Intro

Our conception of musical form

Before starting to annotate musical forms, it is useful to familiarize oneself with a working definition of form. Although there is no universally shared understanding of musical form, we wish to propose a working definition that contains typical components of form:

  1. Segmentation: The auditory stream of events is usually subdivided (or grouped) into larger chunks of events. Segmentation may be based on a variety of factors such as gestalt principles or closing patterns (e.g., cadences). Furthermore, adjacent groups may be disjunct or conjunct to each other (phrase overlap).

  2. Hierarchy: Groups of events can be combined into ever larger groups, thus creating hierarchical relationships between them. Groups located at different hierarchical are related by containment.

  3. Functionality: Groups do not only occur at a certain temporal position within the overall structure; they can also express their form-functionality (i.e., their temporal qualities) through their intrinsic features (see below: interfaces with musical form). They may project a sense of beginning, middle, or end at different hierarchical levels. The formal function conveyed by a group may or may not coincide with the group’s temporal position. For instance, a larger formal unit may start with a cadential pattern (hence an ending function) or with a sequence (hence a middle function).

  4. Repetition structure: Groups may be repetitions or variations of other groups, or they express a contrast to other groups.

  5. Formal templates: the ordering of groups as well as the functionalities they express may be such that typical patterns emerge which we refer to as formal templates or designs (e.g., minuet form, rondo form, sonata form, etc.). Formal designs may also be the result of the combination of two (or more) formal types (e.g., a theme combining elements from a “sentence” and a “period”); this is known as hybrid form.

An important aspect to be included is “reinterpretation”: it is a characteristic feature of music in general that some structural hearings may not be static but depend on their temporal contextualization and the subsequent course of events during listening (which may or may not necessitate a re-evaluation of interpretation). Reinterpretation (or revision) is itself not a genuine aspect of the grammar, but a kind of effect that arises from processing during listening when the grammar allows for ambiguity. In the domain of form, this subdivides into two categories: (a) the functional reinterpretation of entire units based on later events; (b) the idea of “becoming” (i.e. gradual transformation) based on later events (e.g., a main-theme codetta turning into a transition).

The ways in which features (1) to (5) are analyzed depends on style-specific realizations of other structural domains (such as harmony, tonality, voice-leading, texture, instrumentation/timbre etc.) which may be understood as interfaces with musical form.

Negative definitions of form: While it is certainly easier to come up with a list of features that, in one way or another, are associated with the concept of musical form, it is more difficult to set form apart from other concepts which may either be close to, or intersect with, the notion of form. One such related concept is “narrative” unfolding. For example, a piece of music may display the design of a “sonata form”, it may consist of functional units (such as themes, presentation phrases, and basic ideas); but whether the piece exemplifies a “per aspera ad astra” narrative trajectory (or some different narrative scheme), is to be considered distinct from the formal outline (with which it may interact, though). However, it is worth mentioning that, given the fluid boundaries of the theoretical concepts we are dealing with, the relationship between these concepts may turn out to be different at the end of our project.

Annotating musical form

As a fundamental principle we attribute formal functions to particular timespans. An annotation label consisting of a single term is interpreted as a formal function (e.g., cad for cadential). We aspire to design an ontology in which each function can be traced back to either a beginning, middle, or end function.

In addition, we may also attribute a formal type to a timespan, separated from the function by a pipe symbol (|). For example, we read the label 1st_theme|sent as “theme, realized as a sentence”.

Furthermore, a timespan can be named, which is useful for referencing it in other parts of the analysis. A name always precedes the annotation label and can consist of any common printable character that is neither whitespace nor a colon, since a colon (:) separates the name from the label (e.g., superman_theme: theme|sent). Names should be unique within a single analysis so that they can be used to identify a particular timespan unambiguously. For the sake of intelligibility, please try to use descriptive (“main_theme”, “fate_motif”) or at least consistent naming schemes (“a, b”, “α, β”).

Names can be used to indicate distant repetitions of material, be they exact, varied, partial, or mere allusions. A label followed by a name in square brackets ([]) indicates an exact (or transposed) repetition of the referenced material. For example, coda|sent [superman_theme] signifies a coda consisting of a repetition of the material previously named superman_theme. A varied repetition can be marked as [superman_theme'], a partial repetition as [superman_theme*], and a partial varied repetition as [superman_theme*']. In cases where a repetition of musical material entails a repetition of the analytical sub-tree, e. g. for a section enclosed between repeat signs or a da capo. The placeholder symbol % [name] can be used which, in absence of a named reference, stands for an exact repetition of the immediately preceding timespan’s sub-tree.

Sequences are a special type of local repetition. Here The label model (= mod) is used to identify the original part of the music that is then repeated on different scale degrees throughout the sequence. The sequential repeats are labeled as seq and may also be counted using ‘!’, as normal repeats.