# Do secondary Dominants have to resolve to their own tonic?

I am studying modal shifting, borrowed chords and secondary dominants, and I am a little confused of their function, if they do not resolve their own tonic. Would their function then just be a borrowed chord? For example I have the following progression. I V7/vi IV IV I V7/vi V IV I V7/vi VI VI I V VI VI

Since the V7/vi, is not followed by the vi, what is its function. Is it still a secondary dominant? Is it classified as a modal shift, or a borrowed chord? This chorus section is in D. The verse is in Bm. Which I realize is the vi chord of D, which might change the whole analysis, itself. I’m a little lost. Any help would be much appreciated. Thanks!

Secondary dominants often do resolve to their own tonic, but they can also resolve deceptively, an action sometimes hidden with traditional Roman-numeral analysis. This is what happens in your first instance.

We're in D major, as you said. The V7/vi is F♯7, which "should" resolve to Bm. But this chord actually resolves to IV, which is G.

But let's think more globally: in what contexts can an F♯7 resolve to G? Since that F♯7 was briefly in the context of B minor, let's imagine that G chord in the context of B minor, as well.

When we do so, we see that this motion to G is really a deceptive resolution of that F♯7 chord. In other words, the F♯7 to G is really V7–VI within the context of B minor. We can call this an extended tonicization, and we label it by bracketing both chords as "of vi," showing the V7–VI resolution above it.

I discuss a similar concept here.

This does not address the V7/vi to V, nor does it address the V7/vi to VI (it's unclear if that latter chord is B major or B♭ major).

It is quite usual that V7/V resolves as V7->(ii)V7:

e.g. In G the secondary dominant is A7 of D: but between A7 and D we find am7-D7 (kind of a double suspension of the 4-3 and 9-8). But often the new root of the Dominant (D) is already there).

Actually the secondary dominant can resolve back towards the first dominant, let me explain what I mean.

Let's use C major in our example.

If you have C: V/V, you basically have the dominant chord of the dominant note of C Major, which means in essence G: V. This gives you a D major chord with a f# that has to resolve to a G, now this G major chord can be a couple of things. It can be C:V or maybe even B:V I.

This would then be what is called a pivot chord, one which fits in two keys, something that is used to make modulations smooth.

Does the tritone in the 'secondary dominant' resolve properly? Take your example of III7-IV. In C major that's E7 to F. Well, the tritone in E7, D and G♯, would be expected to resolve to C and A in the A minor chord. But those notes are also in the F major chord. So yes, I think we can usefully analyse this as a (deceptive) dominant function.

'Borrowed chord' isn't a function. It's just an excuse for something you can't fit into a nice neat diatonic or 'cycle of 5ths' pattern.