John Adams' newest opera, A Flowering Tree, which has its U.S. premiere this Thursday night at the San Francisco Symphony, represents a welcome return to direct musical storytelling. The libretto, by Adams and his long-time collaborator Peter Sellars, brings an ancient Indian fairytale to life. In order to help support her mother, Kumudha transforms herself, in a secret ceremony, into a tree whose flowers can be made into garlands and sold. A prince sees her performing the ceremony, and his love for her leads to complications that are only resolved through the couple’s steadfast mutual adoration.

The moral fable fits Adams' style like a glove, the music telling the story with straightforward ease and naturalness, and using an extraordinary palette of orchestral color, including, of course, some Eastern percussion. It is, Adams has said, a sort of antithesis to Doctor Atomic (2005). In fact, A Flowering Tree seems like a rededication to simplicity and directness in the relation of words and music. Atomic was abstract and unnecessarily roundabout, the libretto sinking a score that contained numerous glories.

Up to this point, Adams' operatic career has been a little indecipherable, since the composer’s dramatic sense and control of the libretto-writing process has sometimes been questionable. In this, John Adams is hardly alone. When an opera works, the composer often gets all the credit; when it fails, we usually blame the libretto.

Ever since its inception, opera has been a boldly colored, extravagant affair. Its detractors find it overwrought and ridiculous, but those terms go hand in hand with the positive ones, so that, in essence, we’re all agreed. Opera lives at the boundary where charged emotions can easily become bathetic and kitschy. There’s a lot to make fun of in the librettos of the great operas — what aficionado hasn’t giggled at the more bizarre moments of Il trovatore? How many people take the Masonry-tinged philosophy and rituals of The Magic Flute seriously?

And yet, it would be fairly easy to prove that the elements we find laughable in these librettos were integral to the composer’s conception. The gypsy Azucena’s absurdly complicated life story was the catalyst for Verdi’s imagination, and her music tints a good portion of Trovatore. The grandeur of Mozart’s music for The Magic Flute was inspired by the drama’s Masonic theology.

In practice, then, it’s not so easy to separate libretto and score, librettist and composer. Operas are not plays; they are dramas through music. Libretto-making is an aspect of the overall composition, one that may be delegated for technical or efficiency reasons. The drama, and the manner of its presentation, always determine aspects of musical structure and expression, and thus are compositional issues. That’s why reformers like Verdi and Wagner spent so many words on the subject. And that’s why the study of an opera’s verbal text is so important and rewarding. It’s the superstructure for the music, the spark for the composer, the way in for the audience. The history of opera can be read in the changing, complex relationship of music and libretto.

No successful dramatic musician leaves these matters solely to the librettist. The great opera composers tend to have ruthless willpower, theatrical and literary savvy, and business sense in equal measure with the ability to write for voice. Handel, Mozart, Rossini, Verdi, Wagner — each of them micromanaged (or wrote his own), and they rarely got stuck with a bad libretto.

To put those achievements in perspective, consider that composers with plenty of musical skill, like Carl Maria von Weber, Tchaikovsky, even Richard Strauss, turned out a number of operas that delight their fans but hold meager dramatic interest for the rest of us.

One characteristic unites the wide variety of good librettos that have been created through history. They begin with interesting situations and compelling characters who might believably sing extraordinary music. Good opera must take us to the limit of what is sayable (and often beyond), so a few eccentricities, like those already mentioned, are far from the worst flaws a libretto can have. More damaging to a libretto is a lack of interest or originality. On this point, Verdi, the great craftsman, was always adamant. When the Venetian censors had the temerity to offer him a watered-down version of Francesco Piave’s Rigoletto, Verdi shot back:

Someone may say 'A hunchback that sings!' — And why not? … I find it a … beautiful thing to portray this character who is deformed and ridiculous on the outside, and passionate and full of love on the inside. I chose this subject for just these qualities, and these original traits. If they are taken out, I cannot write the music for it. If someone says to me that the notes can work equally well for this [censored] drama … I say frankly that my notes — whether beautiful or ugly — are not written haphazardly, and that I always manage to give them a character all their own.

Reading this letter, you begin to understand how far over the edge Rigoletto was in its time. The enormous power of the central character, the tenor Duke, a highborn character who sings street ditties, and the visceral, violent action would certainly have surprised the average theatergoer.

But Rigoletto also obeyed important conventions that steered the audience through the new work. Most of the arias are in standard forms, and the relationship of closed, lyric numbers and recitative remains largely unchanged from 19th-century Italian tradition. Indeed, Verdi hung onto many of these conventions right up through his penultimate work, Otello. The extraordinary opening, at the height of a furious storm, subsides into the expected (but formally fluid) introduction of the tenor hero, with Otello’s triumphant address, "Esultate."

If conventions bound Verdi and his audience together, his great contemporary, Wagner, made a show of shattering them. Wagner changed the way that words and music related in his operas, so that the distinctions between recitative and lyric pieces were far more fluid, and the way that the music supported and responded to characters’ lines became harmonically wide-ranging and unpredictable. The Wagnerian libretto is a series of dramatic confrontations building to a climax, rather than a distribution of connected lyric spots.

But these are technical aspects. Wagner’s librettos are good because, for example, when Siegmund and Sieglinde fall in love, we are transported along with them (despite the fact that they are brother and sister). Wagner’s operas have flaws, too, but different ones from Verdi, who would never have allowed the extreme verbosity of Wagner’s characters. Wagner’s main subject matter, mythic tragedy purified of any traces of the vulgar crowd, discloses a temperament the opposite of Verdi’s Shakespearean embrace of social variety.

Wagner’s ideas had enormous impact, even on Verdi’s successor, Puccini, who was well-versed in Wagnerian music-drama. You might expect that opera, thus armed, would have continued its dominance among classical genres into the 20th century. But it didn’t. The Italian tradition petered out after the death of Puccini in 1924, and much the same happened elsewhere, so that by the 1950s, only Benjamin Britten among major-name composers (and an Englishman, of all things) devoted a significant part of his career to opera. America’s most famous opera, Porgy and Bess, was written by Gershwin, a crossover composer.

You could cite any number of reasons for the receding profile of opera. Companies became more repertory-dependent and commissioned fewer new operas as movies took over the portrayal of big emotions, even as they transferred Wagner’s musical techniques to the soundtrack. But a bigger deterrent was cultural. After World War I, European intellectuals and the elite classes rejected popular Romanticism and the outsize emotions that opera conspicuously purveyed.

In the 20th century, and our own, driven by suspicion of the sincerity of big emotions and immersed in a culture of irony, many opera producers and composers have tried to cool the flames by creating distance between the audience and the characters. Suddenly, studies in repression — in Berg’s Wozzeck, Britten’s Peter Grimes (and many of his other operas, culminating in Death in Venice), Janácek’s Jenufa and Káta Kabanová — became the favored subjects in opera. Men and women, shackled with insupportable secrets and loves, ostracized and threatened by an intolerant larger community, are the central characters in these works.

But even in Wozzeck, as ironic a work as could be imagined, Berg got around the distancing effect by creating an orchestral peroration for his deceased antihero. Stravinsky’s The Rake’s Progress, taken from a set of 18th-century etchings by Hogarth, with a libretto by W.H. Auden and Chester Kallman in deliberately stylized 18th-century verse forms, still manages to evoke sympathy for its hero in the end. And Stravinsky gave his heroine perhaps the greatest Verdi two-tempo aria of the century. If these operas didn’t affect audiences’ emotions, they wouldn’t be so popular.

Clearly the sea change in the treatment of opera librettos has had some beneficial impacts. The variety of subject matter and the freedom with which it is imagined is a tremendous consequence. Philip Glass’ famously dissociative Einstein on the Beach launched an operatic career in which Glass continues to build on unusual material.

And yet, the winners in the opera world continue to be the composers who can make libretto and music coincide to tell a story that involves the audience in a liminal emotional voyage. Consider Thomas Adés, an English composer who had the audacity to order the complete jettisoning of Shakespeare’s text to The Tempest. The condensation of the story, and the freedom he thus won from a leaden pentameter line, allowed him to create one of the few operatic settings of this play that actually works and speaks to an audience.

And then consider John Adams' operatic career. He likes to take an unexpected route in telling stories. Beginning with a tremendous splash in Nixon in China (1987), an ironic retelling of recent historical events with a number of electric high points, including a chilling aria for Madame Mao, he and Sellars were unable to turn the trick a second time with The Death of Klinghoffer (1991). In this case, the opera needed an emotional focus to the tragedy, but the composer and director chose instead a more detached, oratoriolike approach. The result was an overlong, undercharacterized first act, in an obtuse and dreary staging.

You can’t help feeling that the strangely flat and uncommunicative drama and disconnected verbiage most recently on display in Doctor Atomic is the product of Sellars, not Adams. But it’s Adams' responsibility, because librettos are still part of a larger composition. It’s his opera. A Flowering Tree, however, puts all of the elements together in a convincing whole. Although not conventionally presented, like his other operas, this one at least has a center.

If the 20th century — the great age of irony in the "classical" arts — didn’t kill off opera, we may safely assume that it will continue to exist, 400 years on. It cannot be replaced, not even by the movies, to which it bequeathed so much. Where we go from here is up to the people who are willing to wrestle with the problem of making words fit music, and vice-versa.