|Emotions and Musical Analysis After Meyer
By Michael Spitzer
Philosophers and musicians have speculated on the nature of musical emotion for hundreds, even thousands, of years. Leonard B. Meyer’s Emotion and Meaning in Music (henceforth Emotion), first published in 1956, is the most important contribution to this debate in the modern era. Rather than form and expression lying on opposite sides of the fence, as Hanslick had been understood to have argued, Meyer demonstrated that emotions emerged through the cognitive processing of the music’s formal patterns. To show this at work, Meyer forged an analytical system grounded in psychological principles capable – at first in theory, increasingly in practice – of empirical verification. Meyer’s concise formulation of this idea, repeated ever-after like a mantra, runs: ‘Affect or emotion-felt is aroused when an expectation – a tendency to respond – activated by the musical stimulus situation, is temporarily inhibited or permanently blocked’ (1956, 31). What precisely constitutes an ‘expectation’ is an amalgam of universal perceptual principles (studied in the annals of Gestalt psychology and Information Theory) and culturally specific learned conventions (as codified by style historians). This dualism would become polarised and enshrined within the two branches of the mature analytical system presented in Meyer’s later magnum opus, Explaining Music: respectively, the Gestalt processes of (what Narmour would call) the ‘implication-realization model’ (especially Gap-Fill); and compositional play with stylistic ‘archetypes’ (or ‘schemas’).
Strangely, in the course of refining the analytical model, emotion in itself seemed to become less of an issue for Meyer. As its title suggests, Explaining Music shares with many of his subsequent texts a focus on musical patterns rather than their possible affective meanings. Where questions of emotion do arise, they are ostensibly consistent with the original theoretical underpinnings of the 1956 text. Two of these underpinnings are particularly crucial, in view of the radical criticisms Meyer’s theory has recently suffered. In the first regard, Meyer upheld a ‘deviation theory’ of affect, whereby emotion results through departure from a norm: ‘Hence deviations can be regarded as emotional or affective stimuli’ (1956, 32). Meyer never extensively entertains the possibility that emotion may be produced through states rather than processes; or through the appreciation of regularities (such as ‘grooves’) instead of subversions. Secondly, Meyer’s object was not ‘emotion’ per se but an undifferentiated feeling tone he termed ‘affect’. Emotions proper (‘love, fear, anger, jealousy, and the like’ ) emerged, according to Meyer, only through the ‘differentiation of affect’ in the contexts of specific ‘stimulus situations’ (19). In a brisk series of moves, Meyer interprets ‘emotional behavior’ first as purposive, then as communicative, inferring finally that ‘designative behavior is a cultural phenomenon, not a natural one’ (22). Emotion is thereby parked firmly on the cultural side of the nature/culture (or natural/learned) divide – a dichotomy which governs much of Meyer’s thinking. Just as (cultural) stylistic norms inflect (natural) perceptual processes, emotions – associated ‘through connotation, mood, or the use of a program or text’ – are pertinent only insofar as they ‘color and modify our musical affective experience’ (270). They are thus extra-musical, never penetrating to the heart of musical experience. Emotion’s role is thereby severely delimited. In a perspicuous critique, Stephen Davies argues that Meyer’s position is not essentially different from the formalism of Hanslick:
[Hanslick] thinks that thoughts of emotions prompted by music cannot be of aesthetic/artistic relevance, because the train of such thoughts must fall beyond the control of the development of musical materials within the work. Meyer’s approach is vulnerable to this attack because it treats the music as a trigger that activates the listener’s feeling in an automatic fashion, leaving associations brought from outside the musical context to give that response its emotional individuality (Davies 1994, 290-91).
If Davies were correct, this would be an ironic upshot, given that Meyer’s book begins as a polemic against Hanslick. But I shall argue that Davies misrepresents Meyer in several respects.
Three recent books epitomise a paradigm shift in research on music and emotion since Meyer: Juslin and Sloboda’s edited collection, Music and Emotion (2001); Jenefer Robinson’s Deeper than Reason (2005); and David Huron’s Sweet Anticipation (2006). Broadly Neo-Darwinian in their thrust, these new orientations appear to supersede many of Meyer’s assumptions, particularly what I have termed his ‘deviation theory’, and his separation of emotion from affect. (1) They presuppose a much richer and more plural model of emotion as a package of behavior, physiology, autonomic reactions, and intentionality. This brings with it a notion of expression as adaptive behavior, and the idea that this behavior happens within a complex and dynamic cycle of physiological reaction, appraisal, cognitive reflection, and action. (2) They present an equally diverse model of emotional types. Thus a Discrete Emotion Theory, considering the various discrete emotions as individual systems, rather than aggregating them into a collective phenomenon. In short, within the vast and burgeoning field of current emotion theory, I have found two concepts particularly helpful for my present purpose. First, a two-stage model of emotional perception as a ‘quick and dirty’ affective appraisal followed by a more leisurely and reflective cognitive re-appraisal. Second, a hypothesis that listeners may process musical patterns not in a single way, as Meyer proposed, but in at least five different affective ‘moods’, in line with the five basic emotions of fear, sadness, anger, joy, and tenderness.1 Can emotions, then, cross over from the extra-musical realm, where Meyer had exiled them, into music itself?
To explore this question, I will take the perhaps surprising tack of defending Meyer’s original model against the modernisers. I believe that Meyer’s oeuvre is rich enough to absorb much of its criticism, particularly the two strands I have identified. Reading Meyer to some extent against the grain, I will revisit his most celebrated and extended analytical study, the essay on the Trio of Mozart’s Symphony in G Minor, K. 550.
Meyer’s thoughts on emotion did in fact develop in a crucial direction, despite the appearance of continuity. The new departure was ‘ethos’, a concept discussed sporadically through the pages of Explaining Music, and defined more carefully (albeit still too briefly) three years later in the article ‘Grammatical Simplicity and Relational Richness: The Trio of Mozart’s G-Minor Symphony’. According to Meyer’s 1976 definition: ‘Ethos refers to those aspects of affective experience that remain relatively constant over time and that are the basis for the characterization of all or part of a composition’ (2000, 122).2 Inspired by Saint-Foix’s characterization of the Trio (‘moment of sunshine […] calm, reposeful, pellucid, truly idyllic […] charming […] pure and calm […] so Elysian a grace ),’ Meyer unpacks its ethos in parametric terms as: ‘The absence of extremes (or abrupt contrasts) of tempo and register, dynamics and sonority, together with the use of simple, even commonplace, grammatical/syntactic means – melodies made up of easily grasped intervals, flowing rhythms without marked durational differences, regular meter (with a touch of ambiguity at times), and common triads and chord progressions’ (122-3).
How does ethos relate, then, to the affective level of blocked and realised expectations or implications? We remember that the 1956 theory connected affect with quintessentially musical pattern processing, and emotion proper with fundamentally extra-musical associations. The pairings in 1976 seem to switch round. Now, emotion moves into the cognitive realm of pattern processing: ‘Emotional response, which changes over time, is a direct result of (and consequently congruent with) cognitive activity. It involves intricate patternings of anticipation and tension, delay and denial, fulfillment and release’ (122). Conversely, ethos is now identified with particular character, and yet, confusingly, also with ‘feeling tone’, which in 1956 was a synonym for ‘undifferentiated affect’. This semantic switch has been missed by Meyer’s critics. In Davies’s words, ‘The stimulated affect is, at first, a “feeling tone”; it is not yet a particular emotion’ (Davies, 287). Likewise, Cook and Dibben talk of ‘the kind of undifferentiated affect or feeling tone (in effect a unidimensional variable) that Meyer’s theory predicates’ (2001, 58).
‘Words, words, words’,3 as Meyer might have responded, increasingly fond of quoting the Bard in his later writings. Cutting through the verbiage, the crux of Meyer’s position, I argue, is his subsequent point that ‘Ethos and emotion invariably qualify each other’ (2000, 123). Ethos, or what the 1956 theory called emotion proper (‘love, fear, anger, jealousy, and the like’ [1956, 17]), now has equal standing with the syntax of the implication-realisation model. In fact, ethos slips comfortable into the binary two-tier patterns of Meyer’s model: just as learned, cultural, stylistic knowledge inflects and is in turn inflected by innate, natural, Gestalt processes, so the particular character of ethos now shapes, directs, and qualifies the general tensional relationships of what he now calls ‘emotion’, and is influenced in turn (see Figure 1):
[Figure 1 Near Here]
I will argue, in due course, that this trajectory of Meyer’s thinking converges in part, but not completely, with Discrete Emotion Theory. Ethos is discrete emotion all but in name. Davies’s critique is thus wide of the mark. The emotion felt by the listener does not fall ‘beyond the control of the development of musical materials within the work’; it is outlined by the dynamic contour of these very materials. The listener will only understand the emotion if he or she cognitively engages with the logical and tensional relationships unfolded within the musical material.
Where Meyer is out of line with current thinking is in his suggestion of a regulative relationship between ethos and emotion; i.e., the model of them inflecting each other. Such a view is still quite close to Meyer’s established nature/culture dichotomy, where learned knowledge acts as an external pressure point on innate dispositions. By contrast, the model I will propose in this essay (see Figure 2) sees ethos and emotion as parallel streams albeit at different speeds: a ‘quick and dirty’ channel (ethos) acting simultaneously with a more reflective, finer-grained channel (emotion). Far from being defined by ethos, or even subverting it, musical emotion supports it, relaying the same information at greater length and in finer detail.
[Figure 2 Near Here]
One of the most striking findings of new emotion studies is that listeners are able to ‘catch’ the music’s emotional profile instantaneously, well before its form begins to unfold. Of equal interest is that the intuited emotion conveys information about the piece’s likely form or genre. To a certain extent, the music’s structure is ‘caught’ with the emotion; this contradicts the established ‘information-processing’ paradigm, which holds that form is an abstract mental representation falling at the top of a ‘pyramid of processing’, long after the perception of the basic attributes of sound (see Clarke 2005, 12-16). How, then, does this ‘emotionally intimated form’, as it were, relate to the musical form unfolding leisurely in real time, the arena for Meyer’s analytical principles? Does it regulate it, match it, subvert it? To address these questions, I should consider what I take to be the five most interesting new orientations in emotion theory; that is, those which seem most pertinent to Meyer’s analytical project. As an exercise in extrapolation, this enterprise might have had Meyer’s blessing, since he repeatedly desiderates the need for future research in this area: ‘The analysis must end here [because] the rigorous analysis of ethetic relationships is beyond my knowledge or skill’ (1973, 267).
Five Concepts in New Emotion Theory
1. EMOTION AS MOOD: Musical material presented at the beginning of the piece sets the ‘mood’ of the music, understanding mood as a kind of longer-lasting, non-transitive emotion which doesn’t need to ‘be about’ anything (Robinson, 392-93). Just as emotions and moods ‘regestalt’ the world (Robinson, 128), the listener hears the music through the prism of the emotion/mood presented at its opening. (For simplicity, I shall amalgamate ‘emotion’ and ‘mood’ into a single term, ‘emotion’).
2. EMOTION AS HOLISTIC: The implicative function of emotions is holistic, embracing not just melodic patterns but the entire character of the musical material. Meyer’s own account here is very revealing: ‘Broadly speaking, ethos is delineated both by the disposition of the relatively stable parameters such as tempo and register, dynamic level and mode, and by foreground grammatical/syntactic organization (e.g. the kind of intervals, rhythmic figures, harmonies, and chord progressions)’. He also includes ‘conventional iconic gestures’ produced through their combination (2000, 122), which brings Meyer into the territory of topic and gesture theory.
3. EMOTION AS BEHAVIOR: Emotions entail not just ‘designative behavior’, as Meyer had first written (1956, 22), but the whole spectrum of ethology. Thus ‘tendency to respond’ is extended to what Nico Frijda calls ‘action tendencies, directing people towards one kind of behavior instead of another (Sloboda and Juslin, 87). For example, a fearful person faced by the fear-inducing object may react with a variety of action tendencies: closing the eyes, fleeing, or confronting the object. And all these action tendencies are different from those associated with, say, sadness (e.g. withdrawing in order to mourn or reflect) or anger (e.g. attacking or threatening to attack). Izard and Ackerman list the adaptive functions of the discrete emotions (2000, 257-60). Sadness slows the cognitive and motor systems enabling ‘a more careful look for the source of trouble and deeper reflection on a disappointing performance’ (258). Anger, by contrast, mobilises and sustains ‘energy at high levels’ (259), preparing the organism for a possible confrontation. As a miming of emotional behavior, music can be expressive in a directly iconic manner, and thus not necessarily through the negative thwarting of expectation. The music’s emotion is dynamic rather than static, in spite of Meyer’s habitual reference to emotional ‘states’ (2001, 342). Rather than displaying a sorrowful quality, sad music is imitative of sorrowful behavior. Equally, the listener’s role is not identificatory but participatory, in the imaginative act of following, understanding, and internalising the ‘contour’ of the music’s unfolding.
4. EMOTIONAL SPACE: In apparent contradistinction to music’s ‘behavioral’ dimension, which presumably needs time in order to unfold, its emotional qualities can be appraised instantaneously via music’s sonic parameters. Many researchers have corroborated that performers and listeners are able to ‘decode’ the cues associated with the expression of the five basic emotions. Patrik Juslin has mapped these cues in ‘emotional space’ (represented in Figure 3), along the two axes of ‘valence’ (negative or positive) and ‘activity’ (high or low intensity).
[Figure 3 Near Here]
Thus, for instance, tenderness is associated with slow tempo, legato articulation, large timing variations etc.; anger with fast tempo, staccato articulation, accents on unstable notes, and so on (2001, 314-15). Critically, these cues are shared by vocal expression of these emotions, leading Juslin to propose vocal communication as an evolutionary origin for musical expression (321). In this respect, he follows Meyer in identifying musical ‘behavior’ as essentially ‘designative’ in origin. Mappings of emotional space in music have been confined to its sonic parameters; crucially, Juslin’s analysis pertains to performers and listeners, but not to the form of the music – what I have called its ‘formal behavior’. Gabrielsson and Lindström have pointed out the ‘many gaps, uncertainties, and ambiguities regarding the influence of various structural factors on emotional expression’ (2001, 242).
5. THE APPRAISAL/RE-APPRAISAL LOOP: Darwin’s evolutionary perspective helps us understand the adaptive role of emotions as enabling us to make a ‘quick and dirty’ appraisal of a situation, where more considered reflection might prove fatal (as in identifying a snake). The reality of ‘affective appraisal’ shows that we really do think with our body: decisions can be triggered by physiological and autonomic reactions to novelty or perceived threat. An ‘affective appraisal’ (‘it’s a threat!’) yields to a ‘cognitive appraisal’ (‘it’s a snake!’) leading to reflective re-appraisal (‘it’s behind glass’, or ‘it’s really only a stick’). At a fundamental level, the cyclical interaction (or feedback loop) between affective and cognitive (re-)appraisals is fast and complex, as in Klaus Scherer’s theory of the unceasing ‘evaluation checks’ people carry out monitoring every-day situations (see Scherer 1984). This is true also of Robinson’s ‘process model’ of musical emotion. The point I shall argue, however, is that, on a broader level, the emotions captured by music’s sonic cues in Juslin’s diagram afford us a ‘quick and dirty’ affective appraisal, which we can ‘catch’ with seeming immediacy through what has been called emotional ‘contagion’ (see Robinson, 391-400). If so, then the formal principles theorised by Meyer are the business of the secondary and tertiary stages: of cognitive appraisal and re-appraisal. First, we ‘catch’ the emotions via the music’s tempo, dynamics, articulation, etc; then we reflect on this emotion by following the music’s interplay of pitch, harmonic, and rhythmic patterns.
Returning to Meyer, one may surmise that ‘ethos’, expressed via ‘statistical parameters’, corresponds to Juslin’s acoustic cues:
When listeners or critics describe music as sad, happy, angry, elated, and so on, [these states] are delineated by the action of what I have called the ‘statistical parameters’. These aspects of sound vary in amount or degree – for example, register (lower–higher), dynamic level (louder– softer), speed (faster– slower), continuity (gradual– abrupt), and so forth. (2001, 342)
Conversely, what Meyer ultimately called ‘emotion’ (originally, ‘feeling tone’) is expressed via the interplay of Gestalt principles, unfolded in form’s miming of emotional behavior. Ethos affords affective appraisal; Gestalt rules define cognitive (re-)appraisal. We currently lack a theory for thinking of musical form as emotional behavior. Admittedly, we know even more today than Meyer about the psychology of expectation and anticipation. But little has been written on how patterns of expectancy might mirror behavioral types associated with the discrete emotions. We also need a theory to describe the interaction between these affective and cognitive stages of appraisal. I propose that the music’s ‘formal behavior’ is a synchronic metaphor for the synchronic acoustic cue; that it unfolds the same emotion in time that was ‘caught’ instantaneously from its acoustic cues. I here borrow Juslin’s own adaptation of Egon Brunswik’s suggestive notion of ‘vicarious functioning’, which describes ‘how listeners use the partly interchangeable cues in flexible ways, sometimes shifting from one that is unavailable to another that is available’ (324-35).4 That is, different cues, such as tempo, sound level, timbre, and articulation, are partly redundant in that they may all signal the same emotion – just as different aspects in ecological perception can represent the same phenomenon. Harsh timbre is often expressive of anger; when timbral differentiation isn’t available (as in a piano piece), then staccato articulation, certain melodic profiles, or indeed the minor mode itself, can function ‘vicariously’ to express this emotion. I argue that musical form has a similar ‘vicarious functioning’. In some ways, this recuperates the venerable linguistic and philosophical trope of language as a diachronic unfolding (or unpacking) of ideas which are synchronic within the mind (a model echoed in Schenkerism itself). But it does so in a whole new light.
In what follows, I seek to extend Meyer’s ideas in two ways. First, by refracting what he actually said through concepts from new emotion studies. Second, by contextualising Mozart’s Trio within the Menuetto which frames it.
The Trio (and Menuetto)
As Meyer does in his analysis, it’s helpful to begin with a snap-shot of the Trio’s overall structure, taking in the main sign-posts. The movement is in rounded binary form, approximating to a miniature sonata. A first theme (mm. 6) for strings alone is answered by a bridge modulating to the dominant (mm. 6-14), followed by a cadential codetta figures. The development (mm. 18-26) leads back to a recapitulation of all the material heard in the exposition (mm. 26-42), abbreviated from eighteen measures to sixteen (the two bars are shaved off the bridge).
Mozart’s handling of the material is exquisitely parsimonious, as one would expect. Everything is implied within the opening measures (Example 2):
[Example 2 Near Here]
The first violin’s theme in mm. 1-2 is very rich in possibilities, opened up by its metrical ambiguity. The score says 3/4 with an upbeat, but the ear suggests otherwise, as I will consider shortly. For the present, taking the notation on trust, the quarter-notes outline a broader pattern of a rising step, B-C, which implies a continuation up to D – realised when this pitch is tonicised in the bridge by the flutes (m. 12). D as a goal is reach at a slightly earlier place when the bridge is recapitulated (the horns D at m. 34).
At an intermediate structural level (Example 3), the violin theme traces a changing-note figure, B-A-C-B – what Gjerdingen terms a ‘Pastorella’ (2007, 117-22):
[Example 3 Near Here]
Much of Meyer’s narrative pivots on the remarkable transformation of the theme in the reprise, when it is covered by the horns at mm. 26-30 (Example 4):
[Example 4 Near Here]
The horns’ half-notes on the B and C bring out the changing-note figure, which is somewhat latent at the opening of the Trio. What are lost here are the two third skips, B-D and C-E. The bigger picture, however, is the two triadic motions which support these skips: G-B-E and A-C-E. And this takes us back to the issue of metre.
Putting the score down, what our ears tell us is that the Trio doesn’t necessarily begin on an upbeat; historians of theory have recently focused on the plasticity of classical meter, which typically works against the grain of musical notation (see Maurer Zenck 2001; Mirka 2008). As Meyer points out, the opening G’s tonic status gives it a metrical emphasis. The ambiguity is fostered by the evenness of note-values and the lack of a bass or accompaniment. We are admittedly predisposed to hear the music as 3/4 beginning on G, given the ternary metre of the preceding Menuetto. But I hear a metrical shift to a duple 2/4 metre half way through the measure, after the D. D is a more stable pitch than B, and suggests a new down-beat. The first (notated) measure and a half thereby becomes three (implied) bars of 2/4 (Example 5):
[Example 5 Near Here]
There are contextual reasons for this interpretation, since the Menuetto is infested with hemiolas, as we shall presently see. The pattern is extremely short-lived, however; it is immediately superseded by a restoration of the ‘wrong’ 3/4 metre (with C of m. 1 as downbeat), until this is in turn replaced when the authentic 3/4 crystallizes with the cadential figure of m. 5 (Example 6):