The Gräzer Walzer, with the Valses nobles, represent the late style among Schubert's dances, as they were most likely composed/written down in 1826 or 1827. Of the two sets, the Valses nobles, with their often elongated and asymmetrical forms and leaning toward distinctly pianistic textures and gestures, show much more tension between playing for dancing and playing for listening. The Gräzer Walzer, with the possible exception of the last one (n12), which might have been conceived as the typically extended coda of the set, are all entirely danceable and may in fact easily be strung together to create extended sequences. Many, in fact, have very distinct Ländler characteristics, which seems a bit surprising for such a late period in Schubert's life.
A curiosity of D924 is the large number of minor-key first strains (5), all of which modulate in the second strain to the relative major key. These are n3: c#-E; n6: f#-A; n7: a-C (1st ending), a (second ending); n9: a-C; n11:e-G. All five end in the major key, except for n7, as shown.
In traditional Schenkerian analysis, priority goes to the end, and therefore all but n7 would be read in terms of the ending key, with the opening key situated in the middleground. I'll adopt that view here for sake of discussion, but in general it strikes me that this sort of bald hierarchization misses much of the expressive point of these pieces: their strains and their keys are balanced, two pictures in a locket -- and family pictures at that, as Schubert follows his earlier habit (exemplified in a skewed way in D779n13) of transposing the first strain to serve (with minor emendations) as the second. It seems to me that David Lewin's conception of key change (allied to the double-tonic complex but explicitly transformational) is a much better model.
In D924n3, Schubert plays a simple polyphonic game, flipping the priorities of uppermost and "alto" voice in the right hand. The G#5 (boxed) may sound like a cover tone to ^3 (E5) at first, but the dogged and direct cadence carries the voice leading down from its ^5, not from E. In the second strain the weight is reversed (though of course we have no way of knowing that till the cadence arrives -- but that's often the expressive trajectory of Schubert's waltz strains): B5 does retreat to the cover tone role and the cadence ultimately moves down the G#5 of m. 9 past an incomplete NN (not boxed) A5 through F#5 to E5. Thus the Urlinie design converts c#:^5 to E: ^3, and the c# region becomes a middleground prefix in the bass -- unless you decide that's a bad idea and give the C# bass note the background status it deserves in a double-tonic complex.
I will look at two additional dances, n6 and n9, in tomorrow's post. Both show more extended versions of the same patterns.
