Recent years have seen an increasing influence on music theory of perceptual investigations that can be called phenomenological in the sense of Husserl, either explicitly or implicitly. The trend is problematic, particularly in what one might call its sociology, but it is also very promising. Potential or at least metaphorical links with Artificial Intelligence are especially suggestive. A formal model for "musical perceptions," incorporating some of the promising features, reveals interesting things in connection with Schubert's song Morgengruβ. The model helps to circumvent some traditional difficulties in the methodology of music analysis. But the model must be used with caution since, like other perceptual theories, it appears to make " listening" a paradigmatic musical activity. Composer/ performer/playwright/actor/director/poet can be contrasted here to listener/reader. The two genera can be compared in the usual ways, but also in some not-so-usual ways. The former genus may be held to be perceiving in the creative act, and some influential contemporary literary theories actually prefer members of this genus to those of the other as perceivers. The theories can be modified, I believe, to allow a more universal stance that also regards acts of analytic reading/listening as poetry.
A formal interval system (FIS) is an ordered triple (THINGS, IVLS, int), where THINGS is a set, IVLS is a mathematical group, and int is a function from THINGS × THINGS to IVLS satisfying three conditions: (1) from all r,s, and t in THINGS, int (r,t) = int (r,s)int(s,f) [group product in IVLS]; (2) for all s and t in THINGS, int(t,s) = int(s,t) [group inverse in IVLS]; (3) for every s in THINGS and every i in IVLS, there exists a unique t in THINGS satisfying the equation int(s,t) = i. The FIS is a useful general model for our intuitions about "intervals" between "things" in many specific musical contexts. A "time-span" is an ordered pair (a,x), where a is a real number and x is a positive real. This pair is meant to model a temporal event that begins a units of time after (or – a before) some referential moment and extends x units of time therefrom. A change of referential time-unit and a change of referential moment relabel the events (a, x) and (b, y) as events (au + m,xu) and (bu + m,yu). We seek a FIS with time-spans for its THINGS whose interval function is invariant under such transformations: int((au + m,xu), (bu + m,yu)) = int((a,x),(b,y)). There is in fact exactly one such FIS, up to isomorphism in the pertinent sense. This FIS is discussed and explored.