Recently I have been thinking about situations in which something in a piece of music happens more than once. This encompasses out-and-out repetition—which may be written out or may be indicated by repeat signs—true da capo, the structural element known in some circumstances as “recapitulation,” recurring sections in rondo forms, and any recurrence of a passage identically—even when it is not da capo. This also includes the ubiquitous practice of using recurring motifs—fugue subjects, other motifs treated contrapuntally, any sort of leitmotiv, the repeating bass line of a chaconne or passacaglia, and so on. I want to share some of these thoughts in this column and the next.
These columns will be filled with questions and speculation about a number of different aspects of repetition and how we come to expect it. I will also offer a few practical thoughts about what the concept of repetition and recurrence means for teaching, learning music, and performance. The value of this speculation for teaching lies mostly in the possibility that students may find it interesting, and perhaps it will lead them to further exploration. I also have thoughts about ways in which focusing on repetition and recurrence can help with the practicality of the learning process.
How many times shall we repeat?
Let’s begin with a curious example of repetition. Namely, why did Bach write the first Prelude of the Well-Tempered Clavier like the passage in Example 1 rather than like in Example 2 or Example 3?
In Bach’s composition, the first half of each measure is repeated exactly to form the second half of that measure. (It is so fully exact that he could have used repeat signs.) This pattern persists until the ending, where it is altered to lead to a satisfying cadence. We are deeply accustomed to the piece that is created by this patterning: it is one of the most familiar in the repertoire. But a piece on either of the other two models would have been a perfectly valid musical entity as well. If he had written it one of those ways, we would be deeply used to that.
Or would we? Maybe the “real” version is somehow actually better. I have played through the entire piece with both of these patterns, and it is my feeling that they work. Each one suggests something different about performance. The most striking of these differences is that the three-times-per-measure version feels to me like it should go a lot faster than the other two. It ends up reminding me of this passage from the longer Praeludium in E Minor of Bruhns, at measure 95 (see Example 4). If Bach had written version 3, we might speculate that it was influenced by this passage, which he may have known.
I tend to play the once-per-measure version slower and with more rhythmic freedom than either of the others. It is interesting that a mainstream analysis of the piece, which would be an analysis of harmony, since it is such a pure chord-based piece, would be essentially identical for each of these three very different pieces. I doubt that we can answer the question of why Bach wrote it one way rather than another, or indeed whether one way is better. It seems pretty clear that one repetition is very different both from no repetition and from two or more repetitions. Would writing the chord pattern four or five times in a row be as different from three as three is from two? At what point would it become ridiculous? If each chord pattern were repeated sixteen times, it would be a particular kind of extreme statement: outside the realm of Bachian music or most of what we ever encounter, but valid nonetheless.
So what about repeats as such? We accept it as normal that in many pieces of music, something—a definable passage that has a pretty clear beginning and an end, and not typically just a half-measure—will happen twice in a row. This is a defining trait of late Renaissance and Baroque dance forms: which usually consist of two sections, sometimes more, each repeated, like A-A-B-B. We accept that as routine, but, as with the Bach prelude, we would never expect each such section to be repeated more than once: A-A-A-A-etc.-B-B-B-B-etc. Same question, here applied to bigger increments of music: why not? I mean, why not in both directions? Why is twice in a row effective, and why would more than that not be? The same question applies to other places in common musical structures where repeats are routine, such as the first section (and sometimes other sections) of works in sonata-allegro form.
Tension and release
The most interesting effect and purpose of a repeat does not arise from or during the repeat itself, but rather at the moment when the repeat does not happen that second time. That is, the repeat sets up a tension (about whether it is indeed going to happen yet again, and whether it will end), and the motion to the next section relieves that tension. This tension is a fiction that we allow ourselves to experience, or that the music allows us to experience. After all, we know that the repeated material will not go on forever. We know this for certain if we know the piece, and we know it essentially for certain even if we do not. But there is something in listening to music that allows our reaction—our appreciation—to feel surprise even when listening to something that we know intimately.
I strongly suspect that this dynamic is one of the explanations for the persistent tendency for all of us to feel that “taking the repeats” is a different matter for the first section of a piece in binary form than for the second section. When we come to the end of the second section, first time, the suspense-and-relief scenario has just happened once, and our appetite for it is perhaps satisfied. Also, we know that when the second section gives way, whether after a repeat or not, it will give way to the silence following a cadence, and then perhaps another movement or another piece. That is a less compelling change of direction. This may also tend to explain the almost universal practice of not taking repeats in da capos of the sort represented by the return of the Aria at the end of the Goldberg Variations of Bach.
This may also tend to explain why repeats of sections happen only once (AABB). If part of the point is to set up the moment at which the repeat gives way to something new, it is important for the repeat not to wear out the listener’s interest. Once makes the point, more than once risks irrelevance.
I once heard an experienced concert pianist, playing from memory, repeat the opening section of the first movement of Beethoven’s Waldstein Sonata twice (that is, play the section three times). My impression was that he accidentally started the first ending after the second time through, and that led inevitably to replaying the whole section. I remember having the following reactions: first of all, that it was nice to hear the passage an extra time, just because it is wonderful music; but second to feel a little bit of panic that it would never end!
This experience makes me ponder, how does our enjoyment of a particular passage play a role in interpreting or perceiving repeats? Does this apply in particular to pieces that are short enough that repeats will not try anyone’s patience? In the Bach variation set on the chorale O Gott du frommer Gott the composer follows the structure of the chorale melody as far as repeats are concerned, repeating the first half of each variation, but not the second, at least in most variations. In Variation V and Variation VII he does indeed repeat the second half as well, violating the relationship between the chorale melody and the shape of the variation. Why? Well, I noticed practicing the piece that the opening measures of each of those second halves was especially charismatic, unusual in texture, and harmonically rich. These are moments that I would find myself practicing over and over again just because they are so cool to play and hear. I had to remind myself to practice other bits that needed practicing more. I noticed this before I stopped to think about those repeats. But I wonder whether Bach put in those extra, musicologically “wrong,” repeats just because he liked that bit of music! This is a kind of non-rigorous, taste-driven choice making that we do not associate with JSB. But perhaps we are wrong not to do so.
Do repeats generally have to do with symmetry or logic? I think that we often assume that they do or that they should. If the allemande of a given suite is A-A-B-B, then presumably the courante and the other movements will also be. Otherwise there would be an imbalance. When playing the above-mentioned Bach variations, I have been aware of a pull either to omit the extra repeats in the two variations that have them or to add repeats to all of the “B” sections. And this would be in apparent direct contravention of the wishes of the most august of composers. I once had a student who was working on the Goldberg Variations, and who proposed to omit almost all of the repeats, but to take one pair of repeats, namely in Variation VII. He had an interesting registration idea for that movement, which required four rather than two sections. I have to admit that this bothered me: no logic or symmetry, just going with an aesthetic choice. It is just an interesting light on what I (we?) sometimes want out of repeats.
To play the repeat or not, that is the question.
This brings us to something pretty concrete and specific: “should I take the repeats?” Essentially we never find ourselves wanting to omit any part of a piece that is not a repeat. I have never had a student ask me about a through-composed piece, “should I or should I not play measures 9–16?” or anything like that. Even though we accept repeats, we clearly do not quite accept them. That is, we do not accord the notes indicated by repeat signs absolutely all of the status that we give to other notes. We give this higher status to passages that are a lot like, or exactly like, earlier passages, including actual da capos. Only when a few measures of material are repeated right away, and with such exactness that it can be notated by a repeat sign, do we consider it an issue whether to play those notes or not. In fact, we likely think that way only if the repeat is actually notated by a repeat sign. Once in a while a repeated passage, a literal, full repeat, is written out. In such a case I do not often hear a student, or anyone, suggesting cutting out the similar bits.
For an interesting side note, in his notes for his 1950s recording of Scarlatti sonatas for Columbia, Ralph Kirkpatrick said of his not taking the repeats in the pieces, almost all of them in binary form with repeats indicated, that if listeners wanted to hear them again, they could play them again on the record! This probably indicates as much about his skeptical attitude towards recording as it does anything about his approach to repeats.
As for myself, I recognize a strong tendency to want to take all repeats, in whatever way they are indicated by the composer. I think that this has to do in part with my having spent my life absorbing the ethos of “the composer is always right.” That ethos has grown pretty steadily over the last couple of centuries and has found one expression in certain aspects of the early music movement. And I believe there is a certain logic to that. If a composer put in repeats, why shouldn’t it just be routine to play them? I have this stubborn feeling that there really is not any reason to single those notes out for omission. I also recognize that this could be a different story for pieces that were written in a style that made repeats a matter of routine—early dance movements, say—than for pieces written with no assumptions about repeats—nineteenth-century sonatas, for example—as to which composers made choices about repeats on a custom basis.
I also notice that, of course, repeats make a piece longer. That may seem trivially obvious, but the magnitude of a piece in time is a valid part of its aesthetic. The difference in length between a Haydn or Mozart symphony and a Mahler or Bruckner symphony is of course not even close to the whole difference. But it is not meaningless or insignificant either. If you take no repeats in one of the Bach French suites, it becomes a short piece. There is at least one recording of all six of those pieces, by Thurston Dart from 1961, that managed to fit the whole collection onto one LP. Independent of anything about the playing or the instrument (clavichord) or the recorded sound, that presentation of the whole collection as being that short feels different from a recording twice as long that would have had to occupy at least two LPs. (I say at least because it was a rather long LP.)
Making a piece long by repeating each of its parts is not the same thing as making a piece long by composing a longer amount of new material. And one thread over the long arc of western classical music has been to look for ways to create bigger structures without literal repetition. But in making a choice about whether to take repeats or not, the effect on overall length is part of the equation.
In keeping with some of my thoughts above, I do suspect that when I hear a performance of a piece that features repeats as part of the structure, but are ignored, I usually feel disappointed by the time the second section appears. If the suspense created by the repetition has not been allowed to build, then the resolution of that suspense through moving on to something new cannot have the power that it was meant to have.
To be continued.