Michael Talbot

What does one expect from a critical edition of a baroque opera?

To be sure, most of the questions to be asked and resolved concerning a critical edition, or purely a performing edition, of a baroque opera are the same as those that apply to the equivalent kind of edition of music in any genre. My focus today, however, will be on the problems peculiar, or very nearly so, to the edition of operas.

The first distinguishing feature of an opera is that it is very long, which in terms of a musical score means very bulky. In the baroque period it was common to bind each of the three acts separately, which reduced the size of the object placed by the maestro al cembalo on the music stand before him, and made it less likely to come crashing down on his fingers. (There was of course another practical reason for treating each act almost like a separate work: it speeded up the process of copying and rehearsal.) This lesson about making an edition of an opera compatible in size and weight with the music stand of a harpsichord has not been learnt by all. One remembers with a shudder Raffaello Monterosso’s de luxe edition of Vivaldi’s opera La fida ninfa, which is almost too heavy and bulky to carry under one’s arm, let alone place on a music stand.

The second distinguishing feature is that its user community is smaller and more specialised than that for any other kind of seventeenth- or eighteenth-century music. When one edits a violin sonata or a sacred vocal piece with choir, one attempts to make the result suitable for a wide range of end-users. The violinist may be a top professional specialising in ‘period’ performance; he or she may be an experienced professional, but one playing a modern instrument and treating it in a basically modern way; he or she may be a minor professional (such as a music teacher) or a good amateur; he or she may even be a learner. Then there are the ‘non-playing’ users: scholars and students of musicology. One caters for all these different levels and needs by adding such things as trills and bowing marks, and perhaps a sample realization of the continuo – which one knows will rightly be ignored and even deprecated by the specialist professional but will be a lifeline for the non-professional and the inexperienced. On the other side, even the highest category of professional performer may be totally uninterested in the apparatus criticus and the choices it raises. One could say that such an edition is deliberately ‘multi-purpose’ in nature in order to maximise its usefulness.

When it comes to baroque opera, however, the user community is today unusually concentrated at the ‘specialist’ and ‘professional’ end of the spectrum. Performances of baroque operas are relatively infrequent – this is unsurprising, given their cost, the number and diversity of people they involve as active participants and their dependence on external funding; and also the relative difficulty (some would also say: inaccessibility) of the music means that live performance and recording are dominated absolutely by ‘period’ groups. This is not Hausmusik – or music for the local operatic society that likes to put on Die lustige Witwe. In commercial terms, this narrowing of the market has consequences. National and academic libraries will, if one is lucky, buy a critical edition of a Scarlatti or Vivaldi opera for its musicological value, but few examples will be in circulation among performers, who will often wish to hire the material rather than to own it. Purchases by individual members of the general public will be few and far between.

The paradoxical outcome of this situation is that although opera is the ‘grandest’ of all baroque musical genres, it is not well served, under modern conditions, by having a comparably ‘grand’ edition (as Monterosso’s La fida ninfa so starkly exemplifies). Rather, it is well served by a severely functional edition, perhaps one existing in different variants tailored to a variety of different purposes. One could in fact ‘modularise’ such an edition, treating as separate components the full score, the derived parts, the edition (or photographic reproduction) of the poetic text, the critical apparatus, the editorial advice on performance and the editorial foreword (or afterword). By assembling and binding these components in accordance with customers’ wishes, one achieves flexibility and lowers costs. It is precisely because the user community is small that such variation becomes economic. One feature that one, mercifully, hardly ever has to agonise over nowadays is whether or not to supply a continuo realisation. A performance of a baroque opera is today inconceivable without a harpsichord continuo player (or two players, perhaps augmented by theorbists) well capable of accompanying from the unadorned score.

The third distinguishing feature – perhaps the most challenging of all – is that a baroque opera is rarely a ‘work’ in the sense that a concerto or a Dixit Dominus usually are. It is a somewhat provisional assembly of dozens of small units, some of which can optionally be omitted, some of which are optional additions, and some of which are alternatives to each other. Because, in baroque times, operas were formed around casts, rather than casts around operas, operas were scarcely ever revived without significant alteration. Even during the run of performances following the first night an opera might be hastily amended in response to the public’s, or the singers’, reactions.

One assumes that in most cases, the editor will wish to provide the complete material for the opera (including alternative versions and items for addition or suppression). The cardinal question is therefore: what kind of performing text does the edition wish to propose as ‘standard’ and what kind as a ‘legitimate deviation’ (or, indeed, as something possibly to be studied but never to be performed). The manner in which the numbers (recitatives, arias etc.) affected are positioned in the score usually conveys his or her attitude on this question, or at least that of the series to which the edition belongs.

One approach – which appears attractive since it conforms to the norm where other musical genres are concerned – is to segregate different ‘layers’ of the score carefully, so that the performer will easily be able to reconstruct the opera in the form it took on its first night, or the form it took at the end of its initial run, or the form it took in an identified subsequent production – or even an ‘ideal’ form it might have taken, had the composer’s initial intentions been realised. The obvious way in which to segregate the layers is to present the ‘main’ layer first as a continuous sequence of numbers, and then to group the numbers belonging to alternative layers in one or more appendices. This is what is currently happening, for instance, in a forthcoming edition of Vivaldi’s Teuzzone, to be published by the Istituto Italiano Antonio Vivaldi.

Philologically, this appears scrupulous and logical. However, it does quite clearly subordinate the desiderata of musical performance to those of musicology. The operatic score, one might almost say, has become a gigantic music example illustrating the evolution and vicissitudes of the score as recounted in the foreword. The segregation between the numbers making up the main score and those in the appendices is such as to discourage musicians, implicitly if not explicitly, to make a free selection among alternatives, which might entail for example combining in the same modern production an aria abandoned after the initial ‘historical’ production with another aria belonging exclusively to a later production.

Some scholars and musicians would argue against such an eclectic approach – that of conflating different historical states – on principle. For them, the ideal towards which a modern performance should strive is what I would term an ‘aural facsimile’ of a chosen historical performance (which is usually the first). A variation on this would be a preference for a late version in which the composer achieved, at last, what he considered the ideal form of the work, the Fassung letzter Hand.

But the problem, and the paradox, of such an approach is that, even as it strives for authenticity, it departs radically from one of the most authentic features of all in Seicento and Settecento opera: pragmatism. For impresarios and composers alike, the optimum version of an opera was the one that pleased the public best. Given the choice of three different available versions of a given aria (whether using the same text or a different one), the one to select was the one most likely to please: strongest musically, dramatically most apt, best suited to the singer, or whatever.

This is also the choice facing a modern production. An Alan Curtis in 2006 is in exactly the same position as a Handel in 1736. To refuse to opt for a better aria because it is an ‘import’ from a different historical production than the surrounding material is in my view too great an act of self-denial. It would have appeared incomprehensible to Handel or Vivaldi, who at the drop of a hat would have accepted new or substitute arias from other operas, sometimes including ones even by other composers.

For this reason, my own preference is to group together in the score the alternative arias (or recitatives, or ensembles, or larger units) available for a given point in the opera. The reason is, precisely, to stimulate pragmatic choice among the possibilities. So if Vivaldi at various times composed three exit arias for Scene 3 of Act II, they will appear together – probably usually in chronological order, but at any rate in such a way as to make their philological relationship to the rest of the score clear. (In a recording, however, this cannot satisfactorily be done, since one cannot expect the listener to be alert enough to advance one or two tracks while listening in order to pass over a second or third version, so recourse to appendix tracks is there unavoidable.)

Obviously, some lines have to be drawn – but more out of respect for our modern, ‘strong’ definition of what constitutes a work and a composer’s œuvre than out of purely musical scruple or respect for authenticity. For instance, we would all balk at substituting for a given aria in Handel’s Ottone one that took our fancy in Giulio Cesare even if the musical result, to the innocent ear, was excellent and in no way harmful to the enjoyment of the opera as a whole. In this reluctance, we are superimposing on an artwork of the past a peculiarly modern preoccupation with origin. But we are indeed modern people, and the productions served by our edition are also modern, so this ‘anti-authentic’ superimposition of Werktreue is both condonable and inevitable.

The next aspect of editing baroque operas is one I feel almost passionate about: the handling of the literary text and its relationship to the words underlaid to the notes in the score (which, of course, are the only ones sung in a performance).

It is astonishing how often one encounters, even among experienced scholars and editors, the belief that, in a tidier world, the text underlaid to the notes would correspond word for word to that printed in a libretto for the same production. Such a simple approach forgets the concrete manner in which librettist and composer worked side by side in preparation for the opening night. Long before the librettist (or sometimes the impresario, if he was using a pre-existing libretto) sent his text to the censors to obtain a printing licence, a form of this text, usually manuscript, was in the composer’s hands. From the point of this handover of the text onwards, everything conspired to make the underlaid text and the text eventually printed drift apart. First, the poet continued to polish his verse: after all, a libretto was also an autonomous literary work to be read silently as ‘pure’ literature by lovers of drama, as well as a collector’s item (rather like a modern so-called ‘souvenir programme’) deserving careful preservation, so it was in his interests to continue to perfect it to the best of his ability.

For his part, the composer, working from a less polished version of the drama, could introduce change either consciously or accidentally. Some changes – for example, repetitions of isolated words and phrases within a recitative, or repetitions of ‘no!’ within arias – were made for valid musical or musico-rhetorical reasons. Others arose from pure accident, when lines were misread or errors in the text supplied remained uncorrected. One remembers with amusement how Vivaldi, using the 1714 libretto of his Orlando furioso as the basis for the text to be set in his 1727 version of the same opera, failed to spot a printing error in the French oath ventrebleu! (spoken by the demented paladin), which the printer expressed as two words – ventreb and leu. Evidently knowing little French, Vivaldi perpetuated the mistake in his new score.

I first became aware of this oblique rather than direct relationship between libretto and underlaid text when editing Albinoni’s comic intermezzi Pimpinone in the late 1970s. In this instance, the problem becomes multiplied beyond normal dimensions, since this was almost a repertory work in the modern sense, and over the thirty years or so during which it was played on countless occasions, the score and the libretto continued to evolve partly in step with one another and partly out of step. The ‘best’ text from a literary point of view is the one presented in the libretto for the first performance in 1708 – but it is also one of the most distant from the text of the surviving musical scores.

The cumulative effect of the literary text’s separate evolution en route to its printing in a libretto and the composer’s deliberate or inadvertent alterations (not to mention those brought about by cuts or by interpolations introduced too late to be included in a foglio volante attached to the libretto) is that the libretto as published is far from being an authoritative basis for textual criticism of the underlaid text. Within the stemma of sources, it represents a parallel branch, not a root source. The root source (the text as supplied to the composer), is, alas, in most cases lost forever and has to be reconstructed (as best one can) by working backwards towards a point of convergence between the two traditions, non-musical and musical. Complex as this situation already is, I have grossly oversimplified it. I have not taken into account the effect of the intervention of literary editors and adapters, of composers who fancied themselves also as poets (as Vivaldi did), of typesetters with fixed habits, or of negligent or creative music copyists.

In my view, there is normally no satisfactory alternative to preparing a scrupulous critical edition of the complete literary text according to the form it takes in the best musical sources. For this purpose, the text surviving in the corresponding libretto is not primary but secondary: it requires collation or at least consultation, but it does not override on principle the readings derived from the score. This means more work for the editor and a bulkier and costlier volume, but anything less than this cuts corners and falls short of the highest standards. Needless to say, it also requires the editor to know something about Italian language, poetry and drama of the period as well as about music.

Decisions have to be made in this connection about whether, or to what extent, to normalise spelling and punctuation and to modernise it. (Modernisation in practice presupposes normalisation, because to normalise according to an obsolete practice – for example, routinely to insert commas after every ‘e’ or ‘o’ or to place accents on almost every accented monosyllabic word (rè, frà, mà etc.) – would be self-evidently absurd.) On this point, I know, opinions differ strongly. I place myself fairly and squarely within the camp of the modernisers. For me, there are two very strong pragmatic reasons for modernising. The first is that it provides a familiar reference point for normalisation (in plain language: ‘consistency’), which is desirable on its own account. The second is that it is a democratic gesture aimed at drawing the literary content of the opera towards the modern audience, at demystifying it.

I should add that, here too, there are limits to observe. Where the phonetic quality of a word is altered by modernisation, one leaves it alone. Thus ‘aita’ will never become ‘aiuto’. Likewise, an original ‘de le’ should not be converted to ‘delle’ for the sole sake of modernisation.

My viewpoint on this subject is not shared by many editions. I am ashamed to say that the Editorial Committee of the Nuova edizione critica delle opere di Antonio Vivaldi, of which I am a member, decided not to include such a modern edition of the underlaid text in its editions of Vivaldi operas in progress. Instead, it makes do with a facsimile reproduction of the original libretto. One consequence of this decision is that (as I have had cause to notice when checking editions prior to their finalisation) editors are tempted not to develop and internalise a consistent procedure for the presentation of the literary text underlaid to the notes but tend to make up their rules as they go along, causing many problems of inconsistency. Had they prepared a complete edition of this text beforehand and then replicated it in the score, this shortcoming would have been avoided.

There is still time to talk about a few smaller points.

The first concerns the nature of the critical apparatus. The Apparato critico of a complete three-act opera, as I know to my cost as a translator of many such, can be a colossal, indeed over-long, section. The most common reason for this excessive length is that variant readings in the original sources are presented individually and repetitively – rather than generically and on one occasion only. Which is better: to read on twenty-five separate occasions that a certain bar of recitative contains only two beats in the source, or, on the first occasion when this phenomenon occurs, to have an explanation of why it is so common and a remark to the effect that other such bars will occur without editorial comment further on in the score? The second solution is more informative about what is of importance, as well as being more economical of space. I am strongly in favour of ‘consolidating’ editorial comment in specimen entries in the critical apparatus that cover entire movements and, where suitable, the entire score. In general, critical apparatuses tend to neglect opportunities to bring together analogous cases in a single entry, at once improving the ‘scientific’ value and saving space.

The second concerns the mastery of standard musical notation. In the days before the advent of the personal computer, when experienced typesetters or copyists processed the scores passed to them by editors, many anomalies of notation were tacitly rectified. From the few available published handbooks on music notation (among which is Gardner Read’s much-reprinted Music Notation), one can see what a complex, though logical, matter good music notation is. Although the programs used by music software such as Finale and Sibelius absorb into their default settings most of the classical principles, several default settings are unsuitable and need to be overridden (unless one wishes, of course, to repudiate the inherited tradition and start a new one). For example, traditional notation treats differently three quavers standing together at the end of a bar containing six quavers according to whether the metre is 6/8 or 3/4, whereas the computer default may group the three notes invariably in the ‘6/8’ manner with a beam linking the whole group of three. My fear is that dependence on computer defaults (through ignorance of the rules or through sheer laziness) will lead to a decline in standards of presentation and even sometimes cause problems for users.

Let me end on an even more provocative note. The ‘high noon’ of the ‘interventionist’ editing of baroque operas – by which I mean editing in which the editor’s hand is very evident – was probably in the 1960s, and is represented by Raymond Leppard’s nowadays not very highly regarded editions of Monteverdi and Cavalli operas. Since then, there has been a steady convergence in editorial methods that has made the individual editor’s contribution – at least, within the score itself – more anonymous. My ‘good’ edition of – let’s say – an Albinoni opera is hardly different from someone else’s ‘good’ edition, since the criteria and conventions applied by each editor are so similar. Both of us, for example, may retain original key signatures and note values, since that choice represents the modern trend, has proved itself perfectly satisfactory in practice, and has good musicological foundations.

However, the price paid for this consensus is a depersonalisation of the role of the editor. Add to this the quantum change in the ‘copiability’ of music (via the photocopier and, latterly, the computer and digital camera) since the 1960s, and one clearly sees how difficult it is for an editor not merely to assert intellectual ownership of an edition but even to receive in perpetuity the simple credit due to an artisan. All of us will agree in theory, I hope, that an editor should benefit financially in rough proportion to the frequency of the performance of a opera based on materials in part created by him, whether this benefit is expressed in the form of royalties or in the size of a one-time payment. The difficulty lies in enforcing this principle. Performers in general do not protect the interests of editors who are not members of their group. Indeed, they have even – forgive my apparent cynicism – a financial and strategic interest in not protecting them, since even if they themselves are not responsible for paying the editor (and/or the publisher), the record company or concert hall will pick up the bill, and money will be subtracted from the overall budget, thus reducing margins.

The sad thing is that while the Leppards of this world can prove that their edition and no other has been used, lending muscle to their demand for royalties, a good present-day editor, no less musical but more scrupulous and systematic in scholarly terms, can often find it impossible to prove this to the satisfaction of a court – since, ironically, good editions converge on a common point and become virtually indistinguishable, while bad editions diverge. I believe that unfair advantage is taken of editors because of this fact. I do not, however, approve of the action successfully taken recently by Lionel Sawkins against Hyperion Records – not because the principle was in itself wrong but because the consequences were too extreme (and mostly to the benefit of the legal profession). What we need – and the case of opera is especially relevant, since to produce an opera is a public act par excellence – is a convention over royalties and rewards that balances different interests more sensitively (as baroque opera did so remarkably) and gives no party any great incentive to flout it. An international register of scholarly editions of pre-1800 operas would be a good start.

I have said enough: I am sure I have raised points in this presentation that others will wish to pick up.