Bizet pure – The score of “Carmen” freed from retouches: an interview with the editor Paul Prévost on the differences between the versions and editions.
The new edition of the dialogue version of “Carmen” makes the different stages of the work available for performance: what are the most important stages of the composition?
Put in simple terms, there are three main versions. The first is the one which Bizet composed at his desk in 1874, before he had contact with the director of the Opéra-Comique or his future performers. At that point in time the famous Habanera did not yet exist, the choral parts were full of counterpoint, and the work contained numerous melodramas. The limited abilities of the chorus singers, but above all the wishes of Célestine Galli-Marié, the singer of the title role, led to considerable reworkings. These were made step-by-step and almost continuously during the rehearsals. At the end there was a second version which was probably very close to the version of the premiere, and which Bizet captured in his vocal score of 1875. This was long regarded by musicologists as the “true Carmen”, as it was the last which Bizet oversaw. Nevertheless, it is less substantial than the previous version: the melodramas were lost, the chorus numbers were simplified, and numerous passages which were felt to be less dramatic or simply too difficult to perform were shortened. The second version, however, ultimately contains the Habanera, a number which was hurriedly but brilliantly adapted from an existing song by Sebastián Iradier. In our edition, the main versions of the dialogue opera of 1874 (original version) and 1875 (version of the premiere) and the transformations which took place between are published as the 1874/75 version.
In the 19th century it was usual for a successful dialogue opera to be reworked as a fully-sung work. This opened up the possibility of “exporting” the opera abroad, where spoken dialogue was difficult to translate or adapt to fit the music. As Bizet died on 3 June 1875, his friend Ernest Guiraud took over the task of replacing the dialogues with recitatives. This was the third main version as was created in 1876, and was to make “Carmen” world-famous. Guiraud took over the second version of 1875, altered numerous details and then inserted his mainly very short and conventional recitatives between the musical numbers. The new edition of the recitative version is in preparation.
If an opera house wants to perform “Carmen” in the widely-known “usual” last version – whether with dialogues or with recitatives – which differences does your edition reveal compared with previous editions?
You may indeed ask why a new edition was necessary. Today, sources are available which were not available to earlier editors, including to Fritz Oeser whose commendable edition has been widely used. It is clear that, despite its novelty for present-day audiences, the first version of 1874 remains an “unusual rarity”: without the Habanera, but with another musically and dramatically attractive opening aria. By contrast, the 1875 version, the so-called “true Carmen”, only exists as a vocal score by Bizet himself, for the orchestration had to be reconstructed from the original sources of the Opéra-Comique from the copy of the orchestral score and the complete, rediscovered set of manuscript parts. In earlier editions, all the versions were happily mixed. For example, the tempi were frequently altered: if a conductor finds several different agogic and metronome markings (some in parentheses, others in square brackets), they do not know which to choose. Or, to put it more precisely, it is not clear which were by Bizet, which were by Guiraud or another person, as after Guiraud several more anonymous alterations were made in the vocal scores.
Several modern editions mix the dialogue version of 1875 with Guiraud’s recitative version of 1876, which creates the impression that the recitatives only have to be added or removed in order to change from one version to the other. But the differences between the versions also lie in a multitude of details, such as the articulation in the vocal and orchestral parts and the dynamics as a whole. Therefore, the sources have had to be individually examined, section by section. For example, in Bizet’s “autograph” manuscript, we had to distinguish between the composer’s original handwriting and many alterations to details in another hand. The edition also finally departs from the standardization which some publishers adopted as the only valid approach. My approach has been as follows: the musicologist makes an inventory of the versions, and the interpreter makes the choices.
Can you give a few specific examples of the alterations made by Guiraud?
Guiraud’s orchestral retouches stem from certain ideas of standardization, indeed from an evident academicism. In the Habanera, for example, Bizet accompanied the beginning of Carmen’s aria with just a triangle, only adding the tambourine to the triangle in a second step. Guiraud calls for the tambourine throughout, thereby creating a more unified, but less subtle colour. In the Séguedille Bizet has the last statement of the theme accompanied by the bassoon, which doubles Carmen’s voice: the sound of the mezzo-soprano voice, coloured with the bassoon, is novel. Guiraud omits the bassoon and makes do with the string accompaniment already heard a great deal since the beginning of the number. He undoubtedly wanted to avoid an octave effect between a low register and the voice, which he now offered with a soprano. This is a further special feature of the 1876 version by Guiraud, which suggests numerous ossias and even transpositions for the role of Carmen in order to adapt it for a higher register.
The copy, which was preserved in Vienna for the first performance of the work at the Imperial Court Opera, provides evidence of the desire to entrust the title role to a soprano: the Habanera can thus be sung in E and not in D … as Maria Callas in fact did! Later vocal scores introduce new ossias in a higher tessitura, the authorship of which is anonymous. These naturally distort the role.
How did you, and how have other editors approached the work and its versions?
In fact, I create a critical relationship between our new edition and the “old” sources – with Bizet’s manuscript corrected by Guiraud, the original manuscript sources from the Opéra-Comique, the manuscript copy from Vienna, the Choudens editions from the 19th century, and the manuscript and printed orchestral materials. In his edition, Fritz Oeser mixed all the versions and chose what seemed best to him. Although he used Bizet’s orchestration, he nevertheless mixed this with other options by Guiraud and attempted to resolve everything in one edition, thereby creating a composite work.
After these many years of painstaking work on “Carmen”, do you have a preference for one of the versions?
A music historian doesn’t have a preference, as all the versions have their validity, bestowed either by the composer himself or by history. Guiraud distanced himself from Bizet, yet it is this version which made Carmen famous throughout the world. Instead of a preference, I perhaps have a certain regret. If a version of the dialogue opera is chosen – and here we have a choice! – the melodramas can be reinstated instead of the somewhat dry dialogues without music. In both the 1875 and the 1876 versions, I also regret that a large portion of the choral counterpoint has disappeared. This abandonment evidently took place not for aesthetic reasons, but because of practical difficulties at the Opéra-Comique in 1875; the work entered the repertoire in this form, although the problems are no longer relevant today. In addition, it has to be recognised that the 1874 and 1875 versions are not strictly “water-tight”. The transformations from one to another took place so close together that each hybridisation between them would not contradict the historical truth.
What emanates from Carmen, and what is often forgotten in a somewhat unrestrained Hispanizing enthusiasm, is the refinement, the lightness and the subtlety of the colours – qualities which are characteristic of the dialogue version and which René Jacobs showed to perfect advantage in his 2024 concerts, and which should apply to all the versions.
Interview by Marie Luise Maintz
(translation: Elizabeth Robinson – from [t]akte 1/2025)