Three-Part Textures and Complete Chords

‹-- PreviousNext --›

I have been working with a couple of composers and arrangers recently who have been working in textures with three vocal lines, accompanied by a piano (in one case with several other instruments too, but with the piano at the heart of the band). A question that has cropped up with all of them is to what extent you need the vocal parts to present a complete harmonic texture if the piano is there to fill in the chords for you.

Of course, you can’t actually get complete chords in a three-voice texture unless you only use triads, but you can still make the differences between something that sounds like it is giving you enough harmonic information and something that sounds empty. All this is in the context of the harmonic conventions of western tonality as used in 20th-century popular song traditions; other conventions are available of course, but this was the world to which these particular musicians had made their stylistic commitments.

The generalisations we came up with about how this texture works best are as follows:

  • Unisons (and octaves) work fine: the ear casts the vocal lines as melody, the piano/band as harmony and everything makes sense
  • If you have more than one pitch-class sounding in the voices, there needs to be the interval of a 3rd present for it not to sound empty. Whilst this would usually be the actual 3rd of the chord, it can also be a 3rd between 5th and 7th, though you have to be a bit more careful with both voicing and how you approach/leave the chord for the ear to accept it as harmonically satisfying. An interesting exception is where the prevailing harmony is actually a sus4 chord, in which case you need to substitute the 4th for the 3rd, and also voice it in such a way that the suspended quality is aurally obvious – usually this would involve sounding the major 2nd between the 4th and 5th in the same octave.
  • Open 5ths sound incomplete if accompanied by harmony in the piano. They can work quite dramatically as a special effect if you want to feature a moment without harmony.
  • I have yet to hear an open 4th sound anything other than unconvincing; avoid them.

These points all refer to the sonorities on metrically-strong syllables, using the term metre in the poetic sense: the syllables you would accent in speech. Unless you are in something like the barbershop style that mandates full harmony for every melody note, the little syllables (of, the, -ly etc) don’t carry the same harmonic responsibility. If the salient syllables tell you the chord, the brain carries it over the intermediate details to the next, in much the same way a staccato piano chord on the first beat of the bar can give you the harmonic information you need to last you to the next downbeat.

So that’s what seems to work; I have been finding it interesting to reflect on why this is the case. Why doesn’t the ear accept a 3rd on the piano as filling in an empty chord in the voices?

This is partly about timbre. Analogous to how you’d usually want to orchestrate harmonic structure within a section: woodwind fill, or brass chorale, or string shimmer. Presenting a triad with the root on a flute, the 3rd on a trumpet and the 5th on celesta is gOIng tO sOuND a bIT liKE tHiS.

It’s also about role: in the conventions of song, the voices are about telling the story, representing the human presence, the persona of the song, while the piano/band is basically there to say, ‘Hey, listen to my friends over here, they’ve got something to tell you!’. Accompaniment divides its time between cheerleader and logistical support driver, it helps keep the show on the road while amplifying and sometimes inflecting its emotional content.

In the context of this general division of roles, the voices shouldn’t be leaving key elements of the song’s narrative to the piano. A good test of the texture is whether the voice parts sound like coherent music without the accompaniment; think of it like the melody and bass lines of a trio sonata making sense without the harmonic fill of harpsichord or lute.

(And yes I’m aware that this whole point about roles doesn’t apply to art songs; indeed one of the points of the art song is to subvert this convention and make the piano essential to the musical journey. ‘Mondnacht’ doesn’t make a lot of sense without the piano part, still less the Ariettes Oubliées.)

There’s also a point about consistency between sounded and implied notes that also applies to a cappella arrangers. The generalisations above also work for voice-only textures, because they are about the information you give people about how to listen.

If you only present melody people’s ears go, ‘Oh listen, a tonal melody! My imagination knows what kind of chords would go in there intuitively because I’ve lived in this musical culture all my life’. If you present more than one note at once, people’s ears go, ‘Oh, listen, harmony!’ and then they are primed to take harmonic information from the actual sound they’re hearing. You can switch between these two states mid-song (Four Freshmen, anyone?), but people need to be clear which mode they’re to listen in.

Or, to put it in terms I borrowed from 19th-century aesthetics back in the early days of this blog, the problem with incomplete chords is when they present mixed messages about the Real and the Ideal. The Real is the sounding pitch, objectively detectable; the Ideal is the inferred harmonic context, conjured up in the listener’s brains. You’ll need both to write music, but try not to get them tangled up.

It's an interesting and useful account. Thanks, Liz.

Glad you found it useful Terry!

...found this helpful?

I provide this content free of charge, because I like to be helpful. If you have found it useful, you may wish to make a donation to the causes I support to say thank you.


Archive by date

Syndicate content