There is a moment, in hundreds of pieces everyone knows, when the music takes a run-up to finish and then does not finish. The harmony gets into position, the ear already knows where it is going, and at the last instant a chord arrives that is not the one expected. It is not a mistake, it is a device with a name, an age and a precise function. In Italian it is called cadenza d’inganno: the deceptive cadence.
The mechanics are simple and worth looking at closely, because they explain why it works. A normal cadence puts the fifth degree and then the first: the tension discharges onto the tonic and the discourse closes. In the deceptive cadence the fifth degree is there, identical, but in place of the tonic comes the sixth degree, which in the major mode is a minor chord. The trick lies in the fact that the expected chord and the one that replaces it have two notes out of three in common. So the ear does not hear a leap, it hears a missed arrival: almost everything is in the right place except the foundation, which has moved up a step and pulled the ground from under you.
It is a matter of the physics of expectation, not of taste. The listener has thousands of normal cadences in memory and predicts the next one without knowing it. The composer uses that prediction as material: builds it, confirms it until the last half second, and then defies it. The result is not surprise, it is prolongation. The phrase, instead of closing, has to continue, and in practice this is the real purpose of the device: to extend, to postpone, to keep things open.
The name is old and changes from language to language in a way that says how it was perceived. The Italians speak of deception and, already in the sixteenth century, Gioseffo Zarlino in his Istitutioni harmoniche of 1558 describes the cadenza sfuggita, that is, a cadence that runs away. The French call it broken, cadence rompue, and under that name it enters Jean-Philippe Rameau’s Traité de l’harmonie of 1722. The Germans call it Trugschluss, a deceptive conclusion. English speakers are divided between deceptive cadence and interrupted cadence. Four languages, four different charges against the same chord: it deceives, it breaks, it lies, it interrupts.
In eighteenth-century Neapolitan teaching this was not a rhetorical figure to be studied on its own, it was one of the basic tools for building long pieces. A pupil realising a partimento knew that if he closed every phrase with a true cadence the piece was over after thirty seconds. Avoided cadences were therefore the technical way of keeping the construction standing, and they were learned the way a hand position is learned, not as a notion.
Bach uses them constantly in his harmonised chorales, at the end of interior phrases, and the effect is always the same: the line of text ends, the harmony does not, and the chorale goes on without the breath ever sitting down. In the nineteenth-century repertoire the same thing becomes a way of making the wait for the finale spectacular, and codas stretch by dozens of bars on the strength of missed cadences.
The paradox is that in school this device is still corrected in red. A student who writes a harmony exercise and puts the sixth degree where the assignment asked for the first gets marked down, and in most cases the mark is right, because the exercise was meant to test something else. The problem arises when the student leaves the exercise with the idea that this solution is wrong in itself. It is not. It is one of the few inventions of Western music that works the same way on a listener of the seventeenth century and on one who has never studied anything, because it asks for no culture, only for having heard enough music to expect something.
To recognise it by ear, one exercise is enough. When you hear that a piece is about to end, try to sing in your head the arrival note an instant before it arrives. If the note you hear in your head and the one coming out of the speakers do not coincide, and the music, instead of closing, starts off again, that was the deception. And it is the only case in which the listener’s prediction error is exactly the result the composer had designed.