The literature on Mozart's Idomeneo contains many references to short motives, initially presented in the overture and recurring at critical points during the opera itself. While this practice has not been shown to be widespread in eighteenth-century opera seria, earlier instances may be found in operas known to Mozart and from whose example he clearly profited, notably Gluck's Alceste and Iphigénie en Aulide. The tendency to orchestrate most or all of the recitative and to elide aria cadences contributes to increasing continuity of musical thought in all these works; the denouements of both Gluck's Iphigénie operas become nearly symphonic.

Fidelio has always posed difficulties for operatic interpreters. On the literal level it is straight-forward enough, much more so, certainly, than its great Mozartian predecessor, The Magic Flute. All three of its versions tell the unproblematical story of a wife, Leonore, who disguises herself as a boy in order to rescue her unjustly imprisoned husband. In the prison she obtains a job with the jailor, Rocco, whose daughter proceeds to fall in love with her (in spite of being promised to the prison gatekeeper). The central action is triggered when the governor of the prison, Don Pizarro, learns that the Minister (representing the central monarchy) has set out to visit the prison, suspecting that it harbours several ‘victims of arbitrary force’. Pizarro is terrified that the Minister will discover one particular inmate, Florestan, who had threatened to expose his crimes and whom the Minister believes to be dead (we, of course, have no difficulty recognising him as Leonore's husband). Pizarro thus resolves to murder Florestan. In the second act Rocco and Leonore precede Pizarro into the dungeon to dig the victim's grave, and there Leonore ascertains that the condemned prisoner is indeed her husband. When Pizarro descends for the kill, he is confronted by Leonore, who tells him he will have to kill Florestan's wife first, and pulls a gun on him. At exactly this moment, a trumpet call announces the Minister, whose arrival dissolves the dramatic situation with breathtaking suddenness: Florestan is rescued, husband and wife are reunited, and Pizarro's tyranny is broken. In the final scene all the prisoners are liberated, Pizarro is banished (presumably to face imprisonment himself), and the Minister, learning that his friend Florestan has been saved by his wife's courage, invites her to unlock his chains. The opera ends with a choral tribute to wifely devotion.

The use of semiotic methods and concepts in analytical studies of opera has not yet produced the results that the variety of communicative levels in musical theatre might lead us to expect. If, to repeat a frequently cited formulation by Pierluigi Petrobelli, ‘various systems work together in opera, each according to its nature and laws, and the result of the combination is much greater than the sum of the individual forces’, it seems likely that the difficulty of applying this principle may in fact be directly related to the multiplicity of ‘systems’ involved. Only theoretical enquiries that go beyond differentiating expressive levels can hope to arrive at a more satisfactory concept of this ‘system of systems’, and thus apply it in a useful way. The problem is of course too vast to be developed here; however, because the following reading of Aida involves such theoretical considerations, it may be useful to make explicit some basic difficulties involving polytextuality in opera.

The train of thought that I wish to pursue here was initiated by Edward T. Cone's recent essay ‘The World of Opera and its Inhabitants’. To the extent that my views diverge from his I suppose I may be taken for a critical adversary. But I prefer to think of the present effort as more a continuation and development of Professor Cone's ideas than an attempt to refute or criticise them. It is in the spirit of ongoing research rather than the more common one, in my profession, of philosophical confrontation that I offer remarks on the general questions, as posed by Cone: ‘How does the world of opera differ from other dramatic worlds? Who are the people that inhabit it, and what sorts of lives do they lead there?’ More particularly, my question is: What is the nature of operatic utterance? How are operatic characters ‘saying’?

In the first section of my paper I will present Professor Cone's answers to these questions. In the second I will go on what will appear, no doubt, to be a completely tangential excursion into R. G. Collingwood's philosophy of art. But in the final two sections I will try to weave these two seemingly disparate strands together into an answer of my own to the questions that Professor Cone has so insightfully raised. Perhaps ‘an answer of my own’ is too strong a phrase to use, implying something more like disagreement than is actually the case. So a better way of describing my whole enterprise, and the final sections especially, is ‘variations on a theme by Cone’.