Chords / Melody ?

November 4, 2006 9:10 AM

What's the relationship between the melody of a song and its chord structure?

I've been playing guitar for some 20 years and when I'm playing a song I understand on some pre-rational level why the chords change at that precise instant, and how the song fits into it, but I've never figured it out on an abstract level that I could explain to somebody else or use to write my own songs.
I can play some blues, and know that at certain chords in a progression, certain intervals sound better than others, but I've never figured out how to go the other way: given a melody, figure out the chords that will fit it.
When I have tried to write songs, the chords tend to match the melody almost exactly (if I sing a C, I play a C chord), which of course sounds like crap.
I'm interested in the theory of this, no matter how dry, and I'm thinking of a pop/rock context.

posted by signal to media & arts (32 answers total)

Key signature has a lot to do with it-how many sharps or flats a particular scale has. If you play a song in C, it generally has certain basic chord structure-commonly for that key, C, F and G-but not necessarily. There will be chords that do NOT fit into a particular scale.

You can get a booklet of piano chording which might make it simpler for you to think about this. As to your songwriting, you might have to transpose your melody to get it to work with certain chords, OR figure out what keys fit your voice, what chords go with them, and go from there.
posted by konolia at 9:22 AM on November 4, 2006

It's not just in the pop/rock context.

You're already most of the way there. You sing a C during a C chord. Now, what other notes are in a C chord? Try singing an E or a G, and see how it changes things.

Generally, the rule is that you sing a note that's in the chord. If you're singing a note that's not in the chord, it's because you're making a new one. So, for example, if you decide to sing a B during a C chord, in total you've got a Cmaj7 chord.

So how do you figure out which chord to play at which time? Write out your melody. Next, figure out what key you're in. Then, underneath each note, write what chords in the key that note is in. Then it's trial and error. I wrote a lot more about this at this link.

Have fun.
posted by kingjoeshmoe at 9:24 AM on November 4, 2006

Oh, I should mention that with the link I attached, the part you want starts where it says, "Starting with a melody".
posted by kingjoeshmoe at 9:27 AM on November 4, 2006

There's not too much to it: melody generally uses notes which fit within the chord structure. Start learning some older standards on guitar, even just the chord structure. Soon you'll begin to get a sense of how melodies play with and against the prevailing chords and key signature.

There's no real answer to your question - melody follows chords to the extent that the songwriter wants them to.

Obversely, coming up with chord structures to strictly melodic tunes is a great exercise in figuring out what "sounds right" without being totally pedestrian. Give it a try! Come up with some alternate chord changes to your favorite tune and see for yourself what works and what doesn't.

Good luck! The chord vs melody relationship can take a lifetime to appreciate!
posted by Aquaman at 9:28 AM on November 4, 2006

Generally, the rule is that you sing a note that's in the chord. If you're singing a note that's not in the chord, it's because you're making a new one. So, for example, if you decide to sing a B during a C chord, in total you've got a Cmaj7 chord.

Sort of. It's like this: there are two kinds of melody notes. Chord tones and non-chord tones (or non-harmonic tones). This lesson is a good introduction to non-harmonic tones.

A chord tone is a note that's part of the harmony being played. If the chord is A major, A, C#, and E are chord tones. kingjoeschmoe is talking about extending the harmony with the melodic instrument. If you have a guitar playing A, C#, and E, and then you have a clarinet play G natural, you can say that the overall harmony is A7 and the clarinet is just filling it out.

But the thing that distinguishes non-chord tones is that we expect them to resolve. If you play A, C#, and E, you can sing a D natural, but you'll create the expectation that it's going to resolve to C# or E. Of course, sometimes confounding those expectations in different ways is part of good composition. Maybe you hold the D natural until the next chord change, where it becomes a chord tone. That would be called an anticipation. Maybe the D natural was part of the last chord; that would be a suspension or retardation.

Anyway, that's about it. Melodies do not have to stay within the song's key signature, and most complex melodies indeed contain accidentals. I recommend checking out that musictheory.net lesson as well as some of the others if they seem interesting. Understanding this kind of stuff can go a long way towards demystifying composition.
posted by ludwig_van at 9:34 AM on November 4, 2006 [1 favorite]

There's no real answer to your question

Sure there is, you just didn't know it.

Also, I recommend Alan W. Pollack's notes on The Beatles canon. Extremely in-depth analyses of every Beatles song, complete with lots of good melodic analysis and references to this kind of chord tone vs. non-chord tone stuff.
posted by ludwig_van at 9:37 AM on November 4, 2006 [2 favorites]

The Wikipedia entry on harmony has a few books and links at the bottom - maybe look at those and work from there?
posted by Opposite George at 9:40 AM on November 4, 2006

I'm in the no real answer crowd actually. Music is really a tension between the predictable and the unpredictable. Sometimes you sing your melody within the chord and sometimes you don't in order to vary it up and to make the music interesting. Take Dylan's Leopard Skin Pill Box Hat. When he sings the title line in the chorus he goes off-key when he sings the Pill part of the line.

You don't have to follow the rules every time.
posted by Ironmouth at 10:06 AM on November 4, 2006

I'm in the no real answer crowd actually.

Alright, I'll stop at this point, but come on. Just because you don't have a complete handle on the subject doesn't mean that there's no answer to the question. The "tension between predictable and unpredictable" also in no way implies the lack of an answer to this question.

You don't have to follow the rules every time.

No, you don't, but people who say things like that don't tend to have a very good grasp of "the rules" to begin with.
posted by ludwig_van at 10:28 AM on November 4, 2006 [1 favorite]

Like the others are saying, this goes to the heart of what songwriting is. If the question is what chord progression goes with a certain melody, the answer is really whatever works best. It's almost completely subjective.

But for the more general question of how melody interacts with harmony, I'd think about it this way: the best composers typically don't conceive of melody and harmony separately, but rather simultaneously. That is, a composer will create a melody in his or her head and just intuitively hear the chords that accompany it. To do something like that, you have to be either innately gifted or be well-versed in scales, modes, etc.

That skill can only come with extensive practice, but it has to be perhaps a different type of practice than you're accustomed to. Start by doing something like this: pick a short, very basic melody (think children's songs, holiday tunes, classic showtunes, etc.), and for each note, find as many chords as you can that fit. For instance, if the melody starts with C, there are options like C maj, A min, F maj, and Ab maj, but also D7, Db maj7, G sus4, not to mention inversions. Do this for each of the notes in the melody, and you'll start to get an idea of the range of harmonic options.

Once you start to hear these different chords along with a given note, then you can start to connect the chords in a logical progression, usually centered around a "home" key -- this is the key signature concept mentioned above. But at this point I think the key is experimentation: play around with notes and chords as much as you can until it becomes more natural, then work from there.
posted by padjet1 at 11:30 AM on November 4, 2006

I wanted to add that I got an amazing melody book from Berklee at my Barnes & Noble. It codified at least some rules for writing a melody. I don't have it with me but it was green.

Also, if you've haven't heard anything of Ludwig van's I would be rushing to do that. His music is as good as his advice in this thread (which is damn good).
posted by Brainy at 11:39 AM on November 4, 2006

I'm a little perplexed why l_v came out swinging from the outset. Does this claim touch a nerve based on other conversations you've had that go beyond the posts above?

When I read the "no real answer" comments, I took them to be saying that although there is a classic set of pop formulas that bear a relationship to older western rules of tonal harmony, in pop/rock, these rules are a moving target--made to be broken--and to pretend otherwise just doesn't tell the whole story.

Take certain sonic youth songs, as one example. How much of the tension in their songs is based on the Alan Pollack-style melodic analysis (which is undoubtedly insightful), and how much if it is that they aren't using traditional tunings, but introducing quarter-tones and even smaller changes in tuning.
posted by umbú at 11:47 AM on November 4, 2006

One more on ludwig_van's side. The fact is that the "rules" of composition weren't arbitrarily selected by a man with big hair to make it difficult for everyone: they arose out of patterns that became convention because of the fact that they sound good.

For this reason, it's good to have a strong handle on what these rules are if you're interested in understanding how melody and harmony work. Yes, it is a general truth that melodies and harmonies that employ non-chord tones will sound more complex and interesting, but it is also a general truth that a lack of understanding of how consonance (a state of a pitch in which it is stable in relation to the underlying chord) and dissonance (the opposite) work will result in a song that sounds disjointed and fails to lead the listener anywhere.

The gist of it is that your question is answerable but complex. Read up on harmony and melody– there are lots of books on the topic. Opposite George has a good suggestion, and you might try your local library or bookstore. A lot of these books will probably be excessively complex, so good luck!
posted by invitapriore at 11:49 AM on November 4, 2006

Brainy: <3 33br>
umbú: I'm a little perplexed why l_v came out swinging from the outset.

I didn't "come out swinging," I just objected to people giving answers which were unhelpful and incorrect. And nothing personal of course, but your comment isn't particularly coherent or apropos either.

This isn't some kind of subjective question about what makes a good song. It was a question about the relationship between chords and melody in a pop/rock context, which is quite definable and comprehendible. Saying "there's no real answer" is wrong and gets the OP nowhere.
posted by ludwig_van at 11:59 AM on November 4, 2006

Aw, I should've known that wouldn't work. What I meant was:

Brainy: <3<3<3
posted by ludwig_van at 12:00 PM on November 4, 2006

To clarify just a little more: music doesn't have rules. That's something that people who know a little bit of theory think. Music theory is descriptive, not prescriptive. There aren't rules in music, only idioms and expectations.

If you're trying to write in the Baroque idiom, or the jazz idiom, or the Serialist idiom, for example, there are certain things you do and don't do in order to achieve that goal. If you want to sound like Bach, you don't write parallel fifths. If you don't care or if you explicitly don't want to sound like Bach, by all means, go for it. Not using parallel fifths isn't a "rule," it's just something that wasn't done in a particular musical idiom.

Similarly, listeners approach music with a set of expectations based on their listening experience. If you're writing music for people who mostly have experience with western tonal music, which I took to be the context of this question, those listeners can be assumed to have certain expectations about consonance and dissonance, tension and resolution, etc. As I said earlier, sometimes a composer will want to satisfy those expectations and sometimes he'll want to confound them.

Of course, idioms and expectations are always evolving; new ones have to come from somewhere. But that does not imply that ignorance is the best path to innovation; personally I think the opposite is true.

Of course, loads of natural talent/intuition can make up for a lack of knowledge of music theory/structure (like it did for The Beatles, etc.; but I'd argue that The Beatles must've understood the complex musical devices they used because they employed them so cleverly and consistently - it's just that their understanding was intuitive rather than formal), but I prefer not to leave things to chance, so I like to learn all of the tricks and techniques that I can. The OP's question seemed to be coming from a similar place.

So there's plenty of freedom to be had. But that doesn't mean that there's no answer, or no way to analyze and categorize these things.
posted by ludwig_van at 12:16 PM on November 4, 2006 [4 favorites]

[+1 to ludwig_van; I'm siding with him]

The fact is that, for whatever reason, certain chord progression can sound more pleasing to the ear than others. Next time you pull out your guitar, play the following progression: [F Dm G7 C].

Doesn't that sound nice? It's partially because you recognize the progression (IV - ii - V7 - I] as an extremely common one in Western music, for better or for worse. This is (obviously) not to discount the billions of musicians from other cultures and genres of music who don't rely on these common progressions, but the fact is that we're used to them, and it's one of the reasons that they are pleasing to our ears.

Now, to tie this in with melody writing. You want to create a melody that hits a lot of these chord tones, especially on strong beats. Passing tones (non-chord tones on weak beats) are fine, and depending on the genre of music, there can be whole different "rules" (heavily air-quoted) for what notes do and don't sound good against that progression.

There's a lot more I can say on this (I'm a music student, as, I think, is l_v) - email in profile if this is actually interesting to you.
posted by rossination at 12:22 PM on November 4, 2006

Yeah rossination, I study composition. The moral of the story here, from my perspective, is like something one of my teachers used to tell me: you can do absolutely anything you want, as long as you do it deliberately.

(Unless you make aleatoric music I suppose, but that's another discussion for another thread)
posted by ludwig_van at 12:29 PM on November 4, 2006

I'm another music student, and I agree with ludwig_van on pretty much everything he said.
posted by danb at 2:36 PM on November 4, 2006

Wow, who'd have thought this question would turn out this way?

signal, here's how I understand music theory from the point of view of someone who's played rock and pop music for many years, but not studied music.

All songs can be said to be in a particular key.

Say the key is C. That means the melody will probably start and finish on the note, and the chord, C.

The song will almost certainly use the chords F, G, and A minor. The other chords you're "allowed" are D minor and E minor, maybe a B flat. Pretty much any sequence of those chords will sound like a conventional western pop song.

The way to fit chords to a melody is just to figure out which key it's in, and figure out by trial and error which of those chords fit the notes.

The other main concept I've taken in is "tension" -- the brains of everyone listening know, in an instinctive way, that your song is "supposed to" end on the note and chord of C. That's where it starts, and then it goes somewhere, and then in the end it's supposed to go back there.

Take a very conventional chord sequence. C, F and G. Just play a bar of each, it and sing the notes C, F and G with them. Your brain wants the next note and chord to be C, because that's the natural progression "back" to the starting point.

So do it again, and this time wait twice as long on G. More tension. And the second time, change the G to G7, more tension. And sing the note B, more tension. You should be feeling that the song "needs" to go back to C.

All these rules, are of course, made to be broken. Swap major for minor with any of the chords mentioned, see what happens. Swap halfway through a bar. "Trick" the ears of your listeners by making them think you're about to resolve back to C but go somewhere else.

But the variations occur mostly within those basic structures. You will never hear a pop song on the radio which starts with C and then goes to C sharp or F sharp.

Rock music might use a set of chords based on the blues scale instead. My quick recipe for a conventional metal song in C would be C, E flat, F, G and B flat only. No minor chords.
posted by AmbroseChapel at 2:47 PM on November 4, 2006 [1 favorite]

One very important part of the melody vs. chordal structure question:

Melody is developed over time against changes.

Take some of what's been said already: a bit of melody over a given chord can have harmonic and non-harmonic tones in it. So if the chord is C, a relatively harmonic melody will have lots of c, e, and/or g in it, and little tension. Non-harmonic tones can be added without creating notable tension, if they're used between tension-relieving harmonic tones—a walk down from a g to an e would use, say, an f-natural as a passing tone.

Harmonic tones falling on a beat with non-harmonic passing tones on the off beat will generally create little tension. Non-harmonic tones on the beat will create more tension.

Add to that the fact that some tones are more non-harmonic feeling than others in a given context. Against a C chord, an f is non-harmonic but not as glaring as an f-sharp. What does and doesn't feel jarring depends on the chord and the context.

So: melody over time:

A melody is not just a set of harmonic and non-harmonic notes strung over a given chord, followed by another set of notes over the next chord, and so on. A melody is a cohesive set of notes over a whole chordal structure, and that is where things get interesting. Where a particular non-harmonic note might just sound jarring against a single chord out of context, it can have a much more satisfying and complex place if the same note appears as, say, a non-harmonic note in one chord and then as a harmonic note of a subsequent chord. That f-natural that sounds so awful in a C-Major ends up sounding sweet in a D-Major. So the construction of a song where you tease and tense with a melodic phrase over the C, and then repeat that melodic phrase of the D: that's the sort of thing at the heart and soul of melodic construction.

Roughly, pop/rock theory is an understanding of what chords work against what melodies. It's a much more condensed, repetitive form than some older classical stuff, which makes it more approachable, which is fun.
posted by cortex at 3:53 PM on November 4, 2006

AmbroseChapel:
You will never hear a pop song on the radio which starts with C and then goes to C sharp or F sharp.

Half step
http://www.youtube.com/watch?v=Cg5cUIU7oHs

http://www.youtube.com/watch?v=W8Hi7c8cKPw

Tri tone
http://www.youtube.com/watch?v=7BjpuyRzmn0

OK, they prove your point more than not, but I just thought, never??

cortex:
That f-natural that sounds so awful in a C-Major ends up sounding sweet in a D-Major.

Depends on what you mean by "sweet"! (Or maybe you meant f-sharp?)
posted by nonmyopicdave at 5:08 PM on November 4, 2006

Shit. Yes, I very much meant an f-sharp there.
posted by cortex at 5:12 PM on November 4, 2006

nonmyopicdave I'm at something of a loss to know what you meant by that last link.

The first two are jazz, and yes, in case anyone thought differently, what I wrote above definitely doesn't apply to jazz.

But what was going on in that samba music that you felt contradicted me? It has a flattened fifth in the melody, was that it?
posted by AmbroseChapel at 8:43 PM on November 4, 2006

The first two are jazz, and yes, in case anyone thought differently, what I wrote above definitely doesn't apply to jazz.

Actually, it does. I have a decent knowledge of jazz tunes and I certainly had a hard time coming up with ones that contradicted your statement!

But what was going on in that samba music that you felt contradicted me?

I was just thinking of tunes that start on the root and move a tritone away for the second chord. "Summer Samba" was all I could come up with. In that link, there is an intro, but when the tune starts (vocals in) the first chord is a B and the second an Fm7.

Incidentally, after I posted I also recalled the intro to this awesome tune:

http://www.youtube.com/watch?v=5n3c8IX2WKM
posted by nonmyopicdave at 2:02 AM on November 5, 2006

Say the key is C. That means the melody will probably start and finish on the note, and the chord, C.

The song will almost certainly use the chords F, G, and A minor. The other chords you're "allowed" are D minor and E minor, maybe a B flat. Pretty much any sequence of those chords will sound like a conventional western pop song.

Eh, I can't really vouch for this. First of all, in the key of C, the song is fairly likely to begin and end with the chord C (but this is not a requirement), but that doesn't mean at all that the melody has to begin and end on C.

Secondly, you're talking about the chords which are diatonic, or naturally ocurring, in a key. None of them contain any notes which aren't found in the key signature. In any major key these are: I ii iii IV V vi viidim. So in C major that's C Dm Em F G Am Bdim. Now these are just your plain vanilla chords. If you only wrote songs with these, they'd get boring pretty quickly. There are many ways to work different chords into the mix, but again, that's something of a separate topic.

Take a very conventional chord sequence. C, F and G. Just play a bar of each, it and sing the notes C, F and G with them. Your brain wants the next note and chord to be C, because that's the natural progression "back" to the starting point.

So do it again, and this time wait twice as long on G. More tension. And the second time, change the G to G7, more tension. And sing the note B, more tension. You should be feeling that the song "needs" to go back to C.

That's pretty minimal tension, though. That's everything that's expected. Serious tension is created when instead of resolving that G to C, you go to F#m, then B7, then Em, then F#dim7, then G, then G7, and then to C, or something far out like that. (For those following along at home, that would be analyzed in C as: V ii/iii V/iii iii vii/V V V7 I). Or even just going from G to another diatonic chord besides C (which is called a deceptive cadence). Or again by using the right non-harmonic notes in the melody. I tend to think of those as two sides of the same coin - musical interest/complexity can be created by either the relationships between one chord and the next (and the relationship between any chord and the key center as well) and also by the relationship between a given chord and its simultaneous melodic content.

My quick recipe for a conventional metal song in C would be C, E flat, F, G and B flat only. No minor chords.

This seems fairly dubious to me as well, but I can't really claim expertise on metal.
posted by ludwig_van at 3:05 AM on November 5, 2006

After two posts rather unhelpful to the OP...

I've never figured out how to go the other way: given a melody, figure out the chords that will fit it.

I gained this skill not by studying theory but by simply attempting it over and over again. Eventually I started to figure it out. My dad used to play the piano and I would play along on bass. I had no idea what notes were what, what keys were, or what the tunes were. I just would fish around for notes that sounded good. And he never stopped, or even gave me any instruction. We just plowed our way through tunes. Within each song, by the end I'd have some of the structure figured out. And from session to session, I'd start to remember parts of songs we'd played before. This went on for years and eventually I got good enough so he'd take me on gigs with him.

If you don't have a similar dad available to you, you can do the same thing with the radio. Turn on an oldies station--you probably know the melodies to most of the songs--and play along. Try to pick out the chords. Do it regularly for enough time and eventually you should be able to play a song you've never played before (but have heard a millions time, so you know it) the first time through without error. I've done this many times right on the stage.

Interestingly, my wife and I had our first child recently and as a result I've done almost no playing in several months. I played a gig last night, and the ears felt very rusty. I really had to concentrate on the songs to get the changes right. So picking out changes appears to be a skill you have to maintain!
posted by nonmyopicdave at 12:59 PM on November 5, 2006

Thanks for all the answers. FWIW, this is one of my favorite questions (of mine).
posted by signal at 4:27 PM on November 5, 2006

>Eh, I can't really vouch for this.

I have to say, I'm in the anti-ludwig_van camp now, after that post. For some reason you think the question was "are you cleverer than everyone else here?".

The guy is a beginner. He doesn't know the first thing about music theory. He doesn't want to know abstruse theoretical details, he wants to know the absolute basics. He doesn't want to understand atonal pieces by Stockhausen written for forty-three vacuum cleaners and a piano full of nails, he wants to understand how conventional popular music works. I tried to give him that, in simple language. And your contribution is to sneer at me?

When someone wants a very basic introduction to a subject, it's not helpful for you to say "ah yes, but" and show off your more detailed knowledge when they get one. The thing is, I know more detail than I put in my post too. But I didn't put it in. Because that's not what the OP said he wanted.

I'm not saying you're wrong, by the way. I'm saying you're patronising, and displaying your knowledge seems far more important to you than answering the question.
posted by AmbroseChapel at 7:48 PM on November 5, 2006

Well, I have a degree in music theory, and all I have to say is this:

Think of songwriting like filmmaking. Think of the chords as the frame of the camera, and think of the melody as the people or objects traveling in and out of that frame.

Just like in filmmaking, conventions exist, and were developed because they are successful and effective. In music theory, lots of what has been said above and linked above applies here. You should know what the conventions are and how they work, if you appreciate learning.

But, don't let the rules guide what you do. If a particular shot looks good to a filmmaker, she films it and makes it work. Likewise, if something sounds good, play it and make it work. Even if it doesn't fit with theoretical conventions (which are descriptive and not prescriptive, as someone mentioned above), go with it.

That's all I have to say, because I want some chocolate.
posted by billtron at 9:17 PM on November 5, 2006

Ambrose, I'm not trying to patronize or show off. This being Ask Metafilter, I'm just trying to make sure the asker gets accurate and complete information like I wish someone would've provided me when I was first learning. Beginner or not, he said he's "interested in the theory of this, no matter how dry." This is not rocket science; it only seems difficult when the waters start getting muddied. That's all from me; feel free to e-mail me with questions/complaints.
posted by ludwig_van at 7:25 AM on November 6, 2006

The Simpsons theme song and "Maria" from West Side Story both start with a melodic tritone (C to F-sharp or some transposition thereof).

Just sayin'.

It might help you to further classify non-harmonic tones, since "there are notes in the chord and notes outside of the chord" might seem too vague to be very helpful. You can try experimenting with:

passing tones - notes that move from one chord tone to another (for example, if you were moving from G to E in a C major chord, F would be a passing tone you could throw in between)
neighbor tones - notes that move away from a chord tone and then return (e.g. E-F-E)

There are sub-types of passing tones and neighbor tones, like incomplete passing tones, which are moved to by leap and resolved by step (e.g. C-F-E) or accented passing tones, which is when the non-chord tone lands on a downbeat. This can be an interesting, tension-generating effect because you initially hear a note that sounds "wrong" but is then resolves.

Other non-harmonic tones include:

anticipations - when a note from the next chord appears over the previous chord. For example, if you're moving from G major to C major and you throw a C over the G major right before the chord change. Another good tension-release device.
suspensions - notes held over from a previous chord which then resolve. Similar to the previous example, if you held a D over the G major chord and then continued to hold it over the C major chord before resolving to a C or E.

This is by no means comprehensive, but it should give you some things to experiment with. Play around with these and figure out what sounds good to you.
posted by speicus at 4:06 PM on November 6, 2006