vibrato both tail off. To judge by the Violinschulen of Spohr (1832), and later David (1864), the linking vibrato speed and loudness goes back a very long way, ¹⁹ but both authors are discussing single notes requiring special treatment, not a continuous vibrato that varies from moment to moment as the melody changes in pitch and loudness, which seems to be a later development. Whether this generation of players born in the 1870s was really the first to adopt it we cannot tell without more and earlier recordings.

¶11 Because this continuous change in vibrato extent happens in coordination with changes in the musical surface one hardly notices it in normal listening. This is a phenomenon comparable to perceptions of rubato: research by Bruno Repp has shown that people do not easily notice it in conventional locations (for example phrase-ends). ²⁰ But with visualisation tools such as spectrum analysis software (which we’ll discuss in more detail in chapter 8) it’s much easier to see. So what strikes one above all about the development of Kreisler’s performances as one listens to them is not so much their flexibility coordinated with the score, but rather the way in which his vibrato becomes a lot more even. The following images show the same note from the 1912 and the first of the 1926 recordings of Kreisler’s Liebesleid (approx. 1’4”–1’6” into each). Sound File 19 (wav file) plays them in chronological order.

¶12 The difference may not look great, but consider that the uneveness in the height of the cycles in the 1912 note differ by almost the depth of half a cycle, and the depth of the cycles themselves varies by almost the same amount, and you can see why the 1926 note sounds more regular. ²¹ It’s not that the uneveness is ever great enough to sound rough―on the contrary, as we’ve seen it sounds rather colourful―but rather that Kreisler seems to have developed the skill to be much more regular, and presumably preferred the results. The downside, and of course this is a matter of opinion, is that the sound is just a bit less interesting, a bit more mechanical and perfect, rich but unvarying. ²²

¶13 How reliable this sort of chronological snap-shooting is as evidence for the development of vibrato remains to be seen, of course; a much more thorough study is required. ²³ Birth date can only be one factor, and seems unlikely to be more important than teacher, or the years and place in which a student was first exposed to a range of other fine players. Clearly there were plenty of tendencies leading towards wider, more even, and continuous vibrato. But we simply don’t have enough evidence to show that it became a norm as late as the early 20^th century, nor that Kreisler was responsible for its universal adoption.

¶14 In that case, what is it about Kreisler that makes him so convenient a father-figure for modern violin sound? He was born in 1875, which makes him somewhat younger than any violinist mentioned so far, and hmade began to record in earnest in 1910, by which time he already had a glowing international reputation (he gave the first performance of Elgar’s concerto in that year). For anyone already familiar with the Kreisler legend these early recordings come as a bit of a surprise; for, like Joachim, Kreisler has a variable style. ²⁴ His very first Victor recording, made on 11 May, of Smetana’s ‘Bohemian Fantasie’ shows some of the sound that one expects from his later recordings. ²⁵ It’s unnaturally dull, thanks to the limitations of the relatively early acoustic technology, but still hints at the silvery top which is so characteristic. We also find a continuous vibrato of much the width that we hear on his later, better-known discs. On the other hand, his Bach Gavotte (from the third partita), recorded two weeks later on May 24, has a lighter vibrato, hardly more than Auer in Tchaikovsky and less than Viardot or Powell, and like Joachim Kreisler doesn’t use it on the shorter notes. A Schubert ‘Moment musical’ recorded on May 18 has very slightly more. Kreisler’s own Liebesleid, recorded the following year, has more again, but still less than the Smetana. ²⁶ So at first Kreisler was really not so unlike his contemporaries in so far as he modified his use of vibrato according to his perception of the character of the piece he was playing. (It is a character judgement he’s making, incidentally, not a historical one, because he plays Gluck with at least as much vibrato as Smetana.) If there’s a difference it’s that he clearly is working several notches further along a line towards continuous wide vibrato. Nothing he plays lacks it entirely; and at its widest it’s somewhat wider than that of his contemporaries in 1910. But it’s hard to believe that at that date he would have seemed extraordinary simply on that account. ²⁷ Other factors must have been in play.

¶15 I suspect that it had to do with timbre, which is partly a function of vibrato but contains more complex elements to do especially with pressure and speed of bowing. The sound itself, even in these quite early acoustic recordings, is coloured in a very particular way which does seem to mark it out from others. It sounds as if it has some bright upper partials and a glowing middle, ²⁸ but in fact there are no sounds recorded above about 5000Hz on the 1910 discs, and so the only way of accounting for this effect is through rapidly varying amplitude among the few partials that are there. Light but very precisely controlled bowing, generating a high rate of slipping and catching of the string by the bow, could be its cause. ²⁹ Add to this the continuous vibrato, slightly wider than was common, and you get a lot of movement among the partials at a level too fast for the brain to register as anything other than an impression of liveliness in the sound. And that is exactly what is so characteristic of Kreisler’s playing, characteristic too of singing, and it was the vocal quality of Kreisler’s musicianship that so struck his contemporaries. Louis Lochner, in his biography, quotes a remarkable 1908 review by the Chicago critic W.L. Hubbard of a Kreisler concert which seems to be describing a vocal recital until Kreisler’s name is revealed at the end. His sound, Hubbard writes, has a low register of ‘wondrous warmth and richness, its middle portion brilliant and vibrant, ... and its upper tone being of a clarity, a sweetness and an exquisite finesse that ravish the sense.’ ³⁰

¶16 The Bach double concerto recording with Efrem Zimbalist, from 1915, provides a convenient test for our ability to hear the differences listed in Table 1. Though both players come from broadly the same stylistic world, it’s easy to distinguish between them by ear, without a score, because Zimbalist’s sound is duller (probably because he bowed with higher pressure on the string) and his vibrato is noticeably slower: 4/100ths of a second for each vibrato cycle may not seem much (and incidentally it’s extremely approximate, not least because both players still use uneven speeds and depths) but it is enough for us to be consciously aware of the difference.

¶17 Kreisler’s discography is huge for his time, judging by Creighton’s listing larger than any other violinist born before 1890. ³¹ So it’s easy to understand why commentators like Flesch felt that he was overwhelmingly influential: he very probably was. But I suggest it was for more complex reasons than just his continuous vibrato. He made a sound that, because its interesting features did not depend entirely on the higher partials, was transmitted reasonably well on record. Nevertheless, when we get to his electrical recordings we realise how much we’ve been missing. If you recall my warnings in the last chapter about the kinds of things we can and cannot determine from recordings on 78rpm shellac discs you’ll have noticed that I’ve been quite careful not to try to talk about Kreisler’s sound in terms of the balance of the harmonics. So much depends on the recording and the transfer. And so we need to continue to be cautious here. The American electricals (transferred on RCA) sound much brighter than the acoustics, and one would expect that; but they also sound thinner, the 1949 disc almost tinny. By contrast, the HMV discs (transferred by Mark Obert-Thorn for Naxos) sound quite different, much warmer, an altogether richer and more beautiful sound no one could ever have expected. Which is more like Kreisler? At the moment it’s impossible to say. The HMVs on Naxos sound more lifelike to a modern listener; but then ‘lifelike to the modern listener’ is the aim one can reasonably expect of Obert-Thorn in making the transfer. It’s a circular process from which there is currently no escape. That said, a comparison of the transfers of my next example, based on (presumably) different copies of the same issue, shows Obert-Thorn producing a far superior result to his unnamed rival on the Symposium label, with a greater range of frequencies present and greater clarity among the instruments. In the Kreisler case, though, we’re not comparing like with like—the RCA and HMV discs carry different recordings. All this shows is that an accumulation of different transfers of the same originals over time might amount to some kind of evidence for Kreisler’s sound in so far as it was recorded on shellac. You can see, then, why I recommend caution.

¶18 I’ve spent several pages talking about Kreisler’s sound both because it’s loomed so large in the mythology of 20^th-century violin playing, but also to show just how much work there is still to do before we can begin to feel confident that we understand how string playing changed in the early years of recording. It’s also allowed me to discuss some of the aspects of sound, and ways of comparing sounds, that will be useful in other contexts, especially when we come to look in more detail, in Chapter 8, at expressive gestures. For the moment, though, I want to move on through the century, taking as my examples of later violin playing a sequence of concerto performances. Once we get into the electrical era, and orchestral pieces can be adequately recorded, concertos begin to function as the shop window for violinists, and we can think of them as offering fair displays of each player’s style. ³²

Changing vibrato

¶19 It’s helpful to begin by going back to Flesch, since he continued to record through the 1930s and was so influential as a teacher and writer. His electrical recordings (see Table 1 and 2) show that his vibrato changed not at all from 1905 to 1936, and that it was always deep and slow compared to Joachim or Kreisler. And if we compare his electrical recording of the Brahms concerto, preserved on acetates as yet undated, with that of Bronislaw Hubermann, born in 1882, we can see an even more ‘extreme’ case in another of the older recorded violinists, suggesting that any notion that vibrato was narrow and fast at the beginning of the 20^th century is an over-simplification.

¶20 In fact vibrato characteristics change far less clearly across Table 2 than one might have expected, and if we compare Hubermann with, for example, Mutter it’s hard to say who is the more ‘modern’. More interesting, if we want to trace developments, is Hubermann’s varying use of vibrato, which tends to get deeper as a note becomes louder, and also as phrases rise towards a peak, while lower notes tend to have shallower vibrato, sometimes reduced to almost none as phrase-ends come in to land. For Hubermann, then, there’s a clear connection between vibrato depth and emotional weight; which is something we discerned on a considerably smaller scale in Joachim, and to some extent in Flesch, but only rarely in Kreisler (Table 1). From Hubermann onwards, though, every player up to the present day makes this relationship absolutely consistently. Is it possible that the tying together of expressivity in loudness, vibrato and the composition’s surface—which has seemed so obvious to players ever since that it was not questioned again until so-called ‘Historically Informed Performance’—really did not happen before players born in the 1880s? It seems extraordinarily unlikely, and it may be simply that the generations represented at the top of Table 1 recorded too little for us to hear examples of it, or that it temporarily fell out of use during a more ‘puritan’ phase in the history of performance style. It’s a question for further research. Whatever the reason, this change looks and sounds very significant, and signals a development towards a more overt expressivity, one that in Hubermann is particularly obvious because his vibrato is so slow and so deep.

¶21 Hubermann studied first at the Warsaw conservatory, and then with Joachim’s assistant in Berlin (among others). Josef Szigeti, born a decade later in 1892, followed a similar path, studying as a child in Budapest under Hubay, but making his debut in Berlin (1905), where he heard Elman, Kreisler and Ysaye; he lived in Britain for six years in his teens. So in his case we’re hearing the playing of a relatively young man (as recorded violinists go) who’d grown up with Kreisler as one of a number of available models. There is an early Bach recording, of the Prelude to the E major partita, made in 1908, ³³ with virtually no vibrato at all until the closing bars, ³⁴ where its characteristics are much the same as Joachim’s. But it is hard to imagine that the Brahms Concerto (Table 2) is the work of the same player; and in a sense, when you consider his age in 1908, it isn’t. So naturally enough it is with the later Kreisler that he can be most sensibly compared. A tendency noticeable in Kreisler’s recordings of the 1910s and 20s is taken a small step further here, namely an increase in rubato, and especially in the use of rubato, crescendo, portamento, and widening vibrato together to mark the most expressive notes of a musical phrase. ³⁵ In other words, just as we saw in Hubermann’s vibrato use, there’s a clear inflation in expressivity; and when one considers the date and the sorts of things that singers were doing at the same time, it’s not surprising that for a young musician in the 20s a more intensely expressive style seemed entirely natural. (Also noticeable is that the speed of Szigeti’s vibrato is much more regular, hardly varying at all.)

¶22 Jascha Heifetz is another decade younger, born in 1901, and studied with Auer in his native Russia before moving to America in 1917. Like so many in this survey, he made his debut in Berlin; and so despite their varied origins, it's clear that these players gravitated towards an international career and an equally international style. What differentiates Heifetz has much to do with his extremely flexible vibrato usage. ³⁶ In his Brahms slow movement, for example, high notes have the deepest and fastest vibrato, low notes the most shallow and slow, all of which forms a more complex picture than one might think. Deep, fast and slow can all be used to signal feeling; what kind of feeling depends on the combination: deep plus fast tends to suggest excitement, while slow plus shallow suggests heartfelt feeling but of a more restrained sort. The low notes add into the mix the richest sounds Heifetz makes. In other words, he has a number of different ways of producing intense expressivity, and tends to make different effects in different registers, giving a sense of lively responsiveness to the changing surface of the music. Individual notes tend to be quite even, so his playing sounds regular and controlled and yet intensely engaging, which matches well with the many reports of a striking contrast between his inexpressive appearance and highly expressive sounds. ³⁷ In fact, while commenting on how he looked in performance they were, without realising it, talking about the sounds too.

¶23 Joachim Hartnack, writing from a more conventional, perhaps even chauvinist perspective, ³⁸ and for a moment recalling Adorno, was one commentator who loathed the evenness of Heifetz’s style, and his denunciation of it is worth quoting both for its tone and also because it points to one way in which performance style can seem to relate to wider cultural tendencies.

It is the consummate expression of a widely-accepted American ideal of beauty, fulfilled in the pure aestheticism from which all impurities, but also all problems have been eliminated. The norm of this ideal corresponds to the average of the modern consumer society. Its lower limits are delineated by the chrome and glass of automobiles, the middle range shaped by pretty and boring advertising models who woo the consumer with their sterile smile to buy Cola-Cola or toothpaste, and the upper limit is formed by the aesthetics of cubism as expressed in the architecture of skyscrapers or by the smoothness of sentiment and tone of a Heifetz. ³⁹

¶24 Nathan Milstein, just two years younger than Heifetz, followed close in his footsteps studying with Auer at the St Petersburg conservatoire, and it may not be too surprising, then, that his vibrato is very similar in its speed and range of depth, ⁴⁰ although Milstein varies depth according to expressive intensity without tying that so closely to tessitura. As with Heifetz, individual notes are very even, though Milstein shapes each by taking several vibrato cycles to reach its full extent: each note, in other words, begins with a vibrato ‘crescendo’ which functions as a form of articulation, giving each a subtler quality of yearning than can be achieved by a loudness crescendo. Both players use vibrato even on the short triplet semiquavers, which earlier players (Hubermann, Kreisler, Flesch) more often left partly plain.

¶25 Neveu’s two performances of the Brahms are very similar, at least in those details that are wholly under her control, and her tendency to deep vibrato and to wide tempo variation (not always successfully reined in by Issay Dobrowen, the conductor of the 1946 performance) gives her personal style a particular intensity which, as we shall see when we analyse the Table 2 statistics more closely, is exceptional for her generation. The unevenness of vibrato we saw in Kreisler’s playing is characteristic of hers as well, but of no one else in this comparison, and it too may contribute to the passion and sense of engagement that commentators particularly sensed in her playing.

¶26 Stern’s vibrato shows a very clear link between speed, depth and expressivity. As in all these players, but to a marked degree here, faster = deeper = more expressive; and so in this Brahms extract the first, ‘winding-up’ half intensifies in each dimension, while the second, winding-down, relaxes. Interestingly, and exceptionally, Stern doesn’t use portamento on the way up, only on the way down, and it’s possible that it’s used to compensate for the winding down effect of melody and vibrato, so that latter part of this passage, ending the slow movement, doesn’t sound any less expressive overall.

¶27 As we work though these performances I hope it’s becoming clearer how individual styles differ within a really rather consistent extended period style, so extended in fact that were it not for the emerging influence of HIP in mainstream playing during the 1990s, one might almost suppose that violin style became largely set in stone with the generation following Kreisler. In fact, the picture that emerges from players born after 1940 is surprisingly varied, and it may be that it was precisely the stability of general style that made this possible. Kremer, Mutter, and Shaham have deeper and (less so Kremer) faster vibrato than became the norm. Mutter in particular goes to the extreme end of her range of speed and depth a lot, and her vibrato is very closely tied to dynamics. As a consequence she stands out far from the orchestral accompaniment, very much the soloist. It’s also a slow performance so a lot of vibrating gets done on each melody note. If intensity is to be communicated through vibrato and dynamics it’s hard to imagine a performance of the Brahms Concerto getting more intense than this. She also uses a lot of portamenti—slightly more than most early recorded players, in fact—but on the other hand hers are much shorter. Portamento, one feels, is for Mutter simply a tool to ratchet up the expressivity that she seems so desperately to need. Some similar points could be made about Shaham’s performance. It might be tempting to argue that these two owe something to the Berlin Philharmonic, with which both are playing on these recordings, but not only are the conductors different (Karajan and Abbado), these features of personal style are far too ingrained to be turned on and off according to the accompaniment.

¶28 Among the youngest Brahms players here, Vengerov uses almost no portamento and rather slow vibrato, while Barton uses a lot of portamento—the most since Mutter and Kremer—but rather fast vibrato (mostly, despite her wide range, staying at 0.15s per cycle, together with Shaham the fastest since Stern). Barton writes historically in her CD booklet and pairs the Brahms with its supposed model, Joachim’s 2^nd concerto, as well as using a violin Brahms knew, so it’s quite likely that her increased use of portamento is also an attempt to be historical, an early sign of the growing influence of historical recordings in mainstream playing, while Vengerov is old enough to have been influenced by HIP—amusingly generating more or less opposite tendencies.

¶29 Turning to the Beethoven Concerto, it’s worth noting first of all how similar are the 1926 and 1936 performances of Kreisler. It’s a cliché that no one can give the same performance twice, but in fact there are plenty of examples on record of a popular performer in a popular piece playing in a very similar way. Neveu’s Brahms is one example. (Another is the Myra Hess case mentioned already, and more will be noted later.) This is another way of saying that performances become (procedurally) routine when they’ve been given often enough. Compared to Kreisler’s Beethoven, Georg Kulenkampff’s sounds not dissimilar in many ways. Kulenkampff was born a generation later, in 1898, and made his recording in 1936; and while his vibrato can on occasion be deeper than Kreisler’s it is only in his much less frequent use of portamento that he sounds noticeably more modern.

¶30 Heifetz again varies speed and depth of vibrato; here, with fewer opportunities for lingering than in the Brahms slow movement, it’s more noticeably related to intensity. Relaxed, lower notes tend to have slower shallower vibrato; more intense, louder. Often higher notes tend to have faster and deeper vibrato, but it depends on the character of the performance from moment to moment; in other words it’s tied to Heifetz’s reading of the score as much as to the score directly. He uses changes in the dimensions of his vibrato, in other words, to characterise passages as he wishes. For example, in the first movement of the Beethoven Concerto the passage from bars 335-9 ‘modulates’ from 0.3st to 0.6 and back to zero as the line ascends and then descends; in the decorated restatement that follows one particularly intense note (bar 343i) almost reaches a semitone depth and 0.12secs speed. Heifetz shows, then, an extremely skilful and flexible matching of vibrato depth, speed, dynamics and rubato to respond sensitively to the momentarily changing character perceived in the score: and the relation between these elements is constantly changing.

¶31 Although they are of very different generations, Perlman (born 1945) like Milstein (born 1903) uses ‘hairpin’ vibrato on the main melody notes, especially in the slow movement, which tend at the most expressive moments to be matched to dynamic hairpins. The notion of shaping a note by increasing and then decreasing one of its dimensions is something that recurs in different forms throughout the recorded history of performance—it’s a defining feature of HIP, of course—but its application to vibrato depth is relatively unusual, and in mainstream violin playing a feature of personal rather than period style.

¶32 Chung (born 1948) is especially interesting because her vibrato speed varies a lot, and so seems to play as important a part in the changing emotional surface of her playing as changes in depth. And she seems able to separate them to allow a wider palette of effects. At the start of the Beethoven slow movement extract, bar 45, for example, a very slow and very shallow vibrato combines with the slow speed and low amplitude to produce intense stillness which gradually becomes more active with each subsequent note as the vibrato speeds up and becomes slightly wider, and the loudness increases through to bar 47. The same happens again through bars 48-9, with the chromatic a’# shallower and even slower than in bar 45 but louder, which makes the effect distinctly less restful: we expect speed, depth and loudness to go together, and here when they don’t it’s appropriately unsettling. This is perhaps a symptom of the increasing control achieved by players later in the century—it’s certainly hard to imagine Kreisler, with his unsteady vibrato, managing this in any consistent way. Chung’s Rondo-Allegro is much narrower in range of effects, reflecting a simple view of the character of middle and final movements.

¶33 Joshua Bell (born 1967) is conducted by Roger Norrington, so speeds are fast and vibrato is presumably discouraged; but Bell occasionally forgets himself and reverts to normal modern practice at especially expressive moments. Thus although most of his vibrato is confined between 0.3 and 0.5 semitones, there are some peaks at 0.7. Nothing else distinguishes his performance, apart from some scrambling for notes, poor editing and more than average portamento in the Beethoven 3^rd movement episode theme, perhaps encouraged by Norrington in the light of early recordings.

¶34 Hahn, the youngest player here, born in 1979, shows a more recent trend, in which (influenced without doubt by HIP) vibrato is reducing and narrowing again, although the equation ‘louder = faster = deeper = more expressive intensity’ still applies. Like almost everyone else she uses faster vibrato in the faster movement, again probably because she’s more excited (as opposed to moved) by the more lively composition. One might think there’s a contradiction here, but in fact much of the expressivity of slow movements comes from their stillness. ⁴¹

Portamento

¶35 Focussing on vibrato use has shown up quite a number of significant points of style in violin playing. We’ve seen how speed and depth can be fixed or flexible, how when flexible it can respond to changes in melodic direction and tessitura, and how it can be linked to changes in loudness, so that small changes in this one dimension can have very significant effects for the perception of expressivity. What we’ve not seen very much, at least not since players born in the 1880s, is fundamental change in general period style. And in this sense violin playing looks rather unlike singing. We’ll return to this in a moment, but first let’s look a little more closely at portamento, that other means of being expressive with pitch for which stringed instruments are so well adapted.

¶36 Figures 3 and 4 show parts of the extracts from the Beethoven and Brahms concertos used as a constant in Table 2. The numbers indicate the (very) approximate length in hundredths of a second of each portamento for each player. ⁴² A chart like this can be surprisingly informative. Given enough performances it allows us to distinguish between traditional and individual choices of portamento placing, and gives a sense of any trends. In Figure 4 (Figure 3 is less use because there are so few players) one sees at once how the lengths diminish quite suddenly after Hubermann. The number of portamenti used in each passage, however, remains strikingly stable across the century, with few exceptions (for example Hubermann who likes a lot, Stern who doesn’t); and again this suggests a much less consistent change in performance style over time than has been suggested, or than we saw in singing. Portamento remains common in violin playing throughout the century; we just don’t notice it so much.

¶37 For more precise observations, though, we need to treat these figures, and those for vibrato, as raw data and graph them. Trends will then become clear that are far harder to see in tables of numbers.

¶38 Figure 5 plots speed of vibrato (slower at the top, faster at the bottom) against birth date of the performers, and shows first how varied it was earlier in the century (among players born up to the 1920s) and how much more consistent it is now (players born from the 1940s onwards), and secondly how the overall trend has been towards slower vibrato. It’s worth bearing in mind that the speed given is a rather crude average of the range used by each player, and doesn’t account for tendencies to remain within a limited part of that range for most of the time. Thus Barton, for example, is misrepresented as having an average of 0.165s per vibrato cycle when in fact, as we’ve seen, she mostly sticks to 0.15. These sorts of details should ideally be improved in future research, although at the moment, given the current state of the software, to collect such detailed data would be immensely time-consuming.

¶39 Figure 6 plots the depth of vibrato (deeper at the top, more shallow at the bottom) against birth date, and presents a more complex picture. Overall there seems to be little change, but within that overall picture are a number of changes that may perhaps be related to other developments in violin playing. The increasing consistency among players born in the 1890s and 1900s—Szigeti, Kulenkampff, Milstein, and also Schneiderhahn—may relate to the growing internationalisation of playing as more violinists gravitated towards the same centres for study and their early careers. Leaving aside Neveu as exceptional, which she was in other ways too as we’ve seen, the tendency through Shumsky, Stern (born in the late 1910s), Perlman and Chung (the late 1940s) is towards shallower vibrato; but overlapping that, one can also see a deepening from Perlman (born 1945) through to Shaham (born 1971), reflecting the intensification of overt expressivity through the generations for whom the Amadeus Quartet and Karajan’s Berlin Philharmonic, in their different ways emblematic of professional emotionalism, were the acme of recorded musical excellence. And then from Shaham through to Hahn there is a very obvious trend towards ever shallower vibrato, surely brought about by the influence of HIP and the spread into mainstream music-making of its determined reaction against exactly that kind of heavy, almost plastered-on emotional gesture.

¶40 There are plenty of other stories one could tell using the graphs in Figures 5 and 6. What we really need is more hard evidence, and it should be one task of future research to gather it. Turning to portamento, Figure 7 plots the number of portamenti in the chosen passages against birth date. For these purposes the Beethoven extract is really too small to compare well with the Brahms, but there are enough similarities to suggest that it’s not useless. Again, a targeted research project—as opposed to an illustration of a method, which is the main intention here—would generate much more data. ⁴³

¶41 Nevertheless, some trends are very obvious and—surprisingly—are clearly related to vibrato practice. Portamento seems here to be used increasingly among players born up to the 1900s, at the same time that vibrato was deepening; then it falls off rapidly for players born in the 1910s, just as vibrato become shallower; portamento use increases again for players born in the 40s and 60s, like vibrato depth; and it falls off a lot among the youngest players (except Barton who, as we’ve seen, may have had reasons of historical principle, rather than general period style, for using more). ⁴⁴ But why is there this link with vibrato? Presumably the two are perceived as generators of more overt expressivity, and it’s this underlying desire for more or less intense expression that is changing. What underlies that, in turn, is likely to be extra-musical, a more general trend in styles of emotional communication in society reflecting something really quite fundamental and widespread about the way people think and behave.

¶42 Figure 8 inflects this story in a rather interesting way, plotting the average length of a player’s portamenti against their birth dates. This is simply a graphic representation of the data in Figures 3 and 4. There was a sudden drop after Hubermann, then a steady decline throughout the century until the sudden leap with Shaham, Vengerov and Barton (all born in the 1970s). Long portamento became an embarrassment until very recently, but it still has quite some way to go before it reaches the style practised at the start of the 20^th century. Indeed, the fact that it’s still generally supposed that portamento is not much used any more—despite the wealth of evidence here that it is—suggests that there is some kind of perceptual threshold between 150ms and 180ms that has yet to be crossed. Below it portamento is perceived as an ornament, above as an exaggeration. It will be interesting to see what happens next.

¶43 Table 3 shows the standard deviations for portamento lengths. Standard deviation represents the variety of lengths used by each player: the higher the number the greater the differences among lengths used. We need to be careful not to read too much into these figures, because differences in lengths are small enough in real time not to be perceptually very striking, and so it’s no surprise that those who use longer slides will show more variation among them. But even so it’s striking the extent to which Flesch made more varied use of portamento than anyone else. Shaham, with his idiosyncratic taste (among modern players) for long portamenti, comes a not very close second. It’s also interesting that players born in the 1910s, whom we saw using less portamento and shallower vibrato, nevertheless make relatively varied use of portamento compared to both earlier and later violinists. I hope this emphasises the subtlety of changes in performance style. It’s all too easy, as I’ve unfortunately demonstrated again and again, to summarize performance styles in crude generalisations, but in fact what’s going on when styles change is a lot more complex, with the constituent elements of style changing in different ways. It’s why performance styles never recur, even when we try to make them.

¶44 Table 4, finally, shows for all the Brahms extracts the relative loudness of the slide compared to the main notes on either side. We’ve already seen how the idea that portamento was dropped early in the 20^th century is contradicted by the concerto evidence. Certainly it was reduced, but there has been plenty of it ever since. Could it be, I wondered, that our perception of portamento depends not just on the number and length of slides, but also on their loudness relative to the notes they join? The point of Table 4 was to test the hypothesis. It seems to me that the answer is a clear no. There is no consistent pattern of increased, decreased or steady loudness either across time, or according to musical context, or within the work of any player. A graphic representation of these numbers, using a spreadsheet’s graphing facility to turn them into directional lines coded by colour for each performer helps make this even clearer. (Figure 9) Portamento loudness, on this very limited evidence, seems entirely accidental, not a factor controlled by players.

¶45 That’s not to say, however, that we don’t perceive it as playing some part in the effect. If we compare the portamenti that end bar 100 in the Hubermann and Busch performances of the Brahms extract we can hear that Hubermann uses a long soft slide where Busch uses a short loud one, yet the effect feels rather similar. A possible explanation is that there’s some sort of balancing of length and loudness, so that an increase in one dimension can be counteracted by a decrease in the other, or emphasised if one wants a stronger effect. At any rate, examples like this could be studied in greater quantity and with greater accuracy in order to test the extents to which we perceive loudness and length as active constituents in the effect that a portamento has on us as listeners. So even though players seem to control only length, both length and loudness may matter, and may be able to work together or against each other. But this is a tiny sample, and clearly measurements need to be made and analysed over much larger pieces. The main technical obstacle is that, so long as analysis software is unable to match the ear’s ability to focus on the loudness of solo line regardless of orchestral accompaniment, it’s going to be necessary to take solo pieces or long solo passages (cadenzas from concerti would do if enough players used the same one, which here they don’t) recorded by many players over long periods of time. It would be a worthwhile experiment, though.

¶46 One other factor in our perception of portamento may very well be the shape of a slide—for example rising fast at first and then tapering off, or vice versa—and its relation to the kind of slide (from the first note or to the second) that’s being used. Since other factors seem not to, it may be this that makes early-20^th and early-21^st century portamenti sound significantly different, even on occasions when the lengths of slides are similar. At present I’m not aware of a manageable way of mapping and quantifying those details, short of differential calculus, but it would be worth further investigation.

Conclusions

¶47 What can we conclude about changing styles of violin playing on record, and their relationship to singing? Early in the 20^th century vibrato speed appears to slow down, but only if we include Joachim. We can’t know, because we have no other players of his generation on record, but if he were exceptional then we’d see vibrato speeding up through to players born in the 1900s, which just goes to show how fragile any conclusions about 19^th-century violin playing must be. With or without him, vibrato depth increases through to players born in the 1900s, and then on average remains fairly steady until recent times. Vibrato was continuously present on all but the shortest notes from at least Auer (born 1845) onwards, but became noticeable and discussed only once it got deeper, from Flesch and Kreisler (born in the 1870s). From Flesch sometimes, and regularly from Hubermann (born 1882) for ever after, vibrato was varied in depth from moment to moment, coordinated with changes in loudness, in order to give expression to changes in musical character.

¶48 What do all these changes add up to? It seems safe to assume that deeper vibrato signals deeper expressivity. We’ll look more carefully at these kinds of associations in chapter 8, but the analogy with voices trembling with emotion seems fairly uncontentious here. ⁴⁵ There are acoustic reasons for using vibrato—it makes a solo instrument stand out better from an orchestral accompaniment—but there are clear emotional connotations too. Portamento works through more complex associations—with infant-directed speech—that I’ve examined in a separate study. ⁴⁶ I won’t rehearse the findings here: suffice it to say that there are good reasons why we perceive portamento as emotionally moving. But what of vibrato speed? Is it more expressive when it’s fast or when it’s slow? I’ve already suggested that either is possible when one considers it alone. But the brain is promiscuous in making relationships between signifiers and signifieds, and one sign alone is rarely enough. You need several pointing in the same direction, for example vibrato speed, vibrato depth, tessitura, melodic shape, harmonic direction, loudness. ⁴⁷ If all the others point towards greater expressivity (for example a rising line, increasing loudness, harmonic crux, vibrato depth) then it doesn’t matter whether vibrato is faster or slower than usual, either will be taken to signify greater expressive intensity. This is a principle with important ramifications for our understanding of musical performance, and one to which we’ll return. ⁴⁸ The possibilities for combining details in new ways are far greater because of it, so style can change more radically and in more directions from era to era.

¶49 How does it work in practice for violinists? Probably rather simply: the linking of vibrato rate, depth, loudness, and melodic shape signals the performer’s wholehearted physical and emotional involvement in the music; genuine excitement, one imagines as an innocent listener, causes the finger to move faster, the bow to press harder, as the music gets more intense, and to relax when the music relaxes. Of course we can see that this is hopelessly naive: ‘the music’ is not just the score, but what the player does to it, so what is really happening involves the player’s learned response to certain kinds of melodic and harmonic configurations which they choose to read as requiring certain levels of intensity for their most effective expression in performance, and these levels are then signalled by physical gestures that cause sounds with particular associations for knowledgeable listeners. The experience, though, depends on listeners, and at their most involved performers too, believing and feeling that these kinds of sounds most naturally represent feelings expressed by the composer in the score.

¶50 Given that context, it’s reasonable to understand the changes we’ve seen in violin playing as reflecting a gradual inflation in expressivity through the first few decades of the century, which does indeed parallel that seen in singing, and in fact in all genres of classical music-making on record. But it’s much less obvious than it was in singing, and overall violin playing has been much more consistent over the last 100 years. Why?

¶51 With all respect to instrumentalists, any instrument is more limited than the voice in the means at its disposal. And so the possibilities for recombining the physical capabilities (in this case continuously changing pitch and loudness) are more limited. The relative inability of instruments to vary the spectrum from microsecond to microsecond, causing a relatively fixed tone and making it impossible to produce vowels and consonants, and hence text, inevitably limits the expressive options. So, it’s particularly interesting that it’s singing that violin playing has always been said to imitate, with violinists considered the divas of instrumental playing. The ease with which a violinist produces portamento and vibrato is, of course, the main reason. Just as in singing, vibrato and portamento seem to say, ‘see how intensely I feel this music’. But whereas in singing there are expressive reasons (in the text) why some notes can stand out from their surroundings, or disappear into them, allowing a much more variable surface, whose manner of variation can change over time, in instrumental playing there is no such reason, and notes that are suddenly louder or softer, or are attacked in very different ways, just seem awkward and out of place. With fewer opportunities for changeability, there are fewer possibilities for general style change over time. Instrumental playing is necessarily more consistent than singing.

¶52 It’s often said that there’s been a homogenisation of instrumental performance style, ⁴⁹ and in a way that’s true. We’ve seen that there was much more variation between players born before 1900. Whether we look at the graphs of vibrato speed (Figure 5) or portamento length (Figure 8) or (to a lesser extent) vibrato depth (Figure 6) there has been a narrowing of options. But in no case, except perhaps vibrato speed, has playing become really consistent. And we have too few players on record born before 1900 for us to be confident that we really know how people typically played before about 1920. I’m not sure, therefore, whether things are as bad as is sometimes suggested. But if there is one thing these examples do tell us it is that there are a lot more ways of making music on the violin than we’ve heard recently, and it might be rather interesting to hear some of them reappearing in new combinations in the future.

¶53 My other purpose here, of course, has been to illustrate methodologies: describing performers’ sounds and styles in words, both in detail and in general, gathering and analysing data, using graphs, thinking about how we perceive details and their interrelation when we listen to a performance. I hope readers have enough ideas, now, about ways in which one might study style change for me to focus, in an introduction to 20^th-century piano playing, on other, perhaps more contentious issues.