Voice-leading Anomalies in J.S. Bach's Mass in B Minor, BWV 232

September 14, 2016

I have owned the score of Bach's B-Minor Mass (BWV 232) since I was 18, and have studied it from time to time along with recordings, sometimes also during live performances. For decades it has remained for me one of the most inspiring of Bach's works. Two days ago, I had the opportunity to attend a performance of the masterpiece by Collegium Vocale Gent, conducted by Philippe Herreweghe, at the Jahrhunderthalle in Bochum, Germany (my new home since July 21, 2016). Unfortunately, I missed this opportunity, because I missed the bus which I had planned to take to the concert.

I was faced with a choice: sulk all evening, or open the rehearsal score of this work and read through it with a fine-toothed comb at the piano. I chose the latter. In so doing, I noticed several interesting things in this marvelous music that somehow had escaped my attention for decades.

Music students are often taught that Bach never wrote parallel octaves or parallel fifths, but that isn't exactly true. If you look carefully at Bach's music for large ensembles, you can find these voice-leading anomalies. They are few and far between, but they do exist.

All music students are taught to avoid writing parallel 5ths and parallel octaves. The reason is that voices moving together harmonically by these intervals tend to fuse into one sound (as in organ registration), and unless the voices are moving consistently together in parallel as a technique of orchestration, the occasional parallel octave or fifth between two parts will leap out of the texture and sound like a mistake.

It is true that you won't find those kinds of errors in Bach's chorales (through there are rare instances similar to those I'll show you here), but the fact is that Bach did occasionally write parallel octaves and parallel fifths in his music scored for large groups. I'm not talking about octave doublings in terms of orchestration. I am actually talking about voice-leading instances where independent parts all of a sudden move from only one note to the next in forbidden parallel motion. Bach did write such parallel octaves and parallel fifths occasionally. And guess what — they do not sound wrong. On the contrary, they sound just fine. Why? Because he knew what he was doing. In almost every case, I wouldn't characterize the parallel 5ths or octaves that Bach wrote as "mistakes"; hence I have titled this post "Voiceleading Anomalies" rather than "Voiceleading Errors". This is a huge topic, for sure, and probably someone has written a book about it already, but I have never seen it. So let's take a look at these examples, from the Mass in B-Minor BWV 232, one of Bach's most famous compositions.

Christe eleison

The first anomaly appears in the Christe, a duet for two sopranos with violin and continuo. From the last beat of bar 15 to the downbeat of bar 16, there is a parallel octave between the violin and the continuo.

Bach could have avoided the parallel octave by changing the A to a C# in the violin, but he didn't. Why? Apparently he wanted it that way. In any case, there is a simple reason this doesn't sound wrong, and that it needn't be considered a mistake. The reason is this: the eighth notes in the continuo (the bass line) are to be played non-legato, or detached. Anyone interested in early music knows this; it is part of the style. The result in this case is that the space between the written notes provides room for the violin to move against an unwritten rest in the bass. In other words, what looks like a parallel octave on the page is not actually played that way. What we hear instead is the B in the violin against the A in the bass, and then the B again with the bass – in effect oblique motion.

Laudamus te

Another parallel octave occurs at bar 56 of the Laudamus te (from the Gloria), an aria for Soprano with continuo, a trio of strings, and violin solo.

This instance may be a bit harder to see, and is also shorter in duration than the previous example. This is usually the kind of parallel octave or 5th that appears in Bach's writing. The parallel is a result of an ornamental note, an upper neighbor, in a figure that appears as a motive in the piece. The actual voice-leading between these two voices is tricky. On the downbeat, the upper part has a D, and so does the continuo, so that overall there really is structural voice-leading of parallel octaves here. On the other hand, the upper part (violin) leaps away from the D down to a B and moves up from B to C#, against the continuo (bass) which is moving down from D to C#, so structurally there is also contrary motion. Again, there is space between the D and C# in the bass, so that the upper part is actually moving against a rest. In some ways there is more of a parallel octave here than in the previous example, but again it doesn't sound wrong, because it's hidden, obscured — all but non-existent, so that nothing leaps out of the texture.

et resurrexit

The et resurrexit, (from the Credo) is scored for 5 voices with continuo, strings, woodwinds, and brass. Here we find the first example of a parallel fifth, between bars 53 and 54.

The upper voice in play here against the continuo is the second oboe. In this case the parallel is also an instance of a motivic lower neighbor tone, moving against an unwritten rest in the bass. By now you understand how this works.

pleni sunt coeli

The pleni sunt coeli, from the Sanctus, is scored for 6 voices with continuo, brass and timpani, 3 oboes and strings. A parallel fifth appears at bar 116 between the second violin and the continuo.

This example is more problematic, and one might be tempted to call it an error, because both the violin and the continuo are moving in 16th notes. In this case, the effect of the parallel is minimized since it occurs in an inner part. The outer (soprano) voices are moving in contrary motion at that point, and several voices hold a long note in oblique motion to the bass. Attention is drawn to these other parts, and the parallel basically vanishes within an active texture.

A second parallel fifth occurs later in the same movement in bar 134, this time including the soprano.

If we look at what is happening in the bass instruments here, we find that the upper parts complete two first-inversion triads: Am/C and G/B. The image below makes this clearer.

Parallel first-inversion triads sound good, regardless of what voicing is used. To avoid parallel fifths, the voices are supposed to be written up from the bottom in the order: 3rd, 5th, root, so that the upper interval is a fourth instead of a fifth. Here we see that Bach did it the other way, in the order up from the bottom: 3rd, root, 5th. And it sounds fine. In my own music from time to time I've also done this. In fact, this one special use-case of parallel fifths should probably be exonerated from the list of forbidden voice-leading, because it almost always sounds perfectly fine. But, not always. If you're a student, don't do it.

et resurrexit – a jazzy pre-dominant chord

The last example I found is not a parallel 5th or octave, but an instance (in et resurrexit) of a harmony that includes the 3rd and the 4th of a chord together. I've mentioned before that Bach does this quite often using suspensions. This is a more unusual case.

The harmony here without brass and timpani would be a second inversion tonic with the major 7th in the key of A, which is not unusual as a pre-dominant function (although the major 7th is spicier than normal). With the brass and timpani added, the 3rd and 4th of the harmony sound together. It's normal in a jazz style to voice chords with both the 3rd and 4th together, so this harmony becomes like a jazz chord. It sounds good because of the way Bach has orchestrated it. The D in the timpani (written C) is there for its weight more than its pitch. The D in the brass (also written C) is a repeated note which rings out very clearly on the valveless baroque trumpet in D. It may be an obvious point, but Bach did not write chords like this in his keyboard music. There are certainly examples of unusual dissonances to be found in the keyboard music, but as I hope you can see from these examples, Bach wrote for ensembles quite differently than he wrote for keyboard.

Conclusion

In a score that I thought I knew well already, I was delighted to find these things. In fact, as much as I love this music, I had never played through the entire work at the keyboard before. It's amazing the things one can discover in the space of a few hours just by looking into a piece of music. I highly recommend it as a way to spend an evening or two, or three. The piece has become new to me again. The main thing to take away from all this might be the same observation we started with, that Bach really knew what he was doing (big surprise).

Has this made you curious to find more anomalies in Bach's music? There are many other interesting passages in the passions and cantatas from which to learn.

Happy searching!
Aaron