Images 2e série: 

Poissons d’Or 

Claude Debussy

Original recording by Ricardo Viñes:

My recreated performance:

Cross-cut video:

Annotated score:

[…]Debussy sat down at the piano and started to play, in his supple, velvety style: Poissons d’Or. Then he showed me, laughing up his sleeve, that he had dedicated it to me. I thanked him, overcome with emotion. The only other pianist he awarded this favor to was Chopin, in whose memory he had dedicated a collection of pieces for the piano.[1]  

The dedication of the Poissons d’Or composition to Ricardo Viñes is a testament of Debussy’s appreciation for the pianist who premiered such a large portion of his piano solo works. Having established both that there are several fundamental romantic performance practices in the playing of Viñes, and that we also find these same practices in the playing of contemporaneous pianists, I set out to recreate the four Debussy recordings. Are the same “romantic” practices also present in the recordings of “impressionistic” repertoire?

The recreating experience

Except for a few bars which shall be discussed, there are not substantial deviations between the recording and the original score. The exceptions that do exist have been indicated in my annotated score. As before, the annotated score should be used as a supplement to this text and my recreated recording.  

Having recreated this Poissons d’Or recording, I decided to divide the “unmarked” tendencies of Viñes into four categories: 

• Tempo modifications (rushing/slowing)
• “Dislocation”
• “Polyphonic rubato”
• Sonority (pedaling and brisk articulation)

Firstly, Viñes is using rushing/slowing extensively (I have marked the most relevant examples of this in the score with thick red arrows). The general tendency of his tempo is that it is “forward moving”, much like the more virtuosic pieces of Borodin and Scarlatti. As with the earlier recreated pieces, there is a balanced use of rushing and slowing. We also hear sections that fit well into my “rubber ball” metaphor. Bars 46-51 in the annotated score is a good example of this. 

The second aspect is the “dislocation” of voices, mostly by playing the melody after the accompaniment. This is heard throughout the piece and is therefore a general tendency. It is interesting to note that there are few added arpeggios to the chords. I suspect the reason for this is that the chords in this piece seldom consists of individual voices and are therefore played “un-rolled” like the non-polyphonic chords in other pieces. 

The third aspect is the “polyphonic rubato” between the two hands. Bars 14-15 is a good example of this, as illustrated in these figures:

Original notation of bars 14-15

Viñes’s execution of bars 14-15

Rather than “waiting” for the left hand to finish, Viñes plays the right hand much like he did in bars 10-11. In other words, the right-hand timing is independent from the left. There is also a subtle use of “dislocation” happening between the two hands, meaning that Viñes’s execution consists of more layers then I have written out in the above example. This is also an example of chords that Viñes does “arpeggiate” (on the first beat of both bars). In my experience, this is a typical example where “arpeggiation” is used as an expressive tool – emphasizing a tenuto feel on the first beat of the bar. 

The fourth aspect is how the legato in the main themes are realized through a staccato-like finger attack together with a rich use of the sustain pedal. In a closer bar-by-bar analysis of the Viñes performance versus the original score, I have found the following deviations, also marked in the annotated score:

The first F# is played tenuto,[2] followed by a series of tremolos with two distinct «swells» before the melody starts. Viñes does not play the F# at the beginning of bar 2, but he does play it at the beginning of bar 3. The Bb and F (perfect 5th) at the end of bar 7 is played an octave below what is written in the original score. The high C# at the end of bare 8 is omitted so that it is not played twice, but only once before the passage starts to descend. Se above for the deviations in bars 14-15. In bar 37, the last three notes (G + D + G) are played more so as a triplet. In 51, he skips the last appoggiatura figure, ending one octave lower than written (omitting the last 5 notes). In 53, he does the opposite - playing this passage in 3 octaves. The D (independent voice) in bar 86 is not emphasized. In bar 93, he plays eight demisemiquavers in the last figure (not 7 as written) – identical to bar 91, but with ritardando. In bar 96, he adds an arpeggio at the end of the bar before playing the last notes of the piece in bar 97.

As an interesting side-note, according to Mauruce Dumesnil, Debussy wanted the opening tremolos to be played “almost immaterial” and to let the “two clarinets” be heard above.[3] This can be heard in the performance by Viñes. The tremolos are “gestural” rather than literal in a rhythmical sense. Indeed, this “immaterial” way of playing virtuosic passages is found throughout the piece, for example in the aforementioned bars 51 and 53. 

Comparison to other recordings

In Seong-Jin Cho’s recording, there is an evident practice of slowing down the tempo at the end of the phrases.[4] This can be seen as a contrast to Viñes’s “forward-moving” and even rushed pace. There is a lot of rubato in his playing, but as we have seen in other modern-day recordings, this seems to be in favor of slowing down rather than rushing. This performance has great clarity and articulation, but because of the range of different tempos and the very coherent use of slowing down towards the end of the phases, his recording seems more fragmented, as opposed to the larger “sweeps” in the case of Viñes. 

In the recording of Claudio Arrau, we also hear a tendency of slowing down, but here with less general rubato and played in a more even tempo.[5] His general timing is also on the heavier side compared to Seong-Jin Cho, which results in several of his demisemiquaver notes being played as semiquavers instead. 

The recording of Marcelle Meyer is very interesting because she was for a period a student of Ricardo Viñes.[6] Her recording is evidently much faster than Cho’s and Arrau’s, and the duration of the recording is in fact very close to Viñes’s. Compared to her former teacher, Meyer has a rubato which is balanced in the sense that she is using equal amounts of rushing and slowing. We can hear several sections that fit well into my “rubber ball” metaphor, for example in bars 10 and 14, where she greatly accelerates after the tenuto on the first beat of the bars. In Meyer’s recording, we also hear a subtle use of “dislocation” and polyphonic timing, for example in bars 74-75 and 78-79. The practice is however not as evident as in the Viñes performance. At the end of the piece, Meyer does not follow Viñes example when he ads an extra rolled chord before the final bar. There are several traces of Viñes’s playing to be found in Marcelle Meyer’s recording, but the main difference between them is that where Meyer more often slows down, and thus divides the piece into several sections, Viñes’s performance is made up by several longer “sweeps”.

Final thoughts  

As I have demonstrated, we find several of the “romantic” practices of Viñes also in this Debussy piece. The most un-typical aspects that are seldom heard in modern performances of this piece is the use of rushing, “dislocated attacks” and “polyphonic rubato.” We hear several examples of both rushing and “dislocated attacks” in the recording of Meyer, but it is only in the Viñes recording that we clearly hear a practice of “polyphonic rubato.”

Previous                                                                                                                                                                              Next