Wednesday, September 8, 2010

goldberg variations, ct'd

Okay, enough "Handel-lambasting" (as Brett accused me of in his last comment). Let's do some Bach-lambasting instead.

Wha??? But I thought GBV was, like, the greatest? Well, it is....but Bach does some nasty things in there that just make it really hard to pull off, on harpsichord or piano. Especially on piano....but especially on harpsichord. One of the difficulties in performing, listening to, or understanding GBV is voice-crossing.

When two separates lines of music "cross" pitch paths, we call it a voice-crossing; one starts out above the other in pitch, but ends up below. If one person started singing a scale up from below, and another person down from above, there would be a voice-crossing somewhere in the middle.

Voice-crossings are, in general, a big no-no in counterpoint. Why? In counterpoint--or music or two or more distinct lines--the ear, or more precisely, the brain, has to keep track of more than one "object" at a time, while also fusing them into one coherent whole. But since the brain can't attend to more than one object at any specific moment, it goes back and forth between objects quickly (or slowly, or not at all, in which case you're not a good listener). In order to parse two lines of music perceptually into two different objects (or "streams"), the brain needs a way of distinguishing them. One of the easiest ways is pitch level or fundamental frequency; in our natural environment, streams of sound separated by large gaps in frequency usually come from different sources. If one line of music is consistently above another, then the brain reckons, it comes from one source, while the other line comes from another. Perceptual separation achieved!

But what if one line of music isn't consistently above another, and they cross once, twice, or over and over? In the case of different singers or instruments, we can also segment by tone or timbre. And there are also other, more subtle ways of parsing counterpoint, by the consistency of note lengths, articulations and volume. But on one instrument, parsing voices accurately is much more difficult because the timbral cues are gone.

Unfortunately, Goldberg Variations, unlike most of Bach's keyboard music, is full of 'em! They come in two distinct types: within-hand and between-hands, and the two types present different challenges for the pianist as well as the listener.

Above I've pasted an example of the first type (second half of variation 6), which is easily, and I suspect, nearly universally, misheard, when played on the piano or harpsichord (yeah I can't get the image down here for whatever can listen to it on piano here [variation 6 starts at 7:52 and the part I pasted at 8:17]). Starting in the 8th measure of the image, the voices cross back and forth, with each voice alternating between sixteenth notes and a dotted quarter held through the bar. But it's easier to hear a sequence where one voice carries a rising melody on top with an alto voice playing sixteenth notes underneath. Now listen to and watch this (variation 6 starts at 4:10, the image at 4:42). The selection is much easier to perceive correctly because the violin and viola, which have distinct tones, carry the two lines separately. In addition, watching the separate instruments bow correctly adds visual cues which the brain can integrate with auditory information to stream the two lines separately. (Visual evidence will turn out to be of critical importance even for performance on the piano when we get to between-hands crossings and issues of "cheating.")

If Glenn Gould doesn't voice this section "correctly," what hope do I have???! On the one hand, maybe a "correct" voicing--ie, one that induces accurate contrapuntal perception in the listener--isn't especially necessary or desirable. But maybe it is. I've got some ideas. One thing that for some reason I've never heard any pianist do, and seems relatively simple, is to articulate the two voices differently, or play one of them more loudly. More on that later. This little section is just the tip of the iceberg of course....lots more to consider!


  1. I guess this means you've come around to the string trio version? It's at least interesting, right?

    (But especially Bart)

  2. another reason the two-manual piano is the ticket!! you could use different "stops" so that the two voices were distinctive, no?
    and goodness, why does GG sit so-o-o-o low on that piano stool?!

  3. @sarah: not totally sold yet, but it's definitely interesting!
    @dad: no "stops" on the two-manual piano, although this article, which is the best info I found, was a little unclear about the differences between the two manuals:

    does the top manual really always play each note in two octaves? i guess so. which, by the way, is really weird.

    also, the two manuals only helps with "between-hand" voice crossings, either visually or timbre-wise, because "within-hand" voices have to be played with the same hand on the same keyboard, as in variation 6 mentioned above in the post. for the same reason, harpsichord is no help for those variations; in fact, it's much HARDER to distinguish those voices on a harpsichord cause you can't vary the dynamics at all.

    interestingly, the taylor guy who's been playing the crazy piano notes the advantage of the visual effect (2nd paragraph from bottom), without mentioning the octaves thing, which would also be a powerful mechanism for parsing the two hands as well.