VILLA-LOBOS' GUITAR MUSIC: Alternative Sources and Implications for Performance
by Stanley Yates, March 1997
© Stanley Yates, 1997
This article first appeared in Soundboard, Journal of the Guitar Foundation of America, Summer 1997, vol. xxiv, no. 1, pp. 7-20; and, as "Die Musik von Heitor Villa-Lobos, Alternative Quellen und Aufführungspraxis," in Gitarre & Laute, vol. xix, no. 6, 1997, pp. 55-65.
Technical Note: the low resolution gif images contained in this document are intended to assist in reading the article online. You may also download Postcript files, in zip format, for high resolution printing.
The most intriguing of the manuscripts is the 1928 autograph copy of the Etudes, which adds yet another twist to the sketchy history of this seminal and germinal work for the guitar. Each page bears the publisher's stamp "reproduit par les soins des Editions Max Eschig," and is in all respects a finished version. It is meticulously written in the hand of the composer, and contains considerable detail of expression markings and fingerings. This manuscript, however, predates several others and clearly was not the version used by Eschig as the basis for publication-the several manuscripts dated Paris 1929 are much closer in their details to the Eschig publication than is the 1928 manuscript and derive, therefore, from a revised version (or versions) of that year [Note 2]. The Eschig edition is copyrighted 1952 (with the exception of Etude 1, which is copyrighted 1953), and Andres Segovia's preface to that edition is dated January 1953. The first edition did not appear, therefore, until twenty-five years after composition [Note 3]. The publication manuscript upon which the edition was based has yet to surface [Note 4].
Why did Villa-Lobos revise his 1928 manuscript-ostensibly a finished work? And why did virtually none of his fingerings make it into later versions? It is interesting that the 1928 manuscript makes no reference to Andres Segovia who, by 1929 was the acknowledged dedicatee of the Etudes, and who later supplied the preface to the Eschig publication; it is not unlikely that Segovia suggested revisions to Villa-Lobos.
The issue of the fingerings is even less clear-in his 1953 preface Segovia defends the integrity of Villa-Lobos' fingerings, though very few are actually found in the publication:
I have not wished to change any of the "fingerings" that Villa-Lobos himself indicated for the performance of his works. He understood the guitar perfectly and if he chose a particular string or fingering to produce a certain phrasing, we must strictly obey his wish, although it be at the cost of greater technical effort.
It is possible that, since the 1928 manuscript alone is significantly fingered, Villa-Lobos may simply not have bothered to copy out the fingerings again when preparing his revision(s). Whatever the case, the 1928 manuscript provides much valuable information that, inevitably, will inform future interpretations of the music.
In this article, I discuss the major differences between the manuscripts and the published versions, and address the interpretation and performance issues they inform. I also attempt to distinguish between those differences that clearly reflect errors or omissions in the published score (and which may therefore reasonably be adopted without further discussion) and those differences that seem to be revisions on the part of the composer (and which should therefore be treated more cautiously)-the issue of a composer's final intentions is not always a simple one, as in the case, for example, of the composer who is persuaded into revisions by his editor.
In the discussion that follows, measure numbers refer to the published score; measures, beats and subdivisions of beats are identified in the form m1.1.2 ( measure1.beat1.subdivision2).
The 1928 Manuscript of the Twelve Etudes
Figure 1. Etude 4, mm. 46-47.
The technical subtitles ("de arpège," etc.) attached to many of the Etudes in the published score do not appear in the 1928 manuscript, although the first Etude does bear the provocative subtitle Prelude. It is tempting, though fanciful, to imagine this piece to be the "lost" sixth prelude from the set of 1940.
Also noteworthy is that this first study, along with numbers 2 and 9, is written without any the repeat markings found in the published score.
Solutions to Ambiguous Passages
At the end of Etude 1 an open-string is used instead of a harmonic in m32 beat 3.3; and in mm33-34 the notation confirms the use of harmonics on the e and b strings (fig. 2):
Figure 2. Etude 1, mm. 32-34.
Although the intended execution of the ending of Etude 2 has led to some controversy, the 1928 manuscript does appear to support the explanation found in the "Carlevaro" manuscript. In this latter source the following Portuguese annotation appears at the point indicated in figure 3: "Pizz. tos simultaneos da máo direita e máo esquerda na mesma" ("pluck simultaneously with the right and left hands on the same [string]") [Note 6]. Although the 1928 manuscript contains no such description, the words harm duple that appear in the published score are not present either. However, the circled asterisk does appear in the 1928 manuscript and, although devoid of any annotation, is positioned on the pitch e-perhaps indicating the string upon which the term pizz mg applies. Furthermore, the diamond noteheads are provided parenthetical accidentals which align them with the sounds produced on the first string behind the fretting finger. It would appear that Villa-Lobos' earlier intention, then, was to fret and pluck the upper notes as written and simultaneously pluck behind each fretted note with another left-hand finger pizz m.g. (in French: "pluck with the left hand")—a potentially witty conclusion to a virtuosic study! (fig. 3):
Figure 3. Etude 2, mm. 26-27.
The ending to Etude 3 is another notoriously ambiguous spot. The 1928 manuscript, however, clearly indicates that in m30 the lower pitch is not a harmonic but a normal note. Following Villa-Lobos' usual notational practice, everything then makes perfect sense: d on the a-string is played with the third finger; and the harmonic at the fifth-fret of the d-string is played with the fourth finger (and sounds at the pitch indicated above it) (fig. 4):
Figure 4. Etude 3, m. 30.
In general, Villa-Lobos' fingerings for the left hand show concern for legato connection and clarity of voice leading, as well as specific effects of phrasing. For example, a wonderful effect is produced in the opening section of Etude 11 through a combination of glissando and ligado (fig. 5):
Figure 5. Etude 11, mm. 1-3.
A technical aspect of Villa-Lobos' fingerings for the left hand is a tendency to connect distant positions by shifting rapidly along a single string, treating the relative strengths of the fingers with apparent impunity (fig. 6):
Figure 6. Etude 3, mm. 9-10.
In chordal passages, Villa-Lobos sometimes uses an unconventional second or fourth finger barré (although a third finger barré is not employed for the numerous half-diminished chords such as those found in mm11-16 of Etude 4) (fig. 7 and fig. 23, below):
Figure 7. Etude 4, mm. 8 and 31.
When present, Villa-Lobos' fingering indications for the right hand generally are orthodox: i-m alternation is used for scale passages; i-m-a are otherwise assigned to the the treble strings. In five-note chords the thumb plucks two adjacent bass strings simultaneously-when more definition between the lower voices is required, or when the basses are not adjacent, the lowest note is performed as a grace-note (as in Etude 4, mm5-6, 29-30 and 35).
However, Villa-Lobos also employs several less orthodox right-hand techniques. In the central section of Etude 12, for example, he calls for the index and middle fingers to pluck two strings simultaneously (fig 8):
Figure 8. Etude 12, mm. 38-40.
More interesting are the right-hand indications that Villa-Lobos provides in conjunction with slur markings. Etudes 10 and 11 both contain passages which involve a single finger, or the thumb, playing across one or more adjacent strings (noting that Villa-Lobos always uses a slur to indicate this technique). In the passage shown in figure 9, the thumb plays across the lower two strings and the index finger across the top four strings-both are indicated with a slur (fig. 9):
Figure 9. Etude 10, mm. 72-72; and Etude.
In the following example, the index finger plays across the top five strings, as indicated; the four-note ascending group almost certainly is intended to be played with the thumb (fig. 10):
Figure 10. Etude 11, m. 19.
Take a look now at the passage from Etude 1 shown in fig. 11 (and bear in mind that the 1928 manuscript does confirm the right-hand fingering that appears at the beginning of the piece in the published score). Do the fingerings suggest that in m24.3.2 the index finger plays across the second and third strings? (fig. 11) [Note 7]:
Figure 11. Etude 1, mm. 24-25.
If intentional, this technique may help explain some of the ambiguous slurring found in both the manuscript and the published score, as discussed below.
In many cases, the intended function of the slur is obvious-the grouping slurs in mm33-37 of Etude 8, for example (which serve to divide the phrase into the two parts implicit in its construction) (fig. 12):
Figure 12. Etude 8, mm. 33-37.
Also obvious in this example is that the first phrase is to be performed ligado. Analogous passages (ones in which a large grouping slur encloses several internal ligados) may also be found at mm56-57 of Etude 8 and mm20 and 45-50 of Etude 10. But what are we to make of the slurring found in the following example? (fig. 13):
Figure 13. Etude 7, mm. 8-10.
In measures 4-11 of this Etude, ascending slurs are independently marked within the larger grouping slur (clearly, a short ascending slur should also be present at mm8-9). Comparing these scale passages with the articulated scale that appears at m56, there is the implication that the notes falling under the large slur should be performed as ligados. However, a short ascending ligado followed by an articulated scale seems the more likely interpretation-the scale at m56 being articulated metrically, those under slurs as a single gesture.
Ambiguity is also present in shorter slurred groupings, especially in Etude 2. Although the 1928 manuscript contains several divergent slur markings for this study, their interpretation still remains uncertain. Of those that appear on the first and third beats of each measure, most seem to be ligados. Others, however, could not possibly be performed that way (fig. 14):
Figure 14. Etude 2, mm. 14-15 and 18-19.
Perhaps these slurs indicate that the thumb or a finger be "dragged" across the indicated strings (as noted above)-this works reasonably well in some ascending groupings, but seems entirely unnecessary in most descending ones. There is also the strong possibility that the slurs define melodic grouping-that is, the notes under the slur should not ring over one another. And perhaps some are oversights, inadvertently added under the momentum of the slurring in preceding measures.
Although consistently marked, the slurs in Etude 9 are also ambiguous. In the 1928 manuscript, mm30-59 are slurred as shown in figure 15: the slurring shown in m30 appears to combine a grouping slur with a re-articulated descending ligado, while that at m51 indicates the articulation of repeated notes on the second string. In the first case the lower slur suggests performance as a single gesture that combines arpeggio and ligado (as typically performed). However, the 1928 tempo indication Moins, as well as a slower initial tempo (Un peu animé), may well indicate a literal realization of the figure.
Figure 15. Etude 9, mm. 30 and 51.
A table of divergent slur markings has been provided at the end of this article.
Divergent Pitches and Rhythms
Figure 16. Etude 4, mm. 17-18.
A table of divergent pitches and rhythms may be found at the end this article, along with an opinion in each case as to the legitimacy of amending the published score. Among the numerous inconclusive divergent pitches and rhythms listed there, I find the examples that follow particularly interesting.
In Etude 5 (fig. 17): in m9.4.2 the melody note f seems to fit well with the circolo character of the melody thus far; at m10.1.2 the ostinato requires the pitch b, but e has been substituted-perhaps to relieve the dissonance otherwise produced; at m22.4 again b is the required note for the ostinato, but b@ does provide more movement over the barline; in mm27-28 the additional first-string e results from a double ligado in which the first finger of the left hand plucks both first and second strings; and in m48.3.2 e@ does seem to be the correct note, harmonically-with or without additional basses.
Figure 17. Etude 5, mm. 9-10, 22-23, 27-28 and 48-49.
In Etude 6: f natural is indicated in mm2 and 3 beat2.2 and at mm28 and 29 (although not at mm56 and 57); and Villa-Lobos originally had a different texture in mind for mm33-41, both fifth and sixth strings probably to be played with the thumb (fig. 18):
Figure 18. Etude 6, mm. 33-34.
In Etude 8: the somewhat elusive character of the published opening section is transformed by the consistent glisssandi and triplet rhythms employed in the 1928 manuscript-the effect is almost jazz-like; and in mm47-48, a simpler harmonic texture does not incorporate the appoggiaturas of later versions (fig. 19):
Figure 19. Etude 8, mm. 1-4 and 46-49.
In Etude 10, mm63-64, the sixteenth-note figures substitute the open e-string. Again, the character of the passage is altered by the more relaxed effect (fig. 20):
Figure 20. Etude 10, mm. 60-64.
Tempo Indications and Expression Markings
The 1928 manuscript also shows differences for internal tempo changes in several of the Etudes.
Expression markings (dynamics, fluctuations of tempo, and articulation) are more detailed in the 1928 manuscript than in the published score, and often clarify form, phrase structure, texture, and motivic character. A good example is the opening section of Etude 8 (which is devoid of dynamic markings in the publications) , as can be seen in figure 21: the upper and lower parts are given independent dynamic and articulation markings; written decrescendos shape the opening glissando motive in mm1-4, as well as its expanded version that follows through m14; the lower voice is independently shaped sf-p and sfz-mf in mm10 and 12; the subtle contrast between rallentando and ritardando is exploited in mm13-14.
Figure 21. Etude 8, mm. 1-15.
In addition to being more detailed, the markings in the 1928 manuscript are often more expressive than those found in the published score-rallentandi and crescendi are applied over longer spans, and a greater number of dynamic contrasts are applied. Although it is not possible to list every divergent expression marking found in the 1928 manuscript (they simply are too numerous), a table found at the end of this article lists the more pertinent ones.
Figure 22. Etude 10, new material [mm.1-4].
Quite considerable redistribution of material is also found through measures 29-47 of Etude 11, and includes some material that was not retained in later versions (fig. 23).
Figure 23. Etude 11, new material inserted at mm. 39 and 42.
Although the omission of this extra material in all other sources does serve to increase the concision of the Etudes in question, the material is of such interest that reintroduction seems a justifiable option.
In Etude 7, one measure is added-measure 10 is inserted before measure 40 (agreeing with the opening section).
Manuscripts of the Five Preludes
Figure 24. Preludio No. 2 para Violão, mm. 33-35.
In the fair copy of Prelude No. 1, the passage at mm43, 47, 122 and 126 has a harmonic notation for the top three strings only the second time it appears (i.e., in m47). However, in the first appearance of the passage in the compositional sketch harmonics are present (and are followed by an indication to play in the 7th position).
Figure 25. Preludio No. 1 para Violão, compositional sketch, m. 43.
The compositional sketch of the first Prelude, very obviously written at the moment of inspiration, provides us an interesting glimpse of the genesis of the piece. The initial tempo is marked All' agitato (which later became Andantino espressivo), and the middle section is marked Meno (this later became Più mosso)-these earlier tempos enrich our interpretation of the revised ones. Here is the opening of the Prelude as notated in the sketch, along with Villa-Lobos' initial idea for the central section; note the prime importance of the melodic material as the structural base of the piece, and the simple waltz-like conception of the accompaniment (which later became a syncopated "catch-all" to the revised melody) (fig. 26):
Figure 26. Preludio 1, compositional sketch, mm. 1-9 and .
Along with a fair copy manuscript of Prelude No. 5 dated September 1940, which agrees with the Eschig publication in virtually all matters (with the exception that it is written without expression markings), is another fair copy which reveals an earlier completed version of the piece. In addition to several small changes in the first two sections (fig. 27), the third section employs an entirely popular-sounding harmonic style (fig. 28).
Figure 27 Preludio No. 5, compositional sketch, mm. 9, 11, 15-16 and 24-25.
Figure 28. Preludio No. 5, compositional sketch, mm. 33-42.
A Final Thought
Table 1. Divergent Pitches and Rhythms found in
the 1928 Manuscript of the Twelve Etudes
I have established Villa-Lobos' calligraphy with reference to
several signed documents and autograph scores, among them the compositional
sketches of the Five Preludes (Museo Villa-lobos P.201.1.4), and the "Lubrano"
manuscript-a 1929 manuscript of Etude No. 5, advertised and authenticated by
the antique firm J & J Lubrano (see Matanya Ophee, "How does it
end?" Classical Guitar, May, 1995, Vol. 13, No. 9, pp. 14-22). Several
characteristics show these manuscripts to be Villa-Lobos autographs: the
calligraphy of Villa-Lobos initials and signature (particularly the shape of
the letter "H") and the crossing of the letter "T" (which
increases in pressure as it ascends); the calligraphy of Villa-Lobos' treble
clef and sharp sign (which he crosses, unusually, downwards from left to
right); and peculiarities of Villa-Lobos' music notation such as the notation
of strings by letter instead of by number, circled right-hand fingering
indications, and harmonics indicated at fretted rather than sounding pitch.
In his article "Villa-Lobos: New Manuscripts" (Guitar Review, Fall
1996, 22-28), Eduardo Fernandez refers to additional manuscript copies of the
Etudes housed at the Museo Villa-Lobos, but does not offer information
relating to their chronology. Among these is a set included in the Guimarães
collection-a manuscript collection donated to the museum by the family of
Villa-Lobos' first wife (the Museo Villa-Lobos has not been able to supply me
with any information regarding the chronology of these scores). A number of
additional manuscripts, including one in the possession of Abel Carlevaro and
a Museo Villa-Lobos manuscript of Etude 10 (P.200.1.19), dated 1929, appear
not to be in the hand of Villa-Lobos. The "Lubrano" manuscript of
Etude 5, cited in footnote 1 above, is dated 1929 and almost certainly is an
H. Villa-Lobos, Douze Études, preface by Andrés Ségovia, (Paris: Éditions
Max Eschig, 1953?). An edition of the collected solo guitar works of
Villa-Lobos with a "correction of obvious typographical errors" was
published as Heitor Villa-Lobos Collected Works for Solo Guitar, (New
York: Amsco Publications, 1990).
At the time of writing, Editions Max Eschig have not responded to my requests
for information relating to the manuscript used for publication; nor has the
Museo Villa-Lobos been able to supply any information.
The Eschig edition of the Douze Études does make an attempt to distinguish
between the two sizes of noteheads, although subtly (see, for example, page 10
of the publication). The 1990 Amsco Publications edition, obviously
re-engraved from the Eschig publication rather than the manuscript itself, is
oblivious to them.
The passage is reproduced in Abel Carlevaro, Guitar Master Class, vol.
III. Heidelberg: Editions Chanterelle, 1987, p.12; but also see Matanya
Ophee's article, loc. cit. In the opinion of the present writer, and judging
from the available evidence of the 1928 manuscript, the confusion surrounding
the ending of Etude 2 most likely is the result of a revision - in 1928 the
intention had been for a "bi-tonal' ending but was later changed to
harmonics. The confusion arises from the unfortunate inclusion in the
published score of both the original indication "pizz mg" and the
new indication "harm duple" (which, incidentally, is set in a
different type face).
In the Fernandez article, loc. cit., the position indication VII is displaced
two notes over to the right (over the b); in the 1928 manuscript, however, the
sign is clearly
over the g.
Copyright © 1997 by Stanley Yates