Repeats in Baroque music / Leaving out repeats to fit something onto discs

Juozas Rimas wrote (February 28, 2002):
< Bradley Lehman wrote: “Then he omits the Allegro's repeats, throwing off the architectural balance of the whole thing. " >
Abstracting myself from the recording in question (frankly, I usually find the omission of repeats unwanted myself), I'd like to ask the following questions:

Are repeats marked by Bach or is there a firm Baroque tradition to play them
obligatorily?

Or maybe it was considered obligatory to repeat Gavotte I after Gavotte II but the performer was given a free hand to omit the repeats inside one piece (eg a Goldberg variation)?

Also, are repeats meant to be played exactly the same or do they imply a certain initiative of improvisation from the performer? I have heard both seemingly identical repeats and very diverse ones (eg Gould's - playing a voice staccato in the first play and legato in the repeat). As there are less possibilities to make such big changes when playing on the harpsichord, should we draw the conclusion that, from the historical point of view, repeats should be played as closely to the first play as possible?

Thanks!

Bradley Lehman wrote (March 1, 2002):
< Juozas Rimas wrote: Are repeats marked by Bach or is there a firm Baroque tradition to play them obligatorily? >
Typically they're marked.

< Also, are repeats meant to be played exactly the same or do they imply a certain initiative of improvisation from the performer? I have heard both seemingly identical repeats and very diverse ones (eg Gould's - playing a voice staccato in the first play and legato in the repeat). As there are less possibilities to make such big changes when playing on the harpsichord, should we draw the conclusion that, from the historical point of view, repeats should be played as closely to the first play as possible? >

Objection: counsel is leading the witness! I disagree with the premise of your question.

A seriously-trained harpsichordist playing on a good harpsichord has AT LEAST as much articulative range of expression as a pianist...perhaps more.

Pianists are trained to do their expression primarily with dynamic contrast: playing individual voices more loudly or quietly than others. If they use the sustaining pedal, the releases of notes are not controlled by the fingers anyway. Pianists can of course use articulative nuances also, but that's typically secondary under dynamic shaping.

On harpsichord, articulative control is everything: the beginning and end of every note is crucial, with the fingers on the surface of the keys to control attack and release precisely. Rhythmic and articulative subtlety takes the place of bringing out notes by playing them more loudly or quietly...the notes can be brought out by articulation and timing. They can be somewhat longer or shorter than notated, and within the musical texture they can start or end at slightly different times from one another. (Even more range is developed by playing clavichord and organ, also...every instrument teaches the player additional techniques of expression.)

Pianists can develop such a range of subtlety in articulation and timing, of course, but usually it's not done to that extent. More typically, pianists use articulation to change the prevailing tone color of entire passages, rather than trying to differentiate that articulation from note to note within a passage. Broader strokes of tone color, with fewer changes from moment to moment...the moment-to-moment dynamic expression is done with volume.

Now, applying that to the question of repeats: a good MUSICIAN (on any instrument) is sensitive to the musical and dramatic differences of playing the same notes multiple times. That's true whether it's the repeat of a section, or multiple presentations of the same (or similar) phrases within a passage. The performer's musical sensibility and imagination guide the way s/he gives each note or phrase a convincing amount of expressive weight within the whole structure. There's no way this can be notated. The performer might change the amount of legato, or point the rhythms more sharply, or change dynamics (on some instruments), or change tone color, or change the tempo slightly, or even change some of the notes, all in the interest of making the music more expressive to fit each moment...and it could (I'd say "should") be vastly different from performance to performance.

Really it all comes down to questions of tradition and "good taste." How much should the musician inflect each section of music, and with which techniques? That's musicality and taste and experience. A good musician has a large range of possibilities available, and selects a convincing expression for every moment between the beginning and end of a composition. Time is the structure. The good performer fills up that time convincingly.

Trevor Evans-Young wrote (Maarch 1, 2002):
[To Juozas Rimas] I don't know what the tradition was or is(there are arguments about the repeats in Brahms symphonies and Mahler 6th!) but I do know that Gould made multiple recordings in the Suites; 1^st half, 1st time; 1^st half, 2^nd time; 2^nd half. He then would listen to see how he could make the best mix from the set. For example he would use Take 2 for the 1st time, Take 5 of the repeat and Take 4 of the 2nd half. He probably did 4 or 5 takes of each part. Are there artists who would not play the repeat again and just use the one take over for the repeat? I don't know but, I think if you are going to repeat you should do something different.

Tjako van Schie wrote (March 1, 2002):
Sometimes a repeat has a function, sometimes it is merely extending the music to a longer time span (for entertainment purpose ? or better timing in a mass?). Remember also that the music was 'new' in the days it was written. So a composer might choose to put a repeat in place so the listeners would be enabled to understand better what they have heard before when they hear it a second time. This would be a pledge to play 'all' repeats.

However, sometimes the playing of only the first repeat will suffice, and the second half contains more elaboration on the presented material from the first half. In that case a repeat is less necessary.

There were many traditions regarding repeats. One was to play the repeat with ornamentation, another was to play the repeat with a different 'color' (e.g. shift in register on harpsichords) . Also sometimes the repeats were played with different articulation and/or timing.

In baroque music the repeat often was an invitation to play heavily ornamentation on a melodic schedule (the written notes were only a 'skeleton': as we can see e.g. in the beatiful Sonata for oboe and b.c. by Geminiani. Look at the slow movement!)
Conclusion: there are no strict rules...

(fyi: In my Bach Goldbergs CD I chose to get rid of some repeats as the CD lenght prohibited more than 75 minutes.)

Bradley Lehman wrote (March 1, 2002):
< Trevor Evans-Young wrote: Are there artists who would not play the repeat again and just use the one take over for the repeat? I don't know but, I think if you are going to repeat you should do something different. >
In the Rousset recording of Bach's French Ouverture that I reviewed here a few days ago, there is some obvious regeneration (using the same take twice) in the first movement: the entire fugal section from bars 21 to 143. It's 3'23" of music both times: 2:27 to 5:50 and 7:02 to 10:25.

I noticed it first in the bars 104-123 (4:43 to 5:16 and 9:17 to 9:50): a section played on the upper manual. Rousset lurches into a different tempo and character here, it doesn't blend with the rest of the fugue very well, and it's clearly from a different take, an insert. He does some strange interpretive things in it, too, enough to make me notice immediately (first time) it's an insert. Then in the repeat, 9:17 to 9:50, it's the same insert, easily recognizable, equally obtrusive.

On that first listen through the recording I also noticed immediately that the fugue's return at 7:02 is an edit: in performance cit doesn't go well at all with the ending Rousset has built up in bars 160-162 immediately preceding it. (If he was intending to go into a repeat of the fugue, he would have shaped bars 160-162 differently; due to the editing here, the return of the fugue comes as a big surprise, an abrupt shift of mood.)

So then I wondered: is the whole fugue regenerated? It's clear that the first section, 7:02 to 7:43 (bars 21-47), is regenerated from 2:27 to 3:08; I checked that after checking the 104-123 above. Sure enough, exactly the same. So, we now know that 21-47 and 104-123 are regenerated, and the whole fugue takes 3'23" both times. Hmm.

If I cared enough about this performance to finish the investigation with direct comparisons of every section, that is if I could stand to listen to this "non-performance" that closely another time today, I know what I'd find anyway: the whole fugue second time around is just a digital copy. It's much less work for the producer and editor that way, rather than trying to re-edit different sections together to make the second time sound different. Just copy the whole thing and hope nobody notices.

The point was, the "performance" didn't sound like a real performance even on my first time listening to it a few days ago: the digital shenanigans were obvious enough to block my enjoyment of the music. What we've got here is a bunch of short disconnected sections assembled end to end, making up 3'23" of fugue. Little bits of cloth stitched together, getting all the notes accurately. Then that whole collection of sections is copied exactly for the repeat. All it does here in musical effect is make the music twice as long without giving us any new perspective on it. It's insulting. This recording would have sounded more musical if Rousset and his producer (Wadland) hadn't taken that repeat at all.

When does anyone ever say the same thing EXACTLY the same way twice? One always re-frames things slightly to emphasize different things, otherwise there's no point in saying something twice. "There's a bee on your arm. Hey, there's a BEE right there on your arm!" It's basic communication. If a three-year-old said something exactly the same way twice in a row, we'd start getting suspicious that she's been replaced by robots or aliens. Why then is this practice considered OK in the production of recorded music? It insults the intelligence of the listeners.......

-----

Regeneration and inserts are of course useful editing techniques to fix small blemishes in a recorded performance. I've used them, and seen them used, in several recordings I've worked on. The goal is to make an edit completely transparent...it's like typing a paragraph and then going back to fix a typographical error or two, without changing the paragraph. The integrity of the passage is paramount; things start to sound mechanical and lose their freshness if too much editing is done. (In one of mine, I had played a piece with all its repeats and the whole performance was fine, except for a slight smudge of two notes in a fast passage. So I regenerated as few notes as possible--less than one second of music within a phrase--from one of the repeats in the same performance, and dropped them in. The notes go by so fast that it's I think impossible to know which ones are copies, and I'm not telling which ones they are. This, to my mind, is much different from copying whole bars or sections as repeats.)

How much time and money does one want to spend in the editing studio fussing with details, before deciding the recording is good enough to go out the door? Those are artistic and practical questions. The musician and the producer should remember that some of the people out there listening to the recording will also be producers and musicians, and will notice any work that is either too-fussy or too-sloppy. When the musical expression starts to sound too regular, or when the character changes too abruptly during a performance (revealing an obvious edit), it gives the impression that the music has been replaced by robots or aliens. :) When that happens, it might as well not be played at all...nothing meaningful is being communicated through it.

"You can fool all the people some of the time, and some of the people all the time, but you cannot fool all the people all the time." -
Abraham Lincoln

Thomas Braatz wrote (March 1, 2002):
< Bradley Lehmann comments: Little bits of cloth stitched together, getting all the notes accurately. Then that whole collection of sections is copied exactly for the repeat. All it does here in musical effect is make the music twice as long without giving us any new perspective on it. It's insulting." >
I know that you are referring to actual performances or recordings in a recording studio here, but it is interesting to consider from an entirely different vantage point that fact that Bach constructed his major choral movements in this very way, thus making it possible for him to economize on the effort expended on the actual task of composition. I have demonstrated this in a number of cantata mvts.

Recently, however, on a thread concerning the structure of opening mvts. of the orchestral suites, just the opposite occurs in that Bach deliberately changes the A' repeated section (after the fugal 'B' section). After reexamining and comparing the 'A' sections, I was amazed how extensive these changes were. To a listener (and I have been listening to them without a score for years) it appears that Bach is simply repeating the first 'A' section, but the score shows numerous, considerable changes in the repeated section. Perhaps when given the opportunity to do so, Bach preferred this type of treatment of repeated sections.

Comparable features occur in the opening mvts. of the cantatas. If Bach has a fugue beginning with the bass voice and moving up step by step to the soprano or even a trumpet, he may repeat the same subject, but reverse the order of the entries later in the same mvt. If the alto and tenor voices have a short duet followed by a similar pairing of soprano and bass, he will reverse the pairing when the section is repeated. The Stollen of the chorale in the opening mvt. of a chorale cantata is a mandatory repeated section in which Bach is forced to use the same music with changing words. Here the musical repeat suffers because very often the words are no longer directly connected with the music (word painting, etc.) and any special emphasis of the these new words in the repeated section or an attempt to change the interpretation of the music the second time around would only serve to underline the disparity between the words and the music - a dilemma of sorts that Bach had to live with.

Mario Zama Escalante wrote (March 1, 2002):
[To Tjako van Schie] I think repetitions are intended, as I have considered, to allow dance partners to retake a passage with another dance partner.

Juozas Rimas wrote (March 1, 2002):
[To Mario Zama Escalante] Probably, but the dances in Bach's suites aren't dances anymore :)

Bradley Lehman wrote (March 2, 2002):
[To Thomas Braatz] An interesting case of repeats comes up in one of the trio sonatas for organ. Bach lifted the slow movement from the D minor trio BWV 527 and rearranged it as the slow movement of the Triple Concerto, BWV 1044.

The organ version simply has the three lines, and repeat signs. In the concerto movement, Bach has four voices to work with: flute, violin, harpsichord right hand, and harpsichord left hand. He composes a new fourth voice (the pizzicato accompaniment). And the repeated sections are fully written out, with the flute and violin exchanging parts on these repeats! The repeats are thereby given a different tone color.

Now, should one do this also when playing the organ version? Change registration on the repeats? Or simply consider the arrangement a later thought about the composition (since there's a new part), and play the organ version for itself as if the concerto version doesn't exist? There are no provable answers here; the performer can choose whatever his/her artistic sensibility suggests in making the piece convincing. The situation suggeststhat Bach the arranger, revisiting his own music, thought it would be lovely to vary the tone color here in the context of the concerto.

Does the arrangement let us extrapolate back: (1) the repeats MUST be played? (2) they should be varied somehow, for the listener's interest? I suspect so...it could be as simple as playing the hands on opposite manuals, without changing any stops. Meanwhile, Bach doesn't tell the organist what to do, explicitly; he merely expects the organist to do something intelligent, beautiful, and musically convincing.

Take another example: Anton von Webern's arrangement of the six-voiced ricercar of the Musical Offering. Six lines that can all be played on one keyboard by one player; now for large orchestra, and the lines being swapped to different instruments _every few notes!_ Does this tell us anything about how a keyboard player should play the music? Does it matter that the arranger was someone other than Bach himself?

Bradley Lehman wrote (March 6, 2002):
< Riccardo Nughes wrote: A "damaged" recording is the Leonhardt 's Seon French Suites, where the absence of repeats damaged the final result. For major labels it's always so "hard" sell 2 cds for the price of 1..>
That Leonhardt set of French Suites, his English Suites (EMI Reflexe, then reissued on Virgin), and Partitas (ditto) were each issued as 2-LP sets. The English Suites and Partitas are even more damaged than the French Suites by that commercial "necessity" of getting by without a third disc. Leonhardt's first set of the Partitas (on dhm) was on 3 LPs.

The French Suites set by Thurston Dart on clavichord is another example: all six suites had to fit onto ONE LP! The playing is lovely, but again I really miss the repeats.

Pablo Fagoaga wrote (March 6, 2002):
[To Bradley Lehman] It is a sad paradox that an artist devoted to the HIP concept as Leonhardt is, has to drop out repeats just because of a commercial, economical reason. On one hand, there is a deep study of the scores, with impressive research, even with subtleties that people without formal musical education like me don't even notice.... and suddenly, a decision as noticeable as the skip of repeats (no more, no less!!!) is guided by a completely non-musical reason. Does this kind recording remain HIP after such a restriction?? I don't think so. It's a sad thing, given the outstanding artistry of Leonhardt among others, of course.

Thomas Boyce wrote (March 6, 2002):
Bach Recordings:
"Warning: Does NOT include repeats"

Mario Zama Escalante wrote (March 6, 2002):
[To Riccardo Nughes] You are right. I shouldn't have posted the OS as an example of tempi, but, aside of repetitions, I find some passages a little bit faster.

Mario Zama Escalante wrote (March 7, 2002):
[To Thomas Boyce] Good idea Tom.

Re-registration in repeats: an arrangement example from Bach himself

Bradley Lehman wrote (January 2, 2004):
Neil Halliday wrote: < (...) I don't like this 'damped' registration on a harpsichord, which removes its main asset, namely, a 'rich' timbre - and the keyboard score simply played without the already highly decorative arpeggiated style of the first exposition. Combined with the interesting pizzicato of the cello, this more restrained approach is how I would look for contrast in the repeats, for a quieter, more introspective version (in the repeats), etc etc.)

Did Bach have anything to say about major variation of the keyboard thorough-bass part in repeats (as demonstrated by Brad in this example)? >
I have here an example that is not about the improvised realization per se, but about the re-registration and the use of pizzicato as a contrasting sound on repeats, by Bach himself.

Listen to the middle movement of the [organ] trio sonata BWV 527; and then to Bach's own rearrangement of it (for flute, violin, and harpsichord) as the middle movement of concerto BWV 1044. Indeed, he gives the violin (and even the flute!) that pizzicato touch alternately on repeats, completely re-registering the music by swapping parts around, and--in my opinion--it is a winning, charming sound.

My thinking on this is: if Bach himself was this free with his arrangement technique, even to adding a fourth part as here, why would he have any objections elsewhere to musical decisions made with similar intentions of sounding good?

This example suggests to me that, if any single approach to performance is "wrong," it would be one where the same music is played exactly the same way twice in succession. That's basic musicality, as illustrated here by Bach. It's also basic human communication. When saying the same statement to someone, twice in succession, there is always some difference of inflection (at least, and maybe also changing some of the words, or amplifying the point, or leaving something out)...to give a different perspective on the material, bring out different features of it, give a better chance at communicating. Indeed, in normal human communication, the second time 'round of saying something is most frequently more emphatic than the first time, not a more subdued or introspective echo.

Think about it this way. Why should music, a language, not follow that same pattern of being differently inflected, and generally more inflected, on a repeat? Think about it this way.

=====

And in the Art of Fugue, a pair of his rearrangements is much more extensive than this: recasting a strict three-voiced fugue as a four-voiced (and occasionally more!) texture where the parts are broken up "cubistically" and passed all round, to both hands of two keyboard players and terrific stereo effect, along with all manner of free material given to whatever hand is unoccupied at any moment? That is what he does to both the three-voiced mirror fugues there. The strictness of the fugue, the integrity of each voice and consistency of rhythms, is cast to the wind in favor of something much wilder! And notably, these two end up not being strict mirrors of one another anymore; but that seems not to have bothered Bach here.

Why would he do this, if we are not welcome also to do likewise in a similar spirit of adventure, and good taste? Indeed I have been asking that question for years; see my review of this recording: http://www.amazon.com/exec/obidos/tg/detail/-/B00000083B

=====

And yes, I checked the authenticity of the 1044 arrangement, so no cynic can automatically jump on it trying to negate everything I'm saying here, and "prove" that Bach never did this to his own work. According to Alfred Duerr and his colleagues in the 1998 edition of BWV, the concerto arrangement 1044 is secure. "Echtheitszweifel an der Bearbeitung sind wohl unbegruendet." (There remains no seriously grounded doubt of the arrangement's probable authenticity by Bach, and they cite especially a 1978 article by Kilian, who later also did the NBA's editorial work on this concerto.) The only outstanding dispute, as noted here, is whether the ensemble version or the organ version came first, citing a 1970 article by Eppstein.

Either way, they're both by Bach. And perhaps the ensemble version has some things to suggest to thoughtful organists about registration, whichever version came first.