What you're training
The same notes can be arranged completely differently. Stack them and strike them together and you get block chords — one thick, homophonic voice. Break one part into a fast repeating figure under a slower held line and you get melody-plus-figuration — Alberti bass and its relatives. Let two single-note lines move independently, never quite landing on the same beat, and you get polyphony. Same pitches, three different textures, three different listening jobs.
Two levels, one skill
Level 1 asks the broad question: what's the texture — chordal, figuration, or polyphony? Level 2 zooms into one specific case: a chord repeats every beat while a single line moves against it — is that moving line riding on top, or is it underneath, moving in the bass?
Go deeper
The corpus by texture — the classifier's own labels → shows how real pieces actually split across these textures, straight from the corpus's own labeling. A different kind of "same notes, different arrangement" → trains the melodic-color cousin of this skill — one altered scale degree instead of one rearranged texture.
Where it fits
Comfortable with chord quality and chord inversions already? Musical texture zooms out from individual chords to how they're actually arranged in real music — the difference between reading a progression and hearing an arrangement. The full drill ladder has everything else.