Schumann's Kreisleriana

[(poco) Sforzando]

All right, let's see if these links work for you guys. I'll put this in the Bruckner thread too, but here's my little experiment with the scherzo to the 7th:

https://www.dropbox.com/s/6k1hvxjcmk6rn3e/Bruckner%207%20piano%20MP3.mp3?dl=0
https://www.dropbox.com/s/jygw6ldd1oxptus/Bruckner%20start%20single-dotted%20MP3.mp3?dl=0
https://www.dropbox.com/s/o7rku3uxx5837jr/Bruckner%20start%20double-dotted%20MP3.mp3?dl=0
https://www.dropbox.com/s/iccoi6nnzqye035/Bruckner%20end%20MP3.mp3?dl=0

#1 just outlines the main motifs from the scherzo on the piano, with double-dotting
#2 plays the opening 16 bars with single-dotting (as Bruckner did not write)
#3 plays the same 16 bars with double-dotting (as Bruckner wrote)
#4 plays a later passage from the movement with double-dotting

First, let me know if the files play for you; otherwise I'll have to try a different technology. Second, do you hear a difference between 2 and 3, and if so, do you know a recording that sounds like 3? And is the double-dotting more pronounced in 4?

Perhaps next I'll try to generate a passage from the Beethoven 7th. I don't think Tchaikovsky 5 causes the same rhythmic problems, because the "siciliano" rhythm is broken up by other patterns, unlike the B 7th where it is maintained over and over without any rhythmic contrast.

[Sergeant Rock]

First, let me know if the files play for you; otherwise I'll have to try a different technology.

They play, no problem.

Second, do you hear a difference between 2 and 3, and if so, do you know a recording that sounds like 3?

Not really. The difference is so subtle I would not be able to find a recording (of the twenty-one I have) that matches 3. Sorry.

conductors almost invariably play it as single-dotted quarter

That  makes me feel a bit better. Apparently most conductors can't hear a difference either 

Sarge

[(poco) Sforzando]

They play, no problem.

Not really. The difference is so subtle I would not be able to find a recording (of the twenty-one I have) that matches 3. Sorry.

That  makes me feel a bit better. Apparently most conductors can't hear a difference either 

Sarge

I agree that the distinction is subtle. But before anyone is ready to dismiss it, consider that Bruckner explicitly wrote the rhythm this way in hundreds of cases, where writing a single-dotted quarter - eighth - quarter would have taken a lot less effort. Neither am I much impressed by the mention of 21 recordings. I wouldn't be impressed if there were 121 recordings that played the rhythm inaccurately, and deviations from a composer's score are far more common than most listeners recognize. We have Imogen Cooper playing Kreisleriana 8 and getting the rhythms completely wrong for a start. I refer you all again to Gunther Schuller's book on conducting for further evidence.

Take also the issue of a composer's tempo indications and metronome marks. The opening of the Bruckner 7 first movement is written as 2/2, Allegro moderato, 58 to the half. This is from the critical edition, which presumably has some authority, and this means that the first 24 bars of the symphony should need about 50 seconds to play. Sampling four performances on YouTube today, I found only one that came close. Celibidache needs 85 seconds to complete the passage in a plodding 4/4 Andante, and so does Jochum, both substantially slower than the indicated tempo. Klemperer takes the passage in 67 seconds. Only Abbado takes 60 seconds, putting him within striking distance of Bruckner's metronome mark. Anyway, here's a computer-generated file of these opening 24 bars taken at Bruckner's own tempo, a nicely flowing Allegro moderato:

https://www.dropbox.com/s/qc5cgqcilcdtpso/Bruckner%207-1.mp3?dl=0

As for the rhythm of the scherzo. a piece which someone I know called a sea shanty (where, in landlocked Vienna?). I would be inclined to play that small quarter note almost as a grace note. I was hoping the computer files would register this distinction more sharply than they did. You might assume a computer file will have a high degree of accuracy, but in those examples of mine from full score there are so many computations that the machine may have a slight degree of lag. For this reason, the simpler outline file played on the piano is where I think you can hear the double-dotted rhythms best.

[amw]

I agree that the distinction is subtle. But before anyone is ready to dismiss it, consider that Bruckner explicitly wrote the rhythm this way in hundreds of cases, where writing a single-dotted quarter - eighth - quarter would have taken a lot less effort. Neither am I much impressed by the mention of 21 recordings. I wouldn't be impressed if there were 121 recordings that played the rhythm inaccurately, and deviations from a composer's score are far more common than most listeners recognize. We have Imogen Cooper playing Kreisleriana 8 and getting the rhythms completely wrong for a start. I refer you all again to Gunther Schuller's book on conducting for further evidence.

Take also the issue of a composer's tempo indications and metronome marks. The opening of the Bruckner 7 first movement is written as 2/2, Allegro moderato, 58 to the half. This is from the critical edition, which presumably has some authority, and this means that the first 20 bars of the symphony should need about 50 seconds to play. Sampling four performances on YouTube today, I found only one that came close. Celibidache needs 85 seconds to complete the passage in a plodding 4/4 Andante, and so does Jochum, both substantially slower than the indicated tempo. Klemperer takes the passage in 67 seconds. Only Abbado takes 60 seconds, putting him within striking distance of Bruckner's metronome mark.

It should actually be ~41 seconds according to my calculator. It's 46 in Herreweghe's recording, although due to articulation he doesn't sound very "Allegro moderato". Norrington's right on the nose at 40, but for some reason, most people haven't yet warmed to 15 minute B7 openers. >_>

I don't know of anyone who plays the trumpet rhythm accurately either! (At dotted minim = 80 it would indeed sound like a grace note, but no one does that.) Or for that matter the string rhythms later on, where they're supposed to play double-dotted crochet—two demisemiquavers—crochet. It's always dotted crochet—two semiquavers—crochet. Can't blame them, at Bruckner's speed

During the B6 blind comparison I noticed that metronome marks appear in the manuscript (though not in the score). For the first movement, it's minim = 72, so the first 24 bars (until the first tutti) should take 40 seconds. Almost everyone—including Norrington!—takes either the same basic tempo that gets them to the first tutti around 50-54 seconds, or a slower basic tempo that gets them to the first tutti in a minute+. This time the prize for following instructions goes to Marcus Bosch and the Aachen Symphony Orchestra. Then at the start of the second theme he slows down to minim = 50—and doesn't indicate a return to the original tempo until the recapitulation. (Almost everyone goes back to it at the "third theme", including Bosch, who therefore loses his prize. Only Dennis Russell Davies & the Bruckner Orchestra Linz do it properly, but they're not following the metronome marks—main tempo closer to minim = 60, secondary tempo closer to minim = 40—so it sounds really forced, like that 100 minute Klemperer Mahler 7 some people love.)

[(poco) Sforzando]

It should actually be ~41 seconds according to my calculator. It's 46 in Herreweghe's recording, although due to articulation he doesn't sound very "Allegro moderato". Norrington's right on the nose at 40, but for some reason, most people haven't yet warmed to 15 minute B7 openers.

Typo on my part above, now corrected. I should have written not "the first 20 bars" but "the first 24," which is the number of bars I entered into my file. I stopped just before the full orchestra begins its restatement of the main theme; hence 49-50 seconds for the first 24 is correct.

[Jo498]

I can hear the double-dotting well in the "piano version" but only a very slight difference in the "orchestrated" one.
Some time ago I read that the metronome markings in the 7th may not be by Bruckner himself. But the scherzo is supposed to be one of the faster ones and in whole bars, so 80 does seem a quite appropriate tempo (it does not sound too fast).
And as the melody has also something of a trumpet signal, I can't really see that the 16th should sound like a grace note.

Strangely, I find the character of the "fast" beginning of Bruckner 7 not terribly different from the considerably slower recordings I have. (The most flowing I have is probably Gielen (1980s SWR orchestra) or Horenstein ca. 1929.)

[(poco) Sforzando]

I can hear the double-dotting well in the "piano version" but only a very slight difference in the "orchestrated" one.
Some time ago I read that the metronome markings in the 7th may not be by Bruckner himself. But the scherzo is supposed to be one of the faster ones and in whole bars, so 80 does seem a quite appropriate tempo (it does not sound too fast).
And as the melody has also something of a trumpet signal, I can't really see that the 16th should sound like a grace note.

Strangely, I find the character of the "fast" beginning of Bruckner 7 not terribly different from the considerably slower recordings I have. (The most flowing I have is probably Gielen (1980s SWR orchestra) or Horenstein ca. 1929.)

Then perhaps I should quickly put together a "piano" version where the d-d will be more apparent. The software I am using (Finale) is intended primarily as a notation vehicle, not a substitute for an orchestra, and the greater the number of "instruments" I include, the more likely that the playback will suffer as the software and processor have to work harder to perform their calculations.

[Brian]

It should actually be ~41 seconds according to my calculator. It's 46 in Herreweghe's recording, although due to articulation he doesn't sound very "Allegro moderato". Norrington's right on the nose at 40, but for some reason, most people haven't yet warmed to 15 minute B7 openers. >_>

It's his orchestra's sound, really, that's the issue. Since I have the first three movements pretty much memorized, the orchestra in my head has been playing the first movement nearly at the proper speed since last summer (that clip Sforzando posted is still a little too fast for me to adjust to) - around 16:00-16:30. The key, though, is to keep that expressive warmth and aching beauty - if you can pull that off, this is among Bruckner's most passionate and humane creations. It feels youthful, rather than feeling like something from a classical sculpture gallery. But for me the Norringtization goes a little too far in the stripping-down.

[(poco) Sforzando]

It's his orchestra's sound, really, that's the issue. Since I have the first three movements pretty much memorized, the orchestra in my head has been playing the first movement nearly at the proper speed since last summer (that clip Sforzando posted is still a little too fast for me to adjust to) - around 16:00-16:30. The key, though, is to keep that expressive warmth and aching beauty - if you can pull that off, this is among Bruckner's most passionate and humane creations. It feels youthful, rather than feeling like something from a classical sculpture gallery. But for me the Norringtization goes a little too far in the stripping-down.

For a variety of reasons affecting tempo relationships (both within the 1st movement and in relation to the others), I think the metronome marking in the score valid. Consider the big C minor outburst in the 1st movement; Bruckner marks it Tempo Primo and most versions I know take it at about half=60. But Tempo Primo means also that this should be the tempo of the opening theme, and if that theme is played at the usual droopy Andante the relationship is lost.

Where the 1st movement is marked half=58, the slow movement is eighth=63 - almost an identical pulse (and which in practical terms means that the main theme of the slow movement, which has rhythmic parallels to the main theme of the first, is to be played twice as slowly as the opening of the first movement). And the finale is half=63, virtually the same tempo as the first theme of the first movement. In this way, when that theme returns at the very ending of the symphony, it will be heard (if the score's markings are observed) in almost the identical tempo.

That said, I won't want to buy the Norrington. Norrington was the next big thing in the 80s when his LCP version of LvB's 2-8 was given a review by Richard Taruskin in Opus Magazine that made it sound like the second coming. But the chinks in the armor became readily apparent even to Taruskin with his next Beethoven, the 9th, and then we got some bizarre things like a Fantastique that managed the almost impossible, making Berlioz sound boring. That first Beethoven recording remains outstanding, but based on other things I've become wary of listening to ol' Rog.

[(poco) Sforzando]

One last try with the Bruckner scherzo, though I may work up a bit of the Beethoven 7th tomorrow. These excerpts use the piano sound only, and isolate only the motif using the dotted rhythm. One of the skeletons is as Bruckner wrote the piece, with double-dotting, the other is as I believe it's usually heard, with single-dotting:

https://www.dropbox.com/s/az4y1ijxl2bhjd8/Bruckner%20skeleton%201.mp3?dl=0
https://www.dropbox.com/s/jebatrxq6xs3qc7/Bruckner%20skeleton%202.mp3?dl=0

[(poco) Sforzando]

Well, no further interest on my Bruckner samples, it seems. But if it is assumed that I am trying to hijack this thread to turn it away from the Kreisleriana and use it to discuss unrelated topics from Bruckner and Beethoven, that is not the case. I am taking my cue from the discussion of rhythm in Kreisleriana 8, and using that to discuss the problem of dotted rhythms, specifically in quick 6/8 time. As has been said, Imogen Cooper does not get the rhythms in 8 right, she doesn't even seem to try to fit the shorter notes in correctly, and so I'm uploading her Kreisleriana 8 so you can compare it to Kuerti and to my computer sample. Gunther Schuller in his book The Compleat Conductor calls these fast dotted six-eighth rhythms among the hardest to play, and we've seen already what problems they cause in the scherzo to Bruckner 7.

https://www.dropbox.com/s/yflvfnr32uf3ex9/Cooper%20Schumann%20op.%2016-8.mp3?dl=0

Today I tried to generate a computer file of a passage from LvB 7, a notorious example of the problem. But it backfired. I notated the passage correctly, but the playback mechanism from my Finale software tended to distort the rhythm unexpectedly. (Specifically in certain bars it transformed the rhythm into a duple, something like quarter-eighth-eighth, in an effort to fit all the notes in at the metronome mark I coded.) So unfortunately my experiment was not quite a success, although if anyone is interested I'll upload it. Suffice it to say that according to Schuller, Toscanini gets the rhythms almost completely wrong, and Carlos Kleiber gets them almost completely right.

A word about Gunther Schuller. He was brilliant, and something of a lunatic. His book The Compleat Conductor analyzes numerous recordings of eight works (Beethoven 5 and 7, Brahms 1 and 4, Schumann 2, Till Eulenspiegel, Daphnis Suite 2, and Tchaikovsky 6). You will need scores to follow his discussions. His premise is that accuracy to the score is essential, and he finds flaws in almost every recording out there. In fact, he rarely recommends a recording in toto; instead he will praise individual passages but almost never a performance as a whole. Still, I regard his book as essential reading for anyone concerned with how music is performed and the relation of a performance to the score. His 20 pages discussing the phrase structure of the first movement of LvB 5 are in my opinion brilliant. And he was a superb practical musician as well. His main concerns in the book are balance, tempo, phrasing, and dynamics, and his own recording of the LvB 5 is the best I have ever heard. (Many have disagreed with me. But for one thing, he balances the finale so that you can actually hear the piccolo, trombones, and contrabassoon. How often can you say that?)

The question he does not consider is whether total fidelity to the score is always essential or even possible, and whether departures are inevitably fatal or even deleterious. And since all his criticisms pass through Gunther Schuller's ears, how accurate is his own hearing? One case in point concerning score fidelity, Charles Rosen again: he recorded two of the Boulez sonatas in the presence of the composer, and Boulez wanted a particular passage slower. "But I'm already slower than your metronome mark," protested Charles. "Never mind," replied Pierre. "Follow the sonority." And if Schuller's premise is accepted without question, numerous beloved and celebrated recordings would be tossed out the window.

I'm waiting for my Charles Rosen CD to arrive. Let's see how he does with Schumann's rhythms.

[kishnevi]

Hijack away.  I am finding your discussions fascinating.

Which Rosen CD is this, btw? The Sony box contains a recording of the Davidsbundlertanze and Carnaval, but no Kreisleriana (but does have what I presume to be that Boulez recording)

[(poco) Sforzando]

Hijack away.  I am finding your discussions fascinating.

Which Rosen CD is this, btw? The Sony box contains a recording of the Davidsbundlertanze and Carnaval, but no Kreisleriana (but does have what I presume to be that Boulez recording)

See upthread. The Rosen is OOP on the Globe label, and was originally issued on LP by Nonesuch.

[(poco) Sforzando]

I had a hunch that if I slowed down the tempo for my LvB7 file, the computer would have time to fit in the rhythms properly. And so it came to pass.

https://www.dropbox.com/s/nz0pm8sgakm5te3/LvB%207%20fast.wav?dl=0
https://www.dropbox.com/s/9mtfzgnyjw57xsk/LvB%207%20slow.wav?dl=0

Of course, your processor, and how much else your computer is doing at the moment, may affect the way you hear the file. In any case, the "fast" version is at Beethoven's dotted quarter=104, and the slower one is, well, slower - about dq=84.

[Jo498]

Thanks!
Please do not interpret lack of immediate reply, commentary (or even download) as lack of interest.

[(poco) Sforzando]

Thanks!
Please do not interpret lack of immediate reply, commentary (or even download) as lack of interest.

Thank you for that. One never knows.

[Pat B]

Today I tried to generate a computer file of a passage from LvB 7, a notorious example of the problem.

Thanks. Is the tendency is to play them swung instead of dotted? For me this is much more difficult to mentally play than the Kreisleriana excerpt.

I put the Schuller book on my wishlist.

[(poco) Sforzando]

Thanks. Is the tendency is to play them swung instead of dotted? For me this is much more difficult to mentally play than the Kreisleriana excerpt.

I put the Schuller book on my wishlist.

I can't quite relate to those terms, but let me try to create a very brief skeleton example that will demonstrate the difference between playing as written (dotted eighth-sixteenth-eighth in 6/8 neter) and as sometimes happens when the meter breaks down into duple (dotted sixteenth-thirtysecond-eighth in 2/4 meter). My LvB files, as noted above, got too ambitious and I didn't reckon with the distortions introduced by the computer playback. I'll also upload passages from Toscanini and Kleiber, so you can test for yourself if Schuller's ear is accurate. This will be a little later in the day, however.

[(poco) Sforzando]

On the LvB7, let's start here, to test our own ears and the accuracy of Schuller's analysis. Schuller claims that Carlos Kleiber's rhythms are correct, while Toscanini NBC and Gardiner are incorrect most of the time. I've extracted the development section from the first movement from each of the recordings. Compare these to the "slow" version of my computer file, which is rhythmically exact. What do you think?

For good measure, I've included a clip from Schuller's LvB5, so you can hear the impact of the "special instruments" for yourself, as well as the exceptionally clear and strong bass lines in the finale.

https://www.dropbox.com/s/v9ievpozulnyjjl/LvB7%20Gardiner.wav?dl=0
https://www.dropbox.com/s/cc6j1hp9s2hly6j/LvB7%20Kleiber.wav?dl=0
https://www.dropbox.com/s/wurdkslz02w8p5p/LvB7%20Toscanini.wav?dl=0
https://www.dropbox.com/s/7y58flncec4fmd7/LvB%205%20Schuller%203-4.wav?dl=0

[Jo498]

I downloaded all the samples and the "piano" LvB play fine. (I have yet to try the excerpts from actual recordings) You were probably right when somewhere above you wrote that the problem here is the persistence of the same rhythm that leads to a shortening/distortion.

As for the Bruckner, I cannot help thinking that the difference between simple and double dotting will hardly ever be playable clearly with full orchestra at the appropriate tempo (I don't know for sure, as I don't play the trumpet), so I don't know what he thought. I don't really care, it sounds o.k. to me in the usual way.

I do care about the Beethoven and Schumann because they are certainly playable as written and today's pianists, conductors and orchestras who should be able to play far more difficult music should be able to do it right.