Playing Bach on the Guitar (by Frank Koonce)

Frank Koonce has been long established as a preeminent voice in the Classical Guitar world in regard to transcription & interpretation of the works of J.S. Bach. As the owner of Strings By Mail, I am personally gratified that Frank, a friend since out days together at North Carolina School of the Arts studying under Jesus Silva, is our first guest author on the new Strings By Mail Blog.

Frank’s own works were on the top of my list as we moved aggressively to add more excellent music to the offerings at SBM. We were soon very pleased to also be able to add works he has published under Sound Set and with Les Productions d’OZ such as The Frank Koonce Series. His work has brought us many excellent and important works including those of Nikita Koshkin. I encourage everyone to explore the fruits of his labor, both in those many publications and in this new article regarding the playing of Bach.

~John Wunsch

J. S. Bach: The Solo Lute Works 2nd Edition, ed. by Frank Koonce

J. S. Bach: The Solo Lute Works 2nd Edition, ed. by Frank Koonce

Playing Bach on the Guitar
Frank Koonce

Transcription vs. Arrangement

To adapt music written for one instrument so that it may be played on another involves making either a transcription or an arrangement. Transcriptions, strictly speaking, are more faithful to the original, while arrangements are more interpretive. For example, most modern guitar editions of Bach’s lute works are close enough to the original that they may be called transcriptions, whereas most guitar editions of Bach’s violin and cello works might better be called arrangements because of added notes and subjective reconstruction of the original. The only guitar edition of the cello suites I have seen that is a true transcription, with minimal editorial intervention, is by Richard Wright, who notes:

Although the guitar is in some respects the ideal instrument to realize the cello suites’ unique polyphony, the predominantly single-line texture of the originals has been preserved throughout…. The player must instead develop his or her own clear understanding of where one voice gives way to another. [1]

This excellent edition holds true to its stated intent to provide an alternative to pieces by Bach that are more difficult, either through Bach’s original design, such as the lute works, or through arrangements with added (and, as Wright argues, unnecessary) counterpoint. Furthermore, the player is given the opportunity to make his or her own decisions in realizing possible “implied polyphony” of the original, and this is without question a very worthwhile exercise.

When Bach recycled his own compositions, however, he did not simply transfer the same music from one instrument to another; instead, he modified it by taking into account the strengths and weaknesses of the new instrument. There is a particularly revealing comment by Johann Friedrich Agricola, a Bach pupil from 1738 to 1741, regarding the sonatas and partitas for solo violin. Agricola tells us, “Their composer often played them himself on the clavichord, adding as much in the way of harmony as he found necessary.” [2]

We are very fortunate to have pieces within the standard guitar repertoire that are examples of how Bach adapted his bowed-string music for performance on the lute or keyboard. For instance, the third in a set of four suites that are now usually attributed to the lute is Bach’s arrangement of a cello suite, and the fourth is his arrangement of a violin partita. Furthermore, two of Bach’s violin sonatas, commonly played by guitarists, may be compared to lesser-known keyboard arrangements of the same pieces. (The keyboard versions survive in the handwriting of one of Bach’s pupils, Johann Gottfried Müthel, and therefore may be his instead of Bach’s; nevertheless, they are entirely within Bach’s style.)

Compound Melodic Notation

The single-line texture of the originals that Professor Wright refers to is a style of composition and of notation that is often referred to as “compound melody,” in which two, three, or even four voices may be derived from what appears on paper to be a single voice. This is the notation Bach used in writing his solo cello works and solo violin works. Another term that is sometimes used is “implied polyphony.” Compound-melodic notation is like musical shorthand that greatly reduces the clutter of rests, ties, and separate stems and beams that would be required if the work were instead written in polyphonic notation. It is especially helpful for music that is compressed on a single staff, such as the violin and cello works. A limitation of this notation, however, is that a specific voice to which a note belongs is at times unclear, thus making this the responsibility of the performer to decide. Notwithstanding, a “silver lining” is that it can also result in different interpretations, which keeps the music fresh and challenging to performers.

In compound-melodic notation, larger intervals within what otherwise is mostly a stepwise line may suggest the presence of a second melody, or of rudimentary structures for harmonic and bass support. For instance, the exerpt shown in example 1a, from Anna Magdalena Bach’s manuscript of the first cello suite is fully realized in example 1b to show a three-voice interpretation in D minor for the guitar. The notes that are lengthened to overlap other voices, and the suspension shown in measure two, are subjective reconstructions.


Example 1a: BWV 1007, Menuet II, mm. 1–4.


Example 1b.

Example 1c is a compromise, but it is more consistent with Bach’s own practices in making his “lute” arrangements. For the latter, he expanded the bass into a fully independent voice line with added notes and rests. However, he left the upper voices notated as a compound melody for the most part, with some added chords and contrapuntal lines “as he found necessary.” This is the way I have chosen to make my arrangements of Bach’s bowed-string compositions.


Example 1c.

The performer is free to sustain and overlap the upper notes (or not) as desired to selectively bring out, or “realize,” different voices so that they are perceptible to the listener. One of the best ways is to allow the final note of one voice to overlap the entrance of another. [3]

Let’s look at some examples to show how and where, and to what extent Bach chose to change the compound-melodic notation for bowed string instruments into polyphonic notation for the lute or keyboard. First, example 2a is an exerpt from the Prelude of the third violin partita.


Example 2a: BWV 1006, Prelude, mm. 1–4.

Example 2b is from Bach’s arrangement of the same work for the fourth lute suite. Notice that the original notes are still written as a single line, and that new bass notes have been added and separated by rests.


Example 2b: BWV 1006a, Prelude, mm. 1–4.

Example 2c shows how a middle voice could have been realized in the notation. As you can see, however, the added rests and separate stems clutter the score, which is a strong argument for keeping much of the two upper voices written as one compound line as Bach chose to do. The added layer of notation is not needed.


Example 2c.

Example 3a shows an excerpt of the Bourrée from the same violin partita.


Example 3a: BWV 1006, Bourrée, mm 1–4.

Example 3b is the same passage from Bach’s version for the lute. Notice that he reassigned the original melody notes on the downbeats of measures two and three to become part of a separately notated bass line. Even though these notes (G# in measure two and F# in measure three) still function as part of the melody, as well as bass notes, he chose not to write double stems. Bach’s usual practice was simply to write rests in the upper part in places like these. He also added a few new bass notes at the end of the phrase to provide harmonic support at the cadence.


Example 3b: BWV 1006a, Bourrée, mm 1–4.

Example 4a, the Sarabande from the second lute suite shows another passage in which Bach chose compound-melodic notation over polyphonic notation, this time in an original work for the lute––one that was not arranged from the violin or cello. Even though he could have separated the upper line into two parts by notating a separate middle voice, as shown in example 4b, he chose not to do so, possibly because players can still easily see the linear connections through the overall layout of the music. Again, his notation is perfectly clear, without the added clutter of a separately notated third voice.


Ex.mple 4a: BWV 997, Sarabande, mm. 6–8.


Ex.mple 4b.

With compound-melodic notation, one’s interpretation may be affected by whether large melodic intervals are believed to be the entrances and exits of a separate voice that is in dialogue with another, or whether it is simply an expressive rhetorical gesture of a single voice. This ambiguity is illustrated in example 5, from the Prelude of the third lute suite, Bach’s arrangement of his fifth cello suite (BWV 1011).


Example 5a: BWV 995, Prelude, mm. 19–21.

The larger intervals can suggest two distinct voices, as shown in Ex.mple 5b; however, the upper notes, become nothing more than voice fragments if separated from the lower notes, and both lines lose the intensity that they have when thought of as a single voice.


Example 5b.

When all the notes are played together, they work better both melodically and rhythmically. In fact, you can make a logical and conventional stepwise line simply by changing some octaves, as shown in example 5c, but this is far less interesting and lacks the intensity of Bach’s original. Therefore, I hear this passage, with the notes in Bach’s original octaves, as one voice with dramatic rhetorical gestures.


Example 5c.

Even though Bach usually retained the compound-melodic notation of the original composition when arranging for the lute or keyboard, this did not prevent him from making other changes. Besides adding new bass notes, he would add chord tones, especially to avoid having open fifths and octaves. He would also sometimes create new contrapuntal lines, or add ornaments and melodic embellishments to take advantage of the new instrument’s idiomatic resources. example 6a, from a facsimile of the fifth cello suite prelude, is particularly interesting for a number of reasons, especially when compared to example 6b, a modern engraving of the same passage from the lute version. First, notice that Bach changed the cello’s compound-melodic notation into polyphonic notation for the first four measures by importing the lower notes into the newly created bass voice. In doing so, he lengthened the downbeat notes of measures two and four so that they sustain a little under the treble voice. The original notes in measures five through seven, on the other hand, seem to function best as a single voice; therefore, Bach added new bass notes under them to provide harmonic support. Similarly, he turned the final note of the phrase into a full chord by adding the third, fifth, and octave.


Ex.mple 6a: BWV 1011, Prelude, mm. 102–109.


Ex.mple 6b: BWV 995, Prelude, mm. 102–109.

Because the low notes in the original cello version are not part of an extended bass line, they don’t need separate stems and beams, thus also eliminating the need for rest signs. With regard to this, a Baroque notational convention, which unfortunately has been lost to time, was to change stem direction within the same group of beamed notes, as shown in the first four measures of example 6a. Bach did this frequently, especially when adjacent notes within a group involved large intervals. It seems that the main purpose of this practice may have been to keep note stems contained within the staff as much as possible. I haven’t seen any evidence that Bach applied this notation as a deliberate way to set apart different voice lines. Instead, the reason may simply have been to squeeze more staves on the same page by reducing the spaces needed between systems.
Baroque music characteristically increases the harmonic rhythm and energy immediately before important cadences, as had been shown in example 3b. Bach achieves this effect in the present example with the added basses and with a hemiola rhythm spanning across two measures. The hemiola emerges as a sequential four-note melodic pattern (sometimes referred to in English as an inverted “turn” or in Italian as a grupetto) that is repeated three times. Musicians in Bach’s time were expected to understand such things; therefore, no slurs or special notation were considered necessary. A composer today, however, might choose to temporarily change the meter to combine two measures of 3/8 into one measure of 3/4 or, alternatively, to use cross-bar beaming as shown in example 6c.


Example 6c.

The second violin sonata is another especially useful work from which to learn aspects of Baroque style because it also survives as a version for the keyboard, BWV 964. Example 7a, from the third movement of the violin version, is a revealing example of the limitations of compound melody when compared to the polyphonic notation of 7b, from the keyboard version (transposed to match the violin key).


Ex.mple 7a: BWV 1003, Andante, mm. 3–4.


Ex.mple 7b: BWV 964, Andante mm. 3–4.

Example 7c is a compromise for the guitar, but done in a way that still allows the overlapping of notes so that the different voices can be distinguished.


Ex.mple 7c.

Melodic Embellishment

Example 8 is also from the third movement of the second sonata. The main staff is the violin version, while the “ossia” staff is taken from the keyboard version. (Ossia is Italian, meaning “or it may be,” a term used in music to indicate an alternative way to do something.) Here, you can see a simple way to embellish a melodic by dividing longer notes into shorter ones with neighboring tones or passing tones.


Example 8: Andante, mm. 1–2.

There are similar melodic divisions and other ornamental changes throughout the same movement, some more complex than others, as shown in example 9. The first is the violin version and the second is the keyboard version.


Ex.mple 9a: BWV 1003 (violin), Andante: mm. 9–10.


Ex.mple 9b: BWV 964 (keyboard), Andante, mm. 9–10.

Technical Concessions

In the second movement of this sonata, Fuga, we can see an instance where Bach had to accommodate a technical limitation of the violin. Measure 239, the fourth measure of example 10a, has A instead of F for the third bass note. An F would follow the natural sequence of notes in the fugue subject, but that pitch is outside the range of the violin. The keyboard version restores the F, and this is certainly the note that guitarists should use, as shown in 10b, since this pitch is available to us as well.


Ex.mple 10a: BWV 1003 (violin), Fuga, mm. 236-239.


Ex.mple 10b.

Going back to the Andante of BWV 1003/964, the chord on beat three in the first measure of example 11a has D in the tenor voice, as played on the violin, whereas the keyboard version in example 11b has Eb. Although Eb is possible to play on the violin, it is a four-finger compressed chord that makes B, the note that follows, very awkward to reach.


Ex.mple 11a: BWV 1003, Andante, mm. 24–25.


Ex.mple 11b, BWV 964, Andante, mm. 24–25.

Both chords are, in a sense, secondary dominant chords, that resolve deceptively to a cadential 6/4. The D7 chord in the violin version is the dominant 7 of G, but it resolves to a second inversion C chord: the bass goes where you would expect, but the top is a surprise. The Eb in the keyboard version makes the chord a viiº7 of G. Both chords are theoretically “correct,” although the D7, to my ear, is less pleasing, probably because it is a second-inversion chord that is not a neighbor, passing, or cadential 6/4 chord. The viiº7 in the keyboard version (F#-A-C-Eb) is in first inversion, with A in the bass. I prefer this harmony because I think it is not only more intense, but also is more effective when resolving to the next chord. Nevertheless, since each harmony has its merits and each is valid, I would show both in a guitar arrangement with a small ossia staff above the main staff, as shown in example 11c. Notice, also, that I have re-voiced chords from the keyboard version and adjusted time values so that these notes are playable on the guitar.


Ex.mple 11c.

The Fuga from the first violin sonata (BWV 1001) also exists in a version for the lute (BWV 1000), as well as for the organ (BWV 539). Only the violin version should be regarded as authentic; however, the lute version, written in tablature by Bach pupil Johann Christian Weyrauch, may also stem from Bach’s influence. Nevertheless, some of the changes found in the tablature seem to result from idiomatic concessions for the lute. For instance, the range of the violin version extends to the high F, as shown in example 12a. This note on the lute, however, is beyond the neck length of a standard ten-fret instrument; therefore, D# is used instead, as shown in example 12b. Again, it is important to recognize that a change like this is often not a matter of choice, but of necessity, and they do not apply to an instrument without those limitations.


Ex.mple 12a: BWV 1001 (violin), Fuga, mm. 40–42.


Ex.mple 12b: BWV 1000 (lute), Fuga, mm. 40–42.

Example 12 also provides an opportunity to discuss a major difference in how most musicians, other than guitarists, perceive written note values. Most instruments are designed so that the player must continue to activate a note for the length of time it is intended to last. A violinist must keep drawing the bow across a string, and a woodwind player must continue to blow air, or else the note will end. For these instrumentalists, sustaining a note is an active expenditure of energy, while ending a note is a passive release of energy. For guitarists, however, it is often the opposite: an open string or a note within a chord position will continue to vibrate unless physically stopped by the player.

Most written music notation is designed to show the active sustain of notes, but not the residual decay of those notes in a natural way. Look again at example 11a. The sixteenth notes in the bass correspond to when the violin bow must leave the string to play notes on other strings; they do not indicate that the violinist should stop the string from vibrating. This may be one reason why Bach did not include rests in his string music except when needed to clarify the rhythm of a line. When writing in polyphonic notation for instruments like the harpsichord and the lute, however, he still often wrote an articulate bass line with short notes separated by rests. We do not know to what extent musical and technical concerns may have factored into his choice for this articulation; but one possible explanation is that it was to reflect a technical rhythm in releasing notes on the keyboard.

My personal perspective is that we should not ignore the written rests, as some guitarists still do, but we should also not feel obliged to follow the rests explicitly, especially when doing so requires a physical action that could abruptly stop the vibration of a note in an unnatural way. I think latitude can be given to the performer to determine the articulation that most effectively suits the music, the instrument, and the ambient acoustic of the performance venue. To be specific, I think that notes on the harpsichord have such a strong attack compared to their sustain, that it works well on that instrument to have short, well defined bass notes that end exactly on beats. But the guitar sounds better when you let the basses overlap a little into the next beat. Bass notes that are stopped precisely on the beat often sound “clinical” on the guitar. They draw attention to themselves because they don’t have the right amount of natural decay.

Returning to example 12a, it is particularly interesting to see how Bach chose to notate this passage and the surrounding phrases. Here, the bass is functioning as a pedal tone; but it is impossible to sustain the repeated D as half notes on the violin. In passages like this, the longer bass note values are not to be taken literally, but may reflect an ideal sonority that the player should try to achieve by whatever means desired. Some violinists, including Baroque music specialists who play on period instruments, alternate the bass note between chords, as in example 12c, while others arpeggiate these chords such as in 12d.


Ex.mple 12c.


Ex.mple 12d.

In summary, coming full circle, transcriptions and arrangements both have a place in our study of Bach’s music, and both serve useful purposes. Good transcriptions make his music accessible on other instruments through thoughtful choices made by the editor, such as suitable keys and modern notational standards while still being as faithful as possible to the original. Good arrangements, while having more editorial intervention than transcriptions, should be based on a thorough understanding of stylistic and theoretical values espoused by Bach himself, and without the editor succumbing to temptations to overreach and over-interpret. By studying Bach’s own models, we endeavor to understand how and why he did things a certain way and, in trying to apply his methodologies, we come to fully appreciate the true depth of his genius.

[1] Richard Wright, J.S. Bach, Cello Suites 1 – 4 (London: Cadenza Music, 2008), 2.
[2] Bach Dokumente III, 293, cited by Richard Troeger, Playing Bach on the Keyboard: A Practical Guide (Cambridge: Amadeus Press), 91.
[3] This overlapping also is implicit in the so-called style brisé (“broken style”) of some lute music, or in keyboard music that is designed to imitate the lute, in which arpeggiated textures are allowed to sustain and blend together. Pieces of this type do not have fully developed individual voice lines, but instead combine various melodic figures, and therefore are best notated as a single line.

7 thoughts on “Playing Bach on the Guitar (by Frank Koonce)

  1. I am just beginning my study of playing Bach on guitar, and this is a marvelous introduction to the subtleties involved–and the abundant possibilities. Thank you so much.

  2. While it is joyous to play Bach on guitar, it is self-indulgent to offer a mere transcription to an audience. The cello or violin’s bowed notes contain worlds that we guitarists’ plucked ones can barely suggest. Arrangements that exploit the guitar’s genius for harmony can justify our attempts.

  3. The article is tiresome, pedantic, and states and restates the obvious.
    In his own publication of the first three Cello Suites, suites one and three are little but Duarte’s transcription with many of the bass notes removed. It seems more effort went into the 12 pages of liner notes at the end of the book than in coming up with something closer to what Bach, himself, might have done.

Leave a Reply

Your email address will not be published. Required fields are marked *

Please fill in the number below: *