Latest articles from "The Sondheim Review":

Remember?(July 1, 2015)

Sondheim's inspiring compositional voice(July 1, 2015)

Sweeney made me do it!(July 1, 2015)

A wish come true(July 1, 2015)

Listen to me(July 1, 2015)

Singing the same way(July 1, 2015)

Sondheim's almost-operas(July 1, 2015)

Other interesting articles:

The Swan Song for Robin Williams
Syracuse New Times (December 24, 2014)

Syracuse New Times (November 12, 2014)

The Stranger (March 25, 2015)

The Horn Book Guide to Children's and Young Adult Books (April 1, 2015)

The Horn Book Guide to Children's and Young Adult Books (April 1, 2015)

The Historian
The Antioch Review (January 1, 2014)

The Stranger (August 3, 2011)

Publication: The Sondheim Review
Date published:
Language: English
PMID: 78651
ISSN: 1076450X
Journal code: SOND

Editor's Note: In 1978 Stephen Sondheim responded to questions at a session of the Dramatists Guild Special Projects. His ad-lib remarks were transcribed and published in The Dramatists Guild Quarterly.


I'm a lazy writer. My idea of heaven is not writing. On the other hand, I'm obviously compulsive about it. And I don't really look for properties. I'm usually dragged in kicking and screaming by somebody. Occasionally, as in the case of Follies, I've gone to a writer and said, "Let's do a musical"; I went to James Goldman because I'd read a play of his called They Might Be Giants that bowled me over. And I said, "Have you any ideas?" In the case of Forum, the same thing; I wanted to work with Burt Shevelove because I admired his work.

Generally, however, people ask me to do shows. I have a reputation for being intimidating and for writing rather arty work, and so I don't get an awful lot of offers - which is all right by me, because it's hard for me to say no. I'm afraid of offending people. I also take a long time to write.

The case of Pacific Overtures is, I suppose, typical. Harold Prince handed me the script. It was a straight play by John Weidman, his first play. Hal had been trying to do it as a straight play for two years and had decided that it needed a kind of epic theater expansion, which meant music. And so he asked me. I couldn't have been less interested in politics or in this particular kind of theater, Japanese theater, which I'd always found just silly and screaming and endless and slow and boring. As usually happens - I'm sure it's true of most of you - the more you get into something, the more in love with it you become, and by the time I was three months into it I thought it was just the best idea in the world. But the first three months I thought, why am I doing this? What usually happens is, I resist everything and then try to work on it because I know that the only way to give a notion a chance is to plow into it. By the time you're inside of it you then know whether it's a feasible piece or not, whether it can work or not as a piece of writing. Obviously, you can't tell whether it's going to work on the stage.

When I found the musical style for Pacific Overtures, I was hooked. When I started it the obvious worry was, how do you find a style of both lyrics and music that doesn't violate the whole spirit of Japanese theater and at the same time is neither pretentious nor coy (because that's what that kind of ritualized theater tends to promulgate)? I decided I would attack the musical aspect of it first, and I thought, now what kind of sound is going to work here? For a month I just kind of fiddled and did some research into Japanese music. And I made the, for me, remarkable discovery that the Japanese pentatonic scale (which is unlike the Chinese pentatonic scale) has a minor modal feeling and kept reminding me of the composer [Manuel] De Falla, whose work I admire a lot. A Spanish guitar principle or modality that underlies his music was precisely the Japanese modality. I know De Falla's music; so I just started to imitate him. I took the pentatonic scale and bunched the chords together until they resembled that terrific Spanish guitar sound. And then I was able to relate to it, because suddenly it had a Western feeling and at the same time an Eastern feeling. I became excited. I had seldom written in minor keys; it seldom occurs to me to write in the minor, but because I had to have the feeling of Japanese tonality, this afforded me the opportunity to do it.

Finding the lyric style was much less difficult than I thought it would be: a kind of translatorese, parable sentences, very simple language with very simple subject-predicate structures, and very little in the way of rhyme. The only heavily rhymed song in the show is the admirals' song, which is all about foreign powers.

The show starts in the middle of the 19th century, when the Americans came in, and then it's about the Americanization of Japan. I decided that as the score went on it would get a little closer to Broadway, until finally it would become an industrial show. We were very heavily criticized for this. In fact, Variety, which didn't like it, said, "The last number looks like nothing so much as an industrial show." I took it as a high compliment.


This gave me a progression, both musically and lyrically, particularly musically. Once you discover a kind of framework for a piece, it helps a lot. You get into terrible trouble if you don't know what your end is going to be. What made Pacific Overtures easy -and in fact it's true of all shows - was knowing the beginning and the end and then finding the style of diction and the style of music. You either fall in love with the material or you don't, and I've seldom abandoned a show in the middle. I find that generally the love affair lasts.

Follies started out as a kind of murder mystery - a who'll-do-it rather than a whodunit. The four come together for the party, and all the old emotions start to recur. Until Hal Prince came in on it near the end of our series of drafts, we never had the four characters as young people. It was always the older people, and when they had their slip into the past, the actors themselves remained the same, they just behaved and spoke like 21-year-olds reliving the traumatic incidents, including a key semi-rape scene played up in the old office that used to look down on the stage. At the end of the first act they each had a reason to hate and be murderous towards each other, so it was who-will-do-it-to-whom in the second act. It turns out Sally tries to kill Ben, although he can't even see that she has a gun because he doesn't have his glasses on. It was all done as a sort of black comedy. She missed, and then they all went home unhappy.

We found as we wrote the first draft that for the first 10 minutes, when people were collecting for the party it seemed terrific. As soon as incidents started to happen, it all seemed a little in the wrong way melodramatic. We thought, "Maybe we won't have anything happen until halfway through the first act," so we wrote a draft - and the first half of the first act was terrific. It took us four drafts to learn the lesson that in this particular piece, nothing must happen. We finally ended up taking all the incidents out.

Finally, Hal suggested that we were being foolishly scrupulous keeping the four characters' young selves off the stage. He said, "I think there's a way to do it - not flashbacks but a cinematic simultaneity of action." So at about the 18th draft, Jim started to write two scripts, on the right-hand page everything that was going on among the middle-aged people, on the left-hand page, passages of dialogue among the young characters. And Hal said, "Don't worry about where everything occurs. I will figure out how to make it work on the stage." We didn't have to worry about the staging of those particular scenes because Hal wanted the challenge. It became a schizophrenic script, even in print. I think this was the right choice, but I'm not sure. I'd like to see it the way we wrote it at about the ninth draft, just for fun, with all the incidents out but all being done by the middle-aged characters.

Musical collaboration: the hardest part

The so-called integrated musical in which the story could be told through its songs, not just with its songs, was the contribution of Rodgers and Hammerstein in 1943. Before that, integrated shows were not musicals, they were operas or operettas. The development of the so-called integrated piece over the last 30 years has come from close collaboration.

The musical, more than any other kind of theatrical piece, is a collaborative effort; and what is required is that everybody sit down together from the first day of the inception and talk about what the show should be. The hardest aspect of writing a musical is to be sure that you and your collaborators are writing the same show. Now, that sounds like sophistry but it is very difficult.

I'll give you an example. We worked for four years on Forum, which was based on some Plautus comedies all mashed together. Burt Shevelove got me the usual academic dreadful translations, but the plays are still funny. I couldn't figure out why I was laughing, because the language was so terrible. But the plays are that good, and so we determined that we would make a musical out of them. He got a collaborator to work with them on the book, Larry Gelbart. Larry and Burt started working on a sort of general outline, and then Burt and I would make a list of songs that were actually used in the Plautus plays.

Over a period of four years, we worked on Forum for three of the years because farce is, I think, the most difficult form of playwriting. I've stated it before and I don't see any reason not to state it now: I think that Forum is the best farce ever written. I think it makes Feydeau look like a piker. Forum is much more elegant than anything Feydeau ever wrote and much, much more tightly plotted. There is not a wasted moment in Forum, and the truth and the test of it is that the play is just as funny when performed by a group of high school students as when it is performed on Broadway. It is never not funny. The reason is, it is based on situations so solid that you cannot not laugh.

We were about to go into production when I got a funny feeling in the pit of my stomach that something was dreadfully wrong with the score. I thought maybe it was just the usual writer's neurosis, you know, everything I write is terrible, everything everybody else writes is wonderful. I had James Goldman come down to the house and first I had him read the script. He said, "It's delightful and it's brilliantly put together and it's very elegant." And then I had him listen to the score, and he said: "It's a terrific score." I said, "Why do I feel peculiar? Am I just getting nervous?" This was about two months before rehearsal. And he said, "No, it's just that the score and the book have nothing to do with each other. The book is written on a kind of low comedy vaudeville level with elegant language, and you have written a witty score, a salon score. Either the score should be lower or the book should be more of a salon piece."

Well, of course it was too late to do anything about it. The test of it was, when we opened out of town the score didn't work at all. There was only one number that the audience enjoyed at all, "Everybody Ought to Have a Maid." It was a throwaway song - I had this whole intellectual notion that all the songs would be. I thought that would be the right thing for a farce. One of the problems with writing a score for a farce (and I kept complaining to Burt about this all the time) is that it interrupts the action instead of carrying it on. It interrupts because it's not about character, those are not songs that develop people and story. Burt said, "You must use songs in a different way than you have used them before." He said, "There are other ways to write songs besides the Rodgers and Hammerstein school," which is what I was trained in by Hammerstein himself. He said, "There are songs that just act as respites. That is the way they were used when the Romans used them, and that's the way you should use them."

It was very hard for me to do. It was like a series of nightclub numbers. It was precisely in trying to do it that I made the error, and the result was that the numbers didn't hold up on their own because they were written in a different style. There was a big difference between what was going on on the stage and what was going on in the songs. And that's really why the audience only enjoyed that one number, because it was the only number written in the right style. I then resuscitated a number that had been thrown out in the four years that we were writing Forum, called "Impossible," and that's also in the right style. But everything else in that show - and I'm proud of some of those songs - is really wrong for that show. "Pretty Little Picture" is a perfect example of what's wrong. In fact, we dropped it in the revival. It's a very elegant, rather witty song, very verbal, and has nothing to do with the way the play is written.

Burt Shevelove used to say about Forum, "It's a scenario for vaudevillians. I want to give five terrific vaudevillians a chance to cavort on that stage. I want to give them a structure that is so rigid that the plot will keep them going, but within that, if somebody wants to take a page and just smell mare's sweat, he can do it." Zero Mostel could do that for 15 seconds or 30 seconds or a minute and a half, smelling the mare's sweat on Davey Burns. And Davey Burns could do it. So they had a routine worked out. Now, you know, in the East Keokuk High School, they don't do a routine. They just do the lines, but ... maybe if they have an inventive young man there, he will do a routine on smelling mare's sweat. That's the idea of it. Burt wanted every performance and every production to be different.

You have to be careful about what demands you make on an audience. How much do you want them to listen? How much do you want them to just look and receive? In a show like Forum, it's very important that they don't have to listen a lot. In a show like Company it's very important they do have to listen a lot, because you're making points. Forum is not about making points. It's a situation comedy, you know, it's the basis of everything from Sergeant Bilko to I Love Lucy. Plautus invented it all.

Integrating a score - the secret metaphor?

To make a score sound like a score when it's interrupted by great chunks of dialogue is difficult. In a sense, opera is easier to write because it's through-composed. The libretto and the spoken text take much less precedence. I have used arbitrary methods, although not as arbitrary as they seem, to hold scores together. Certainly leitmotifs are useful (but a very bald device, when every time a character comes on you get the Jet whistle and you know he's a Jet). In Night Music I put everything in some form of triple time so that the whole score would feel vaguely like a long waltz with scherzi in between so that no song would seem to have come from another texture.

In the case of Pacific Overtures I kept a very limited harmonic language, with very little harmonic motion in the songs. Pacific Overtures had static songs, harmonically. They don't go anywhere. "Someone in a Tree" is an example of a song that's built almost entirely on two chords and an endless rhythmic vamp that bored the audience to death in some cases. But I found that since Japan's music is relentless, you've got to have some relentless songs. The score of Pacific Overtures holds together because it has the same harmonic texture and the same lack of variety within the songs, as opposed to a score like Follies, which is built on variety.


Nothing whatsoever holds the Follies score together. All I tried to do was keep the so-called book songs and the so-called pastiche numbers entirely separate in style so that you could have the sense of seeing two shows at once. I'm a firm believer in content dictating form, and when you're faced with a play in which you have two casts, one of older people, one of younger people, then you've got to have two kinds of music going. And there were two kinds of music going in Follies. I was splitting the attention, but that was the idea. It's a schizophrenic piece, and it's supposed to be.

I tried to hold the score of Company together through subject matter. I thought, OK, I'm going to write 13 songs, all about marriage, and the only thing to avoid is monotony. What will hold it together will be that everything is dealing with one-to-one relationships. (With one exception: We cast a girl named Pam Myers, and we wanted to write something in for her.)

Incidentally, in every show there should be a secret metaphor that nobody knows except the authors. As in Elgar's Enigma Variations, you only hear the over-structure, but something is holding together underneath that's inaudible or invisible. In Company, we were making a comparison between a contemporary marriage and the island of Manhattan - in fact, it was even spoken about at one point. We made a vaudeville joke about it in the middle of "Side by Side by Side," and then we took it out because we decided never to let anybody know that that was what we were about. But it justified my writing a song about Manhattan, "Another Hundred People," which is the only song that doesn't deal with one-to-one relationships.

Anyone Can Whistle is sort of a music student's score. That whole score is based on the opening four notes of the overture, which is a second going to a fourth. All the songs are based on seconds and fourths and the relationship between a D and an E and a C and an F. The fallacy is that the music isn't continuous, so that it doesn't mean anything to the audience's ear really - except to some musicians who could possibly hear it. But the seed doesn't germinate the way it does in a symphony with a continuous spread. It germinates, and then there is dialogue, and then I remind everybody that it's based on the second or the fourth ... it's all too late. But it helped me make the score.

Sweeney Todd will, I think, be almost continuously underscored. You can't really underscore - unless you are writing an opera - until you have staged it. You see, for underscoring in an opera, the director has to fill in those moments where the composer is making tiny little intermezzi. But in a musical you do a lot of underscoring during rehearsal because it would be foolish, when there are so many cuts and changes to be made, to try to write everything in advance. You can't. That's also why composers can't orchestrate their own work unless it's a very small band - because you don't know what the keys are going to be until you are in rehearsal. You can say, "We will have underscoring and bridge from there to there," but until you put it on the stage, you can't outline it or write it. You make so many changes out of town, it would become prohibitively expensive to keep changing all the underscoring, because you can't just keep underscoring with single instruments. If you design it carefully enough, however, I think you can get a lot in, and that's what I'm going to try on Sweeney Todd. The basic structure of the play is so tight, I don't think we'll be making a lot of changes. I may be eating my words in a year, but ...

There is some danger that total underscoring would be distracting to an audience. The difference between sung and spoken is a huge difference. It's one of the reasons why there is such trouble making rock work on the stage - the contrast between that kind of singing, which is, you know, highly sung, so to speak, and the dialogue - the drop is so much. The same thing is true of underscoring. You might think, "Oh, oh, here comes a song," and then you'd be unconsciously waiting for the song, whereas it's just going to be underscoring.

In collaboration, we talk about the material until we eventually get down to that terrific moment, after a number of weeks, when the secret metaphor becomes clear. I work with Hal Prince because he's so creative and so stimulating. I sit with him and the librettist and sometimes the set designer because Hal always goes to the set designer first. Like most directors, he gets a visual concept before he even gets an oral one. But Hal also has an impeccable sense of what a whole evening should be, what the totality is supposed to be. And there comes that moment, usually somewhere between the second and fourth week - meeting twice a week or three times a week - where you all look at each other and somebody says, "You know what this show is really about? It's really about X." And everybody says, "Yes!"

Once you know that, you can start writing. Now, that X is never stated baldly. I want a certain mystery for the audience, something that they can sense and even discuss and have to discover for themselves. Or they don't. If they don't, that's OK, too. It's something ... something that holds the evening together, gives it a reason but isn't stated because as soon as you state it, it's like bursting a balloon.

There are certain things that, the minute you talk about them, lose their charm. Certainly in a sustained piece of any weight or seriousness, there should be a mystery. Sometimes you have to deliberately create that mystery, but usually it's there anyway. The thing that makes you write the piece is quite often something unspoken that you only discover as you're writing it. Why does this piece appeal to me? Why do I want to tell this story? And then one day you say to yourself, "Oh, I see, it's about blank, blank, blank, and that's something I relate to" (but I think it's important not to state it). It grows on you, and if it doesn't, the work is probably too shallow. I want to know what's going on underneath, but I don't have to know in a sentence. It doesn't have to be like English II where you reduce Macbeth to one sentence. It's not exactly subtext, I understand that to be something else, although I suppose you could call it a subtextual matter.


If you find the underlying metaphor and then don't state it, it helps shape the whole work. But it's your secret, yours and your collaborators'. You may find that you all think the show, on certain levels, means something different. If you asked Hal Prince to state what Company is about in less than one sentence and asked George Furth and asked me, we would probably come up with different sentences. But we all know about this underlying metaphor of Manhattan, which is after all the handiest locale for the inhumanity of contemporary living and the difficulties in making relationships.

That comparison, incidentally, wasn't in George's mind when he wrote the plays, which were about California couples. They were the first plays he'd ever written, a series of seven one-act plays, each of which had two or three people in it, all variations on marriage. Hal read them and said, "I think they could be a musical." And I said, "They're so unmusical. They are so unsinging." George writes non-singing people. And Hal said, "That's what's interesting about it." And I said, "You've got me hooked." I was so stimulated, it sparked something in my head. Hal wanted to change the locale to New York, and George kept asking why, or saying, "Maybe the pot scene could be in Westchester," and Hal said, "No, I think everything ought to be about Manhattan." In discussing why, we discovered, the three of us, that there was a metaphor that was pertinent; not just laid on, not arbitrary, but something inherent in the material.

This text is excerpted from a longer essay, "The Musical Theater: A Talk by Stephen Sondheim," originally published in 1978 (Volume 15, Issue No. 3) in The Dramatists Guild Quarterly. The Sondheim Review is grateful to the Dramatists Guild and Stephen Sondheim for permission to reprint this selection. (Reprinted by permission of The Dramatist: The Journal of the Dramatists Guild of America.

The use of this website is subject to the following Terms of Use