The graphic matches the figures of the valse à trois temps and a rhythmically dissonant hemiola pattern. The first iteration of this hemiola (mm. 3-4) may be taken as a courtesy to the woman, whose figure runs in four steps followed by a two-beat pivot turn that does not shift weight, or a rhythm q-q-q-|q-h, which aligns nicely with the hemiola, while the man's mirroring figure initially fights against it: q-h-|q-q-q. (btw, I am using the gender labels appropriate to the period. Although other combinations and roles were certainly possible in couple dancing, I am focusing on the mode for public dancing, which is also what Schubert's friends report in their reminiscences of house balls and parties.)Whether portrait (rhythmic depiction of the dance) or not, this figure of the dance can also be taken as the basis of a reading following Marion Guck's metaphor-based method. What she calls the "music-literal"--and uses to guide analysis--might be taken as "shapes in space" or "qualities of movement." I have not carried out student surveys of the kind that support Guck's claims about the arch figure as a controlling metaphor in the Chopin Prelude in B Minor, but the experience of playing D779n13 and working out other analyses suggests strongly that the "music-literal" here is a sense of pairing (generally, lower against higher). Apparent immediately in the separation of voices on the downbeat of m. 3, this idea works itself out through the hemiola patterns.
The eighth-note figures (circled) are paired, moving from lower voice to upper voice. This latter pair is also the basis of the entire C#-major section, after which the eighth-note pairs resume (mm. 30, 32) and we eventually hear one last iteration of eighth-notes to rising quarters (mm. 34, 36). A rising gesture, or a pattern of low-->high, thus obtains consistently throughout the waltz, and I would take that as equivalent in function to the arch that Guck finds in the Chopin Prelude.
The suspensions work in a larger time-frame to express the same gesture. As the brackets and connecting line show, the suspensions remain in the lower voice (and emphasize not only "low" but "descending") through all of the first phrase except the close, where a suspension-like effect is achieved with the 6-5 over I. The same pattern of staying-low-then-ascending-at-the-end follows in the second phrase, the F#-E now forming a 9-8 suspension-like effect in m. 15 in combination with a true suspension in the alto. In the second strain, the suspensions remain firmly in the lower voice (mm. 21- 25, 31, 33) until the repetition of m. 15 as m. 35.
The abstract echoes of these patterns resound in registral motions across the first strain (F#5 and E5 till the cadence that rises to G#5 and A5). At the level of the entire waltz, the figure is not "perfect" (the waltz does not end in the highest register), and I am tempted to link the final retreat from the sixth octave to the suspensions (the suspension being a classic instance of a recessive gesture). If so, one might argue for the final integration of the two spatial metaphors, a single metrical group in which the long, asymmetrical low-to-high pair is motion toward an accent and the recessive gesture falls (literally) after the downbeat at m. 31.
Reference:
Guck, Marion. "Analytical Fictions." In Adam Krims, ed. Music/Ideology: Resisting the Aesthetic, 157-77. Amsterdam: G + B Arts International, 1998. Originally published in Music Theory Spectrum 16/2 (1994): 217-230.
