Magdalena Duhagón comments: “I think of the manuscripts of composers of whose works we have recordings in which they play their own works, and their versions are quite different from the editions we know: Barrios, Villa-Lobos, Dilermando Reis. Besides some technical problems in the recordings, sometimes the works are so different in their hands, that some rhythms are completely different from the score, and sometimes entire passages. In these cases I follow the recordings. I’d like to know what do you think about these cases.”
Good point and it deserves to be considered. It is really an extension of the “Urtext mania”, only Magdalena seems to consider the recording as even more “Ur” than the text. It is interesting that the composers mentioned are all from South America, and it is no accident, because there is a big chunk of the repertory from SA that comes from folk traditions. I have addressed this issue in a series of articles for “Il Fronimo”, but it does no harm to insist on some basic ideas.
We have basically two very different situations here. There are composers who come from “folk” or “traditional” music, and composers who come from the “classical” tradition. The first ones (Barrios, Reis, also Gentil Montaña, and many others in South American guitar literature, but by no means all of them) tend to think of the manuscript (or edition) as capturing one possible version of the work, just like a photography captures one version of a person. They feel free to take liberties and make changes with the details, because for them a piece is a kind of Platonic archetype, which will always be incarnated somewhat imperfectly when it is played; they do not see the differences between versions as differences in text, but as variations in the interpretation of the work, much as it would happen with the ornamentation of a Baroque piece. The same goes for a recording, obviously; it has nothing sacred, it is merely the register of a version at a particular time and place. The edition, when and if it exists, might be quite different; there might be multiple authentic manuscripts with significant differences. What to do? I would say, do the same you would when choosing a photograph of a relative: choose the one you like and never mind if it is printed, recorded, or not. In this tradition, the text is not necessarily sacred, and the latest version is not necessarily the best one. But just stick with one version, because you cannot be sure if the author would have combined two or more different versions. Don’t create a Frankenstein monster combining all the little bits you like, unless you really know the context and the genres and the mannerisms associated with them. For an edition of this type of composers, the best solution would be to print all available manuscripts (yes, I know, it will never happen).
A special case of this is when the only editions we have are taken from recordings, because the composer never wrote out the piece. The most notorious case is Barrios’ “Caazapá”, and to a lesser degree some of his works which use the folk genre “estilo”. No edition has been able to present a convincing written version of Caazapá, not because it is not possible to do so, but just because the editors had no idea of the original folk genres Barrios is using. In Caazapá, the underlying basic pulsation is always the same, and it is quite simple to write it out using proportional notation. It might look complicated on paper, because there will be measures or sections oscilating between ¾ and 6/8, and/or between 12/16 and 4/4, but it is perfectly doable. With “Luz Mala” and other estilo pieces, it is just a question of knowing the forms of the genre: there is a “recitativo” section, based on octosyllables, and instrumental interventions between each strophe, usually of 10 verses. Here, context is more important than text, and listening to some good estilo singer (including the early Carlos Gardel) is more helpful than a million words. By the way, Llobet made also mistakes of notation when trying to write an estilo. So if you see indications of “rubato”, beware. Most likely, the editor did not know what to do with the notation.
In the “classical” tradition, things are quite different. Text trumps context, basically. The composer makes the decisions and the final version is the definitive one. The Villa-Lobos question is surely about the “recording” of Prelude Nº 1, available (I think) at the Museum VL. It is a home recording, and it seems to me pretty clear that VL was just fooling around – maybe he had not yet decided on a final form for the piece. But once the composer decides, fixes it in writing and publishes, that’s it. And Heitor did decide. To use a recording in preference to the published version is not a good idea. Of course, the composer might change her/his mind (more on this later), do a new version and maybe even have it published, if the publisher agrees. Basically, the latest version is the valid version. So forget about manuscripts and recordings: Prelude Nº 1 was published – and actually, the published version is a lot better than the semi-improvisations of the home recording. Here there are no questions of editing by a guitarist: VL knew his way around the guitar very well. A recoding by a composer is always interesting: it might give you some useful ideas on the character of the piece, the intended tempo and other things, but it is not sacred. Composers are not always the best interpreters of their own works.
Even in the “classical” mode, things can get more complicated, and sometimes a work might exist in two different and (more or less) equally possible versions. This is why every string quartet which records Beethoven’s op. 130 includes the “Grosse Fuge”, to give you a chance to listen to the quartet with the “original” ending. But when they play the quartet, they use the last movement that Beethoven composed as substitute for the Fuge; they do not end with the Fuge in the name of “authenticity”, because they know the rules quite well. There are cases in Bruckner or in Schumann where there might be more than one “legitimate” version of a work. Or take the case of Boulez’s works, where sometimes several successive versions of a work were published – of course the final one is “the” version, but it might be interesting to play an earlier one, as long as you announce it as such. But the rules of the game are not the same as when playing folk-inspired music, even when the work happens to be inspired in traditional genres, as it is in Guastavino or Ginastera. In these cases, the composers are just using material from folk sources, but the material is inserted in a completely “classical” mode. So, it is the composer who decides, not the context, and you cannot take liberties with the text as if it was a folk song. You should not mess with the text in Ginastera in the name of context any more than you would with Bartók or Stravinsky. But knowing the context will probably help you to play it better.