In his magnum opus, The Corded Shell, the American Peter Kivy, a remarkable contemporary music aesthetician, argues that a theory of musical expressivity can only be developed if we first clarify what exactly we mean when we say that music is expressive.

When we speak of expression, according to Kivy, we usually assume that the emotion is actually present. Downward-turned corners of the mouth are only an expression of sadness if the person in question is actually sad. The downward-turned corners of a St. Bernard’s mouth merely have the appearance of sadness without the St. Bernard actually being sad. When Kivy talks about the expressiveness of music, he means the latter: „Music is sad,“ for example, means that music has the expression of sadness without being sad itself.

The reason music cannot be sad in the same sense that a human being (or a St. Bernard) can be sad is that only sentient beings can have feelings. The absurdity of the assumption that music itself is capable of sadness is also evident in the fact that we do not respond appropriately to such a finding when we listen to music: we do not go and try to comfort it.

How accurately can music express feelings? Is it possible, for example, to establish detailed connections between musical works of art and the biography of their creator, as Cooke does with Beethoven? Kivy discusses as an example a passage about the first movement of the Hammerklavier Sonata from J.W.N. Sullivan’s Beethoven biography: „These cold harmonies, so characteristic of Beethoven’s late work, no longer convey the warm, human confidence of a man who knows that victory will ultimately prevail. They express a strong, sober determination, fearless but lacking the liveliness of joy in contradiction.“

Kivy cites this passage as an example of the belief that music is an expression of certain feelings. At the same time, however, it illustrates the unyielding precision with which the expressive content of music is often described, and the unprovable effort to pin down biographical details in music.

Kivy deliberately remains vague in his classification of feelings. Vagueness, he believes, is entirely appropriate to the subject matter. In response to objections raised by his critic Anthony Newcomb, he distinguishes – in a stylistically somewhat awkward manner (as he himself admits) – between general expressive properties (Geps, from: „gross expressive properties“) and subtle expressive properties (Seps, from: „subtle expressive properties“). Geps refer to general feelings such as sadness, happiness, joy, anger, and so on. Seps refer to very specific feelings, such as the determined coldness of a disillusioned old man in the example above. This distinction largely coincides with Hanslick’s.

If only Geps are used in descriptions of music, they risk remaining trivial. Seps, on the other hand, cannot be rationally defined, at least in pure instrumental music. To avoid triviality, Kivy also allows so-called Meps (from: „moderate expressive properties“) as possible descriptions of music. Meps form the middle ground between Geps and Seps.

As mentioned above, Seps cannot be identified in instrumental music. However, lyrics that accompany music (i.e., song lyrics in particular) can transform Meps into Seps because the concrete content of the lyrics results in a concretization of the corresponding expressive content of the music. In other words, words create Seps, they do not reveal them. For example, the program of Berlioz’s „Symphonie fantastique“ can give the Meps found therein more precisely definable content. The „Symphonie fantastique“ with a program is therefore a different work than the one without.

What guarantee do we have that our description of music in Meps is actually accurate? Kivy remains vague on this question as well, but this time the question is whether the vagueness is justified—we are back to the question that Hanslick avoided and Cooke failed to answer.

According to Kivy, the undisputed, intersubjectively discussable experience of the listener is proof that a piece of music actually possesses this or that meps. Kivy points out that there is usually consensus about whether a piece of music has this or that meps, thereby exposing himself to a similar suspicion of circularity as Cooke, who claims that if someone does not feel the music correctly, then they are simply unmusical. However, Kivy’s argument may also be circular when he defines meps as those properties of music that are subject to intersubjective consensus.

A more detailed analysis of Meps would require theoretical considerations of the logical, psychological, and epistemological mechanisms of musical expressiveness; however, Kivy does not provide this, which he himself seems to regard as a shortcoming. His theory of musical expression merely provides „a description, but not an explanation, of musical expressiveness.“

As already mentioned, Kivy rejects the idea that music possesses feelings in the same way that humans do. Music is not sad in the same sense that a sentient being is sad. Further preliminary work is therefore needed to formulate a theory of musical expression, taking into account the relationship between a feeling G and the music that expresses G.

Kivy discusses two possible positions. The first, which he calls arousal speech theory, makes three claims:

• Music has the expression of G because it arouses the feeling G in the listener.
• It arouses the feeling G in the listener because it resembles the human voice when expressing G.
• Listeners recognize the similarity and feel a similar emotion due to empathy or compassion.

Listeners therefore behave similarly to children who start crying when they hear someone else crying.

Kivy raises the usual objections to the arousal theory. First of all, it is not clear why we would expose ourselves to music that expresses feelings of suffering, since it must cause us suffering. In fact, however, we also find pleasure in the experience of such music.

The argument is important, but it stands or falls on the precise characterization of the nature of the alleged pleasure (to avoid the accident trap: people are similarly fascinated by accidents and harm to others that actually express suffering). Kivy seems to be insufficiently aware of this problem. Secondly, it can often be observed that the feeling evoked by music in the listener is not identical to the feeling expressed by the music, which is incompatible with point 3. Thirdly, Kivy cites an objection raised by Richard Wollheim: A prerequisite of the linguistic theory of evocation is the view that the expressive characteristics of music are not consciously transferred to the listeners. It thus shifts characteristics of music that we consider to be obvious to the level of hidden or dispositional faculties.

For these reasons, Kivy opts for the second theory, which he calls cognitive speech theory.

Two points characterize this theory:

• Music has the expression of G because it represents the expressive characteristics of the human voice.
• The listener recognizes and identifies the signs (icons) created by the music for this purpose. These two points are compatible with a third: the identification of the represented emotion G triggers an emotion in the listener that is not necessarily identical to G.

At the heart of speech theory of perception is the concept of representation. In fact, the question of how the representation of feelings is possible in music is extremely complex, and Kivy devotes a separate book to it, which continues the project of „The Corded Shell.“

Kivy does not reject the idea that we recognize certain feelings in music because it imitates nature (mimesis), but he points out that „The Corded Shell“ cannot provide a theory of how this imitation is possible. He explains the existence of the phenomenon in a biological way: we hear shapes in music because we are predisposed to see general shapes in forms. In human history, this predisposition has provided an evolutionary advantage – but the corresponding creative process remains unconscious.

The unconsciousness of the way in which we create similarities should not be confused with the unconsciousness postulated by the language theory of awakening in relation to the feelings expressed by music. The latter are only conveyed with the help of similarities and are therefore only to be found at a higher level.

As prerequisites for formulating a theory of musical expression, Kivy outlines possible modes of representation, thereby approaching Deryck Cooke’s conception: The first – which he calls contour theory – is the one already presented, which assumes that music expresses emotion G on the basis of some similarities. The second – which he calls convention theory – sees habitual associations between certain musical and emotive features as primarily responsible for representation. Such features need not be structurally analogous. Both contour and convention play a role in musical expression. There is also the genetic hypothesis, which states that expressiveness based on convention can be traced back to original expressiveness based on contour.

This provides the building blocks of the theory of musical expression. We formulate it in four points: The music of the Western tradition consists of patterns (features) that belong to at least one of the four categories:

i) those that show similarities to expressive behavior of any kind.

ii) those that do not in themselves show similarity to expressive behavior, but in certain contexts contribute to forming similarity.

iii) those that at an earlier point in history showed similarities to expressive behavior and at a later point are regarded as expressive due to convention or habit.

iv) those that never bore any resemblance to expressive behavior but which, at an earlier point in history, contributed to creating resemblance due to their syntactic function and which, at a later point, following a change in syntactic function, are regarded as expressive due to convention or habit.

Points i and ii concern the contributions of contour theory to musical expression, while points iii and iv concern the changes in contour into convention postulated on the basis of the genetic hypothesis: iii refers to the historical change in the characteristics covered by i, iv to those covered by ii.

An example illustrating point i of the theory: Where do we find similarities between musical characteristics and expressive behavior? Kivy cites a well-known example: the melodic line that imitates a plaintive sigh. It can be found, for example, in the last act of Mozart’s „Le nozze di Figaro.“ In her F minor aria, Barbarina laments the loss of a brooch („L’ho perduta, me meschina! ah chi sa dove sarà“) with a falling minor third on „sa dove sarà,“ which emphasizes the characteristic, pain-evoking minor sixth of the minor scale on a heavy beat.

According to Kivy, the same figure can be found in the secondary theme of the first movement of Beethoven’s Pathétique Sonata (which is even in F minor in the recapitulation) and in Desdemona’s aria in the fourth act of Verdi’s Otello („Salce! Salce!“). However, the strong sighing third in Otello’s aria seems appropriate to the text, but not to Barbarina’s lament over the loss of her brooch.

Point ii can be clarified with the changed function of the diminished triad. In classical harmony, it remains in need of resolution; it is active in the sense that it must lead somewhere. Syntactically, it remains incomplete in traditional contexts, like a predicate without a subject. However, it does not have this function in itself, and in the broader harmonic understanding of later times, it is largely lost. Nevertheless, we still hear it in such contexts as restless, tormenting, and tense. The example is poorly chosen, however, because the diminished triad originally belonged to category ii rather than category i.

Kivy’s example to illustrate point iii is the chromatic scale in certain melodic contexts. He does not elaborate further, but he is probably thinking of formulas such as the chromatically falling Baroque sigh motif.

Finally, point iv can be illustrated with the help of the minor triad and the major triad in certain contexts. The minor triad used to have a similar function to that of the diminished triad between 1700 and 1900 (as demonstrated, for example, by the Baroque custom of ending music in a minor key with a major triad that can stand alone – with the so-called Picardy third). However, we tend to associate the minor triad with sadness today precisely because we perceive echoes of its incompleteness as an expression of grief in earlier music.

As already mentioned, despite his defense of musical expression, Kivy sees himself as a formalist in the Hanslick sense. This position is consistent because his theory does not claim that music has feelings as its subject matter, but merely that it carries the expressive character of certain feelings. The linguistic theory of perception asserts that the feelings experienced by the listener when listening to music are not necessarily identical to those expressed by the music. There is music that represents feelings, and even music that tells stories. However, the vast majority, especially instrumental music, is expressive but not representative.

The question therefore remains as to what purpose music serves in terms of Geps and Meps. According to Kivy, the usual analyses of pure instrumental music can be divided into three types (for reasons of clarity, Kivy’s point 2 is considered point 3 in this list and vice versa):

• They are purely formalistic and consider the quality of musical expression to be non-existent or irrelevant.
• They are purely emotional and reduce instrumental music to secret programs.
• They exhibit mixed forms, but do not treat the quality of expression as formal structures, even if they are embedded in such structures.

Kivy proposes what he believes to be a completely new alternative: in his view, qualities of expression are those of musical structure (insofar as they are constitutive of form). This means that an analysis of pure instrumental music in Kivy’s sense has the task of breaking down the structural and syntactic function of the qualities of expression.

Geps and Meps therefore do not have the function of communicating feelings, but are grammatical components for constructing music, just like harmony, rhythm, and so on. Viewed in this light, formal analysis without analysis of expressive qualities is incomplete; formal analysis that assigns expressive qualities a merely semantic or representational function is even wrong.

The end of Sound and Sentiment forms the outline of a research program. Two points are at the forefront: The task of the expressive qualities of music in pure instrumental music must be clarified in the sense of the above alternative to conventional analyses. Secondly, the genuinely aesthetic experience of beauty (and how it enters into music) requires closer examination. Kivy is convinced that this beauty exists and that it moves us. However, even he is left puzzled as to how this is possible.