I think the basic theory idea you are looking for is that each tone of a diatonic scale (exluding
MI) can be chromatically altered, raised a half step, to become a temporary leading tone to the diatonic tone above. And importantly, only that one altered tone is needed to create the secondary dominants. These alterations can be inserted into ordinary step-wise voice leading patterns. The result will be lots of simple step wise motion, but a fair amount of chromatic complexity. It's sort of "process driven" - meaning you just follow mechanical procedures - so while it sounds complex, perhaps "chromatically rich" is a better description, it isn't really too difficult to understand.
Take a simple diatonic line...
...insert chromatic half step as leading tones to the step above and add some harmony above enough to clarify the secondary dominant role of the chromatic leading tones...
Your example song used a descending line so let's just make the line descend...
...notice that the leading tones are now in the treble part and the descending line of the bass repeats tones, that's just how it works out when we procedurally insert secondary dominants into an ordinary descending diatonic line.
You can insert secondary dominants in this way for the whole scale. However, when the diatonic line moves through the areas of
MI FA and
TI DO those are already half steps diatonically so it complicates our procedure of inserting chromatic half steps.
Let's take a descending bass and stack up diatonic seventh chords...
...take note of where the diatonic half steps in the bass occur, then insert the secondary dominants...
...notice that things are a bit odd at the points of the diatonic half steps. The chord rooted on
B is a diminished chord so we can't really make a secondary dominant for it, and the secondary dominant moving the to
E chord - noted with
! breaks the smoothly descending bass, because of the rise to
F# necessary for a dominant seventh chord.
...if we wanted to avoid that
F# we could do a number of things, but one that I like is a tritone substitution so we can use the
F natural in the bass...
OK, we've worked out the procedure in a fairly mechanical way, and it should be clear that all we have is a series of diatonic seventh chords preceded by secondary dominants of some kind. In the results we can note that:
- all the movements from tone to tone are simple steps, whole or half
- almost all tones are diatonic, there are a total of 56 tones to play all the chords but only 5 are chromatic (I highlighted the in blue)
So, on the one hard the procedure gives us chromatically rich music, on the other hand it's still mostly diatonic steps.
My example isn't the progression of the Norah Jones song, but keep in mind I'm only trying to outline the basic theory idea: you can insert secondary dominants into most diatonic progressions. The Norah Jones progression is basically portions of the circle of fifths with secondary dominants inserted. Same concept applied to two different diatonic progressions.