Scholarly article on topic 'Cortot's Berceuse'

Cortot's Berceuse Academic research paper on "Art (arts, history of arts, performing arts, music)"

Share paper
Academic journal
Music Analysis

Academic research paper on topic "Cortot's Berceuse"

DOI: 10.1111/musa.12054


An increasing quantity of music-analytical work in recent years has drawn on recorded performances, including (perhaps especially) recordings from the past. It is now widely agreed that the things performers do with notes can influence more than simply the local relationships listeners perceive among them, and to that extent they offer an analytical perspective on a composition.1 In some recent studies a stronger claim has been made,2 which it will be helpful to restate before we go on to look at an example. This claim has three parts. The first is that most music analysis cannot be done without implicitly calling on performance.3 There may be exceptions, for instance certain kinds of statistical or symbolic analysis in which the sounding result is not considered; but any kind of analysis that speaks of relationships between notes, or of the meanings or effects of elements within compositions - that deals, in other words, with music as it is or might be experienced - depends on assumptions about how notes sound. And those assumptions can only be based upon the analyst's experience of performances.

The second part of the claim draws on the evidence of the history of recorded performance. Now that we have over a century of recordings, we can hear that the way performers sound notes changes greatly over time. What is changing is the way performers are expressive, the ways in which they do not sound a score literally but adjust timings, loudness, pitches and timbres in the process of joining notes together into performances that make music. A particular set of such habits of adjustment defines the performance style current at a particular place and time; as they shift, so does performance style as a whole. Performance style provides the context in and through which music seems to have meaning or to create a sense of communication:4 it constitutes a language of musicianship. As the language changes, notes in scores acquire different significance and scores as a whole come to sound different, they acquire different flavours, sometimes even different characters, and in the process different associations, implications and meanings.5 The compositions we conceptualise as lying behind current soundings of the scores (and often imagine as musical works) therefore seem also to change.6

Thirdly (following from the previous two), any 'reading' of a score - reading it through, imagining it, thinking about how notes function in relation to one another, for example - involves imagining sounds it might make, and that cannot avoid being subject to the performance style assumed by the reader. As analysts, then, what we think about pieces of music depends on when we think it. This is obvious enough in broader cultural terms: after all, the style of writing about music changes. But we can now see that that is not just a matter of local literary

Music Analysis, 34/iii (2015) 335

©2015 The Author.

Music Analysis © 2015 John Wiley & Sons Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

and social communicative style; it is not just that listeners heard the same thing we do but wrote about it differently. It arises also from the fact that music sounded and meant differently. Following the same logic, it seems probable that when we listen to early recordings, or imagine performances drawing on their performance styles, we hear something different from what contemporaries heard. So the picture is greatly complicated by the fact that performance styles change so much. But however complex the picture, there is no escaping the imagined hearing of scores within some performance style. Typically, that will be the performance style within which one matured as a musical performer and thinker,7 although it may well adjust as time passes, either as general performance style changes around one or as one absorbs and develops a taste for performance styles from the past. But whatever its characteristics, the style on which our musical thoughts rest will determine their nature.

So, analysis depends on performance as it is practised at particular places and times. We need, therefore, to free ourselves from the delusion that we are telling truths about compositions. We are offering views of them that rest on the way they are or have been played. Performance affords analysis.

This conclusion is perhaps not the best way to win the hearts of readers of Music Analysis, but what is the alternative? How might one's auralisation of a score bypass one's experience of sounding music? Jonathan Dunsby's proposal, of which Joel Lester reminds us, that ' [m]ore often than not what the analyst is working on is his or her own "performance" in his or her head, and more often than not this performance will be second-, third- or worse-rate' will scarcely be more welcome as an escape clause. I think we should assume more of a skilled music analyst than this.8 She has, after all, throughout her professional life been intimately engaged with 'listening', whether to performances imagined or real. Analysis in turn seeks to enhance listening,9 to deepen understanding, to intensify appreciation. It hardly lies beyond listening. The more accurate one's aural imagination, the more fully rooted one's sense of the music must be in a fully specified manner of performance.10 But at the very least, the analyst's sense of how notes matter will, I suggest, have been significantly mediated - indeed mediatised - by the ways of thinking about, playing and singing music with which she is most comfortable. As analysts we need to be much more aware of this and explicit about it.

One may object that 'granted, the way people imagine and play scores changes, but the structural fundamentals - which are the important notes, where the phrases go - remain agreed, they are just differently marked.' This remains to be seen. We have not had recordings for long enough or been able to free ourselves from our own period styles thoroughly enough yet to know how many of our beliefs about musical form and musical meaning are more or less permanently true. Certainly, performers often mark evident structural boundaries with changes of tempo, but if you go back far enough - as Cook has recently shown with d'Albert (and indeed Schenker), and Slattebrekk and Harrison have shown just as surprisingly with Grieg (that most four-square of composers) - in the more

distant past performers (and in Grieg's case the composer) covered over phrase boundaries to make them as inconspicuous as possible.11 In this respect their attitude to structural fundamentals was quite opposite to ours. The very oldest players on record, therefore, really should cause us to doubt that we can rely on much about the relationship between form and performance being self-evident.

Cortot's 1920 Berceuse

When we come to a performance that does something very unusual, it may be worth stopping to see what can be learned from it about the potential of the composition, which may be greater than we realised. This is the case, I think, with Alfred Cortot's 1920 recording of Chopin's Berceuse, Op. 57, which is the focus of this article.12 The opening, transferred from an original 78 rpm recording by Ward Marston, is available in Media File 1 ( This opening has some rather interesting properties. At the bar level the timing is quite regular - unusually for Cortot, who was famously flexible. This is easy to see in a video of a Sonic Visualiser analysis (Media File 2, Sonic Visualiser is software that annotates sound files,14 in this case with a curve representing Cortot's rubato at the bar level, where a rising curve maps Cortot's slowing down, a falling one his speeding up. Cortot was meticulous in collecting and considering evidence for Chopin's practice, and so in playing these opening bars with (by his standards) relatively little metrical rubato he may have had in mind Kleczynski's suggestion of the Berceuse as a good example of Chopin's reported notion of a strict left hand supporting a free right.15 In fact, as soon as the variantes (Chopin's preferred term for the variations, and his original title for the piece)16 become complicated, Cortot's rubato gear engages, after which the variantes with most notes are played much faster, amplifying the composed activity.

But already at the start there is plenty of rubato, just at smaller levels than the bar, and to that extent very much in line with Kleczynski's suggestion. Media File 3 ( provides a video of the half-bar dotted-crotchet rubato, indicated by the purple curve, against the red bar lines and blue variation divisions (bars 3-9). The music expands during the second half of the bar as Cortot slows, and contracts during the following half bar as he speeds up again, inhaling and exhaling; and because the rubato is so strong it seems to be breathing deeply, at the level of the bar, in a strikingly human way.17 We shall return to the humanity of Cortot's performance later. The pattern can also go the other way when melodic activity increases in the second half of the bar, as for example in bars 22-24. There is little sign that Cortot shares Schenker's intuition that the last three beats of each variante bring with them a sense of speeding up. Schenker's unpublished analysis has been considered in depth by Antonio Cascelli.18 Here, at the start, Schenker's middleground neighbour-note

motion 3-4-3 is shown as speeding up towards the G 4 on the sixth beat of the opening variantes, holding back on it and then speeding again back towards the next 3 on F.19 Sometimes Cortot accelerates here, sometimes he slows. Nonetheless, in variantes 4-14 Cortot uses his strongest ritardandi to mark the variation divisions, mostly (except in 7, 11 and 12, all of which are preceded by rapid triplet figuration in need of resolution) by lengthening the previous quaver beat (Media File 4, And in variantes 4-10 'breathing', achieved by speeding up and growing louder, then slowing down and becoming quieter, is deployed very clearly at the variante level (Media File 5, where the green curve maps changing loudness; However unusual his rubato may seem locally, then, he is clearly deploying it at a formal level as well, and with greater consistency than his local rubato might lead one to expect. So we can both hear and see that Cortot is highly responsive to changing compositional surface and also to some extent structure. Whose structure is a question we can return to later.

But that is not the most surprising feature to observe here. Media File 6 ( shows an orange curve that maps quaver-beat rubato and a white one that follows the melodic line. And one can see immediately that Cortot is responding to the rising and falling of the melody by slowing down and speeding up the quavers. Cortot slows down whenever the melodic line rises and speeds up when it falls. Not only does this mutually empower the rubato and contour by reinforcing one with the other; more tellingly, it leads us to experience the contour of the melody through our embodied knowledge of what it is like to move up and downhill in a landscape - a classic example of Mark Johnson's musical landscape and music as moving force schemata.20 Cortot induces in us the sense that it takes more effort for the line to move upwards, and less for it to come down. The cycling of tension and release engages our bodies as listeners through our experience of moving in the world. Cortot achieves this rubato, moreover, by slowing for the upward leaps and waiting on the note before a leap up, generating a sense of gathering energy before reaching up. The whole effect is extraordinarily lifelike.21

Does this pattern continue over the whole performance? Not so consistently, but to a surprising extent, given the more complex demands of the composition as it develops. We can see this by following the same graphic analysis over the next 20 bars or so (Media File 7, As we saw in Media File 5 at the variante level,22 Cortot consistently gets louder while accelerating and quieter while slowing, one his favourite habits as a musician. The whole performance, of course, is relatively quiet, so the effect is less marked than usual for Cortot, but one has to take it together with the other respects in which the performance seems to breathe, to expand and contract from bar to bar and section to section.

Of all these observations the most surprising, surely, is that the rubato and Sonic visualiser melody graphs (the white curves in Media Files 6 and 7) match. So let us ask whether this is (1) unique to this recording, (2) unique to Cortot or (3) characteristic of performances of this score. To test this I have used Sonic visualiser to map the melody and rubato in sixteen recordings dating from 1918 to 1998, including three by Cortot: 1920, 1923 and 1926. To keep the example manageable, I consider only the opening statement of the theme.

As is easily seen in Fig. 1, Cortot's recordings are indeed quite similar to one another, but even so neither the 1923 nor the 1926 recording is as consistent in matching rising melody to lengthening beat as was the 1920 one: comparing them, we can see the match drifting apart. We can confirm this by correlating the melody and quaver-beat lengths across all sixteen recordings (Table 1). Cortot's match does not reach the level of statistical significance, yet is quite obvious to a listener once observed. Those matches that do (vasary and especially Pires) display the opposite correlation: they get faster as the melody rises and slower as it descends. We shall return to the variety of approaches to this score in a moment. Cortot in 1920 has the closest match: the nearest behind him are pianists of more or less the same generation (Cortot, b. 1877; Ignacy Paderewski, b. 1860, and Ignaz Friedman, b. 1882, both Poles; and Moriz Rosenthal, b. 1862), save for Howard Shelley (b. 1950) whose performance of the opening of the piece is very like Cortot's of 1923 (Fig. 2).

At the other end of the scale the pianists are mostly at least 50 years younger than Cortot, but they include Cortot from 1926, emphasising how much and how quickly his approach changed. And indeed his late recording (1940) uses rubato rather differently, as one might expect given the fourteen-year gap. Another study has revealed a similar pattern of change in his recordings of the Chopin preludes.23 And this is perhaps what one would expect to find in any imaginative performer: readings of scores changing as the years pass, very probably without the performer's being aware of it.24 But this is also entirely consistent with the gradual change in general performance style which recordings document over the past 100 years.

We can begin to answer our three questions, then, by suggesting that the matching of melody and rubato in this score may have been somewhat characteristic of pianists born around 1860-80 - we have no earlier evidence - and possibly therefore an aspect of a performance tradition for this score. We can say with more confidence that for Cortot in his earliest (1920) recording, this matching is stronger than for others and that it becomes less a feature of his performances as time goes on. How he played the Berceuse before 1920, of course, we cannot know: the match between melody and rubato may have been stronger still. Certainly it was not characteristic of Cortot, despite the mystique surrounding the improvisatory feel of his playing, to improvise his performances. A study of his multiple recordings of the Chopin preludes shows that Cortot learned every tiny detail of his apparently spontaneous expressivity

Table 1 Correlations of melody and rubato in recordings of Chopin's Berceuse, bars 3-12 (Spearman's test)

Cortot 1920 Paderewski 1922 Friedman 1928 Rosenthal 1930

.212 .162 .161 .016

Shelley 1991 Cortot 1923 Backhaus 1927 Ashkenazy 1975

-.028 -.038 -.042 -.088

Demidenko 1992 Joyce 1940 Kissin 1998 Cortot 1926

-.093 -.101 -.144 -.149

Perlemuter 1960 Hofmann 1918 Vasary 1965 Pires 1998

-.158 -.181 -.262* .445**

and reproduced it exactly, with gradual change only becoming measurable after

a year or more.25

Our performance data also allow us to ask how consistent performances of Chopin's Berceuse are in general. This may seem a curious question, because one thing that's very obvious about the recorded performance tradition is that the pastoral, consoling character of the piece seems unchanging. Given the nature of the score, one might think it impossible for this to change, though that remains to be seen (we shall return to this at the end). But a close examination of pianists' rubato across this same initial thematic material presents a surprising picture. The chaotic appearance of the graph shown in Fig. 3 illustrates well just how different these performances are in their idea of how to relate the melodic and harmonic content to the metre: the variety of approaches seems far to exceed that normally found in multiple recordings of a classical score. We can test this by taking a similar stretch of initial thematic material from another piece by Chopin and comparing the variety of approaches taken by a similar selection of pianists. The Mazurka Op. 17 No. 4 seems a suitable candidate, both because detailed timing data for multiple recordings of the mazurkas exist,26 and because mazurkas are also subject to considerable variation in the handling of accent, so it seems a fairer comparison from among Chopin's triple-time scores than, for example, a nocturne, despite the somewhat greater difference in character between the Berceuse and a mazurka. Fig. 4a shows the distribution of correlations between sixteen performances of the thematic material of the Op. 17 No. 4 - by pianists offering a similar range of birth and recording dates, and again including three recordings by the same pianist (Rubinstein) - which can be compared with Fig. 4b showing the distribution for these sixteen Berceuse recordings. The difference is striking. The Mazurka shows a normal distribution clustered around the centre, while the Berceuse is far from normal. This indicates that the range of approaches to this score is exceptionally wide.

So for the initial thematic material, at least, the Berceuse seems to suggest no particular pattern of rubato - or, to put it another way, it is amenable to a great many different patterns. This is a rather interesting characteristic for a score to have. We have not had recordings for very long, so, as observed earlier,

¡2. o

O* ' > J

£ü tr >

8« Cfl o 3

Fig. 2 Chopin, Berceuse, bars 3-12: quaver-beat lengths in recordings of Cortot (1923) and Shelley (1991)

A A A f \ A K N Is \ r* 7

\ J] < / V J -7 / ■V I 1 v. 9 V k \ \ A A \ V ft y 7 r V

i V- y - V --

- - - -- N -

-Shelley -Melody

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61

i tfl o X

Fig. 4a Chopin, Mazurka, Op. 17 No. 4, bars 5-362: distribution of correlations among sixteen recorded performances

we do not know if this is going to be true indefinitely; but on the whole (and excepting the very oldest players on record, who as we have seen did things differently) performers of classical scores largely agree, over fairly long periods of time, which are the longer and which the shorter beats, even though the extent of lengthening or shortening beats changes greatly in general performance style over many decades (typically increasing from the very earliest recordings up to about 1930, decreasing very rapidly around 1950, then gradually increasing once again from about 1990). But that is certainly not the case with this score: there are very few patterns here.

It almost seems necessary to ask at this point: What is Chopin's Berceuse? What is the unchanging core, and where does the rest come from? To be strict about it one might need to say that only the pitches, their intervallic relationships, and their metrical positions remain stable; and if performers change notes, which they not infrequently did before modern times, then even those may come partly into the domain of performance style.27 I think we have to question whether that is enough of a core for us to think sensibly about the music made with one of his scores as substantially Chopin's. He has provided a starting point, but most of the musical work is done later. This may be true of most music composed

Fig. 4b Chopin, Berceuse, bars 3-12: distribution of correlations among sixteen recorded performances

for performers:28 it is just that this piece shows it particularly clearly. There is much less work being done by the score and much more by the performer than is implied by the way we habitually talk about scores. Of course this is consonant with much recent rethinking of music's ontology,29 and in that sense I am pushing at an open door, but perhaps that door has not opened far enough yet.

Having said that, it remains true that the Berceuse does have a couple of unusual properties, and they may be contributing to the exceptional lack of agreement among performers as to what kind of musical creature it should become. First, unlike most triple-metre pieces, it tolerates lengthening on most beats, regardless of metrical position. Readers may wish to play or sing through the first section, lengthening different beats at each pass: while the third is hardest to lengthen consistently in a persuasive performance, even that is not impossible. Lengthening the first or second throughout is relatively easy to make convincingly musical. Add to that the possibilities, typically exploited, for applying longer beats irregularly from bar to bar, according to one's personal sense of the melodic and harmonic weight, and the possibilities for rubato are just as wide as we see in these performances.

Secondly, Jeffrey Kallberg has likened this piece to a music box.30 Kallberg argues that Chopin was interested in musical machines (discussed in his letters soon after the appearance of the Berceuse) and that the score has characteristics of a music box. He points also to the mechanical layout of Chopin's sketch, which uncharacteristically exposes the working parts and which numbers (and renumbers as he reorders) the constituent variantes. The end of the sketch also has sets of numbers which Chopin has heavily covered up but which Kallberg goes some way to decipher. They seem to have to do with numbers of bars. 'Might Chopin', asks Kallberg, 'have been calculating, in the manner of an engineer, aspects of the configuration of the piece?' And Kallberg goes on to describe Diederich Nicolaus Winkel's Componium:

The design of this machine allowed it to improvise on a theme in an apparently (if not mathematically) random way. Not only did its complex mechanism engender vast quantities of melodic variations, the Componium also varied the theme timbrally as well. The constantly shifting perspective of the variants against the mechanically recurring bass seems to foreshadow Chopin's achievement in the Berceuse.31

Certainly the score has some interestingly mechanical characteristics, leaving it a neutrality that offers many possible interpretations. Along with Kallberg's points we may note that its theme is roughly symmetrical, filling out the pitch space around the initial F. This is especially clear in MIDI notation (Fig. 5). And MIDI also gives us a vivid sense of the music box-like character of some of this score (Media File 8 []). Fig. 7 - which represents the whole melodic line in MIDI notation - shows that in fact there are many quasi-symmetries on various levels: more symmetry in this final form of the score, incidentally, than in Chopin's original ordering of the variations in the sketch. Kallberg's intuition that the piece is a mechanism seems the more justified the closer one looks.

At the centre is a W-shaped melodic line spread over three variantes (8-10). On either side are three variantes (5-7 and 11-13) with rapid ascents and descents variously decorated. Before these are the four opening, circling variantes (1-4); and after, the closing sixteen bars, balancing the opening four variantes in length, in which the line winds down. Here the ostinato pattern is broken and extended, while the melody, via the crucial seventh scale degree on O at the start of variante 14, leads the piece back towards its close (variante 14 to the end).

This was not quite the original form of the piece. The surviving sketch, which incidentally Cortot owned and issued in facsimile with commentary in 1932,32 shows that the order of variantes in the first state of the composition was 1-7, 11, 8, 9, 12-end.33 Variante 10 was added, and the rest renumbered, in the sketch. The autograph still began at (final) bar 3, with the introductory bars 1-2 (introducing the ostinato) added only at the last moment in the Breitkopf engraver's copy.34 From the defaced calculations at the bottom of the sketch, Kallberg 'can see "18" (or "8"), "4", "6", and "3" amid the thatches of ink[;]

© 2015 The Author. Music Analysis, 34/iii (2015)

Music Analysis © 2015 John Wiley & Sons Ltd

Table 2 Duration of variantes in Cortot's 1920 recording of the Berceuse

Variante Duration Shorter/longer

1 15.3

2 15.8

3 15.8

4 15.3

5 11.9 S

6 14.1 L

7 14.7 L

8 12.4 S

9 14.9 L Central

10 11.5 S

11 20.0 L

12 17.4 L

13 13.2 S

14 16.8

15 13.8

16 17.1

17 19.4

but some of the numbers remain fully visible, if not fully decipherable: an "18" (or "84", "18c", "/8c" or "/8g"?) near the right edge of the leaf, and - separate from the others - a "72" after the concluding bar of the piece.' The final number, unconcealed, is 84, which in the completed (revised) sketch is the number of beats preceding the golden-section point.35 In this state of the composition it falls on the first note of bar 45, immediately following the long trills over which the melody line lingers for an unusually long time. However, the later addition of bars 1-2 and the shortening of the final chord show that if Chopin calculated the form of the piece, he was willing to abandon his scheme for musical improvements. Cortot's rubato patterns at the variante level, though emphatically not their extents, are roughly symmetrical around the centre (Table 2).36 It is the extent of rubato, however, that the listener perceives, and the management of that extent seems to be at least somewhat under Cortot's control. His rubato climax, which comes at the trills of bars 43-44, just after the melodic climax of the score, is prepared from the beginning of variante 10 (bar 39) and then winds down right through to variante 13 (bar 51, or perhaps - for he is still accelerating on average - until bar 53 or so). This can easily be seen in Media File 9 (, where the quaver- (orange), beat- (purple) and bar- (red) level rubato curves are all shown. In fact, looking at Cortot's rubato at the bar and variante levels suggests that he had considerably less trouble identifying a middleground structure for this score than did Schenker, who, as Cascelli shows, struggled through variantes 7-13 to find plausible prolongations of his motivically crucial 3-4-3 figure.37 If we look at the whole performance (Media File 10 []) we can see a slight acceleration towards the centre and a slight deceleration away

from it, but this hardly counts as a shaped background structure: it is much more likely to be an artefact of a growing sense, past the halfway point, that the end is increasingly nigh. While for Schenker the key to the piece is tracing a 3-4-3 prolongation of 3 from bar 3 until the very last gesture in bars 69-70, for Cortot what matters is to pace the variantes, and on other levels the crotchet and quaver beats within them, so that his performance makes sense in relation to the foreground material both locally and over only somewhat longer stretches of the score, at the level of the few bars that a listener can easily sense as a unit of shaped time. Both analysts, one might say, are concerned with the relationship between the local and the global, but Cortot shows no signs, until the piece starts to wind down towards its close, of any focus on underlying melodic structure. For him the ecological significance (for the listener) of the local earns it his most careful attention. Schenker studies the present moment in order to understand the whole, which, once clarified, leaves the present moment as of no more than passing significance, a vehicle in which to reach a destination. But for Cortot it is the relative weight of each note to its immediate neighbours that really matters. The global need be no more than a vague sense that all is well once we reach the end, the sense of closure and catharsis that classical music always seeks to provide. No wonder, then, that Cortot hardly ever agrees with the arrow markings for accel. and rit. in Schenker's own copies.38

Cortot becomes noticeably more Schenkerian, however, in his reading of the closing page, from variante 14 onwards. Media File 11 [ watch?v=S8sac8D87PU] adds to the previous (green) loudness curve - which was smoothed to give a general trend at the bar level- a second loudness curve (also green) to show detail from attack to attack. Both indicate, of course, the loudness of the recording, not necessarily Cortot's performance in the studio. The 1920 disc was recorded using the acoustic process with one or more horns directly driving the cutting stylus, and in a single take with no editing. No means existed for the levels to be adjusted during the recordings, but on the other hand, recording horns tended to favour (unpredictably) some frequencies over others. So we have to be careful not to depend too much on the loudness of any one pitch. Furthermore, it is important in using Media File 11 to distinguish critically between those notes that look loudest and those that sound loudest. Judging by looks, in other words in absolute acoustic terms, Cortot is making a point about 5 At, which almost every time from bar 56 to the end is consistently louder than its surroundings, and so might be an artefact of the recording horn.39 On the other hand, Cortot's Polish contemporary Josef Hofmann notoriously turned fifteen left-hand Ats in the eight bars before the final cadence into a loud tolling bell in his 1937 recording (to the audible displeasure of his live audience),40 and it is remotely possible that some kind of tradition lay behind both pianists' interest in Ats here. But when one simply listens, it is rather the melodic descent towards 1 that seems stronger. No doubt this is an impression one brings as an analytically minded listener, but I suggest that Cortot is using this aural illusion rather carefully.

Cortot makes less than do some other pianists of the large descent from O in bar 55 to Bt in bar 59, despite Chopin's having drawn attention to it in the sketch through his notation of the fingering.41 Cortot's interest seems rather to be in the descents from Bl> beginning in bar 61. There the Bl> and At are acoustically louder, drawing attention to the start of a middleground descent at four-semiquaver intervals - Bl^A^Gt-F-El^Dl» - taking us to the end of bar 62. After the first two notes the rest are not emphasised, but they do not need to be in order to seem prominent. After this initial descent (6-5-4-3-2-1) over the new dominant-implying ostinato (in play since the prominent Bt in bar 59), Cortot treats bars 63-64's F-Et-Dt as a 3-2-1 descent that fades out. He draws more attention to the new 6-5-4-3-2-1 from bar 64, now in the lower register and with the tonic ostinato restored. For this he uses loudness and ritardando at the last quaver of bar 64 to catch the listener's attention, and ritardando through the 6-5 and 3-2 pairs to hold it. As a result, the final structural 1 seems, to this listener (as to Paderewski),42 as much that in bar 67 as that over the final chord of the piece, which belongs almost to an afterthought. In his study score Cortot notes, 'These two last chords exhaled, sighed, more than played.' And he separates them with a comma, which he asks the student to observe only sparingly ('Menager la legere interruption suggeree par la virgule').43 By contrast, Schenker's reading takes the Fs in bar 68 - where, as Cascelli argues, for Schenker the accompaniment and melody merge - as providing the Urlinie's 3,44 linked locally to its repeated scalar descents through from its habitual register at the starts of variantes 14 and [15] (bars 55 and 59) and ultimately back at bar 3. Cortot's concern is much more local, to realise convincingly the more meandering winding down composed into the closing sections' descents; and to achieve that, the repeated 6-5-4-3-2-1 figures are far more pertinent. The Fs in bar 68 are no more than an echo of a note once important but already fading in the memory.

Cortot is giving us here a view of the winding down of Chopin's mechanism, which may not be 'properly' Schenkerian but which has just as much logic and, from a listener's point of view, considerably more structural and aesthetic pertinence, not to mention its symbolic strength in relation to a view of Chopin's composition as a mechanism that winds down. In performing his reading in sound, shaping time, Cortot specifies a more concrete melodic structure for the piece than the notes unperformed imply. One could counterclaim that Schenkerian analysis, or indeed another approach, argues for a somewhat different structure as correct. But that is to say nothing more than that an analyst, hearing through Schenkerian principles, performs an analysis guided by those principles: the score is heard through that analysis-performance. Only the claim that Schenker is right offers a tactic for choosing one over the other. That claim, in turn, calls on repeated intuitions about the rightness of Schenker's reading of the relationships between notes, formed through the analyst's experience of hearing those relationships in imagined (or actual) sound. It is always open to another performer to offer different relationships and make them more persuasive. And that is what Cortot does.

Like all such attempts at structural analysis, however, the preceding view of Cortot's closing page fails to achieve significant ecological value. In relation to our experience of listening to the piece, it is overspecified. We do not, except when analysing, hear music in this way; Wollner, Ginsborg and Williamon find that even music researchers do not.45 And since such analysis claims, at least by implication, to reveal something about how we hear music ('structural hearing', in Salzer's hopeful phrase), analysis using real performances, keeping the sound and experience of music in the forefront of our minds, has the dangerous potential to reveal the incompatibility of theory with practice.46 Cortot is arguably doing more than most to specify a melodic structure. But here it is hard to claim honestly that the experience of his performance, or anyone else's, is the experience of an analysis in sound. Analysis is better understood, as Cook long ago proposed, as a parallel activity, a way of imagining music,47 at best a performance of the score in another domain. Faced with real music, in time, analysis is not what most of us, players or listeners, do.

Looking at the relationship between structure and experience from this more realistic angle, it becomes clear - if we accept the findings of analysis concerning the multilevel symmetries and interrelationships within scores - that compositions in the Western classical tradition are massively overengineered. The frameworks, spans and ties composers construct go far beyond anything the listener needs in order to feel a sense of consistency and plausible continuity. What kind of work do they do, if they are there? The more studies appear that find that listeners do not perceive them,48 the more this question needs a fresh look. Cortot's belief was that what students needed from music history was not structural analysis ('chemical analysis', he is reported to have called it) but biography, in particular stories about the origins and associations of pieces that would offer clues to historically informed interpretation, prompting them to conceive the character for their performances that the composer intended.49 We may not subscribe to that either. But that approach is at least much more concerned than the structural analyst's with style and moment-to-moment continuity, and it is that to which the listener responds.

If we want some understanding of what Cortot is making out of this score, it may be more productive to look at the kinds of associations and interactions his sounds make with our bodies.50 I suggested in a previous study that by matching rising and falling melody with slowing down and speeding up (as he does here), Cortot habitually draws on our embodied sense of motion - that is to say, a sense we have acquired through a lifetime's experience of moving in our environment - giving melodic lines 'a sense of physical motion in a lifelike gravitational field, where going uphill is more effortful than going down; the climbing line generates tension, descending brings release'. And I showed that Cortot was reported to have used in his teaching similar metaphors for the effect of rubato ('moving through resistance' and 'running down hill').51 Moreover, '[t]he sense that the leap takes longer because there is further to travel adds another dimension to

the lifelike quality of this performance. The sense that the music involves space as well as time, gravitational forces, a set of bodily processes, contributes vitally (quite literally vitally) to the power of this performance.'52

The nature of these experiences and the ease with which music models them are well explained by Daniel Stern's theory of 'vitality forms',53 the contours given to the ostensible content of our moment-to-moment experiences by our feeling response to them. The vitality form, the shape of the feelings we experience, is immediately modelled, indeed sounded, by music as its frequency, timing and loudness are modulated by the performance of a score. Through shared vitality forms, music sonifies feeling (in much the sense that data is now routinely sonified).54 And so a performer like Cortot, who is able to modulate timing and loudness constantly, deeply and yet idiomatically within their performance style, is exceptionally well able to model human experience.

Taking this 1920 performance as a whole, therefore, what we seem to be experiencing is the expansion and contraction pattern from the opening ostinato

- speeding through the tonic first beat, slowing through the dominant second -worked out on various levels: from note to note, as the melody rises and falls; from half-bar to half-bar, as the ostinato breathes; through to the variante level; all by means of loudness and timing. This is important not so much because of the multilevel patterning, which may have quite low cognitive value: it remains to be shown that motivic or structural patterning has a role in music cognition. Rather, I am suggesting, the value of Cortot's approach to rubato lies in the way in which the expansion and contraction of time and intensity engage our bodily responses by behaving like breathing, and particularly like intensifying and fading feeling states, so that the music models the dynamics of human feeling in just the manner suggested by Stern's vitality forms. And by 'the music' - such a common expression and yet so underdefined - I mean the experience of responding to the performance of these notes, the music Cortot causes us to perceive.

Listening impressionistically, there is a striking mismatch between the quasi-mechanical musical structure put together by Chopin and the kind of flexible, living performance that Cortot ostensibly gives it. But what is so interesting about Cortot's 1920 performance is that it manages to be at the same time meandering ('les adorables meandres d'une musique imponderable', as Cortot put it),55 reflecting the decorative turning around itself that Chopin's melody seems to enjoy, while at a deeper level making considerable structural sense, marking out the variantes with great clarity and winding down the melody with meticulous attention to the underlying line. Moreover, while highly expressive it certainly is

- more so than anyone else's perfomance56 - it is also uniquely mechanical in a further respect, namely the matching of rubato to melodic direction. What Cortot has done with this technique is to find a way to play much of the piece - including the most evidently thematic variations - as a mechanism and yet, as Chopin is assumed to have done,57 to play it with all his characteristic expressivity signalled through rubato. It is meandering and yet strict; it is Cortot, unpredictable as ever, and yet it is a music box.

That contrast is entirely characteristic of the contradictions in Cortot's personality and behaviour. In his teaching, technique had no role other than to enable interpretation,58 yet he wrote a terrifyingly systematic book on piano technique requiring machine-like dedication to repetition. In the service of interpretation alone, he insisted that a music-historical and music-theoretical education was essential for a performer and co-founded a rival school to the Paris Conservatoire, the Ecole Normale de Musique, in order to ensure it.59 Cortot presented himself as a dreamer in sound; indeed, he says to a student in a masterclass on film, of a piece of Schumann, that 'one shouldn't play this piece, one should dream it'.60 And yet, far from being a dreamer, Cortot was a formidably competent administrator, conductor, organiser, planner and director. He was a willing senior functionary in the Vichy government, to be found in the Paris salons where Nazis were entertained, a willing performer in Germany in 1942 and a huge admirer, one suspects, of systematic and thorough organisation.61 His performances always sounded spontaneous and improvised on the spot. And yet, as I've shown elsewhere, these famously inspired, improvised performances were worked out in advance and reproduced so exactly in performance, down to the most minute hesitations, that you can take two recorded within a couple of years, play them simultaneously, and hear almost one, with the phrase divisions appearing at exactly the same moment in clock time despite the rubato varying in between.62 Cortot's ability to sound like one kind of person while being another is dizzying, and also disturbing. And that perhaps helps to explain what is going on here. The lifelike quality of his performance, its breathing, its tensing and relaxing, its humanity, is real for the listener and intended by Cortot, yet is achieved to some extent systematically by arguably mechanical or, one might say, inhumane methods.

Some Conclusions

Let us come back now to the distinction between Chopin's Berceuse and Cortot's. Neither, of course, is mechanical except in a metaphorical and partial sense. But at the same time, neither is irregular or unplanned. Both offer an appearance of fantasy masking a quasi-automatic procedure. On the face of it they seem rather alike, so that one might almost be tempted to trust Cortot as a representative of Chopin's imagination. Cortot identified deeply with the Chopin he created as a projection of himself, and he collected remains and memorabilia - portraits, autographs, letters, a cast of the composer's hand, a lock of his hair, drawings made at his deathbed - to bring himself morbidly close.63 But his Chopin is still imaginary and self-serving, not to be trusted despite his being so much closer in time. Neither he nor we know Chopin's Berceuse, nor ever will. The notation Chopin left is all that remains of music he imagined, and all it can do is serve as a starting point for new performances and through them for newly made Berceuses. Nonetheless, Cortot makes the score into an experience that is persuasive, one


worth understanding more about, and through it gives us a Chopin who feels plausible, contributing memorable ingredients to our own versions of 'Chopin' (and indeed to those of a great many pianists who followed).64 If one is going to analyse anything of this, it might better be, rather than an imagined amalgam of ideas from recent Chopin playing, a beautiful, thought-provoking, even alarming performance like this one. At least we then know what the object of study is, and we can examine it as minutely as we wish in a form that we can share fully with one another.

This is not all that early recordings do for us, however: their implications go further. We may never know how Chopin sounded, but performances of late nineteenth-century music, recorded in the few decades following, show that in modern performances we very significantly misrepresent the character of compositions by playing them so differently. An 'RIP' (recordings-informed performance) practice, carefully implemented, would make these scores sound to us quite radical.65 And that would have implications for analysis. For it is not just alternative structural readings of moments in a score that different performances afford. When the very nature of a piece changes (for example, a reverential treatment giving way to something more like salon music, as happens in some Brahms recordings by his circle) then what it is that analysis seeks, in adding value to our estimation and experience of a score, requires rethinking. Analysis has developed in tandem with a particular approach to performance, one in which everything of classical music worth playing or studying is reverentially handled, and in which precision and objectivity are owed to the composer and his creation. What is analysis to think of performances that don't share those values, in which notes are skated over, cut out, added, quartered or quadrupled in length? What does it seek to show when pieces are no longer subservient to their texts?

What also follows from the largest differences between early recorded and modern performances is that other, radically new approaches to making music from scores must also be possible. If they can sound as different as we hear on record, how else might they be shaped and yet still make sense to us? Once we let go of the apron strings of objectivity and precision, and of the delusion of faithfulness to the composer (so intimately bound up with the way we play now), what becomes conceivable? Performance style will gradually change, whether we like it or not, but knowing how much it has already changed in just one century opens up the possibility of changing it by design - an alarming notion maybe (despite its having happened already with historically informed performance), but one very much in keeping with developments of all sorts today.

Recordings show that there are many ways of being musically persuasive. What a persuasive performance brings, startlingly in this Cortot case and potentially still more so in others yet to be made, is the interaction of 'the music' as sound with our bodies, allowing us to talk relevantly, and with all the sounding evidence to hand, about music's dynamics in the full motional and emotional senses of that word. The changing quantities and intensities of the music take their meaning from their action on our bodies, including involuntary responses and

our rich stock of memory and association, some shared, some individual, in a process that brings together inherited, embodied, cultural and personal responses competing in our brains to give rise to a complex, multi-sourced and mixed-up sense of meaning that we can discuss and attempt to explain and to share. Cortot was exceptionally effective at triggering this kind of embodied response and associative meaning. But performers unlike Cortot and unlike any we know will achieve this too. Our understanding of compositions will change at the same time, and we shall need to rethink our analytical premises as we go.

I gratefully acknowledge support from the Arts and Humanities Research Council, via the AHRC Research Centres for the History and Analysis of Recorded Music (CHARM) and Musical Performance as Creative Practice (CMPCP), which provided the context for the work written up in this paper. I owe special thanks to Antonio Cascelli, David Fanning, Jeffrey Kallberg, Mats Kiissner and John Rink.

1. Previous analytical uses of specific performances range widely across the literature of music analysis and music psychology. A few examples include Daniel G. Barolsky, 'The Performer as Analyst', Music Theory Online, 13/i (2007); Bryony Buck, Jennifer MacRitchie and Nicholas J. Bailey, 'The Interpretive Shaping of Embodied Musical Structure in Piano Performance', Empirical Musicology Review, 8/ii (2013), pp. 92119; Eric Clarke, 'Expression in Performance: Generativity, Perception and Semiosis', in John Rink (ed.), The Practice of Performance: Studies in Musical Interpretation (Cambridge: Cambridge University Press, 1995), pp. 21-54; Nicholas Cook, 'Off the Record: Performance, History, and Musical Logic', in Irene Deliege and Jane W. Davidson (eds), Music and the Mind: Essays in Honour of John Sloboda (Oxford and New York: Oxford University Press, 2011), pp. 291-309; Alan Dodson, 'Performance, Grouping, and Schenkerian Alternative Readings in Some Passages from Beethoven's "Lebewohl" Sonata', Music Analysis, 27/i (2008), pp. 107-34; and John Rink, Neta Spiro and Nicolas Gold, 'Analysis of Performance', in Anthony Gritten and Elaine King (eds), New Perspectives on Music and Gesture (Aldershot: Ashgate, 2011), pp. 267-92.

2. Daniel Leech-Wilkinson, 'Compositions, Scores, Performances, Meanings', Music Theory Online, 18/i (2012).

3. This has become increasingly clear since Nicholas Cook, 'Words about Music, or Analysis versus Performance', in Nicholas Cook, Peter Johnson and Hans Zender (eds), Theory into Practice: Composition, Performance and the Listening Experience (Leuven: Leuven University Press, 1999), pp. 9-52.

4. The wording here acknowledges (without wholly accepting) Imberty's interesting proposal that musical communication 'is not communication, but it is a representation of our capacity to communicate [ . . . ] communication without an object [.. . ] proto-narration.' Michel Imberty, 'Can One Seriously Speak of Narrativity in Music?', in Alf Gabrielsson (ed.), Proceedings of the Third Triennial ESCOM Conference, Uppsala, 7-12 June 1997 (Uppsala: Department of Psychology, 1997), p. 22.

5. For a definition and fuller discussion of performance style, see Daniel Leech-Wilkinson, 'Recordings and Histories of Performance Style', in Nicholas Cook, Eric Clarke, Daniel Leech-Wilkinson and John Rink (eds), The Cambridge Companion to Recorded Music (Cambridge: Cambridge University Press, 2009), pp. 246-62. As an illustration of the extent of these performance stylistic differences and of their implications for interpretation of scores, see Leech-Wilkinson, 'Sound and Meaning in Recordings of Schubert's "Die junge Nonne"', Musicae scientiae, 11 (2007), pp. 209-36.

6. The thinking behind this formulation is sketched in Leech-Wilkinson, 'Compositions, Scores, Performances, Meanings', §§4.1 and 5.

7. Ibid., §2.3.

8. Some research on this is reported in ibid., §2.2.

9. Caution may be in order. Nieminen, Istok, Brattico, Tervaniemi and Huotilainen, in a survey of existing work, include among the questions still to be addressed empirically, 'Does musical expertise have an effect on aesthetic experiences of music?' S. Nieminen, E. Istok, E. Brattico, M. Tervaniemi and M. Huotilainen, 'The Development of Aesthetic Responses to Music and Their Underlying Neural and Psychological Mechanisms', Cortex, 47/ix (2011), p. 1144. Most of the literature on expert listeners' ability to recognise music-theoretical structures has tested undergraduate music students, largely with negative results. A selection of studies is cited in Leech-Wilkinson, 'Compositions, Scores, Performances, Meanings', §4.8, to which a useful addition is Elizabeth West Marvin and Alexander Brinkman, 'The Effect of Modulation and Formal Manipulation on Perception of Tonic Closure by Expert Listeners', Music Perception, 16/iv (1999), pp. 389-407.

10. Leech-Wilkinson, 'Compositions, Scores, Performances, Meanings', §2.2.

11. Cook, 'Off the Record'; Sigurd Slattebrekk and Tony Harrison, 'Edvard Grieg: Chasing the Butterfly: Recreating Grieg's 1903 Recordings and Beyond' SIMAX CD PSC 1299 (recorded 2007 and 2009, released 2010), and In Beyond the Score: Music as Performance (Oxford and New York: Oxford University Press, 2013), p. 72, Cook quotes Schenker as recommending just this approach to phrase endings in Heinrich

Schenker, The Art of Performance, ed. Heribert Esser, trans. Irene Schreier (Oxford and New York: Oxford University Press, 2000), p. 55.

12. I offered a brief preliminary discussion of this recording in Leech-Wilkinson, 'The Emotional Power of Musical Performance', in Tom Cochrane, Bernardino Fantini and Klaus R. Scherer (eds), The Emotional Power of Music: Multidisciplinary Perspectives on Musical Arousal, Expression and Social Control (Oxford and New York: Oxford University Press, 2013), pp. 41-54, from which I quote several points here. Previous studies of other recordings of the Berceuse are Wojciech Nowik, 'The Expression of Form and Forms of Expression: Fryderyk Chopin's Berceuse in D major, Op. 57, in the Interpretations of Josef Hofmann', in Artur Szklener, John Comber, Kinga Tarka and Wojciech Boiikowski (eds), Chopin in Performance: History, Theory, Practice (Warsaw: Narodowy Instytut Fryderyka Chopina, 2004), pp. 273-85; and Krystyna Juszynnska and Piotr Rogowski, 'Acoustic Analysis of Recordings by Josef Hofmann of Fryderyk Chopin's Berceuse, Op. 57', in ibid., pp. 287-310.

13. Victor 74623, matrix C 22502, take 5; digital transfer by Ward Marston. I am most grateful to Mr Marston for allowing me to use his transfer to accompany this article.

14. The software is freely available, together with online and video tutorials, from The videos in these media files are made by recording the screen activity of a Sonic Visualiser session. Red vertical lines mark the bars in Chopin's score, while blue lines mark the divisions between the variantes: each set is numbered, the numbering of red lines matching the numbering of bars in the score. Other numbers can safely be ignored.

15. Quoted in Jean-Jacques Eigeldinger, Chopin: Pianist and Teacher (Cambridge: Cambridge University Press, 1986), p. 51. Cortot would have known the 1880 French translation: Jan Kleczynnski, Frederic Chopin: de l'interpretation de ses œuvres.

16. The philological implications of 'variants' for Chopin's thinking about the composition are discussed in Jeffrey Kallberg, 'Chopin's Music Box', paper delivered at the joint meeting of the Society for Music Theory and American Musicological Society, 6-9 November 2008. I am most grateful to Professor Kallberg for sharing it with me.

17. Leech-Wilkinson, 'The Emotional Power'. Marion Guck offers an interesting discussion of the breathing metaphor applied to the performance of Chopin in 'Musical Images as Musical Thoughts: the Contribution of Metaphor to Analysis', In Theory Only, 5/v (1981), pp. 29-42.

18. Antonio Cascelli, 'Schenker, Chopin's Berceuse Op. 57 and the Rhetoric of Variations', Ad Parnassum, 2 (2003), p. 62.

19. For the rubato indications see ibid., Table 3, p. 78.

20. Mark Johnson and Steve Larson, '"Something in the Way She Moves": Metaphors of Musical Motion', Metaphor and Symbol, 18/ii (2003), pp. 63-84. A wider context is given in Mark Johnson, The Meaning of the Body: Aesthetics of Human Understanding (Chicago: University of Chicago Press, 2007).

21. Leech-Wilkinson, 'The Emotional Power'.

22. This is an acoustic recording, so the recording machinery was insensitive to the quietest notes, but owing to the nature of recordings made before the advent of tape it was not edited in any way. On recordings as sources for musicology see Daniel Leech-Wilkinson, The Changing Sound of Music: Approaches to Studying Recorded Musical Performances (London: CHARM, 2009), Ch. 3,

23. Daniel Leech-Wilkinson, 'Making Music with Alfred Cortot: Ontology, Data, Analysis', in Heinz Loesch and Stefan Weinzierl (eds), Gemessene Interpretation - Computergesttitzte AujfUhrungsanalyse im Kreuzverhor der Disziplinen, Klang und Begriff 4 (Mainz: Schott, 2011), pp. 129-44.

24. Leech-Wilkinson, 'Recordings and Histories'.

25. Leech-Wilkinson, 'Making Music', pp. 136-7.

26. Data produced by Craig Sapp for the Mazurka Project within CHARM, the AHRC Research Centre for the History and Analysis of Recorded Music. Data available from

27. Most early recordings of the score play many of the left-hand bottom Dts an octave lower or higher than notated. Grunstein provides a good selection of other kinds of alterations in her comments on another selection of recordings. Sarah Grunstein, 'Improvisatory, Compositional and Performance Issues in Chopin's Berceuse, Op. 57' (DMA diss., City University of New York, 2005).

28. Leech-Wilkinson, 'Compositions, Scores, Performances, Meanings'.

29. See especially Cook, 'Words about Music'; see also Cook, 'Music as Performance', in Martin Clayton, Trevor Herbert and Richard Middleton (eds), The Cultural Study of Music: a Critical Introduction (London: Routledge, 2003), pp. 204-14; and, more recently, Cook, Beyond the Score, passim but especially p. 175. See also Lawrence Kramer, Why Classical Music Still Matters (Berkeley and Los Angeles: University of California Press, 2007), especially p. 76.

30. Kallberg, 'Chopin's Music Box'. There is a brief discussion in his 'Mechanical Chopin', Common Knowledge, 17/ii (2011), p. 270.

31. Kallberg, 'Chopin's Music Box', p. 4.

32. Alfred Cortot and Edouard Ganche, Trois manuscrits de Chopin (Paris: DorbonAîne, 1932).

33. In the sketch everything from bar 55 onwards is labelled '14'. For machine convenience, the Sonic Visualiser displays number the remaining four-bar units as 14-17 and mark them with blue bar lines.

34. Wojciech Nowik, 'Fryderyk Chopin's Op. 57: from Variantes to Berceuse', in Jim Samson (ed.), Chopin Studies (Cambridge: Cambridge University Press, 1988), pp. 25-40. Much valuable detail is also considered in Grunstein, 'Improvisatory, Compositional and Performance Issues'.

35. Zofia Chechlmska makes a case for golden-section planning by Chopin in the etudes, albeit with exceptions allowed for missed targets; Chechlmska, 'Chopin in Proportion', in Artur Szklener (ed.), Analytical Perspectives on the Music of Chopin (Warsaw: Narodowy Instytut Fryderyka Chopina, 2004), pp. 157-65.

36. The central peak of the central melodic W shape in Cortot's performance sounds at 2'11" of the 4'22'' duration from the start of the first to the start of the last note. I have suggested (in 'Making Music', p. 131) that Cortot may have had a surprisingly accurate sense of absolute duration at the phrase level; but, even given his very neat timing for the central point peak, one could not seriously suggest controlled planning to this level of precision.

37. Cascelli, 'Schenker', pp. 65-8.

38. Valuably tabulated in ibid., pp. 78-9.

39. See especially bars 563, 573, 575ii, 583ii, 586ii, 595ii, 602ii, 614, 635, 646, 653, 665, 672 and 682, where the superscript arabic numeral signifies the quaver beat and the roman numeral 'ii' the second half of a quaver beat.

40. Nowik, 'The Expression', pp. 282-3; and Juszynnska and Rogowski, 'Acoustic Analysis', pp. 303-5. Hofmann's tolling bell reappears in Demidenko's 1998 recording.

41. Grunstein, 'Improvisatory, Compositional and Performance Issues', pp. 100-3.

42. As illustrated by Cascelli, 'Schenker', pp. 68-9. This is not, of course, to say that the piece is persuasively complete in bar 67, only that in this performance its final has been definitively reached.

43. Frederic Chopin, Pieces diverses, 1st ser., ed. Alfred Cortot (Paris: Salabert, 1936), p. 46.

44. Cascelli, 'Schenker', p. 69.

45. Clemens Wollner, Jane Ginsborg and Aaron Williamon, 'Music Researchers' Musical Engagement', Psychology of Music, 39/iii (2011), pp.


46. Leech-Wilkinson, 'Compositions, Scores, Performances, Meanings', §§4.4-4.5, suggests that it was the impossibility for Schenker, and separately for Adorno, of reconciling their taste in performance style with their theories of music that prevented them from bringing to completion their books on performance. Cook makes a similar point concerning their view of performance as reproduction, with much more context, in Beyond the Score, pp. 88-90.

47. Nicholas Cook, Music, Imagination, and Culture (Oxford and New York: Oxford University Press, 1990), especially pp. 2-4 and 242-3.

48. See n. 9 above.

49. Jeanne Thieffry, Alfred Cortot's Studies in Musical Interpretation (London: Harrap, 1937), pp. 19-20.

50. A key text here is Johnson, The Meaning of the Body.

51. Thomas Manshardt, Aspects of Cortot (Hexham: Appian, 1994), p. 120.

52. Leech-Wilkinson, 'The Emotional Power'.

53. Daniel Stern, Forms of Vitality: Exploring Dynamic Experience in Psychology, the Arts, Psychotherapy, and Development (Oxford and New York: Oxford University Press, 2010). Stern's work provided a central reference point for work on music and shape developed within the AHRC Centre for Musical Performance as Creative Practice between 2009 and 2014. The results will be published in Daniel Leech-Wilkinson, 'Shape and Feeling', in Daniel Leech-Wilkinson and Helen Prior (eds), Music and Shape (forthcoming). Stern's and Johnson's earlier work was related to music in Hallgjerd Aksnes, 'Music and Its Resonating Body', Danish Yearbook of Musicology 29 (2001), pp. 81-100, which includes a discussion of Guck, 'Musical Images'. On Stern and his precursors, Hausegger and Truslit, see Jin Hyun Kim, 'Shaping and So-Shaping Forms of Vitality in Music: beyond Cognitivist and Emotivist Approaches to Musical Expressiveness', Empirical Musicology Review, 8/iii (2013), pp. 162-73.

54. Thomas Hermann, Andy Hunt and John G. Neuhoff, The Sonification Handbook (Berlin: Logos, 2011).

55. Cortot, in Chopin, Pièces diverses, p. 37.

56. In terms of the extent of its rubato this is measurable - the standard deviation of his beat lengths at 0.23 is off the top end of the range of the sixteen performances used here as representatives of the Berceuse tradition - though there is of course more to expressivity than that, and I am speaking here impressionistically. Pires is the only modern pianist here who comes close, at 0.21, but in a much slower performance (ï = 77) which affords much more variation. More typical of modern performance are Demidenko (0.15) and Kissin (0.14). Closest to Cortot's average tempo (ï = 95) is Shelley (ï = 96, 0.16). Most of the early recorded pianists play the score much faster than Cortot, with less space for rubato (Hofmann, ï = 136, SD 0.09; Friedman, ï = 126, 0.12; Paderewski, ï = 110, 0.14; and Backhaus, ï = 109, 0.13).

57. George Hogarth's comments on Chopin's performance of the piece in London in July 1848 are pertinent: 'It is highly finished, new in its harmonies, full of contrapuntal skill and ingenious contrivance; and yet we have never heard music which has so much the air of unpremeditated effusion. The performer seems to abandon himself to the impulses of his fancy and feeling, to indulge in a reverie and to pour out unconsciously, as it were, the thoughts and emotions that pass through his mind'. Jean-Jacques Eigeldinger, Chopin, Pianist and Teacher, as Seen by His Pupils, trans. Naomi Shoet with Krysia Osostowicz and Roy Howat (Cambridge: Cambridge university Press, 1986), p. 294.

58. Karen M. Taylor, Alfred Cortot: His Interpretive Art and Teachings (PhD diss., Indiana University, 1988), especially pp. 391 ff. and Ch. 8, pp. 458-557. See also Thieffry, Alfred Cortot's Studies, and Manshardt, Aspects.

59. Taylor, Alfred Cortot, pp. 456-81.

60. Christian Labrande and Donald Sturrock, The Art of Piano: Great Pianists of the 20th Century (NVC Arts, videotape 3984-29199-3, DVD 3984 29199 2, 1999), Ch. 19, 56'11//-58/47//.

61. Myriam Chimenes, 'Alfred Cortot et la politique musicale du gouvernement de Vichy', in Chimenes (ed.), La vie musicale sous Vichy (Paris: Editions Complexe, 2001), pp. 35-52.

62. Leech-Wilkinson, 'Making Music'.

63. Alfred Cortot, In Search of Chopin (London: Nevill, 1951), Chs 1-2, pp. 9-23.

64. Cortot's annotated editions are still widely used in piano teaching, and in conversation many pianists still speak admiringly of his playing while not wishing to copy it. See Labrande and Sturrock, The Art of Piano.

65. For such a practice see Anna Scott, 'Romanticizing Brahms: Early Recordings and the Reconstruction of Brahmsian Identity' (PhD diss., Leiden University, 2014). See also Slattebrekk and Harrison, Chasing the Butterfly.


Daniel Leech-Wilkinson is Professor of Music at King's College London. He has led AHRC-funded research projects on early recordings, performance expressivity and the ways in which musicians use notions of shape. His other publications include The Modern Invention of Medieval Music (Cambridge University Press, 2002), The Changing Sound of Music (CHARM, 2009; and, with Helen Prior, Music and Shape (forthcoming).


Alfred Cortot's 1920 recording of Chopin's Berceuse has some unusual properties - illustrated here in a discussion of the relationships among rubato, loudness, variation form and melody - which shed new light on the score and exemplify the pianist's ability to trigger embodied metaphor with unusual intensity. Comparisons with other recordings are made, and Jeffrey Kallberg's image of the Berceuse as a music box is considered in relation to the layout of the score and Cortot's performance. Drawing on the work of Antonio Cascelli, I compare Schenker's and Cortot's readings of melodic structure, which demonstrate the ecological validity of Cortot's construction. Some of the many respects in which analysis depends upon performance are discussed, as is the likelihood of very different performances in the future and the expectation that analysis will adapt itself to changing approaches to performance. The article is illustrated by Sonic Visualiser analyses presented as YouTube videos.