“The writing of words is already the writing of sounds, the writing of a text is already the outlining of structures: the librettist participates in the work that the composer is used to doing alone, it is therefore a delicate position that necessarily requires trust.”
Kaija Saariaho about Reconnaissance, 2021
I see a certain type of handiwork and craftsmanship being neglected in the way a lot of opera librettos / sung texts are conceived, and believe it would be useful to get a little nerdy about this. Sung text doesn’t work exactly like printed or spoken text, and it’s also not an inferior use of words simply because it is destined to music – on the opposite, it has to satisfy the normal needs one would assign to text in a literary and theatrical context and also serve a musical purpose. Since the ways in which a libretto does the latter are poorly documented and discussed, I want to talk a little about this aspect specifically. I have shared in other texts thoughts on the more dramaturgical dimensions of opera and libretto-making, and will focus here on something more technical and apparently much smaller, but equally important.
The libretto forms a musical work’s first musical material, and it is actually a tremendous delegation of responsibility from the composer to let someone else make decisions about forms and sounds that they normally develop themselves. This is true even in those extreme cases where the text is entirely dismantled in the process of setting. However in operas this is usually not the case: the prosodic and phonetic qualities of the text are the backbone of the rest, and essentially all the phonetic material is provided to the composer by the librettist. There would be a lot to say about the librettist as a storyteller, dramaturge, and inventor of forms, and again these are topics I address in other contexts, but today I will speak specifically about sung text craftsmanship, and take the example of the opera INNOCENCE, which is great to illustrate this point, as after collective work on dramaturgy and the macro-form, my personal focus on that project was on creating singable text.
In the making of INNOCENCE, it was formative for me to be able to focus intensely on this part of libretto-making. When they are set to music, written words are inflated with literal air – the smallest details become oversized and the effects they produce (good or bad) amplified. As with the other levels of a libretto that bear responsibility in the end result on a musical level, if no attention is being given to the musical nature of text itself, the composer will be trapped with elements that are difficult to use and that trigger few musical inventions.
THE INNOCENCE PROCESS. INNOCENCE (premiered in 2021) was conceived within a three-way collaboration with composer Kaija Saariaho and writer Sofi Oksanen that started in 2013. After we had made joint decisions about form, voice types, and storyline, Sofi wrote the entire script for the opera, complete with dialogues and stage directions, in Finnish. I was then tasked with creating the final libretto that Kaija would compose and that the singers would sing, in nine languages (which we selected among others on the basis that I had at least minimal notions of them). About 4% of the Finnish text was used in its original form.
The main languages – English, French, German, Spanish – I translated myself and had native speakers proofread. For the languages I didn’t know as well – Czech, Swedish, Greek, Romanian – I proceeded the other way around, asking native speakers to produce a first draft that I then reworked. All of this entails fascinating topics from the perspective of translation alone, including the unusual situation that the Finnish original would disappear and never be accessible to the audience, the new ‘original’ being whatever was being sung (in the following I will quote Sofi’s unpublished original Finnish text only to make my process clearer). Both dramaturgically and musically, the creation of the multilingual text was then not bound to the usual translational standard, where one is aware that one reads an imperfect rendering of an original, but had the responsibility of being the definitive text. Once I started creating the libretto and we agreed with Sofi on the meaning of each sentence, and who the characters had become in their knew languages, I took over the last stage of the libretto-making process, which as a whole is normally concentrated in one individual: the creation of the sung material, and later its discussion and tweaking with the composer throughout the composition process (which in this case spanned the entire years 2016-2018, not including proof-readings), and its correction in rehearsals with the help of the performers. Most of the things we worked on with the performers were not translational matters, but rather ways to make the sentences both sound better to their ears (as they were native speakers of the languages they performed) and sit better prosodically. This is the crux once the form and script are established: how can a text work best, as sounding material and dramaturgy of sounds, for the composer, performers, and ultimately audience?
One first example of this: surprisingly, given the apparent simplicity of the sentence, it took us many tries, first with composer Carlos Pérez Tabares with whom I reworked the first version of the Spanish text, then with tenor Camilo Delgado Diaz who premiered the role, to make a functional version of Jerónimo’s line:
(…) naapuri tuijotti minua
kuin en olisi saanut tehdä mitään,
mitä olisin tehnyt normaalisti.
> (…) el vecino me miró
como si no debiera estar haciendo
lo que normalmente habría hecho.
In addition to the phonetically favorable features of the Spanish language, the prosodic solutions illustrate some of the issues I would like to single out in the following.
CUSTOMARY WISDOM AND BEYOND. Although operatic text and music are no longer as constricted by set forms as in earlier stages of their shared history, there are some customary rules about creating any sung text: constructing rhythms that are interesting to build on, that help build a sense of acceleration, deceleration or stability; avoiding long words and sentences that become clunky when dilated into singing lines; dispensing with consonant clusters and other tongue-twisters, unwanted alliterations; favoring open syllables, especially for text meant to be sung in the high register, etc. Of course, like all rules, these are meant to be bent. But being aware of them makes their breaking more potent. Besides, the point of a libretto is not to be transparently coated in music – it can also fight for its own autonomy, remind of its own presence as text. Such is the power of unexpected word choices and syntactic solutions, alliterations and wordplays. Without going that far, by standard rules a sentence like the Bridegroom’s
They would give me an incriminating look.
is less than ideal to sing, with all its nasal consonants and closed vocals (to make matters worse it also turned out to be composed loud, fast, and high), but the sheer lingering and insistent expressivity of the word ‘incriminating’ makes it worth it.
The fact is that every language provides its own expressive solutions, that are not naturally present in others, and in INNOCENCE it was a delight to work with this. For instance the various German voiceless fricatives, of which Kaija made abundant expressive use in her setting of the character of Anton:
Sinä aamuna heräsin pahoinvointisena (…)
> An jenem Morgen fühlte ich mich schlecht (…)
Or
Ampuja ampui häntä vatsaan.
> Der Schütze, der Schütze schoss ihm in den Bauch.
[Repetition added during the composition process.]
This was something that excited us in particular with the Czech language, as Leoš Janáček’s work on natural prosody was one of our foundational sources of inspiration. The expressive qualities of Czech never disappoint, and the character of Markéta makes full use of the crammed violence of the word mrtvá (dead). But when with translator Linda Dušková we proposed the following tongue-twister for the climactic Scene 24, Kaija had to politely decline after her first attempt at setting it:
Ten střelec mě střelil do srdce.
We came to think that the hissing alliteration of the English would be sufficient:
The shooter shot me in the heart.
SYNTAX AND PROSODY. I have generally built the multilingual text upon the contrasted syntactic structures crafted by Sofi in Finnish, most of which are easily transposable from one language to another: long/poised vs. short/percussive sentences, anaphoric repetitions that underline a rhythm, or the messiness of a confused character’s run-on sentence. Sofi took it to heart to constantly question whether there was too much text and made it a personal challenge to chisel the text into its most concentrated and efficient form. This was the necessary groundwork for the next step: to create the multilingual text that would actually be suitable and inspiring for the composer and performers.
Prosodically the performable text was not meant to mimic the Finnish, and in a way Kaija’s wish was the very opposite: she wanted to be able to capture the prosodic idiosyncrasies of each of the many languages to use them as musical material. Trying to transpose the Finnish strictly would therefore have been counterproductive; each language had to appear in a natural, if stylized, guise. However this doesn’t mean that one simply has to go for the ‘default’ form of a sentence in any given language. Specific length and stress patterns are always a way to invite a certain type of music. In the fragile final utterance of the Father-In-Law in Scene 14, the proposed sentence was very natural in a colloquial sense, and would have worked perfectly as spoken text, with its hushing and slightly macho quality:
Anna kun vien sinut kotiin.
[syllable-length pattern: — u / u — / u — / u —]
In tone and rhythm, in English it could be something like: Come now, I’ll take you home. But I liked the possible nuance of begging (from someone who really wants the other person to not ruin the party she crashed), and in line with this also found that a gasping rhythm would be more effective for the sung text, while additionally cleaning up the colloquial turn of phrase:
Please let me drive you home.
[stress pattern: — / — u / — u / —]
Kaija seized the opportunity and used the heavy prosody, weighed further down by its trochaic pattern, to compose the sentence as a painstaking descending line spanning almost two octaves.
In general, any syntactic feature is amplified by its musical setting, and therefore needs at the very least to provide the composer with interesting options.
An additional observable artifact is that, even without a traditional prosodic structure, the endings of sentences tend to be automatically endowed with more weight than in another context. This effect can be utilized in various ways, for instance to balance the perceived stress pattern of a sentence. By default, a sung sentence often seems to work better when ending on the word most deserving to be stressed:
… paikassa, jonne kuka tahansa olisi voinut tulla.
> … un sitio donde pueda entrar cualquiera.
Esikoiseni oli myös lapsena kuin enkeli.
> My firstborn was also like an angel as a child.
Ne harvat, jotka meillä kävivät,
halusivat vain kuulla,
miten tutkinta etenee (…)
> The few who would visit us
only wanted to hear
about the investigation (…)
… että joutuisin häihin hänen tilalleen.
> … that I would replace her at a wedding.
In this last case, the reworking is a little deeper, removing the interesting double meaning of joutua, which has the connotation of both an obligation and of a certain sense of fate; the verb used in English is more factual, and more emphasis is put on the ominous ending. Kaija set this by detaching ‘at a wedding’ (sung softer) from the rest of the sentence, as if it had been preceded by an ellipsis, and by having these words repeated by the off-stage chorus like an echo chamber. (In this example, note also the removal of the Finnish assonance for a more intelligible result.)
VOCAL TECHNIQUES. One exciting feature of INNOCENCE is that, building on my experiences with hybrid forms in the field of contemporary music theatre, we decided early on that we would use different vocal techniques on the spectrum between speech and operatic singing, including folk singing. This was liberating for many reasons: in terms of script, it allowed for a larger quantity of text material, as not everything had to take into account the dilation of classical singing, in particular in its melismatic Saariaho version; in terms of music, it widened Kaija’s palette considerably, and favored stronger musical characterization of each person on stage; it also has widely enriched the fabric of each production, by calling for performers of different backgrounds and trainings. And in terms of the present discussion, although this is not the first thing that is usually underlined, the use of different techniques also expands vocal and prosodic possibilities on the libretto-level, which makes it more interesting for everyone.
Technically, even in a classical opera setting, a libretto should always take into account the voice type for which a character is intended, as the physiological and acoustic realities at play are implying different areas of strength in projection and formant alignment. However, like most things related to the musical dimensions of a libretto, this is usually left for the composer to deal with – most often with problematic results, since the foundations are unreliable. The vocal writing might then lean more into trying to salvage the result, rather than taking full advantage of the possibilities of any given voice. It turns out uncomfortable for the performer, and also generally doesn’t live up to the creative potential of the text/music collaboration.
In many projects I have tended to seek variety in vocal styles, not in order to avoid the specific limitations of operatic singing, but to have more dimensions to explore also in the musical component of the libretto itself, even if each voice comes with its own limitations matching its idiosyncratic strengths.
One inspiring example was the character of the Teacher, which was intended to sing in a form of Sprechgesang stretched through various extended vocal techniques. Kaija herself took a very open approach to the character, and for the first round of rehearsals composed the part with just basic rhythms and pitches, with the intent of crafting the vocal effects together with the singer Lucy Shelton. The character, an ex-teacher devoured by the guilt she feels over the rampage carried out by her former student, was thought of as a broken person – which vocally would not point towards the open projection characteristic of operatic singing. This subtraction (from an operatic perspective) is much more interesting when superimposed with an expansion of the vocal language in other directions. To create the material for this, I was first unsure about treat an utterance such as:
Outoudet hänen aineissaan
olivat olleet merkkejä,
jotka minun olisi pitänyt huomata.
The roundness of the diphthong-heavy first line and well-balanced rhythm of the whole, if transposed in English, could have fit a different character, especially one singing classically. So as opposed to some examples quoted earlier, there was an attraction to favor unbalanced rhythms that would suit the intended musical treatment, fast alternation of open and close vocals, and the normally odd accents of the z sounds piling up, breaking in the end into a transparent, sharply lucid regret:
Some peculiarities in his essays
[stress pattern: u — u u — u — / u u u —]
had been signs
[u — —]
I should have noticed.
[u — u — u]
An extreme last example of not writing for a normal singing voice would be the character of Iris, the perpetrator’s accomplice. We selected notated speech as a technique for the character, because we knew she needed to deliver a lot of information in a limited time in order to create the reveals in the last third of the piece; and to connect her with the Bridegroom, we chose French as her language of expression. This choice of French was interesting for a character who lives with repressed rage, as the absence of lexical stress allows for an extremely flat sound result with occasional eruptions. Kaija expanded this by composing the part as an alternation of a dilated and ‘airy’ form of Sprechgesang, notated with relative pitches, and of normal speech assumed to be a more fluid version of the same. For me this was an opportunity to work very differently than in the sung parts, leaning on the floating quality of French in the ways I would for spoken text in general, i.e. not putting stress on the end of sentences, favoring suspension with more fragmented syntax than in Finnish and by using feminine endings and closed and nasal sounds:
Sinä iltana äitini piti television kiinni
ja kuvitteli varjelevansa minua uutisilta,
mutta minä luin niitä peiton alla puhelimesta.
> Ce soir-là maman n’a pas allumé la télévision,
elle croyait me protéger des reportages,
mais moi je lisais tout sur mon téléphone
en secret dans mon lit.
Everything that in a normal operatic setting is limiting about French – the lack of lexical stress, the nasals, the guttural r sound… – could be used to its full potential and become an expressive tool. This goes on to illustrate how, in terms of working with the musicality of text, INNOCENCE was something of a dream project, due to the variety of vocal techniques and languages, and how broad the resulting palette (for us) and experience (for the audience) could be.
CLOSING REMARKS. The reason some plays written to be spoken have become great opera librettos is that they were chosen by the composer also for their musical qualities: Maeterlinck’s Pelléas et Mélisande and Büchner’s Woyzeck, in addition to being great texts on their own right, contained the specific potential for the music (both in terms of vocal expression and the functional status of the orchestra) that Debussy and Berg wanted to create, and this made up for the fact that they were not conceived to be sung. Additionally, while the latter is true, these are texts that did beg difficult aesthetic questions as to how they should be spoken and staged, to which the music – a music that extended beyond what the playwrights might have imagined – could offer visionary solutions. These precedents should be humbling to any librettist: the text they provide should not compromise with their own standards, but should also be as helpful to the composer as an existing text that the composer would have themselves handpicked (a situation many composers find preferable), and yet challenge the composer in ways that cannot be entirely anticipated.
It takes more than a good story, good dialogue, and good music to make a compelling piece of music theatre. This commonplace axiom is usually followed by the acknowledgement that opera is storytelling and theatre by means of music. This cannot be emphasized enough, but should not be divorced from the fact that it is also music by means of words. It is time we gave more visibility to everything this can mean beyond old-fashioned platitudes, and explored in depth what the art of the librettist in all its dimensions: as a writer, as a key co-creator and artistic partner to the composer, but also as the artist whose medium is the invention of new intermedia forms suited for new narratives and experiences. This endeavor takes place on all levels of creation of a new piece, from the macro-form to syntactic and phonetic level decisions – decisions that are already music-making.
>marginalia* 