So You Want To Write A Song

Composition is a mysterious thing. I mean this in the typical artsy sense of “I don’t actually know the precise mechanism by which things ‘come to me’ when I’m composing”, but also in the sense that I don’t think most non-composers know what we do when we disappear into a practice room for hours at a time. To be sure, some of it really is just staring into space and trying to imagine something musically compelling, but some of it’s considerably more plodding, methodical, and mundane. It’s art, sure, but that doesn’t mean there’s no reason to the rhymes. So today I want to open up the hood and give you a peek into my compositional process.

(As always with these sorts of posts, when I say “my compositional process”, I really do mean my compositional process. Other composers definitely go about these things in wildly different ways, and I think that’s ultimately a very healthy thing for the diversity of the art that results.)

Our vehicle for this will be the first song that I wrote for my Tisch application, a setting of a sonnet by Edna St Vincent Millay. The text runs as follows:

Time does not bring relief; you have all lied
Who told me time would ease me of my pain!
I miss him in the weeping of the rain;
I want him at the shrinking of the tide;
The old snows melt from every mountainside,
And last year’s leaves are smoke in every lane;
But last year’s bitter loving must remain
Heaped on my heart, and my old thoughts abide.
There are a hundred places where I fear
To go — so with his memory they brim.
And entering with relief some quiet place
Where never fell his foot or shone his face
I say, “There is no memory of him here!”
And so stand stricken, so remembering him.

So that’s the poem. How did I turn it into a song?

Because I’m a formalist at heart, I like to start by hashing out the piece’s form. This is a lot like outlining a prose essay before sitting down to write: I’ll start with this idea, then I’ll transition into that one, and then I’ll bring the first one back, but in a new context that makes it feel a little different than the first time . . . It’s creating a roadmap of where I want to go, giving me a sense of what I’m going to need to fill in when.

Sonnets, it turns out, don’t play terribly nicely with standard songwriting forms.

Arguably, the Most Iconic songwriting form (to the point that it’s often just called “song form”) is the AABA form. As the letters in its name suggest, it’s built from a first phrase (called A) that’s repeated twice, a middle (usually contrasting) phrase B, and then a return of the A phrase to bring things to a close. Examples of this are everywhere, but you can’t go wrong with a familiar icon:

Somewhere over the rainbow
Way up high,
There’s a land that I heard of
Once in a lullaby.

Somewhere over the rainbow
Skies are blue,
And the dreams that you dare to dream
Really do come true.

Someday I’ll wish upon a star
And wake up where the clouds are far
Behind me,
Where troubles melt like lemon drops
Away above the chimney tops
That’s where you’ll find me.

Somewhere over the rainbow
Bluebirds fly.
Birds fly over the rainbow,
Why, then, oh why can’t I?

Unfortunately, since formal sonnets developed largely independently of popular song lyrics, the sonnet doesn’t fall cleanly into divisions like that. It breaks down quite nicely into a chunk of eight lines and a chunk of six, as is standard for many sonnets, but even if you break the first eight lines into two sets of four, it’s still an awkward, kind of shoehorned fit. As we’ll see later, that doesn’t mean it can’t be done, just that it’ll take some doing.

Another very common form is verse-chorus form, where there’s a fixed chorus that sits between verses that use the same melody to set different words. (Examples are thick on the ground here, too, but to pick an example that keeps with lovesickness, “Greensleeves” illustrates the form quite tidily.) Now, obviously there aren’t any repeated lines in the Millay, but if you’re willing to be loose with the text, you can make some. The obvious candidates would be the first two, since they’re not actually a bad summation of the sonnet as a whole, but you quickly run into problems.

For starters, it’s not entirely clear where you’d break the rest of the lines into verses. The next two (“I miss him.../I want him...”) make a decent unit, but the next four really need to be taken as a chunk, and the last six probably should be too, tho I suppose you could split them into 2+4. Either way, it’s not a particularly elegant arrangement.

The other problem is rhyme. In a verse-chorus form, it’s usually a good idea to keep the chorus from rhyming with anything in the verses. This helps keep the two sections distinct and marks each out as its own thing. But since the sonnet follows an ABBAABBA/CDEECD rhyme scheme, using the first two lines as a chorus means that you’re inevitably going to wind up with a chorus that rhymes with quite a few of the verse lines, and, compounding matters, a chorus that doesn’t rhyme with itself. (Splitting off the first four lines as a chorus at least gets you a chorus with its own rhymes, but it doesn’t solve the problem of the rhymes with the lines in the verse.) Again, it’s definitely not impossible to make this work, but it’s not something you can do on autopilot.

This leads to the last formal possibility I’m going to discuss here, which isn’t strictly speaking a form at all. That option is to just take the poem as it is and set it line by line, writing the music that seems to fit each given moment in turn. This is commonly called “thru-composition”, and it tends to show up more in concert settings than in pop or theatrical ones, but it only takes one listen to Aaron Copland’s setting of Emily Dickinson’s “Heart, We Will Forget Him” [YouTube] to see that it can still be emotionally devastating. The danger with a thru-composed setting, tho, is a lack of cohesion. With verse-chorus or AABA form, you’re guaranteed a certain consistency due to the sections that repeat; with thru-composition, you run the risk of introducing new idea after new idea after new idea and producing something that resembles a meandering potpourri rather than a focused, unified whole. Unsurprisingly, composers and theorists have expended a considerable amount of energy over the years trying to figure out how to produce the latter rather than the former when taking this approach.

I ultimately took kind of a hybrid approach, largely thru-composed, but with hints of an AABA form lurking in the background. I took the first four lines as an introduction, an announcement heralding the core idea of the song. (This itself is actually a pretty common trope, tho the section in question is often ditched. Did you know that the full version of “Fly Me To The Moon”, for example, actually begins like this? [YouTube]) With a little finagling*, the next four lines can be the two A sections (with two lines apiece), the two after that a B section, the next two a (highly modified) A, and then the final two a climactic coda at the end. It’s by no means perfect**, but it’s a place to start, and at a certain point, the job of composition is to write music compelling enough that the form seems inevitable, not constructed.

With the form hashed out, the next big step is settling on the appropriate musical language. Now, obviously each composer has their own individual voice, a voice that has limits in how far it can stretch, but those limits can cover quite a range. Some of my music is dense and thorny, bristling with unresolved dissonances and miasmatic melodies; some of it is light and boppy, easily singable and sometimes even danceable. Settling on a specific point on that continuum is a largely intuitive and holistic process, but here are some of the factors I explicitly consider:

  • How florid is the text? A florid text doesn’t always require florid music — Lin-Manuel Miranda’s treatment of George Washington’s farewell address [Spotify] is a beautiful example of how a (relatively) florid text can sometimes best be served by plainspoken music — but there are times when a simple setting of a florid text can be downright jarring, and that’s an important thing to be aware of. The Millay is pretty florid, especially towards the beginning, and my setting reflects that.
  • What’s the emotional tenor of the text? Long, flowing melodies can be great for expressing joy or longing, but they’re less well suited to expressing rage or despair. Dance rhythms work well for anticipation, but less well for anxiety or dread. The really fine-grained, line-by-line distinctions will obviously be done at the local level, but I find it helpful to go into things with some idea of what I’m aiming for. It would be strange to say the least if you went off to set the Millay and came back with something that sounded like “Call Me Maybe” [YouTube]. (That said, I would dearly love to see someone produce a screaming, full-throttle heavy metal setting of this text.)
    • Music can get very specific with the emotion it expresses, which is both a blessing and a curse. One of the joys of good poetry is that it can have a different emotional valence each time you read it — is Millay deeply bitter in that sonnet, or is she more wryly accepting? Is she forlorn, or is there a seething mass of resentment under the surface? There’s obviously some wiggle room for interpretation when it comes to music, but usually vastly less, and text setting often winds up flattening the possible emotional worlds into just one of the many options.
  • How directional is the text? This is related to the above. Some texts live inside a single emotional moment, turning it this way and that and examining it from every angle. Others have more of an arc, more of a linear story that builds to a climactic point. Likewise, some musical languages are better at holding a mood and staying with it, while others are more restless and want to drive to a conclusion. I very much read the Millay as building to the crushing defeat of the final two lines, which pushes me towards a more restless music. (And also, incidentally, towards a more ad-hoc form. I’m presenting these as separate steps, but they both feed back into the other. What can I say? Creative processes are seldom super cut and dried.)
  • Who is singing? This is frequently less pertinent to art songs than theatrical songs, but it can still be there in the background. A teenager going thru their first breakup calls for different music than an octogenarian with a full life of experience under their belt. In a standalone song like this, there’s no dialogue or other text to tell us who the singer is; the music has to carry that weight alone. You could write entire books about how music represents character (and people have!), but for now suffice it to say that this is an important factor in the planning stages of a song.

If there were a lot of possibilities in the section on form, there are vastly more here, and different composers are going to have substantially different lists of the factors most critical to their approach to this process. But that’s the heart of my list, and feeling my way around those concerns led me to a point in the more convoluted, harmonically ambiguous regions of my vocabulary.

And now, at last, we come to the actual writing of the music. And now, regrettably, as I said at the outset, we come to the part that I can’t explain. I can tell you that I sit down at a piano and play around with things until I find something that sounds “right” for the moment and the piece in question, but I can’t really tell you how I do that, any more than I can explain how the thought “I want to get a glass of water” translates to the specific muscle contractions that heave me out of my seat and over to the kitchen sink. Having decided what I want to write, I write it, in fits and starts, easily or effortfully, and sometimes going back and throwing the whole plan out the window and starting over from scratch. And then, finally, there’s a song:

So yeah, that’s the story of how I got from a bunch of words to a song. I’ve heard a few of the other composers’ takes on this, and they are all delightfully, deliciously different. I love that. I love how many ways there are of responding to a text, and the variety of the art that results. (I also love hearing how other people cross this gulf; process is fascinating, and if you want to share yours, I’m all ears!) For those of you who aren’t composers, I hope that what we do is maybe a little less mysterious to you now. There are a lot of choices for composers, but they’re choices, not arbitrary flips of a coin.

*There isn’t really room to go into this here, but the rhyme scheme also makes this a little dicey. The melodic tendencies that have developed over the centuries in Western music tend to be more accommodating for ABAB patterns, and the ABBA pattern obviously requires reversing the second half of that. My approach was basically to make the melodic patterns almost identical, so that it wasn’t as obvious that this contortion was going on.

**Just for starters, it will always irk me that the second A (“But last year’s bitter loving...”) doesn’t contain place imagery while the B section (“There are a hundred places...”). It would be so tidy if all the As talked about place while the B didn’t, but alas, such formalist dreams are not to be.