In J. L. Austin's theory, the signal trait of performative utterances is that they are primarily pragmatic rather than denotative. Performative utterances can either inform (these are illocutionary) or induce a certain state of affairs (these are perlocutionary). To offer my own scenario, imagine I say to you, across the kitchen table, “It's hot in here.” I state this not primarily to denote a fact (the temperature) but rather to inform you that I am hot, and, even more, to get you to do something about it—to open a window or come outside with me for some fresh air. In other words, a performative utterance harbors a subtext, one that informs or elicits a particular action or response.

My aim in this chapter and the next is to argue that it is more fruitful to conceive of music-analytical statements as performative than as factual. Take a structural attribution in the form “musical entities x and y are related in z way” and assume the relation z is somewhat nonobvious. I argue that one does better to view this attribution not as an objective fact but as an implicit directive: “hear x and y as related in terms of z; imagine that relation obtains” (rather than presupposing that one already hears x and y in this z way, which may or may not be the case). That is the primary perlocution. The secondary perlocution might be “play y as if it were related to x by z.” Of course, there might be several ways to do that, depending on what x, y, and z are.