• 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

7. Style change: causes and effects

¶1 It’s clear that performance style has changed hugely, and also that the mechanisms underlying style change must be hugely complex. We can achieve some clarification if we try to work out what elements make it up. ¹ One approach would be to think about what performers do. A very rough map is set out in Figure 16.

¶2 According to this model, a performer’s style is defined by an interaction of the properties of their instrument (potential sound and the ways it can be produced), what they can physically do, and what they choose to do. The sound of the instrument, however, results also from an interaction of potential plus period preferences for sound production. What performers can physically do, is a mix of the mechanics of their body plus practice, the way they have taught their body to make music; and practice is shaped by period and personal taste (especially by current ideas about what kinds of sounds and sequences of sounds are musical and appropriate), encouraging the development of certain habitual physical movements, for example particular kinds of phrasing and the fingering, bowing or breathing that produce those movements. ² What performers do by conscious choice is an interaction between their physical ability (as just defined), period taste and personal taste.

¶3 What you can see from that set of definitions is that period taste comes in at almost every point. It’s there because we know from recordings that performance style changes. Why it should change does not seem on the face of it a difficult question for a society that prizes individuality and variety to any extent. How it changes is much more interesting. It seems hard to imagine—and once we’ve seen how much musical expression draws on other kinds of expressivity it will seem inconceivable—that musical practice is self-contained, changing independently of changes in taste in the world around it. So one area for future investigation has to be the relationship between performance style and other kinds of styles, especially styles of communication, including staged communication.

¶4 Making a very modest start, comparing performance style only to styles of writing about music, I looked in chapter 4 at the case of Dietrich Fischer-Dieskau as a singer of Schubert, arguing that his approach to the character of the music led to writers seeing Schubert songs from a new, less innocent perspective. And in an earlier article I added to some of those points the case of Pierre Boulez, whose conducting career brought with it exposure he’d avoided as a composer both to problems of musical continuity and to wider performance traditions. This in turn led him to rethink his compositional style. And I showed how in response to later, more melodious performances writers on modernist music came to hear it in a quite different way: where once they had heard points, now they heard lines. ³ In both these cases the influence seemed to come from the performance to the writings, ⁴ and in both the change of view was profound, so radical in fact that the music appeared in retrospect to have been thoroughly misunderstood a generation before, even by those who composed it. As I proposed in chapter 2, as performances changed, the music they made—our view of the compositions, in fact our conceptions of these ‘works’—changed. This gives us some clues, I think, to the extent of change that we can expect in the understanding of many kinds of music, at any rate across the last one hundred years (probably, given the huge and rapid changes in composition style, across centuries before).

¶5 The changes I’ve mentioned so far were all intra-musical, but performance style can change in response to extra-musical changes in outlook as well. In an article on portamento, mentioned above in passing, I drew on the possible relationship outlined in chapter 4 between Fischer-Dieskau’s Schubert interpretation and the effect on people’s attitudes to artistic expression brought about by the horrors of the Second World War. ⁵ We’ve already seen how styles in all genres we’ve examined changed after that war, and the discussion of Schwarzkopf and Fischer-Dieskau gives some idea of why that might have happened. In the article I look at how that brought a more or less immediate end to an expressive technique that had been declining only quite gradually up until 1939. Where portamento had for a long time been charming, helping to signal uncomplicated feelings of security and love (though its association with ‘motherese’), it now seemed undeservedly self-indulgent, pathetically naive in the face of so much evidence that people could delight in the Austro-German musical classics, and the performing styles that before the war had seemed to belong to them, without being fundamentally well-intentioned towards fellow human beings. All that portamento, which had seemed to signal empathy, a willingness to be moved by the feelings being portrayed in music, was a sham. After the War it became an embarrassment.

¶6 Here, then, is one instance—but I hope a powerful one—of the dependence of musical performance style upon much wider changes in outlook. But note where the interrelationship is made. It comes very specifically in the modelling of emotional expressivity. As music is similar to other kinds of human communication, through its ability to sound like anger or love or calm or agitation and so on, so it is capable of being influenced by changes in the way in which those states of mind are expressed in the wider world. Let’s take an imaginary example. In a society in which triumph was felt to be tainted and its public expression inappropriate, for example after militaristic triumphalism had led to nuclear holocaust, grandiosity could play little part in current performance style. Because music can model grandiosity quite easily this would have consequences for playing and singing: much Beethoven and Berlioz, Wagner and Verdi would sound quite different. By contrast if society were to take violent exception for a while to television it would have no stylistic impact on music-making whatsoever: music can’t model television, because although TV transmits styles of human communication it does not generate them: it’s a thing, indexing a technology but not a person in the way that, as Watt & Ash have shown, music indexes the movements and emotional experiences of humans to the extent that it is perceived as behaving as if it were a person. ⁶ Portamento is a much more complex phenomenon than either of these examples, and consequently the factors that led to its rejection will be much more complex than I have suggested. But the interrelationship with wider culture is clearly operational.

¶7 We can reasonably conclude, therefore, that performance styles change not just internally, for reasons that make sense only to practising musicians, but in a relationship (sometimes more causal, perhaps, sometimes more responsive) with other complex patterns of change in expressive communication. A good place to look, then, for parallel changes ought to be theatre and film. Because both those reflect very directly current modes of communication between humans in a particular cultural context we ought, all other things being equal, to be able to see similarities between actorly and musical performing styles. Of course, all other things are not equal and it’s not as easy as it sounds. So tempting as it might be to see links between the extreme visual expressivity of faces in silent cinema and musical performance styles of the 1910s and 20s (rubato, portamento etc), the fact that film was silent obviously caused extreme exaggeration in facial expressivity. One could argue, perhaps, that the exaggeration shows all the more clearly the sorts of emotional responses people had at the time, but unless one could show that music was also an exaggerated form of emotional representation (which might be interesting, since it could be so) it would be hard to correlate the two convincingly. A better parallel might be between, say, the intense expressivity of acting style seen in the great Hollywood blockbusters of the 1930s, for example ‘Gone with the Wind’ (1939), and the performances of the most actorly singers of the period, such as Lotte Lehmann, or between performance styles in, say, Ingmar Bergman and Fischer-Dieskau. At any rate, music in performance is in this way much more like theatre or film than perhaps we have wanted it to be. Actors use the physical signs of emotional states in their performances, and we are moved by watching and listening to them. ⁷ Exactly how that works is a much more complicated issue, one that has been very little studied by film theorists so far. ⁸ Indeed, music cognition is far more advanced at the moment than film cognition. It seems reasonable to see their problems as alike but it is too early to know on what level these relationships can be most profitably explored. The subject has not yet been researched, but it would be surprising if relationships were not there to be found, given that both art forms are concerned above all else with communicating information about emotional states as they change over time.

¶8 Coming back to musical performance per se, another approach to trying to analyse style change would be to think about what performers do to scores. If one were to think again in terms of a hierarchical diagram like Figure 16 the universals would remain Time, Frequency and Amplitude, but immediately beneath them would come those aspects of music-making that are likely always to have been determined by naturally selected responses to sound. This is a topic that we’ll discuss in more depth in the next chapter, but it should not be too controversial at this stage to include among them imitations of sounds in the environment (rain, thunder, wind), of internal (pulse, breathing) and external bodily movements (running, walking, jumping, caressing), and of pre-linguistic communication through sound (shouts, murmurs, pitch glides and so on). Through these responses one could explain associations such as ‘faster and/or louder playing suggests more excitement, slower and/or quieter suggests more calm’ and so on. At this level, in other words, choices of style (whether composition or performance style) refer to universal physical/emotional states: one would expect to find these processes at work in the music-making of many cultures at most times, to the extent that their music was designed to influence states of mind. They give us, for example, most of the elements we need to conceive and perform motoric dance music.

¶9 But if we want to get into the model those features of western performance style that we hear changing over recorded performance history then we need to start to sift through a great many habits that are hard to disentangle. Relatively constant, one might suppose, is a tendency to slow down at the ends of phrases or to articulate them by altering the sounding lengths of notes. We can’t know how far this goes back in time. Although it’s documented already in early plainchant notation it’s not present in all world musics (nor in much western popular music). Clearly music can manage very well without it, but equally clearly it is so natural that, as I mentioned when discussing Kreisler’s playing, in experiments listeners are often not aware that it’s happening. ⁹ Is this an extended period feature or a universal therefore?

¶10 The difficulty of knowing how to place such practices in a nature—culture hierarchy arises because such features, although their use varies over time, nevertheless draw on innate responses to sound. They are not simply codes chosen for social reasons. In semiological terms they are not symbolic but rather iconic, they resemble the thing they describe (in this case slowing down in order to finish an activity). Similarly vibrato, which one might think should go quite late on into a tree-diagram of performance style, being a feature that one gets at some times and not others, in fact may be iconic, for example of a trembling voice and might thus index vulnerability (among other things, and apart from its acoustical benefits for projection); ¹⁰ tempo flexibility may be iconic of fluctuating emotional intensity and may thus index emotional involvement; portamento, as I’ve suggested, ¹¹ may be iconic of motherese and so may index love and security. ¹² And so on. How do we arrange these hierarchically, and considering how deep they go into our physical and psychological makeup would it make sense to try?

¶11 A better way of conceiving these constituents of performance style might be to imagine them laid out over a field of possibilities, some of which is visible to people at a particular time and some of which is not, the limits of visibility being set by general period beliefs and tastes. The elements interact in very complex ways because of the huge number of possible combinations. What we’re really talking about here are the ‘period taste’ elements of Figure 16, but as we can see they are in fact very much more than just period specific. What is characteristic of particular places and times is particular choices and combinations of elements (habits of timing, dynamics and pitch adjustment); it’s these combinations and interactions that change over time on a broad scale (period style) and from person to person on a smaller scale (personal style), from performance to performance on a still smaller scale, and from moment to moment at the musical surface. In principle, therefore, the same process operates on all time-scales: what we do in varying timing from moment to moment in performance is in principle not different from what happens to habits of rubato, for example, over centuries. All options are theoretically available at every level, though in practice most are masked by social constraints. From second to second physical changes in sound may be quite large or quite small in absolute terms, depending on the current style (lots of rubato, for example, or almost none), but the habits change little. Over larger spans of time (decades) they change a lot more, and quite new ways of listening have to have been developed in tandem.

¶12 If this seems artificially schematic consider some of the examples we’ve seen in the previous chapters. As Craig Sapp’s graphical analyses of Chopin mazurkas show so clearly, ¹³ expressivity operated typically from moment to moment earlier in the 20^th century, and at the next level up, from phrases to phrase more commonly later on. It’s not difficult to see different performances by the same person as a manifestation of expressivity operating on a larger scale: the performer’s expressive profile, their personal style, is mapped by just those differences and by the frame within which they remain. Similarly a period’s style is mapped by the differences between many musicians’ performances and by the limits within which they are confined. They too constitute together collective expressivity through music. It doesn’t make good sense to insist on a fundamental difference between any of these layers of expressive activity.

¶13 Conceiving them all together is another matter though. We are very good at adapting to changes on the small scale. In fact it’s that process that makes musical performances musical as small irregularities in the sound surface communicate some kind of meaning (however abstract or emotional) between performer and listener. But on a large scale it’s extremely hard for us to understand what on earth was going on. How could audiences in 1900 have taken seriously Lilli Lehmann’s singing or Leopold Auer’s playing? There can hardly be a listener to early recordings, unless brought up on them exclusively from birth, who has not initially had this reaction. The difficulty of this first encounter, and the time it takes listeners unfamiliar with them to enjoy early recordings, clearly indicate that people then understood what they heard quite differently from the way we understand it now.

¶14 To see this more clearly let’s return now to the example that opened chapter 4. We found there in the reactions of one modern critic, uncomfortable (I think we can assume) with early recordings, a degree of alarm at the idea that performances like Elena Gerhardt’s might be taken seriously today. Why alarm? What makes them not just alien but threatening? Presumably it’s that at some level we do recognise what it is that the singer is doing to be expressive, and that indeed she is expressive, powerfully so. We hate the way she does it, but that it works on its own terms makes it potentially challenging to the way we would do it today. Suppose young singers started to copy her? Would we then begin to experience a manner of emotional communication that threatened to rewrite our current rules of personal interaction, that might require us to feel more deeply, or to feel deeply more publicly about the power of this music? What would it be like to be in an audience that was collectively as deeply moved by ‘An die Musik’ as Gerhardt seemed to be? By current standards it would be embarrassing. It’s our habit to admire the beauty and poise and subtlety of Schubert, and to find sexual subtexts, ¹⁴ but not to be moved to intense feeling by his music in a way that others can see or share.

¶15 So we recognise what early recorded singers are communicating, by understanding the means they use, yet we don’t experience it as they did. We resist their experience in favour of our own, at any rate on first exposure to it. The question, then, is whether, in adapting to it as we become expert listeners to early recordings, we develop in ourselves the same responses that originally gave rise to that style and which it generated in listeners. Do those of us who now like the way Gerhardt’s generation sang and played make the same sense of it as they did? Well, yes and no. When we first hear a performance like this, and laugh at it, it’s because those signs of emotion that she makes now signal parody. Somebody singing like that today would be in pantomime, mimicking, with slapstick humour, wildly exaggerated emotional posturing. We can learn gradually to alter our responses so as to become sympathetic to her approach, recognising the sincerity of her response to the text and music. But we can never belong to a culture in which that expression is natural in an artistic situation. We wouldn’t sing like that ourselves. And this is the acid test, which so far very few modern musicians have come near to passing. You can’t reproduce these styles of performance by calculation, but only by feeling those sounds work musically and translating that feeling into the motions required to make them. If you can do that you’re half way towards understanding how it felt then. But the other half of the journey would have to consist of becoming so much a part of that expressive world that all your playing or singing had that style as its default, as what came naturally to you.

¶16 Something of this magnitude has happened to baroque violinists and their colleagues specialising in period performance. The difference here is that the style they have made is a modern style, and so of course fits perfectly with current taste. ¹⁵ And so it’s not really the same thing. Perhaps performers will come to specialise in early recorded styles so exclusively that they will reach the same stage, and then we shall indeed be able to learn a huge amount from them about early 20^th-century performance by studying them. But it remains to be seen whether there will ever be the market for this sort of development, especially given that we have the recordings to listen to and methods of cleaning them up are getting better all the time.

¶17 All this said, not everything has changed. Readers will have noticed, both from the examples used here and also from their everyday experience of listening to music (and this has been confirmed in more empirical work), ¹⁶ that across the recorded century it is generally the same points in a score that get emphasised, whatever the expressive means for giving emphasis may by at any one time. Phrase-ends, the highest notes, the metrically strongest beats, harmonic cruxes, new sections, tend in all recorded performance styles to be emphasised one way or another. So clearly there are aspects our response to musical structure that remain relatively stable across centuries. How many centuries we can’t say because we’ve not had recording for long enough. Features dependent on tonality won’t go back as far as features dependent on melodic structure, for example, but there must have been continuity at the level of musical structure over long periods of time, much longer than periods in which performance style was stable. And that means that some of the things performers do performers have always done, while others differ from generation to generation, even while tending to occur in the same places in which performers did other things before. We just can’t know, save for the evidence of recording, which are which.

¶18 To take one example, but a significant one, consider the changes in portamento and vibrato during the first half of the 20^th century. We’ve seen that at the same time as portamento gradually declined vibrato speeds gradually slowed and vibrato widths deepened. It seems reasonable to suggest that some of the expressive work done by portamento was gradually transferring to vibrato. Other factors were changing too, for example speeds were slowing, so it’s not safe to assume that the transfer was exclusive: patterns of change must have been more complex. And we can’t assume either that the expressive effect on listeners was the same after the transfers as before, nor, therefore, that performers deployed vibrato in the same way they’d deployed portamento. But some exchange probably occurred, presumably because portamento seemed less appropriate in a larger communicative context (perhaps for reasons to do with the rise of modernism), ¹⁷ and vibrato more so. It’s certainly striking that the greatest increase in vibrato width comes exactly at the point (the later 1940s and early 50s) at which portamento ceases to be routine. ¹⁸ Of course vibrato doesn’t do exactly what portamento did: you can’t use it to join two notes, and it doesn’t model baby-talk; but it does similar expressive work, and that’s the level at which expressivity remains more stable.

¶19 How stable? Is modern performance overall as expressive as early recorded performance? That’s very difficult to answer impressionistically, and it would be even harder scientifically. One’s first thought is ‘no’, but it’s also possible that expressivity is simply fragmented in modern performance styles, distributed in smaller amounts over a greater number of parameters or even a greater number of places in a score: tiny changes at the microsecond level contributing cumulatively a similar effect to larger changes that once took longer and so happened less often. It certainly seems unwelcome to suggest that music is less moving now that it was 100 years ago to the extent that listening would engage one less or leave one less satisfied. Nevertheless, these are questions that need to be considered and to be investigated more thoroughly, and preferably more scientifically in the future if we want to understand more about what style change means and about how music works in performance.

¶20 What are the mechanisms by which style changes? It’s hard to avoid proposing a neo-Darwinian evolutionary analogy because selection of the environmentally fittest provides such a powerful explanation for a process of change so gradual as to be unnoticeable at a local level and yet transformative over time. ¹⁹ As we’ve seen, there are a lot of factors that may influence the nature of change in performance style, many of them from beyond music, so any proposal that hopes to be practical or comprehensible needs for now to look at just a small part of the whole picture. Let’s assume, therefore, for the sake of clarity, that performance changes mainly through influence from one player to another. It can’t be that simple, but that must constitute part of the story. How would that work? Presumably musicians develop in their first two decades (substantially less in the case of prodigies like Patti or Rubinstein) through modelling themselves on players (or singers: let’s take that as read from now on) whom they hear around them and according to templates set out by their teachers. But by the end of their training certain individual characteristics will begin, in the most able, to become audible. ²⁰ There cannot be too many of these, nor can they depart by very much from the norm—otherwise the playing style will be rejected by critics, work will dry up and they will no longer be heard and be able to influence anyone else. Throughout their lives musicians are also hearing other players and are adopting from them, no doubt often without realising, features of personal style that they find particularly effective. Habits that seem useful to many players will spread quite fast, habits that don’t will die out. How fast depends on how many other players are heard, especially during the early years. Hence hothouses of musical education, such as Berlin, will have a disproportionate effect in diversifying performance style. And when one looks at the number of musicians discussed in this book who trained there in the decades around 1900 it becomes less surprising that style changed so much in the early 20^th century.

¶21 Recording disseminated performance styles too, but what would its effect have been? The common assumption at the moment is that it would have tended to homogenise playing by spreading a norm far and wide. Yet was a norm recorded in the early years? No, on the contrary, we are surprised by the diversity of styles represented on cylinders and early 78s. If anything, early recordings probably diversified more than they reduced personal styles. But not for long: there is no question that over a longer period of time, once listening to recordings became common—and very noticeably by the second half of the century—the mingling of styles produced something very much more uniform overall. Even so—and this is the interesting point, not always given enough emphasis—it remained notably varied at the micro level and continued to change rapidly.

¶22 A genetic analogy can explain how this works. In small and isolated populations a few powerful individuals can ensure that their genes come to dominate the population as a whole. It may be very different from other populations, but there will be little change within it except by chance mutation, which is much easier at the genetic level than in society, where artificial rules tend to inhibit change over time. An isolated musical society, therefore, would tend to maintain its traditions rather strictly, producing a recognisable ‘school’ of playing.

¶23 But musical populations were never all that isolated. Musicians have always travelled in search of work. So even though a national or local tradition might be confined there’s no reason to suppose that many became moribund or decayed, whatever that would mean in musical terms—perhaps a form of mannerism so extreme that no one outside the immediate circle would find it persuasively musical. Nevertheless, exchange would have been constrained: most musicians would be local in most traditions.

¶24 Recording, however, would function almost as effectively to cause musical styles to coalesce as did increased mobility. When migration leads gene pools to mingle, two things happen: on the large scale, features of both races mix and after a few generations it becomes increasingly hard to say whether an individual belongs to one or the other; but on the small scale, because the population is genetically more varied, a much wider range of mutations occurs than would have been possible with fewer and less genetically diverse individuals. Some of these will be useful and will survive through natural selection, and over time there will be more rapid evolution than is possible in smaller groups.

¶25 What this means for musical performance—though one can’t check this reliably against evidence, since before recording there isn’t enough—is likely to be that while general features of performance style have become homogenised, there is much more variation between individuals than there was before, and there is more rapid change in performance style over time. Styles would be more similar from place to place at any one time, but they would change more quickly because, through recordings, individual mutations were disseminated so much more rapidly. What little evidence there is suggests that this is so. We know that performance style has changed hugely during the past hundred years, because we can hear it on record. We can’t hear it any earlier, but the treating of 18^th- and 19^th-century performance practice together in single studies based on the surviving teaching manuals tends to leave the impression that there was much more consistency during the 19^th and 18^th centuries. ²¹ There may, as I suggested earlier in chapter 4, be an element of wishful thinking there: it would certainly be very convenient for students of historical performance practice to be able to argue that recordings c.1900 tell us much about earlier performance; but if recordings and the democratisation of travel contributed to a pooling of performance styles then the genetic analogy suggests that they may have a point. Styles may have changed more slowly the earlier one goes back in time.

¶26 On the other hand, as I’ve already suggested, there is good reason to believe that styles were already changing quite rapidly by the time recording began. We’ve seen how few performers like Patti or Joachim seem to represent the old, simpler, pre-verismo style, and how old they were when recording began, and we’ve seen that even middle-aged performers were already using a more modern style. It seems very possible, then, that a significant shift had taken place before recordings began. This should come as some relief to proponents of all but the most rigorous HIP, because if styles were still changing slowly in the second half of the 19^th century we shall have to consider the possibility that Beethoven and Mozart sounded more like Paderewski and Patti than Levin and Kirkby, and I’m not sure that that is what historically informed performers at the moment really want to hear. We looked at Mozart played by Carl Reinecke in chapter 6 and I doubt that many hope it may tell us much about Mozart. We shall never know. And this a peculiarity of our present situation. We have had recordings for long enough to have difficulty with the past but not for long enough to know how it works out over centuries. Do performance styles change continuously, or in leaps, or do they go in cycles? Will future generations develop performance stylistic worlds in which Gerhardt’s singing again makes perfect sense? A hundred years of evidence isn’t nearly enough for us to hypothesise.

¶27 The theory of runaway sexual selection offers some insight into this. According to Geoffrey Miller, ‘Music is what happens when a smart, group-living, anthropoid ape stumbles into the evolutionary wonderland of runaway sexual selection for complex acoustic displays.’ ²² Like the proverbial peacock’s tail, the ability to sing well is taken as a sign of power—the animal has the strength to survive despite putting himself (it is almost always males that provide the display) at risk of attack from predators by stopping and singing extravagantly, and so females are successfully attracted. Runaway selection, the process by which ever more extravagant vocal displays are required to out-bid rival males, leads vocalisations to develop into ever-more sophisticated forms leading over time to human music. Miller’s theory could benefit from much more detailed substantiation, but as a hypothesis it is surprisingly fruitful. Runaway selection is rather a plausible explanation for some extremes of human composition and performance style, for instance the 14^th-century ars subtilior or 20^th-century atonality, or, indeed, the ultra-expressive singing and playing characteristic of the first thirty years of recording. Whether any of these was an effective way of attracting sexual partners is perhaps no longer the point. Runaway selection could be fired almost as powerfully by economic gain or popular acclaim or (in the case of composition) by peer-group esteem (which provides an almost complete explanation for integral serialism).

¶28 If sexual competition drove the development of musical skills in the past it’s unlikely to have gone entirely away. One of its consequences, paradoxically, is ritualisation, the tendency for strongly attractive behaviours to become fixed, and as a result of becoming fixed no longer to be noticed as exceptionally attractive. Smart animals are able to gain advantages by modifying ritualised behaviour in order to attract attention to themselves from smart, novelty-seeking mates. As Geoffrey Miller points out, drawing on Huxley, ²³ among modern humans ‘neophilia is so intense that it drives a substantial proportion of the global economy’; while among our hominid ancestors he suspects that ‘it favoured not so much diversity of sexual partners but selection of highly creative partners, capable of generating continuous behavioural novelty throughout the long years necessary to collaborate on raising children.’(346)

¶29 In the light of this tension between ritualisation and creativity we can perhaps think of musical style change (both performance and composition style) as bouncing around between extremes reached through runaway selection while trying to improve on a widely-popular mean of ritualisation. Thus style would tend towards a ritualised mean with which many felt comfortable, which might be characterised by such traits as relative faithfulness to the score, restrained rubato, and so on. Yet from time to time, as it became just too comfortable to be interesting to younger musicians, the more adventurous among them would aim to be more attractive by extending expressivity in one or other of the available dimensions, either pitch, rhythmic, or amplitude modification, or some combination of the three (as for example in neo-baroque articulation which deals in modified note-lengths and loudnesses). After which others might bounce style off in another direction, only for it to return in due course towards the mean, but no doubt a slightly different mean than before. Extremes would include such states as atonality in composition, or in performance hyperexpressivity as in the 20s or inexpressivity as in Bach playing in the 50s; and the bouncing back from an extreme happens whenever that extreme reaches such a point that its costs outweigh its benefits. For example, atonality has clearly reached that state by the mid-70s when minimalism, which wouldn’t have had a hope earlier, came to many as a timely relief. The hyperexpressivity of the 20s and 30s, as I’ve argued, ²⁴ was too costly in cultural terms once the horrors perpetrated by its German contemporaries made it seem unforgivably complacent. The inexpressivity of hardcore HIP was far too costly in terms of lost musical meaning for it to survive the introduction of new expressivity by Brüggen, Harnoncourt and others.

¶30 Another way of understanding style change, entirely compatible, is to think of elements of musical culture as viruses (with Dan Sperber) or with Richard Dawkins as memes, ²⁵ units of culture that spread among human populations. ²⁶ To make any quasi-genetic analogy work it is essential to allow for the fact that in cultural transmission nothing physical is passed from one person to another. Rather, something that one person does is imitated by another. Imitated, not copied; or copied only in the humanities sense that copies are never exact. It’s the imperfection of the copy that causes the far more rapid evolution of culture than the evolution of life. A gene may be copied exactly over tens of thousands of generations before a useful mutation occurs. But a meme—some cultural habit or concept—is remade slightly differently every time it is picked up by one human from another. There are many more potentially useful variants in circulation, and there are probably many fewer constraints on ‘usefulness’ as well. And so culture changes very fast indeed, yet through ‘units’ of habit or concept that may be unnoticeably indistinct or blindingly obvious or anywhere in-between.

¶31 Musical performance is an extremely fertile environment for the exchange and modification of habits of expression. The extent to which listeners (including performers) can apply varied meanings to sounds, and so to find newly-shaped sounds meaningful (newly-shaped as a result of some quasi-mutation in performance style), even though they may be unfamiliar, allows a great many mutations to be found useful and so to be developed and passed on in turn. The main brake on the process is the professional advantage to performers in not modifying their performance habits once those have been satisfactorily established. Performers have quite enough to do in recreating at the highest possible level of excellence, time after time, compositions that are very hard to perform well, without also opening up their performances to new influences by listening to others and adapting and incorporating their ideas. Hence the predominant tendency of performers to maintain a fairly consistent style over their lifetimes. But as we’ve seen there are exceptions, and we need to be sensitive to the possibility of influence at any moment.

¶32 So how informative individual performers are about the styles of their youth has to depend on their personal susceptibility to influence. It isn’t hard to suppose that some would be less susceptible than others to infection, their quasi-antibodies strengthened by disposition and perhaps by particularly strict training. At the opposite extreme, some performers would be infected unusually easily. We can see Rubinstein as one example of that, Arleen Auger (chapter 4) another. A more challenging example is that of Lotte Lenya. ²⁷ Lenya’s performance style changed more radically than any other performer I can think of, greatly altering the reception of Kurt Weill’s compositions in the process. But rather than taking in memes from around her and allowing them to transform her singing it seems far more likely that she became an exceptionally powerful cultural virus herself (powerful both politically and musically) infecting others with unusual virulence, so that her later style was universally adopted by other singers for performing Weill. Most players are only lightly inf(l)ected by exposure to others, however, so that symptoms are small enough to be very hard to spot, yet the accumulation of mutations across populations over a generation can still produce noticeable change in general performing style. It’s harder to recognise as it happens around one than when one can look back on it from a distance, but there are good reasons to think that style is still changing now, and in a more expressive direction, perhaps influenced in part by the easy availability of reissued early recordings. But this is only an impression, and in retrospect the situation in the early 21^st century may look quite different.