A Majestic Trill from Schumann

Take a look at that bass trill in measure 6 of the Sempre marcatissimo movement of Schumann’s Symphonic Etudes, op.13:

This movement is confusingly subtitled “Etude VIII (Variation VII).”

This trill seems rather benign on the page. Schumann draws no special attention to it among the several other trills in the surrounding measures. In fact, the hairpin decrescendo just above the trill would seem to relegate it more to the background. And this trill is usually treated rather benignly in performance as well, even by the greatest of the greats like Géza Anda, Alfred Brendel, Alfred Cortot, Evgeny Kissin, Ivo Pogorelich, Maurizio Pollini, Sviatoslav Richter, Arthur Rubinstein, András Schiff, Rudolf Serkin, and Daniil Trifonov. (Links to audio excerpts.)

None of these obedient interpretations of this trill captivate me. (BTW, not one of these performances is obedient to Schumann’s brisk tempo marking. Rubinstein comes closest but is still a good clip slower.) But thankfully there is another interpretation, one that is less obedient to Schumann’s score. And it’s majestic and spellbinding. (It’s very likely that I overstate things here, but I often get caught up in details like this to the point where their effect seems to grow beyond reasonable proportion. You probably know what I mean.)

In order to get the full effect of this trill execution, you must listen from the beginning of the etude since Schumann sets it up so beautifully by the diatonicism of measures 1-4, by the replacement of the bass’s B# (Ti) downbeat in m.3 with B-natural (Te) in m.6, and even more so by the replacement of the soprano’s D# (Re) at the beginning of m.2 with the D-natural (Ra) trill in m.6 (which the bass imitates), thus preparing the tonicization of the submediant (VI) that is fully achieved by the bass trill and resolution. (By the way, that m.6 D-natural is just a perfect note, isn’t it?!)

And now listen to Mikhail Pletnev’s wonderful and disobedient performance. (The final portion of the score is below.) AUDIO

If you want to hear just the trill again, click here.

While we’re at it, listen to Pletnev’s performance of Etude XI/Variation IX. I can’t imagine a more perfect performance.

That tonicization of the Neapolitan! That climax! And the way in which he pulls back dynamically at that “arrival IV” (2:17 mark)!  (From what I can tell, he combines the two versions of this etude. He plays the two-bar introduction that appears only in the first version but takes the repeat that only appears in the second, which is shown below.)

 

“Correcting” Bach’s Parallel Fifths: Breitkopf Editorial Practices

In an earlier post I examined 20 instances of consecutive perfect fifths and octaves in the Bach chorales (not counting 26 “fermata” consecutives occurring between the final chord of one phrase and the first chord of the next).

Since writing that blog post, I discovered something else about these consecutive fifths and octaves as they appear in the Breitkopf edition of Bach chorales which was posthumously published in the 1780s. The discovery reflects on the editorial practices of those involved in putting that edition together (namely, C.P.E. Bach and, to a lesser extent, Friedrich Wilhelm Marpurg) and it has significant consequences for analysis of the Bach chorales in general. I’ll explain all this, but first, a word about the background of the Breitkopf edition (which served as the basis for the ever-popular Riemenschneider edition) is necessary. If you know all about this edition, feel free to skip the bracketed passage below.

[After J.S. Bach’s death in 1750, there was a growing interest among his family and his admirers to both preserve (and in some respects even improve) his legacy as a composer and to present his works to a broader public in a manner that facilitates their study. Attention turned to the chorales in particular given their primacy as models of part-writing. The first efforts to produce a collection of Bach’s chorales were undertaken by the publisher Friedrich Wilhelm Birnstiel who engaged Friedrich Wilhelm Marpurg as chief editor of the project. However, soon after embarking on Volume 1 of the projected three-volume collection (each volume containing 100 chorales), Marpurg secured a more desirable position that required him to step away from the project. Birnstiel was forced to approach C.P.E. Bach, who demanded a fee that tripled the rate paid to Marpurg, to complete Volume 1, which was published in 1765. After C.P.E. Bach himself took a new position in Hamburg, Birnstiel turned to yet another editor, this time Johann Friedrich Agricola, a former student of Bach’s at the Thomasschule. Agricola, unfortunately for Birnstiel, made a mess of Volume 2, which was published in 1769 and sold very poorly, due in no small part to a brutal critique from C.P.E. Bach for the volume’s numerous errors. Birnstiel decided to cut his losses and abandon the project.

The hope of preserving Bach’s chorales in a single collection dimmed, that is until Johann Philipp Kirnberger stepped in to resurrect the project. Kirnberger went to great lengths to persuade the publisher Breitkopf to take on the project with C.P.E. Bach back at the editorial helm. Emmanuel’s collection of chorales by his father had grown to nearly 400 by then, and Breitkopf finally agreed to proceed on the condition that enough advance subscriptions were secured. The Breitkopf edition of “370” Bach chorales was finally published in four volumes between the years 1784 and 1787 (one volume per year). A more thorough account of the history of the Breitkopf edition is available here. The Breitkopf can be viewed in its original publication here, and is presented in modernized notation in volume III/2.2 of the authoritative Neue Bach Ausgabe (NBA).

This Breitkopf edition served as the basis for the Riemenschneider edition of Bach chorales, often referred to simply as “the 371”. (The discrepancy between 370 and 371 can be explained by the fact that two chorales in the Breitkopf, the final chorale in Volume 3 and the first chorale in Volume 4, received the same number, 283.) Riemenchneider, for reasons unknown to me, rearranged a few of the chorales as they appeared in the Breitkopf (as shown in the sortable table of chorales here), and, for all its faults as an edition (don’t get me started!), also made corrections based on the original Bach manuscripts.]

What is important to remember about the Breitkopf edition published in the 1780s is that the individual chorales BWV 253-438 have survived by way of this collection (although about a fourth of these chorales also appeared in the important Dietel collection from around 1735, a collection to be discussed more later). That’s important for this reason: while any editorial revisions done during the preparation of the Breitkopf edition can be checked against the original manuscripts of Bach’s larger choral works (the cantatas, passions, motets, etc. – i.e. BWV 1-252 works), they cannot be checked against BWV 253-438 chorales since there are no original manuscripts.

So why is this fact important? It all relates to my discovery. I discovered that a number of the parallel fifths that occur in chorales for which we have original manuscripts were “corrected” (presumably) by the Breitkopf editors! By today’s standards, an editor taking such liberties would be unthinkable. But the late 18^th century was a different world. And we should remember that C.P.E. Bach took considerable measures to ensure that his father’s legacy was bolstered as much as possible, even if that required exaggeration, as is demonstrated in the great composer’s famous obituary (Nekrolog) written by C.P.E. and, ironically enough, Agricola.

List of “Corrections”

Below is the list of 20 instances of consecutive fifths or octaves as taken from my previous post, this time with Breitkopf (B) numbers included and a brief description of the kinds of consecutives involved. Highlighted are the ones we are concerned about – that is, chorales which 1) are in the Breitkopf and 2) come from larger works for which we have original manuscripts. (Decimals represent movements of larger works. “R” numbers refer to the chorales’ position in the Riemenschneider edition.)

7.7 (RX, BX), measure 2/6.1 – (Consecutive fifths in contr. motion in B. and T.)
22.5 (RX, BX) measure m.27-28 – (Consecutive octaves in contr. motion in B. and T.)
26.6 (R48, B48) measure 4.2 – CORRECTED (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)
33.6 (R13, B13) measure 14 – not corrected (Consecutive fifths in contr. motion in S. and T.)
40.8 (R8, B8) measure 2.2 – CORRECTED (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)
40.8 (R8, B8) measure 4.2 – CORRECTED (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)
40.8 (R8, B8) measure 6.2 – CORRECTED (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)
40.8 (R8, B8) measure 16.2 – CORRECTED (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)
48.7 (R266, B266) measure 14.2 – CORRECTED (Parallel fifths in A. and T. involving a p.t.)
86.6 (R4, B4) measure 13-14 – not corrected (Chordal parallel fifths in S. (Re-La descent) and T. (Sol-Re descent))
99.6 (RX, BX) measure 11 – (Parallel fifths in S. and A. created by lower n.t.)
146.8, (RX, BX) measure 10.3 – (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)
167.5 (RX, BX) measure 11/25-12/26 – (Parallel octaves(!) in T. and B.)
244.40 (R121, B121) measure 4.3 – CORRECTED (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)
248.33 (R139, B139) measure 2.2 – CORRECTED (Parallel fifths in S. and A. involving p.t.)
251 (R329, B328) measure 14.1 – not corrected (Consecutive octaves in contr. motion in S. and T.; S. involves voice-crossing w/ A.)
263 (R128, B128) measure 6.2 – (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)
308 (R27, B27) measure 9 – (Consecutive fifths in contr. motion in T. and B.)
323 (R320, B320) measure 8 – (Chordal parallel fifths in S. and T.)
361 (R264, B264) measure 12.2 – (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)

Eight of the eleven highlighted instances of consecutive fifths or octaves were “corrected” in one way or another. The specific ways in which each instance is corrected is shown below. But first, let’s examine the implications of this a little bit.

Observations and Implications

Given the fact that nearly three-fourths of the consecutive fifths and octaves appearing in these chorales were corrected, it would be only logical to assume that other instances among the BWV 253-438 chorales, for which we have no original manuscripts, were also corrected. Furthermore, the fact that 16 instances of consecutives appear in the roughly 225 chorales from BWV 1-252 while only four appear in the 186 chorales from BWV 253-438 leads one to suspect that many more instances were “corrected.” (The disproportionate number of corrections among the BWV 253-438 may have other explanations that would take far too long to get into here. If you really want to know about these possible explanations, ask me. Suffice it to say that I am sympathetic to the theory that many of the BWV 253-438 chorales came from a compilation of chorales Bach owned rather than coming from larger choral works (e.g. cantatas) that are now lost.)

Other questions emerged in my mind from this discovery:

• Are there patterns in the ways in which consecutives are corrected by Breitkopf editors that can be applied to BWV 253-438 chorales in the opposite direction? I’m guessing that our sample size is a bit too small for this.
• Though impossible to know, were the five chorales from extant cantatas listed above that do not appear in the Breitkopf (BWVs 7.7, 22.5, 99.6, 146.8, and 167.5) excluded from the collection because they contained consecutives? My guess is probably not, for a couple reasons. First, with the exception of the parallel octaves in BWV 167.5, the kinds of consecutives appearing in these chorales are no more egregious than those that were corrected and even no more egregious than those that were left uncorrected. Second, the fact is that more than sixty chorales were left out of the Breitkopf. There is little reason to think these five are exceptional.
• Is it possible that these “corrections” were made by J.S. Bach himself at some point later in life? I doubt this is the case as well. First, such an idea would suggest that these instances of parallels were “mistakes” that needed correcting, something that I cannot bring myself to believe, for a variety of reasons. Second, given the fact that C.P.E. Bach was so willing to take measures to enhance Bach’s legacy, reasonable suspicion falls on him.

The Corrections

26.6 (R48, B48) measure 4.2 – CORRECTED (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)
40.8 (R8, B8) measure 2.2 – CORRECTED (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)
40.8 (R8, B8) measure 4.2 – CORRECTED (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)
40.8 (R8, B8) measure 6.2 – CORRECTED (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)
40.8 (R8, B8) measure 16.2 – CORRECTED (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)
48.7 (R266, B266) measure 14.2 – CORRECTED (Parallel fifths in A. and T. involving a p.t.)
244.40 (R121, B121) measure 4.3 – CORRECTED (Re-Do anticipation over Sol-Fa motion in S. and T. at a PAC.)
248.33 (R139, B139) measure 2.2 – CORRECTED (Parallel fifths in S. and A. involving p.t.)

Six of the eight corrections involve what I called “cadential consecutives” in my original post. These are created by the simultaneous appearance of a Re-Do anticipation in the soprano over a delayed arrival of the seventh of a V7 (Sol-Fa) in an inner voice at an authentic cadence, by far the most common type of mid-phrase consecutives in Bach’s chorales. In one case, BWV 26.6 (Example 1), the Breitkopf editor staggers the parallel fifths by delaying the tenor’s Sol-Fa passing motion by a sixteenth – a miniscule revision, but a “correction” nonetheless. In one instance (m.2) from BWV 40.8 (Example 2), the editor removes the soprano’s Re-Do anticipation altogether and in the three other instances in this setting staggers the parallels by delaying the soprano’s Re-Do anticipation figure. Finally, in BWV 244.40 (Example 3), the editor removes the tenor’s Sol-Fa passing motion altogether.

EXAMPLE 1 (BWV 26.6, B/R 48)

EXAMPLE 2 (BWV 40.8, B/R 8)

EXAMPLE 3 (BWV 244.40, B/R 121)

The other two “corrections” occur in BWV 48.7 (Example 4) and BWV 248.33 (Example 5). In the former case, the alto part is simply rewritten: Bb-A is replaced with an A-G suspension figure. As a result, the parallel tenths between alto and bass (which I highlighted in my original post) is interrupted. The parallel ninths that are created by the editor’s alteration stand out from among the string of tenths as a kind of obstruction, partly due to the weak-to-strong repetition of the G into beat 2. More than the others, this example demonstrates well how, for Bach, a strong musical idea can override contrapuntal “correctness.”

EXAMPLE 4 (BWV 48.7, B/R 266)

In BWV 248.33, an editor has changed the alto’s E to F#, thereby eliminating the parallel fifths with the soprano. Harmonically, the F# works fine, converting a root position A minor chord (ii in the beginning key of G) into a first inversion F# diminished (viio6). And the alto’s F#-F chromatic descent might be thought to match that of the bass (A-Ab). Still, taken as an entire phrase, the alto’s F# impedes significantly on its directed upward motion – the line D-F#-F-G is certainly less compelling than D-E-F-G. For this and other reasons that are well stated by Fitsioris and Conklin (see pp.7-8), the parallel fifths Bach has written here offend not the ear and therefore need no correcting.

EXAMPLE 5 (BWV 248.33, B/R 139)

A Note on the Riemenschneider Edition

The reason these “corrections” eluded my attention until now is because these “corrections” have been eliminated (corrected the “corrections”!) in the Riemenschneider edition and in subsequent editions of the Breitkopf. So while the Riemenschneider edition is based on the original Breitkopf collection in terms of its organization, it does not duplicate its editorial revisions. (In his edition’s appendix, Riemenschneider discusses some of these updates.)

However, Riemenschneider could obviously only check the Breitkopf settings against the original manuscripts for those chorales which have survived in the original manuscripts. That excludes 185 chorales, nearly half of the entire collection. (And unfortunately, due to the frustratingly random organization of the Riemenschneider, one has no way of knowing which chorales are among these 185 chorales without searching the commentary in the appendix. Ugh.)

More Research: The Dietel Collection

In the brief description of the background of the Breitkopf edition given above, mention was made to the collection of 149 chorales created around the year 1735 by Ludwig Dietel, a student at the Thomasschule during Bach’s time in Leipzig. Scholars are fairly certain that Dietel copied these chorales directly from the original manuscripts of larger works. The importance of the Dietel Collection lies in this fact as well as in its early date – it predates the Breitkopf by 50 years! The unfortunate fact about the collection is that Dietel was not terribly accurate as a copyist, which resulted in numerous errors.

Despite these errors, one could easily argue that the Dietel Collection deserves priority over the Breitkopf A) given that the level of certainty that Dietel copied these directly from original manuscripts, and B) given that C. P. E. Bach took such editorial liberties with the settings. Because of this, a next logical step in this research project would be to compare the approximately fifty chorales from BWV 253-438 that appear in the Dietel with their appearance in the Breitkopf.

[11/5 Update:] I have found five instances of parallel fifths in the Dietel that have (presumably) been corrected by C.P.E. Bach:

Dietel Nr. 7 = Breitkopf Nr. 252 (R252) = BWV 362, measure 5/13, beats 2-3: the alto’s A-Bb has been changed to A-G in order to prevent parallel fifths with the tenor.

Dietel Nr. 23 = Breitkopf Nr. 274 (R274) = BWV 397, measure 18, beat 4: “Cadential parallels” are staggered rhythmically by delaying the tenor a sixteenth, in a manner precisely like several of the “corrected” cadential parallels mentioned above.

Dietel Nr. 57 = Breitkopf Nr. 275 (R275) = BWV 393, measure 5, beats 1-2 : the tenor’s downbeat A-B has been changed to C#-B to prevent parallel octaves with the soprano.

Dietel Nr. 104 = Breitkopf Nr. 349 (R350) = BWV 360, measure 12, beat 2: “Cadential parallels” are staggered rhythmically by delaying the soprano a sixteenth.

Dietel Nr. 109 = Breitkopf Nr. 36 (R36) = BWV 385, measure 6, beats 2-3: The bass’s running eighths have been changed from A3-G3-C#4-E#3 to A3-G3-F#3-E#3 to prevent loosely disguised parallel fifths with the tenor’s E-G# quarters.

A Final Note

After writing everything up to this point, a friend of mine came across a citation to a 1983 article by Gerd Wachowski entitled “Die vierstimmigen Choräle Johann Sebastian Bachs: Untersuchungen zu den Druckausgaben von 1765 bis 1932 und zur Frage der Authentizität” (“The Four-Voice Chorales of Johann Sebastian Bach: Studies on the Published Editions from 1765 to 1932 and the Question of Authenticity”) (Bach-Jahrbuch, 69 (1983), pp.51-79). In that article, Wachowski discusses some of the same Breitkopf editorial practices I discuss here. He points to a couple of very specific musical characteristics that frequently appear in the BWV 253-438 isolated chorales that do not frequently appear in chorales from BWV 1-248 works as evidence of editorial liberties. Included in his essay are the corrected “cadential fifths” I refer to above, though not to the corrected BWV 48 parallels, nor to the Dietel comparative research. He also gives a very thorough account of the nearly two dozen duplicate chorales appearing in the Breitkopf (duplicates which were also included in the Riemenschneider) as well as his own final commentary on the authenticity of the BWV 253-438 individual chorales.

3-Second New Music Quiz

Here’s a “3-second” quiz for you composers and new music enthusiasts out there. Identify fifteen 20th-century composers and works by listening to a 3-second excerpt. (Okay, some excerpts are more like 5 or 6 seconds, but who’s counting?) It’s indeed remarkable that some composers have so individual and unique a sound fingerprint that identification is possible in 3 seconds or less, even if the piece itself is an unfamiliar one. I suspect that some of these, then, will be fairly easy for those who are familiar with modern music. In fact, some excerpts come from iconic pieces that will be instantly recognizable. Others are more obscure and, therefore, much more difficult. It would make sense to assign more points to the more difficult ones, but for convenience sake, let’s just say 5 points for each composer, piece and movement (when applicable). 200 points are possible, though a score of 100 would itself be impressive. If you’re really good at this kind of thing, try guessing the year each composition was written (which is included in the “Piece” section). Better yet, don’t keep score and just play.

#1 Audio
(Highlight boxes below to reveal answers)

Composer	Piece	Movement
Pierre Boulez	Le marteau sans maître (1955)	3. L’artisanat furieux

#2 Audio

Composer	Piece
György Ligeti	Artikulation (1958)

#3 Audio

Composer	Piece	Movement
John Adams	Violin Concerto (1993)	3. Toccare

#4 Audio

Composer	Piece	Movement
Luciano Berio	Sinfonia (1968-69)	2. O King

#5 Audio

Composer	Piece
Arvo Pärt	Beatus Petronius (1990)

#6 Audio

Composer	Piece	Movement
John Cage	Sonatas and Interludes (1946-48)	Sonata 1

#7 Audio

Composer	Piece	Movement
Igor Stravinsky	Symphony in Three Movements (1942-45)	Movement 3

#8 Audio

Composer	Piece	Movement
Claude Debussy	La Mer (1903-05)	2. Jeux de vagues

#9 Audio

Composer	Piece	Movement
George Crumb	Makrokosmos I (1972)	8. The Magic Circle of Infinity

#10 Audio

Composer	Piece	Movement
Arnold Schoenberg	Pierrot lunaire (1912)	20. Heimfahrt

#11 Audio

Composer	Piece
Helmut Lachenmann	Mouvement — vor der Erstarrung (1983-84)

#12 Audio

Composer	Piece
Steve Reich	Music for 18 Musicians (1974-76)

#13 Audio

Composer	Piece	Movement
Gerard Grisey	Vortex Temporum (1994-96)	1. Interludio I

#14 Audio

Composer	Piece
Morton Feldman	Durations 1 (1960)

#15 Audio

Composer	Piece	Movement
Olivier Messiaen	Turangalîla-Symphonie (1946-48)	2. Chant d’amour 1

---

One side note on the Stravinsky excerpt:
In his article “Stravinsky’s Orchestrational Style” (Juilliard Review, IV, No.2 (Spring 1957), pp.10-19), Jacob Druckman points to this precise gesture as bearing the quintessence of Stravinskian orchestration. He writes:

“The key to Stravinsky’s orchestral clarity lies in his ability to produce a sound which can probably best be illustrated by that of striking a bell. The original impact or ictus is sustained, not in its original quality, but by a purer and softer ringing of the original tone… In its manifestations it allows the most forceful forte to exist in a transparent texture; it allows incisive rhythmic emphasis of any notes in a given line, the delineation of contrapuntal entrances, even the addition of tiny excitements in an otherwise Mozartean accompaniment.

[Rehearsal 157] from Symphony in Three Movements illustrates the bell sound in its most obvious form. Imagine this sweep to a D major chord in the hands of another composer. With Wagner there would probably be a rush of strings and woodwinds to a solidly-based tutti; with Ravel, probably a sustained crescendo chord in horns and trombones over which the harp would sweep up to the D major tutti with strings divisi on every possible chord tone. Stravinsky chooses not harp, but piano and horns for the glissando. Besides being more incisive, the piano can be quickly dampened after the third beat, whereas the harp certainly could not execute an étouffé over three octaves. The horns, in order to accomplish the glissando, must force to the point where the high D will sound cuivré. Incidentally, the unison sound of high horns and piano is remarkably bell-like. The ictus is reinforced in the upper partials by the marcato flutes and piccolo in their most incisive range, and by the high pizzicato. The only sound that remains after the first striking of the chord is the clarion triad in the trumpets, piano [dynamic].” (p. 11)

How Bach Prevented Consecutive Fifths and Octaves

In a previous post, I examined 46 instances of consecutive fifths and octaves found in the Bach chorales. Here is a summary of what we learned:

• Bach did on relatively rare occasion write consecutive fifths or octaves involving non-chord tones (13 instances). Also, seven chordal consecutives appear in the Bach chorales, all of which involve leaps rather than steps, most of which involve fifths and not octaves, and none of which involve outer voices.
• Consecutive fifths are far less objectionable than consecutive octaves.
• Consecutives involving non-chord tones are less objectionable than chordal consecutives.
• Consecutives in contrary motion are far less objectionable than consecutives in parallel motion.
• Consecutives involving inner voices are far less objectionable than soprano-bass consecutives.
• Parallel fifths created at authentic cadences involved a Re-Do anticipation in the melody over Sol-Fa movement adding the seventh of a V7 are not at all objectionable.
• Bach simply does not write stepwise chordal parallels.

Developing a clearer picture of Bach’s approach to consecutives in the chorales, however, should involve looking not only at the consecutives that he did write, but also the consecutives that he seemingly intentionally prevented. The difficulty of this task lies in the rather obvious fact that we cannot always know the pains taken by Bach to avoid consecutives just from examining Bach’s scores. We have no log of Bach’s thought process through the compositional act. (Robert L. Marshall has done extensive work examining corrections in Bach’s manuscripts, and while his essays on these corrections do suggest that Bach occasionally made corrections in order to fix consecutives, unfortunately Marshall does not expound upon these general statements, nor does he specifically catalogue the manuscripts in which these corrections occur.) Nevertheless, certain musical devices do seem to be regularly employed for what seems to be the explicit purpose of preventing or masking consecutives.

Certain overriding questions are of interest here: First and foremost, does Bach intentionally prevent the kinds of parallels that in other cases he was comfortable writing (as presented in the previous post)? If the answer is no, then our enterprise of developing Bach’s approach to consecutives is further solidified. If the answer is yes, then further questions follow. Are there contextual factors that led Bach to consider certain kinds of consecutives to be allowable in some cases? Does this suggest that Bach’s approach to consecutives developed over time? Does this suggest that the consecutives that he did write were the result of oversight – i.e. “mistakes” – as Malcolm Boyd has suggested? If the answer to this last question is yes, then the question of why such mistakes were made follows.

So what musical devices are employed with any kind of regularity to prevent or mask parallels? Let’s get to it.

Suspensions

The 7-6 Suspension

Suspensions are occasionally used to prevent or offset parallels. In particular, the 7-6 suspension frequently masks parallel fifths involving upper voices. In fact, a significant proportion of all 7-6 suspensions do this. Looking at just chorales from the first 100 cantatas (BWVs 1-100), nine of the 38 total 7-6 suspensions (24%) prevent parallels. Remove the suspension and blatant chordal parallel fifths appear. Below are just three examples. Other instances (just from among the first 102 cantatas) include BWVs 7.7 (not in Riemenschneider), m.3/7; 28.6 (R23=R88), m.9; 56.5 (R87), m.21; 66.6 (Not in R), m.7; 78.7 (R297) m.3/7; 84.5 (R112), m.13; and 102.7 (R110), m.8.

EXAMPLE 1

With the exception of BWV 104.6, all of the instances of parallel-masking 7-6 suspensions mentioned here either involve suspension preparations that are off the beat (as in BWV 72.6 below), several involving suspension chains. Only 104.6 above prevents chordal parallels occurring from beat to beat. Does this suggest that Bach did not generally consider 7-6 suspensions to be sufficient in masking beat-to-beat parallel fifths? Possibly. Unfortunately, the incomplete sample size here and the fact that Bach is not averse to using other kinds of figuration to mask beat-to-beat parallels, as we will soon see, make it difficult to arrive at any firm conclusion. If further research confirms that the vast majority of parallel-masking 7-6 suspensions involve off-the-beat preparations, perhaps this suggests that Bach felt more comfortable finding other ways of preventing beat-to-beat parallels.

EXAMPLE 2

In the first two instances above, as with most cases of these kinds of 7-6 suspensions, the parallel fifths that are masked are chordal parallels, not the kinds of parallels that Bach occasionally allowed in his chorales. Example 3, however, features a 7-6 suspension that masks parallels involving a non-chord tone (passing tone in the alto), parallels that are occasionally found in the chorales. Unfortunately, since it is impossible to know whether Bach added this suspension for the intent purpose of masking the parallels or whether he added it for purely musical reasons, making any assertions regarding Bach’s approach to NCT-involving parallels based on this or similar passages is next to impossible.

EXAMPLE 3

Other Suspensions

Among the four basic suspension types (4-3, 9-8, 7-6 and bass 2-3), it perhaps makes sense that the 7-6 is used most frequently in masking parallels since the 7-6 suspension almost always involves a delayed chord root, and parallel fifths almost always involve the movement of root and fifth of one chord to root and fifth of another. Both the 4-3 and the bass 2-3 suspensions generally involve a delayed chord third, a chord member rarely involved in objectionable parallel motion. So these two suspensions cannot generally be used to mask parallels. As for the 9-8 suspension, Bach does not use the 9-8 to mask parallel octaves.

Below, however, is a rare instance of the use of the bass 2-3 suspension preventing parallels, and it actually features two! (The sequential force of the suspension chain actually helps in attenuating the effect of the staggered parallels.)

EXAMPLE 4

To summarize, Bach does indeed use suspensions to mask parallels. In the vast majority of cases, the suspensions are prepared off the beat further attenuating any negative effect of the staggered fifths. While it seems obvious that in these cases Bach has intentionally inserted suspensions to prevent chordal parallels given the frequency with which such suspensions appear, in the one case (BWV 72.6) of non-chordal parallels involving NCTs, parallels that Bach occasionally allows, it is impossible to know whether the 7-6 suspension was added for the intent purpose of masking the parallels or whether it was added for musical reasons.

Voice-Crossing

Naturally, consecutive perfect fifths and octaves are considered objectionable only if they involve the same two voices. A perfect fifth between soprano and alto in one chord followed by a perfect fifth between soprano and tenor in the next is not considered objectionable in any way. Bach frequently uses this principle in conjunction with voice-crossing to prevent objectionable parallels. This seems like a feeble way of getting around a potentially inevitable instance of parallels, a kind of cheat. Yet, Bach resorts to this technique regularly! In the passage below, another instance of a 7-6 suspension masking parallel fifths is immediately followed by a set of non-objectionable parallel fifths involving three voices. The natural resolution of the tenor’s B3 would most logically be the C#4. Since this would create objectionable parallels with the soprano, Bach simply gives it to the alto and has the tenor melodically descend a diminished fifth to E#3, the most logical resolution of the alto’s F#3. The alto’s octave descent crossing the tenor perhaps helps to delineate the two inner voices, while at the same time attenuating any negative effects of the disjunct melodic leaps.

EXAMPLE 5

In all honesty, this resembles the kind of clumsy work-around often encountered in beginning part-writing exercises of first-year theory students (at least the observant ones who detect the problematic parallels!). Yet, Bach apparently considered such methods legitimate. And, of course, Bach never fails to create strong, even beautiful, melodies in such contexts.

Example 6A demonstrates the voice-leading challenges that often accompany the Phrygian half cadence. The most logical place for the needed B-natural is in the tenor. Yet, what is the alto to do? Moving to B-natural is doubly problematic (for both melodic and chord doubling reasons) and moving to G4 produces parallel fifths with the soprano. The only solution is D4, thereby doubling the chord fifth, a solution not entirely objectionable, though perhaps not as desirable as a doubled root (G).

EXAMPLE 6A

However, instead this doubled-fifth solution, Bach achieves a doubled-root V chord through the use of voice-crossing (Example 6B). On beat 1, tenor and alto simply switch notes in preparation for switching back at the moment of cadence. The objectionable parallels imagined in Example 6A are still there, only three voices are involved, thereby legitimizing the voice-leading.

EXAMPLE 6B

Here is another problematic scenario:

EXAMPLE 7A

In Example 7A, a half cadence to a close-structure V chord is achieved. However, the chorale melody (which Bach is not at liberty to simply change) causes issues moving forward. If the V-vi progression that begins the next phrase is desired, what are the alto and tenor voices to do that don’t cause objectionable parallels with the bass? Take a moment to imagine a solution before seeing Bach’s version in Example 7B below.

EXAMPLE 7B

Bach breaks one of the fundamental “rules” of first year part-writing. He writes voice-crossing involving an outer voice! One might argue that the aural effect of the passage essentially does change the chorale melody, and that the pseudo parallel octaves involving both alto and soprano with the bass does not sufficiently attenuate its negative effect. Apparently, Bach felt otherwise.

Here’s a third passage that needs fixing. Blatant parallel octaves and fifths appear in the ascending voice leading of this IV-V progression (in C major). Were the bass in the lower octave providing more space for the tenor, alternative solutions are more easily imaginable. But leaving the bass where it lies, what solutions are possible?

EXAMPLE 8A

Bach’s solution is rather brilliant (Example 8B). The swapping of inner voices here in no way changes the chord structure of the problematic passage above. The precise notes of these two chords are identical in both passages! The inserted passing tone in the alto (E) draws attention to the voice-crossing, further attenuating any negative effect of the pseudo parallels. The alto’s octave leap that follows is combined with the tenor’s octave leap of its own, though in the opposite direction. The final result is a passage featuring two wonderfully striking inner voice melodies, a feature so common in Bach’s chorales.

EXAMPLE 8B

In each of the cases cited here, voice-crossing prevents choral parallels rather than the kinds of parallels involving NCTs that Bach occasionally allowed. This seems to true for all other instances of voice-crossing preventing parallels that I have found. A couple other such passages are BWV 67.4 (not in Riemenschneider), m.2; 86.6 (R4), m.13; 330 (R33), m.2.

Chordal Leaps

Does Bach use chordal leaps to prevent parallels? Yes, he does, and frequently. Perhaps the most common type of parallel-preventing chordal leap is the anticipation figure that staggers the consecutives, as in Example 9. Notice that the anticipation figure, E4 in this case, is a chord tone within the first chord (E major). It is an anticipation figure, not a true non-chord tone anticipation. And here, a point is worth stressing. With the exception of suspensions, Bach does not typically use surface non-chord tones to prevent chordal parallels. A simple passing tone, anticipation, or incomplete neighbor does not sufficiently mask objectionable consecutives fifths or octaves. Chordal figuration, on the other hand, is used by Bach to prevent parallels, in most cases staggering them.

EXAMPLE 9

This kind of chordal leap anticipation figure can be found in numerous chorales. Here are just a few other examples: BWV 70.7 (not in Riemenschneider) m.28; 95.7 (not in Riem.), m.12; 99.6 (not in Riem.), m.2; 144.3 (R3), m.2; 229.2 (not in Riem.), m.7; 244.54 (R74), m.15.

In the previous examples, consecutive fifths are staggered by the early arrival of one of the voices. In the next example, parallel fifths are staggered by the delayed arrival of one of the voices.

EXAMPLE 10

Once again, notice that the tenor’s D4 which carries over into beat 2, is a chord tone in the beat 2 chord. Thus, no non-chord tones are involved here.

Chordal leaps other than the simple staggering of parallels feature in Examples 11 through 16 below. The first four mask parallel fifths while the last two mask parallel octaves. In all cases, these chordal leaps prevent chordal parallels.

EXAMPLE 11

EXAMPLE 12

EXAMPLE 13

EXAMPLE 14

EXAMPLE 15

EXAMPLE 16

While the previous chordal leaps occur off the beat staggering beat-to-beat parallels, Bach also occasionally interrupts parallels by inserting a different chord member on the beat. The following two examples break up parallels (fifths in both cases) with on-the-beat chordal leaps.

EXAMPLE 17

EXAMPLE 18

A few other examples of chordal skips masking consecutives are BWV 20.7=20.11 (R26), m.1; 22.5 (not in Riem.), m.6/15; 41.6 (R11), m. 44; 124.6 (not in Riem.), m.13; 281 (R6), m.5; 334 (R73), m.10.

Anticipation Figure Creating a New Chord

Earlier, I stated that Bach does not typically use basic surface non-chord-tone figuration such as passing tones or anticipations to mask parallels. However, Bach does not object to using anticipation figures (not true NCT anticipations) that create new chords to break up parallels. Perhaps the most common situation in which this occurs is the ever-challenging iv7-V progression with its potential parallel fifths – chord third and seventh of the iv7 moving to root and fifth of the V. As in Example 19, the parallel fifths are staggered by resolving the seventh early, thereby transforming the iv7 into a iiø65 chord. Two other instances of this particular anticipation figure are BWVs 162.6 (not in Riem.), m.10; 177.5 (R71), m.11.

EXAMPLE 19

Other Methods

Two additional examples of Bach using surface devices to prevent parallels are given below. These instances are individual, not representing a broad category, but they demonstrate other creative ways in which Bach prevented parallels.

In one sense, Example 20 features a leap like the many other chordal skip examples already seen. The difference here is that the leap is to a chord tone of a new chord. The F chord, V of Bb, that appears on the beat of beat 2 moves deceptively to a moves to a vi chord (G minor) on beat 3. The chordal parallels that occur from beat to beat are broken up with the interpolation of a viio7/vi between the two chords. The chromatic ascent of the bass further removes any negative effect of the beat-to-beat parallels.

EXAMPLE 20

Example 21 comes from Bach’s motet, Komm, Jesu, Komm!

EXAMPLE 21

This curious passage features two instances of parallels that are, in a way, staggered by each other. Remove the tenor’s D on beat 1, and parallel octaves occur. Remove the G, and parallel fifths occur. But the curiousness of the passage doesn’t end there. Is the soprano’s F-E motion a suspension? If treated so, then the harmony is not iv7, but iiø65. Yet this interpretation would suggest that the tenor’s D is the chord seventh, which should by “rule” resolve down by step, a resolution that would problematically double the soprano’s C# leading tone on beat 2. The alto’s accented passing tone (A) perhaps contributes to the harmonic ambiguity of the moment, further attenuating any sensed pull of the tenor’s D to C#. The manner in which Bach has avoided all potential problems here is ingenious. (Side note: This “chorale” is, in fact, not a true Lutheran chorale. Its melody is (presumably) composed by Bach, so one might suggest that Bach could have changed it here to allow for the tenor’s movement to C#. Given Bach’s general approach of composing outer voices first, perhaps Bach did not want to disturb that outer framework after the fact.)

Conclusion

To summarize our findings here, Bach uses certain musical surface devices to mask or prevent parallels. These devices include 7-6 suspensions, voice-crossings, chordal leaps, and anticipation figures into new chords. He does not, however, use 9-8 suspensions or other surface non-chord tones like retardations, anticipations, incomplete neighbors (escape tones or appoggiaturas), or passing tones (accented or unaccented) to prevent parallels.

What does this information contribute to our overall understanding of Bach’s approach to consecutive perfect intervals? Considering that, by and large, the kinds of chordal parallels that are prevented in these examples are chordal consecutives, are consecutives in parallel motion rather than in contrary motion, and thus, are not the kinds of parallels Bach occasionally allowed in his chorales, the data here largely coincides with the findings from the previous post regarding the consecutives Bach did write. There is no evidence from this research, then, that the NCT-involving consecutives occasionally found in Bach’s chorales are “mistakes,” (as Malcolm Boyd has suggested).

For those of us who teach part-writing, might all this data change the way we approach the “rules” regarding parallel or consecutive perfect intervals? Instead of discouraging students who attempt creative (or not-so-creative) solutions to problematic voice-leading, perhaps we should encourage them to do so! After all, if Bach is our supreme model for part-writing (as he should be), shouldn’t we also model our ways of cheating after him as well?

And for you students who are currently studying part-writing, the next time your professor marks your parallel-masking suspension or voice-crossing as wrong, feel free to show him or her any one of the Bach examples presented here. Only don’t tell them where you found them!

Consecutive 5ths and Octaves in Bach Chorales

So did Bach actually write consecutive perfect fifths or octaves in his chorales, breaking the First Commandment of part-writing? The short answer is yes, he certainly did. The long answer, however, will be far more informative if it addresses additional questions regarding consecutive perfect intervals: What kind of consecutives or parallels did Bach write and in what specific contexts? How many consecutives are in parallel motion as opposed to contrary motion? Did he only write parallels that involve non-chord tones (i.e. parallels that disappear with the removal of NCTs)? Conversely, what do the chorales tell us about how Bach prevented parallels? What kinds of parallels did Bach intentionally prevent? These last two questions will be taken up in a later post. For now, let’s concern ourselves with the data relating to the first set of questions.

An interesting paper on this topic has already been written by George Fitsioris and Darrell Conklin who used a computer program to locate parallel fifths in the Riemenschneider edition chorales. It gave me a great head start on the data presented here. However, two problems with the paper limit its scope considerably. First, it deals only with consecutive P5s and P8s in parallel motion, ignoring any consecutives in contrary motion (which are generally considered to be categorically the same, at least in most music theory textbooks). Second, only the chorales found in the Riemenschneider edition were included in the data, thereby excluding nearly 60 chorales. For the research to be more informative and comprehensive, the scope must be broadened.

So how many consecutive P5s/P8s did Bach write in the chorale?

Well, technically, there are at the very least 46 instances of consecutives, which are listed below. The first number is the chorale’s BWV number, with the decimal indicating movement number. The “R” numbers are Riemenschneider edition numbers. And an important note about measure numbers: the Aufgesang measures of all bar form chorales (which constitutes the majority of Bach’s chorales) are counted twice in my research since the Aufgesang is always repeated. Thus, measures within the Aufgesang are numbered as “4/8.” Decimals, when used with measure numbers, refer to beats.

7.7 (RX), measure 2/6.1
22.5 (RX) measure m.27-28
26.6 (R48) measure 4.2
33.6 (R13) measure 14
40.8 (R8) measure 2.2
40.8 (R8) measure 4.2
40.8 (R8) measure 6.2
40.8 (R8) measure 16.2
48.7 (R266) measure 14.2
60.5 (R216) measure 6
78.7 (R297) measure 4-5
86.6 (R4) measure 13-14
92.9 (RX) measure 14
99.6 (RX) measure 11
99.6 (Rx) measure 12
108.6 (R45) measure 4
111.6 (RX) measure 15
115.6 (R38) measure 10-11
146.8, (RX) measure 10.3
157.5, (RX) measure 10.1
167.5 (Rx) measure 11/25-12/26
174.5, (R58) measure 23.4
183.5 (R123) measure 14
190.7 (R327) measure 18
244.40 (R121) measure 4.3
244.44 (R80) measure 14 (2 instances)
244.54 (R74) measure 14 (2 instances)
244.62 (R89) measure 14 (2 instances)
245.40 (R107) measure 23
248.33 (R139) measure 2.2
251 (R329) measure 14.1
263 (R128) measure 6.2
266 (R208) measure 5
301 (R134) measure 3.3
308 (R27) measure 9
323 (R320) measure 8
329 (R212) measure 5
333 (R226) measure 12
340 (R277) measure 21
347 (R2) measure 14
361 (R264) measure 12.2
385 (R36) measure 6
436 (R278) measure 18

Excluding consecutives following a fermata…

As I said, “technically” there are 46 instances of consecutives. However, we can eliminate over half of these (26 to be exact) if we exclude consecutives that occur between the last chord of one phrase and the first chord of the following phrase, as in BWV 244.54 below, their exclusion perhaps being warranted by the fact that new phrases constitute a syntactical restart to a significant degree.

See the Manuscript: Image Link:  D-B Mus. ms. Bach P 25

The 26 instances of “fermata consecutives” are:

60.5 (R216) measure 6
78.7 (R297) measure 4-5
92.9 (RX) measure 14
99.6 (Rx) measure 12
108.6 (R45) measure 4
111.6 (RX) measure 15
115.6 (R38) measure 10-11
157.5, (RX) measure 10.1
174.5, (R58) measure 23.4
183.5 (R123) measure 14
190.7 (R327) measure 18
244.44 (R80) measure 14 (2 instances)
244.54 (R74) measure 14 (2 instances)
244.62 (R89) measure 14 (2 instances)
245.40 (R107) measure 23
266 (R208) measure 5
301 (R134) measure 3.3
329 (R212) measure 5
333 (R226) measure 12
340 (R277) measure 21
347 (R2) measure 14
385 (R36) measure 6
436 (R278) measure 18

However, before moving on and completely ignoring these “non-syntactical” consecutives, let’s consider them further, as they are not completely devoid of information. In fact, they may not be entirely non-syntactical after all. First, while 26 sounds like a high number, it’s actually not. If we estimate (conservatively) that there are on average six phrases per chorale (which means five internal cadences), then there are more than 2,000 internal cadences in the ~410 chorales of Bach. Even if we set the number at that low threshold of 2,000, then the 23 cadences represented here constitute only 1.15% of internal cadences feature consecutives between the phrases, a minuscule number. This extremely low figure would seem to suggest that Bach did not completely disregard the voice-leading between phrases as being syntactically insignificant. It should also be pointed out that not all phrase junctures are equal. Some cadences, like those internal within the Aufgesang, are more open-ended than cadences ending the Aufgesang. The more open-ended the cadence, the more syntactically connected with the subsequent phrase. Of the 23 cadences, eight consist of a half cadence followed by a phrase that begins in the same key. The other 15 either consist of a cadence after which a phrase begins in a new key or they end the Aufgesang. 14 of these 15 are authentic cadences. (In one case, a half cadence is followed by a phrase that starts in a new key area.)

[Note: I did not consider cadence transitions between the end of the Aufgesang and the repeat to the beginning of the chorale. Doing so would have almost certainly increased the number of “fermata consecutives,” thereby decreasing the proportion of “fermata consecutives” at open-ended phrases.]

If we put these 26 “fermata consecutives” aside, we’re left with 20 consecutives to deal with. In each case, I have examined original manuscripts to ensure that the consecutives were not the result of a copyist error. (Fitsorios and Conklin found this to be the case with BWV 355 (R169) measure 15. Parallel fifths result from a mistake in the melody – the soprano’s beat 1 B should be an A.) Images and links to original documents are provided. Here are the remaining 20:

7.7 (RX), measure 2/6.1
22.5 (RX) measure m.27-28
26.6 (R48) measure 4.2
33.6 (R13) measure 14
40.8 (R8) measure 2.2
40.8 (R8) measure 4.2
40.8 (R8) measure 6.2
40.8 (R8) measure 16.2
48.7 (R266) measure 14.2
86.6 (R4) measure 13-14
99.6 (RX) measure 11
146.8, (RX) measure 10.3
167.5 (Rx) measure 11/25-12/26
244.40 (R121) measure 4.3
248.33 (R139) measure 2.2
251 (R329) measure 14.1
263 (R128) measure 6.2
308 (R27) measure 9
323 (R320) measure 8
361 (R264) measure 12.2

Cadential Consecutives created by anticipation + delayed seventh…

By far the most common type of mid-phrase consecutive 5ths one finds in the Bach chorales is created by the simultaneous appearance of a Re-Do anticipation in the soprano over a delayed arrival of the seventh of a V7 (Sol-Fa) in an inner voice at an authentic cadence. Nine of the remaining 20 consecutives constitute this very specific figure.

26.6 (R48) measure 4.2
40.8 (R8) measure 2.2
40.8 (R8) measure 4.2
40.8 (R8) measure 6.2
40.8 (R8) measure 16.2
146.8, (RX) measure 10.3
244.40 (R121) measure 4.3
263 (R128) measure 6.2
361 (R264) measure 12.2

The resulting parallel fifths are considered to be “non-structural” since they involve a non-chord tone (NCT) in the soprano combined with a passing figure, the chord seventh. Of the nine instances, four of them appear in a single chorale: BWV 40.8 (R7).

See the Manuscript in Bach’s own hand: Image Link:  D-B Mus. ms. Bach P 63

That leaves us with 11 consecutives yet to examine:

7.7 (RX), measure 2/6.1
22.5 (RX) measure m.27-28
33.6 (R13) measure 14
48.7 (R266) measure 14.2
86.6 (R4) measure 13-14
99.6 (RX) measure 11
167.5 (Rx) measure 11/25-12/26
248.33 (R139) measure 2.2
251 (R329) measure 14.1
308 (R27) measure 9
323 (R320) measure 8

Other consecutives created by NCTs…

As mentioned, these nine consecutives are considered non-structural since they involve non-chord tones. Eliminate the non-chord tones and the consecutives disappear, leaving a foundational harmonic framework that features faultless voice-leading. Of the 11 remaining consecutives, four others also result from NCTs. Three involve a passing tone and one involves a neighbor tone. Let’s take these one by one.

BWV 48.7 (R266) measure 14.2

See the Manuscript in Bach’s own hand: Image Link:  D-B Mus. ms. Bach P 109

Here, the alto’s passing tone creates parallel fifths with the tenor. In terms of scale-degrees, these parallels resemble the nine examined in the previous section: Re-Do motion over Sol-Fa. In this particular context, the strength of the parallel tenths between bass and alto overrides the negative effect of the parallels (much as parallel tenths between outer voices in a I-V43-I6 progression overrides the effect of the unresolved seventh). Yes, Bach could have left the alto’s passing tone out, thereby eliminating the parallel fifths, but the parallel tenths were perhaps simply too compelling a musical idea. The soprano’s upward movement contrary to the parallels further masks any negative effect.

BWV 167.5 (RX) measure 11/25-12/26

See the Manuscript: Image Link:  D-B Mus. ms. Bach P 46, Faszikel 2

This passage represents the only instance of midphrase consecutive octaves that occur in parallel motion. All other instances of consecutive octaves occur in contrary motion, and all other parallel consecutives are parallel fifths. Two observations are worth pointing out in this particular context. First, both voices involved have NCTs. Second, this chorale features an elaborate instrumental accompaniment that is not included in the example. Still, the parallels are curious. The tenor could easily have stayed on the D instead of leaping to the F#. Alternatively, the bass could have gone to D# instead of F# just as the continuo part does (not shown in the example). So baffling are these parallels that I checked the original manuscripts for both the score and the parts. Both show that Bach indeed wrote the parallels. At least I am not alone in my bewilderment. The scholars of the BGA (Bach Gesellschaft Ausgabe) apparently were so puzzled that they decided to insert a question mark by the bass’s F#:

BWV 248.33 (R139), measure 2.2

See the Manuscript in Bach’s own hand: Image Link:  D-B Mus. ms. Bach P 32

Here, perhaps the bass’s chromatic descent combined with the use of mode mixture sufficiently attenuates the negative effect of parallels. Perhaps the passing tone in the melody had become a standardized feature of the chorale tune. Or perhaps neither of these sufficiently excuse these parallels!

BWV 99.6 (RX), measure 11

See the Manuscript in Bach’s own hand: Image Link:  PL-Kj Mus. ms. Bach P647 [früher D-B Mus. ms. Bach P647]

Another head-scratcher. The parallels are created by a needless neighbor tone. In none of the other six chorale settings of this Gastorius tune did Bach add a lower neighbor in this spot.

These four consecutives (BWVs 48.7, 167.5, 248.33, 99.6) are, like the prior nine examined, created by the insertion of NCTs. The remaining seven consecutives, however, constitute legitimate midphrase chordal consecutives:

7.7 (RX), measure 2/6.1
22.5 (RX) measure m.27-28
33.6 (R13) measure 14
86.6 (R4) measure 13-14
251 (R329) measure 14.1
308 (R27) measure 9
323 (R320) measure 8

The rest, involving chordal consecutives…

BWV 7.7 (RX), measure 2/6

See the Manuscript of the parts: Image Link:  D-LEb Thomana 7

These consecutive fifths in contrary motion are also a bit curious since an easy fix presents itself. The tenor could easily move from its F# to G, thereby doubling the root of the VI chord, preparing its G in the iiø65 chord that follows, and, of course, eliminating the consecutives.

BWV 22.5 (RX), measures 27-28

See the Manuscript in Bach’s own hand: Image Link: D-B Mus. ms. Bach P 119

Again, curious. Why have the tenor leave its D before moving to the C? The chordal leap to G is entirely unnecessary since both the bass and the alto have the G and since movement directly to C from the D would constitute strong voice leading. It should be noted that this chorale, too, has a more elaborate instrumental texture. Yet, nothing within that texture would seem to require the tenor’s leap to G from a voice-leading standpoint.

BWV 33.6 (R13), measure 14

See the Manuscript: Image Link:  D-B Mus. ms. Bach 1023

Downward leaps in the chorale melody play a role in four of the remaining six instances of consecutives, including this one. The tenor’s leap up to Do results in a doubled root. But certainly a doubled root is not reason for writing consecutive fifths. We’ve already seen Bach go out of his way to double the chordal fifth in another occasion. So there’s no clear explanation as to why the tenor doesn’t simply remain on the A here. (The beginning of the phrase that follows in no way requires it either.)

BWV 86.6 (R4)  measures 13-14 & BWV 251 (R329) measure 14

See the Manuscript: Image Link:  D-B Mus. ms. Bach P 157

See the Manuscript in Bach’s own hand: Image Link:  D-B Mus. ms. Bach P 123

These two instances of consecutives are taken together since they occur in the exact same spot in two settings of the tune “Es ist das Heil uns kommen her.” The melody features a descending leap from Re to La following a scale down from Fa. This descent of a sixth (Fa to La) poses voice-leading challenges. In BWV 86, imagining an easy fix to eliminate these consecutives is difficult. Perhaps the most easily achieved option would be for the tenor to leap down to the F# directly from the D# without the chordal skip to B. In BWV 251, Bach places the alto above the soprano in order to alleviate spacing limitations, something he does in this spot in another setting of the tune (BWV 9.7, R290). Yet, consecutive octaves still occur. The effect of these are attenuated by the voice-crossing and by the soprano’s quick movement away from the E (the only time in Bach’s five settings that the melody features this). If examining only the voice parts, a fix is easily achieved by taking the tenor up to E at the ii65 chord (see below). However, this would result in parallel unisons with the lower horn part.

BWV 308 (R27), measure 9

Chorale survives by way of the Breitkopf edition (pub. 1780s): Image

Taking this passage as evidence, consecutives in contrary motion are much less objectionable to consecutives in parallel motion, even if the former involves chordal consecutives and the latter involves consecutives created by NCTs. The tenor here could have moved to C instead of G, thereby creating parallels with the soprano’s passing figure. Yet, Bach opted for the contrary motion consecutives.

BWV 323 (R320), measure 8

Chorale survives by way of the Breitkopf edition: Image

Another descending soprano leap occurs here, though a simple early arrival of the B as a chordal skip in the tenor would stagger the fifths sufficiently, a simple “fix” that Bach employs frequently.

Quick Summary of the data

While Bach certainly did write consecutive fifths and octaves in the chorales, we see patterns emerge in the kinds of consecutives that appear and in the contexts in which they occur.

Consecutive fifths are far less objectionable than consecutive octaves. Of the 46 consecutives, only 10 involve octaves. If we eliminate “fermata consecutives,” only 3 of 20 involve octaves. Considering only chordal consecutives, 2 of 7 involve octaves.

Consecutives involving NCTs are less objectionable than chordal  consecutives. Of 20 consecutives not occurring after a fermata, 13 involve NCTs.

Consecutives in contrary motion are far less objectionable than consecutives in parallel motion. Of all the 33 consecutives that involve leaping voices, only six are in parallel motion. Even with “fermata consecutives,” only four of 26 involve parallel motion.

Consecutives involving inner voices are far less objectionable than soprano-bass consecutives. Not a single one of the non-fermata consecutives occur between soprano and bass. This may partly be due to the fact that chorale melodies feature far more conjunct motion than disjunct. Since Bach considers contrary motion consecutives to be less objectionable than parallels, and since contrary motion consecutives virtually always involve leaps, the lack of soprano-bass consecutives is perhaps understandable. Nontheless, chorale melodies do feature occasional leaps, so the complete lack of outer voice consecutives remains significant. Of the 26 fermata consecutives, only six involve outer voices.

Parallel fifths created at authentic cadences involving a RE-DO anticipation in the melody over SOL-FA movement adding the seventh of a V7 are not at all objectionable. Nine of the 13 consecutives created by NCTs represent this specific parallel motion. For the record, he writes them in both major and minor modes – five are in major, four in minor.

Bach does not write stepwise chordal parallels. There simply are none. All stepwise parallels involve NCTs rather than being chordal consecutives. All chordal parallels (there are only 2, and both are fifths) involve leaps.

The question of why Bach wrote the consecutives that he did, particularly those chordal ones that are quite fixable, is difficult to answer. Malcolm Boyd has suggested that Bach, who often quickly tossed off his cantata-ending chorales at the end of the week prior to Sunday services in Leipzig, simply overlooked these consecutives. They are truly mistakes. Fitsioris and Conklin, on the other hand, challenge this idea. (“After detailed research, carefully avoiding ‘wrong’ scores found in certain printed editions of Bach chorales, we came up to the conclusion that in 18 passages Bach seemed to be tolerant with such ‘forbidden’ successions.” p.2) If Boyd is right, then this research tells us something about the kinds of consecutives that evaded the great composer’s ear and eye. (Honestly hard to imagine.) If Fitioris and Conklin are right, this research tells us about the kinds of consecutives he considered allowable. Of course, both could be right to some degree, with some consecutives being intentionally allowed (an idea which seems to be supported by the patterns we’ve seen) and others being mistakes. Perhaps we can learn more by looking at the kinds of parallels Bach intentionally prevented via suspensions, chordal leaps, NCTs, voice-crossings, and other devices. Does he in some cases prevent the same kinds of parallels that have been observed here? If so, might this lend support to the idea that these consecutives are indeed mistakes? This will be the topic of a subsequent post.

Intertextual Connections: Mozart, Beethoven & Rachmaninoff

Last week when revisiting the sonata-form first movement of Mozart’s Symphony No.40 in G minor from 1788, one brief passage began triggering intertextual thoughts. I had heard music very much like this before, but where? After some mind-scrambling, I came up with two passages, though I’m still not sure these two exhaust the intertextual connections buried in my subconscious.

The passage in Mozart’s symphony that instigated this chain of events comes from the moments just before the Recapitulation. In bars 160 to 165, the main motive of the movement (short-short-long descending figure with a weak-strong metric placement) is sequenced in tandem between flute and oboes descending chromatically over a dominant pedal just before the violins enter with the opening theme.

LISTEN

Notice that the main motive (labeled X below) is at times modified through this passage into a simple chromatic descent (labeled X’).

Now jump ahead from 1788 to 1811, the year Beethoven completed and premiered his  Fifth Piano Concerto. The passage below appears in the initial statement of the main theme of the concerto’s third movement. Aside from the change in meter and mode, the two passages are so similar as to be virtually identical. The voice leading of Beethoven’s right-hand top voice corresponds exactly with Mozart’s flute part: Both contain motive X and move from scale degree 5 to 1 by way of a lower neighbor to 7 (5-4-3-2-1-7-1) with chromatic steps inserted along the way. Likewise, Beethoven’s left hand inner voices correspond exactly to Mozart’s oboes, moving in parallel thirds from scale degrees 2/7 down chromatically to 3/1. Both resolve harmonically to tonic at the passage’s end.

LISTEN

Skip ahead more than a century to 1940, the year Rachmaninoff completed his final composition, the Symphonic Dances. The Mozart excerpt conjured up a particular passage from the first movement of Rachmaninoff’s Dances (shown below). The similarities aren’t as exact as with the Beethoven, and the passage is stretched out considerably, but the prominent features are still there. A pedal tone runs through most of the passage (though on tonic rather than dominant). Motive X is present, and while through much of the movement the short-short-long rhythmic element is usually combined with an outlined triad rather than a stepwise descent, here the motive appears in the form of X’, a chromatic descent. Furthermore, Rachmaninoff’s assignment of the X’ descent to double-reeds (bassoon and English horn) brings a timbral connection to Mozart’s oboes. (Was it this timbral connection that initially brought this passage to my mind?) The descending chromatic voice-leading over the pedal also connects with the Mozart, with the tenor voice 7-6-5-4-3 line connecting with the lower oboe part, descending chromaticism being one of the hallmarks of the Rachmaninoff style. (Was the fact that this line begins on the same note (D) as Mozart’s prominent flute line the reason this passage came to mind?) Finally, it may be worth noting that the prevailing key of this passage is Eb major, the same key as the Beethoven excerpt. (Or was it this key relationship to the Beethoven that triggered the connection? I really don’t know anymore.)

LISTEN

Why that note?!? — Bach French Suite No.1 Sarabande

“Why that note??” is a new series featuring musical pieces or passages in which a single note (or other musical detail) seems unjustified, out of place, or just simply wrong, within a given context. I suspect that in most cases, if not all cases, no answer will be provided to the main “why” question, which is not to say that there is no answer.

 

First, a confession. From my youth, I developed a habit of skipping over sarabandes of Bach suites. Please get me past those trudgingly slow triple meter dances to the much more exciting gavotte or gigue that follows! I prefer to lay the blame for this at the foot of Glenn Gould whose sarabande renditions I generally find to be as lethargic as his faster dances are thrilling and ebullient. (This is certainly unfair to Gould, but can you think of anyone else to blame? Neither can I.)

Then one day, I fell into love with is the sarabande from the C minor keyboard partita after learning that there is such a thing as a stirring, inspired performance of a sarabande.

French Suite No.1 Sarabande

More recently, I’ve been taken in by the sarabande from the first French Suite in D minor, BWV 812 (above) after spending time in class analyzing it. Together, we discovered how brilliantly the piece is constructed with its bass melody in measures 9-13 being a nearly exact duplication of the soprano melody in measures 1-5 but with completely different harmonies, with its descending chromatic voice-leading (mm. 1-5 and mm. 17-20 in the bass, mm.10-11 in all voices) giving it a spirit of lamentation, with its Neapolitan chord in an unusual second inversion (m.7), with its strategically placed accented dissonances (particularly in m.8, m.18 and mm. 23-24) adding angst to lament. We also spent time trying to figure out why the arrival of the low m.13 E-flat is so striking, especially since E-flats were already present in the immediate context.

And not only is the sarabande brilliantly constructed, it’s stunning in its beauty. Just listen.

Why that F?!?

However, I can make no sense whatsoever of the measure 21 F4 in the tenor voice. Not only can I not make sense of the note, I’ll go further and say that it’s simply a wrong note! Bach has it wrong. (Did I just say that?) It should be a D4 rather than an F4.

Listen again to this passage and pay attention to the effect of the F.

So why is the F wrong?

The primary reason has to do with chord structure. The chord implied here is a first inversion D minor chord, a tonic triad in the key of D minor, the immediate resolution of a V42 chord that preceded it. In first semester music theory, we learn that the strongest and much preferred chord structure for first inversion triads is to avoid doubling the bass. So why has Bach doubled the bass here? (F in tenor doubles the F in the bass.)

Of course, weak though it may be, doubling the bass in a first inversion chord is not in itself a part-writing “error” and is, in fact, commonly seen. However, such weaker doublings should only be used when voice-leading considerations dictate that they be used, and Bach follows this approach consistently in his music. But here, the most logical resolution of the V42 chord that preceded it is to the strongest doubling, with the C# resolving to D (doubling the soprano’s implied D that is never actually realized). This means that Bach has gone out of his way to achieve a chord structure that is weaker than it could be. And the fact that the D actually does appear on beats 2 and 3 of the measure adds to the confusion.

Possible reasons for the F

Perhaps Bach has inserted the F to create some kind of motivic gesture (F-D-D) that features prominently earlier in the piece. But this descending third gesture is found nowhere else in the piece. Or perhaps Bach is intentionally delaying the arrival of the D to beat two to match similar delayed arrivals in the G—F—E—D descending line beginning in m.17. The F is delayed to the second part of beat 1 in m.18, the E is delayed to beat 2 in m.19, and here the D is delayed to beat 2 in m.21.

Relatedly, perhaps Bach is intent on never keeping three separate voices on the same note through an entire measure. See measures 1, 5, 6, 9, and 13 – even when two accompanimental voices are stationary, the third is moving.

Alternatives Explored

But do these reasons really explain or warrant Bach’s decision to put an F on the downbeat? He easily could have created movement in ways that do not result in obviously weakened chord structures.

For example, he could have created motion in the tenor by putting a C# on beat 1 creating a retardation figure like the F#—G in measure 5, like the implied F#—G retardation figure in m.13, and like the C#—D soprano figure in the final cadence. In short, the C# on beat 1 of measure 21 would fit the musical features of the piece with its accented dissonance and delayed resolutions, and it would result in a much improved chord structure.

Or, Bach could have created motion by inserting a neighboring E on beat 3. This would increase harmonic tension on beat 3 (creating sevenths against both the soprano and bass), and would once again restore the ideal chord structure on beat 1.

Perhaps the best of these alternative solutions would be to move to F on beat 3 thereby creating a voice-exchange with the soprano. Voice-exchanges feature prominently throughout the sarabande, including another soprano-tenor voice-exchange in the immediate context (m.20).

Far be it from me to suggest improving Bach (though apparently not too far), but I simply cannot understand Bach chose the F for the downbeat. It goes against his usual practice of preferring stronger chord structures, it goes against the most logical voice-leading in the immediate context, and furthermore, plenty of alternatives are achievable. I was so flummoxed that I considered the possibility of a scribe error. But after checking three early manuscripts (here, here, and here), two of which were completed during Bach’s lifetime, I found the F present in all three. (The image below is from document P 418 dated from 1720-1739 and copied by the hand of Bernhard Christian Kayser.) This doesn’t entirely rule out the possibility of a scribe error, but it provides no reason to suspect it.

So I’m left scratching my head about that F. Why that note??

 

Liszt’s Nuages gris: a Tristan parody?

Post–tonal theory class, opening class agenda, two musical excerpts: the iconic opening to Wagner’s Tristan und Isolde Act I Prelude and Franz Liszt’s remarkable little piano piece, Nuages gris. The former, completed in 1859, is iconic for signaling the limits of chromatic saturation within common-practice tonality, and its much-debated harmonic structure provides plenty of discussion fodder. The latter, completed in 1881, is remarkable for its radical innovation especially considering the source from which it came. (It’s fun to play the “name that composer” game with this piece.) Tonal, harmonic and metric ambiguities abound throughout, leading to an open-ended final pseudo-cadence that contains elements of both resolution and non-resolution simultaneously. (Liszt’s “cadence”, with its whole-tone final chord, brings to mind the “in-between” music of Alexander Scriabin beginning around the op.53 Piano Sonata No.5. Compare, for example, Liszt’s cadence with the final cadence of Scriabin’s op.56, no.4 etude from 1908. The fact that Scriabin’s cadence has more tonal closure with its descending fifth bass line indicates just how radical Liszt’s piece was, being written 27 years earlier.)

Alexander Scriabin, Etude, op.56 no.4 (ending)

Liszt, Nuages gris, ending (Errata: B-natural in R.H. in final chord)

At any rate, as I worked through Nuages gris once again, I heard something I had never noticed before. There is a striking resemblance between the final 16 bars of the Liszt and the opening 13 bars of the Tristan prelude, though Liszt’s “cloudy” setting prevents the connection from being blatantly obvious. Listen 

Wagner, Tristan (1857-59), Act I Prelude opening

Liszt, Nuages gris (1881), mm.31-48.

The most obvious resemblance, the one that initially caught my ear, is Liszt’s rising chromatic soprano line. Like the Wagner, there are three main phrases consisting of 4 to 5 notes, the first notes of each phrase have longer duration relative to the final notes of each phrase, the final notes of each phrase fall on a weak beat with the penultimate note serving as a kind of accented dissonance, and the entire ascent occupies the same register, ultimately arriving at the pitch F# (which, in both cases, eventually leads to G). Listen to the Liszt passage again (starting at the 2:13 mark in the above YT video). With these resemblances in mind, it’s easy to hear the Tristan prelude in there.

The resemblances don’t end with the melodic ascent, however. The left hand of Liszt’s piece crawls downward chromatically matching the chromatic descents that pervade the Wagner accompaniment. Furthermore, the bass arrival on A connects with what is generally considered to be the initial key of the Wagner prelude, A minor.

Could this be Liszt borrowing from Wagner, whether consciously or subconsciously? Could he be commenting on or even parodying the Wagner passage from two decades earlier?

I investigated a bit and things got pretty interesting. Kenneth Hamilton’s fascinating book chapter entitled “Wagner and Liszt: Elective Affinities” from Richard Wagner and His World explains just how complex the relation between the two in-law composers was at this time. Hamilton, admittedly not interested in refuting the traditional view that “Wagner exploited Liszt — both financially and artistically — and that Liszt allowed himself to be exploited” (p.27), devotes his attention to showing the dramatic impact each had on the artistic endeavors of the other. Below is a timeline of relevant events drawing primarily from events Hamilton’s essay.

1841: Liszt’s song “Die Lorelei” completed.

One instance of outright thievery is obvious: Wagner’s Tristan prelude opening is right there in the opening bars of Liszt’s song “Die Lorelei” from 1841, revised in 1856. Take a listen. In addition to the obvious aural connections, notice the E#-F# motion mimicked by Wagner’s prelude.

Franz Liszt, “Die Lorelei” (1841, rev. 1856)

(So is Nuages gris an instance of Liszt stealing from himself, filtered through Wagner?)

1844: First version of Liszt’s “Ich möchte hingehn” completed.

The infamous “Tristan chord” (bar 2 of the Wagner prelude) can be found in Liszt’s song “Ich möchte hingehn” originally composed around 1844, thus predating the opera by about 15 years. However, the truth is that Liszt inserted the chord as a quotation of Wagner much later, after Liszt had become familiar with the opera.

1847/?1849: Liszt composes the first of many Wagner transcriptions to come, the Overture to Tannhäuser.

1856-59: Tristan composed

1865: Tristan premiered, led by conductor Hans von Bülow (from whom he stole Cosima, Liszt’s daughter, to be his wife). Von Bülow is also responsible for the piano arrangement of the Tristan prelude excerpted above.

1867: Liszt’s piano transcription of the “Liebestod” from Tristan completed

1881: Nuages gris completed, a few months after an accident falling down stairs left him bedridden and ultimately suffering from dropsy.

1882: Liszt’s composes his Solemn March to the Holy Grail (from Parsifal). Beginning in the mid-1870s, Liszt’s arrangements of Wagner’s music veered further and further away from their models. What began as embellishments led to “downright distortions.” Such distorted renderings reached an apex with this Parsifal March, which Hamilton refers to as a “twisted parody rather than a transcription from it, as if Liszt is trying to remember the music but can’t quite figure out how it goes.” (p.42) Hamilton points to this twisted march as the probable cause of Wagner’s sharp attacks on Liszt’s late music. According to Cosima’s diaries from November 1882, her husband referred to these late works of her father as evidence of “budding insanity.”

1883: Wagner’s death

1883: Liszt’s Am Grabe Richard Wagners completed, a “disjointed, nostalgic sketch” that even further extends the tendencies of the Parsifal march.

1886: Liszt’s death. That Tristan was on Liszt’s mind at the end of his life is evidenced by Cosima’s claim that her father’s dying utterance was “Tristan!” — that is, if we are to take that story as being anything more than legend.

Aleatory Quiz

John Cage and Karlheinz Stockhausen, c. 1958

It’s Aleatory Week in my Music Theory IV class, and I always give a fun aleatory quiz for the occasion. Aleatory (derived from the Latin alea, meaning “dice”) is “a term applied to music whose composition and/or performance is, to a greater or lesser extent, undetermined by the composer.” (Grove Music) Aleatory is thus synonymous with the term indeterminacy. Two broad types of aleatory music can be distinguished: music in which elements of the compositional construction are determined by chance (resulting in fixed compositions), and music in which elements of the performance are undetermined by the composer and must be determined by the performer(s) (e.g. through the use of graphic notation or “mobile form”). Certainly it is possible for these two types to be combined in a single work.

The below quiz contains six excerpts from the following three piano works: Pierre Boulez’s Piano Sonata No.2 (1947-48), Karlheinz Stockhausen’s Klavierstücke, set 1 (Nos.1-4) (1952), and John Cage’s Music of Changes (1951). Only the Cage work is aleatory. It is a composition in which each element was determined by the use of chance procedures. “Each detail of his score was determined by the toss of three coins six times, which directed him to a specific number in I Ching (the Oriental “Book of Changes”); this in turn sent him to a numbered position on one of twenty-six charts he had devised. Thus a single pitch was determined. The procedure was then repeated in the determination of duration, timbre, and other parameters.” (Watkins, Soundings, p.560) Cage’s Music of Changes is then aleatory of the first type described above – that is, chance operations were used in constructing its fixed, carefully notated score.

The quiz, then, essentially requires the listener to aurally identify Cage’s chance music from among the other two highly-controlled works. Each of the three works listed above are represented at least once on the quiz. If you are up for a real challenge, try to distinguish the Boulez excerpts from the Stockhausen excerpts as well. (No fair cheating with your Shazam app!) Answers are posted at the bottom of the page.

THE QUIZ
Number 1
Number 2
Number 3
Number 4
Number 5
Number 6

The obvious point of this quiz is the ironic fact that aleatory of the kind represented by Cage’s Music of Changes often produced results that are strikingly similar to works of composers in the modernist avant-garde, composers that came to loathe Cage’s aleatory. Boulez, who initially recognized an aesthetic affinity between Cage and himself, eventually attacked Cage mercilessly (without naming him) stating that the adoption of chance procedures only “conceal[s] a fundamental weakness in compositional technique… It is an artificial paradise, comfortably arranged, whose dreams are, I suspect, never very miraculous: a narcotic which protects against the needle-prick of invention.” (Boulez, “Alea” 1957, tr. Stephen Walsh)

#	Answers (highlight boxes below to reveal answers)
1	Cage, Music of Changes
2	Boulez, Piano Sonata No. 2, mvmt 4
3	Stockhausen, Klavierstück II
4	Boulez, Piano Sonata No. 2, mvmt 4
5	Cage, Music of Changes
6	Stockhausen, Klavierstück I

 

Glenn Gould’s Singing Transcribed – Bach Sinfonia 4

So, what to make of Glenn Gould’s incessant singing and muttering while performing? Some find it intolerable. Others find it charming. Some of us may even have favorite Gould vocalized passages. (One of mine is the 17^th measure of the 14^th Goldberg Variation from the 1981 recording.) Personally, I find Gould’s singing in no way distracting. I am no longer the big fan of Gould’s pianism that I once was, but that has nothing to do with his eccentric vocalizations. Most of the time, I even forget they’re there. (The subconscious mind must be at work here distinguishing the intentionality of musical tone from the non-intentionality of extraneous sounds.) If anything, Gould’s singing authenticates and humanizes his performances. It reveals a performer so entirely absorbed in the music’s moment and reminds us that this is a performance, even if within the confines of a recording studio. Gould’s mutterings distinguish his recordings from those countless note-perfect recordings available today that take on a fabricated, sterile, and even robotic quality. (Is perfection ever very interesting?)

What I find a bit more off-putting than Gould’s vocal eccentricities are the middle-register hiccups that emanate from Gould’s beloved “CD 318” Steinway piano on a couple of his recordings. The hiccups are perhaps most pronounced on Gould’s 1964 recording of Bach’s Inventions and Sinfonias (the first recording project following his well-known withdrawal from the concert stage) and are apparently the result of Gould’s constant tinkering with CD 318 in order to achieve as harpsichord-like a quality as possible. Gould found the hiccup effect charming enough not to abandon his dear CD318, which traveled with him throughout his concertizing career. (In the early 1970s, CD 318 suffered irreparable damage in a moving accident, which devastated Gould.)

Below is a transcription of Glenn Gould playing and singing the fourth Sinfonia in D minor. The score reflects exactly what Gould played and not necessarily what Bach notated (i.e. embellishments are written out), exactly what Gould sang (notated to the best of my ability), and the hiccups that ring from CD 318 (notated with boxed noteheads). Gould’s notated sung part is obviously an approximation in certain passages given his use of vibrato, frequent portamenti (some before the beat, some after), and occasional intonational imprecision. Whatever you make of Gould’s singing, at least now you can follow along. (By the way, is that Gould himself in the final measures or a passing bus?)

Listen here. (Headphones are an obvious advantage.)

Click to embiggen

The difficulty of transcribing Gould’s singing depends on several factors such as how fast and loud the piano is, how vocal and loud Gould is, whether the lower piano part falls in Gould’s baritone singing range, etc. (This fourth Sinfonia then is an ideal piece to transcribe as it is a slow work that lies above Gould’s singing range much of the time.) What makes transcribing Gould even possible, however, is that he truly does sing, as opposed to, say, the loud, guttural, untranscribable noises that Keith Jarrett makes while performing. (I am a big Jarrett fan and have learned to ignore all of his noises as well (though jazz improvisation is admittedly a different animal than playing Bach). To me, Jarret’s physical gyrations, which were worse in his younger years, are more distracting than his vocalizations, but perhaps that’s only because I don’t often watch him play.)

Now on to the very important research made possible by this project. Which part does Gould prefer to sing? Outer voices? Inner voice? And how often does Gould provide an added part as has been claimed? Well, Gould sings the top voice 23.1% of the time, the middle voice 24.5% of the time, and the lower voice 51.4% of the time. The fact that the lower voice falls nearest his voice range may be the most logical explanation for this preference. Only 2.2% of the time (for a total of 2 beats in mm. 13 & 14) does Gould sing an added part, and even these moments are more added notes that true parts.

So there you have it.