Also if I'm looking for interesting non-diatonic chord progressions, should I primarily look at mode mixing?
That's one way. But borrowed chords mostly function the same way in the key as the diatonic chords. So they provide colorful, but not necessarily non-functional harmony. It depends on what you really want, but I think the simple comparison is Romantic era harmony versus Impressionism like Debussy where you might have non-diatonic, non-functional dominant seventh chords, etc.
I recently have been exploring this topic myself. I identified the basic diatonic seventh chord types: major seventh, minor seventh, dominant seventh, and half-diminished seventh, and then looked for various non-diatonic harmonic patterns I could apply to them.
One of the first things I took note of was succession of chord qualities within the diatonic realm and then make contrary patterns. For example, diatonically we have two adjacent minor seventh chords a whole step apart: ii and iii
, but we do not have similarly adjacent dominant seventh chords. So, I can get a simple non-diatonic, non-functional progression with something like G7
to A7
. To avoid that A7
suggesting a V7/ii
I could use G7 A7 B7
for an emphatically non-functional progression.
I also looked at all the chord qualities for the fundamental root progression by descending fifth. Going through the whole diatonic circle of fifths I listed the two chord pairs (which I called bi-grams after a linguistic analysis term) and compared that with the 16 possible bi-grams of chord qualities. I eliminated the bi-grams that were diatonic and selected the others. A diatonic root progression by descending fifth from viiø7
goes to iiim7
. I could alter the iiim7
chord by lowering the fifth and get a succession of chord qualities that isn't diatonic: Bø7
to Eø7
.
Instead of using a circle of fifths I tried a circle of chromatic mediants. I tried both root progression by ascending minor third and ascending major third. One thing I thought was interesting is how when diatonic seventh chords move through a circle of fifths each progression has two common tones and two tones changing by step. With a circle of chromatic mediants - ascending minor third, using any of the four seventh chord qualities - the smoothest voice leading resulted in two common tones and two tones changing by half step. Both of these sequential progressions had similar voice leading re. common tones.
Finally some interesting progression can be found by holding the chord root while changing the chord quality and then moving sequentially by ascending or descending half step. Something like: start with a minor seventh chord, lower the fifth to create a half diminished seventh on the same root, then drop the root a half step and form a minor seven chord above it by dropping the root, third, and seventh, then continue that sequence. Like this Cm7 Cø7|Bm7 Bø7|...
. There are many other possibilities ascending and descending.
Some rules of thumb emerge:
- multiple successions of the same chord quality are mostly non-diatonic
- changing chord quality on a shared root is non-diatonic
- number of common tones in voice leading helps determine harmonic 'pull'
I don't really know how to word that last point concisely. Basically, it seems that regardless of the pitches being diatonic or chromatic, if the number of voices changing is only one the progression is 'weak' or 'gentle', if about half the voices change the progression is 'strong' or 'dynamic', if all the voices change, the progression is 'emphatic' maybe even 'abrupt.'
This was my approach. I was aiming for a fairly traditional sound. Most of it sounds jazz-ish or Impressionistic. You could do something different. I mostly want to say you can try to create your own tonal logic. When you dispense with diatonic+secondary+borrowed harmony there really aren't established conventions.
Another thing to try is progressions and voice leading through any non-diatonic scale. Try the double harmonic as an example. It has all kinds of interesting options with simple triads and seventh chords. By definition all of the progressions will be non-diatonic.