This paper exposes problems with the diegetic/nondiegetic distinction as a means of describing film-musical numbers. Tracing the use of these terms from Plato to present-day cinema studies, the author identifies a divergence of meaning between scholarship on film musicals and that directed toward non-musical films. Film-musical scholars' idiosyncratic use of these terms not only poses obstacles to effective scholarly dialogues across film genres but also leads to logical problems when the standard criteria for diegetic status are combined with the realism criterion presupposed by most scholars of film musicals. As an alternative means of describing differences between film-musical numbers, the author proposes two scalar concepts. One tracks the number's degree of realism in terms of its dramatic context, performers' skill levels and preparation, identification of a fictional source for its accompaniment, and visual effects. The other measures the performance's degree of formality and intended function. At the formal end of the spectrum, there are numbers with a strict separation of performers and audience members, where the performance is intended primarily for aesthetic appreciation or entertainment. On the other end are spontaneous performances that are intended primarily for communicative or expressive purposes, which often have no fictional audience aside from the performers. In between these extremes lie communal performances, in which audience members tend to become performers in the course of the number. Such songs and dances are intended as much for community building or cheering loved ones as for aesthetic appreciation or entertainment.

This paper concerns the ontological status and authorship of the music in opera, refining and expanding the work of Edward T. Cone (1989) and Peter Kivy (1991, 1994). Their proposal that opera characters live in a world fundamentally different from ours, a marvelous place where one’s every thought and deed passes to music—and where song rather than speech is the normative mode of communication/expression—has not received the attention it deserves in opera studies. According to the prevailing understanding of operatic metaphysics, proposed by Carolyn Abbate (1991), the majority of an opera’s music is not part of the ontology of the opera’s fictional world. Music is used as a medium to represent non-musical communicative/expressive acts. Abbate’s theory is predicated upon the assumption that it is possible to separate the linguistic and musical components of characters’ utterances and accord them different ontological statuses. The former is a part of the fictional world, but the latter is not, and thus characters do not have epistemic access to it. The problem is that song is the fusion of words and music. An adequate account of the meaning and illocutionary force of such an utterance can only arise from the consideration of both constituent components. Denying characters epistemic access to the musical portion of the utterances they and others make hinders their ability to understand these utterances, and leaves the interpreter unable to explain how they gain the knowledge that causes them to act in the manner that they do. My extension of Cone’s and Kivy’s work offers a more comprehensive examination of the orchestral music, and addresses several neglected phenomena that are important to the study of modern opera.