Different Kinds of Fidelity: The Canzonetta from Mendelssohn’s First String Quartet

Guitars International recently released a beautiful video of Hao Yang playing the Canzonetta from Mendelssohn’s First String Quartet, op.12.

The transcription Hao plays is one that I made for her: in fact, some passages in the final version reflect her preferences among various options I proposed. As we went through the string quartet original, I was quite surprised at how many passages present difficult choices that affect how the music hangs together: you could say that each decision represents a different way of being faithful to Mendelssohn’s score. The publication of Hao’s video seemed like a good opportunity to share the transcription and reflect a bit on the choices made along the way.

To start with, then, here is the transcription for download in PDF. The original quartet score is available from IMSLP. This mischievous piece was an immediate hit after Mendelssohn published it in 1830. In the nineteenth century, it was already sometimes performed on its own in versions for different instruments: many of these historical transcriptions are available on IMSLP (look under the tab ‘Arrangements and Transcriptions’).

As for the guitar, Tárrega was the first to make a transcription (the score, published by Alier in Madrid around 1925) is also on IMSLP, both in its first edition and in a manuscript copy by Llobet); Segovia recorded it, adding personal touches as he always did; Julian Bream goes back to the original – as he always did – to make a more complex and virtuosic version of Tárrega. (To listen to these various versions, here is a playlist on YouTube.)

‘My transcription’ might be a bit of an overstatement. A transcription is not really new unless it demonstrates a new concept, and I’m not sure if it would have occurred to me that this piece could work on the guitar without Tárrega’s version to start from. But as I went through the score, I did in fact see some opportunities to try something new, particularly with regard to the role of the bass. Let’s dive in. 

The Outer Sections

This canzonetta (a light song or ditty – nowadays, even ‘pop song’) combines with a trio section to make a traditional ternary form: Song – Trio – Song (reprise). The reprise varies the song in delightful ways, some involving timbre. In the original quartet version, Mendelssohn uses pizzicato frequently and cleverly – often mixed with arco on other instruments, as at the very beginning. Here, with the viola plucking along to the melody, Mendelssohn is likely thinking of the guitar, evoking the sound of the street:

We guitarists have our own pizzicato effect, of course, where we muffle the strings with the side of the hand. Trying to use it to mimic Mendelssohn’s mixed effect is probably not worth the trouble, but our pizzicato certainly has a place elsewhere in the transcription. In this respect, Segovia’s transcription is instructive: instead of using it in the same passages as Mendelssohn, he applies it to passages that suit the guitar, such as the beginning of the song’s second section (bars 15–18). I find it fascinating when transcribers respond to the idioms of the original in an imaginative way, but in this case I wanted to see what would happen if I followed Mendelssohn’s use of pizzicato as closely as possible, at least where all the instruments of the quartet use it together. Here is the ending with Mendelssohn’s pizzicato applied:

The Trio

Turning to the middle section, the trio, the topic switches to harmony and voicing. Tonal harmony relies on the framework of the outer voices: melody and bass. When adapting complex textures for the guitar, we generally try to preserve the outer voices and thin out (or even remove, at times) the inner voices. If you’re having difficult making a passage work in transcription, it’s a good idea to reduce it to two voices and then see what else you can manage in between.

Unfortunately, that approach doesn’t work here. If fluency is your goal, it’s not really practical to preserve Mendelssohn’s treble and bass together throughout: here and there, one or the other has to be sacrificed. But which? 

Tárrega has a clear policy: wherever necessary, he rewrites the bass. Listening to the trio, however, I wondered what would happen if one did the opposite. After all, the upper voices contain figuration made out of little scraps of motives that dance over the strong structure of the bass line. This makes it possible to reinvent the treble a little while keeping the bass unchanged.

For a spectacular example of how Mendelssohn uses the bass to steer the harmony, let’s look at the middle section of the trio. Here, the cello ascends a full octave from F♯ to F♯ (in G, the original key):

Above each bass note, Mendelssohn places either root-position or first-inversion sonorities. This is easier to see if we take out all the figuration and show only the chords and voice-leading:

As an aside, notice the two roman-numerals underneath the reduction – only two. We might say that a scale such as this prolongs the dominant harmony, D major: that is to say, Mendelssohn has found a way to keep the D-major harmony – and its tension – in force without just sustaining it or repeating it literally. Instead, he sounds the harmony once at the beginning of the section and once at the end, an octave higher. Then he fills in the space, with a scale, and each note of the scale supports a chord. In an early theory class, we might be required to label every chord with a roman numeral as an exercise; but when preparing a piece for performance, it can be helpful to use roman numerals only for harmonic pillars, omitting them for chords that fulfill a more connective function.

It is worth comparing this progression with the standard renaissance schema for an ascending bass that alternates 5ths and 6th, checkerboard fashion, above every bass note:

Evidently, Mendelssohn is playing with this foundational pattern, and so I wanted to try everything possible to preserve it. Along the way, I stumbled on an additional benefit, for if the bass starts this passage on G♯, following Mendelssohn, it becomes possible to imitate the dialogue between the violins and the lower strings. See again example 3 above (for the quartet original) and compare it with the rendering below:

Tárrega approaches this passage quite differently, putting all of the chords in the first eight bars of the sequence into root position:

The outer sections of the trio are no less challenging to fit onto the six strings. They again come out very differently if Mendelssohn’s bass line is retained, but to achieve this, one has to be very free with the passagework above. The most crucial moments are the end of each section – that is, the cadences. Here’s the first section in the original score with the bass’s cadential figure highlighted:

In his transcription, Tárrega has just a low E in the bass, giving the player the freedom to play the passagework in the treble: 

Did Tárrega have much choice? After all, the upper part consists of virtuosic scales: if we try to put in Mendelssohn’s bass line, a good deal of freedom is lost just when you want to be at your most open and relaxed. Hao and I tried it anyway, but we quickly decided that another solution had better flow: we keep the bass, we play a typical cadential formula in the treble, and we delay the scales by one bar. Here’s a direct comparison between the two choices: the upper staff, in small notes, is what Hao plays in her video:

On paper, it might look rather high-handed to displace the violin line in this way, but we felt that in performance, the differences get lost in the virtuosic flow of the music. Of course, it’s important that these sections still have the same number of bars: the phrase lengths can’t be changed.

What’s at stake here is competing notions of fidelity. Comparing Tárrega’s version with the small staff in example 10, it might seem as though Tárrega is closer to the original; after all, he has thrown away just one element (the bass), while I’ve taken liberties with several elements. On the other hand, I’ve preserved the harmonic structure – directed by the bass line, with its activity, its voice-leading and its way of saving low E for the final chord – without creating awkwardness. I resort to the same trick at the end of the second repeated section of the Trio.

Another moment worth noting occurs in the trio’s reprise, where the E-minor sonority of the first section gets transformed into a luminous E major (see example 3 above, three bars before the end; transposed to A major on the guitar, this becomes an F♯-major chord). Here I wanted to capture the resonance of the original, so I simplified the top voice (violin 1), and incorporated a bit of violin 2:

For comparison, here is Tárrega’s realization of this passage. Here, actually, one sees that in bars 66–67, Tárrega has done some recomposing of his own, introducing a tinge of E minor with the B–C♮–B bass figure. Mendelssohn has C♯ here and a very bright harmony, but that would not sound good with Tárrega’s voicing. For these bars I was able to keep C♯ by making small changes to the figuration (see again the end of example 11 above). 

Keeping the Faith

Only the simplest pieces can be ‘imported’ from another medium onto the guitar with no changes. As we develop the transcriber’s skill, it’s natural at first to have a rather quantitative view of fidelity to the score: if one version changes only 5% of the notes of the original, whereas another changes 10%, we might consider the former more ‘faithful’ than the latter. As we gain in experience, however, we have to contend with different kinds of faithfulness. Fidelity doesn’t always reside in the notes on the page, and sometimes a little rewriting better captures the flow and character of the original. As Dryden puts it, taking about the parallel art of translation (in the same year that Bach was born):

For, after all, a Translator is to make his Author appear as charming as possibly he can, provided he maintains his Character, and makes him not unlike himself (Preface to Sylvae, or the Second Part of Poetical Miscellanies, 1685)

This transcription was one of several I made for Hao Yang of romantic pieces, to make a ‘romantic album’ that might precede a performance of Nicholas Maw’s Music of Memory, which is based on the Allegretto from Mendelssohn’s Second String Quartet. I’ll share some of these other transcriptions in future posts. 

Further Reading

Chapter 10 of Guitar, ‘Unstable Texts’ looks at editions and transcriptions, teasing out some of the issues discussed above using Albéniz’s Granada as an example.

Some of the best writing I’ve read on transcription is by Katalin Koltai. Two of her articles, ‘Breaking the Matrix’ and ‘The Tradition of Illusion’, are published in Soundboard Scholar. David Harvey sometimes writes about transcription on his blog, always insightfully, as in his article on Albéniz.

But the best way to learn about transcription is to play through transcriptions and compare with the original. There are so many interesting publications out there it would be invidious to single out a few, but one learns the most from approaches that are provocative and deal with questions of fidelity in an imaginative way, whether or not you agree with the results. Thinking of guitarists with whom I’ve discussed transcription in person, I’d recommend studying transcriptions by Stanley Yates, because his editions often contain prefaces and notes that help one understand the philosophy behind his choices. And now Alan Mearns is demonstrating a similarly unfettered approach in which each piece is treated as a unique problem to be solved. Roland Dyens’s arrangements of Chopin in his collection Mes arrangements à l’amiable show what happens when resonance is put above the details of the score.