Challenging Performance: The Book. 3 Performance changes

3 Performance changes over time

Classical musicians are brought up to believe that there is broadly one proper performance of any score, the performance its composer imagined; and that it is their job to produce that; and also that that is, as far as humanly possible, what they are doing. I say ‘broadly’ because musicians are also encouraged to believe that individuality in a performer is desirable, that they should each have something unique to offer in their reading of a score. But the constraints around that freedom of interpretation are very narrowly drawn. Go far from the norm, even in quite small details of timing (rubato) or emphasis,[1] and you will be criticised.

Performers know this all too well, for they’ve been pushed back at the boundaries of those norms—by teachers, conductors, examiners, adjudicators, critics: the gatekeepers to the profession—all their lives. The critic Joan Chissell in 1968 found even Jaqueline du Pré and Daniel Barenboim—whom one might imagine were about as safe from this kind of boundary-drawing as one could hope to get—‘self-indulgent enough in rhythm and tempo to be un-Brahmsian’ in the Brahms cello sonatas.[2] This is particularly ironic given Brahms’s own preferences for far greater flexibility than this, as we shall see in a moment. Chissell’s comment suggests that what seemed Brahmsian then is not what seems Brahmsian now. And it gives a vivid sense of the way in which any performer at any point in their career can be censured for stepping even slightly out of the narrowly-defined norms imagined as proper to each composer, sometimes even to each score. We must remember, too, that all the while musicians, reinforced by the rest of the gatekeeping community, are teaching the same beliefs about what is proper—albeit with silently shifting practices—to the next generation, aiming to ensure that strict norms are passed strictly on.

I say ‘silently shifting practices’ because of the awkward fact, that’s become increasingly obvious thanks to easy availability of an abundance of recordings made over the past 120 years, that performance style is actually constantly changing.

Performers, and many listeners, are increasingly aware of that now; but not everyone has yet appreciated the huge implications of performance style change for WCM beliefs and practices. First of all, what we think is proper to a composer or a score is already slightly different from what our teachers’ generation thought. And over a century, as recordings show, these differences accumulate to such an extent that musicianship becomes in some respects unrecognisable. (There’ll be examples of this in a moment, and in Chapter 4.) Secondly, while there will certainly have been a manner of performance expected by the composer, and during their composition the notes will necessarily have been imagined with that in mind, it’s clear that these scores we feel we know and love have been performed in many other ways, ways that if you go back far enough are radically other; and each generation has found these increasingly different manners to be perfectly suited to what they hear as the essential nature of the music. The notion of a style proper to a composer and a score looks decidedly shaky when one takes this into account: it looks as if there must be many different ways in which these same scores can make convincing music.

In that case, what is all this gatekeeping for? Why are critics (not to mention teachers and all the others) damning performances as unsuited to the composer when clearly whole generations of musicians find them (or have found, or will find them) highly suitable? (We’ll consider this in depth in Parts 2 and (from psychological perspectives) 4.) Only if you believe that there are ethical reasons for playing or singing exactly what the composer expected his performers to play and sing is your choice of performance style a matter on which gatekeepers have any business trying to rule, if many other styles work equally convincingly. That ethical case for strictly historical performance is hard to make (we’ll look at it more closely in Chapters 6, 11 & 19), and if you don’t accept it—which one hardly can and still believe in the value of current performances of any but the latest scores—then it’s clear that musical meaning must be contingent on period culture.

Performance, in other words, inevitably changes in order for scores to continue make sense to new generations. Which is not to say that other generations’ styles cannot still be appreciated. The enthusiastic market for early recorded performances proves that even styles very different from our own can make excellent sense to modern listeners, with a bit of exposure. In any case—and this is the third key conclusion from performance style change—whatever you believe about the special value of original performance, it’s clear from the extent to which the character of scores has changed over time, with changes in performance style, that performers are always doing much more of the musical work, and the composer much less, than we’ve been led to believe.

To make all this more tangible let’s listen to a genuine historical performance, Mary Garden singing Debussy’s ‘L’ombre des abres’, recorded in 1904.


Debussy is playing the piano, accompanying one of his favourite singers in one of his own songs. You can’t get much more historical than that. But who in our culture of historically informed performance today is singing Debussy in anything like this manner? You couldn’t get a booking if you sang like this. So much for performing the composer’s expectations! Whatever HIP enthusiasts may claim about the performance of Baroque and Classical scores, it seems that once we reach 20th-century repertoire, for which recordings actually survive and for the first time we really know how music sounded, suddenly no one wants to hear historically informed performance any more. Why not? Why does it suddenly become unacceptable the moment we can hear it?

Where shall I begin? The voices aren’t synchronised, which is one of the most crass faults in any modern performance, chords are spread, there’s wide rubato and portamento, none of it notated, which to us would seem to rule it out as intended. But there it is. These features that we now consider unmusical were essential to ideal musicianship a century ago. People make excuses for ignoring that evidence: the performers were old, they say, or the recording wasn’t very good. But these excuses don’t stand up when one looks more closely. The performers were often, as in this Debussy case, the best and most authoritative; the recordings are quite good and numerous and consistent enough to show that the rules of musical performance 100 years ago were simply different from today. You can make the same case for the singing of Verdi and Wagner. There is ample recorded evidence. But no one wants to copy it in the modern opera house.

Here is some Mozart playing that in one sense is closer to Mozart than anything else we shall ever know.


Again, there is wide dislocation of left and right hands, with the bass notes well in advance of the melodic line. (In the YouTube video it’s possible to see them punched separately near the right-hand edge of the roll.) Chords notated in the score are arpeggiated more often than not. In the melodic runs there is inégalité quite as strong as we expect to hear in a modern performance of French Baroque music, and there is also improvised melodic decoration, which is perhaps the least shocking feature to a period-instrument performer though still far from usual in modern pianism. All this in Mozart, for whom we think a very regular, even style is most appropriate.

The pianist, Carl Reinecke, was born in 1824, the year in which Beethoven’s 9th symphony and Schubert’s Death & the Maiden Quartet were first performed. It’s hard to know which era Reinecke’s pianism represents; but he became a mature artist in the 1840s, and as most players don’t radically change the way they play to follow their younger contemporaries,[3] Reinecke’s playing probably tells us something about pianism around the middle of the 19th century. And so if you want to believe that modern fortepiano playing has accurately recreated Mozartian, Beethovenian or Schubertian playing then you have to argue that there was an undocumented revolution in pianism between the 1820s and the 1840s; a revolution that takes you from nice, neat playing to this, all in twenty years. I don’t think that’s very likely. What does this imply for the historical playing of late Classical and early Romantic composers? That it was a lot more like Reinecke that we might wish it to be.[4]

Just how radically the character of a score, and of a composer, has changed over time is shown vividly by these next examples of Brahms playing. First listen to Arturo Benedetti Michelangeli (1920–95) in Brahms’s Ballade Op. 10 No. 4 in B.


The video is particularly interesting because of the way Michelangeli prepares himself to play, raising his hands to the keyboard and then taking a further 13 seconds to put himself into the right frame of mind for a performance of this little piece that looks almost like a religious ritual, praying over the text before sounding it as deeply serious and profound. But listen now to a pianist, Ilona Eibenschütz (1872-1967) who knew Brahms, whom Brahms admired, to whom he first played one of his last sets of piano pieces, allowing her to give their first performance in London. He is reported to have said, “She is the pianist I best like to hear playing my works”.[5] This is her performance of this B-major Ballade, made much later in life (in 1952); but, given that ‘her 1962 recording of Brahms’s Waltz, Op. 39 no. 15 is remarkably like her 1903 recording of the same piece’,[6] it’s reasonable to suppose that this performance of the B Major Ballade reflects the way she played to Brahms.

It’s hard to imagine any score having a more different character from the modern ‘Brahmsian’ reading reverentially presented by Michelangeli. It’s much faster and lighter, almost salon music. That Brahms himself played with great flexibility is suggested by what can still be heard on the very damaged recording of him playing his Hungarian Dance No 1 in g minor that survives from 1889[7] and is confirmed by comments of his contemporaries.[8] Anna Scott has shown in great detail and beyond any doubt, that the modern Brahms sound characterised by control and gravitas is far away from the Brahms known to his contemporaries. And yet both seem to have been felt as entirely suited to the nature of his scores, as have intermediate approaches to Brahms in the intervening 130 years (Chissell’s, du Pré’s, and many others’). Once again we have to admit that what we think of as proper to a composer is a modern fabrication, and that other approaches have worked just as well, including, of course, the very different approach of the composer and his contemporaries. It seems implausible to argue (and very few modern Brahms performers would wish for a moment to argue) that the way Brahms was played in the 1890s is to be preferred. But nor does it make any sense to argue that it was worse, or misguided, unless we want to argue the same for the way we play our own contemporaries today. You could make that case, I suppose, and claim that it takes 100 years to find the ideal way to play a composer, or even that we get better and better all the time, in which case roll on 2100’s view of Brahms. But I’m not sure that it’s very helpful.

In any case, the enthusiastic CD and download market for early recordings has shown that, with exposure to the performances of earlier times, we can learn to love them, understanding their rhetoric, their habits of style and expressivity, and broadening our own sense of how scores can be made beautiful and meaningful. And that should by now be starting to spread respect for late-19th/early 20th-century performers and for the very different approaches to expressivity that they and their audiences loved. (There are many more examples and more discussion in my online book, The Changing Sound of Music.)

A couple of modern pianists have made thorough attempts to copy early recorded styles, Sigurd Slåttebrekk in Grieg and Anna Scott in Brahms.[8] Rather more have borrowed from early recorded styles without thoroughly changing their own. Copying exactly—which you might think would be essential for anyone who actually believes in Historically Informed Performance (HIP)—is hugely demanding: it involves giving up important aspects of one’s own musicianship in order to embody another’s. For it does have to be embodied, learned by one’s own body so that it feels natural as one plays, in order to produce convincing performances. It’s an act of self-denial and generosity, but one that allows us to hear performances of scores, as if by their contemporaries, that were never recorded at the time (both Slåttebrekk and Scott offer examples, and it’s wonderful to have them.) Again it teaches us that it has been and still is possible to make persuasive music with radically different sets of assumptions about what is musical. And that therefore—and this is the point that really matters for us—there must be many other sets of assumptions, radically different again, that we could use to make persuasive performances of these same scores today.

What you cannot do (though many teachers try), is to argue on the one hand that early recorded performances are worse than modern performances, and argue at the same time that being faithful to composers’ intentions produces the best performances. You can’t have it both ways. Either Debussy knew best how his music should sound, and we should be copying it, or he didn’t, and we shouldn’t. It’s obvious that Debussy expected his scores to sound a certain way – something like we heard earlier, shaped by the performance style around him. But we also know, from 100 years of performance since then, that those same scores can work very well performed very differently. And this is obviously true for all music. It follows that Debussy’s intentions or expectations would not necessarily produce better performances of his scores than any others. So it’s not a question of which performance style you use, set in which historical period: what matters is how well you use that style – any performance style – to make music with a score. Each generation finds new ways of being musical, and there’s nothing inherently better or worse about any of their solutions. In every generation superlative performers make superlative music with these same scores.

But, of course, as taste in performance and therefore performance style changes, so the character of the notes changes, and thus the meaning of the scores changes, and our sense of what the composers are like changes with them. That’s the nature of classical music. It changes over time. It’s not like a painting or sculpture, that looks the same but is understood differently. Music sounds different too, because it’s made in performance, and only in performance.

With all this in mind, and taken seriously, there are simply no grounds, historical or ethical or musical, for disapproving of this:

Monteverdi, ‘Zefiro torna’, Paul Derenne and Hugues Cuénod (tenors), Nadia Boulanger (piano), unnamed (cello). Recorded, Paris 11th February 1937. HMV DB 5039


The style of singing is unlike anything that a contemporary early music group would tolerate for a second. And then there’s the piano accompaniment, in Monteverdi! And yet, it’s very moving; or so I find it, and it’s been continuously admired, and available in reissues, for over eighty years. A great performance of a score produces deeply rewarding experiences in unprejudiced listeners, regardless of when or how it is made.

So I think it’s impossible to continue to delude oneself about what is right or wrong, or to allow a priori beliefs about right and wrong to inflect one’s responses to performance, once one has thought carefully about the implications of early recordings.

Here is another example to which, I suggest, there are no good grounds to object. (Just listen to the miraculous triple layering of the sound, and the balancing act that’s involved in keeping it steady.)

Bach/Siloti, Prelude in b minor (derived from Bach Prelude in e minor, WTC I, BWV 855a). Emil Gilels (piano). Recorded as part of a concert in Moscow, 1978.


The temptation with Gilels’ performance of Siloti’s rescoring of Bach’s notes (I’m choosing my words carefully here) is to call it an arrangement, and thereby try to put it in a safe place that doesn’t threaten beliefs about right and wrong. But you’ll recall that there are two different scorings of this piece by Bach. Are we supposed to believe that either of those is original, whereas any other is an arrangement? And that each of those by Bach will somehow inevitably produce a more powerful listener experience than any other? I think it’s very hard to argue that persuasively in the face of a performance like this. Was there ever a more riveting and moving performance of a score based around those notes?

Musicians don’t spend twenty years training to do history; they make music, the most powerfully expressive music that they can. That’s what music is for. To generate deeply engaging listener experiences. And, as we’ve seen, that can be done in an unknowable number of quite different ways, most of which we have not yet heard.

It’s also becoming clearer from these examples how much of the creative work that’s involved in making music is done by the performer, not the composer. We’ll look at this in more depth in Chapter 5, but whatever this music is depends on how it sounds; how it sounds has changed over time, because it has so much more to do with the performance, and so much less is inherent in the notes and the relationships between them, than music theory and music teaching have been claiming since at least the rise of music analysis in the late 19th century.

It’s tempting to suppose that a great composition is one whose score can give rise to innumerable radically different and deeply persuasive performances.[9] And in that case (in any case) musicians should be allowed to be much more interested than they are in making radically different but deeply persuasive performances from the scores we admire. Continuing to make basically the same performance over and over again, which is what performers are trained and paid to do now, tells us nothing more about the piece. We have enough typical performances now. More than enough. It’s time to explore, to ask what else these scores can do.

I think we’ve heard enough, and I’ve said enough, to show that the way we think about what is proper in musical performance, and the way we train musicians to provide it, is seriously at odds with the evidence and with reason. It is simply wrong to train students to believe that there is a broadly correct way to perform these scores, or a best approach to being musical. In fact about the only thing that’s true about current ideology in classical music training is that you’d better play the recommended way if you want work. And that itself shows the measure of control and coercion that infects the music business. That if you play or sing differently you won’t get work is the sign of a ruthlessly policed state.

That being the case, it’s time to take a fresh and sceptical look at a wider selection of the many delusions that help to maintain the belief-system, or ideology, that underpins our current teaching, practice and gatekeeping of WCM.

Continue to Chapter 4: ‘Identity’

Back to Contents menu


Further reading

Da Costa, Neal Peres. 2012. Off the Record: Performing Practices in Romantic Piano Playing (New York: Oxford University Press).

Day, Timothy. 2000. A Century of Recorded Music: Listening to Musical History (New Haven: Yale University Press).

Hamilton, Kenneth. 2008. After the Golden Age: Romantic Pianism and Modern Performance (New York: Oxford University Press).

Kennaway, George. 2014. Playing the Cello, 1780-1930 (Farnham: Ashgate).

Leech-Wilkinson, Daniel. 2009a. The Changing Sound of Music: approaches to the study of recorded musical performances (London: CHARM).

Philip, Robert. 1992. Early Recordings and Musical Style: Changing Tastes in Instrumental Performance, 1900–1950 (Cambridge University Press).


[1] ‘…any good young instrumentalist knows how each piece is expected to be played, right down to bowings, dynamic marks, and places to breathe.’ Haynes, Bruce. 2007. The End of Early Music: A Period Performer’s History of Music for the Twenty­-First Century (New York: Oxford University Press), 6.

[2] The Gramophone, December 1968, 854.

[3] The case for this is made in Leech-Wilkinson, Daniel. 2009b. Recordings and Histories of Performance Style. In ed. Nicholas Cook et al., The Cambridge Companion to Recorded Music (Cambridge University Press), 246-62.

[4] For an excellent and much more detailed discussion of another Reinecke recording, and of its implications now, see Peres Da Costa, Neal. 2019. Carl Reinecke’s Performance of his Arrangement of the Second Movement from Mozart’s Piano Concerto k. 488. Some Thoughts on Style and the Hidden Messages in Musical Notation. In ed. Thomas Gartmann & Daniel Allenbach, Rund um Beethoven: Interpretationsforschung haute (Schliengen: Argus), 114–49.

[5] Allan Evans at

[6] Leech-Wilkinson, Daniel. 2009a. The Changing Sound of Music: approaches to the study of recorded musical performances (London: CHARM), chapter 6, paragraph 18.

[7] Musgrave, Michael. 2003. Early Trends in the Performance of Brahms’s Piano Music. In ed. Michael Musgrave and Bernard D. Sherman, Performing Brahms: Early Evidence of Performance Style (Cambridge University Press), 302-26.

[8] Scott, Anna. 2014. Romanticizing Brahms: Early Recordings and the Reconstruction of Brahmsian Identity (PhD thesis, University of Leiden), 135–43.

[9] So tempting that I said it myself in an earlier draft! On reflection I don’t see the evidence yet: it’s a nice idea, but a way of testing it needs to be worked out.

Be the first to comment

Leave a Reply