Tuesday, June 19, 2012

Theory of Pop

I don't know who is behind Hooktheory, but a post is up doing what seems to be a rather broad analysis of harmony in popular songs. That seems interesting, so I had a look. Here is the post. This will make more sense if you go have a read. Don't bother with the comments, though.

Two things strike me about this. Firstly, it is a statistical analysis of how often chords appear, though they don't seem to measure how long chords are held for, which seems pretty important. Secondly, it is a probability analysis of the likelihood of one chord following another. If you scroll down you will see a chart titled "Chords following em [E minor]". That looks pretty wacky too. 59% of the time E minor is followed by F major in the key of C major? That's a pretty weak progression. But maybe that is, statistically, true. The problem is that all this seems quite oblivious to the function of harmony.

I've been raging about pseudo-science lately and how researchers doing research into music who know absolutely nothing about music are just embarrassing. This site is a little different. For one thing, it is digging into the music itself instead of coming at if from left field somewhere. Remember the yellow-bellied marmot in this post? Hooktheory has a page where they put all their analyses. Here that is. Here is the analysis of "California Gurls" by Katy Perry.

The first thing I did was look for a Beatles song to see how that came out. Take the analysis of "Hey Jude" for example. The authors know something about music theory, which is refreshing, but not quite enough. Look at the "outro" (I think "coda" might be the better word) for "Hey Jude". They give the progression as I, IV of IV, IV, I. The idea of a secondary dominant is a long-standing one in music theory, but it really can't be extended to the idea of a secondary subdominant which is what is being claimed so blithely here. IV of IV? S'existe pas!

The idea of a secondary dominant is a powerful one because the relationship of V or better, V7, to I is so strong that it can be transferred to any chord. There is no strong relationship of IV to I. IV, in traditional theory, prepares V, that is its role. So the progression I, IV of IV, IV, I just isn't plausible because the relationships claimed are not plausible. The progression is actually I, bVII, IV, I. This is interesting because it is so very coda-like. In common-practice harmony, the coda has the role of lowering the tension and it very commonly uses subdominant harmony to do so. Interesting that this song, with its huge coda, does the same. But what is also happening here is the creation of an ambiguity. Any time you have an ambiguous progression that fades out, it can be an undecided question just what the final chord is. For example, suppose that the key is not F but Bb? We then have the progression V, IV, I, V. That's a lot more probable than the one they suggest and just as probable as the one I first suggested. What is actually happening here is something that always happens when you have IV I progressions. It either sounds like a plagal cadence, which is inherently weak, or it sounds like the IV is actually the I. The coda to "Hey Jude" is interesting, I suggest, because it floats between both possibilities.

Well, that's my stab at doing some popular music theory. I would suggest to the authors of Hooktheory that instead of doing a statistical analysis of 1300 songs (and not coming up with much), that they pick a few good songs and try and figure out how the harmony functions.


Dave said...

Hey Bryan,

I'm actually the author of this article (found you via Google Alerts). Thanks for having the interest in writing about us. The article was purposely very light on the "theory" side. We didn't want to scare anyone away. We plan on gradually easing into using more advanced stuff as the series progresses.

I had one comment regarding your beef with the IV/IV. It turns out that for popular music there are actually very strong theoretical reasons for functionalizing the bVII as IV/IV. Basically IV chords, unlike in classical music, very often have a dominant cadential function.

We're not the first to use this terminology and it's actually ironic that you brought up a Beatles song for your example. I Don't know if you've seen musicologist Alan Pollack's excellent analyses of every Beatles song every written (wow!), but here's a quote from him regarding this chord from one of his analyses:

"I've been often tempted to label that A-flat chord a "IV-of-IV" when used in this context. I was gratified to recently learn that Beatles musicologist Walt Everett coined the term "double plagal" to refer to this."

Anyway, thanks for the comments. Don't hesitate to leave us more feedback good or bad. We can take it!

Bryan Townsend said...

Hi Dave,

Thanks for being a good sport! I'm delighted to get feedback too!

Yes, I do know the work of Walter Everett and have read a bit of Alan Pollack's discussions--thanks for reminding me of him. Sure, I know the "double plagal" cadence that Everett identifies.

I think that what we are running across here are some interesting limitations on our current theory. We have an extremely highly developed theory of tonal harmony and my remarks are based on that. But, as you point out, the way harmony works in popular music is a bit different. As Everett and Pollack have both pointed out, it is often better characterized as "modal" rather than tonal harmony. The link you put up, to Pollack's analysis of "She Said She Said", points out that this song (and a number of others by the Beatles) has a strong mixolydian flavor. For this reason, his analysis of "She Said She Said" shows the Ab chord not as "IV of IV", but as bVII.

The upshot I think is that the modal harmony of popular music is not as well understood as the tonal harmony of the common practice period. The "double plagal" cadence of the Beatles is a useful notion, but two things occur to me: it is an inherently weak cadence and it does not quite equate to "IV of IV". They obviously wanted a weak rather than strong cadence as V7 - I sounds corny in pop music. A subdominant can have a cadential function, but not a 'dominant' cadential function.

Anyway, more power to you Dave. You might be making a great contribution to the understanding of how harmony works in pop music. I'll look over some other analyses when I get a chance.