Wits and Interpretation

Bengt Edlund

**Wits and Interpretation** 

Keyboard Thoughts

Bengt Edlund

# **Wits and Interpretation**

Keyboard Thoughts

#### **Bibliographic Information published by the Deutsche Nationalbibliothek**

The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available online at http://dnb.d-nb.de.

#### **Library of Congress Cataloging-in-Publication Data**

A CIP catalog record for this book has been applied for at the Library of Congress.

Open Access: This work is licensed under a Creative Commons Attribution CC-BY 4.0 license. To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0/

© Bengt Edlund, 2023

Peter Lang – Berlin · Bruxelles ∙ Istanbul · Lausanne · New York · Oxford This publication has been peer reviewed. www.peterlang.com









# **Preface**

Music cannot be the art of sound unless someone plays or sings it, and this applies even when we read a score. Notated music bears the implicit demand that it must be performed, audibly or silently, and this in turn means that someone has to understand, to interpret, what is written. No matter whether we contribute to our musical culture as listeners or as musicians, interpretation emerges as a core activity, engaging a wide range of cognitive abilities – our intuition as well as our wits.

The sixteen texts presented in this book do not make up a whole, but there is a common thread leading from philosophical issues via music analysis to artistic decision-making. Some of the essays take up a critical standpoint, and some proposals may appear controversial. The music to be discussed is piano music since this is a domain of which the author may claim to have hands-on knowledge, but most of the thoughts are valid beyond the keyboard.

Adopting the musician's point of view, the first essay makes up a contribution to musical ontology while the following one discusses the prerogatives of the composer and the duties of the musician. The next three texts present various aspects of the musicians' freedom vis-à-vis the score.

Then follows two excursions dealing with two fundamental conditions of music making: the way you feel the music with your body, and how you hear it.

It is often held that tonal reduction is of great value when it comes to interpretation. Four Schenkerian analyses are subjected to critical scrutiny, and the outcomes indicate that this view is contestable. There are other, less theoretically committed approaches that emerge as more productive if you want to probe into a piece of music.

It seems that a core aspect of interpretation is to find out how the music continues from moment to moment. This approach to analysis, opening up for a sharpened sensitivity to musical change and for the inclusion of elements of human import, is applied to three works.

The two final texts are case studies, showing how analytic observations of various kinds may lead to insights of relevance for interpretation.

I wish to express my gratitude to Sten K. Johnssons stiftelse which has generously supported the printing of this book.

Lund, 19 August, 2022

*Bengt Edlund* <046.131466be@gmail.com>

# **Chapter 1 On scores and works of music. Interpretation and identity**

# **Introduction**

According to a widely held view in 20th-Century aesthetics, a music work is equivalent to the performances that conform to a certain score. And this notion complies with at least three requirements of a satisfactory ontology of the music work, or so it seems. 1

Thus, although a Beethoven manuscript would command a very high price, it is not very interesting as a unique physical object from a musical point of view. It may of course be valuable for collectors or have affection value, and it is indispensable as a source, but unlike, say, a van Gogh painting it has no value that any uncorrupted copy of it does not have as well. Beethoven scores in general, on the other hand, are musically crucial because they are records of compositions, and because they may give rise to performances.

The association between a score and its performances means that the focus of the ontology is transferred from signs on paper to the domain of sound events. This is certainly a step in the right direction since (leaving some varieties of esoteric music out of account) an association with sound appears to be necessary. 2

Finally and perhaps most importantly, the reference to the class of conformant performances amounts to acknowledging, albeit by implication, the need in

<sup>1</sup> There is much music that should not be categorized as music works, however; cf. Zofia Lissa, *"Über das Wesen des Musikwerkes", Die Musikforschung* 21(1968), 157– 182. It would be problematic for any ontology of the music work to mix up Beethoven sonatas, jazz jam-sessions, gamelan playing, and Eskimo songs of insult, and due to historical changes qualifications must be introduced even within specific traditions and genres. The present essay, as well as the writings to which it refers, is concerned with instrumental art music from the last few centuries, music taken down in standard musical notation.

<sup>2</sup> This is not to say that music is only an auditory matter. There are residual qualities in scores, things that can be seen but not heard, and qualified readers do not need to actually hear the music. Generally, if the musician is granted an aesthetic status on a par with that of the listener (which seems reasonable), further properties emerge that the listener knows little or nothing about; cf. Bengt Edlund, "The Phenomenology of fingering", ch. 7 in *Chopin. The Preludes and Beyond*, Frankfurt 2013, Peter Lang Verlag, and "A comprehensive approach to musical idiomatic", Chapter 6 in this volume.

music ontology to take account of interpretation. A music work must have some *musical* properties, but scores, unless they are somehow interpreted, are devoid of such qualities.

Nevertheless, the current idea of the ontology of the music work is deeply problematic.

It seems that an unreasonably heavy burden is placed on notation when it is both used to secure the correspondence between a score and its performances, and to guarantee the identity between the members of the class of performances that are to constitute a certain music work. On closer consideration it appears that this kind of ontology entails some serious mistakes as regards the purpose and nature of musical notation.

Furthermore, whether due to a fear of getting too involved in mental issues or to ambitions to provide a water-tight argumentation, the musical properties that are actually allowed to enter into the interpretation process, and thus are included into the performances making up the music work, are quite restricted. And this restraint turns out to be well-advised, indeed necessary, since if interpretation in the current and quite comprehensive musical sense were allowed, the identity between the performances constituting the music work would be at risk. The reluctance to open Pandora's box indicates that the notion of 'identity' used when construing the ontology of the music work is unduly impoverished from a musical point of view. It seems that we are dealing with the "identification" of the work involved, rather than with its "identity".

# **Nelson Goodman's ontology of the music work**

The most discussed and also most influential version of the score/performance notion of the ontology of the music work is no doubt the one advanced by Nelson Goodman. 3 According to his account, paintings are "autographic" while works of music – being recorded by means of a notational system – are "allographic". And further: "In music, the work is the class of performances compliant with a character [i.e. a score]". (p. 210) In order to understand this properly it is necessary to recall the specific meanings that Goodman's theory assigns to the terms "notation", "score", and "performance".

In western standard musical notation it is only the subsystems specifying pitch and duration that qualify as "notational". Only these inscriptions satisfy the syntactic and semantic criteria that Goodman lays down; only these signs

<sup>3</sup> Cf. Nelson Goodman, *Languages of Art*, Minneapolis 1968.

have sufficient precision to define a structure, and consequently "scores" are exclusively made up of such signs. And it is the inscriptions specifying pitch and duration that must necessarily and with no exception or deviation be respected in order to produce a "performance" that exemplifies the work; indeed, there are no "performances" of a work but such strictly compliant, exemplificative ones. 4 This implies that what a "score" defines, and thereby establishes as the basis of a music work, is a pitch/time structure.

Other inscriptions to be found in actual scores, for instance signs that refer to dynamics, are imprecise and therefore not "notational". Such "characters" are not definitive of music works, and do not belong to the "scores"; consequently, they must not necessarily be respected when generating the class of compliant "performances" that make up a certain music work.

This essay aims at showing that Goodman's ontology of the music work is untenable – the role assigned to notation is musically counterintuitive, and so are several consequences of his line of reasoning. As will become evident, the core of the criticism is that the preoccupation with notation and pitch/time structure conceals more important issues. Whereas it appears that Goodman merely accounts for a set of strict identification (rather than identity) conditions applicable to performances of a certain composition, a musically comprehensive approach will be proposed that includes the essential, constitutive traits that make up music works as the kind of phenomena we actually encounter and store in our minds. 5

<sup>4</sup> Jerrold Levinson avoids this unfortunate, excluding sense of "performance" by stipulating a distinction between "exemplifications" that must conform perfectly to the score, and "performances" that may feature deviations from it and that thus represent the far more common case. Levinson also individuates, gives a richer identity to, music works by taking account of various contingent facts, such as historical context and medium of performance, an approach that is different from and yet, with respect to its general aim, consonant with the one to be pursued here; cf. "What a Musical Work Is", *The Journal of Philosophy* 77(1980), 5–28. According to Kendall L. Walton, mistakes when playing give rise to bad performances – if there are deviations from the pitch/time structure specified in the score, this will necessarily harm the musical relationships that the musician wants to convey; cf. "The Presentation and Portrayal of Sound Patterns", *In Theory Only* 2(1977, 11/12), 3–16.

<sup>5</sup> This means that two different ontological problems and two kinds of "identity" are involved. Apart from the way in which Goodman chose to deal with the ontology of the music work, nothing prevents that he might have had some sympathy for a more inclusive approach to musical ontology.

### **Notational/non-notational vs. structural/interpretative signs**

Beyond being an exaggeration, it amounts to introducing an alien principle when Goodman claims that the "primary purpose" of the "notational" subsystems found in a musical score is to make "possible recovery of score from performance" and to "ensure identity of work from performance to performance". (p. 183) It is true that the standard musical notation has been used prescriptively by composers, but in order for the notation to serve as a key concept when defining what a music work is, the crucial and unexceptional agreement between a score and its performances taken for granted must correspond to a very strong normative authority on the part of the scores and a complementary, very loyal attitude on the part of the musicians – conditions that have not always been met. The composers have sometimes been quite permissive, and current artistic practice has from time to time and to various extent accorded the musicians the right to deviate from the scores.

In addition to such tendencies we must take into account the obvious fact that not all details and aspects of a composition are equally important. The exact reproduction of certain details of the score is immaterial, and this holds both with regard to their musical importance and their pertinence for correct exemplification/identification of a certain music work. 6 And the distinction between work-defining and less important inscriptions is not congruent with the distinction made by Goodman between "notational" and "non-notational" signs. In other words, whether the signs to be found in a score are ontologically crucial is not a matter of their precision of reference.

Dynamic marks, for instance, may very well be constitutive of a work despite the fact that they are approximate, whereas signs exactly prescribing pitches or durational proportions may be non-structural. Two examples from Chopin's nocturnes clarify this point. The dynamic marks in mm. 27–30 of Op. 32, No. 1 may very well have structural significance since they introduce a sense of dialogue; cf. Ex. 1a. Playing dotted note-values in the initial phrase of Op. 9, No. 1, on the other hand, cannot reasonably be said to change the structure and hence the identity of the work; cf. Ex. 1b. Whereas Goodman for his purposes must

<sup>6</sup> Although (particularly) the pitch and time characters are considered to be nonnegotiable no matter why the composer wrote them down as he did, one might suspect that there are passages that, notwithstanding their seemingly definitive and intentional appearance in the score, actually contain elements of arbitrariness. When it comes to language, we all know and acknowledge that there are different, but virtually equivalent and interchangeable ways of putting things.

"categorically require full compliance with the specifications" (p. 187), neither musical reflection, nor musical tradition/practice demands quite that much.

One might think that what Goodman wanted to capture with his idea of "notational" and "non-notational" characters is the distinction between "structural" signs, which only composers are entitled to inscribe and which musicians are obliged to observe, and "interpretative" signs, which specify how the musicians are to render the structure and which are less binding since they may be regarded as being inscribed by the composer as an interpreter of his work. This distinction is very attractive for people involved in practical interpretation work rather than beset by ontological worries. 7

That Goodman may have had an association between "interpretative" and "non-notational" signs in mind seems to appear from his remark that "tempo specifications cannot be integral parts of the defining score, but are rather auxiliary directions whose observance or non-observance affects the quality of the work but not the identity". (p. 185) But while the imprecise tempo indications – or for that matter exact metronome specifications – are (correctly or mistakenly) regarded as "interpretative", it should be pointed out that the tempo might in fact alter the identity of the work. A faster or slower tempo may suggest another pulsebearing note value, and the changed metric framework may in turn give rise to perceptible, or even quite decisive, qualitative differences. Indeed, such tempodependent changes in character can be induced by the structure itself without any change of the actual speed of the music. The middle section of the fourth movement of Beethoven's Piano Sonata Op. 26 suggests a doubled pace rate, making for a fiery *Presto* within the otherwise smoothly flowing *Allegro*; Ex. 2.

Thus, Goodman's distinction between "notational" and "non-notational" characters is not congruent with that between "structural" and "interpretative" signs. The musical structure must be taken to include more than just the "notational" pitch/time structure since, for instance, inexact "non-notational" dynamic signs may crucially contribute to define the composition's structure, whereas, say, the exact rhythmic signs sometimes merely suggest transient and inessential inflections in the performance.

<sup>7</sup> The distinction between "structural" and "interpretative" signs is advanced and discussed at some length in Bengt Edlund, "*Sonate, que te fais-je?* Towards a theory of interpretation", Chapter 2 in the present volume, and also appearing in *The Journal of Aesthetic Education*, 31(1997), 23–40.

# **Categorical production and categorical perception**

William E. Webster has shown that not even the "notational" systems within the standard musical notation fulfil the criteria of water-tight reference required by Goodman. 8 But devising enharmonic or rhythmic dilemmas is to miss a more fundamental problem, a problem that goes deeper than the annoying shortcomings pointed out by Webster. Goodman's theory obviously takes the notation to be a sequence of discrete marks, but although musical notation may look atomic, music – essentially made up of relationships between sound events – is not.

There is no need to enter into niceties of intonation to bring out that the signs for C♯ and D♭ are non-redundant, and that even the seemingly non-equivocal character of, say, C(-natural) turns out to be ambiguous. To realize this, it is necessary to consider the semantic job done by the "notational" system specifying pitches. Jointly these characters do not denote the physical frequencies of sound events, but refer to tonal positions building up musical relationships and *Gestalts,* patterns that are mediated from composers to musicians, and from musicians to listeners by means of deeply ingrained conceptualizations.

Hence, in a consequential notation the character C♯ is chosen for good reasons, and it can very seldom be exchanged for D♭. Although the same key is to be played on a piano, these inscriptions do not refer to the same *musical* entity, and what they mean cannot be abstracted from the musical context. C as tonic is a world apart from C as leading-note to D♭, and the interval E/G, for further instance, may function as the upper third of a C-major triad or as the lower third of an E-minor triad; it may top an A7 chord or it may be part of an ambiguous diminished-seventh harmony. The interval F♯-over-C is an augmented fourth "wanting" to expand, whereas G♭-over-C is a diminished fifth promising contraction.

If this musical view is adopted, it becomes much less important (for musical ontology, not for music instruction) whether the pitches actually played on (say) a violin comply exactly with the pitch symbols or not, and whether the compliance classes intersect, thus violating the requirement of semantic disjointness, when the intonation is free, as it is when playing the violin. What matters (at least in tonal music) is how the tones are apprehended as *musical*, phenomenal entities; whether, say, an F♯ in a C-major context is understood and rendered so as to be

<sup>8</sup> William E. Webster, "Music Is Not a 'Notational System' ", *Journal of Æsthetics and Art Criticism* 29(1970/71), 489–497, and "A Theory of the Compositional Work of Music", *Journal of Æsthetics and Art Criticism* 33(1974/75), 59–66.

about to rise to (and perhaps transiently tonicize) G, or played so as to suggest a forthcoming descent to F, which would be the proper thing to do for a G♭. Correct musical "spelling" is replete with meanings, and it cannot be simplified without loss or destruction of information.

Turning to the subsystem of durational notation, a similar picture emerges. Needless to say, there are many different and yet score-compliant ways to play a certain sequence of note values: just like pitch intervals, durational proportions is not only a matter of categorical perception, but also of "categorical production". And just like tonality, rhythm is a relational phenomenon. Individually, i.e. when each sign is read in relation to the immediately preceding or following sign, every note value represents a certain, strictly defined durational proportion, but jointly the *musical* significance of these signs, and in turn the actual temporal values giving rise to the rhythmic qualities of a performance, stems from other sources, such as the pace of the music, the character of the melodic motion, the metric position, and the rhythmic grouping. Consider, for instance, the sequence♩ ♪|♩. It makes a great difference if the eighth-note is conceived of as, rendered as, and heard as an afterbeat or an upbeat.

#### **Recovering scores from performances**

Goodman's theory of the music work requires full compliance with the characters of the "notational" sub-systems of notation, otherwise "all assurance of workpreservation and score-preservation is lost", otherwise the "score" (i.e. the pitch/ time structure) cannot be recovered from the performance. (p. 187) But since our understanding of music is not a matter of identification of sound fragments, but operates in terms of meaningful musical patterns, this view must be qualified.

The importance of musical conceptualizations and particularly the effects of categorical perception/production are understated in a notation-based view such as Goodman's. 9 If musically literate listeners are able to write down a passage of music correctly after hearing it – as they sometimes can –this is not due to any unexceptional fidelity to the score or to any neat semantic disjointness of the signals, but to the triple fact that the listeners are assisted by the musical context, that a number of concepts basic to notation are shared by composers, musicians, and listeners, and that the actual values corresponding to the various categories of pitch intervals and durational proportions intersect.

<sup>9</sup> Cf. Eric F. Clarke, "Categorical Rhythm Perception. An Ecological Perspective" pp. 19–33 in Alf Gabrielsson (ed.), *Action and Perception in Rhythm and Music,* Stockholm 1987.

It is crucially important that the musicians take part in this communication by means of shared concepts. As long as their playing is a response to the notated configurations understood in terms of shared concepts, the musicians are likely to supply cues guiding the listeners' apprehension of the musical structure. Indeed, even if a performance is unusual or atypical in some respect, a clever listener, striving to understand the music, will still hear the music correctly and be able to recover the score. The musical context and the pertinent concepts will organize the situation, and the proper sequence of signs will be selected as the only possible or reasonable notation if the listener is assigned the task of taking down the music.

It does sometimes happen that even highly competent listeners having heard fully compliant performances commit "errors" when recovering the score. But more often than not such discrepancies are indicative not of mistakes, but of genuine and meaningful structural alternatives inherent in the musical structure. A listener might, for instance, have notated the voice leading in another way, which makes for a non-trivial difference between the transcription and the original score. But generally speaking, it seems that discrepancies of this kind are seldom sufficient to give rise to transcriptions that disagree with the original pitch/time structure to the point of specifying another work.

After all, what any good listener apprehends is a phenomenal musical structure, including relationships and ambiguities that the notation (including its "nonnotational" signs) is not suited to specify. And what else than a phenomenal structure was the composer trying to write down, to describe, by means of an imperfect prescriptive notation? It appears, then, that the understanding required when it comes to successful music dictation is a fairly shallow one, and that concordance in terms of pitch/time characters may obscure ambiguities with regard to the phenomenal relationships inherent in virtually any sequence of individual sound events when heard as a whole by a musical mind, relationships that make up the core of the music work as an experienced entity.

#### **Metric signs are non-notational and structural**

Goodman has nothing to say about the characters in musical notation that refer to metre (time signatures, bar-lines, beams), and yet they appear to be crucial for his notion of the identity of the music work.

The meaning of metric signs is in fact quite intricate. Most often time signatures, bar-lines, and beams merely confirm the hierarchy of accents and metric formats that is inherent in the structure when apprehended and interpreted by a competent person; the metric signs may be said to represent inferences drawn from the music's phenomenal appearance. But sometimes the signs indicate a metric organization that further specifies or even contradicts the inherent metre, and in such cases these characters are all but redundant – they carry a very strong normative authority.

Turning to performance, the picture turns even more complex. Sometimes the metric signs do not seem to demand any specific or perceptible "measures" from the musician; sometimes they invite or urge him/her to interfere in quite conspicuous ways. But it is not at all clear what to do; as long as the result sounds right, it is left to the musician to decide upon the nature and proper degree of the interferences. The notated metre may in fact be reflected in several ways that can be variously combined: music as performed is characterized by patterns of stress, onset timing and articulation. 10

Lacking specific and also exact reference, metric characters do not qualify as "notational" in Goodman's sense, and yet these signs, evoking quite definite concepts, may bring about crucial, differentiating structural properties – properties that can be indicated in this way only and that have to be present in the performances, properties that must be understood and taken down by means of metric signs when recovering a score.

When the inherent metre is ambiguous, for instance, it must (normally) be clarified according to the metric notation. And if in such cases the metric signs are neglected or violated (perhaps by just happening to play in a way that is "over the edge"), the difference can be substantial to the point of effecting a change of the musical identity. It may be quite difficult to recognize such a passage if it is disguised by being played according to a wrong metric scheme. Turning to Goodman's idea that it must be possible to recover the score, the original score – including its correct metric notation, however "non-notational" it is – would probably not be recoverable from such a performance.

Exs. 3 x/a/b illustrate the point. The pitch/time structure in Ex. 3x is inherently ambiguous, but it may be clarified by two different performances (*a* and *b*)

<sup>10</sup> Cf. John Sloboda, "The Communication of Musical Metre in Piano Performance", *Quarterly Journal of Experimental Psychology* 35A(1983), 337–396, and Bengt Edlund, "Representation of Metre in Performance. A Study of Bach Melodies", "Communicating Musical Metre. An Expanded Restudy", "The Tyranny of the Bar-lines. Encoding Notated Meter in Performance", pp. 84–88 in Friberg, A. et al. (eds.) *Proceedings of the Stockholm Music Acoustics Conference 1993*, Stockholm 1994 and "Making Meter Evident. On the Playing of an Ambiguous Bach Melody", *Musikpsychologie* 12(1995), 28–41, and *Performance and Perception of Notational Variants. A Study of Rhythmic Patterning in Music*, Diss. Uppsala 1985.

according to the notations in Exs. 3a and 3b, in which the normative metric signs do away with the ambiguity. Performance *a*, whatever its actual physical properties, complies with Ex. 3a, whereas performance *b* does not; it is in fact not acceptable as a performance of it. Ex. 3a, the fugue theme from J. S. Bach's Chromatic Fantasy and Fugue, cannot be recovered from performance *b*.

Hence, structurally and ontologically essential properties may reside in characters that are not "notational"; the identity between performances as well as the possibility of recovering scores from performances may be crucially dependent on signs that are imprecise, signs that are both genuinely structural and genuinely interpretative.

# **The beginning of Alban Berg's Piano Sonata**

Goodman construes the ontology of the music work in a way that combines misunderstandings of the nature and purpose of musical notation with musically counterintuitive lines of reasoning. But an even more serious failure, already hinted at, is the poverty of his concept of music and the concomitant slighting of the scope of musical interpretation. 11 In the previous discussion of the "notationality" of pitch and time characters, examples have been presented that demonstrate different musical meanings of "the same" sign, and it has also been shown that the possibility of recovering scores from performances turns out to be illusory. These observations throw doubts on the notion that compliant performances are musically identical, that they are identical in a way that allows us to lump them together to form a class that expediently defines what a music

<sup>11</sup> "It is, however, a fundamental tenet in our understanding of performing arts that a composition is *essentially* subject to interpretation", Stefano Predelli, "Against Musical Platonism", *The British Journal of Aesthetics* 35(1995), 346. Or, quoting again Kendall L. Walton ("The Presentation and Portrayal of Sound Patterns", p. 11): "A performance not only presents a pattern [specified by the score], but portrays it in a certain light. It interprets, parses, organizes, the pattern in some way or other, as well as indicating what it is. And how a pattern is portrayed and interpreted, as well as what it is, is musically significant." Walton's views are largely consonant with those of the present writer, although his notion (p. 15) of the music work as a hierarchy of Schenkerian levels, portraying each other all the way up to the musical surface (the score) and then to the performance, which interprets the score, seems superfluous and questionable. There are, after all, other ways of conceiving musical structures, and Schenkerian analysis is perhaps, after all, not that fundamental when it comes to the art of interpretation; cf. Chapters 8–11 in the present volume.

work is. How can (even compliant) performances be identical, if the scores are inherently ambiguous, and if the same performance can be heard differently?

This emerges as evident if we consider interpretation as the activity that musicians actually devote themselves to. The very first bars of Alban Berg's Piano Sonata Op. 1 (cf. Ex. 4) may serve to illustrate the scope of interpretational decisions. 12 It will appear that the musical consequences of such decisions range from minute details of articulation to crucial matters of form, with concomitant changes in associative content.

A reasonable attitude to notation entails (among other things) judicious treatment of passages in which signs seem to be absent. Does the long slur preclude separate articulation of the two falling "thirds" in the upper voice in m. 1? Such descending skips appear later on in the sonata, where they are marked by two-note articulation slurs. But separating these sighing intervals from each other might conceal the motivic kinship between m. 1 and the melodic motion that resumes the musical process in m. 3; cf. y and y'.

Turning to the middle voice in m. 2, it is possible to bring out a repeated poignant motif featuring a dotted falling semitone. This can be done either in concurrence with or at the expense of the vanishing upper-line melody, which actually brings a rhythmic augmentation of the very same motif; cf. z and z'.

It is essential that the proportional exactness of the durational notation does not impede the rhythmic diction, which may be crucial when it comes to conveying musical content. Therefore the dotted rhythm of the active, somewhat acute rising fanfare opening the sonata cannot be played in the same way as the dotted rhythms of the inner-voice sighing motifs in m. 2; cf. x and z1 .

If the pianist so wants, he/she can understand the left-hand dotted motif in m. 3, featuring a perfect fifth and fourth, as a calm and resigned, literally and metaphorically transfigured, answer to the initial dissonant and distorted righthand fanfare, and try to convey this affinity to the listeners, cf. x and x'. 13 Again, to convey this association, this sense of resolution, it is necessary to carefully adjust the articulation and the rhythmic diction.

The first three bars may either be smoothly attached to the rest of the sonata, or be rendered so as to stand out as a separate utterance. If this fragment is played

<sup>12</sup> Edition Universal, Wien.

<sup>13</sup> To let a sense of motivic resolution stand out at the very beginning of the sonata makes sense because this one-movement work ends quite demonstratively with the dissonant fanfare motif, repeated in ever-higher octave transpositions. The discord is then soothed by minor thirds played in ever-lower registers – the same third as appears in m. 3.

as an opening-then-closing gesture, and if the *ritardando* is pronounced enough to give rise to a gap in the temporal continuity, the first three bars will emerge as an introduction to the sonata, as a summarizing motto for the music to come. 14 Thus, depending on the choice of the pianist, one might say that we have two works, distinguished by an important formal difference, suggested by the "nonnotational", structural/interpretative caesura sign: there is one Berg Sonata Op. 1 with an introduction (motto), and one without. 15

# **A musically inclusive ontology of the music work**

From a musical point of view it is very disappointing to associate the identity of the music work with its "notational" pitch/time structure, and the problem is not just that elements like dynamics, tempo, and metre are left out of account although they are important and sometimes even structurally crucial. The most unfortunate thing is that the notation, which is devoid of *musical* relationships, is assigned a decisive role in the ontology of the music work. The disappointment derives from the fact that when establishing criteria of identity we are reduced to consider trivial concordances in terms of pitch and time between a score and its performances instead of taking account of essential musical phenomena.

Goodman's idea is understandable and effective if the main task of the ontology of the music work were to prevent Beethoven's Fifth Symphony from step by step deteriorating into *Three Blind Mice.* (p. 187) Then a number of exact signs to be strictly observed would be what you need, and the notated pitch/time structure serves this purpose (fairly) well. Truly exemplificative performances can be generated in order to constitute the work, which amounts to what these compliant performances necessarily must have in common, namely the pitch/ time structure making the performances identical.

Although sounds and performances are involved, the music work has ultimately been reduced to two elements in its graphic representation. And although required to complete the music work, the class of compliant performances does not contribute anything of musical interest; these exemplifications just provide the sound that after all is considered necessary when defining the music work. When it comes to the decisive core of Goodman's ontology of the music work, the optimal situation would be to engage one computer that can reproduce the pitch/

<sup>14</sup> For evident reasons, this way of playing (indicated by Berg's caesura sign) is unsuitable when the exposition of the sonata is repeated; the gap would be devoid of meaning, and later on the effect of the similar transition to the development would be ruined.

<sup>15</sup> This drastic conclusion is provisional and will be modified.

time "score" correctly and another one that can recover it from these deadpan realizations. If considered necessary, a third computer might be used to check whether these "performances" really exhibit the perfect pitch/time conformance stipulated to obtain between members of the class of exemplifications defining the work. In other words, interpretation is absent in Goodman's ontology.

But shouldn't the ontology of the music work amount to something less barren and reductive, something more musically rewarding? And shouldn't it, in order not to arrive at counterintuitive conclusions, acknowledge musical realities such as the fact that there is no music in a score until it has been understood/ interpreted in one way or the other, and the fact that actual performances tend not only to be different, but significantly different. 16

To conclude this essay, the outlines of another, musically inclusive and musically informed, ontology of the music work will be briefly sketched.

Composers produce scores, and the characters in a score – excepting signs that obviously and exclusively refer to matters of execution – contribute to define, to specify a certain composition. But this specification is necessarily imperfect and incomplete in various important respects: musical relationships and functions, for instance, generally fall outside the notation.

Scores are associated with the imperative to be heard. 17 A score gives rise to performances, and given the respect that musicians usually pay to what the composer has prescribed, these performances tend to resemble each other to a high degree. 18 But since interpretation is necessarily involved, this similarity, reduced as it will inevitably be by non-trivial *musical* differences, does not

<sup>16</sup> It seems that a musically inclusive notion of the music work leads to interesting further questions. For instance, when evaluating various performances, announced as being renditions of a certain music work, simply referring to full compliance with the pitch/ time inscriptions in the score means putting an all too swift end to the discussion. Whether, say, the sound medium chosen for the performance, the arrangement of the music, or some traits of the interpretations are disqualifying, should be determined with respect to whether or not these properties are compatible with constitutive *musical* properties of the work in question – issues that would productively tax both our intellect and our musical intuitions.

<sup>17</sup> Normally this means that the score must be played/sung, but a competent musician can read a score and imagine the music without any acoustic realization.

<sup>18</sup> This holds by and large, but there are exceptions; cf. Bengt Edlund, "Directions and compliance" and "Loyal disobedience. When is it OK not to play as written?", Chapters 3 and 4 in the present volume.

amount to identity. And nor is identity satisfactory or sufficient as a basis for a definition of what a music work is.

What do the signs in the scores refer to? It turns out that two levels of signification must be distinguished. As separate inscriptions they specify sound events or indicate what properties these events should have. The signification of structural signs is prescriptive and – leaving the bewildering variety of the actual correlates in real musical performances out of account – also exact as far as pitch and duration are concerned, but less so when it comes to other aspects of the sound events.

In addition, and in a less trivial, emphatically musical sense, the signs jointly refer to musical relationships. This reference is not prescriptive and it cannot be since scores are fundamentally indeterminate or ambiguous in this respect, and since this kind of meaning is not explicit but implied and suggestive as to its nature. 19 In order to understand the phenomenal musical structure inherent in a score, interpretation is required, just as it is for any worthwhile performance of music. 20 And this kind of interpretative understanding is also needed if a listener is to grasp the *musical* relationships that are embodied in the sound events of a performance.

It seems imperative that a music work must be musical in an emphatic sense and thus much richer in structure and content than it becomes when strict compliance with the "notational" aspects of the score is required as the necessary and sufficient condition for all performances constituting a certain work. When defining what a music work is, we cannot abstain from the relationships that stem from the music as an interpreted phenomenon, and that make up the core of our musical experience; we cannot rest content with just the fragmented signto-sound-correspondences that remain when the work is identified with what strictly compliant performances have in common.

Hence, the music work is not something that you establish by reduction, but something that you approach by means of ever more comprehensive syntheses based on your encounters with a certain composition in its various emanations.

<sup>19</sup> These matters are discussed in Bengt Edlund, "*Sonate, que te fais-je?* Towards a theory of interpretation".

<sup>20</sup> This agrees with the position taken by Thomas Carson Mark; cf. "Philosophy of Piano Playing. Reflections on the Concept of Performance", *Philosophy and Phenomenological Research* 41(1980/81), 299–324. He grants the performance of a piece of music status as a work of art in its own right, as a second work of art issuing from and illuminating the composer's work.

Berg's two significantly different piano sonatas should therefore not give rise to any ontological worries; they are compatible aspects of his Piano Sonata Op. 1.

If this non-reductive perspective is adopted, an altogether different kind of identity emerges; what defines a music work is not identity as a condition, but identity as a result. An important consequence of this shift as regards the key concept of 'identity' is that the identity of a music work includes essential qualities that performances of the work do *not* have in common.

A work of music is something that is slowly formed in the beholder's mind as the residue of the various ways in which a composition has been rendered and understood. An assimilative ontology of the music work explains how we in fact manage to discriminate good performances from bad ones. A good performance is one in which many – but hardly all – of the potential, aesthetically important musical properties have been realized; in a bad performance many of these crucial facets of aesthetic interest are absent or have been misrepresented. It goes without saying that it takes much experience to tell informed interpretations from shallow playing, and that our minds may be cluttered up by ideas deriving from inferior performances. 21

As so often, the misconception of music criticized in this essay is rooted in our inclination to let the notation stand for the music – Goodman ultimately associates the music work with its imperfect and incomplete written record. Let us instead regard the composer's imagination of the work and the musician's rendering (or the listeners' experience) of it as the two holders of a sandglass. The music must pass through the narrow funnel of notation, but it is wrong and degrading to define what a music work is on the basis of this state of transition, this precarious passage between birth and rebirth. 22

<sup>21</sup> As long as there is full compliance with the "notational" signs, bad or mediocre interpretations are as good as excellent ones in Goodman's ontology. But, given the element of transcendence that often permeates great interpretations, they entail a substantial risk that the works fail to be exemplified.

<sup>22</sup> A first version of this essay has been read by Professor Jerrold Levinson and by Professor Thomas Anderberg, and a later one by Professor Peter Kivy. They have given me valuable suggestions, for which I am very grateful. An earlier version of this essay appeared in *The British Journal of Aesthetics* 36(1996), 367–380.

# **Chapter 2** *Sonate, que te fais-je?* **Towards a theory of interpretation**

## **Introduction**

When discussing matters of interpretation with musicians, they sometimes express a quite dogmatic ambition to strictly observe the composer's intentions as inscribed in the score. This attitude is not altered by arguments to the effect that playwrights do not by far enjoy the same privilege to be obeyed beyond criticism, and that it can be gathered from the history of musical practice that unconditional submission has not always been compulsory. The current code of the profession bids musicians to pride themselves on respecting notation to the letter; indeed, it is sometimes considered a moral obligation to do so.

It seems that this ideology of fidelity rests on two foundations. Excepting ornamental details, it is difficult, sometimes impossible to violate or ignore the notation (or at least the signs that specify the musical structure) without impairing the music. Generally speaking, one might say that music in virtue of its closely-knit formal nature tends to defend itself against most attacks on its integrity – substitutions, additions, and omissions are seldom musically acceptable. Secondly, fidelity when it comes to musical reproduction is often grounded in the idea that there is an unwritten "contract" between composer and performer. The composer's prerogative to determine how the music is to be played follows from the notion that the score (or at least a great majority of the signs in it) can be likened to a set of testamentary stipulations. When a musician decides to perform a work, he/she takes over a legacy and commits himself/ herself to observe the stipulations prescribed by the testator. 1

It is obvious that these two reasons for fidelity mutually support each other, and also that the prohibitive nature of the musical structure is a stronger reason for fidelity than the metaphoric parallel between a composer's score and a person's last will. Dramatic texts usually have structures that do not preclude interferences to the extent that most music works do, and consequently we have a tradition of freedom in staging that has prevented the idea of a binding contract to take root.

Musicians adhering to the ideal of strict fidelity tend to be unwilling to exercise textual criticism – many scores do contain details, even vital ones,

<sup>1</sup> Many years ago philosopher Peter Kivy read a paper analysing this view.

that might be questioned – and they are also disinclined to rank notational prescriptions according to function and importance. 2 It furthermore seems that many musicians do not fully acknowledge their own role in the process of interpretation: sometimes they deny that their performances involve any interferences, and they are often convinced that "it's all in the score". There are of course musicians that penetrate deeper into the problems associated with the ideal of fidelity, but it is remarkable that so many musicians so enthusiastically adopt an ideology that minimizes their own contribution to the result, and that obscures what actually happens in interpretation. And yet some of the most loyal of them are great artists – there is often a salutary discrepancy between principles and practice. 3

But it would be desirable to have a theory of interpretation without contradictions, a theory that emancipates musical interpretation from an overly strict observance of the musical text as given in the score. We have, mostly for good reasons, a tradition of musical interpretation that in various respects is characterized by fidelity, and therefore it is essential to clarify the nature of reproductive fidelity and to reflect on the problems that this attitude entails. The obligations *vis-à-vis* the composer must be formulated in a fruitful way, and no musician should feel forced to sign a fictive "contract" with unreasonable stipulations.

Since the identity of a music work does not reside in the score and in the exact reproduction of its inscriptions, but emerges as the cumulative result of many encounters with a certain composition, as the balanced sum of different performances revealing various aspects of the potential richness of the work, the crucial role of interpretation in musical ontology is obvious. The musicians are vitally important agents when the idea of a music work is formed in our minds. 4

<sup>2</sup> Randall R. Dipert discusses these matters in his essay "The Composer's Intentions. An Examination of their Relevance for Performance", *The Musical Quarterly* 66(1980), 205–218. Although his argumentation runs differently, Dipert's general outlook seems to agree with the views of the present writer. Cf. also Bengt Edlund, "Loyal disobedience. When is it OK not to play as written?", Chapter 4 in this volume.

<sup>3</sup> Cf. Bengt Edlund, "Directions and compliance", Chapter 3 in the present volume.

<sup>4</sup> Cf. Bengt Edlund, "Scores and works of music. Interpretation and identity", Chapter 1 in this volume, also published in *The British Journal of Aesthetics* 36(1996), 367–380. In this text the relationship between interpretation and ontology is demonstrated on the initial bars of Alban Berg's Piano Sonata Op. 1, an exemple that makes up a complement to the remarks on the second-movement theme of Beethoven's Piano Sonata Op. 13 in the present essay.

The relationship between score and interpretation may be analysed in terms of a number of "fidelities" that the musician is supposed to observe. They are quite useful since they help to distinguish some crucial elements of musical interpretation.

# **Fidelity to the notes**

"Fidelity to the notes" involves strict observation of the individual signs in a score, and you must not just respect the "notes" but also each and every other inscription. This seems to be an obligation that is elementary and inescapable at least if the consequence of infidelity is put in rigorous terms: those who deviate from what is written by the composer are playing another piece.

We will not dwell on the question of how many wrong notes you can play in a sonata and still be considered to have performed it. Although some ontologicallyminded people may be prepared to say that just one single wrong note is enough to ruin the identity of a work – no matter which note, no matter whether the deviation was intentional or not – it is a fact that listeners and other members of the musical community are quite tolerant, and also that it makes some difference whether the wrong note is a mistake or played on purpose. 5

But the fidelity to the notes must be reasonable in a more important respect. A vital and, as will be argued here, legitimate distinction must be made between different kinds of signs appearing in the scores. It may be held that strict fidelity particularly or exclusively applies to signs specifying the "pitch/time structure" of the work, whereas signs concerning matters of interpretation are exempted from this duty. 6

The point of distinguishing between "structural" and "interpretative" signs is that the former are inscribed by the composer as a composer, while the latter are added to the score by the composer, acting as the first interpreter of the work, a domineering, meddlesome, anxious, or just helpful fellow musician as

<sup>5</sup> The irksome problem of wrong notes may be elegantly solved – or circumvented – by stipulating a distinction between the exemplification of a music work on the one hand, and the performance of it on the other. A performance can absorb quite a few wrong notes before it loses contact with the work, whereas the notion of exemplification (or instantiation) does not tolerate even the slightest deviation from the score, at least not as far as its pitch-time content is concerned; cf. Jerrold Levinson, "What a Musical Work Is", *The Journal of Philosophy* 77(1980), 1–28 (p. 27).

<sup>6</sup> These formulations are preliminary, and the distinction will be further developed and discussed.

the case may be. Structural signs are accorded normative validity – they are "authoritative" in the literal sense of the word. Interpretative signs, on the other hand, are not binding for the performer since they are, or can be regarded as, proposals as regards the execution of the music – very interesting proposals, certainly, proposals that you should seriously consider, and that you may eventually decide to leave out of account. 7

Musicians do not copy each other's interpretations, nor should they feel obliged to imitate the interpretation of the composer, having provided the score with signs indicating how the work is to be, or might be, played. Generally speaking, it would be unfortunate to regard the composers' interpretative ideas as strictly binding. The more uniform the interpretations of especially the masterpieces in the standard repertoire, the greater the risk of wearing this music down. 8 Furthermore, since the presence of interpretative signs tends to block the musician's creativity by barring alternative ideas out of consideration, pieces with few or no such inscriptions appear to be richer than pieces with many interpretative marks. But this is of course a delusion. Just as a "clean" score does not refuse various interferences, a "pre-interpreted" one does not preclude the non-indicated interpretative options.

The notion of strict and indiscriminate adherence to every sign in the notation gives a deceptive aura of definitiveness and even inevitability to composed music. When playing from our neat printed editions, we do not suspect that hesitation or arbitrariness may have been involved when writing down some of the individual signs, whether structural or interpretative. Until we discover deviating parallel passages or learn about divergent sources, we tend to take the text for granted in a wholesale manner.

One can easily think of some composers who would have protested fiercely against the distinction just proposed, either declaring that all signs in their scores are structural, or referring to some "contract" with the musician granting the composer the right to decide also upon matters of interpretation.

Whereas it is true when it comes to serial music that all (or virtually all) parameters are strictly included in the compositional design, it is most doubtful

<sup>7</sup> It should be pointed out that the interpretation suggested in the score is not the only possible one, and that it might perhaps be less than optimal.

<sup>8</sup> Leonard B. Meyer has pointed out the vitalizing effect of interpretational variability when pieces are listened to once and again; cf. "On Rehearing Music", *Journal of the American Musicological Society* 14(1961), 257–267, also published as ch. 3 in *Music, the Arts, and Ideas*, Chicago University Press 1967.

whether this is generally the case. Serial music appears to be the extreme point of a development towards doing away with the performer as an independent agent in the musical communication. Until (say) 1700 the composers usually added very little or nothing to the signs prescribing the pitch/time structure. Since then a growing number of signs regulating the execution of the structure can be found in the scores, but the intentions of the composers have varied: sometimes the signs are probably to be taken as prescriptions, sometimes they are rather to be understood as proposals.

As to the testamentary "contract", it can be argued that what this notion primarily, but inexactly captures is the normative nature of the structural signs. The object that we actually take over is the (pitch/time) structure, whose integrity we must respect if we want to inherit the work at all. But the "last-will" metaphor does not explain why musicians necessarily have to submit to the interpretative signs, which – irrespective of what the composers' intentions actually were when writing them down, and irrespective of what analytic reflection discloses as to their function and importance – are supposed to attach to the heritage as obligatory stipulations. 9

#### **More on structural and interpretative signs**

But which signs are structural, and which are interpretative? The current answer is based on the all too simple implied notion that music in essence "is" a pitch/ time structure. Hence, the signs for pitch and duration make up the category of non-negotiable structural inscriptions. But since these signs are not always structurally decisive, since musical structures may have other essential properties, and since the pitch/time symbols are in fact supplemented by other inscriptions, such as metric signs and dynamic markings, that (may) contribute significantly to specify the musical structure, it is obvious that the current, simplifying generalization can at most serve as a handy approximation. It will always be

<sup>9</sup> Perhaps we should do away altogether with the metaphorical "contract" between composer and musician? Far from giving any insight into the relationship between the parties involved, it amounts to a persuasive definition of it. If the notion of a contract is accepted, the musician has by implication made a concession to the effect that the composer has also the prerogative to prescribe matters of performance, which in turn means that the distinction between structural and interpretative signs turns more or less redundant from a practical point of view. Also from a practical point of view, it should be observed that the "parties" actually involved when a work is to be played are the musician and the score, not the composer (who is dead, usually).

incumbent upon the musicians to decide whether inscriptions are structural or interpretative. An informed and critical musical judgement is indispensable for any musician who accepts this distinction and wants to employ its consequences.

In many cases the status of a sign can be informally tested. If an inscription is interpretative, it is likely that it can be disregarded, or that other interpretative signs can be substituted for it, with an acceptable musical result. This test is based on the observation that there are often several possibilities to render a musical passage in a convincing way, and that genuinely interpretative signs are used to indicate one of these options.

But there are also passages in which no meaningful interpretative alternatives can be devised, and in such cases it makes no practical difference whether signs specifying aspects of performance are taken to be structural or interpretative. If the signs in question obviously refer to the execution of the music rather than specify its structure, they are of course interpretative, but they are also redundant since the qualities or actions asked for are inherent in the structure, which only allows of one way of playing. In many such passages the composers could just as well have abstained from regulating the performance. But structurally necessary, redundant interpretative markings are far from rare, and as a musician one tends to miss them if they are absent – there are, for instance, many redundant articulation slurs in Classical scores.

It would be a mistake to think that structural signs, just because they prescribe the (pitch/time) structure, always specify traits that are more important and conspicuous than those indicated by interpretative signs. Examples to the contrary are not difficult to find. Whether a passage is doubled in the octave or transposed by an octave, or if a rhythm, say a pair of note values, is even or dotted, may emerge as immaterial when compared with a *crescendo* or a sudden dynamic contrast. Taking the existence of non-essential structural signs and quasi-structural interpretative signs into account, it seems unwise to condemn all offences against the composers' inscriptions with the same severity.

Furthermore, just as (say) dynamic marks may be truly (or quasi-) structural, there are – particularly in genres associated with improvisation – signs specifying pitch and duration that in fact rather indicate matters of interpretation. In some cases it may, for instance, be more appropriate to regard ornamental figurations and rhythmic details as hints at suitable ways of execution than as normative prescriptions specifying the melodic and rhythmic structure.

It should finally be pointed out that the distinction between structural and interpretative signs is different from a distinction proposed by Nelson Goodman. Using syntactic and semantic criteria, only the sign systems indicating pitch and duration, being orderly and exact, qualify as "notational" according to Goodman. 10 When it comes to the distinction between structural and interpretative signs, on the other hand, the crucial issues are musical function and normative validity.

Musicians often hesitate to think of notation in terms of "structural" and "interpretative" signs – or rather, they hesitate to admit that they now and then actually resort to this distinction when dealing with musical scores. This attitude is quite understandable since thinking in such terms introduces an undesirable element of uncertainty into the study of the works and makes for situations requiring delicate decisions. But every freedom has a price, and taking full command of interpretation entails both responsibility and risks. Structural signs might be mistaken for marks pertaining to the execution of the music, which means that you run the risk of arriving at a deteriorated or corrupt version of the work. And whereas it is legitimate to treat interpretative signs with some discretion, it is embarrassing to have presented a performance that is inferior to the one indicated by the composer.

Whether a musician ventures to accept the distinction between structural and interpretative signs, or abstains from it, is his/her private affair. But in any case the "fidelity to the notes" must be reasonable. It is necessary to consider the precision of the signs, to realize that they are sometimes ambiguous and always dependent on the context, and to understand them according to relevant stylistic conventions.

# **Fidelity to the style**

"Fidelity to the style" is nowadays so widely accepted and (sometimes) pursued an ideal that we seldom give it a thought. But earlier, when preoccupation with historical performance practice was an exclusive, pioneer activity, strict adherence to stylistic norms was equivalent to a controversial attack on prevalent habits

<sup>10</sup> Cf. Nelson Goodman, *Langauges of Art* (Indianapolis 1968). Goodman aims at a clarification of the ontology of the music work (which is not the primary purpose here). But his theory is flawed by the fact that he does not take account of the insight that "non-notational" signs may specify structure, and that the "notational" signs for pitches and (relative) durations do not always do so. There is more to be said about Goodman's ideas, and about the relationship between ontology and interpretation; cf. Bengt Edlund, "Scores and works of music. Interpretation and identity". In addition there is, it seems, an intermediate category between structural and interpretative signs; cf. below.

of performance. 11 The difficulty (and challenge) associated with the ambition to play in a stylistically correct way partly derives from the fact that musical notation is incomplete in many respects. The composers did not (and do not) bother to inscribe self-evident things that they assumed that all players knew. 12 But performance traditions change, and habits of execution tend to be forgotten; hence the later-day efforts to carefully document old manners of making music, studies that (among other things) have produced textbooks urging musicians to play "authentically".

In order to understand properly what fidelity to the style amounts to, and to give it its proper place within interpretation, it is necessary to keep in mind that performance style involves conventions as to how certain signs in the notations should be read and how certain typical configurations should be executed. Stylistic insight is generalized knowledge derived from the study of theoretic treatises and specific repertories, and it is valid for works belonging to a certain period, national tradition, or genre. Insights relating to individual works, on the other hand, are not matters of style, and they play an altogether different, contrary, role in interpretation.

Unfortunately, these two kinds of insights have sometimes been mixed up to the detriment of the latter. Stylistic requirements, which seem to be possible to establish and satisfy, have been given priority at the expense of the demands of the individual works, implicit demands that appear to derive from more or less subjective interpretative considerations. This state of affairs is understandable but not desirable since it is evident that at the time when fidelity to a certain historical (i.e. then contemporary) style was self-evident, the musicians were free to give precedence to interpretation.

<sup>11</sup> It appears that we can now afford to admit that the contemptuous dismissal of "unhistorical" music making as merely unthinking routine was unjust. Lack of historical knowledge or interest does not necessarily imply that "uninformed" performances are aesthetically deficient and without insight. These musicians, issuing from *their* tradition and using *their* intuitions and wits, rely upon what *they* have found in the individual works.

<sup>12</sup> The growing number of signs referring to performance in scores from (say) 1700 and onwards may be taken to indicate that the composers were not just anxious to obviate interpretative misunderstandings, but also that they suspected that some lay musicians were perhaps more or less ignorant of the proper style. Thus, some of these signs are not to be understood as interpretative in the core sense of the word; they rather enjoin the players to observe certain stylistic conventions.

It may furthermore be argued that the core of a truly artistic interpretation lies in bidding defiance to conventions and rendering habits suspicious. Keen listeners do not just want another exemplification of general rules; they prefer a revelation of the unique.

Not only have the styles of composition and performance changed through the ages, so has also (although it is more difficult to establish) the style of listening. 13 Therefore it is not surprising that "authentic" performances may run contrary to the taste of some, perhaps backward, present-day listeners, that people, who no longer find the cherished values in the music they love, get very disappointed. 14

Frankly speaking, fidelity to the style tends to have an intolerant side: there are musicians who can be proved to play the music incorrectly, just as there are listeners who are appreciating it in the wrong way. This attitude seems to be rooted in the notion that there is one and only one way to play, and to apprehend, a certain kind of music. But whereas it is true that there is a relationship between the manner of performance and the possibility to hear certain things in the music, the history of musical performance and reception shows that the same compositions have been played and listened to in quite different ways and yet with great profit. Apparently, the diverse performance styles have brought into relief different qualities inherent in the music. 15 While it may be established that there was once a certain "authentic" style of playing a certain kind of music, it seems that there may be several adequate, i.e. rewarding, ways to make this music speak.

The conviction that "historically informed" performances necessarily produce the greatest aesthetic profit is not self-evident, and therefore it is hard to maintain that strict adherence to style is a necessary obligation for musicians. Historically founded knowledge of performance styles is highly commendable since it may prevent mistakes, and compelling interpretations can certainly arise within the confines of the appropriate style, but "fidelity to the style" is not compulsory; it remains an option.

<sup>13</sup> For an investigation into past listening habits, cf. for instance Melanie Howe, *Pleasure and Meaning in the Classical Symphony*, Indiana University Press 2007

<sup>14</sup> It is against this background that the following tirade, conservative but not conservative enough, and too magisterial to be translated, must be understood: *"Mechanisch zirpende Continuo-Instrumente, bettelhafte Schulchören dienen nicht der heiligen Nüchternheit, sondern der hämischen Versagung, und daß etwa schrille und hüstende Barockorgeln die langen Wellen der lapidaren grossen Fugen aufzufangen vermöchten, ist purer Aberglauben.*" (Theodor W. Adorno, *"Bach gegen seine Liebhaber verteidigt", Gesammelte Werke*, *Band* 10(1), p. 150)

<sup>15</sup> One way of playing reveals "long waves" in "lapidaric fugues", other ways do not.

There is more to say about stylistic observance, but we must proceed to the final duty on the path towards the consummate interpretation: the fidelity to the work. But before dealing with this obligation, it will be divided into two separate but intimately related components that are at the core of musical interpretation: fidelity to the text, and fidelity to the content. 16

### **Fidelity to the text**

"Fidelity to the text" will not be used as synonymous with "fidelity to the notes", as often happens in current musical parlance. The latter term has already been used in this essay to refer to the obligation to respect the individual signs of a score; it is now stipulated that "fidelity to the text" involves combinations of signs, from local configurations all the way up to the entire notated structure. It is necessary to make this distinction since combinations of signs raise issues of an altogether different scope and character. Whereas individual signs must be read with careful attention to their function and context, combinations of signs – in addition to taking account of their function within their larger contexts – must be understood as musical entities. Fidelity to the text concerns the musical structure, the "text" that the individual signs jointly specify.

In a simplified and when it comes to a theory of interpretation altogether inadequate sense, it may be true that the signs for pitch and duration taken together unambiguously specify a certain pitch/time structure, but in order to grasp the musical structure it is indispensable to take thorough account of all interrelationships involved, and to realize that a structure does not become a *musical* structure until it is provided with phenomenal qualities. The structure that a musician has to assimilate in order to understand the music is replete with things like formal configurations, gestalt qualities, functional complexes, inner dynamics, and local or global characteristics of various sorts. This phenomenal structure, this "text", is implicit, something that must be read out of the notation, which only specifies the (pitch/time) structure, the substratum of the "text". 17

Musical experience shows that, even if the "notes" and the style are strictly observed, several *musical* structures, being different in important respects,

<sup>16</sup> This distinction is made in order to expose some important issues within the theory of interpretation, and it does not amount to an affirmative answer to the worn-out question whether music has both form and content.

<sup>17</sup> To avoid misunderstandings it should be added that also the pitch/time structure is "phenomenal" in the sense that it can be perceived if it is, say, executed by a computer

may be inherent in a given notation. Thus, one and the same score may contain different "texts", and notations are open in a crucial sense: they do not specify how the prescribed events are to be apprehended as a *musical* structure, they do not specify the phenomenal qualities of the music. In order to find its "text", a score must be understood, "interpreted".

But interpretation should not be taken to mean that the musician offers aspects of the music by expressing his/her views on it. The interpretation presents what the musician has read out of the score, what he/she has found that the notation signifies: this is how the music is. To persistently search for the phenomenal, *musical* structure inherent in the notation is the attitude that the fidelity to the text enjoins the musician to take up. It must be stressed that it is necessary to find a text (not *the* text) beyond the notation. One cannot, perhaps attempting to avoid decisions that may seem arbitrary, refrain from interpretation (i.e. understanding) and stick to some kind of "objectivity". Such uncommitted performances will soon expose themselves as crude. It is impossible to play convincingly something that you have not understood, but there is a scope for artistic decisions, and you must not necessarily understand the music in the same way as the composer (presumably) did. 18

The *musical* structure is a multi-faceted, transient, and delicate thing that largely eludes prescription as well as description, and the notation is very incomplete with respect to the "text". There are just a few signs that to some extent clarify the intended *musical* structure: slurs can be used to indicate events that belong together or are to be kept apart, constituents of form or structural contrasts can be suggested by means of dynamic marks, etc. And most often these signs are quite vague: the precise phenomenal character is not specified, nor is it exactly indicated how the effect is to be achieved. The musician is incited to play in a way that lets a certain phenomenal structure come to the fore. 19

and heard by a guinea pig. But the notion 'phenomenal structure' includes the specific organization and the additional qualities that the notated pitch/time structure is bound to get when it is played, heard, or read – i.e. understood – as *music*.

<sup>18</sup> The stance expressed here agrees with, and is encouraged by, the arguments advanced by Thomas Carson Mark in his musically sensitive essay "Philosophy of Piano Playing. Reflections on the Concept of Performance", *Philosophy and Phenomenological Research* 41(1980/81), 299–24.

<sup>19</sup> It might perhaps have been possible to develop signs indicating *musical* structure and character in more detailed ways, but apparently the composers were either less interested in prescribing or suggesting such matters – leaving it to the musicians to

Such signs occupy a crucial, intermediate or borderline category between structural and interpretative signs: they are interpretative signs that indicate *musical* structure. They refer to and influence the performance of the music, and in many cases they can be taken as inscribed by the composer assuming the role of the interpreter. On the other hand, they may substantially modify the appearance of the (pitch/time) structure, which belongs to the composer's domain. Some of these signs, bringing only slight structural consequences and seeming more or less arbitrary, may be understood as interpretative marks, while others, crucially important and deeply rooted in the structure, should be regarded as structural indications – slurs marking sighing motifs, for instance. It is evident that musicians wanting to distinguish between structural and interpretative signs must pay inscriptions belonging to this intermediate category particular attention, both in order to gain freedom and to avoid mistakes.

Ultimately the musician has to make up his/her mind and take a stand on a fundamental issue. How much, beyond pitches and their durations, and beyond metric and certain dynamic marks, does reasonably belong to the "basic" structure prescribed by the composer, and where does the interpreted *musical* structure, the "text", begin? The decision to submit or not submit to interpretative signs that are more or less associated with the structure depends on the extent to which the musician considers it to be the composer's prerogative to determine the *musical* structure. 20

#### **Fidelity to the content**

"Fidelity to the content" refers to the obligation to discover and convey the content inherent in the music, "content" being taken to mean such things as emotional mood, associative potential, and dramatic development. These aspects must agree with the "text" that you consider to be valid, and generally the content has to be compatible with the "spirit" of the work as judged by both internal and external evidence; cf. next section.

Scores seldom bring inscriptions that obviously refer to content. Composers are not often as helpful as Scriabin when he added the exhortation *"avec une joie débordante"* over a passage that either embodies the quality of exuberant joyfulness, or perhaps needs just a little push in the right direction to suggest it. Examples of this kind reveal that the content that a loyal musician looks for is

grasp the music on their own – or they were discouraged to do so due to the difficulties involved.

<sup>20</sup> We will return to this question later on.

normally to be found close to the *musical* structure, the "text" that he/she strives to establish. Some people would even say that "text" and "content" are one and the same thing – or perhaps two aspects of the same matter – while others would rather hold that this kind of content is not "extra-musical". Be that as it may, here it will only be claimed that it might be a useful working hypothesis in musical interpretation to assume that text and content are sometimes distinguishable.

When musicians claim that "it's all in the score", they apparently mean two things. Firstly they announce that they belong to those who strictly adhere to the "notes", secondly they express their conviction that everything you need for interpretation is revealed if you just study the notation thoroughly. But this view seems to entail two questionable implications. It suggests that the notation allows of only *one* phenomenal organization, one "text", or that there are decisive reasons for preferring one such organization – the correct one – to several others. It also suggests that fidelity to the "text" turns attention to the content superfluous: either there is no content beyond the painstaking articulation of the *musical* structure, or there is for each "text" *one* and only one content.

But just as there is no one-to-one correspondence between notated configuration and *musical* structure, there is no one-to-one correspondence between "text" and "content". The truth seems to be that the notation in essential respects leaves both "text" and "content" open, and that it is the task of the musician to understand both aspects as precisely as possible. Therefore it is necessary to transcend the text in order to find the content that must inform the phenomenal structure.

But even this might be understood in too simple a way. It must be observed that the relationship between "text" and "content" is dialectic. Both aspects are open and ambiguous, and the content may be of help when searching for the *musical* structure as well as the other way around. ("Isn't this after all a funeral march?") A musician who is aware of both possibilities and has a flexible method of interpretation, is likely to penetrate deeper into the music, and is less liable to make mistakes or end up in superficiality. 21

#### **Fidelity to the work**

"Fidelity to the work" as understood in general musical parlance is not simply synonymous with the kind of careful and searching integration of "text" and

<sup>21</sup> Donald M. Callen has convincingly argued for the idea that consideration of content should be seen as an essential and integrated part of a genuine and mature musical interpretation; cf. "Making Music Live", *Theoria* 48(1982), 139–168.

"content" described above. It refers to the belief that beyond the score there is an "authentic" compound of *musical* organization and associative content, a message from the composer that the musician is obliged to find and express.

But fidelity to the work in this sense is an awkward ideal to live with since it presupposes that the musicians are able to select the right *musical* structure and the right content out of the ambiguities offered by the score. By and large, this does not seem to be possible – scores do not provide decisive reasons to settle for a certain reading rather than other possible ones when we are confronted with the interpretative challenges posed by the works. And we can never know for certain whether we have succeeded in being loyal to the work, whether we have actually interpreted the music according to the composer's intentions.

It seems, then, that musicians, irrespective of what they say and even if they do not draw the full conclusions from it, are actually adherents of what we may call an "emancipated" fidelity to the work, rather than being devoted to the pursuit of the one and only message. All conscientious musicians search until they have found an interpretation, whose *musical* structure and content make up a coherent whole that is compatible with the notation according to *their* musical judgement, and that also agrees with what *they* know of the work and its composer. Then some of them claim that this is also the "authentic" message of the work, and that they have been loyal to the composer's intentions.

But there is a further argument for why it is not advisable to identify the meaning of a musical work with what we suppose, guess, or even know to be the composer's intentions. In general, there is no reason to maintain that the composers necessarily had the deepest and most interesting insights into their music. There may very well be more and other things to a work than what the composer saw in it, and this holds both for the "text" and the "content". If one accepts the idea that the notation allows of several different readings, giving rise to performances revealing substantially different *musical* structures and contents, why should the composers have the prerogative to prescribe what the musicians – and the listeners – are to make out of their music? It would simply be disadvantageous to define fidelity to the work in a way that blocks the ambiguity and richness of the music. It is neither reasonable, nor wise to let the composers control the musician's and the listeners' understanding of the *musical* structure and the associations it evokes; this cannot but reduce the aesthetic potential of the work.

When advancing a distinction between structural and interpretative signs, we observed that the composers have tended to take over the task of the musicians.

The notion of fidelity to the work, if understood as a quest for the "authentic" meaning, prepares the ground for a further and more momentous restriction of the musicians' importance and independence, a restriction that also affects the listeners. It seems that there are good reasons for denying the composer the dual role of normative interpreter and ideal listener. Whereas the composer has the prerogative to prescribe the "basic" structure (i.e. the essential pitch/time specifications together with some structural or quasi-structural interpretative directions and important qualifications of other kinds), the interpreter should preferably be responsible for the *musical* structure and its content, for the "fidelity to the work".

#### **The contribution of the interpreter**

Is it enough to observe these fidelities in order to interpret a work of music? The general view – even among those who consider the musician to be merely the composer's tool – seems to be that something more is needed, at least in great interpretations.

Some people would say that the musician must also be loyal to himself/ herself, while others would put it that it is legitimate, desirable, or inevitable that the interpreter leaves his/her stamp on the music. "The contribution of the interpreter" seems to be a suitable heading for a section presenting various aspects of artistic performance falling outside the scope of fidelity.

As has been made evident, already the pursuit of the fidelities, reasonably understood, allows of a good deal of freedom and brings in decisions of many kinds. The fidelity to the notes entails distinguishing between signs of various importance, function, and authority. The demands of the individual work may from case to case dispense from the generalizations of style. And the two components of the fidelity to the work presuppose that the musician acquires insights, discovers ambiguities, and makes decisions.

But in addition we tend to require that a truly great interpretation, beyond a rich and as it may seem incontestable revelation of a work's *musical* structure and content, should also bear the unique stamp of the musician. This quality may come about by spontaneous elements in the performances as well as by persistent musical reflection, but no matter whether momentary inspiration or profound thinking is involved, the crucial thing seems to be that the work is enriched by something unexpected or even to some extent distorting. The discriminating listener demands from great performances that the current and foreseeable "truths" of the musical works are not invariably or exclusively displayed.

Another kind of contribution bears no immediate relationship to the specific music work, but emanates directly from the musician's personality or behaviour. We say that some musicians display a strong "temperament" or that they show a deep "feeling", while others appear to be exceptionally "concentrated" or have a very strong "presence".

Finally we must mention the element of technical mastery, the degree of naturalness and perfection of the performances. Sometimes the command is such that extreme difficulties turn into supreme idiomatic, a transcendence that instils a particular sense of delight and confidence into the listener. 22

# **Remarks on the interpretation of a Beethoven theme**

There may be readers to whom this critical account of the various fidelities appears subversive or licentious, but the intention has been to present some aspects of interpretation and to argue for a better balance in our views on the joint endeavour of the composer and the musician. To further clarify this more reasonable and flexible approach to notation, an example will be provided that illustrates the nature of the decisions that musicians make, and are entitled to make, when they interpret a score, when they do what they must do with a sonata.

The eight-bar period beginning the slow movement of Beethoven's Sonata Op. 13 is not "periodic"; cf. Ex. 1. There is a melody and a harmonic progression running from tonic to tonic, but no symmetry between an opening four-bar antecedent and a closing four-bar consequent, and no unifying motivic work; the phrases as slurred by Beethoven are of unequal lengths. 23 And yet the music

<sup>22</sup> Thomas Carson Mark, arguing that musical performances should be regarded as separate works of art, also claims that virtuosity in performance amounts to a positive aesthetic quality, a quality that is fully accessible only to the initiated; cf. "Philosophy of Piano Playing", and "On Works of Virtuosity", *The Journal of Philosophy* 77(1980), 28–45.

<sup>23</sup> The slur is an ambiguous sign. Slurs may simply be used to show what belongs together, but they also suggest general *legato* or prescribe *legato* articulation of short groups of notes – in the latter case they may also indicate short demarcating moments of silence and initial dynamic emphases. This ambiguity notwithstanding, slurs often seem to divide musical passages in summary ways that apparently exclude alternatives and intermediary options. In Classical music the slurs were often inscribed according to metric conventions in ways that suppress – or that (perhaps incorrectly) may be taken to suppress – upbeat relationships. Slurs may decisively influence the rhythmic grouping, but it should be noticed that musical structures may form groups irrespective of articulation signs.

is in perfect repose, and the continuity is well provided for: the second beats of the bars bring various dominantic seventh-cords – except m. 4 having instead an e♮<sup>1</sup> striving upwards –and mm. 5–7 are linked together by a falling fifth/rising fourth sequence, f1 –b♭–e♭<sup>1</sup> –a♮–d♭<sup>1</sup> –g.

There are two peculiarities in the melody. The sense of a local melodic upbeat at the end of m. 3 has no suitable downbeat to attach to: the rising gesture lacks an organic conclusion, and m. 4 starts what will eventually turn out to be the consequent phrase. One may imagine an added fourth bar, bringing (say) a b♭<sup>1</sup> – e♭<sup>1</sup> conclusion of the antecedent, and postponing the actual m. 4 to its proper barfive position within a regularized period. (Ex. 2a) The swift filled-in rising fourth b♭–c1 –d♭<sup>1</sup> –e♭<sup>1</sup> in mm. 5/6 might be understood as a rhythmically contracted and melodically extended quasi-imitation of the preceding slow ascent e♭<sup>1</sup> –e♮<sup>1</sup> –f 1 . As a result of this contraction the pace of the metric strong/weak alternation seems to have been hastened; the downbeat of m. 6 has a sense of arriving too early. This imbalance can be ironed out by expanding the consequent to six bars. (Ex. 2b)

The first bar allows of two options as regards the rhythmic grouping, and the pianist's choice will slightly but perceptibly influence the character of the initial phrase. Should the weak-beat b♭ – and the d♭ in the bass, suggesting a free inversion of the melody – be grouped together with the first beat so as to form the afterbeat of a trochee, or should it be connected as an upbeat to the following strong beat making for a iambic group?

Similar, but more consequential overlapping options can be found at the next rhythmic level. It may seem most natural to conceive of and play m. 1 as metrically strong, but m. 2 may also (its initial instability notwithstanding) be thought of as the strong unit within the first pair of bars. 24 If m. 1 is strong, m. 3 will appear to be strong as well. Since it soon betrays a striving for completion, m. 3 will in retrospect turn out to be weak, but it cannot join m. 4 to form a iambic pair due to the melodic hiatus and the starting quality of the latter bar.

The slurring in mm. 1–3 also requires interpretative decisions. Several different options for the transition between the phrases can be distinguished. The eighth-note d♭<sup>1</sup> in m. 2 may, as Beethoven's notation seems to demand, be

<sup>24</sup> The nature of higher-level metric accents is an elusive aspect of musical structure, and it cannot be accounted for here. It is sufficient to acknowledge that we have a propensity to group bars in strong/weak or weak/strong configurations. From a phenomenal point of view this means that the second, weak bar of a strong/weak pair seems to grow out of the preceding strong bar as a kind of opening, whereas in a weak/strong pair the initial weak bar appears to stretch towards the following strong bar.

rendered so as to exclusively emerge as an afterbeat. This makes for a passive, open ending of the first phrase, and means that the following c1 should be played in a manner indicating a fresh start. But it is a delicate matter to suggest this, and only this, phrasing, and perhaps it is not even desirable to render the transition between mm. 2 and 3 as if it unequivocally involves a demarcation. Due to the temporal distance back to the e♭<sup>1</sup> and the proximity to the c1 , and due to the falling dominant-seventh implication of d♭<sup>1</sup> , this eighth-note is likely to be heard as having at least a double function, being concurrently an afterbeat of the first phrase and an upbeat to the second; a fact that cannot but make for a sense of elision.

Disregarding the actual slurs – or rather regarding the first slur as an articulation sign but not as indicating a phrase unit – it seems quite natural to close the first phrase on the c1 in m. 3. This opens up for another elision since the c1 suggests that it has a dual function as both a conclusion and a start.

The shift from m. 3 to m. 4 may be interpreted in three phenomenally different ways. The (c1 –)e♭<sup>1</sup> –a♭<sup>1</sup> –b♭<sup>1</sup> rising gesture, demanding a melodic arrival, can simply be abandoned since it is interrupted by what seems to be a new initiative that abruptly introduces another, quite unrelated idea starting from e♭<sup>1</sup> . But it is also possible to let the new idea emerge less suddenly, but inevitably, from the rising alto-voice motion a♭–b♭–c1 –d♮<sup>1</sup> – in m. 3, a connection that may be gradually brought to the fore concurrently with a withdrawal of the soprano. In addition to these readings, suggestive of two quite different shifts in a virtual dialogue, the melody in m. 4 may (notwithstanding the shift of slurs eclipsing the falling fifth) be formed as the actual, but quite unexpected and very odd, continuation of the rising motion in m. 3. If this monologue reading is chosen, the beginning of the consequent phrase (if any) will either appear to be postponed until the b♭ in m. 5 or seem to start already from the c1 in m. 3: two quite interesting and very different options.

Turning to mm. 4–5 four alternative interpretations can be distinguished; three of them stem from the quasi-imitative relationship between the motivic particles. If the antecedent phrase is extended up to f1 in m. 5, the following free imitation serves to link the close of the antecedent to the beginning of the consequent by means of a quick resuming gesture. On the other hand, if you start the consequent phrase in m. 4 (or m. 3), the most straightforward way to understand the motivic relationship is to conceive of the passage as made up of an initial groping for f1 followed by a swift but densely accented affirmative motion towards e♭<sup>1</sup> , from where the regular fifth/fourth sequence starts. But the first gesture may of course also be played as the more prominent motion, reducing the transformed imitation to a vague reminiscence. Finally, the element

#### Conclusions 49

of imitation can be altogether disregarded, a reading that implies that the two rapid notes in m. 5 as well as the sixteenth-notes in m. 7 are rendered so as to give the impression of subordinate filling-in motions within a fifth/fourth sequence issuing from f1 .

## **Conclusions**

It will have appeared from these remarks on the slow-movement theme of Beethoven's Op. 13 that reading musical notation, i.e. interpreting the music, involves deciphering the phenomenal structure, the *musical* "text", and finding the content associated with it. This is what being faithful to the work amounts to, and it implies a respectful *and* critical attitude to the "notes", but also an open mind for productive ambiguities.

Taking "performance" in an emphatic, qualifying sense, one is inclined to agree with Thomas Carson Mark's statement that "one cannot perform a work without attributing to it some meaning". According to his analysis, performing a work of music is like asserting something by means of a quotation. And just as you cannot assert anything with a sequence of words that you do not understand, it is impossible to play a passage of music convincingly without understanding it. 25

But what is it that must be understood? What kind of meaning is essential in music? It will also have emerged from the discussion of crucial points in the Beethoven theme that we have sometimes characterized the available choices in terms of modes of musical continuation that can be associated with human behaviour. 26 In this sense (and no doubt in others as well) music brings meanings of deep significance, but there is nothing in the notation that reveals or clarifies these inherent messages. To do justice to music we need interpretation, a kind of interpretation that identifies and finds ways to express the various modes of continuation that are exemplified in the musical process.

This is, then, what should be done with a sonata. The primary obligation seems not to be fidelity, but the endeavour to convey the animated form that makes up the essence of music. 27

<sup>25</sup> "Philosophy of Piano Playing", p. 317.

<sup>26</sup> More extended analyses exploiting the idea of "modes of continuation" are to be found in two other texts in the present volume, "Prelude to the art of continuation", and "Interpretation as continuation".

<sup>27</sup> An earlier version of this essay appeared in *The Journal of Aesthetic Education*, 31(1997), 23–40.

# **Chapter 3 Directions and compliance**

# **Introduction**

Whereas analysis of compositions has been practiced and developed for centuries, no such persistent efforts have been made when it comes to describing interpretations. Compositions are laid down in scores, a fact that seemingly implies that there is only one thing to be studied whereas musical interpretation, the activity that diversifies the composed objects, turns out to be an evasive matter. This is regrettable since a music work is bound to be apprehended as a confluence of composed and interpreted structure. 1

"Interpretation" is ambiguous in current musical parlance. Sometimes the word is used to refer to the ideas that a work has evoked in us, but it may also signify the final, audible result of a musician's endeavours as well as the analysis or the intuitive understanding that precedes any performance of artistic value. The latter two aspects of interpretation are of course intimately related, and if one wants to describe interpretation-as-result in a meaningful way, one cannot escape saying something about interpretation-as-analysis.

Musical performances are awkward objects of study since they are evanescent temporal sequences of sound, and the fundamental difficulty remains no matter the invention of sound recording more than a century ago, and no matter many decades of sophisticated registration techniques. It is of course an advantage to be able to listen to a certain passage as many times as you want, and so it is to see various aspects of the musical sound being displayed on a screen and/or printed out. But the former opportunity has not been used very often to study the details of performances, and as to the visual representations they still require attentive, musically involved listening in order to be informative. While it cannot be denied that valuable insights have been gained by means of modern techniques and statistical methods – important principles of intonation, metric encoding, rhythmic characterization, polyphonic co-ordination, and expression of formal properties have been discovered – it must be observed that this knowledge is of a general nature: we have begun to understand what musicians usually and normally do.

<sup>1</sup> Cf. Bengt Edlund, "On scores and works of music", Chapter 1 in this volume; also published in *The British Journal of Aesthetics* 36(1996), 367–380.

Thus, there is a gap to be filled, a need for insights concerning musical performance when it comes to specific situations in particular compositions, and this is a kind of knowledge that cannot be obtained without careful listening to individual performances. But if such investigations are to be productive, the ear must be guided by, and the findings related to, thorough analyses of the musical texts, analyses aiming at traits that, whether due to their structural implications or their (quasi) extra-musical content, may be relevant for interpretation. 2 Much of what there is to be heard in interpretation-as-result can be anticipated by focussing on interpretation-as-analysis – a fact that, far from detracting any value from the performances, testifies to their musical solidity.

What are the sources at the musician's disposal when devising an interpretation?

First and foremost there is of course the score. But "the score" in the present context cannot refer to anything else than the musical structure as understood by the musician reading the text. There is a wide range of analytic observations that may influence our interpretational decisions: tonal progressions, voice-leading tendencies, melodic implications, harmonic tensions, motivic correspondences, patterns of rhythmic emphasis and grouping, formal demarcations. Any score is replete with such information, and it may be argued that interpretation is tantamount to choice. When preparing an interpretation or when studying interpretation-as-result, it may be productive to look especially for structural traits suggesting different options of musical continuation. 3 It of course happens that structural cues of pertinence for musical continuation go unnoticed or are deliberately disregarded – there may be too many of them, they may be contradictory, and they are not compulsory. But such cues indicate hot spots where differences between performances are to be expected, and they can also give clues as to what interpretational interferences that are likely to appear.

It may appear from what has just been said that analysis, which is indispensable when working out penetrating interpretations, and which should precede descriptions of interpretation-as-result, is thought of as dealing only with the structural aspects of the music. But it must be pointed out that structural cues indicating options of interpretation are often charged with expressive content

<sup>2</sup> This is not to say that the process of discovery cannot proceed the other way around: inspired and original interpretations may reveal structural properties and relationships that might otherwise have escaped analytical attention.

<sup>3</sup> For two studies based on the relationship between options of continuation and interpretation, cf. Bengt Edlund, "Prelude to the art of continuation", and "Interpretation as continuation", Chapters 12 and 13 in the present volume.

#### Introduction 53

when apprehended by a musically perceptive mind. The fact that music analysts tend to maintain a neutral silence in matters considered to be impossible to speak about, does neither preclude that interpretations may be inspired by inherent content, nor rule out that inherent content may disclose otherwise neglected aspects of musical structure.

The immediate source, disclosing the composers' intentions when it comes to performing their music, is of course the interpretative marks in the scores, the signs that incompletely and in frozen form, as it were, specify the envisaged, or indeed the ideal performance – or just one out of several possible and acceptable renderings. Although the question of whether these signs are imperative cannot be settled once and for all, but should be answered from case to case, musicians tend to declare, and to tell their students, that the composers' directions should always be complied with.

Recordings made by composers are sometimes used as additional sources, offering "authentic" solutions to interpretational problems. Or abused, rather, since resorting to ready-made musical clothing comes close to preferring submission to creativity. And although it may seem more controversial, this applies also to the interpretational marks in the scores. As a matter of principle, there is no difference between certain traits in a composer's performance and his "authoritative" interpretative signs in the score, and yet, while many musicians are prepared to ignore the former, they most often feel obliged to conform to the latter. It seems that, unlike a composer writing down a score, a composer playing the piano or conducting an orchestra can be regarded as just a fellow performer. 4

Musicians sometimes talk about their musical intentions, and interpretational matters are of course dealt with in advanced teaching. But these ideas are seldom written down, and when they are, they are often put in disappointingly vague terms. The musical tradition, on the other hand, offers a wealth of examples passing on (more or less) positive convictions of uncertain origin as to how certain passages in the standard repertory should be played.

For more than half a century "performance practice" has frequently been consulted in order to shed light on matters of interpretation, but it appears that this involves a mistake. Performance practice, although ultimately deriving from encounters with individual compositions, makes up a body of generalized knowledge that cannot really tell you very much about interpretation, about how

<sup>4</sup> For a discussion of these matters, cf. Bengt Edlund, *"Sonate, que te fais-je?* Towards a theory of interpretation", Chapter 2 in this volume; also published in *The Journal of Aesthetic Education* 31(1997), 23–40.

to play specific music works. Whereas a "historically informed" sense of style may or should serve as a frame for your artistic endeavours, it cannot replace the musically vital decisions that are occasioned by the demands of unique passages in unique works, decisions that may bring about exceptions to the rules.

The purpose of the present investigation is twofold. Firstly, it makes up a demonstration of how the description of interpretations may benefit from prior analytical observations. Secondly, taking as a point of departure the reasonable assumption that interpretation as currently practiced includes an element of negotiation between the interpretative marks in the score and the musician's own convictions as regards how the musical structure should be rendered, the investigation is also a study of compliance. Turning to recordings by renowned musicians, to what extent is the frequently declared obedience *vis-à-vis* the composer's directions born out when it comes to actual musical practice?

# **Two passages from Beethoven's Piano Sonata Op. 110**

Departing from an account of structural traits pertinent for interpretation, 29 recordings of the first movement of Beethoven's Piano Sonata in A♭ major Op. 110 will be examined. A few preliminary words are due, motivating the choice of the two passages to be studied. 5

Starting with a poignant enharmonic modulation, mm. 66–70 veer off from the region of the subdominant D♭ major and settle in E major. Even more peculiar is the rectification of the tonal course in mm. 76–79; it sounds like turning the pointers of a clock back by half an hour. A diatonic sequence starting in A major is drastically curtailed by falling semitones, and then an extra bar is inserted that demonstratively lifts the music up to a renewed start in D♭ major.

These two extraordinary passages, provided with a few, but crucial performance directions and charged with structural tension and extra-musical content, cannot but challenge a pianist's intellect and imagination: even at a cursory glance they disclose a wealth of interpretative options. And yet, their joint effect is just to carry out a basic duty within the tonal machinery of the sonata form, that of keeping the recapitulation to the tonic no matter the excursions.

<sup>5</sup> The short and quite static development of the movement has also been dealt with in the same way; cf. Bengt Edlund, "Directions and Compliance. The Development" in Varia 1

The sudden enharmonic shift from D♭ major to C♯ minor in mm. 66/67 is preceded by two complementary melodic phrases in the subdominant; cf. Ex. 1. They have identical rhythms, and while the rising phrase eventually descends to c3 , the lower neighbour-note to d♭<sup>3</sup> , the falling phrase starts from the upper neighbour-note e♭<sup>3</sup> and ends with an ascending fourth so as to reach d♭<sup>3</sup> . The underlying (applied) tonic/dominant/tonic harmonic progression lacks closure due to the final first-inversion chord.

Depending on whether the pianist decides to counterbalance the proximity of the eighth-notes to the following accented notes, the rhythmic grouping within the two phrases is either a dactyl followed by a trochee (as indicated by the slurs) or a dactyl overlapped by an amphibrach. As to mm. 66–67, dactyls will come to the fore if the unexpected enharmonic shift is played so as to suggest a clear demarcation; otherwise the eighth-note d♭<sup>3</sup> will join the c♯<sup>3</sup> , giving rise to an amphibrach group across the crucial bar-line. In m. 69 the afterbeat of the amphibrach is extended and attenuated by a cadenza-like figuration (highly reminiscent of the one appearing in m. 4) until the very last notes of the bar, notes that tend to attach as an upbeat to the g♯<sup>2</sup> of the E-major harmony, tonicized by the long applied dominant. If, on the other hand, m. 69 is played so as to clearly start with a contracted replica of the trochaic sigh motif in m. 68, the entire cadenza will emerge as a prolonged upbeat.

The passage is characterized by the arrest of the melody at d♭<sup>3</sup> /c#3 , subtle grouping ambiguities, and (excepting m. 69) a highly uniform rhythm. Indeed, the uniformity allows of a four-bar unit starting in m. 66, competing with the regular division of mm. 63–66 into two-bar phrases. To support the former impression, the rising fourth a♭<sup>2</sup> –d♭<sup>3</sup> in m. 66 has to be played, not as an ending, but as the beginning of an intruding expansive phrase – an expansion to be inhibited by the sudden enharmonic change in m. 67, introducing a minor sonority. Turning to the very agent of the modulation, the bass line in mm. 65– 68, it features a slow, chromatically falling motion g♭<sup>1</sup> –f1–e♮<sup>1</sup> –d♯<sup>1</sup> , overlapping whatever demarcations there may be in the treble.

The most crucial question as regards the interpretation of this passage is whether and how m. 67 is to be foreboded. The left-hand e♮<sup>1</sup> may, in virtue of the *crescendo* and the rising fourth in mm. 66, be rendered as the result of an accumulation of tension, and it may even emerge as expected since it is implied by the descending bass line started in m. 65. But the enharmonic shift can also be highlighted as a surprise if it is somewhat delayed or, contrary to the *diminuendo* mark, if it is played *subito piano* – indeed, the e♮<sup>1</sup> in the bass may even be played as if it were caused by the preceding, repeated eighth-note d♭<sup>3</sup> in the melody. In any case it must be observed that the enharmonic shift occurs one bar before the upper line makes it manifest by finally bending downwards to b♮<sup>2</sup> in m. 68. Beethoven's *diminuendo* notwithstanding, it might therefore also make sense to maintain the tension and the *crescendo* until the melodic turning-point has arrived and to subdue the moment of enharmonic change, turning it into a secondary, retrospectively understood event by just treating the e♮<sup>1</sup> in m. 67 as a chromatic passing-note.

The shocking events in mm. 77–78, precipitately effecting a change from A major to D♭ major, make up the most challenging passage in the entire movement, but in order to discern the options for interpretation, the nature of these events must be thoroughly discussed; cf. Ex. 2.

The rapid shift happens around the double bar. The gentle dropping of single notes is suddenly replaced by simultaneous octaves in both hands. The unexpected and too early e♯/e♯<sup>1</sup> -instead-of-e1 -to-e in the left hand for a short moment offends the ear as an apparent dissonance, and the quasi appoggiaturaresolution e♯/e♯<sup>1</sup> –e♮/e♮<sup>1</sup> almost sounds like a correction of a mistake. But e♮/e♮<sup>1</sup> also makes up a passing-note from e♯/e♯<sup>1</sup> -read-as-f/f 1 towards the goal e♭/e♭<sup>1</sup> . Turning to the right hand, g♯<sup>2</sup> /g1 (or rather a due-time g♯<sup>2</sup> ) is expected, but this is not the case with the ensuing falling inflection to g♮<sup>2</sup> /g♮<sup>1</sup> . The slur suggests a sense of appoggiatura, but retroactively g♮<sup>2</sup> /g♮<sup>1</sup> is rather heard as an anticipation of g2 /g1 . From a harmonic point of view, this peculiar instant modulation is effected by two minor thirds, nominally representing E♯ minor and E minor. Rhythmically g♮<sup>2</sup> /g♮<sup>1</sup> makes up either an afterbeat or an upbeat, heading for E♭ major after the bar-line.

Which of all these aspects do pianists select for expression, given the means at disposal: articulation, modification as to rhythm and tempo, and dynamic emphasis, balance and contrast? Is this drastic modulation prepared in some way, and is m. 77 joined with m. 78 or demarcated from it by, say, a sudden change in dynamics or a short moment of silence?

M. 78 brings a gesture leading steeply upwards. The score tells the pianists to play first *crescendo* and then *ritenente/zurückhaltend.* Whether this means that the rising gesture in spite of its resurging momentum is to lose the power to pursue its course, or that it has the power required to reach its goal in a vehement uphill effort, cannot be positively established. But no matter whether there is local continuity in sound and timing across the bar-line or not, m. 79 comes up with a fresh D♭-major start, replacing the A-major one in m. 76.

The rising parallel motions in m. 78 – the notes are to be articulated in pairs – make up a series of consecutive thirds of an E♭-major chord expanding upwards, a series that can be heard as surviving the *ritenente*, and whose last member, f 3 /f 2 -over-d♭<sup>2</sup> /d♭<sup>1</sup> , might still represent the ninth and the seventh of E♭-major. Indeed, m. 79 has a sense of augmented reversal – while restating the material from mm. 76–77, mm. 79–80 slowly and quietly withdraw downwards along the E♭-major scale/triad. Understood in this way, mm. 78–79 exhibit a tight structural continuity, and the new start in m. 79 represents not only the subdominant D♭ major, but also the upper components of the E♭-major dominant, a fact that opens up for the possibility of a subtle and interpretatively challenging difference in relation to m. 20 of the exposition.

When comparing mm. 76–77 with mm. 79–80 two differences come to the fore. The former passage, featuring two right-hand slurs but only one slur in the left hand, asks for being played *molto legato*, while the latter, having two-bar slurs in both hands, is marked *espressivo.* After the enigmatic modulation in mm. 77– 78, must not the "same" material be rendered differently? The post-modulation atmosphere of mm. 79–80 might be expressed in various ways: by another dynamic nuance, by another tempo, by another treatment of the leading-note d♮<sup>3</sup> connecting the two bars, by a different balance between the four voices, or – which perhaps amounts to almost the same thing, but seems more active – by a different manner of emphasizing either the strong or the weak notes within each quarter-note unit.

# **Observations from the recordings**

The following recordings were examined; performances on the hammerclavier are marked with **H**. 6 To give an idea of the tempos, a metronome value (calculated for mm. 44–47, a quite uniform passage of the development) is added after each entry.

```
Arrau 1987 (Philips 422067-2) 54
Ashkenazy 1980 (Decca MCPS 417152-2) 63
Backhaus 1967 (Decca SXL 6300) 77
Badura-Skoda 1980 H (Astrée AS 49) 58
Barenboim 1 1969 (HMV HQS 1181) 49
Barenboim 2 1984 (DGG 413772-2) 54
Binns 1981 H (Oiseau Lyre D 185 D3) 57
Bishop 1974 (Philips 6500764) 51
Brendel 1 1966 (Vox VBX 417) 71
```
<sup>6</sup> Virtually all recordings stem from the collection of the Swedish Broadcasting Corporation, Stockholm. The author is very grateful for the courtesy of making them available for study.

```
Brendel 2 1974 (Philips 6768004) 65
Brendel 3 1983 (Philips 412789-2) 65
Browning 1967 (RCA Victor LSC 2963) 58
Demus 1 1966 H (Harmonia mundi HMS 30833) 59
Demus 2 1984 H (Fono FSM 123015) 58
Eschenbach 1979 (HMV 153-03628) 49
Földes 1968 (DGG LPM 18636) 68
Gilels 1985 (DGG 419174-2) 54
Gulda 1968 (Amadeo ASY 906444) 69
Gould 1956 (CBS M3K 39036) 57
Kempff 1964 (DGG LPM 18945) 62
Nat 1954 (Disque français DF 730.013) 71
Pollini 1975 (DGG 2530645) 68
Richter-Haaser 1959 (Columbia 33 CX 1666) 53
Rosen 1 1966 (Epic LC 3900) 56
Rosen 2 1971 (CBS M 30941) 56
Schnabel 1932 (HMV COLH 63) 68
Serkin 1 1972 (CBS M 31239) 59
Serkin 2 1987 (DGG 427498-2) 62
Siki 1959 (Columbia 33 CX 1185) 63
```
Before dealing with the modulation in mm. 66–67, it is of interest to describe some performance characteristics of mm. 63–64 and 65–66. As regards the rhythmic grouping of the melody, most renderings feature stressed eighthnotes that no matter the slurs give rise to a sense of amphibrach groups across the bar-lines. A few pianists play trochees in the even-numbered bars, and this happens in m. 66 in no less than eight recordings: the following eighth-note upbeat signals that something new and unusual has started. In some renderings the rising-then-falling neighbour-note motions in the bass are brought out so as to suggest complementary pairs of bars: in mm. 63–64 the first e♭<sup>1</sup> as well as the preceding d1 (or d♭<sup>1</sup> 's) are stressed in some recordings, and this way of playing is even more frequent in mm. 65–66. Prominent f1 's in m. 66 are often used to build up tension and expectation.

Turning to the very locus of the chromatic/enharmonic modulation in mm. 66–67, very few (if any) of the pianists play the shift inconspicuously or refrain from preparing for it, but the strategies for making it stand out are quite diverse. In almost all recordings m. 67 is slower than m. 66; it is either played *ritardando* or in a sudden *meno tempo*. Excepting Rosen 1, amphibrach groups come to the fore in mm. 66/67 and then in 67/68; sometimes the eighth-notes are quite perceptibly stressed, sometimes not. Some twenty pianists observe the prescribed *cresc./dimin.*, but in several of these performances the *diminuendo* is rather exchanged for a *subito piano* effect, which is most often associated with a short moment of silence at the bar-line. Two pianists bring a slight *ritardando* just before the bar-line as a preparing cue.

But in several of the recordings the increase-decrease in dynamics is displaced or radically changed. Badura-Skoda starts his *crescendo* already at the beginning of m. 65, Backhaus plays *diminuendo* already in m. 66, and Schnabel reaches his dynamic peak at the second beat of m. 66. Brendel 1 and 2 as well as Serkin 2 withhold the *diminuendo* until the soprano c♯<sup>3</sup> –b♮<sup>2</sup> ♯ resolution in m. 68, thus postponing the turning point of the section and diminishing the focus on the preceding enharmonic change. In Demus 2 there are diminuendos in each bar in mm. 65–68 whereas Gould brings *crescendo*-*diminuendo* waves around the bar-lines.

In the majority of the recordings the bass is rendered prominent; it seems to give rise to the *crescendo* and in some cases even to produce the harmonic shift. At least four different patterns of bass prominence can be found. Sometimes all quarter-notes in mm. 66–67 (or 65–68) come to the fore making for a continuous chain. Another quite frequent option is to stress only the third and first quarternotes of mm. 66/67 and 67/68, suggesting iambic groups underscoring the falling bass line. To the same effect three quarter-notes seemed to lead to a dotted halfnote in some recordings; this pattern appears either in mm. 65–66 and 67–68 or only in mm. 66–67. Richter-Haaser, playing the entire passage mm. 63–69 virtually without using the sustaining pedal, marks the *crescendo* and directs attention to the modulation by pedalling just the two quarter-notes around the crucial bar-line.

Most pianists play progressively slower in mm. 68–69, and in most of these recordings the amphibrach group in mm. 68/69 seemed to reach until the Emajor downbeat in m. 70. This sense of dissipation is also present in several recordings that do not feature any *ritardando*. However, there are also a few pianists (Barenboim 1 and 2, Rosen 2) who rather appeared to start m. 69 with a short trochee, as well as some (Browning, Gould, Földes, Kempff) who infuse energy by breaking up the rapid motion into smaller units.

Turning to the passage mm. 76–80, the variety is very great indeed, and we will first consider the passage from the second beat of m. 77 to the first beat of m. 78, and then that from the first beat of m. 78 to the first beat of m. 79. Finally, the surrounding parallel passages mm. 76–77 and 79–80 will be compared.

Since the abrupt modulation by means of descending thirds is quite extraordinary, all pianists make it stand out, and some of them also prepare the listener for it. Browning, Eschenbach and Gilels begin a *ritardando* already at the second beat of m. 77, and in several recordings the unexpected g♯<sup>2</sup> /g♯<sup>1</sup> -over-e♯/ e♯<sup>1</sup> is announced by treating the second beat of m. 77 as an upbeat. Rosen 2 gives weight to both f♯ in the left hand and to the right-hand a1 , while Ashkenazy, Backhaus, Gilels and Kempff stress only the left-hand f♯ – the first three of them start a clearly perceptible chromatic descent from this note to e♭.

The parallel thirds making up the third beat in m. 77 tend to be highlighted by dynamic means, and various patterns emerge. Five pianists play suddenly louder while Bishop and Gilels bring a *subito piano*. *Diminuendo* is frequent between g♯<sup>2</sup> /g♯<sup>1</sup> -over-e♯/e♯<sup>1</sup> and g♮<sup>2</sup> /g♮<sup>1</sup> -over-e♮/e♮<sup>1</sup> , and considering also g2 / g1 -over-e♭/e♭<sup>1</sup> , several pianists play an overall *decrescendo*. There are also some recordings featuring an overall *crescendo* or a *crescendo* just at the bar-line. Nine pianists begin m. 78 *subito forte*; Backhaus, being quite loud already, plays almost *fortissimo*. Brendel's recordings, on the other hand, feature a *subito piano* at the g2 /g1 -over-e♭/e♭<sup>1</sup> *.* Binns and Demus seemed to play the shift g♯<sup>2</sup> /g♯<sup>1</sup> -over-e♯/e♯<sup>1</sup> to g♮<sup>2</sup> /g♮<sup>1</sup> -over-e♮/e♮<sup>1</sup> *diminuendo* in the right hand and *crescendo* in the left – perhaps a way of reflecting the sense of anticipation and passing leading-note, respectively. Some pianists appeared to give equal emphasis to both hands while others more or less clearly favour one of the hands, strengthening the sense of an upper- or lower-line connection.

A majority of the pianists play this modulating transition more or less *ritardando*, but in several cases the retard comes to the fore as a sudden lengthening of g♮<sup>2</sup> /g♮<sup>1</sup> -over-e♮/e♮<sup>1</sup> or as a short delay at the bar-line. Gould, on the other hand, lingers on g♯<sup>2</sup> /g♯<sup>1</sup> -over-e♯/e♯<sup>1</sup> and then plays g♮<sup>2</sup> /g♮<sup>1</sup> -over-e♮/e♮<sup>1</sup> *à tempo* and *subito forte*. While *legato* articulation prevails, Schnabel plays the two eighth-note chords *portato* in both hands; in Serkin 2 this articulation turns up in the right hand, while Brendel 2 has it in the left hand.

All but two pianists play m. 78 *crescendo*; Brendel 3 features constant-level dynamics while Backhaus, starting very loud, renders the ascending gesture *diminuendo* (and just slightly *ritardando*). A quite common trait is that the increase in dynamics is replaced by a sudden *diminuendo* when reaching the last two sixteenth-notes. This feature is probably related to the *ritenente* marking in the score, a prescription that is respected (to various degrees) by all pianists, and that in most cases is taken to apply to the last two chords only; Schnabel taxes both the increase in dynamics in m. 78 and the final retard to the utmost. In some recordings the *ritardando* is preceded by a perceptible *accelerando*. In Gould and Brendel 1 a♭<sup>2</sup> /a♭<sup>1</sup> -over-f/f 1 (being a metrically weak, passing chord) is played softly, postponing the *crescendo*; Gulda separates the four sixteenth-note chords by playing them *subito forte*.

#### Conclusions 61

As to the transition to f3 /d♭<sup>2</sup> in m. 79, nine recordings suggest a demarcation by means of either a short fermata on the preceding sixteenth-note or a short moment of silence at the bar-line. Eight pianists, on the other hand, tightly attach the ascent to the beginning of the descent. But none of these renderings seemed to suggest that E♭-major is prolonged over the bar-line.

No less than fourteen of the recordings exhibit perceptible differences between the passages mm. 76–77 and mm. 79–80. In two cases (Ashkenazy and Gilels) the second, transposed statement is played more softly, whereas the other interpretations involve tempo differences; Barenboim and Schnabel play faster the second time while Gilels, Eschenbach, and Brendel 1 and 2 are slower. But a change of meaning is also suggested by minute differences in rhythm and dynamics. Seven recordings (Backhaus, Binns, Brendel 2 and 3, Gulda, Schnabel, and Serkin 1) first feature upbeat-like stresses on the second and fourth notes within each unit; after the modulation this patterning is replaced by downbeat stresses on the first and third notes. Such a difference is also to be found in Ashkenazy's recording, but he rather emphasizes the third and fourth notes in mm. 76–77. Barenboim gives prominence to the second and fourth notes in both passages, but adds a strong third note in the second statement. Demus 1 and Serkin 2, finally, feature still another configuration: emphases on the first and second notes of each unit in mm. 76–77, and on the first and third notes in mm. 79–80. Whether the various ways of playing accounted for are meant to observe the distinction between the prescriptions *molto legato* and *espressivo* is an open question. 7

#### **Conclusions**

A generalizing summary of the disparate findings will not be offered; in fact, it cannot be given. Besides, since interpretation is a matter of how musicians understand and play specific passages of particular works, it would be contradictory to make a contribution to what may be called "interpretation practice".

But the outcome yields a general methodological point. Listening repeatedly to recordings in order to find out what happens is a quite difficult and bewildering task. Yet it turned out to be feasible due to the fact that it was preceded by and based on analyses bringing out structural traits relevant for interpretation, and

<sup>7</sup> All these observations as to the relative dynamic prominence of the notes within the metric units are somewhat uncertain since different registers are involved.

perhaps even making for a hypothetical understanding of the intentions of the pianists.

Turning to the second aim of the investigation, very few pianists played according to the principle "the whole truth and nothing but the truth". The renderings featured various interferences not indicated by the composer, a fact that is neither remarkable, nor objectionable since all prescriptions (or proposals) as to interpretation cannot very well be entered in a score. In addition, a significant number of deviations from the interpretational marks – interferences involving neglect, change, or even reversal – were also present in the recordings. But if you entertain a reasonable notion of what interpretation amounts to, there is no reason to be very upset. Such transgressions or "violations" appear legitimate if you think that the musical passages in question allow of different ways of playing, and if the interpretations make sense.

To a varying extent and in various ways the 29 pianists have shown how *they* have understood the two crucial passages: interpretations, and not just compositions, are worthy objects of our attention. Even when playing Beethoven's Piano Sonata Op. 110 – a composer and a work instilling respect – pianists do not always comply with the interpretative directions given in the score, an element of non-compliance that runs contrary to what many musicians claim when it comes to a traditional principle in the art of interpretation, that of strictly observing the given text. Turning especially to mm. 76–80, some of the 29 pianists perhaps did not think that it was their duty to reproduce what might after all amount to a composer's idiosyncratic idea of interpretation.

# **Chapter 4 Loyal disobedience. When is it OK not to play as written?**

# **From fidelity to loyalty**

Musicians devoting themselves to Classical music usually maintain that it is their duty to strictly observe everything in the score; indeed, some of them act as if they had signed a contract with the composer to that effect. On the other hand, the very same musicians, wittingly or unwittingly, do not always and fully live up to their professed ideal – and sometimes they fail to do so for good reasons. 1 Nevertheless, there is no doubt a strong and persistent agreement within the community that you should be unswervingly loyal to the work.

This attitude is grounded in the laudable ambition to present the works in all their grandeur and subtlety, and the musicians shoulder the responsibility for respecting the *integrity* of the works. The straightforward, but arguably too radical, too narrow-minded conclusion is that a prerequisite for being loyal to the work, for respecting its integrity, is that you must pay unconditional respect to what the composer once wrote in his score.

It should be added that musicians are not very worried by ontological concerns, by the idea that any deviation from the "pitch-time" structure specified in the score is bound to jeopardize the work's *identity*. Music works do survive a fair amount of deviations of various sorts; it takes a good deal of grave liberties (let alone unintentional wrong notes) before it is reasonable to say that one has failed to play a work. 2

On second thoughts, however, the current ethos of fidelity/loyalty, bordering on submission, is not self-evident. A comparison with what goes on in the theatre is enlightening. When staging a drama, lines may be changed, omitted, or added; indeed, entire scenes are sometimes taken away. It has also come to be accepted

<sup>1</sup> Cf. Bengt Edlund, "Directions and compliance", Chapter 3 in this volume.

<sup>2</sup> Cf. Nelson Goodman, *Languages of Art*, Minneapolis 1968 and Bengt Edlund, "On scores and works of music. Interpretation and identity", Chapter 1 in the present volume**.** It might be claimed that intentional deviations from the score are ontologically fatal whereas occasional wrong notes are not. But listeners are not always capable of noticing the difference, and what about performances in which the musician intends to play the right notes but has read the score incorrectly? Furthermore, scores may for various reasons be corrupt.

that plays can be transplanted as to time and/or place, and that the tendency of a drama may be altered in ways that obviously go far beyond, or even run against, what the author apparently or presumably had in mind. And very few attempts are made to stage old plays so as to achieve stylistically correct productions, which is fairly frequent in music. Why have we accepted that it is legitimate to interfere with the plays of Shakespeare, Schiller, and Schnitzler in ways that we do not allow – and that we willingly abstain from – when it comes to the works of Bach, Beethoven, and Brahms?

One explanation is that it is easier to deviate from the text when staging a drama; due to the tightly knit structure of many music works it is difficult to make changes and get away with it. Another, more valid reason is that dramatic texts are usually deeply rooted in their historical context. Plays grow old, and therefore it may seem necessary to modify them in order to make them relevant or even understandable for modern spectators. Music works, on the other hand, tend to seem here and now as soon as they are performed and listened to, no matter where and when they were composed. There is no need to adapt them since, according to a widely held belief, they are not about anything specific, and since their original context has become less relevant.

But there is a twist in this. The adaptations made by directors, actors, and stage designers are considered necessary in order to revitalize the dramas, and in this light the difference between theatre and music-making emerges as less clearcut. It is likely that musical interpretation as currently practiced has assimilated various piecemeal adaptations undertaken throughout the centuries, although these changes, when compared with the altering practices within the theatre, may be different both as to kind and degree. Behind the seemingly unfettered art of staging dramas there is often and after all an ambition to be, in a way, loyal to the works.

The idea of strict adherence to the score as entertained by many present-day musicians may to an appreciable extent be anachronistic. In the 18th and 19th centuries the score was not accorded a normative status *vis-à-vis* the performances in the way we tend to take for granted nowadays – for one thing, making music often included various elements of improvisation. Furthermore, the scores were not necessarily regarded as definitive; the idea of the fixed, unalterable work had not yet taken firm root, hence the profusion of arrangements.

It was suggested *en passant* that "there may sometimes be good reasons" for not playing as written in the score. Given the prevailing culture of loyal submission to the text, how can such a standpoint be defended?

First of all, it seems reasonable to adopt a distinction between signs in the score that are essential since they define the musical structure, on the one hand, and signs that may be considered contingent since they rather refer to the execution of the music, on the other. The former are inscribed by the composer as a composer whereas the latter stem from the composer as the first (actual or virtual) interpreter of the work. 3 And venturing to be reasonable once again, while musicians are obliged to respect "structural" signs, they may sometimes be relieved of the duty to observe "interpretational" signs, i.e. the signs inscribed by a fellow musician.

This is merely to acknowledge the fact there are nowadays two professions in Classical music, a division of labour that has developed gradually, and that implies that we should expect to find traces of the composer as a musician in the scores we have inherited; many composers were quite accomplished performers. This division of labour makes for a further conclusion: it may be legitimate to question and sometimes deviate from certain indications in the score if you feel that they infringe upon your rights, your freedom as a musician. When it comes to the performance of written-down music there is always a hidden conflict between two complementary and sometimes competing endeavours – that of the composer and that of the musician – endeavours that must be in balance if the musician is to keep a reasonable degree of artistic independence, is to maintain his/her right of devising interpretations.

Turning to practice, there is of course a problem involved: it may be difficult to tell structural from interpretative signs. This distinction is not simply the same as that between the exact signs specifying pitch-time events and the approximate signs indicating "additional" properties of the music. Whereas, for instance, many dynamic marks can be regarded as interpretative, some others are clearly structural, and if you treat signs of the latter kind as optional, you run the risk of exposing yourself as a fool if your interpretation turns out to be inferior to the one specified in the score. On the other hand, seemingly structural signs are not necessarily work-defining; there may be cases where signs referring to pitch and duration are non-structural and can be considered negotiable. As to the supposedly non-negotiable pitch-time structures to be found in the scores, one should take into account the possibility that some "composed" solo works (or parts thereof) may in fact be the final, or perhaps not-so-final, results of improvisation.

<sup>3</sup> For more on this distinction, cf. Bengt Edlund, "*Sonate, que te fais-je?* Towards a theory of interpretation", Chapter 2 in the present volume.

In addition, there are further, even more imponderable factors. Are the interpretational marks in the score to be understood as imperative – and hence as infringing upon your freedom as a musician – or just to be taken as helpful suggestions to amateurs not knowing what to do? And quite a few interpretational signs are redundant since they specify things that a competent musician would do anyway. Being also busy as performers, many composers realized that music by necessity involves a give-and-take relationship between composers and musicians. There is a chance, then, that some composers were in fact quite reasonable people, providing hints rather than orders.

If you want to find out what a work affords in terms of interpretational opportunities, and particularly if you want to negotiate with its composer, it is more productive to think of him, not as "Herr van Beethoven", but as "Ludwig". But suppose that he was not at all a reasonable person, and that he will posthumously be very angry at you. All that you can say in your defence is that in the long run his works might fare better if he were kind enough to allow the musicians some freedom, if he allowed his works to display a greater degree of interpretational variability.

Finally, we may assume that composers (like all of us) were sometimes at pains to make up their minds – hence, perhaps, the existence of divergent parallel passages and the revisions of entire works. We all know that there are different, but often virtually equivalent, ways of putting things when we speak and write, and that goes for composing music as well (although music may be more form than content).

Approaching the subject of the present essay, it might be argued that there are no infallible composers – like all humans they had both bad days and bad ideas – and hence no perfect compositions. Consequently, all works are likely to benefit from loyal *and* critical attempts to read and bring life to the scores. The best results are arrived at if the composer (usually dead) works in tandem with the interpreter, trying to understand what the composer once wrote and presumably wanted. But this is not possible if the composer is put on a pedestal, if the musician is always the underdog. In a task presupposing intimate cooperation, it is very difficult, and also somewhat humiliating, to feel forced by your partner to do things that you do not quite understand or find detrimental to the final result.

Since composers sometimes do make mistakes, someone must assume the responsibility for amending the flaws, and – leaving here the important job made by editors out of account – those who have to do so are the musicians undertaking to play the music. Needless to say, amending things cannot but involve a risk of committing errors or doing stupid things. 4 But what is the point of displaying over and over again the very same imperfections to ever new listeners? If there are "problematic" (unsatisfactory, exaggerated, unnecessary, out-of-taste, incomprehensible) and yet sacrosanct passages in the Masterworks, there is also a strain of masochism in the current practice of Classical music. Perhaps the quasi-contract idea of the relationship between work and musician should be re-negotiated, be exchanged for, say (and no offence), the tacit understanding between a client and his/her make-up artist, the former having entrusted his/her face (including its imperfections) to the latter, wanting him/her to do the best out of it, using some discretion when necessary.

This essay is not an attempt at persuading musicians not to bother about the score whenever they think that they have hit upon a better idea. The modifications to be proposed will be as small as possible, and they should only be undertaken in order to amend passages that arguably do involve real, perceptible flaws. The ambition is to improve the works by taking critical care of passages that have become stumbling-blocks to you. And presumably not only to you; it is likely that there is, by and large, a consensus among musicians, and often among their listeners as well, that there are certain passages that feel like having a stone in your shoe.

As regards signs that arguably have a structural function, one should observe restraint – without very strong reasons one must not infringe on the composer's prerogative, without imperative reasons one cannot as a loyal interpreter allow oneself to (perhaps) compromise the work. Treating interpretational signs with some discretion is less controversial since it may be argued that musicians should be accorded a reasonable degree of creative freedom. In what follows we will therefore mostly deal with signs that, one way or the other and to varying extent, may be considered as structural; as a consequence and a bonus, some of the cases to be brought up and some of the interferences to be proposed will emerge as debatable.

All examples will derive from the literature for solo piano; unsatisfactory passages may turn up in every genre, of course, but when several musicians are involved, it may be difficult to reach an agreement and to implement the modifications. The examples are selected so as to represent a wide variety of situations that perhaps call for, or at least invite to, amending interventions of different kinds.

<sup>4</sup> Recall the amended *al fresco* painting of Christ in the church of Borja in Spain.

Each passage will first be studied in order to find out in what way it perhaps makes up or involves a flaw demanding critical attention. The next thing to be considered is whether the problem can be solved or at least circumvented by generally accepted interpretational means. Only then will more radical solutions (deletions, additions, re-compositions, etc.) be proposed, which – if accepted as legitimate – may perceptibly improve the situation or entirely do away with the problem.

When presenting the examples, the first person will often be used, thus admitting that it might be "I" who have the problem, not necessarily the passage in question. Using "I" also invites "you" to disagree, but the following discussions will of course be more interesting if you agree that there is something in the passages that after all may be less than optimal – and if you disagree about the proposed amendments.

Three things must always be kept in mind. The cure must never be worse than the illness. Loyalty counts for more than fidelity. The worst way of being disloyal to a work of music is not to (say) add or leave out a note, but to be unresponsive, to neglect the work's aesthetic potential.

#### **Adding and omitting notes**

Starting with a small problem, what is wrong with the very end of Chopin's Impromptu Op. 51? (Ex. 1a) Arguably, the dynamic marks in the last few bars seem exaggerated; this delicate piece should not bid a heavy-handed farewell. But this is clearly a matter of interpretative signs: start the last phrase *mezzoforte* and close it *forte* if you like.

But the passage involves two more interesting problems. As a pianist, I do not like the gap in the prescribed *legato* of the melody: the penultimate melody note a♭<sup>1</sup> has to be left for a short moment while catching the deep G♭<sup>1</sup> with the pedal. Furthermore, keen to clarify the voice leading, I do not want anyone to hear a melody closing with a false a♭<sup>1</sup> –f 1 –g♭<sup>1</sup> motion instead of just a falling step a♭<sup>1</sup> –g♭<sup>1</sup> as patently indicated in the score' a misunderstanding that may be hard to avoid even if you take care to play the a♭<sup>1</sup> louder than the inner-voice resolution to f1 .

There is a simple remedy that does away with both problems, and it is so natural that the right hand barely can resist it: just add an anticipatory, "smallprint" eighth-note g♭<sup>1</sup> before the bar-line. (Ex. 1b) This extra melody note can be held while you secure the left-hand G♭<sup>1</sup> , and it lets the g♭<sup>1</sup> –g♭<sup>1</sup> anticipation cliché steal the listener's attention from the unwanted f1 –g♭<sup>1</sup> leading-note connection.

The musical situation is changed to the better, but is it acceptable to add this g♭<sup>1</sup> ? Well, Chopin certainly knew about melodic anticipations, and hence I have not done anything that Chopin himself could not have come up with. Furthermore, the extra note does not give rise to any great musical difference. Indeed, the fact that the two formulations are virtually equivalent exemplifies the observation that the very notes prescribed in a score are not always and necessarily structurally essential, do not always define the work. Additions, omissions, and changes with respect to pitch-time, "structural", signs are sometimes possible; they do not as a matter of principle affect a work's identity, and even less its integrity.

In Rachmaninov's Prelude Op. 23, No. 5 there is arguably one note too many in the right hand in mm. 1–2 and 10–11. (Ex. 2) Since the melody, divided between the hands, obviously proceeds in deep octaves in mm. 1–6 and 10–16, the fourthbeat b♭'s in the right hand are superfluous. These notes are disturbing in a way that the triple tonic notes starting the bars and giving emphasis to the primary downbeats are not. Should one take these b♭'s away? Yes, why not? But since the right hand must have something to do on the fourth beats – in fact, this seems to be the *raison d´être* of the musically redundant b♭'s – use the right hand to play the B♭'s. (This applies also to the g in m. 6.)

Later on in the prelude there is a passage featuring a bewildering difference. Why does the right hand play octaves within the fourth beat in m. 17 but not in mm. 18–20, where only the first and third sixteenth notes are reinforced by octaves? (Ex. 3) Since it is not likely that Rachmaninov felt any need to make the prelude easier to play, we have to figure out his musical intentions. It seems that it was important to him to achieve a tight *legato* in both hands, and also that he wanted to treat the fourth beat of these bars like the other beats, i.e. so as to suggest two stresses. In other words, the reading of mm. 18–20 is the one to be preferred, and therefore – if you want to get rid of the annoying difference – you might take away the c2 and e2 ♭ in m. 17, which opens up for the *legato* fingering 3454. Signalling two sub-accents within the fourth beats is a good idea since it helps to bring out a sense of quadruple time in mm. 17–22 and lends a firm, accumulating quality to this passage, which introduces a metric contrast to the rest of this march-like prelude with its latent *alla-breve* character.

Two categories of omitted notes will be mentioned but not exemplified since we deal with matters that should be treated with discretion.

It sometimes turns up chords that outsize your hands. In such cases it may be preferable to leave out a note rather than to break the chords. Some passages simply have too many notes to be played effectively or at the required tempo; a discerning use of the red pencil might help you out.

# **Compensating for the lack of keys**

We all know that composers like Bach, Mozart, Beethoven, and Chopin wrote music for keyboards lacking treble and bass keys that we nowadays take for granted. We may also safely assume that they would sometimes have used these keys, had they been available. This is at least what a number of passages, exhibiting awkward emergence solutions, strongly suggest – melodies, bass motions, or figurations that are obviously contracted or otherwise pursued in unexpected and less than optimal ways. But what was then utopian is now possible, and many present-day pianists simply extrapolate the notes needed to restore such passages to what they in all likelihood were meant to be.

A self-evident example is to be found in m. 315 of Chopin's Scherzo Op. 39. (Ex. 4a) After having already met with this figuration quite a number of times, it seems absurd to play this bar as written, i.e. to start with a silent second beat in the right hand and with no left-hand part on the second and third beats. (Ex. 4b) The fact that Chopin's keyboard lacked the key for g♭<sup>4</sup> is the only reasonable explanation for this flaw, and it is of course quite OK to add the missing notes. (Ex. 4c)

In addition, this second-theme actualizes what might appear as clear violations of the text, a rhythmic liberty that is so ubiquitous that we do not notice it or complain about it. Unless having played the preceding chords slowly or with a generous amount of *ritardando*, no pianist starts the rapid figurations in due second-beat time since it is virtually impossible to change hand positions fast enough, and since it would sound quite unnatural if you really managed to do so – listeners do not expect or require such radical changes in register and/ or musical substance to happen with metronomic exactness. Chopin might have notated these passages as in Ex. 4c but he didn't, perhaps because he knew that the pianists were going to play something like this anyway. He might also have wanted them to slightly emphasize the fifth right-hand note so as to clarify its strong metric position.

There are quite a few left-hand passages that insistently beg for keys that were not available at the time of composition, and that are still not played by strict pianists. Generally, it is of course a bad habit to add deep notes every now and then just because you want to have a big sound. But in bars 221–232 in the third movement of Beethoven's Sonata Op. 53 the final statement of the theme cannot very well start less impressing than the two preceding ones. (Ex. 5)

Another case in point is m. 45 in Chopin's Etude Op. 10, No. 1. Of course you should allow yourself and your audience the structural pleasure of completing the bass sequence with the octave B2 /B1 – to hear mm. 45–46 twice bringing the same bass B1 /B is not satisfactory. (Ex. 6)

Sometimes such decisions may be controversial, however. In Beethoven's Sonata Op. 27, No. 2, there is a passage that, after having played the first movement for almost sixty years, struck me as somewhat odd. (Ex. 7a) First comes a slow, but boldly rising sixth in the left hand – a motion promising great things to happen in the bass – but then m. 57 turns up, merely offering a fairly indifferent double neighbour-note motion. Perhaps is this disappointing outcome an unfortunate compromise caused by the compass of Beethoven's 1801 keyboard? How lucky we are nowadays when we are in the position to reconstruct a better melody out of the old!

But unlike in the Chopin Scherzo, there is no preceding model to rely upon, and therefore it takes some courage to let the first half-note rising sixth be followed by a descending sequence of three more urgent quarter-note ones, thus retaining Beethoven's pitch-class essence but using keys that he did not have at his disposal. (Ex. 7b) (To get used to the idea, invert the initial, half-note rising sixth so as to introduce a series falling thirds.) It should be pointed out that this interference with Beethoven's text, making for a descending sequence of rising sixths in the left hand, is (perhaps excessively) supported by the fact that the right hand might be conceived of as a descending series of falling seconds.

Admittedly, the three left-hand octave transpositions bring a radical melodic change in the bass, and in addition the soprano melody tends to emerge as a sequence of iambs, rather than (as convention bids) as a series of note-repeating trochees. Nobody has heard the passage being played in the descending-sixths way before, and some listeners might be very upset, wanting this famous movement to sound as it has always done. 5 As to myself, I must confess guilty of altering structural signs in a way that may perhaps compromise the integrity of the movement. But is my loyal interference really that bad? I can hear Ludwig

<sup>5</sup> It is a good idea to play the lower notes of the left-hand octaves softly in order to avoid a too thick sound. Also, playing iambs in both hands may amount to over-explicitness; therefore, balance the left-hand iambs against right-hand trochees as indicated in Ex. 7b.

shouting (using his ear-trumpet as a megaphone) "*Entlich, nach mehr als zwei Jahrhunderten!*"6

But there is a snag. Look up mm. 29–32 (124–127) in the third movement, and you will find two right-hand double neighbour-note motions over two falling steps. (Ex. 7c) Far from being a very obvious similarity, this amounts to a perhaps significant affinity that would be ruined if the crucial passage in the first movement were changed as just proposed. (I don't want to hear what Ludwig, having taken care to integrate his emergency solution (?) into the motivic design of the sonata, might shout.)

#### **Matters of proof-reading**

Let's compare the last ten-bar part of the final thematic section as it appears in the exposition and recapitulation of the first movement of Schubert's Sonata D. 575, studying these passages as they are printed in the Wiener Urtext (Universal) and the Henle Urtext editions. If you are a pedant – as you should be when setting about learning a sonata – you will notice quite a few differences. (Ex. 8 a/b)

Starting with the Wiener Urtext, the articulation slurs in the right hand are different in mm. 50–51 and 137–138, and in mm. 56–57 and 143–144; the main difference is that in the exposition the last two notes are slurred separately, as to the eighth-note upbeats they are sometimes slurred. The left-hand slurring in mm. 52–55 and 139–142 is also different: the recapitulation features fournote slurs leading to the downbeats whereas the exposition brings an irregular pattern. Turning to the tonal substance, the right-hand figuration is not the same in the fourth beat of mm. 50–51 and 137–138, and the first beats in the left hand in mm. 56–57 and mm. 143–144 are different – m. 144 has two eighth-notes.

In the Henle Urtext most of the articulation differences, but not the ones involving pitch, have been ironed out (probably) by the editor, choosing one of the options and resolutely keeping the pianist out of the problems.

This Schubert example represents a host of other cases exhibiting similar discrepancies, some of which have been removed by sharp-eyed editors before the pianists see the scores. And yet, as pianists we often meet with parallel passages that are not exactly alike, and we cannot shirk from critically evaluating what we read. Are these differences intentional or due to carelessness; are they immaterial or musically important; does the work gain if we respect or neglect

<sup>6</sup> This movement is further discussed in Bengt Edlund, "Navigating in Moonlight", in Varia 1

them? (And if we respect them, are we following the composer or perhaps just the editor?)

At any rate, such discrepancies are annoying if we try to learn, say, the Sonata D. 575 by heart. (I can hear Franz exclaiming "*Mensch, spielen Sie meine Sonate auswendig?!*") Why, we cannot but ask ourselves, should we waste time and concentration on memorizing lots of, and perhaps not very significant, differences that few listeners are likely to notice, let alone appreciate, differences that often seem arbitrary and unintentional?

What can be done to clear up this particular D. 575 mess? (We will avoid getting lost in source matters.) First of all, just add an eighth-note C♯<sup>1</sup> in m. 57 – we have this key nowadays and a soft, deep thud sounds nice here – and play an octave in m. 56 if you want. As to the difference in the right-hand figuration, it is arguably musically insignificant, and therefore it is up to you to choose; due to the lower register it may be a good idea not to double the third when the passage returns. The articulation differences may bring perceptible musical effects. Regular four-note slurs in the left hand giving rise to long upbeats make immediate and pleasant sense, and they should be adopted throughout. Turning to the melody, the main alternatives are a smooth seven-note *legato* or a more lively articulation with separate upbeats and/or afterbeats. If you think that variety was what Schubert wanted, you can play the exposition and the recapitulation differently, or choose one option in the high register and the other one when the passage returns in the low register.

#### **Voice-leading and manual rearrangements**

If one listens critically to the first four bars of Chopin's C-minor Prelude, one may notice that the melody of mm. 3–4 does not quite live up to that of mm. 1–2. (Ex. 9a) Two motifs issue from the same note, d1 , and the first of them brings a strange, contracted motion featuring first e♮<sup>1</sup> and then e♭<sup>1</sup> (rather than e♮<sup>1</sup> as in the autograph) while the second merely offers a fairly bland G-major triad. 7

But two more satisfactory and quite similar melodic motifs are hidden in the right-hand chords of mm. 3–4, namely b♮–c1 –g1 –f 1 –e♭<sup>1</sup> and c1 –d1 –b♮<sup>1</sup> –a♮<sup>1</sup> –g1 , motifs which after the low turning point on b♮ bring the melody back to e♭<sup>1</sup> and then to g1 , thus reversing the motion in mm. 1–2. It is not impossible to bring this alternative line out if you refrain from playing mm. 1–4 very loud (which

<sup>7</sup> For a thorough discussion of this prelude, cf. Bengt Edlund, "Reconsidering the Cminor Prelude", ch. 6 in *Chopin. The Preludes and Beyond*, Frankfurt 2013 Peter Lang Verlag.

often happens), if you support the motion upwards with the *crescendo*, and if you break with the habit of pianists to give dynamic precedence to the top voice of right-hand chords. (Ex. 9b)

The question is if you can allow yourself to meddle with the chords in this way, whether or not Chopin as a composer has indicated top-voice dominance. But pianists often excel in hidden polyphony, not least when playing Chopin. And what is so especial with chords, don't we have the right to "orchestrate" them in order to bring out something that we have discovered and found worthy of expression?

What is wrong with the passage from Liszt's *Nuages Gris*, where the "theme" starting the piece returns together with a melody above it? (Ex. 10a) The flaw is a matter of idiomatic, and by extension (and in a quite private way) of the musical structure as well. When starting the piece the theme is very well adapted to the right hand, but when it turns up as a secondary voice in the left hand the manual delight is gone. It is uncomfortable to play c♯<sup>1</sup> with the thumb, and then you have to make a disturbing finger substitution or glide with the thumb in order to play the d1 . (Although it feels awkward, it is also possible to achieve a sense of *legato* by playing the c♯<sup>1</sup> with the second finger.) But the real problem involved is the phenomenal change that occurs when the theme is transferred from the right hand to the left – to the pianist, it is no longer identical with itself, which it of course should be from a musical point of view.

The solution is simple: no matter Liszt's notation, play the "theme" with the right hand also when it is used as an accompaniment, and let the left hand take over the melody. (Ex. 10b) Not a single note has been altered, and yet the structure, as experienced by the pianist, has turned radically different, different in a way that makes the passage more comfortable to play and that accurately reflects what has happened, a fresh melody has been added above the "theme", which is exactly what the manual rearrangement suggests. A "third hand" has been introduced.

Fingerings tend to subtly alter the phenomenal structure for the pianist,8 and the distribution of the notes between the hands, causing great differences, can be regarded as a kind of fingering. Being a matter of execution, fingerings

<sup>8</sup> Cf. Bengt Edlund, "The phenomenology of fingering. Structure and ontology in the F-minor Etude from Méthode des méthodes", Chapter 7 in *Chopin. The Preludes and Beyond*, Frankfurt 2013, Peter Lang Verlag, and "A Comprehensive Approach to Musical Idiomatic", Chapter 6 in the present volume.

are generally considered to belong to the domain of the musician, but what about the quasi-structural differences brought about when interfering with the prescribed division of labour between the hands? Generally speaking, does the music's phenomenal structure, as experienced by the performer, fall within the competence of the composer, or does the musician have a say?

It may be argued that Liszt straightforwardly specified the notes to be played, but didn't necessarily prescribe how they were to be executed. No listener is likely to be aware of the fact that (if the version in Ex. 10b is accepted) the left hand plays the new melody, and it is hard to imagine that Liszt, a master of keyboard tricks, would have filed a protest against the proposed rearrangement. There are quite a few examples in the piano literature where the score shows the intended, but technically awkward musical structure, but not how the music should preferably be played. In other cases the notation is idiomatic, and it is left to the pianists to figure out what the musical structure is to be.

Sometimes changes in the distribution of the music between the hands give rise to audible differences. It is questionable whether Scriabin's notation of the first-movement main theme of the Sonata Op. 23 is the best one. (Ex. 11a) The core idea obviously consists of thirds running in contrary motion, and while the notation in mm. 1, 3, and 9 does suggest this, it also prescribes a convenient and yet less than optimal way of execution.

When playing these bars, I thought that it was desirable to bring out the contrary motion with full force by relegating the descending line to the left-hand thumb and the ascending motion to the right hand, adding the notes required for parallel octaves. (Ex. 11b) In m. 4 it feels fine to play both a1 and b1 with the left hand.

If you don't like to take risks, the exposed left-hand octave leaps might be divided between the hands. Scriabin's one-hand notation (and way of playing?) is more "*drammatico*" whereas the safe two-hand solution has the advantage of supporting the auditory imagery: the seemingly uncoordinated, too-early downbeats bring quasi-first-beat notes below the roots, and the sound recalls that of big bells.

The melody-to-become-accompaniment starting Chopin's Impromptu Op. 36 may benefit from a supporting harmonic fundament, and it can easily be provided for. Just play the F♯<sup>1</sup> or F♯, or both notes, of the initial tonic root with the right hand and then keep the key(s) depressed as long as suitable; the resonance drone can be gradually filtered away in m. 4 or 5. (Ex. 12) Less demonstratively, you can use the middle, sustaining pedal for prolonging F♯<sup>1</sup> /F♯. But don't use the right pedal clinically; the point of using the right hand to sustain the root is to more readily achieve an aura of sound that would otherwise require great skill (and some luck).

But no matter how you do, you may come in for some criticism: Chopin didn't write this, nor had he any sustaining pedal at his disposal. Turning from restrictions to opportunities, must it necessarily be argued that embedding the melody in rich resonance necessarily runs against the composer's intentions? Notice Chopin's pedal mark suggesting that at least to begin with he wanted the sounds to be blended. Maybe preserving the tonic root for several bars exceeds rather than runs contrary to his intentions?

## **Two problematical transitions**

In Chopin's Impromptu Op. 36 there are two, apparently inserted bars that at first may emerge as incomprehensible: the alto motif in m. 59 and its offshoot in the next bar seem so irrelevant. (Ex. 13a) These bars do effect the modulation back from the D-major-then-G-major march section, but the music can in fact do without them: the left-hand rise in m. 58 from G1 /G with its d/d1 anchor in the middle register leads perfectly into m. 61, eventually coming to rest at C/c/ c1 . And if you take a closer look, continuity is abundantly provided for also in the right hand of mm. 59–61. The theme is already in place; if you are not entirely absorbed by the doings of the middle voice, you will notice that the first bar of the theme is "pre-repeated" twice in mm. 59–60 – arguably a quite odd feature. 9 As to the alto, preparing for the accompaniment to come, g♯-then-a♭ rises to a♮, and the upper motion e1 –d1 , e♭<sup>1</sup> –d♭<sup>1</sup> descends to a dragging d♮<sup>1</sup> –c1 .

Obviously, you have to treat the alto as a two-voice subordinate chromatic transition and to play mm. 59–61 as a third-time-lucky structure. This is greatly facilitated if you let your hands confirm your musical intentions – or conversely, if you let your hands inform your mind. Play the alto voice in mm. 59–60 with the left hand, play it so as to show where it belongs, so as not to disturb the prerepeats in the melody. (Ex. 13b) While musically enlightening, this arrangement is more awkward to play.

<sup>9</sup> If you have missed the presence of the soprano theme, it may be due to the fact that you have started learning the impromptu by reading it with your fingers, i.e. by playing it before taking a close look in the score. The problem is that mm. 59–60 feel very different from the initial, right-hand presentation of the theme.

The transition taking place in mm. 60–62 in the first movement of Schumann's Fantasy Op. 17 is no doubt a stroke of genius. (Ex. 14a) But the notation is flawed since it indicates a way of execution that, although it is the most straightforward one, is not musically optimal; indeed, it might lead to a subtle and yet farreaching misinterpretation. It is all a matter of the left-hand C starting m. 61.

In the local perspective it is crucial to notice that the melody is released only in m. 62 with the slightly delayed f1 /a♮<sup>1</sup> third that almost imperceptibly introduces F major. 10 This means that the C-major root-position harmony at the beginning of m. 61, urged by the ripe G-major applied-dominant sonority in m. 60 and confirmed by the deep C, makes up a kind of false start. There is no C-major section starting in m. 61 since the note a♮<sup>1</sup> suggesting F major is immediately introduced, and then comes an undermining b♭<sup>1</sup> turning C major into an applied dominant – but an F-major section beginning in m. 62.

This being said, it is also true that, along with the falling-third motion c2 / a♮<sup>1</sup> –b♮<sup>1</sup> /a♭<sup>1</sup> –b♭<sup>1</sup> /g1 –a1 /f 1 , a return of the complete second theme furtively begins already in m. 61 with the (seemingly) missing notes c2 –b♮<sup>1</sup> –c2 . But Schumann hides these notes in the accompaniment; he rather wanted an apparently new idea starting with three c2 's to emerge in mm. 61–62. It would not only be overambitious, but also disloyal to the composer and his notation to bring out this hidden start, prematurely disclosing the transposed recurrence of the second theme, at the expense of the chromatic transition towards F major.

Turning to the long-term perspective, the first movement of the "C-major" Fantasy is remarkable because it avoids introducing the C-major tonic until the very last page. But the passage under discussion is on the verge of destroying this bold tonal plan, unless the pianist subdues the transient sense of a C-major tonic in m. 61, unless he/she refrains from slowing down before m. 60 and from playing a firm C suggesting the arrival of an important harmonic root.

But the very sight of the low left-hand C beginning m. 61 is treacherous since it suggests that after the G-major-dominant preparation this deep bass note has to represent a root of structural importance. But it should be tucked away – try for once to play the transition without it – and the best way to achieve this is to sidestep Schumann's notation by softly touching the C with the right hand while the left hand takes over the descending motions before and after the bar-line. (Ex. 14b) When playing the piano (and other instruments), there is much to

<sup>10</sup> This melody is not actually new since it brings a (seemingly) without-the-initialnotes variant of the D-minor second theme introduced in m. 41. A further qualifying observation will follow.

gain if you, whenever possible, let what your hands are doing correspond to the musical idea that you want to convey. 11

But there is another thing to consider. Ex. 14a reproduces the transition according to the Henle Urtext edition. While the very long slurs suggest sweeping motions downwards, they are not very informative as to the details of the passage. In the widely spread Peters edition by Emil von Sauer we can see several shorter slurs; cf. Ex. 14b. If we take account of them, we might be following von Sauer rather than Schumann, but this being said we can afford to admit that von Sauer's phrasing is musically perceptive: the slurs starting in m. 63 and 65 bring out the element of imitation in the passage. Of special interest is the c2 –f 2 – slur starting in m. 61 that completes the disguise of the second theme by expressly starting it after its hidden beginning. But you may also let the "new" melody emerge gradually out of the three c2 's.

# **Disregarding the composer's indications**

Composers have various means to specify how a certain passage is to be played. These indications may sometimes be merely suggestions; in other cases they are perhaps to be thought of as strongly held convictions: "this is how I urge you to play this passage". In the two examples to follow we will (in order to sharpen the conflict) assume that we are dealing with indications of the latter kind, and these very examples are chosen since they may have far-reaching consequences although they merely seem to involve local and fairly insignificant details of expression. 12 They also suggest that it is simply not true that the composers always possess the only artistically valid way of interpreting their music.

The first seven bars of Chopin's Ballade Op. 52 make up a self-sufficient section, demarcated from the following main theme by a fermata sign. (Ex. 15) According to the numerous hairpin marks what the composer had in mind was an introduction with an expressive top-voice melody, initially complemented by a most telling middle-register line.

But suppose that a pianist hits on the idea to play this passage *wie aus der Ferne* – i.e. very softly and *sans expression* – to make for a strong contrast to

<sup>11</sup> For a controversial example, the closing bars of an *Intermezzo* by Poulenc, cf. Bengt Edlund, "A comprehensive approach to musical idiomatics", Chapter 6 in the present volume.

<sup>12</sup> For another example, from the Fantasy Op. 49, cf. Bengt Edlund, "Chopin themes", Chapter 15 in this volume.

the lamenting main theme, or perhaps to follow up the introduction by playing also the main theme in a very restrained manner. Much later in the work, the introduction turns up again, now in A major. If the pianist sticks to his/her quiet attitude also in mm. 129–134 – which is what consistency bids, and which would suit the following groping transition eventually finding the way back to the main theme – the form and content of the ballade would be subtly altered. But since this has been achieved by disregarding Chopin's directions as penned down in mm. 1–7, it might, even if the result turns out to be quite interesting, be called into question by some people thinking that such interferences are illegitimate.

In the Nocturne Op. 62, No. 1 there is a five-bar, quasi-improvisatory and eventually closing passage leading from the dissipating end of the transformed main theme to a final section, made up of two similar, four-bar right-hand excursions, in turn followed by two one-bar additions and then three cadencing middle-register phrases. (Ex. 16) In the final bars of the transition you can read *ritardando* as well *diminuendo* whereas the first bar of the final section asks for *à tempo*.

Chopin's indications instructing the pianist how to bring out the formal shift are certainly musically valid. But again, let's suppose that there is a pianist who wants to introduce the transition slowly and almost imperceptibly, and then prefers to erase the demarcation between the transition and the first right-hand excursion, a pianist who is prepared to sacrifice the *diminuendo*, the *ritardando*, and the *à tempo* in order to present a grand, growing gesture bringing the music all the way from m. 75 up to the second-beat b2 in m. 81. This is what the three urging sub-phrases starting mm. 77, 78, and 79 may seem to demand, but it also means that you have to play m. 80 somewhat against the grain. Such an interpretation, reducing the perhaps excessive number of closing passages in the nocturne, and turning the second right-hand excursion into a *pianissimo* addition, may amount to a quite worthwhile listening experience. But is it permissible? On the other hand and as Chopin certainly knew, there are many ways to finish a story.

#### **Structural or interpretational signs?**

To strictly observe all *subito piano* indications has become the very epitome of a true Beethovenian style of playing. However, in as far as these signs are non-structural, in as far as they rather concern the interpretation of the music – which sometimes, perhaps quite often, seems to be the case – you may as a matter of principle consider yourself free to disregard them. Beethoven was no doubt fond of these effects, and he prescribed them frequently, almost excessively. (Indeed, if this interpretational habit of his can be likened to a kind of musical tic, we should not be impolite by using every opportunity to imitate him.)

The passage mm. 5–8 in the first movement of the Sonata Op. 26 is an interesting example since the status of these indications is not clear-cut, and it is also a quite vexing passage since two *cresc.–piano* directions turn up in close succession. 13 (Ex. 17) The second of these signs is apparently interpretational since the firstbeat dissonance can be rendered straightforwardly by an underscoring dynamic stress as well as by means of the prescribed "negative" emphasis. As to the first mark one might think that it must also be interpretational for the strange, even paradoxical reason that it destroys the lofty four-bar phrase. If we assume that both these signs are interpretative, we may for the sake of the long melodic line dispense with one of them – preferably the first one since it is most disruptive – or even ignore both. Try for once to play in this way, and you will find that it sounds quite all right. 14

But on closer consideration the first *subito piano* mark might after all have structural status. Perhaps Beethoven wanted to counteract the growing predictability of the melody by suddenly subduing the dynamics? Given the fourth-up-fifth-down sequence and the resulting stepwise falling tendencies of the two implicit lines, we can envisage where the melody will take us when we are about to hear m. 7. If this observation seems to the point, the *piano* sign (as well as the right-hand slur) starting this bar should be respected; the hushed dynamics is unexpected and keeps up the listener's interest in the otherwise too foreseeable progression. In other words, this interpretational inscription is structural in virtue of being anti-structural. To complete the picture, it must be observed that the left hand starts anew from f; the sudden *piano* may also reflect this turn of events.

But we must also consider the unexpected melodic twist at the *subito-piano* mark in m. 15. At the end of the so far identical consequent the *cantabile* line is suddenly abandoned and replaced by a standard cadence to be played *piano* – perhaps a structural sign rather than just an interpretational one. But if you have already provided a *subito piano* effect in m. 7, the corresponding spot in the antecedent, the listener will be less startled by the sudden *piano* in m. 15. In other

<sup>13</sup> For a detailed study of this theme, cf. Bengt Edlund, "Disciplining reduction and tonalizing interpretation", Chapter 2 in *Questioning Schenkerism*, Frankfurt 2015, Peter Lang Verlag

<sup>14</sup> Or (loyally exploiting Beethoven's tic) add some *subito piano* effects in the first four bars as well, i.e. when beginning m. 2 and m. 3. Didn't that sound a bit mannered?

words, the first of these two *subito-piano* indications cancels out some of the effect of the second. Depending on whether you suppress or take account of the sudden *piano* in m. 7, there are two options: a truly surprising m. 15, and a m. 15 where the listeners are aware that something surprising might occur. Irrespective of this relationship with the antecedent, the *subit*o *piano* in m. 15 brings out the somewhat wanton quality of the structural shift – it adds an element of humour to the theme, an element that it is up to the pianist to underscore or subdue.

# **Matters of embellishment and variation**

Based on overwhelming evidence, it has become an established truth that, as far as Baroque, Classical, and early Romantic keyboard music is concerned, repeats and otherwise recurring material in especially slow movements were often varied by means of additional ornaments or various non-standard inventions. If you choose to apply this insight when playing – you must not, but why shouldn't you – what conclusions can be drawn?

The first and quite obvious conclusion is that such *seconda-volta* additions or changes cannot be thought of as violations of the text – which unfortunately does not imply that your interventions cannot be out-of-style, excessive, or misplaced.

The second, less obvious and more controversial conclusion is that it may sometimes be OK to play sections later to be repeated in a simpler, less embellished way the first time, i.e. simpler than what you can see in the score. Doing so may seem to entail lots of violations of the text since it usually means that several specified events are omitted. Yet it emerges as a quite reasonable intervention if you think that the music as printed in your score is already enough (or even profusely) ornamented. It is quite conceivable that the composers sometimes preferred to show how the music might run when fully embellished – i.e. they wrote down the repeat, not its model – and left it to the musicians to reconstruct the "original", simpler version. Hence, handling repeats does not necessarily mean making the music (even) more complex; sometimes it entails simplification.

For instance, the repeats in the theme of Bach's Goldberg variations BWV 988 (Ex. 18a) may arguably be played in a more naked, but still expressive way when presenting the material the first time. (Ex. 18b) Listening to the richly ornamented version twice in succession borders on an overkill experience. And if you start with a less embellished version of the music, you may avoid choosing an overly slow tempo. When repeating the passage, you are free to imperceptibly slow down the tempo somewhat, just like a clever orator does when the text to be delivered turns more complex.

#### 82 Loyal disobedience

There is a particular reason for selecting this very piece as an example. The Aria closes with a contrasting, wonderfully expansive melody featuring plain sixteenth-notes, and it seems reasonable to assume that this is a "second-time" idea; it is a unique melody that should not be heard more than once. (Ex. 19) Therefore it is a good idea to save it until the second section of the Aria is repeated. The interpretative challenge involved is twofold. For *prima-volta* use you have to retrieve the simple "business-as-usual" version hidden beneath the printed *seconda-volta* melody, and when finally this melody is due, you must start it in a way that evokes a sudden sense of freedom. After having read a long, exquisitely artful poem, the speaker finally puts away the text letting his heart speak in prose, as it were. To underscore the contrast you may abandon all stylish small-scale articulation and give in to unrestrained *legato* playing.

In the (very) slow movement of Beethoven's Sonata Op. 10, No. 1 the second theme, a four-bar idea, is immediately varied, and this happens later on in the movement when this theme turns up again in the tonic. (Ex. 20a) Beethoven applies the current convention implying that restated material should deviate from, and preferably exceed its model. The problem in this case is that what he offers – extremely rapid passages, made up of a recurring cliché and introducing an element of cheap brilliance – is so excessive that the otherwise serious and noble character of the theme is at risk. Indeed, these bars spoil the entire movement so badly that I never cared for playing it, and when listening to it, I try to close my ears when they are due. But I have noticed that there are innumerable professional masochists who can stand these excesses and have polished them to perfection.

But must I, and must you, play mm. 28/75 and 30/77 as prescribed in the score? If Beethoven ("Ludwig") penned down these right-hand derailments as an improviser rather than as a composer, they can be regarded as proposals for an ornamented version of the theme, and if this is the case, a way out of the scrape opens up. 15 We may, after all, be allowed to devise a more modest, but otherwise

<sup>15</sup> The completely mechanical right-hand runs in the second movement of the Sonata Op. 31, No. 1 are even worse, but they are apparently bound up with the very idea of the piece, and therefore we have to live with them. Perhaps this theme and the way it is varied make fun of the stupid piano music composed by most of Beethoven's contemporaries? Indeed, this sonata is perhaps meant to be funny throughout. The main theme of the capricious first movement is unable to settle its left/right-hand co-ordination problems – is someone drunk? – and the third movement starts most abruptly with a closing formula.

equivalent variant of our own. (Ex. 20b) When the second-theme section recurs, a somewhat different stand-in should be invented for the sake of variety. (Ex. 20c)

Turning to the first *Allegro* movement of Beethoven's Sonata Op. 31, No. 3, another speed shock meets our eyes. Four extra bars of right-hand runs are inserted between the second theme and its varied repeat, and the third beat of m. 53 is crowded with extremely fast notes defying execution. (Ex. 21a) Turning to the recapitulation, it features a similar six-bar passage. Some link between the close and the restart of the theme is of course necessary, but what Beethoven proposes, perhaps rather than prescribes, is again excessive – or so it seems.

The basic problem is that Beethoven fails to give a declaration of contents as well as a direction for use. As a consequence of this, many pianists plunge into the runs without any demarcation or any other hint forewarning the listeners of what will happen after the bar-line. And what we hear is a hard-to-understand four-bar (six-bar) bulge in the second-theme section, an anomaly standing out even more due to the sudden cessation of the *Alberti* accompaniment.

The function of the passage is apparently that of a wilfully inserted *cadenza*, or rather that of an *Eingang* after a missing fermata at the B♭-major chord as shown in Ex. 21b. Play it for once, and you will find that the fermata sign provides the declaration of contents while the small notes bring the direction for use.

If the *Eingang* idea is adopted as a model, you have to find a way of arriving at the B♭-major chord making it clear that an effusion of some sort is about to come. As to the many inserted notes, you are free either to distribute them over four (six) bars in a suitable way, or (which is better) to play them as small notes. Turn off the metronome!

Beethoven's runs are exuberant and hence non-repeatable. Since the recapitulation will eventually overbid the exposition, it might be a good idea to invent something more modest when playing the exposition the first time – perhaps a number of small notes that can approximately be accommodated in two bars.

#### **Problems involving developments, codas, and introductions**

Many Classical and also some early Romantic sonata-form movements feature two repeats, a fact that sometimes entails aesthetic problems, one of which will be discussed here. When the developments bring unexpected twists, long excursions into foreign tonal territories, or improvisatory or dramatic passages, it (nowadays?) seems a bad idea to repeat them. What should emerge as unique cannot be played more than once. This may be the reason why the second repeats, getting longer and more adventurous, were eventually dropped.

An obvious but also intricate problem involving a non-repeatable passage is to be found in the development of the first movement of Haydn's C♯-minor Sonata Hob. XVI/36. The long passage mm. 51–64 with its harmonic drift and its uniform design, radically deviating from the rest of the movement, amounts to a kind of quasi-improvised *cadenza*. Consequently, these bars should be left out when the second repeat is played the first time. (Another, straightforward solution is of course to refrain from repeating the second part of the movement.)

Unfortunately, the quasi-*cadenza* is tightly connected to mm. 44–50; if it is taken away, the passage leading up to it must go as well, or be changed so as to lead directly into the recapitulation. This means that the development will become drastically shortened when played the first time, and be presented in full length only when playing the repeat – a quite radical interference, but it does bring some advantages. It makes for an interesting twist in the listening experience when a very succinct development (mm. 34–43) is replaced by a quite expansive one (mm. 34–64), and when the C♯-minor start of the recapitulation (m. 65), heard the first time as "m. 44", is unexpectedly replaced by the same start, now in G♯-minor as in the actual m. 44. (Ex. 22a) Another option is to find up a short transition between m. 50 and m. 65. (Ex. 22b) You can also solve the problem by devising another, much shorter *cadenza* of your own (preferably built on Haydn's) to replace mm. 51–64 when the development is played the first time.

Codas (and other obviously closing material) should of course not be played twice, and yet this sometimes seems to be prescribed when repeat signs are involved. The reason for this anomaly seems to be that Baroque and Classical composers preferred to show how their pieces were to close, but did not always bother to specify how to return to the beginning of the last repeat; apparently they relied on the musicians to make the necessary adjustments.

Consider, for instance, the end of the third movement of Mozart's Sonata K. 282, and compare it with the closing bars of the exposition. (Ex. 23) The last two bars of the movement obviously make up a short, witty coda, which should not be wasted by being used prematurely, by also being played before returning to the development. Thus, when connecting back to m. 40, m. 100 should be changed using m. 39 as a model.

But, it might be objected, in the first movement of this sonata Mozart took care to place the last four coda bars outside the repeat mark, and he could easily have done so also in the finale. The disagreement can be explained by the peculiar form of the first movement: it essential that the recurrence of the main theme is withheld until the coda. 16

Another example of a coda that must be saved until the second presentation of the second part is brought to its end can be found in the E-major Prelude from Bach's *Well-tempered Clavier* II. To avoid the four-bar coda when returning back to repeat the second part, you have to devise a suitable E-major cadence coming to rest on the third beat of m. 50. (Ex. 24)

The first movement of Mozart's Sonata K. 311 presents an interesting problem since it features two corresponding codas: an unexpected idea first turning up after the exposition, then after the recapitulation. 17 The problem is that if you play both repeats as written, the simple but exquisite coda risks being worn-out. (This is not to say that a short four-fold refrain may also have a certain charm.)

You might consider omitting the A-major coda when returning to begin the exposition, using it only when you proceed to the development. (Ex. 25a) This means that the element of surprise is enhanced, and also that the structural *raison d´être* of the A-major coda is clarified – it is immediately imitated when the recapitulation starts. The D-major coda can also be left out when returning to start the development – don't forget to change m. 110 using m. 37 as a model. When it eventually turns up to finish off the movement, the coda is most effective. The two quite unexpected and determined chords in m. 110 bring the listeners' ears on their toes – watching for a weighty, cadenza-like passage, they are instead offered two relaxing bars, a reminiscence of the transition between the exposition and the development. (Ex. 25b)

Turning from codas to introductions, from where should the repeat of the exposition in Chopin's Piano Sonata Op. 35 start? This question has raised a controversy between "traditionalists" and "documentarists". The former hold that after the two long chords in mm. 101–104 you are to return directly to the

<sup>16</sup> For a discussion of the first movement of K. 282, cf. the analytic symposium in *Music Perception* 13(1996), No. 3 and Bengt Edlund, "Keyboard commentaries on K. 282", Chapter 16 in the present volume.

<sup>17</sup> The double-repeat sonata form of this movement is most original in other respects as well: it is impossible (and a waste of time) to establish where the development actually ends and the recapitulation begins, and the main theme is the last constituent to show up in the recapitulation.

86 Loyal disobedience

pulsating B♭-minor accompaniment in mm. 5–8 – this is what (virtually) all printed editions tell us to do – while the latter insist that you should start the music from its very beginning, as Chopin presumably wrote. (cf. Ex. 26a) In short, the fact of the matter is that among the early sources of the sonata the start-the-repeat-from-here dots at the double-bar-sign between m. 4 and m. 5 appear only in the first German edition; yet this way of printing (and playing) has dominated since then. The crucial point of the long story, as told by Anatole Leikin, runs as follows. 18

In Chopin's times there was no established convention to the effect that a start-the-repeat-from-here sign had to be placed between an introduction and the following exposition. The composers as well as the publishers seem to have been confident that the musicians were able to decide when it was appropriate to include or leave out the introduction when playing the repeat. This means that the absence of start-the-repeat-from-here signs in such situations is not conclusive, a fact opening up for later misunderstandings. Turning to the Chopin sonata, the absence of a repeat-from-here sign might either mean that the first four bars are to be included in the repeat, or that they are to be left out, if this is what your musical judgement tells you. In other words, what the German editor (according to his understanding) presumably added between m. 4 and m. 5 was a cautionary start-the-repeat-from-here sign.

So, when it comes to the crunch we are left to use our own, present-day musical discrimination. 19 Before we think that it is a bad idea to include interrupting, slow introductions when repeating expositions, we have to determine whether mm. 1–4 in the Chopin sonata really make up an introduction, how slow this passage is, and to what extent these bars are interrupting. We must also evaluate the two juxtapositions resulting from the two options: the shift between m. 104 and m. 1, and that between m. 104 and m. 5.

When starting the sonata, the first four 2/2 *Grave* bars, dark and dramatic, undeniably sound as an introduction to the frantic *Doppio movimento* music to

<sup>18</sup> Anatole Leikin, "Repeat with Caution. A Dilemma of the First Movement of Chopin's Sonata Op. 35", *Musical Quarterly* 85(2001), 568–582. While Leikin, after reviewing the arguments, finds himself in the "traditionalist" camp, Edward T. Cone and Charles Rosen side with the "documentarists".

<sup>19</sup> Just as we must always do, and as we just did in Mozart's K. 282 Sonata when not playing the short coda before returning to repeat the second part of the finale; cf. Ex. 23. In the best of worlds there would have been a cautionary repeat sign before the last two bars.

follow; on the other hand, the *Grave* passage is quite short. 20 It is not impossible to start the repeat from m. 1, although this may seem to come close to playing two introductions – after the *Grave* passage, mm. 5–8 introduce the accompaniment of the theme.

As to the two juxtapositions, none of them is entirely satisfactory. Traditionalists turning back to m. 5 have to live with a modulation that is less than optimal: mm. 103–104 bring an A♭-major dominant seventh-chord requiring a D♭-major outlet, not a deceptive B♭-minor chord arrived at by means of irregular voice leading. (If Chopin wanted the pianists to return to m. 5, why didn't he provide an impeccable transition?) Starting their repeat from m. 1, the documentarists do land at a non-deceptive left-hand octave D♭/d♭, but the following E♮<sup>1</sup> /E♮ and the right-hand notes bring a C♯-minor chord, leading to another (delayed) harmonic deception. But the listeners have heard this turn of events before, and the falling sixth in the bass actualizes an important interval in the movement's motivic process; cf. mm. 110, 118, and 138–153.

What can be done? It seems that the documentarists must keep strictly to Chopin's tempo indications, i.e. they have to avoid slowing down when starting their repeat from m. 1. This can be done by imagining doubled note-values in the introduction, by inwardly continuing the *Doppio movimento* tempo. Alternatively, they can start the *Grave* tempo during the two long chords in mm. 101–104 by inwardly thinking the introduction in terms of quarter-notes instead of half-notes. Indeed, the excessively long chords mediate between the tempos and may be taken to indicate that the documentarists are right: the repeat should start from m. 1. And when beginning the sonata, it is essential that the *Grave* bars do not sound as an overly brooding introduction. No dragging *ritardando* but immediate *Doppio movimento*! In other words, you have to settle for and keep to a fairly fast *Grave* tempo in mm. 1–4, a tempo allowing you to make a seamless double-speed connection to the hectic music that follows.

Traditionalists who do not want to repeat the introductory four bars might – in addition to clarifying the voice leading by giving emphasis to the right-hand g♭ and c1 in mm. 103–104 – amend the re-modulation by sneaking in a mediating

<sup>20</sup> Very few pianists are likely to consider including the eleven-bar 4/4 *Grave* introduction of Beethoven's *Pathétique* Sonata when repeating the *Allegro-di-molto* exposition . But the Haslinger edition from 1828 in fact lacks the redundant, cautionary begin-therepeat-from-here sign. (But it might be argued that it is not entirely out of the question to start the repeat from the very beginning. The development as well as the coda are introduced by *Grave* passages).

leading-note octave A♮<sup>1</sup> /A♮ in the left hand so as to give the second long chord a last-moment sense of an F-major dominant. (Ex. 26b) In other words, they might do something comparable to Chopin's melodic anticipation paving the deceptive way from the A♭-major seventh-chord to the F♯-minor start of the development. The traditionalists are free to play the *Grave* introduction in a slow, portentous way, turning the *Doppio-movimento* direction into a private, obscure transaction between themselves and the notation. 21

# **Further problems involving repeats**

When approaching the coda in Brahms's Intermezzo Op. 118, No. 1, there is apparently one bar too many since the *seconda-volta* bar makes up an exact replica of m. 28, the bar before the *prima volta*. (Ex. 27a) This quasi-improvisatory Intermezzo does not belong to the kind of music where you expect bars to be exactly repeated, and therefore the duplication emerges not only as redundant but as quite disturbing. Indeed, at first you are prone to think that the *prima volta* should have included m. 28, an unfounded hypothesis presupposing that there is a grave misprint.

To dispose of the problem you might consider to simply skip the duplicated bar when heading for the coda. But it is a fact that this Intermezzo consistently features pairs of bars, and that the problematic *seconda-volta* bar corresponds to the two augmenting bars within the *prima volta*; hence, it is needed to start another pair. In other words, without the duplicated bar the passage, and the piece, would be one bar short. Brahms presumably felt that this "extra" bar had to be there, that m. 28 had to be repeated when heading for the close of the piece. Taking this bar away would disturb the regular periodic make-up of the music – the radical remedy would be worse than the problem to be solved.

Therefore we should rather devise something sensible to do with the duplicating *seconda-volta* bar, something that makes it different from its juxtaposed m. 28 twin. Leaving the notes intact, but changing the phrasing, it can be played both so as to deviate from m. 28 and so as to prepare for the shorter half-bar breath

<sup>21</sup> The discussion between traditionalists and documentarists may be extended to involve further works by Chopin; a discussion that will not be pursued here. Suffice it say that the early mazurkas, waltzes, and polonaises bring examples of all three kinds: introductions that are to be included in the repeat (or *da capo*), introductions that are to be omitted, and cases that are left open; the introductory quality of the introductions varies. Sometimes the repeat of the introductory passage is fully written out as in the Polonaise op. 26, 1.

in m. 29. (Ex. 27b) Whereas Brahms's one-bar *seconda-volta* sign shows that the augmented melodic material of the two-bar *prima volta* is restored to its original size, a virtual two-bar *seconda-volta* notation agrees with the proposed interpretation: the slowing-down *prima-volta* bars are replaced by four urging fragments issuing from the model in m. 28.

Variation II of Beethoven's Diabelli variations Op. 120 obviously lacks the first repeat sign, a fact that cannot but destroy the sense of formal balance. Keen listeners are bound to compare this variation with the preceding one and with the theme, and then with the overwhelming majority of the following variations, comma in which both sections are repeated. And to the pianist the situation is even more disturbing: you set out to erect a truly monumental building, and you have barely started when the architect wants you to wall up one of the windows in the facade. An essential aspect of monumentality is uniformity.

This missing repeat sign evades explanation – otherwise Beethoven did not worry about being stubbornly repetitious in Op. 120 – but in the absence of evidence to the contrary it must be accepted as intentional: the composer does prescribe a non-duplication that upsets the formal layout not only of this variation but of the entire set. Later on in the work, when most listeners are prone to feel safe (or have lost their stamina), there are two further variations, XI and XII, that also lack their first repeats.

Why not let your own better judgement prevail, why not allow yourself to be a pedant and repeat the first part of these variations? Pianists of the world, unite and restore the walled-up windows!

(The first repeat is missing in Variation XXX as well, but a complication is involved. For some late-Beethovenian reason the "start-here" sign of the second repeat appears four bars too late. This means that a normalization operation (if any) requires that a double repeat sign is inserted before the fourth beat of m. 8. Variations XX and XXIX make up exceptional, no-repeat items.)

Apart from Beethoven's Op. 120 and generally speaking, the problem with repeats is that they tend to bee too numerous, some pieces by Schumann, for instance, could do with fewer repeats. Conversely, it also happens that you think that a certain not-to-be-repeated section of a work deserves to be played twice. Depending on the form and character of the piece, it may sometimes be OK to allow yourself the pleasure of repeating such a section. Expert listeners grudgingly accept skipped repeats; why should listeners at large protest against "immediate-encore", bonus repeats?

90 Loyal disobedience

In the first book of Brahms's Paganini variations Op. 35, two closely related *Andante* variations in A major make up a pair which, due to the aggressive brilliance of the majority of the variations, tends to emerge as a *longueur*. This impression can be avoided if the repeats of both variations are omitted in the following way: play the first part of Variation 12 immediately after the first part of Variation 11; the second parts of the two variations are then juxtaposed in the same way. (Ex. 28) This kind of interleaving (used also by Beethoven in Op. 111) gives rise to a "Variation 11/12", in which the more volatile figurations of the "repeats" bring variants of what you have just heard. 22

# **Re-compositions**

Those who have played Schumann's Symphonic Etudes Op. 13 may have wondered why – in contrast to the other comparable variations of the set – Etude V features a second repeat that like the first one ends in E major, thus providing a change of key having nothing to connect to since Etude VI (as usual) starts in C♯ minor. (The final cadence of Etude V would be more appropriate before Etude VII which is set in E major.) The reason for this flaw or peculiarity may presumably be found in the work's extended composition process with several attempts at arranging the sequence of the variations. 23

If you wish to avoid a deviating tonal dead end in Etude V, you have to recompose its last four bars so as to end in C♯ minor, a considerable intervention. But fortunately Schumann has already supplied what you need: there is a sketch of this etude closing in the tonic. (Ex. 29)24 As it were, Schumann posthumously approves of your idea to change the version of Etude V that was (that happened to be?) printed.

This example actualizes a whole category substitutions, of which some raise ethical questions. When musicians for some reason are dissatisfied with a passage,

<sup>22</sup> There are a pair of slow twin variations in Reger's Telemann Variations Op. 134 that may be treated in the same way, thus circumventing the composer's request that the repeats should be ignored throughout. On the other hand, it would be a very bad idea to merge variation 23 and 24 in Brahms's Handel variations Op. 24. These closely related items do not make up a *longueur*; quite to the contrary, a two-stage build-up is needed to prepare for the final variation, releasing the pent-up energy.

<sup>23</sup> For a discussion of how to arrange the 12 etudes/17 variations, cf. Bengt Edlund, "Recycling the Symphonic Etudes", Chapter 5 in the present volume.

<sup>24</sup> This example shows adjustments for various purposes; the sketch is reproduced according to Edition Schott, *Neue Ausgabe sämtlicher Werke*, Serie III:1:3, p. 358.

they sometimes turn to an earlier, discarded (or just abandoned) version of the work that offers a formulation that they can use to replace the one they dislike, or just find less effective. But this is a practice that should be sparingly resorted to. Such substitutions must be occasioned by serious discomforts; more or less collage-like versions of works are not desirable. It also happens that musicians decide to play a non-final version of an entire movement or work.

Prokofiev's fifth piano sonata is published in two versions, the early one Op. 38 and the late one Op. 135. Which of them do you prefer? If you cannot make up your mind, is it advisable to "import" passages from one version to the other?

The slow E-major movement of Schubert's Sonata D. 575 brings a sudden contrast, a five-bar *fortissimo* episode in C major expressive of (say) anger or imperative grandeur. (Ex. 30a) Such passages, suggesting a fierce release of inner tensions, are not rare in Schubert's music. At any rate, some kind of contrast is due at this point in this particular movement.

Yet, and in my opinion, this quasi-orchestral outbreak is not entirely satisfactory – up to m. 26 the movement might have been played by a string quartet. One flaw in the passage is its weak relationship to the preceding section; the recurring bass motif cannot make for a very convincing link since its origin in m. 26 may pass unnoticed. But the root of my dissatisfaction stems from the series of rigid right-hand chords. When looking at old paintings, it is sometimes obvious (or very likely) that some part of the canvas has been left incomplete by the artist, and I get a similar impression when listening to and playing mm. 27– 31; Schubert seems merely to have sketched the right-hand part.

This conjecture can be tested by animating some of the stiff chords, by adding a suitable, already familiar rhythm. (Ex. 30b) This is arguably better, but – if you consider it legitimate to pursue this (still moderate) attack on the integrity of the music – the passage may be provided with a more substantial thematic link. Using m. 1 as a model, the *fortissimo* episode gains considerably in interest. (Ex. 30c) When does loyalty turn into officiousness?

#### **Getting rid of embarrassing passages**

Four bars in Chopin's Polonaise Fantasy Op. 61 are, it seems to me, quite embarrassing. The long, inspired, and impressive *crescendo* passage leading up to the final culmination is suddenly interrupted by a four-bar excursion into foreign keys. (Ex. 31a) The problem with this swaying bravura intrusion is that it brings in a vulgar element in this otherwise most noble work.

Since it seems impossible to save the work by improving these four bars in a way that retains their essence, the only possibility is to simply remove this fit of bad taste, and fortunately this can easily be done. Just proceed directly from the first chord in m. 239 to the second chord in m. 253 where the rising left-hand chromatic motion is continued. (Ex. 31b)25

If you happen to think that the quasi-operatic recitative passage in the Scherzo movement of Schumann's F♯-minor Sonata Op. 11 makes up a strange intrusion, it is equally easy to get rid of it.

There is very strange feature in Liszt's second ballade, a flaw that may perhaps explain why this magnificent piece is so rarely performed. Its first two sections – the ominous and increasingly stormy B-minor *Allegro moderato* issues into an F♯-major *Allegretto* suggesting a state of sensuous tranquillity – are immediately and exactly repeated in B♭ minor/F major. Since the work is called "Ballade", one may assume that there was once a programme, unknown to us nowadays, that motivated the transposed repeat of the initial sections. But this does not wipe out our aesthetic discomfort; we are presumably less prone than 19th-Century listeners to let programmatic considerations excuse formal abnormalities.

But the absence of a programme makes it easier to follow one's impulse to skip the B♭-minor/F-major sections. It may be true that the ensuing *Allegro deciso* episode, starting with the octave a1 /a, follows slightly better after an F-major chord than after an F♯-major one, but considering the gains this cost seems negligible. If you want to make the transition less abrupt, you can keep the F♯ major sound in the pedal until a1 /a brings a touch of F♯ minor. (Ex. 32a)

In order to avoid this juxtaposition you may consider the possibility of sacrificing the entire *Allegro deciso* episode; this would mean landing directly in F♯-minor, the initial key of the following chromatic *agitato* section, associating back to the stormy B-minor start of the ballade. Alternatively, you may skip just the sixteen-bar fanfare-like first part of the *Allegro deciso*, returning to the score at the portentous passage beginning with the *marcato* drone on C♯. It would also be possible to keep merely the two bars immediately preceding the *agitato*; this is perhaps the best solution since they allude to the end of *Allegretto*. (Ex. 32b)

#### **Omitting ill-matching movements**

If you can omit passages and sections, what about leaving out entire movements in cyclic works? Before discarding this idea, one must first find out how "cyclic"

<sup>25</sup> How do you get rid of mm. 23–28 in the first movement of his Sonata Op. 58?

the work in question is. All several-item "opuses" are not cycles, and there may even be "sonatas" that make up aggregates of movements rather than integrated wholes.

The Haydn Sonata Hob. XVI/36 provides a persuasive example when it comes to large-scale loyal disobedience. This three-movement C♯-minor sonata features a second movement in A major, a *Scherzando/Allegro con brio*. This merry Rondo would be very suitable for finishing off an A-major sonata, but it is quite out of place here since it divorces the dramatic *Moderato* movement from its true follow-up, the stoic *Menuet* with its serene Trio. 26 And this is all the more unfortunate since the *Allegro con brio* interferes with the subtle and yet unmistakable thematic kinship between the *Moderato* and the *Menuet*. (Ex. 33)

If you think that Haydn's three-movement sonata is aesthetically flawed, and that a very fine two-movement sonata may be won by just using your red pencil, why not simply leave out its misplaced, unrelated, and arguably less interesting A-major second movement? Haydn might not have disapproved of the new formal configuration; there is a two-movement sonata closing with a minuet, the attractive G-minor Sonata Hob. XVI/44.

In Mendelssohn's Fantasy Op. 28 the melancholic mood and eventually quite stormy improvisatory passages of the initial *In moto Agitato/Andante* movement would have lead perfectly over to the final and very passionate *Presto*. What a pity that all this dark and fiery F♯-minor music is dispersed by an irrelevant and cosy A-major *Allegro con moto* second movement! Why not remove this obstacle for the sake of a more convincing whole? One reason for retaining the interrupting movement is that the second theme of the *Presto* is hinted at in the *Allegro con moto*. But considering the gains this is a negligible loss – a loss that the listeners will not notice since the middle movement is skipped.

For a final example, isn't the fourth, funeral-march movement of Brahms's five-movement sonata Op. 5 an alien and inferior element? No matter why he included it in his sonata, playing the funeral march makes the sonata seem assembled rather than composed.

Looking back at the interventions proposed in Chopin's polonaise, Schumann's sonata, the Liszt ballade, the Haydn sonata, Mendelssohn's fantasy, and Brahms's

<sup>26</sup> In 1780 when publishing the group of six sonatas to which this sonata belongs, Haydn pointed out that the *Scherzando* movement is subjected to a radical make-over and used again as the first movement of the G-major Sonata Hob. XVI/39.

sonata, one might ask whether I have not been disloyal to them, whether I have not behaved like a self-indulgent director staging a classical drama. Even if we were to agree that the flaws in these works are real and quite serious, and that my changes and omissions would have improved them, it may be fair to say that I have not respected their integrity, indeed, that I have not played them, but rather compositions derived from them. Ultimately, then, the question remains whether such interferences can be considered legitimate or not. But we must keep in mind that performances omitting passages, sections, or entire movements were not necessarily condemned in the 18th and 19th Centuries. Apparently, the idea of the composer's prerogatives has changed.

# **Chapter 5 Recycling the Symphonic Etudes**

# **Coherence and unity in cyclic works**

In general we hold that the make-up of cyclic music works is imperatively determined by the score. Much analytic work has therefore been devoted to explain the taken-for-granted coherence and unity of such works, to demonstrate the structural basis for the "fact" that the constituents of cyclic works cannot be exchanged, left out, or re-ordered. And a corresponding attitude is met with when it comes to performance. Since the make-up of cyclic works is considered to be inviolable, they are virtually always played in their entirety and with their constituents in the very order prescribed in the score.

But for several reasons it is hard to defend the belief that cyclic works are always very coherent.

The idea of composing works made up of separate and obligatory constituents that must not be re-ordered – and the concomitant idea of integrating these constituents – is an ambition that has evolved during the last centuries, and the extent to which cyclic unity/coherence has been strived for and attained varies according to genre and composer, as do the means of achieving this goal. It seems that we have often been too inclined to transfer notions of coherence and obligatory sequence from works, in which these ideas are appropriate, to works where thinking in such terms is more or less alien. Thus, in order to assess the actual coherence and unity of a certain cyclic work it is necessary to take account of the period, style, genre, and composer, as well as to analyse the work in question.

But coherence/unity in cyclic works also depends on our ideas as to what a score is, and on our beliefs as regards the complementary competences and duties of the composer and the musician. Current thinking among musicians assigns a strict authority to the score, which is taken to prescribe tiny details of structure (and performance) as well as comprehensive matters of design. 1 But the relationship between composers and musicians has in fact neither been uniform, nor immutable: it has a history disclosing trends of some generality as well as

<sup>1</sup> Cf. Bengt Edlund, "Scores and works of music. Interpretation and identity", and "Sonate, que te fais-je? Towards a theory of interpretation", Chapters 1 and 2 in this volume, also published in *The British Journal of Aesthetics* 36(1996), 367–380, and *The Journal of Aesthetic Education* 31(1997), 23–40, respectively.

plenty of exceptions to go with them. There is evidence that the composers were sometimes quite liberal with respect to the make-up of their cyclic works, and also that the musicians sometimes felt free as regards the selection and sequence of constituents in such works.

Furthermore, the neatly printed scores met with today do not let us suspect the hesitation, or even arbitrariness, that may have been involved when writing down and arranging the music – a suspicion that would militate against our ingrained notion of the unique and carefully calculated masterpiece. But we know that the composers sometimes up to the very last day (and even later) were uncertain or changed their minds about the selection or sequence of the constituents.

As already mentioned, analysts have tried to explain why the constituents of cyclic works are like non-exchangeable links in a chain. The properties considered to make for cyclic coherence and unity have been different, and the analyses have sometimes done justice to the works, sometimes not. It is therefore important to study the criteria of coherence adduced within the musicological and critical tradition. What makes a sequence of constituents optimal, and what kind of affinities between the constituents is required? The traits that have been observed, as well as the properties that have been overlooked, are likely to reveal regulating aesthetic ideas and various aspects of coherence.

But apart from analysis and criticism, and approaching the core of the present study, which criteria of cyclic coherence and unity are relevant for present-day musicians? How would they rank and apply them? And to what extent are these criteria valid from the listeners' point of view?

A study of coherence and unity within cyclic works, and more specifically of why the individual constituents seem to match (or fail to match) each other, is also important since it may influence musical practice. Judging from how musicians usually argue, and from how they tend to compose their programmes, a raised awareness of these matters may lead to artistic emancipation. Cyclic works are most often played from the first page to the last, and this holds even in cases where there is no intended or actual coherence or unity to observe. 2 But

<sup>2</sup> Two examples from the present writer's experience as a concertgoer illustrate this point. A pianist included in her recital ten out of Prokofiev's twenty *Visions fugitives* as well as ten out of Chopin's twenty-four Preludes. No matter the left-out pieces, she played the remaining ones in the sequence to be found in the scores. The point is not whether the played sequences were satisfactory or not, but the fact that the original sequences, although completely ruined, were in a way still respected, and

omitting, adding, and re-ordering constituents may sometimes be legitimate, and if the musicians allowed themselves to use their artistic discernment, they might discover interesting but unexploited configurations in some cyclic works.

Arguably there may sometimes be other worthwhile paths through cyclic works than the one appearing in the score, but the primary purpose of the present investigation is not to persuade musicians into finding these paths and making use of them. The aim is to discuss aspects of cyclic coherence, taking a specific work as a point of departure. It is a fact that the scores of many cyclic works bear implicit, but strongly normative intentions as to the sequence of their constituents, and also that these prescribed sequences often display a quite compelling and demonstrable unity. But this must be weighed against the wish for, indeed the need of, new perspectives on works that may be on the verge of being worn out, and against the artistic merits potentially inherent in alternative configurations.

The idea of re-ordering the sequence within cyclic works is partly a modern one. In the second part of the 20th Century, fragmented "texts" of various kinds were offered, inviting the reader/viewer to take part in the ostentatiously individual experience of composing his/her own story by arranging the sequence of a number of given fragments. And musical "hypertexts" were produced by avant-garde composers already around 1960. In a work like Stockhausen's *Klavierstück XI*, for instance, the pianist is requested to play the printed fragments in any sequence, and all such configurations are considered to be of equal value. Whether the sequences actually turn out to be aesthetically equivalent is an open question, of course, and even less does this Cagean indifference apply to a work like, say, Schumann's Symphonic Etudes. Quite to the contrary: due to the nature of its constituents, the order of the etudes/variations would certainly not be considered as aesthetically inconsequential, and re-ordered cycles would be subjected to keen critical evaluation.

A study of coherence and unity within cyclic works might shed some light on corresponding matters within the other arts; questions of coherence, sequence, and proper matching of constituents turn up in literature and the visual arts as well. The composer's decisions when selecting and ordering pieces to form an "opus" are in some respects akin to those of an author putting together poems for a collection. And composing a programme for a recital can be likened to the curator's work when hanging paintings for an exhibition.

that her approach revealed a refusal to seize the opportunity to devise and take full responsibility for entirely new configurations.

As a further background, three main categories of cyclic works will be briefly presented.

The first category is made up of multi-movement works with titles such as "Sonata" and "Suite": historically grounded, traditional designations arousing specific, and most often satisfied, expectations as regards a certain number of constituents ordered in certain kinds of sequences and featuring certain tempi and characters. Since structural integration lends (or is taken to lend) additional unity and coherence to many or some of these works, it almost never happens that musicians omit movements or change their order.

Sets of variations exhibit structural integration: due to the disguised background presence of a theme or a repeated harmonic sequence, a certain degree of affinity between the constituents is provided for. The order of the variations, on the other hand, is reduced to rely on some external principle of organization, otherwise the make-up of the sets runs the risk of emerging as more or less arbitrary. The printed order of the variations is virtually always observed when playing such works.

Collections of character pieces, finally, sometimes exhibit a cyclic coherence/ unity that is both intentional and structurally provided for, but it seems that many such "opuses" make up fairly loose aggregates. In concert practice, integral performances respecting the sequence to be found in the score are nevertheless quite frequent. But it does happen that individual pieces or selections from the sets are taken out of the published context to be played separately, which agrees better with how many of these collections were once used in their original function as *Hausmusik.*

Neither claiming completeness, nor attempting any ranking in terms of relative importance, some current criteria of coherence/unity will finally be presented.

Cyclic works may have an interesting, satisfactory, or peculiarly ordered tonal layout, and their constituents may be integrated in virtue of some kind of common structural property, such as a recurring theme/motif or harmonic progression. 3 Cyclic works may also emerge as coherent or unified because they exhibit a conspicuous tendency, a consistent development, a commanding curve of tension, a gratifying pattern of culminations, or a psychologically convincing course of events.

<sup>3</sup> In addition to its circle-of-fifths tonal layout, Chopin's Preludes Op. 28 are unified by frequent and, as it seems, intentional reminiscences of *Dies Irae*; cf. Bengt Edlund, "Allusions and affinities. Tracing an ominous motif ", ch. 1 in *Chopin. The Preludes and Beyond*, Frankfurt 2013, Peter Lang Verlag.

Turning to the individual constituents, they may seem to belong together since they display some kind of resemblance or make for a sense of complement or contrast. Other no less important criteria derive from extrinsic characteristics and from evaluations of musical effect. Juxtaposed constituents should preferably be compatible with respect to size and tempo, or feature matching or meaningfully related (or contrasting) emotional characters, or have comparable aesthetic quality – they must not outdo each other.

It emerges from this enumeration of criteria that the notion of cyclic coherence and unity covers a wide variety of phenomena. Some criteria stem from properties that can be analytically demonstrated in the score – thematic relationships, for instance, have often been cited as evidence for cyclic unity – while others involve more subjective things such as mood characteristics or musical content. Or, considered from another point of view: while some criteria of cyclic coherence are matters of large-scale unity, others derive from the individual constituents: their affinity and compatibility, their relationship in terms of meaning, and the effect arising from their juxtaposition.

# **Schumann's "Symphonic etudes/variations"**

Something must be said about the choice of music to be studied, about the work itself and its long process of formation. Why is Schumann's Symphonic Etudes Op. 13 selected when investigating matters of cyclic coherence?

When dealing with cyclic works in modern practice, piano music is advantageous for several general reasons. The three categories of cyclic works presented above are abundantly represented in the Classical and Romantic piano literature, offering a great number of widely known works. Pianists are likely to be interested in these matters since they are most often free to choose their repertory and compose their programmes. Furthermore and as pointed out above, variation works feature some basic affinity between their constituents, allowing you to focus on other aspects of coherence than thematic or structural kinship, such as emotional content, the juxtaposition and sequence of constituents, the emerging form.

Turning specifically to this work, Schumann was apparently at pains to make up his mind as to the configuration of the set. A number of pre-publication sources as well as the early editions disclose that the selection and order of the constituents were open matters for several years. And it may be argued that the coherence within the otherwise magnificent set of twelve etudes, as it was eventually published, is not altogether flawless.

But the most important reason for choosing Schumann's Op. 13 is the fact that many pianists cannot resist the temptation to include some or all of the five further, posthumously published variations into the set of twelve etudes, and that there is no consensus as to where the additional variations should be inserted. A discussion of cyclic coherence and unity with regard to this work and a piecemeal, from-scratch construction of various integral sequences of etudes/ variations may therefore contribute to present-day musical practice.

What follows is not yet another exercise in inserting the five posthumous variations into the published set of twelve etudes making up Op. 13, but a more radical undertaking: a free re-ordering of all seventeen constituents. Since already adding the five extra variations amounts to a fundamental change of the work, it is more interesting and potentially rewarding to construct integral sequences irrespective of the order of the etudes in Schumann's published set – his intentions are disrespected anyway.

How did the Symphonic Etudes come about? The composition process, as far as it can be established, is much more complex and interesting than is generally known. 4

The work was composed in 1834–36, and Schumann's various attempts to settle the order of its constituents can be followed in several sources until the work was finally published in 1837 by T. Haslinger in Vienna under the title *Études symphoniques,* a set consisting of a theme followed by twelve variations, of which the last one is an extended finale in ABABA-form. These constituents will henceforth be called Th, I, II, … XII/F.

In 1852 Schumann revised the work and published a new version with J. Schuberth in Hamburg, the title now being *Études en forme des variations.* This set comprises only ten items; etudes III and IX were omitted, leaving the order of the remaining etudes/variations unchanged.

<sup>4</sup> A thorough investigation of the sources for Op. 13 is to be found in Damien Ehrhardt, "Les *Ètudes symphoniques* de Robert Schumann: projet d'intégration des variations posthumes", *Revue de musicologie* 78(1992), 289–306, or in "Zur Genese der *Symphonische Etüden* von Robert Schumann", *Schumann Studien* 5(1996), 41–54. The research on Op. 13 is brought together in the new scholarly edition of Schumann's works published by Schott Edition, *Neue Ausgabe sämtlicher Werke*, Serie III:1, vol III. More recently, another study of pertinence came to my knowledge, Michael, Seregow, "The Life and Times of Schumann's Symphonic Etudes, opus 13", A lecture-document, presented to the School of Music and Dance of the University of Oregon in partial fulfilment of the requirements for the degree of Doctor of Musical Arts, August 2014. https://scholarsbank.uoregon.edu/xmlui/.../Seregow.

In the third edition, published posthumously in 1861, etudes III and IX were reinstated, and it is in this form, identical with that of the first edition as to the sequence of the variations, that we know the Symphonic Etudes Op. 13 today.

But the source material of the work contains six additional, completed variations, which Schumann decided not to include in the first edition of 1837. Five of these variations were published by Clara Schumann and Johannes Brahms in 1873 with Simrock in Berlin. Later on these five variations were "officially" appended to the main set of etudes in the Breitkopf *Gesamtausgabe*, published 1881–1893. The additional variations will be referred to by the numerals 1– 5 according to the order in which they were first printed. It must be kept in mind that this sequence does not stem from Schumann; the five items are just compiled from a larger set of etudes/variations in Clara Schumann's possession. Johannes Brahms, having access to a different set, decided on the order of the five additional variations to be published, and when doing so he switched the order of the last two variations.

In spite of the great differences as to character, there is a clear affinity between many of the etudes/variations due to the obvious links to the theme – the harmonic layout in many of the etudes/variations is essentially the same, and the initial, falling-triad motif of the theme can often be recognized. In some items, however, the relationship is more remote: this applies to etudes III, VII, VIII, IX, XII/F, and variation 5.

The key of all seventeen etudes/variations is C♯ minor or stays very close to it: VII is set in E major, XI in G♯ minor, and XII/F, 2 and 5 in D♭♯major. Just as the theme, most of the C♯-minor etudes/variations feature a half-way cadence to E major, but in etudes III and V (cf. below) the final cadence also settles on E major, which makes for a (slight) mismatch in terms of key between beginning and end. Etude IV breaks off inconclusively at the dominant; as Op. 13 reads today, the start of etude V, to be played *attacca*, brings the closing C♯-minor tonic.

The main impression of the cycle of twelve etudes from 1837 is one of romantic ardour and pianistic exuberance – only etude XI brings an extended contrast – while the five variations rather represent the dreaming *Eusebius* side of Schumann's mind. Including these variations into the main set of *Florestan* etudes therefore brings a crucial change. Instead of a compact set of twelve mostly powerful and technically demanding etudes, we get a set of seventeen variations with a more varied emotional content. While it may be considered as an advantage, this change in overall character is presumably an important reason why some pianists are reluctant to insert the five variations into the set of etudes.

As far as the two non-posthumous editions are concerned, Schumann gives very few hints of any alternative order of the constituents. The fact that etudes III and IX were simply omitted in the 1852 edition without changing the sequence of the remaining ten items, may be taken to imply that he also accepted the juxtapositions II–IV and VIII–X as valid. As already mentioned, the order of the five variations published in 1873 is gratuitous.

Turning finally to the cyclic coherence of the twelve etudes of Op. 13 as published in the posthumous, restoring edition of 1861, a few critical remarks may be allowed.

Etude III, exhibiting a manifest structural affinity with the other etudes only in its middle section, appears somewhat irrelevant and brings a shift in mood that may seem less convincing. At this point, Schumann's revised edition of 1852, omitting this etude, emerges an amendment with respect to coherence.

As already pointed out, both etudes III and V close in E major, which seems unnecessary (or even disturbing) since etudes IV and VI start in C♯ minor. This means that the standard edition from 1861 provides two potential transitions to E major, but these E-major final cadences are superfluous with respect to the tonal continuity of the set since none of these etudes actually leads to the E-major etude VII. This vehement etude is neither tonally prepared for by the preceding etude, nor psychologically motivated after its stormy C♯-minor close.

But a C♯-minor closing part of etude V is to be found in the source material. 5 It may be argued that the original version of this etude is preferable, and that present-day pianists are free to use its C♯-minor close if they so want. It is shown in Ex. A, where it is provided with endings suitable for returning to the repeat, for immediately attaching to a following etude/variation, and for temporary closure. The C♯-minor close will most often be given priority in the following endeavours to construct integral sequences of etudes/variations.

The emphatic starting qualities – the imitative structure and the solemn, overture-like rhythm – of etude VIII may seem out of place after the previous culmination. The dramatic impetus of etude X should perhaps lead to something else than the sad and harmonically unexpected G♯-minor mood of etude XI, a piece that for all its beauty arrests the cumulative force of the set of etudes. The march-like character of the huge D♭-major finale follows quite harshly after the G♯-minor desolation of etude XI, and it is arguably somewhat too long and repetitious. 6

<sup>5</sup> Cf. *Neue Ausgabe sämtlicher Werke*, Serie III:1, vol. III, p. 358, and Seregow, *op. cit.* p. 53

<sup>6</sup> For this reason, presumably, it happens that some pianists use a variant, discarded by Schumann, when playing the middle A-section.

## **A re-ordering process in three stages**

In what follows an account will be given of how the seventeen etudes/variations were combined, aiming at a number of optimal configurations of the entire material.

Starting from scratch with pairs of constituents, proceeding to groups of constituents, and finally coming to complete cycles, the etudes/variations were systematically combined, and the resulting pairs, groups, and integral cycles were evaluated as to their character and aesthetic merits. This assessment was as neutral as possible when dealing with combinations actually turning up in op. 13 – needless to say it was a delicate task to judge ingrained as well as novel juxtapositions with an unbiased mind.

The procedure was entirely introspective: step-by-step the present author evaluated the various combinations, and particularly the transitions between them, by imagining the music as vividly as possible. Admittedly a most subjective method, but it emerged as the self-evident, natural approach when searching for and applying criteria of cyclic coherence as they appear to a present-day mind. Indeed, the element of introspection made it possible to identify factors making for coherence; the subjectivity of the evaluations did not preclude that supporting arguments of some precision and scope could be adduced.

But these subjective evaluations should be corroborated by other evidence. Some attempts to verify the results will be presented in the final section; meanwhile the validity of the author's assessments has to be weighed against those of the readers.

#### **Evaluation of all possible pairs**

The first stage involves the assessment of the musical plausibility and aesthetic merits of all possible pairs. Ex. 1 shows the beginnings and ends of all eighteen items. In **Table 1** the entries in the column to the very left refer to the first members of such pairs whereas the entries in the uppermost row denote the second members. Turning to the results of the assessments, each row shows the evaluations of pairs starting with a certain constituent, whereas each column gives the evaluations of pairs ending with a certain constituent.

Obviously, some pairs must be excluded for logical or musical reasons. No constituent can succeed or precede itself – hence the diagonal line. And the set must start with the theme and close with the finale, i.e. Th cannot follow after anything, and XII/F cannot precede anything – hence the vertical and horizontal lines. The pair Th–XII/F, immediately finishing off the set, is of course also impossible.

**Table 1**

This means that (18 x 18) – 52 = 272 pairs were to be assessed. The evaluation of each pair is given in Table 1, using the following designations: white = "unsuitable" (or impossible), light grey = "acceptable", dark grey = "good", black = "excellent". Totally 115 "useful" pairs (i.e. 66 excellent and 49 good pairs) were identified, which equals 42 % of the possible combinations. The unsuitable pairs amounted to 112 or 41 % while 45 pairs were judged as acceptable.

It was decided to treat etudes IV and V as independent constituents despite the fact that there is an *attacca* transition between them in Schumann's published versions of the work. They certainly form a most convincing pair, but etude V is not the only item that can follow after etude IV, and in the present investigation it was more important to find out which further pairs these two items might be members of.

In practice, the evaluation was carried out as follows. First each row was treated, imagining as vividly as possible the musical character of the two juxtaposed constituents and the transition between them. (When assessing pairs issuing from etude V, both its E-major and its original C♯-minor final endings were taken into account, selecting for each pair the version of etude V that produced the most convincing transition.) After a few days, the columns were dealt with in the same way. Introspectively, it sometimes seemed to be a subtle difference between "suitable to follow after" and "suitable to precede", but when compared, the assessments were almost always the same. Then, after about half a year, both evaluation procedures were repeated, and the results were compared with those of the previous sets of assessments. Again the agreement was quite substantial.

Thus, the black/dark grey/light grey/white symbols finally entered in Table 1 represent four independent "listenings", weighed together. When evaluating the pairs, musical intuition was given precedence, avoiding rational reasoning as far as possible. Only after the assessment of the pairs was completed, were arguments supporting the evaluations deliberately sought for and taken down.

Before presenting the outcome in some detail a few remarks should be made.

The merits of each pair were judged out of context. This means that the individual constituents making up the pairs were "heard", as far as possible, without any connotations deriving from their positions in the current edition of Op. 13. As a consequence of this disregard of the context within Schumann's set – and excepting Th and XII/F – the Roman and Arabic numerals designating the etudes and variations, respectively, are to be understood as identification labels only.

The ultimate usefulness of a certain pair will of course depend on how it can be incorporated into a larger group of constituents. For instance, while both "IV" and "VI" may follow after "I", "VI" does not make up a convincing continuation after "I" if the latter etude follows immediately after the theme; in such a position, "VI" would mean a too agitated and too closing, premature culmination.

A corresponding qualification also applies to the (°) signs marking transitions that seemed to require, or might profit from, a dividing caesura between the two items. But such matters are of course also context sensitive and depend on broader formal considerations; these signs should therefore be understood as provisional.

When "listening" to the various pairs and particularly to the transitions between them, it could sometimes be observed that the characters and tempos of the juxtaposed items (or just the endings and beginnings joining them) were adjusted in order to make the pairs work – adapting the tempos, levelling out differences, emphasizing contrasts, supporting tendencies, etc. as the case may be. Such adjustments are of course entirely legitimate and make up an important, indeed necessary, element of interpretation when playing cyclic works. 7

The following presentation of the outcome of the first stage of the "recycling" process cannot very well be complete – considering the 272 combinations to be accounted for, the readers are likely to appreciate this compromise. Therefore, and depending on the interest of the coherence criteria involved, only a selection of pairs, preferable ones as well as such that were discarded, will be commented upon.

Excepting VIII, all constituents deemed suitable to follow after the theme (cf. the first row of Table 1) feature a clear affinity to the theme by distinctly alluding to its first falling-triad motif, and all of them bring a sense of increasing motion. These traits are quite appropriate when starting a set of variations whereas a thematically unrelated continuation (III) or a sudden outburst of agitated energy (X) would severely disrupt the evolving form of the work right from its start. The pair Th–VIII emerges as possible since the sturdy dotted rhythms (reminiscent of a French overture) resolutely put an end to the sad mood of the theme, and since an-after-the-theme VIII is quite suitable as a grand introduction to a variety of further constituents.

Turning to context-free evaluation and to items that are suitable to follow after etude "I", the preferred pairs reflect the fact that this etude, beginning quietly in a low register, seems to have a starting function; generally, it leads well to constituents having greater energy and higher tempo. This property also emerges from the constituents that are selected as suitable to precede it (cf. the second column in Table 1): they are all rather slow and are often separated from I by caesuras. Etude I ends openly in a way that is charged with expectations, expectations that are immediately satisfied in the same register by IV, V, VI, IX, and X. The pair I–4 involves an interesting sense of cancelled expansion, and the retained right-hand register makes for a smooth transition.

<sup>7</sup> The characters and tempos of the individual items in any performance of a work like Schumann's Op. 13 are chosen to suit the combinations that actually appear, and they cannot be expected to be optimal when the constituents are used in other combinations. Tempo relationships may be a crucial factor in the interpretation of cyclic works in general, a matter that has given rise to proposals of mathematically regulated tempo differences; cf. for instance David Epstein, "Beyond Orpheus", (Cambridge, Mass. 1979), pp. 75–95.

The powerful sonorities and the emphatic close of etude "II" make for a finishing function, and therefore it may, after a moment of silence, be followed by less passionate constituents, or by the firm etude VIII. Linking II with 2 is quite rewarding since one might hear the right-hand auxiliary theme from II shine through the mists of the improvisatory variation (mm. 11–12). However, if the closing tendency of II is suppressed, it can also be attached precipitately to IV, IX, or VI. In the two former cases the weight of II is dispersed by fasttempo *scherzando* music; when followed by VI, on the other hand, the tension is even more increased. It seems that only the thickening figurations of 1 lead very convincingly to II; variation 3 emerges as too similar. Since it appears that IX must be separated from II by a caesura, this combination indicates that two potentially closing constituents have been juxtaposed.

Etude "III" has a rather weak affinity to the other constituents, and it does not give rise to strong relationships when coupled with them, generally speaking. The items entered as suitable to follow after this light *intermezzo* etude therefore bring new starts – excepting VII, introducing a headlong contrast invited by the preparatory modulation to E major. Turning to preceding constituents, they rely on III for a moderate degree of renewed activity.

Etude "IV" connects convincingly to four constituents, V, VI, IX, and X, which all increase the forward drive started (or mediated) by IV. The transitions to VI and IX are discontinuous in terms of register but may nevertheless be rendered so as to work quite well. If you play the very start of the left hand in X an octave lower, this etude will attach just as seamlessly to IV as does V. The starting and mediating capacities of IV emerge clearly from the column showing constituents suitable to precede it.

Etude "V", beginning as a kind of *scherzando* variant of IV, can lead to more but follow after fewer constituents than IV. If closing in C♯ minor, it may be followed not only by even more energetic etudes, but also by the quiet variation 4. If the modulating E-major final cadence is used, V brings rewarding mood contrasts when coupled with VII and XI. Since V is suitable for mediation, constituents like I, IV, and 3, demanding further agitation, can be placed before it. It should be pointed out that both IV–V and V–IV might be heard as introducing more excitement; the former pair involves an increase as to the density of attacks, the latter a raised tempo in terms of the falling thematic gesture.

While etude "VI" may lead immediately to IX, letting an ardent culmination be followed by a further increase in speed and a retreat into a whirling and capricious fantasy, its resolute ending predisposes it to be followed by constituents either indicating new starts (III, VIII, 1) or bringing contrasts (XI, 2, 4). There are several items that may precede VI, but due to its culminating character and closing function, the usefulness of such combinations when constructing larger groups of items seems limited.

It turns out that it is hard to find constituents matching the ferocious etude "VII", set in E major. But preceded or followed by the deep sadness of XI, it provides an exreme and yet meaningful contrast, and it seems that the initial tonal ambiguity of 1 connects well after the final climax of VII. Consulting the column for VII, it can be seen that it is preferably to be used to provide a stark contrast; the most obvious precursor is III due to its E-major ending.

Having both beginning and concluding function as well as a formal, solemn character, etude "VIII" can be coupled with a majority of the constituents. It may introduce any of the variations, and in virtue of its decisive quality it offers a convincing way out of the lyrical meditations of 2 and 5.

The extremely swift Mendelssohnian *scherzo* etude "IX" may be used in a variety of mediating functions. It can be paired both so as to increase excitement (IV–IX) and so as to dissipate tension (VI–IX); it may be used as a contrasting *intermezzo* (4–IX), and it may lead to culmination (IX–X) as well as to relaxation (IX–4).

Etude "X" with its powerful, *toccata*-like brilliance brings a climax that is difficult to outdo. Rather than resorting to complete contrasts (XI, 2, 4), the exuberance of IX can be used to dissipate the tension. Constituent X is best reached either from items that increase the momentum (IV, V, or IX) or from 3 that finally peaks in massive sonorities and with rhythms similar to those of X.

The G♯-minor distress of the *nocturne* etude "XI" is, it seems, best relieved by the fierce contrast of VII; most other possible resuming items must be held at a certain distance by means of caesuras. XI is preferably to be approached from the vehement VII or from the *scherzando* V (if closing in E major). Initially there is a clear upper-line correspondence between XI and 4 – both melodies are derived from the falling-triad motif of the theme – but it appears that instead of making for a close juxtaposition, this resemblance may emerge as too obvious; these constituents should not form a pair.

Variation "1" is characterized by the *crescendo* tendencies of its two parts, tendencies that can be understated or brought out in order to achieve convincing links. Its rolling figurations may therefore be used either as a relief after some of the most powerful etudes or so as to infuse energy after lyrical items such as the theme and variations 2 and 4. And it leads equally well into the passionate II, as it does (by a sudden change in mood) into the restrained 4.

The improvisatory, dreaming character of variation "2" should, it seems, be brought in as a contrast, and therefore it is preferably preceded by high-tension constituents. For the same reason, it should be succeeded by items introducing firmness and a raised tempo such as the D♭-major XII/F, but it may also with very good effect continue into the broad and calm serenity of 5, also in D♭ major.

Variation "3" embodies a *crescendo* tendency. If started gently, it may for instance follow after Th or 4; if played passionately right from its start, it can match the excitement left in the air after V and IX. Since it ends quite emphatically, it leads well into even more active constituents like IV, V and X – or, by means of a contrast, into the sad mood of 4.

As can be seen from its row, variation "4" has a starting function: the constituents selected to follow after it are characterized by dynamic, ongoing qualities. Its column reveals that it can be used as a contrasting continuation after items exhibiting full drive; it also connects very well to the theme.

Variation "5" has few good links. The ways to approach it seems to be via VIII or (better) via 2, closing in C♯ major. It may be followed by the resuming etude VIII, by variation 2 or by the finale XII/F: the latter combination seems to be the best solution.

The concluding *finale* etude XII/F should be preceded by one of the slow variations 2 or 5, offering good transitions in virtue of the shared key. Depending on the larger context, it may also be introduced by the extremely swift IX; this combination recalls the short, very agitated *Pause* leading to the long finale of *Carnaval*.

#### **Evaluation of privileged groups**

The rationale of the second stage is the observation that the constituents of variation works tend to be clustered into groups. Considering the abundance of "useful" pairs in Table 1, a very great number of groups are possible. For this reason the combination process, while retaining its systematic nature, could not reasonably be pursued in an exhaustive way. In order to make this stage of the recycling process manageable, and also to optimize group coherence, it was decided to leave out of consideration juxtapositions that had been evaluated as merely "acceptable". Furthermore, when assembling the groups, "excellent" pairs were given priority over "good" ones; the latter were used only when no or few better alternatives were available. In practice, then, the groups were mainly constructed on the basis of the black pairs in Table 1. Using such pairs as crystallization points, highly coherent followers as well as forerunners could be found in the rows and columns, respectively.

But assembling groups forwards and backwards from all 66 "excellent" pairs would have entailed many partially overlapping groups and much unnecessary work prior to the third, final stage of the combination process. In order to find groups of vital interest – and to assist in the composition of integral sets – the seventeen items were classified with respect to formal function (beginning, mediating, concluding, culminating, *intermezzo*), character (dramatic, lyric, *scherzando*), emotional content (Florestan, Eusebius), and tonality (home key, unstable key, foreign key). 8

The result of this classification can be seen below. It goes without saying that some items could be placed in more than one category.


Having the goal of constructing integral sets in mind, it was decided to work forwards from the beginning constituents Th, I, IV, VIII, 1, 3, and 4, and to trace forerunners to the concluding constituents VIII and XII/F. Since VII and XI are set in foreign keys, and since III exhibits a strong tendency towards E major, they were brought together to form a group, from which it was worked both backwards and forwards so as to find out how to best reach and then leave these constituents. The groups were not allowed to extend to more than four

<sup>8</sup> This classification emerged as a by-product when studying the properties of the pairs, and several of the categories have already been used when discussing them.

items in backward or forward direction, and less than so when working in both directions, since the offshoots began to be unwieldy and duplications turned up in the chains. Some sequences had to be terminated prematurely since the same constituent occurred twice.

As in the first stage, the groups were evaluated a second time after six months, and again rational arguments were withheld until after this renewed assessment.

A specimen of the second-stage construction of groups is shown in the figure below. The symbols ·, °, and \* for "acceptable", "good", and "excellent", respectively, refer to the coherence and musical potential of the three-member groups serving as germs. Suitable or necessary *caesuras* are marked by the sign '. Less convincing continuations are signified by means of parentheses. When coming to the fourth items of the groups, the branching arrangement is abandoned – the various options are simply lined up – and fourth items are selected only when the third constituent attached in an "excellent" way.

$$\begin{array}{c} \begin{array}{c} \text{I} \ \stackrel{\text{IV\*}}{\longrightarrow} \text{V\*} \ \text{(VI)} \ \text{(IX)} \ \text{X} \\ \text{Y} \ \stackrel{\text{VI}}{\longrightarrow} \text{V\*} \ \text{(VII)} \ \text{(IX)} \ \text{X} \\ \text{Y} \ \stackrel{\text{VI}}{\longrightarrow} \text{IV} \ \text{(V)} \ \text{X} \ \text{4} \\ \text{Th} \\ \text{Th} \\ \text{Y} \ \stackrel{\text{IV\*}}{\longrightarrow} \text{V\*} \ \text{(VI)} \ \text{(IX)} \ \text{X} \\ \text{Y}^\* \\ \text{Y}^\* \\ \text{Y} \end{array} \\ \begin{array}{c} \text{I\*} \ \stackrel{\text{IV\*}}{\longrightarrow} \text{IV} \ \text{(VI)} \ \text{(X)} \\ \text{Y}^\* \\ \text{I\*} \ \stackrel{\text{IV\*}}{\longrightarrow} \text{V\*} \ \text{(VI)} \ \text{(X)} \\ \text{X} \\ \text{Y}^\* \\ \text{Y}^\* \\ \text{Y}^\* \end{array} \end{array}$$

In Fig. 1 are shown branches issuing from Th and then continuing via the first-rate choices I, 1, 3, or 4. All of them lead to gradual expansion and increased activity. The sequences Th–I–VI and Th–4–IX are discarded since VI and IX are too terminating. The groups Th–1–4 and Th–3–4 are unsuitable for a similar reason: the slow, lyric variation 4 prematurely puts an end to the expansion started by 1 and 3. Adding VII after Th–I–VE would mean a too early introduction of a foreign key.

# **Evaluation of integral sets**

In the third and final stage of the recycling process, a number of integral cycles were constructed. First-rate groups culled from the second stage were used as starting points, but as could be expected it often happened that the groups excluded each other. In order to arrive at integral sets without duplicated constituents, accommodations were necessary to bring the pieces of these puzzles to fit.

This bottom/up combination procedure was complemented by top/down considerations securing a convincing large-scale tonal design. The C♯/D♭-major variations 2 and 5 were joined with XII/F to form a final section in the parallel major, and the etudes VII, XI, and III (and/or VE ), were brought together for a midway, foreign-key episode. The remaining constituents were then combined so as to make up forward-driving groups starting from items with beginning function: I, IV, VIII, 1, 3, and 4.

Eventually a fair number of substantially different integral cycles were obtained, complete sets of etudes/variations that could be evaluated and described with regard to their coherence characteristics, **cf. Table 2**. "Good", and "acceptable" juxtapositions are signified by · and °, respectively; all other combinations are "excellent". Caesuras between groups of items are shown by ' signs. Bold letters mark sequences owing some of their coherence to gradual intensification; foreign-key sections are underlined.

The first seven reconfigurations of the entire material pertaining to Op. 13 can be used by performers prepared to take the risk of presenting an entirely new version of the work, a version not restricted by the current order of the twelve etudes.

Cycle (1) prolongs the sad mood of the theme by adding variation 4 after it; then it features two intensification sections, separated by the foreign-key group III–VII–XI preceded by etude VIII.

In cycle (2), bringing three intensification sections rather than two, the thematic resemblance between XI and 4 is exploited when the latter variation is used to gradually resume the activity.


The foreign-key group occupies a later position in cycle (3), in which variations 1 and 3 are used to increase the tension after Th and 4, respectively.

The first intensification section in cycle (4) may initially seem less satisfactory since the sequence Th–I–3–1–II lacks a clear sense of direction. Later on, variation 4 is used as a quiet point of departure for the second section of raised activity.

In cycle (5) variations 3 and 4 start the outer intensification sections, while variation 1 is used to launch an extended central section incorporating the three foreign-key etudes.

Cycle (6) begins with a powerfully expanding group of constituents in fairly moderate tempos; later on etude III starts an intensification section urged by raised tempos. A central section is opened by the sad variation 4 and runs via VE towards culmination, followed by relaxation (VII–XI). In this cycle, all foreign-key constituents are kept together and given a function within the cycle as a whole.

In cycle (7), the slow-pace variations 2 and 5, which may make for a *longueur*, are kept separate. The first three sections are rounded off by relaxing items in slow tempos while the fourth section is driven by a strong impetus leading all the way up to the final march.

It is also possible to construct integral cycles in which the order of the twelve etudes as published in 1837/1861 is respected. Two cycles (8–9) are proposed, in which the five posthumous variations are merely inserted into the main set of the Symphonic Etudes Op. 13.

In cycle (8) variations 3 and 4 are introduced immediately after the theme, prolonging the calm beginning of the work (variation 3 should be played with restraint); etude I issues from variation 4 instead of from the theme. Variations 2 and 5 are used with good effect to provide contrast and relaxation after the culmination of VI; variation 1 attaches well after VIII and leads acceptably to IX. The weak point in this cycle is the unmotivated outburst of etude VII after variation 5.

In cycle (9) the combination 4–1 is inserted between II and III, a solution that disposes of the somewhat problematic transition between these etudes in Op. 13; there is an affinity between 1 and III in terms of the right-hand figuration. Although three slow constituents follow upon each other, the pair 2–5 makes for a good transition to the resuming etude VIII; as a result, there is a contrast to the preceding etude VII that is no less startling, but perhaps more meaningful in terms of mood shift, than the original VII–VIII juxtaposition. Variation 3 is used to start an intensification section after the swift etude IX.

In addition a cycle (10) is proposed, in which the somewhat awkward transition XI–XII/F in Op. 13 is removed by exchanging etude XI for a coherent group made up of the five posthumous variations, a group arranged so as to dilute as much as possible the impression of an accumulation of slow-tempo constituents. Etude XI is placed as a midway slow episode between the energetic outbursts of etudes VI and VII, another less convincing transition in the original cycle. This configuration of the material features three intensification sections, leading up to etudes II, VI, and X, respectively.

Finally, one might also use the insights obtained from the recycling process to change the order of the published set of twelve etudes so as to remove its flaws, if any. The three merely "acceptable" transitions in Schumann's set are avoided in the revised cycle (11). This version of Opus 13 also offers a more orderly tonal design and two quite convincing intensification sections, leading from the theme and towards the finale, respectively.

# **Three attempts at corroboration**

Starting with intuitive assessments and then adducing rational arguments for the preferences, a number of integral cycles have been assembled out of Schumann's constituents – the purpose being to study matters of coherence and unity, and to propose rewarding sequences of all items irrespective of the order of the etudes in Op. 13. With all due respect to intellectual reasoning, this "recycling" has by and large been a subjective process. The value of the undertaking depends on the merits of the proposed cycles, and on whether any deepened insights as regards criteria of cyclic coherence and unity have emerged, i.e. on things that it is up to the readers to decide upon.

Further corroboration is of course desirable, but hard to come by. Three more or less abortive methods were tried in order to confirm the outcome, and they will now be accounted for.

Performances by renowned pianists might support the integral sequences proposed above, provided of course that they include all (or most of) the five supplementary variations. In order to obtain a sample, all available recordings of Schumann's Op. 13 in the collection of the Swedish Broadcasting Corporation were checked. 9

Out of the 28 recordings only 12 included all or some of the five variations. The considerable number of performances comprising only the etudes published by Schumann may of course simply reflect considerations of (LP) space. But it can also be assumed that some of the pianists did not find it appropriate to include the posthumous variations – a choice that may be explained by their respect for the composer's final decision (or, recalling the ten-etude edition of 1852, rather his next-to-final decision), by their wish to keep intact the extrovert and brilliant overall character of Schumann's set, or to avoid arbitrariness as regards the insertion of the five variations into the twelve-etude set.

In all 12 relevant recordings the pianists had simply put in the additional items without altering the order of the main set of etudes. Obviously, none of the pianists felt entitled to re-arrange Schumann's sequence when searching for the best solution. But to the extent that this approach was dictated by respect for

<sup>9</sup> This survey of recordings was undertaken only after my own introspective efforts had been completed. There was thus no risk that the author's assessments were influenced by sequences appearing on the recordings. I want to express my thanks to the Swedish Broadcasting Corporation for their kind assistance. Some further recordings are to be found in Segurow's study (*op. cit*. pp. 32–33), and new performances turn up all the time.

the composer's intentions, it embodies a mistake. The transgression, if any, lies in the choice to include additional material, not in an unprejudiced search for the optimal configuration to be played.

This respect for Schumann's original order is regrettable from a purely methodological point of view. Since the present attempt at recycling the work started from combining *all* possible pairs, no matter the position of the etudes in Op. 13, all that comparisons with the integral recordings can yield is partial corroboration: the sequences proposed here and the recordings may share groups containing the supplementary variations.

Regrettable, again, is the fact that three of the 12 recordings did not include all the posthumous variations, and that another three of them incorporated the five extra variations as a block in the sequence 1–2–3–4–5, i.e. as they once happened to be published in 1873. Furthermore, the corroboration to be gained from comparisons with the recordings was even more reduced by the fact that some of the more or less integral recordings featured pertinent sequences of items that were identical, or nearly so. 10

All this being said, did the recordings give any support for the piecemeal assessments of coherence guiding the "recycling" process and for the evaluations of the integral cycles proposed? Considering that 57 % of the relevant pairs in the recordings were "useful" whereas only 24 % were deemed "unsuitable" according to Table 1, it seems that the pianists preferred "useful" transitions to "unsuitable" ones. As to specific groups turning up in the recordings as well as in the proposed cycles, quite a few correspondences could be noted, indicating that there is some basic consensus as regards preferable juxtapositions. Some of the pianists inserted variations between VI and VII, and between VII and VIII, perhaps attempting to avoid the less convincing transitions associated with etude VII in Schumann's Op. 13.

The second way of validating the results is of course to ask others to assess the proposed combinations/cycles. Evaluations of the order of items in cyclic works may be undertaken in various ways, but the procedure used (or rather the procedure planned to be used, cf. below) deviates radically from the method employed in a study by Gotlieb and Konecni, an experimental investigation

<sup>10</sup> It falls outside the scope of this study, but it seems likely that there are traditions as to how Schumann's work should be arranged if you want to include the supplementary variations.

making up a glaring contrast to the present approach, and a study that deserves to be discussed and dismissed. 11

Determined to test whether the claims made by musicologists, music critics, and musicians with regard to coherence within cyclic works have any perceptual relevance for listeners, Gotlieb and Konecni in one experiment shuffled the sequence of Bach's Goldberg Variations, a work renowned for its intricate compositional architecture. Then they played Bach's composition and two shuffled versions of the work to a number of undergraduate students, asking them to rate their reactions on 200 mm bipolar scales reflecting 15 value dimensions.

The dimensions were: clear-crisp/not clear-crisp, ugly/beautiful, wish to own/do not wish to own, pleasing/not pleasing, simple/complex, cold/warm, exciting/not exciting, spontaneous/not spontaneous, weak/strong, interesting/not interesting, orderly/disorderly, slow/fast, emotional/not emotional, surprising/not surprising, slightly/highly differentiated. No surprise, then, that the researchers' sceptical attitude towards the coherence criteria adduced by experts was overwhelmingly confirmed by the outcome. Only one significant effect was found: Bach's work turned out to be somewhat colder than the two shuffled versions of it.

It may be argued that the experimental design was inappropriate. Excepting orderly/disorderly and perhaps surprising/not surprising, the dimensions used in the test cannot be considered relevant when evaluating the order of constituents in cyclic works. Furthermore, the huge format of the 32-item Goldberg Variations and the many transitions involved made the listening task very complex – and the students' overall [!] reactions correspondingly opaque.

If, for instance, no significant effect was found in the "exciting" dimension, this does not entail that differences in this respect, differences worthy of critical attention, did not actually occur. Apart from the fact that some of the variations are exciting, some juxtapositions in the 32-item sequences may have been quite exciting while some others seemed less so. And if one of all these juxtapositions happened to be, say, very surprising, would that singular experience, crucial as it may be for the quality of this particular combination, be reported as a property of the entire set? As to the orderly/disorderly dimension, it involves a listening task that by far exceeded most of the undergraduates' capacity.

The gist of these objections still applies when the listening task in another of Gotlieb and Konecni's experiments was reduced to "triplets" of variations.

<sup>11</sup> Heidi Gotlieb & Vladimir J. Konecni, "The Effects of Instrumentation, Playing Style, and Structure in the Goldberg Variations by Johann Sebastian Bach", *Music Perception* 3(1985), 87–102.

As already pointed out, most of the semantic dimensions (and the corresponding listening responses) do not emerge as very apt if the purpose of the investigations was to really to find out how cyclic works are perceived and enjoyed. (Perhaps they were just included to hide the aim of the test?) Frankly speaking, it seems that it was Gotlieb and Konecni's primary purpose to deride various music experts. But the irrelevant dimensions and the insensitivity of their approach do not preclude that the experiments may have convinced uncritical and flat-bottomed minds, adverse to analytical and aesthetic sophistication.

On a more fundamental level, one might question the idea of choosing lay listeners to determine the value and justification of analytical and aesthetic observations. 12 While it may be admitted that some coherence criteria favoured by experts are indeed esoteric beyond apprehension (and even credibility), it must be accepted that some coherence criteria are legitimately reserved for the connoisseurs, as well as be acknowledged that some others correspond to properties that are there to be heard by almost anyone. There is arguably a domain between elitism and populism, an interesting domain of musical perception and reasoning that should be studied with respect and suitable methods, instead of being held up to ridicule by means of gross experimental procedures.

Whereas the present writer thinks that there may often be several worthwhile ways to arrange the constituents of cyclic works such as sets of variations, it seems evident that these alternative configurations are bound to be different in non-trivial ways. Some of these re-arrangements may certainly be of inferior value, and this is likely to be both a readily accessible phenomenal fact and a fact for which rational explanations can be found. 13

<sup>12</sup> When did *The Journal of Cosmology* evaluate the string theory, using 200 mm bipolar scales (pleasing/not pleasing, slow/fast, etc.) and engaging undergraduates as subjects? The relationship between listening reactions and analytical observations of various kinds is quite complex; some observations lay explicit claims as regards listening competence while others do not, and what about the implicit claims? Cf. Nicholas Cook, "The Perception of Large-Scale Tonal Closure", *Music Perception* 5(1987), 197–206 and Bengt Edlund, "Tonics and Returns. A Modest Investigation", ch. 8 in *Questioning Schenkerism*, Frankfurt 2015, Peter Lang Verlag, both dealing with the aural support (if any) of Schenkerian theory. Another, most instructive example is brought by a heated debate on the merits of twelve-tone composition; cf. Hellmut Federhofer & Albert Wellek, *"Tonale und dodekaphonische Musik im expermentellen Vergleich", Die Musikforschung* 24(1971), 261–276, Carl Dahlhaus, *"Ist die Zwölftontechnik 'illusorisch'?", Die Musikforschung* 24(1971), 437–440, and several further contributions.

<sup>13</sup> Imagine for instance that chance came up with the following shuffled sequence of Schumann's Op. 13: Th–X–VI–IV–I–VIII–II–V–3–IX–4–1–III–2–XI–5–VII–XII/F,

Avoiding primitive positivism, and yet trying to collect opinions from others – relevant, informative opinions – a systematic validation procedure using highly competent pianists as informants was planned and started. But unfortunately this attempt at corroboration had to be cancelled. There was, in spite of all recruiting efforts, an almost total lack of pianists willing to participate. Some of them had no time to spare, others did not approve of the idea of interfering with Schumann's work. 14 Another attitude was that the undertaking as such might be worthwhile, but that the actual assessment task was very hard, even awkward. 15

It is true that one aspect of great musicianship is the ability to discover and express the meaning of transitions between constituents in cyclic works, even when the juxtapositions at first may seem problematic. The methodological problem is, to put the matter succinctly, that too much understanding makes for too many acceptable combinations, for a too wide grey area between good and bad juxtapositions. When assessing the pairs, groups, and finally the complete cycles, I did not just have to be on guard against preconceived ideas, against disfavouring sequences other than Schumann's, I had also to avoid tendencies to accept sequences on second thoughts. It was necessary not to mix up "feeling a sense of coherence" and "having got used to the juxtaposition".

Thirdly and finally, there is one person who should be asked about his evaluation of my attempts at recycling the Symphonic etudes/variations. One cannot leave Schumann's ideas on how to arrange "Opus 13" out of account. As Damien Erhardt and then Michael Segurow have accounted for, Schumann "proposed"

i.e. with an aesthetic disaster – a bad layout in terms of tempo and tonality, and a haphazard sequence as to mood and character. Is this version of the work, or just an unfortunate "triplet" like 4–IX–3, cold or warm? And how crisp is it?

<sup>14</sup> Jörg Demus declared himself to be quite content with the sequence of twelve etudes in Op. 13, and gave reasons for this. Consequently, he has refrained from any attempt to include the five supplementary variations into his performances, preferring to play these items, preceded by the theme, as a separate minor work – as he does in his complete recording of Schumann's piano music.

<sup>15</sup> In a letter, and then in a valuable discussion, Vladimir Ashkenazy described the basic difficulties inherent in judging coherence, and especially when well-known masterworks are concerned. "Apart from some obviously incompatible sequences, the possibilities of putting together two or three or more variations are enormous. And if you ask why, then the answer is that very often it is because one almost begins to find the sense, reason and logic in lots of different sequences. […] Sometimes I think that if Schumann had presented to us his cycles in different sequences we would find justifications for them just as we do now."

several alternative sets during the extended composition process, and these preliminary sequences should be studied. 16

But unfortunately the usefulness of the pre-publication cycles when it comes to corroboration is diminished by the fact that these preliminary sets do not comprise all 17 etudes/variations. The five posthumous variations were not always selected to be constituents of the work he was groping for, and one unpublished variation was deleted only at a quite late stage. 17 Even more important is the dual fact that six of the etudes (III, VI, VII, VIII, IX, and XI) appear only in the *Stichvorlage* for the 1837 edition, and that the other etudes occur in various positions in the sets.

According to Ehrhardt's analysis, Schumann's various pre-publication sequences can be regarded as a process during which the composer replaced his original layout, based on right/left-hand symmetry as regards the presentation of the initial thematic idea, by a design more and more predicated on cumulative growth of tension. This tendency towards increasing excitement comes fully to the fore in the *Stichvorlage*, in which the five variations are left out and six new etudes are added.

Schumann's re-arrangements are largely a matter of moving a few individual constituents. The two most notable differences between these early sources and the published sequence of twelve etudes are that etudes IV and V were not originally associated with each other, and that variation 5 was strongly connected to etude X – in two of the discarded sets it even forms the middle section in a compound ABA-form.

Thus, the preliminary versions of Op. 13 do not give much ground for conclusions. The late "Wien" cycle, presumably reflecting an ambition to arrange the constituents so as to achieve a sense of increasing excitement, features many "excellent" or "useful" links, a fact that may perhaps be taken as a posthumous approval of some of the evaluations upon which the proposed sets (1)–(11) are based.

Appealing to Schumann's preliminary arrangements of the material for Op. 13 involves a paradox in as far as a main point in the present "recycling" endeavours was to keep the composer out. But considering the difficulties in procuring external corroboration and the scarcity of (explicit or implicit) evaluations from present-day musical minds, this corroboration from a crucially important mind of the past should not be left out of account.

<sup>16</sup> Erhardt (*op. cit*.) and Segurow (*op. cit*, pp. 32–36).

<sup>17</sup> The sketches also feature two further, incomplete variations.

# **Chapter 6 A comprehensive approach to musical idiomatic**

#### **Introduction**

The analysis and evaluation of works of art are practically always undertaken from the perspective of those who enjoy them – the reader, the observer, the listener. But at least when it comes to music this view must be revised.

Few people are likely to deny that music is the art of sounds, but this commonplace is far from the only and whole truth. For prior to, or following upon, its presentation in sound, music may be recorded by means of written symbols. The notation makes it possible to study the music synoptically, but it also means that the music is deprived of time, of its irreversible driving force, its very medium of existence. The visual approach to music involves transcending its temporality – being essentially something that moves, music becomes a static object when seen in print. There are readers who are capable of restoring the motion, but it must be kept in mind that reading remains a potentially deceptive way of enjoying and studying music – you can see things that you cannot hear, and *vice versa*.

Some people are reluctant to attach importance to properties that do not readily, or at all, present themselves to the listener. Others take up the opposite attitude, which means that music analysis may become an affair between the notation and the reader, that analysis turns into an activity that pays too little attention to music as an aural art.

But music is also accessible along still another route. Although we tend to forget this basic fact in our era of passive music consumption, music is also there to be enjoyed by being played and sung. Performing music includes listening to it, of course, but the core of making music is that you control the process, and that you feel the music with and within your body. The latter fact implies that music also speaks to us through the sense of touch and particularly that of proprioception, i.e. the often-neglected inward sense informing you about the positions and movements of your joints and muscles. The whole truth, then, is that music is not only an art that is heard; it is also seen and felt.

Unfortunately, the fact that music is felt has neither been duly acknowledged in music analysis, nor in music aesthetics. As a consequence of the dissociation of music into a relatively exclusive performing art and a common listening art, essential aspects of musical apprehension have been relegated out of consideration. It is a fact that performances are normally needed in order for music to be accessible, but this does not reduce them into just a means to make listening, the "real" enjoyment of music, possible. Venturing a generalization across music history, composers write just as much for the musicians as for the listeners.

The reason for the neglect of the musician's perspective may be that the proprioceptive aspects of music making emerge as irrelevant for, and most often are unknown to, the listeners. Yet, if you want to have a comprehensive understanding of music, the musician's point of view should be the one to prefer: the musician sees, acts, and listens. This is not to imply that the musician's aural enjoyment exceeds or can be placed on a par with that of a person who listens attentively. The musician's deep involvement in the music when directing its course and the complexity of the task of execution make great demands on his/her concentration, and may therefore entail a certain lack of distance to the music as well as a less penetrating experience of it as an aural phenomenon.

Nevertheless, those who enjoy music's bodily sensations are the musicians, and it is a pity that they are seldom consulted when it comes to academic matters, such as analysis and aesthetics. 1

In order to avoid undue simplification, it should be pointed out that devoted music making also includes important non-proprioceptive aspects of experience that likewise are closed to the listeners: the awareness of having chosen between a variety of interpretative options, the strong manifestation of will leading over into effective action immediately reflected in sound, and the feeling of satisfaction when the music is mastered (or the feeling of disappointment when things did not come off as intended). "Subjective" matters, certainly, but accessible to a phenomenological analysis and possible to make intelligible in a rational discourse.

Of core interest in the present context are the proprioceptive sensations that are associated with music making. In as far as music analysis aspires to account for all aspects of structure and meaning in music, and to the extent that performing music is acknowledged as an activity on a par with listening to it, the experience of a piece of music through the body actually producing it must be included as a legitimate and essential source of knowledge. Taking account of proprioceptive matters is indispensable if one wants to gain a comprehensive

<sup>1</sup> Painting does not belong to the performing arts in current sense, and yet paintings are painted. Very few people beyond the artists take an interest in, and are able to appreciate, the delights associated with, say, working with the brush on the canvas.

insight into musical structure, understand interpretation, or teach excellence in performance.

The proprioceptive sensations are of course crucially dependent on the instrument, on its sound and construction, and on the actions required to make it sound – we will shortly touch upon these matters in the final section of this essay. But the proprioceptive qualities of a certain passage also depend on the particular manner of execution that for some reason or other is chosen by the musician. The complex relationships between musical structure, mode of execution, and proprioceptive sensations are in the focus of what follows. In other words, we will discuss matters of idiomatic and interpretation.

# **Idiomatic – a multifarious concept**

In music, the term "idiomatic" has acquired meanings reaching beyond what this word stands for in linguistics. In fact, musical "idiomatic" has not very much to do with "idiom" in language: in music, we most often do not refer to a set of style characteristics, but to a composition's properties with regard to playability. An idiomatic passage, given a certain instrument, is a passage that suits the hands (feet, tongue, etc.) well from a technical point of view, and this definition implies that there are other passages lacking this quality and still others that are really awkward to play. 2 Needless to say, convenience of execution is a good thing, and therefore "idiomatic" is positively charged.

Musicology has of course not failed to take account of this basic sense of idiomatic, being an important factor in the development of musical styles, but the understanding of idiomatic, and its many ramifications, has often been insufficient. Music historians have mostly confined themselves to identifying convenient figurations, prevalent within certain repertories and more or less adapted to the specific nature of various instruments, and to studying the origin, introduction, diversification, and eventual demise of such stylistic clichés. Musicians, on the other hand, have a comprehensive and very intimate knowledge of idiomatic, but unfortunately they have seldom and only fragmentarily formulated their insights in writing.

We have thus learnt how the easy-to-play *Alberti* basses (cf. Ex. 1, showing the beginning of the second movement of Mozart's C-major Sonata K. 545) were introduced in the keyboard music during the 18th century, and how this

<sup>2</sup> A number of connotations associated with the concept of 'idiomatic' will soon be presented.

accompaniment figuration – bridging the gap between the melody and the slow shifts of harmony within a simple homophonic texture, and giving an illusion of fluent motion – was eventually abandoned as too conventional during the 19th century.

We also know how Liszt (and all his imitators) excelled in passages featuring octaves alternating between the hands (cf. Ex. 2a from *Mazeppa*, *Etude transcendentale* no. 4). In comparison with the corresponding, massive but relatively slow two-hand octave passages, this clever invention3 doubles the frequency of attacks without any additional manual hurry (cf. Ex. 2b).

Idiomatic in this established sense bears an interesting relationship to virtuosity. 4 Thomas Carson Mark has analysed the concept 'work of virtuosity', and established that in the musical domain we normally refer to pieces requiring great skill, "virtuosity", in order to be performed adequately. 5 He advances three criteria: a work of virtuosity is actually very difficult to play; this difficulty is an obvious feature of the music (although in some cases the difficulty is only apparent to those knowing how to play the instrument); the composition is such that the listeners' attention is attracted to the performance – indeed, the virtuosity to be displayed by the musician more or less emerges as the "topic" of the music. Mark also points out that a truly virtuoso performance is characterized by the impression that the musician in question seems to have no problems whatsoever to play the music.

The latter observation suggests that works of virtuosity (or at least the successful among them) tend to be highly idiomatic. If a very difficult piece is downright awkward to play, it will hardly give rise to performances displaying supreme mastery; to the extent that such a piece were to be much played at all, its

<sup>3</sup> Or adaptation rather, since Liszt's model was apparently the playing of chromatic scales on the cimbalom, where hand alternation is not just convenient, but necessary, cf. Ex. 2c. There is a snag, however. Due to the design of the keyboard, the hands must sometimes strike white keys, sometimes black keys when playing chromatic octave scales *à la manière de Liszt,* and the shifts of hand position disrupt the motion pattern.

<sup>4</sup> One should not jump into the conclusion that idiomatic, whether conceived of in the current, basic sense just presented or understood in a more subtle and comprehensive way, is something that is only met with in so-called "virtuoso" pieces.

<sup>5</sup> Thomas Carson Mark, "On Works of Virtuosity", *The Journal of Philosophy* 77(1980), 28–45. There are also, according to Mark's analysis, works in which the composer, not the musician, stands out as the virtuoso – this is the case, for instance, in J. S. Bach's *Ein musikalisches Opfer* and *Die Kunst der Fuge*, works excelling in esoteric kinds of canons and fugues, difficult for a composer to master.

difficulty would all too often be obvious in an undesirable way. Therefore Mark's first criterion should be slightly modified. Works of virtuosity are indeed very difficult to play, but due to their idiomatic qualities they are often, and fortunately, not quite as difficult as they sound; this explains why works of virtuosity are so favoured among musicians. Nor are they always difficult in the way that the layman thinks.

It can readily be understood why good idiomatic (no matter whether the works require virtuosity or not) is a positive value: pieces that are convenient to play allow of fluent, relaxed, elegant, and technically perfect performances. And this is an asset that is not reserved just for the musicians. Without fully knowing the idiomatic cause of their delight, lay listeners will profit from the fact that the music is not obscured by any flaws in the execution. Good/bad idiomatic should therefore count as an aesthetic bonus/defect also from the listener's point of view.

In this context and following Mark, it is not out of place to take up the cudgels for the often-depreciated virtuosity. Why shouldn't it be legitimate for a music work to demand extreme technical skill, and to have the virtuosity required to play it as its topic by directing the listeners' attention to the performance? There is undeniably a pleasure in listening to technically exceptional performances, and this is a quality that can be appreciated also by non-experts. Furthermore, one must object to the current view that the virtuoso potential in a music work precludes other aesthetic values. The fact that the listener is made to notice the qualities of the playing does not imply that the music cannot exhibit other, concurrent values. 6

But if we want to fully understand what idiomatic in music implies, the scope must be widened to include much more than what is technically practicable or convenient to play or sing.

Idiomatic passages are often associated with a distinct perception of muscular or kinetic delight. This means that some part of the value of a music work derives from another sense modality than that of hearing, and it also implies that some of the attraction of a music work is accessible only to the musician playing it.

But good idiomatic reveals itself in further ways. Irrespective of the "proprioceptive delight" it gives rise to, a passage is more idiomatic if it allows the musician to feel confident, if he/she has a fair chance of playing it perfectly. 7

<sup>6</sup> Ravel's piano works, for instance, provide a number of examples proving this point.

<sup>7</sup> Intermingled with the impressive and most often quite rewarding pianism in Brahms's Second Piano Concerto, there are some passages that entail risks, no matter how much the pianist has practiced to master them. (Alfred Brendel aptly speaks of "difficulties

All musicians have also made the observation that technically attractive structures are easier to learn, and also easier to memorize. Finally, idiomatic writing means that the music gets a certain kind of durability. Due to its playing qualities, idiomatic music endures sustained and intensive practising as well as many performances; the zest is continually renewed, and even when the music is extremely taxing to play, practising remains paradoxically restful. 8

These observations lead over to another sense of good idiomatic, to another domain where it is manifested. The relative ease with which idiomatic music can be mastered and memorized may in many cases be explained by the fact that the playing motions can be arranged in sequences that meet the demands of a good mental representation. But this should not be understood in a simple, mechanical way. The point is not just that the playing motions as such (or rather the motor commands corresponding to them) form patterns that exhibit a good mental economy with all that this implies in terms of suitable chunks, repetitions, symmetries, and co-ordination. In addition, the mental representation of the motions often turns out to be well adapted to the musical structure, bringing out or revealing meaningful and subtle relationships within it.

This means that idiomatic has much more to do with musical interpretation than is generally assumed. In genuinely idiomatic music, the interpretation and the phenomenology of playing seem to be intimately linked together. Inherent in the playing motions is a "proprioceptive meaning" that metaphorically corresponds to or complements the musical expression. From a practical point of view good idiomatic means that the interpretation and the playing motions are mutually dependent, allowing the musician to work in both directions. The musical interpretation suggests a certain way of execution as artistically suitable; in other cases, it is the optimal way of execution that suggests the interpretation. Indeed, if the music is written by an idiomatically skilled composer, the interpretation prompted by the idiomatic is likely to be a very good one, even "the correct", intended one.

bordering on the perverse".) The last two pages of the second movement in Schumann's C-major Fantasy is another example.

<sup>8</sup> The latter property may explain why certain pieces are discouraging. While, for instance, Beethoven's "Hammerclavier" Sonata as a whole attracts pianists both for aesthetic reasons and for reasons of prestige – it is after all the Mount Everest of Classical piano music – few of them are likely to look forward to the hardships involved in learning to play its huge and almost impractical fugue and to mustering the concentration needed to give a flawless presentation of it in a recital.

But it would be a gross simplification to conclude that good idiomatic allows, say, a pianist to just rely on his/her fingers. There may be other interpretative options worthy of consideration than the one agreeing with the most obvious technical solution, and one and the same structure may keep several satisfactory ways of execution. It should also be pointed out that some passages are latently idiomatic, i.e. they seem technically awkward, and even difficult to grasp musically, until the technical problem has been solved, until it turns out that the passages are actually quite idiomatic.

Such problems, challenging the musicians' technical creativity, illustrate a general methodological difficulty in the study of idiomatic. Just as the aural musical structure is not unequivocally specified in the score,9 one cannot with certainty establish how a passage is to be played by studying the notation: musicians are often able to find several, quite distinct options. They know how to ingeniously use their means (be it fingerings, bowings, changes of register, patterns of breathing, etc.) in order to devise technically as well as musically suitable solutions. Intimate knowledge of the instrument in question and its playing techniques is therefore required if one wants to penetrate into musical idiomatic. And yet it is necessary that the musicians' way of apprehending musical structure is given its legitimate role in analysis. Their perspective is just as important for a full understanding of the music as that of the listeners, and it may be radically different. 10

A further, essential aspect remains to be mentioned before we have arrived at a comprehensive understanding of idiomatic. It seems that idiomatic passages often sound quite well. Their texture and register give rise to beautiful sonorities, or to a sound that is characteristic of the instrument. Idiomatic pieces never expose the imperfections associated with a certain instrument or its way of working. Good idiomatic entails that the technical and musical resources of the instrument are used in a varied and expansive way. Excellent idiomatic is not only rewarding for the musician (and indirectly for the listener), it also flatters the instrument.

<sup>9</sup> Cf. Bengt Edlund, "*Sonate, que te fais-je?* Towards a theory of interpretation", Chapter 2 in this volume and also in *The Journal of Aesthetic Education* 31(1997/1), 23–40, and "On scores and works of music. Interpretation and identity", Chapter 1 in this book and also in *The British Journal of Aesthetics* 36(1996/4), 367–380.

<sup>10</sup> Consider, for instance, a rhythmic sequence played on a drum: the auditory impression is that of a single series of attacks whereas for the percussionist the music is also made up of two interlocking sequences, one for the right hand and one for the left.

Enough has now been said to prepare for the analytical part of this study. In what follows we will discuss a number of examples from the piano literature, selected so as to substantiate the comprehensive notion of idiomatic that has been expounded.

## **Schumann's** *Albumblatt* **and Brahms's variation**

Schumann has composed the figuration shown in Ex. 3a (*Albumblatt* Op. 99, No. 5), and it occurs again, slightly changed, in a work by Brahms; Ex. 3b (Var. IX from his Schumann Variations op. 9). 11 Both pieces are delightful to play, you never get tired of feeling your hands romp around like kittens. The technically exquisite design with its alternation between the hands allows of a very fast tempo and makes for a clear separation of three layers: the swift undulation, the poignant, syncopated motifs in the middle register, and the fluffy thuds of the bass. 12 What you hear is piano music *à trois mains,* but only two hands are busy playing fragments that have but little meaning as such. The amazed listener gets an impression of a *perpetuum mobile,* but the stream of sixteenth-notes is divided; the right hand is inactive at the end of each bar whereas the left has nothing to play at the beginning – a facilitating, relaxing distribution of the material.

These two highly idiomatic, twin pieces are certainly works of virtuosity according to Mark's criteria since they give rise to performances of hushed and yet sparkling exactness, but they are not as difficult to play as the astounded listener believes. Needless to say, in order to exactly deliver the fragments succeeding each other at a rapid pace, an unfaltering concentration is required, and so is a perfect finger technique ensuring that every note is played with minimal motion and maximal precision.

It should also be pointed out how well the manual layout assists the melody. If you play with the same fingering throughout, which makes for mental economy – it seems to be a good idea to let the thumb strike all melody notes – the melody will be brought out by means of new hand positions in a slightly percussive way.

<sup>11</sup> This similarity is not a case of plagiarism, but a quasi-citation and an ingenious *hommage*. The theme of Brahms's set of variations is the preceding piece in Schumann's collection.

<sup>12</sup> Schumann's motif comprises three notes, Brahms's only two, but the difference is minimal: the first note in each bar is likely to get some emphasis anyway, and it cannot actually be held with the thumb as long as the eighth-note demands.

For this reason, the fingerings recommended for m. 1, 3, and 4 in the Peters edition of Brahms's variation (Ex. 3b) are not only inconvenient but also counterproductive. A delaying sub-positioning of the thumb is introduced within the first triplet group; the index finger has to bring out the syncopated melody note without any facilitating shift of hand position; the remaining five right-hand notes must be played with a stretched hand; the ring-finger has to strike two different keys in rapid succession; and in mm. 1 and 3 the right-hand thumb must play the last note under the index finger remaining on its black key. It is far better to sacrifice the water-tight *legato* between the melody notes and to disregard the outdated rule against thumbs on black keys; the unavoidable gap in the right-hand will hardly be noticeable in a very fast tempo. In any case, it cannot be denied that, due to the leap to the thumb in these three bars, Brahms's variation is technically somewhat awkward and risky to play.

Schumann's piece (Ex. 3a) has a minor idiomatic flaw as well. In m. 3 and 4, the second melody note turns up again as the last note in the right hand. This means that the melody note must be left prematurely, ruining the manual sense of *legato* down to the third melody note played by the left hand.

Generally, in Schumann as well as in Brahms, the pedal should be depressed at the first and the third pulses in each bar and be released at the second and fourth. This pattern has several advantages. In addition to concealing undesirable gaps in the melody, the duple metre and the regular harmonic shifts are brought out while the weak second and fourth pulses are given a lighter, but still sufficiently rich, sonority by means of the sustained melody notes. And the release of the pedal at the second eighth-note pulse lends a bouncing quality to the left-hand notes.

It is also of interest to discuss the passage leading up to the culmination in Schumann's piece; cf. Ex. 3c. As can readily be seen, the three-note motif is preserved in m. 23, whereas the sixteenth-note figuration is changed: the rising six notes are immediately sequenced on the second beat, introducing a radical deviation from the incessantly repeated pattern up to this point. The order is restored in m. 24 – the undulating twelve-note figuration returns, but the situation is ambiguous, since this bar may initially be understood as starting with yet another rising six-note motion. The easiest and in current sense most idiomatic way to play this passage would be to use the left/right hand alternation introduced in the second part of m. 23 also in the first part of m. 24, but this way of playing would suggest that a shift to a non-syncopated melody has taken place. This rhythmic mutation should be saved until it actually shows up in the culmination, i.e. in mm. 27–28 and 29–30, unequivocally featuring regular duple units for the first time in the piece.

When possible, the idiomatic structure should correspond to the musical structure, and since it is preferable to suppress the latent regular rhythmic subdivision in mm. 23–24, the ideal solution would be to start m. 23 *and* m. 24 as usual, i.e. with three right-hand notes. But rushing down with the right hand to strike the f♯ beginning m. 24 entails too great a risk. In order not to play in a manner suggesting that the accompaniment pattern from m. 23 is continued, the left hand might take over the first note of m. 24, letting then the right hand carry on as usual.

Turning to the three closing bars of Brahms's variation (Ex. 3d), they are tricky in a quite unidiomatic way. The prevailing pattern of hand alternation is suddenly abandoned in m. 19, and although the triplet motion is pursued, it seems to be replaced by a swarm of quick up-and-down duplets. The second note e1 is unfortunately included in both the right-hand and left-hand figurations. Omitting it in either hand brings an uncomfortable gap in the motion that might cause co-ordination problems; striking the key with both thumbs entails a substantial risk of getting the note too strong. Both hands proceed strictly in contrary motion, and this fact presents itself more clearly to the pianist's mind if the fingerings are devised so as to co-ordinate the thumbs.

As to m. 20, the right-hand descent may be divided into five-note groups by superposing the little finger immediately after the thumb. In order to conceal the gaps, the shifts of position in the left hand should *not* coincide with those in the right – superposing the ring-finger, making for groups of four notes, seems recommendable. But no matter the irregular fingering groups, a sense of metric grouping into six-note units must prevail.

#### **Brahms,** *Intermezzo* **in E**♭ **minor**

The melodic recitative starting Brahms's E♭-minor Intermezzo op. 118, No. 6 (Ex. 4) can be played with several fingerings suggesting different motivic subdivisions and different aspects of extra-musical meaning. This passage demonstrates that idiomatic may ultimately be a matter of expression, and that expression in turn may arise from the encounter between a given structure and a certain way of playing it. 13

<sup>13</sup> Another passage of this kind is analysed in detail in Bengt Edlund, "The phenomenology of fingering. Structure and ontology in Chopin's F-minor Etude from *Méthode des Méthodes*", ch. 7 in *Chopin. The Preludes and Beyond*, Frankfurt 2013, Peter Lang Verlag; it also appears in vol. 2, pp. 88–105 in Poniatowska, Irena (ed.) *Chopin and His Work in the Context of Culture*, Polska Akademia Chopinowska, Narodowy Instytut Fryderyka Chopina & Musica Iagellonica, 2003.

The passage as a whole may superficially be described as a two-bar prolongation of g♭<sup>2</sup> , followed by a relaxing suspension f2 –e♭<sup>2</sup> . But the motivic organization of this long gesture is ambiguous, a fact that contributes to the mood of painful disquietude. The most straightforward reading is probably the one in which a four-note motif, issuing from a long note and continuing with a falling motion g♭<sup>2</sup> –f 2 –(g♭<sup>2</sup> )–e♭<sup>2</sup> , is followed by a rhythmically contracted, less hesitating repetition, and finally by a rounding-off motion, an f2 –g♭<sup>2</sup> –f 2 –e♭<sup>2</sup> neighbour-note figuration with a delayed release down to the tonic note. But the long initial note also tends to split off the two first notes g♭<sup>2</sup> –f 2 , and this falling second is followed by a four-note motif g2 –e♭<sup>2</sup> –(g♭<sup>2</sup> )–f 2 suggesting an inherent rising motion. The ensuing iteration of this motif may be heard as interrupted prematurely by a closing neighbour-appoggiatura motion f2 –g♭<sup>2</sup> –f 2 –e♭<sup>2</sup> ; alternatively, the listener will pick up the final inflection f2 –e♭<sup>2</sup> , recalling the descending second g♭<sup>2</sup> –f 2 starting the piece.

Three notes are involved in the recitative, and the first fingering to present itself, **4**–342 **4**342 343–2, holds together the entire twelve-note phrase and suggests a repeated four-note motif. By means of a final shift to **1**43–2 it is possible to feel (and convey) a four-note closing motif. Beyond matters of idiomatic this is the preferable way to play the recitative since it is congenial with the music's content. Turning up virtually throughout this brooding intermezzo, the recitative melody twice alludes to the four notes beginning *Dies Irae*. 14

Another quite idiomatic fingering, 3–**1**32 3**1**32 **1**3**1**–2, consistently uses the thumb in sub-position playing f2 , and hence it does not give a proprioceptive association to *Dies irae*. If the initial g♭<sup>2</sup> is first played with the index finger and the next g♭<sup>2</sup> with the middle finger, the first two dragging notes will emerge as separated from the following group of four notes, **2**–1 **3**231. Turning to the end of the recitative, the "thumb-fingering" may be finished in three ways, and the deviation from the established fingering pattern gives the pianist a hint of the moment when the releasing break in the motivic sequence takes place. Thus, if the thumb is replaced by the middle finger on the next to last note in m. 2, the shift occurs at f2 , giving rise to a **3**43–2 closing group of four notes; if the last g♭<sup>2</sup> is played with the ring finger instead of with the middle finger, the break is postponed until this note, and you will get a **4**3–2 three-note melodic cadence.

<sup>14</sup> For reminiscences of *Dies Irae* in Chopin's works, including the F-minor etude just mentioned, cf. Bengt Edlund, "Allusions and affinities. Tracing an ominous motif in works by Chopin", ch. 1 in *Chopin. The Preludes and Beyond*, Frankfurt 2013, Peter Lang Verlag.

But the ring finger can also be used already for the second g♭<sup>2</sup> in m. 2, suggesting a **4**2343–2 closing formulation comprising six notes.

If the fingering **4**–342 **4**342 3**4**3–2, always using the ring finger for g♭<sup>2</sup> , is expressive of a delayed fall from g2 towards f2 –e♭<sup>2</sup> , the "thumb fingering" 3– **1** 323**1** 32**1 43–**2 means that this relationship has been enriched by a further aspect. When the thumb, playing f2 , is repeatedly placed in sub-position between the middle finger (g♭<sup>2</sup> ) and the index finger (e♭<sup>2</sup> ), this feels as a manual metaphor for an inherent tension within the melody: g♭<sup>2</sup> suppresses f2 (or g♭<sup>2</sup> and e♭<sup>2</sup> lock up f2 ). The note f2 , being held in suspense, seems to be released only when it is eventually included in the straight fingering 4**3**–2, turning it into an appoggiatura bringing the decisive fall to e♭<sup>2</sup> .

#### **Scriabine, Prelude in G**♯ **minor**

The relationship between idiomatic and interpretation, between the meanings embodied in the proprioceptive sensations and in the musical structure, is certainly intricate and worth studying. The next example illustrates how an analysis of different fingerings, and of the expressive content associated with them, may lead to a solution of a bewildering interpretation problem.

It is likely that many pianists have felt uneasy about the long slur in mm. 2–4 in Scriabin's G♯-minor Prelude Op. 11, No. 12 (Ex. 5). What is wrong with it? Well, the upper right-hand line seems to get stuck at b1 until it finally wags up again to d♯<sup>2</sup> , and the softly dissonant sonority, lacking leading-note tension as well as significance within the large-scale harmonic drift, is quite static. But the corresponding slur in mm. 6–8 is quite meaningful. A falling upper-line motion can be discerned within the figuration, and despite the fact that the harmony is veiled and (just as the melody) eventually turns deceptive, this passage can readily be understood as being the end of the first part of the piece. No matter the peculiar phrase in mm. 2–4, you are prone to understand the entire passage as the antecedent-consequent pair of an eight-bar period.

At the start of the prelude, the right hand immediately presents the six-note gesture that generates the following bars – the long phrases consist of two such gestures linked together by three connecting notes. The gesture is very well adapted to the hand and seems to be made up of two indolent, three-note rising motions, the latter of which is sensually stretched to reach an accented top note; what we hear may be described as two lingering attempts suggesting the presence of a slow upper line played by the little finger. But in the phrase mm. 2–4, it is this very gesture with its inherent meaning – subtly extra-musical and suggested by the motions of the hand – that causes the problem. If you stick to it also at the end of m. 3, you cannot prevent that the phrase (and the whole quasi-antecedent) is finished by a disappointing, out-of-place variant of the initial starting gesture, by a gesture that most inappropriately anticipates the first gesture of the quasiconsequent associating back to the start of the prelude.

The fingering 23**5** 13**5**, in which the stretching of the little finger indicates the motion of the inherent upper line, is likely to be the one that you will first think of when playing the initial gesture. But the fingering **1**24 **1**25 is also quite good, and if it is adopted, the active motion of the thumb suggests a concurrent lower line, describing a descending fourth. So if you want, the six-note figuration may emerge as a two-voice structure: the rising upper line is complemented by a falling counterpart.

In the problematic two-bar phrase, you can gradually turn the upper line c♯<sup>2</sup> –b1 –a♯<sup>1</sup> – b1 – –b1 –d♯<sup>2</sup> out of focus by giving some emphasis to the inherent triad of the lower strand g♯<sup>1</sup> –e1 –c♯<sup>1</sup> –g♯<sup>1</sup> –e1 –. The result will be a figuration that neither promises melodic continuation, nor harmonic development, a nonclosing, dissipating phrase that is not transformed into a furtive start robbing the following six-note phrase of its initiative. This effect is best achieved with a fingering that only uses the thumb for the lower, triad-line – a way of playing that is not very convenient but is quite idiomatic from an interpretational point of view because the proprioceptive feeling agrees with the desired musical effect. 15

In the corresponding phrase of the consequent, on the other hand, it is suitable to highlight the upper line at the expense of the less interesting lower strand. It may also be a good idea to somewhat bring out the left-hand notes e♯–b–d♯<sup>1</sup> in m. 6, i.e. to suggest a rising continuity in terms of imitation between the hands. Offering a rich sound when playing the altered chord – leaving at long last the deep organ-point on the tonic – is most idiomatic. 16

The following four bars of the prelude, twice featuring the thematic gesture in the left hand, warrant a comment. The two entries are identical with the initial

<sup>15</sup> There is of course no need (in the name of consistency) to use the "thumb-fingering" already from the start of the prelude. Another way of giving the lower strand a certain balancing emphasis in the first, germinal gesture and thus of helping the listener to notice its two-voice structure, might be to linger somewhat on its first two notes g#1 –a#1 so as to subtly suggest a false duplet: if transiently heard in this way, the first note, belonging to the lower line, will emerge as occurring on a metrically privileged position.

<sup>16</sup> From a technical point of view it is disturbing that in mm. 3, 4, and 7 the same key is immediately used again by the other hand, an idiomatic nuisance that is rather frequent in Scriabin's works.

model in terms of interval content, but the idiomatic, the proprioceptive feeling, is radically changed due to the transfer from the right hand to the left. Whereas the initial right-hand gesture is a matter of leisurely reaching out for the top note f♯2 with the little finger, the left-hand replica in m. 8 is about moving the thumb from a♯ to c♯<sup>1</sup> , from where a middle-register counterpoint starts.

This change of proprioceptive quality is of course not surprising – our hands are certainly symmetric, but not when it comes to their function on a piano keyboard. And yet this commonplace observation is crucial since it undermines current notions of musical identity. If performing music is acknowledged on a par with listening to it, it is evident that various manners of playing, for instance using different fingerings and transferring material between the hands – which entails radically different fingerings – bring about changes in vital musical respects. What is "musically" identical may turn out to be idiomatically quite different. 17

It is also evident that musical identity, contrary to current understanding, is not immune to transpositions. Playing in different keys is likely to entail substantial idiomatic changes on most instruments, such as new fingerings introducing new proprioceptive sensations and meanings. 18

#### **Chopin, Prelude in B major; Prokofiev, Piano Sonata No. 8**

The optimal interpretation does not always correspond to the technically most convenient way of playing. This, as well as the general observation that idiomatic in a comprehensive sense involves artistic matters, is demonstrated by the beginning of Chopin's Prelude in B-major Op. 28, No. 11 (Ex. 6).

The fingering 3/234314/3 for the initial right-hand melody immediately presents itself and is easy to play, and yet it seems preferable to use a strange and (at first) quite difficult fingering defying the anatomy of the hand, 3/23**545**4/3,

<sup>17</sup> It is an embarrassing fact that I played this prelude over and over again as a teenager before the musical identity between the right- and left-hand gestures dawned on me. This lapse was hardly due to the change as to timbre and musical context, but to the radical proprioceptive makeover; the difference between the signals from the hands was simply more important than the aural similarity. One might conclude that musicians are not always the best analysts. Or perhaps they *are* the best analysts since from their vantage point they have access to an additional, perhaps bewildering, bodily mode of understanding music.

<sup>18</sup> The reader is invited to imagine an E-minor Intermezzo Op. 118, No. 6 and to devise suitable fingerings for its beginning. This transposition does certainly not invite to locking up f#2 between g2 and e2 by playing it with the thumb.

i.e. the fingering that you will use in m. 6 and 18 when the same melody recurs as the upper voice in a series of double-stops. But why choose such an artificial solution already in the solo melody opening the prelude?

The answer is to be found in the following two bars, which may be conceived of as a two-voice structure. There is an upper line, twice moving from d♯<sup>2</sup> to c♯<sup>2</sup> , interleaved with a lower line insisting on the notes f♯<sup>1</sup> and g♯<sup>1</sup> . The upper strand of this swift dialogue is played by the little finger and the ring finger while the lower strand is left to the other fingers. The initial solo melody evidently belongs to the upper voice to be, and this prospective relationship is expressed by the awkward outer-finger sequence 5**45**4 – the melodic fission is prepared in both hand and mind.

But there is a further, paradoxical reason: some difficult motions are pleasant to play. In order to facilitate the precarious sub-positioning of the little finger it is necessary to make a relaxing lateral motion of the hand and arm, and this flexibility must be supported by a certain finger tension, which in turn suggests a certain, desirable qualitative nuance. The difficult fingering seems to give the solo melody a peculiar, somewhat strained intensity. At least to the pianist, it sounds as if it were played on the oboe; the easy fingering associates to a flute.

The next example illustrates quite drastically that a comfortable manual feeling is not necessarily associated with passages that are convenient to play. The passage mm. 22–25 from the first movement of Prokofiev's Sonata Op. 84 features a two-voice writing that is both simple and artful (Ex. 7). The right hand plays a lyrical and expansive melody, while the *quasi fagotto* left hand moves in bold curves up and down, undermining the otherwise unequivocal E♭-major tonality of the melody by transient visits to distant harmonies. This accompaniment is certainly not characterized by good idiomatic in current sense: the left-hand part is difficult in an awkward way, and yet it might be quite rewarding for the pianist.

The fingering that one eventually (and perhaps) finds when trying to surmount the obstacles, includes – in addition to some unusual and partly precarious sub- and super-positioning exercises – a number of extreme contractions of the hand, juxtaposed with extreme extensions. The passage makes up a very stimulating challenge and requires much but always pleasant work, until full and smooth command is attained and an even, sonorous *legato* sequence of tones seems to push the hand to new positions, until the left hand moves with the same astounding ease as an orang-outang transporting itself from branch to branch. An additional source of proprioceptive delight in this passage is that the polyphony emerges so clearly: the motions are so different in character that the hands feel completely independent of each other.

#### **Poulenc,** *Intermezzo* **in A**♭ **major; Prokofiev, Piano Sonata No. 8**

The sight of the long sequence of right-hand chords closing Poulenc's Intermezzo in A♭ major (Ex. 8a) is not likely to instil any presentiments of manual delights. Otherwise this piece (as virtually all Poulenc's piano music) is filled to the brim with idiomatic elegance – it seduces the pianist even more than the listener – and the final passage with its vertiginous gliding-flight zigzagging over the circle of fifths is typical of how exquisitely and absurdly Poulenc often rounds off his pieces. But the parallel, second-inversion triads moved around by the right hand arguably make up an idiomatic flaw.

What can be done? Those who do not give up until the music suits the hands – i.e. pianists who are not willing to let the notation restrict their technical ingenuity – may eventually hit upon an altogether different distribution between the hands. The result is a sequence of double-stops that without any serious difficulties can be mastered with *legato* fingerings, an arrangement that paves the way for a better balance of sound in a passage that now emerges as more linear than chordal (Ex. 8b).

Since the musical result will be subtly different, one should perhaps ask oneself whether this idea conflicts with the composer's intentions. Perhaps it does, but Poulenc was a skilled pianist, and here as well as elsewhere in this Intermezzo his notation is musically lucid rather than officiously over-explicit by showing fellow-pianists how to play. It might furthermore be maintained that the various signs to be found in scores differ as to musical importance and normative power, and that some of them actually concern decisions that arguably belong to the domain of the musician. 19 The distribution between the hands in piano music, as well as fingerings in general, can only in rare cases be considered as highly intentional, and should be counted among the things that are normally left to the player. 20 Differences as to fingering may give rise to appreciable musical differences, but the pianist is responsible for choosing technically and musically suitable fingerings. But apart from what the listener hears and turning to what the pianist feels, playing Ex. 8a is very different from

<sup>19</sup> Arguments to this effect are to be found in Randall R. Dipert, "The Composer's Intentions. An Examination of their Relevance for Performance", *The Musical Quarterly* 66(1980), 205–218, and in Bengt Edlund, *"Sonate, que te fais-je?*".

<sup>20</sup> However, a worrying exception to this principle is to be found three bars before Ex. 8a starts. Poulenc proposes fingerings for several passages in the Intermezzo, but here we find a footnote: "*respecter scrupuleusement ce doigté "*. (Why?)

playing Ex. 8b. Idiomatically, from a proprioceptive point of view, these passages seem to belong to different works.

The odd finishing bars of the Intermezzo make up an example of latent idiomatic. Elsewhere, it has been argued that it is important to distinguish between the notated and the interpreted structure, i.e. the musical structure as apprehended by someone. 21 But in addition to the structure of the music as heard, one must also take account of the idiomatic structure as an essential element of the music although it is present only in the musician's mind. Passages featuring latent idiomatic properties demonstrate how the notated, heard, and performed structures may be incongruent with each other; indeed, this is perhaps and to some extent the normal state of affairs. Music works also exist as something that you play, and that you perceive and understand by means of your body.

Finally, we will see how the sonorous qualities of the music may bring happiness by rubbing the instrument in the right way, by a consummate use of the nature and resources of the instrument. To fully explain this elusive aspect of idiomatic is a delicate task, but perhaps a short description of the first eleven bars of Prokofiev's Piano Sonata Op. 84 (Ex. 9) gives an idea of what is involved.

The excerpt begins with a nine-bar period. It is made up of an antecedent opening towards the dominant and an even more expansive consequent returning to the tonic, and between them is inserted an extra, harmonically redundant but aesthetically indispensable, bridging bar, entirely set in the dominant. The wide melodic arches in the right hand and the accompanying interior voice are convenient to play. On the whole, you can devise a pleasant *legato* fingering for the melody; as to the lower right-hand voice, it can often be played with a gliding thumb. 22

But the great attraction emanates from how the bold expansions in register and harmony produce richly coloured spaces, and from how the melodies, often soaring far above the low basses, are supported by mellow and yet transparent middle-register sonorities. It is also striking how well the texture can be held together by means of just a few changes of pedal co-ordinated with the bass progression, pedal shifts suggesting a sense of breathing, of gradual assimilation of dissonances followed by clarification.

<sup>21</sup> Cf. Bengt Edlund, "On scores and works of music".

<sup>22</sup> Two details may be mentioned as being latently idiomatic since seemingly problematic situations can be elegantly circumvented: the striking with the index finger on an already depressed key (e♮<sup>1</sup> ) at the beginning of m. 4 in order to maintain the feeling of a continued regular fingering, and the sub-positioning of the thumb on e♮<sup>1</sup> at the fourth beat of m. 6, a very convenient solution that clearly signals the deviation from m. 1.

The piano's resources of sensual shades of harmonic sonority are certainly fully exploited in this passage. In m. 1 a softly shimmering dominant complex piles up over the tonic fundament, a dominant that (if you have succeeded in balancing the various notes) hardly needs to be filtered by the pedal. This sonority is for a short moment at the beginning of m. 2 topped and saturated by c2 opening up a new register, a note that (together with is lower octave) is immediately left alone in a suddenly rarefied atmosphere, and that along with the renewed tonic in the bass serves as the starting point for the next sound aggregate and for a further rise. The dissonant upper voice in m. 3, soaring high above the bass, is supported by underlying sixths.

The added m. 5, subtly undermining the regularity of the period, brings a hushed low-register interlude; beginning somewhat like m. 1, it ends as an upbeat to m. 6.

The even more expansive dominant complex in mm. 6–7 is reinforced in the middle register, and urged by its own increasing harmonic mass it leads without interruption up to the culmination and to the point where the bass fundament yields by a semitone. The following decrease in tension and downward transfer of the melody in mm. 8–9 are made poignant by the expressive chromatics and by the ever-darker register of the middle voice.

The beginning of the next period brings a radiant tonic aura made up of two high-register semitone clusters. The following, abrupt descent of the upper strand is accompanied by mildly dissonant seventh chords, vanishing downwards: an image of rapidly subdued light.

Much has been said about technical aspects and interpretative possibilities since it is in the common ground between the motions of fingers, hands, arms, etc. and the artistic options that one should look for the idiomatic of music. These reflections are to be regarded as an attempt at substantiating a more far-reaching claim: it is necessary to widen musical analysis and aesthetic reflection so as to include the bodily perceptions and insights that you get when performing music. Only by taking into account our proprioceptive sensations can we hope to arrive at descriptions worthy of the musical objects and at evaluations allowing of a fair scope for the various ways in which we encounter music.

#### **Idiomatic differences between instruments**

The title of this essay holds out the prospect of a "comprehensive approach" to musical idiomatic, and the account has been comprehensive in the sense that aspects of idiomatic that most often are neglected have been brought to attention. But comprehensiveness is not only a matter of deepening, it is also about broadening, and we have merely dealt with piano music. Making music no doubt has a common ground, irrespective of the instrument used, but it is evident that the basic conditions of piano playing are quite different from the conditions determining the idiomatic when playing other instruments.

To present an account of idiomatic that is comprehensive in the latter sense requires a team of experts and much more space, but it is possible to give an indirect idea of the fundamentally different playing conditions of other instruments by recalling the quite peculiar terms of piano playing. When reading what follows, think of what it takes to make music when playing (say) the violin, the double-bass, the flute, the clarinet, the trumpet, the trombone, the organ pedals, the side drum, and when singing. 23

If you hold out your hands in front of you with the palms directed downwards, the symmetry is obvious: the thumbs point inwards, the right hand is mirrored by the left. But the keyboard of the piano has a lateral layout with the bass to the left and the treble to the right. This clash between biological symmetry and artificial lateralization causes much of the peculiar character of piano playing. If you want to play the same melody with the left hand as you just played with the right, you must reverse the finger sequence, which means an entirely different proprioceptive experience; and if you let the left hand exactly imitate the actions of the right, the melody will be inversed and musically quite different. Beyond awareness for proficient pianists, this paradox is very confusing for beginners.

But whereas the hands are symmetric, the hand is not. However excellent a pianist you are, your fingers are likely to retain some of their anatomically given peculiarities: the slow thumb having its own mode of striking the keys, the strong and nimble index and middle fingers, the not entirely independent and somewhat weaker ring finger, and the little finger which may be somewhat too short. And roll down your fingers on the keys (or just on a table) from the little finger to the thumb, and then from the thumb to the little finger! You cannot but notice the constitutively given difference in ease, speed, and regularity; you have to control the latter motion, but not the former.

<sup>23</sup> For a more detailed study of these matters, cf. Bengt Edlund, "Structural Symmetry vs. Proprioceptive Patterning in Music", *The Quarterly of the International Society for the Interdisciplinary Study of Symmetry*, 7(1996), 139–151. (*Symmetry: Culture and Science*; guest editor Siglind Bruhn), and "Proprioceptive Patterns in Music", in Varia

Furthermore, while your hands look symmetric when held in front of you, they are not alike from a neural point of view. Since most of us have brains with a dominant left hemisphere, our right hands tend to be somewhat more alert, and this difference has in turn determined the lateralization of the keyboard as well as influenced the structure of piano music. High-pitched tones are apt to move quickly whereas the greater (physical as well as perceptual) inertia of lowpitched tones tends obscure fast deep-register passages. No wonder, then, that the keyboard was once lateralized so as to let the right hand take care of the melodies, leaving the bass fundament to the left, and that rapid passages occur more frequently in right-hand parts.

Excepting some genuinely contrapuntal music, the keyboard literature is characterized by the fact that right- and left-hand idiomatic are different. The net effect of the manual demands of the repertoire is to increase the constitutional differences between the hands almost to the point of specialization – the dexterity (!) of the right hand is complemented by the accuracy of the left when it comes to wide leaps, often required in accompaniments. The aim of much piano training, for instance practising etudes composed so as to let the left hand do the job of the right, and *vice versa*, is to level out the constitutional and induced differences, to give the hands the same technical competence. Human hands skilled at playing the piano are, although still made of flesh and blood, a kind of artefacts, shaped by the layout of the keyboard, by the idiomatic properties of the piano literature, and by many years of youthful practice.

Finally, a quick glance at a piano keyboard will disclose that passages do not retain their proprioceptive identity when transposed to another key, i.e. to another set of keys on the keyboard. The irregular distribution of white and black keys means that the same sequence of intervals will get a more or less different white/black "topography" when being transposed, which in turn induces another fingering, another proprioceptive character.

If it is true that music ultimately resides in the mind, musicians playing different instruments cannot but have quite different mental representations of what they are doing. And yet they are able to reach understanding and make music together.

# **Chapter 7 Distant listening**

*Das Ohr ist des Musikers ganzer Verstand.*

*(Robert Schumann)*

#### **Introduction**

It is likely that Schumann referred to the inner hearing that turns sounds into music, and that he, as the ardent romanticist he was, wanted to maintain intuition at the expense of the intellect. Perhaps it can be taken as a token of our more prosaic times if attention is here paid to the conditions determining how musicians perceive the airborne vibrations produced by their instruments.

There are several reasons why musicians are not the best judges when it comes to the sounds out of which they make music. When producing music, the musician is apt to "hear" a confluence of the physical sound waves reaching his/ her ears and the sensations emanating from his/her own music-making body. It is also likely that musicians sometimes confuse the actually emitted sound sequence with their musical intentions – they are likely to hear what they *wish* to hear rather than what there *is* to be heard. Finally and coming to the crucial point in the present context, it is obvious that musicians more often than not listen to themselves from a peculiar and misleading acoustic perspective, quite different from the perspective that really counts in professional music making, that of the audience.

The directions of sound propagation and the frequency-dependent angles of diffusion may be such that some parts of the direct sound are likely not to hit the musicians' ears. This means that their perception may be biased towards low frequencies, and that they are more or less dependent on the reflected sound to get a fair idea of the timbre, a reflected sound that due to absorption tends to be impoverished with respect to high-frequency partials.

On the other hand and due to the very short distance to their instrument, musicians (for instance violinists and flautists) do hear a lot of noise associated with the tone production, sounds that, particularly if they have high frequencies, are not audible at greater distances due to air absorption. The proximity also means that musicians tend to hear themselves as too loud in relation to their fellow players (say the other members of a string quartet) or not as loud as they really are (trumpet players, if you ask the woodwind people seated in front of them). Finally, while musicians are likely to hear a mixture of direct and reflected sound, the direct sound tends to dominate over the reflected sound; for this reason musicians at work are poor judges as regards the effects of reverberation on their playing.

A few further examples may serve to illustrate the problems involved. All brass musicians, excepting French horn players, have their ears behind the bells of their instruments, which is hardly a good position if you want to get full and reliable information as to your actual sound quality and loudness. Singers are even worse off since their sound perspective is dominated by low-frequency sound components transmitted by means of bone conduction. Organists and conductors deal with a multitude of different intensities and sound qualities, and whereas the different locations of the sound sources may help to separate the instruments, it is still very difficult to form a correct idea of the joint effect of, and the balances within, the organ registers and orchestral instruments as they are heard in the auditorium.

Professional musicians have learnt from long experience how to cope with the fact that they cannot trust their auditory feedback while playing. (At least we like to think that they are not at the mercy of the peculiarities of their listening conditions.) But this experience is hard to get; there is a lot of trial and error, and much waste of time, involved in the learning process.

It is true that musicians, ever since tape recorders came into general use, have had equipments at disposal making it possible to listen to themselves at a distance. But it seems that sound recording has been sparingly used to support artistic judgement when practising or in rehearsals. The reason for this is presumably the fact that you listen to recordings afterwards; there is no incitement to corrective change when it would be most effective, i.e. when playing. What you want is to have immediate feedback, to be able to listen to yourself at a distance while playing, which seems impossible. 1

<sup>1</sup> But we must not forget that immediate feedback does not reveal all about your playing, and that it may be deceptive. In fact, recordings are quite valuable precisely since they offer delayed feedback, since they allow of *temporally* distant listening. Being deeply involved in what you are doing, it is hard to be critically aware of how the music is presented, and this seems to be particularly true in the temporal domain. Detached listening afterwards means that you may discover flaws in your interpretation, things like exaggerated agogic (or dynamic) gestures and minute demarcations impeding the flow of the music. When playing you are in command of the musical time, and therefore you run the risk of abusing it.

# **Distant listening – two conditions**

In this report is proposed and evaluated a method that in a number of ordinary situations enables you to listen to your own playing from a distance. It would not be surprising if *distant listening* has already been invented, perhaps several times. The surprising thing is that it does not seem to be widely practiced. 2

In short, the method works as follows. To prevent (as much as possible) the musician from hearing the sound of the instrument in the natural, airborne way, he/she wears high-performance protective earmuffs. The emitted sound is instead picked up by microphones, mounted at some suitable distance in the room, and relayed back to earphones in the earmuffs.

In order to work satisfactorily from a perceptual point of view, distant listening must fulfil two conditions. Firstly, the proximate sound travelling directly from the instrument to the player, and inevitably leaking somewhat despite the earmuffs, must seem to be exchanged for the distant sound fed back from the microphones to the earphones. Secondly, the distant sound must not confuse the player due to its necessarily somewhat delayed arrival.

The first condition entails that the distant sound issued from the earphones must have a substantially higher intensity than the remainder of the proximate sound, finding its way into the ears in spite of the efforts to muffle it. Otherwise the proximate sound will not be properly masked.

The intensity of the direct sound decreases proportionally to the square of the distance from the instrument. On the other hand, and to an extent that depends on the degree of absorption, the room will be uniformly filled with reflected sound. The intensity of the distant sound received by the microphones amounts to the impact of the direct sound, reduced in intensity due to the distance, plus a considerable intensity increment due to reflected sound. In a reverberant room the reflected sound may begin to dominate over the ever-weaker direct sound just one or two meters away from the instrument; at longer distances the reflected sound will in practice determine the total sound intensity, being virtually constant. If the musician is to gain information as to how the music is heard in the auditorium, the microphones must be mounted outside the "reverberation radius", i.e. in the area around the instrument where the reflected sound dominates. 3

<sup>2</sup> Hence this renewed presentation.

<sup>3</sup> The reverberation radius is inversely related to the amount of reflected sound, which in turn is inversely related to the absorption in the room; Johan Sundberg, *The Science of Musical Sounds*. San Diego 1991, Academic Press, p. 176.

Good protective earmuffs of the kind used in the tests to be accounted for reduce the sound level with approx. 16 dB at 125 Hz, 23 dB at 250 Hz, 32 dB at 500 Hz, and 39 dB at 1000 Hz. The masking effect of tones within the same critical bandwidth as the tone to be masked is approx. 20 dB. 4 In practice, and excepting perhaps very low tones (which generally have weak fundamental frequencies anyway), the intensity difference between the relayed distant sound and the muffled proximate sound may often make for proper masking without any amplification of the signal from the microphones – if not, the relayed sound can be amplified until it masks the proximate sound.

The dual fact that the masking implies adding the intensity of the leaking proximate sound to that of the relayed distant sound, and that amplification of the distant sound above its actual intensity level may sometimes be necessary, means that distant listening is not reliable if one wants to check the loudness of the music as heard in the room, arguably a minor drawback in most applications of distant listening. When setting the volume of the relayed sound, proper masking must be the primary consideration; next comes finding a sound level making for comfortable and informative listening. Only in the third place, and if it is of any interest, one might try to adjust the volume so as to equal the "authentic" intensity in the auditorium.

Singers, who would benefit most from the opportunity to judge their voice quality at a distance, have their instruments inside their own head and cannot use distant listening on a par with other musicians. The low-frequency-biased, bone-conducted sound determining the impression of one's own voice can of course not be quenched by ear-protecting devices.

Turning to the second condition, the time interval between the muffled proximate sound and the delayed but stronger distant sound is also crucial. The distant sound having travelled through the room to the microphones is of course bound to arrive at the player's ears somewhat later than the proximate sound. Obviously, double tone onsets must be avoided, and so must any sense of delayed onsets in general – discrepancies between motor and auditory input may be gravely confusing when playing.

If the time interval between the arrival of the proximate sound and that of the delayed distant sound does not exceed a certain, critical value, and if the distant

<sup>4</sup> Johan Sundberg, *The Science of Musical Sounds*, p. 67. The 20 dB value means that in order to be audible a tone must be 20 dB louder than what it had to be if the masking tone were absent.

sound is substantially louder than the proximate sound, a kind of "precedence effect" applies. 5

The precedence effect as currently used in public-address systems means that the amplified, relayed sound is added to the direct sound in such a way that you locate the perceived sound to the place of the original sound source, and that you take it to start when the original sound starts. In order for this illusion to work properly, two limits must be observed. The relayed sound must not be more than approx. 10 dB louder than the original sound, and the time interval between the relayed sound (hitting the eardrums first) and the original sound must not be more than approx. 30 ms.

Distant listening amounts to a modification of the precedence effect, and it means that the 10 dB intensity-difference limit can be disregarded – there is no need to secure correct localization since both the proximate and the distant sound seem to be localized to a place within the earmuffs. Thus, if only the 30 ms time-difference limit is observed, the listener will hear the delayed distant sound as starting at the onset of the proximate sound, and the illusion required to avoid double or delayed onsets has been achieved.

Hence, the second condition for distant listening is reduced to mean a maximum delay of approx. 30 ms, which in turn introduces a limit for how far the microphones can be placed from the instrument. Since the velocity of sound is approx. 343 m/s, the distance from the instrument to the microphones should not exceed approx. 10 meters, which is more than most ordinary rooms measure and also, for all rooms where you make music, far beyond the reverberation radius, i.e. the distance beyond which reflected sound dominates the aural impression.

It should be observed that the masking situation involved in distant listening, and hence the criteria of masking, is somewhat unusual. The (slightly) delayed distant sound from the microphones and the masked proximate sound are the same except for certain crucial quality differences, and this quasi-identity is a necessary condition for the "original-onset" illusion of the precedence effect. In practice, the desired masking is a fact when the ordinary, well-known proximate sound from the instrument seems to be replaced by a distinctly different sound, similar to how the instrument would sound at a distance.

<sup>5</sup> Arthur H. Benade, *Fundamentals of Musical Acoustics*, New York 1976, Oxford University Press, p. 204, and Donald E. Hall, *Musical Acoustics*, Belmont 1980, Wadsworth Publishing Company, p. 363.

# **Evaluation of "Distant listening"**

Distant listening was tested for eight different instruments – violin, violoncello, flute, clarinet, trumpet, trombone, piano, and organ – as well as for a baritone singer in order to find out what insights a singer might gain despite the fact that distant listening as a matter of principle does not work for the human voice. The subjects (teachers or advanced students at a music conservatory) were asked to play a few short excerpts at their own choice, excerpts varying with respect to tempo, dynamics, and articulation. The author took part as well; wearing earmuffs/earphones supplying distant sound, he sometimes accompanied the subjects at the piano in order to try out distant listening in ensemble playing.

The purpose was to find out if and when distant listening works, and to have its merits, if any, assessed by musicians. The core issues of the evaluation were the quality of the distant sound as opposed to the usual close-up sound perspective, and questions of loudness balance and articulation – the latter was likely to be influenced by the fact that the musicians could hear the full effect of the reverberation.

From a technical point of view, distant listening worked well in most situations.

The microphones could be mounted as far away as approx. 7 meters, i.e. at a distance allowing the musicians to be fully aware of the effects of reverberation, but precluding any disturbing double or delayed sound onsets and/or any distracting lack of co-ordination between motor and auditory onsets.

The distant sound relayed to the earphones turned out to mask the residual proximate sound reaching the ears in spite of the earmuffs. The musicians reported that they heard themselves playing in an entirely different, "out-there" way, and this experience turned out to be quite exciting. The distant-listening effect could be achieved in all cases but one without amplifying the signal much above the actual intensity at the microphones.

Turning to the individual evaluations, a number of observations made by the musicians will be accounted for.

The *violinist* sometimes noticed a slight over-hearing directly from his instrument. This effect was not due to bone conduction since it remained when the violin was kept out of contact with the jaw. (When playing the piano, the author could also hear some proximate sound when tipping his head backwards and slightly to the side.) It seems probable, then, that for certain sound-propagation angles or certain positions of the head proximate sound might leak in under the edges of the earmuffs. When using the most distant microphones, the violinist had some difficulties in co-ordinating properly with the piano. He found distant listening particularly useful for improving details of bowing technique – he could distinguish noise components surviving at a distance. Engaged in selecting a new violin, he wanted to use the method in order to compare different instruments with regard to how they would sound to an audience.

The *cellist* found the distant sound with its brightness and transparency quite inspiring, and she was especially interested in the opportunity to get an idea of the acoustic environment; the curtain at the rear wall of the hall was used to test how a varied amount of reverberation influenced the performance. Both equipped with distant-listening earmuffs, the cellist and the author rehearsed the exposition of the first movement of Brahms's E-minor Sonata, evaluating the potential of the method as an aid to achieve a good balance between the instruments.

The first *flautist* was very pleased with the fact that distant listening relieved him of the ordinary way of hearing himself differently with the left and the right ear – to flautists the sound is much louder in the right ear. He also appreciated the possibility to hear to what extent the distant tone was free from noise associated with the blowing. Having brought his baroque and classical flutes in addition to his modern one, he played the same passages on all three instruments and found it quite informative to listen with distant ears. He also played notes in different registers and at various dynamic levels on these flutes, and compared the sound at several microphone distances with the dB-values obtained from the soundlevel meter at these positions.

The other *flautist* used distant listening to evaluate the effects of different (both nearby and distant) microphone positions. Distant listening was found to be quite useful when studying the relationship between various attack articulations and the distinctness of tone onset as heard at a distance. He could also check the balance of multiphonic and whistling effects.

The *clarinet* player did not find that the difference between distant and ordinary listening was very great, but he used the method with profit to evaluate timbre differences associated with various fingerings.

The *trumpet* player did not hear the distant sound as quite representative, but he found it very interesting to play with the relaying microphones as far away as 11 meters although the substantial delay robbed him of the usual immediate auditory feedback. This condition reminded him of the fact that trumpet players (and other musicians seated far back in the orchestra) must play slightly ahead of the others in order to make for a good ensemble precision. [This idea might perhaps be developed into a practising method helping students to acquire a feeling for the proper degree of temporal "push" when playing in the orchestra.]

The *trombone* player sometimes tended to hear both the bright distant sound and some residual amount of low-frequency-biased proximate sound, a situation that made for intonation problems. The intonation was appreciably bettered, however, when the intensity of the distant sound was raised to secure full masking. He used the opportunity to test the rule that the listeners' auditory impression of the trombone is more favourable if the player does not stand exactly face to face with the audience, directly exposing the listeners to all highfrequency components.

The *pianist* was perhaps the most enthusiastic of the subjects. He found the rich, balanced, and transparent distant sound of the grand piano much more attractive than the usual sound heard at the keyboard, being dominated by low-frequency components and including a fair amount of mechanical and hammer noise. He also pointed out that it was of great value to hear more of the reverberation in the hall, and he could immediately use this information to refine articulation and pedalling.

Due to the position of the organ, the distant microphones had to be placed at a great distance down on the floor of the church, but in spite of this the *organist* did not complain about any delayed feedback. (Organists are likely to have acquired a certain tolerance with respect to late onsets.) He considered distant listening to be profitable when judging registrations, particularly such involving the quite strong *Brustwerk* just above the his head and the *Rückpositiv* behind his back. Since it gave him an idea of what it takes to achieve an impression of musically effective silence in a highly reverberant room, he found the distant ears to be very useful for checking articulation and choice of tempo.

Turning finally to the *baritone singer*, he could only hear the distant quality of his voice as part of a mixture of relayed and bone-conducted sound, but he found it quite interesting to sing while wearing earmuffs with the relayed sound turned off. This condition, effecting a substantial reduction of the emitted sound in favour of internally transmitted vibrations, offered opportunities to check the head resonances for various pitches and vowels. The singer and the author at the piano, the latter wearing distant-listening earmuffs, worked with problems of balance: at which accompaniment loudness and in which voice registers does a pianist run the risk of drowning a baritone singer? Distant listening made it possible to gain an objective idea of the relative intensities actually involved.

Tradition bids that a singer is to stand more or less with his/her back towards the piano – an unfavourable listening position not only for the pianist, hearing little direct sound from the singer, but also for the singer, hearing the piano as being quite loud. The lid of the grand piano was sometimes closed or almost so, sometimes it was against current practice left wide open. The sound quality of the grand piano was bettered when the lid was opened, of course, but this did not necessarily result in drowning the singer. [Perhaps the habit of closing the grand piano derives more from the fact that singers, standing in the middle of the acoustic draught, do not like to feel overwhelmed, than from well-founded considerations with regard to sound balance.]

It appears that distant listening may yield valuable musical insights. It can of course not be used all the time or even very often when making music, but frequent use is not necessary in order to gain musical experience and to arrive at musical conclusions. Distant listening, sparingly applied, emerges as an important resource within higher musical education, offering students opportunities for reconsideration of ingrained performance habits. It may also be productive when it comes to solving certain problems in professional work. While distant listening can be refined in various ways to meet specific demands, it can without further development be applied in a variety of musical situations as a complementary, not necessarily very expensive tool for improving training and artistic work. 6

Finally, a word of caution. At least since the demise of cold fusion in 1989 the need for corroborative studies has been increasingly pressing. Especially in psychology, sociology, and biology the craving for new, interesting findings has led to a growing number of investigations that in turn require further

The original report is to be found in the Proceedings of the 6th International Conference on Music Perception and Cognition (ICMPC) and the 4th Triennial Conference of the European Society for the Cognitive Sciences of Music (ESCOM), Keele 2000, August 5–10. ISBN 0-9539909-0-7, CD-ROM, online access.

<sup>6</sup> I wish to express my thanks to Johan Sundberg (Royal Institute of Technology, Stockholm) and Anders Jönsson (Department of Audiology, University of Lund) for their constructive interest. I am also grateful for the open-minded co-operation of these musicians: Anders Frostin (violin), Hege Waldeland (violoncello), Anders Ljungar-Chapelon and Terje Thiwång (flute), Christophe Liabäck (clarinet), Peter Meyer (trumpet), Mattias Cederberg (trombone), Andrzej Ferber (piano), Hans Hellsten (organ), and Johan Weigel (baritone).

investigations. But when such re-studies were carried out, it turned out that (for various reasons) the earlier results could not always be reproduced. The method of distant listening is not exempt from re-evaluation; quite to the contrary, it should be subjected to further tests.

# **Chapter 8 Reduction and interpretation**

*Nicht jedes Ende ist das Ziel. Das Ende der Melodie ist nicht deren Ziel; aber trotzdem: hat die Melodie ihr Ende nicht erreicht, so hat sie auch ihr Ziel nicht erreicht. Ein Gleichnis.*

> *(Friedrich Nietzsche, Menschlich, allzu menschlich, Der Wanderer und sein Schatten)*

*Not every ending is the goal. The ending of a melody is not its goal; but nonetheless, if the melody has not reached its ending, it has not reached its goal either. A parable.*

> *(Friedrich Nietzsche, Human, all too Human, The Wanderer and his Shadow)*

#### **Introduction**

The subject of this text is tonal analysis and its relationship with interpretation. Reduction in a general sense is no doubt an important mode of musical understanding, and it can sometimes be productive when it comes to interpretation. While music – to some extent – may emerge as a quasi-hierarchical structure made up of more and less essential tonal events, my enthusiasm for Schenkerian tonal analysis is very limited indeed. I entertain other notions as to what a "tonal" structure might amount to, and I prefer other ideas as to why and how reductions are to be undertaken.

The aim is not just to criticize Schenkerian analysis; my contribution will be constructive in as far as I venture to propose an alternative way to understand the music to be studied. But my effort is not to be taken as a "post-Schenkerian" undertaking; my analysis will hopefully deviate enough to qualify as a non-Schenkerian reading. Since I do not want to be associated with a dogmatic top/ down approach, and since I think that all too often Schenkerian analysts have disregarded better readings for orthodox (and sometimes quite bad) ones, I wish to avoid all associations with *Schenkerian* tonal reduction. But there is no reason to abstain from the worn-out, but still persuasive qualifier "tonal" since the reading to be proposed is arguably just as tonal as any Schenkerian analysis.

Arguments against my reading to the effect that the criteria of reduction differ from those adopted in Schenkerian analysis, or that the structures eventually arrived at are not acknowledged by Schenkerian theory, are not to the point. My reduction, as well as analyses in general, should not be assessed in terms of compliance with a certain theory supposed to be generally valid, but with respect to whether or not it agrees with and makes sense of the music studied; indeed, it should be evaluated with respect to its usefulness when it comes to interpretation.

The piece to be studied is the slow, A♭-major movement of Beethoven's C-minor Sonata Op. 10, No. 1. The choice of this very piece is coincidental. I came across the theme of this movement when reading about another Beethoven theme, that of the first, variation movement of the A♭-major Sonata Op. 26. These themes have (at least) one thing in common: they feature an event that may, or may not, be a "consonant passing-note" – a slightly contradictory nicety that non-Schenkerians are exempted from problematizing. But turning to an important asset, when dealing with the second movement of Op. 10, No. 1, I had two authoritative tonal reductions of its antecedent at my disposal. 1

Although fairly versed in Schenkerian analysis, I did not want to present a Schenkerian reading of my own. Apart from the awkward methodological issues bound to turn up if I were to use and criticize a home-made Schenkerian analysis, I sometimes fail to come up with the shrewd tricks that may be necessary to make the music exhibit a tonal structure conforming to the theory – tricks that I would not have had the guts to apply, anyway.

When turning to passages beyond the antecedent of the theme and to the movement as a whole, I do not know for certain what the Schenkerian readings might have been. Being often sceptical of the validity of Schenkerian analyses already when small formats are concerned, I tend to find tonal reductions of complete movements unwarranted or incomprehensible. Whereas fundamental structures in Schenkerian sense may perhaps have something to do with the "tonal syntax" of musical "sentences", it seems that the rhetoric governing the tonal design of larger sections and entire pieces is likely to be of an altogether different kind. 2

<sup>1</sup> William Drabkin, "Schenker, the Consonant Passing Note, and the First-Movement Theme of Beethoven's Sonata Op. 26", *Music Analysis* 15(1996), cf. especially pp. 154– 155, and Felix Salzer, *Structural Hearing*, New York 1962, Dover, cf. Ex. 443 in vol. II. There is also a further source: Allen Forte & Steven E. Gilbert, *Introduction to Schenkerian Analysis*, New York 1982, Norton, p. 152. Their views are highly consonant with those of Drabkin and Salzer: "the primary note is C", and the "main melodic motion" involves its neighbour-note, which "is 'covered' by upper-octave doublings". For a thorough discussion of the theme from Op. 26, cf. Bengt Edlund, "Disciplining reduction and tonalizing interpretation", ch. 2 in *Qestioning Schenkerism*, Frankfurt 2015, Peter Lang Verlag.

<sup>2</sup> Cf. Bengt Edlund, "Syntactic vs. rhetoric structure in music", ch. 7 in *Qestioning Schenkerism*, Frankfurt 2015, Peter Lang Verlag.

Yet, in virtue of my general knowledge of Schenkerian theory and methods it was possible to make qualified guesses as to what the results of tonal analyses of the other passages were likely to be. There is, for instance, a rule to the effect that once the primary note (*Kopfton*) of a passage, section, or work has been chosen, it must be retained, or else the tonal unity of the passage/music will be compromised.

### **Salzer's reduction of the antecedent**

Beethoven's theme is to be found in Ex. 1; Felix Salzer's reduction of its antecedent is shown in Ex. 2.

Salzer extracts the first part of an interrupted local fundamental structure issuing from the third degree as well as an expanded hidden repetition of a neighbour-note motif; cf. the brackets. The tonic persists until the last event in m. 7, which means that it incorporates the (slightly) dividing cadence to the dominant in m. 4 as well as the quite prominent excursion to the subdominant in mm. 6–7. The treble line is apparently read as follows: the structural thirddegree c2 in m. 5, reached by means of a two-stage stepwise initial ascent from a♭<sup>1</sup> , is first covered by e♭<sup>2</sup> and then left for its upper neighbour-note d♭<sup>2</sup> , which is approached from above (f2 ) and then covered by a♭<sup>2</sup> – a note deriving from an inner-voice a♭<sup>1</sup> and serving in the sun until the neighbour-note d♭<sup>2</sup> and the thirddegree c2 are back again.

This account of the antecedent is no doubt in accordance with Schenkerian theory: there is a proper *Anstieg* to a third-degree *Kopfton*, having patent tonic support, as well as a falling *Urlinie* interrupted at the second degree over the dominant, and several of the concepts habitually resorted to when analysing melodies are adduced. But Salzer's reading is most disappointing; his description does not match Beethoven's antecedent as an evolving musical entity.

Harmonically, Salzer's prolongation of the initial tonic engulfs not only the internal cadence to the dominant. The D♭-major subdominant, i.e. the harmonic *raison d'être* of the entire antecedent, the only escape from the turn-of-the-mill tonic-to-dominant framework, has all but disappeared in a nested prolongation that makes the tonic-to-tonic jar more important than its subdominant content.

Turning to the melody, the third unit in Salzer's rising chain of *verborgene Wiederholungen* is incommensurable with the preceding ones: its appearance, construction, and function are altogether different. 3 And even worse, in favour of

<sup>3</sup> Cf. Bengt Edlund, "Hidden repetitions and uncovered parallelisms", ch. 4 in *Analytical Variations*, Frankfurt 2020, Peter Lang Verlag.

the insignificant neighbour-note d♭<sup>2</sup> , the rising octave from a♭<sup>1</sup> to a♭<sup>2</sup> , the melodic *raison d'être* of the entire antecedent, is dismembered and packed up like a plastic toy in a *Kinder-Egg*.

According to Salzer's analysis the melodic peak comes about as a by-product of secondary notes recursively covering more essential ones, whereas the impression deriving from the actual musical process is that a♭<sup>2</sup> is actively strived for by the melody itself from its very beginning. Notice how in Ex. 1 the short, vigorous upbeats – disarmed and misunderstood in Ex. 2 – force the melody upwards, and how the rise is underscored by parallel octaves as well as by the fact that the left hand octaves shadow the right-hand melody a tenth below. There is of course no octave transfer producing the a♭<sup>2</sup> in m. 6 as suggested in Salzer's reduction; the ultimate rise of the soprano is certainly not a product of any inner voice. In this Schenkerian analysis, the a♭<sup>2</sup> is something that happens to the third-degree c2 , not a result of the melody's "will" to reach its peak. In Beethoven's music the horse draws the carriage, not the other way around.

Furthermore, as a consequence of this exchange of melodic design for "tonal" structure, the crucial difference between the antecedent and the consequent of the theme – the fact that the consequent, in contrast to what one might expect, merely features f2 , a *less* bold top note, as its peak and goal – is bound to shrink into an insignificant detail. (Cf. Ex. 1 and the account of the theme shown in Exs. 4 and 5)

Turning to interpretation, what can a performer make out of Salzer's reduction? Is Ex. 2 really helpful?

Is there any pianist who wants to sell out Beethoven's glorious and accelerating expansion from the a♭<sup>1</sup> /I beginning up to the a♭<sup>2</sup> /IV6 peak for a "long line", made up of a dull rising sequence of three upper-neighbour-note excursions issuing from a♭<sup>1</sup> /I, b♭<sup>1</sup> /V, and c2 /I, and for an upper-line contour that rams the structural ceiling already with the motion e♭<sup>2</sup> –d♭<sup>2</sup> ? For a piece of paper-work, that not even pays respect to the obvious fact that the third member of the decelerating "sequence" of neighbour-note motions is crucially different – not just different in terms of the surface, but significantly different as to structural content.

Has anyone heard an interpretation managing to express Ex. 2, whose essence is a dubiously well-pruned scheme supposed to impart some kind of tonal unity to the antecedent? If not, it may be due to the fact that pianists are inclined to allow melodies to determine the "tonal" form. Or perhaps their reluctance to play the antecedent "structurally" is due to the sheer impossibility of rendering mm. 5–7 as basically (i.e. merely) an upper-neighbour-note motion issuing from the third degree?

Is it possible to render the tonic chord in m. 7 – its obvious, but very transient quality as a resolution notwithstanding – as anything but a subordinate harmony on the way to the final b♭<sup>1</sup> -over-e♭, preceded by a six-four chord? And how can you clarify that c2 is "in charge" from the beginning of m. 5 until the end of m. 7 when everything you play and your listeners hear indicates that this note is left? Yes, a tonic-supported c2 turns up again at the end of m. 7, but reappearance is not tantamount to prolongation. 4

The chromatic leading-note in m. 4 notwithstanding, is it a good idea or even possible to play the downbeat c2 – having a sense of bringing a third internal start – as an arrival at the note where the melodic "structure" of the antecedent begins? (What else does the *Anstieg* idea suggest?) This c2 is obviously an intermediary point of a flight heading further upwards. Wouldn't it be very disappointing if a pianist somehow managed to clarify that this midway c2 , and not the initial a♭<sup>1</sup> , is the true structural point of departure for the entire antecedent?

No pianist is likely to want to recreate Salzer's reading, with its crowning and yet impotent third-degree *Kopfton*, because it completely misses the musical process as a perceived phenomenon. No musician would want to play a bold melodic rise as something that just happens to the notes, as a series of covering operations. Musicians are exacting customers; they are not bent to buy off the peg but demand tailor-made analyses.

Salzer's analysis is a failure from the pianist's point view since the force of the subdominant is curbed by being encapsulated within a prolongation of a prolongation of the tonic, since the *Kopfton* of the structural descent has already arrived when the boldest part of the melodic rise is about to start, and since the top note a♭<sup>2</sup> is conceived of as the result of nested prolongations and coverings. And yet these counterintuitive observations are exactly what Salzer's reduction implicitly demands that the pianist should take account of and benefit from. His graph is of no use since what is really interesting and makes for continuity in Beethoven's theme is marginalized.

It is often held that the advantage of Schenkerian analysis when it comes to interpretation boils down to the fact that it brings out the "long lines" in the music. This does not apply in the present case. There *is* an interesting long line in Beethoven's antecedent – and this line is quite tall as well – but it is impeded by Salzer's half-way third-degree-over-the-tonic primary note, making for a

<sup>4</sup> There may be a few listeners who notice that there is an A♭-major c2 in m. 7, and that a note of this kind occurred a while ago, but such listeners are not necessarily the best ones, the ones having the most profound or rewarding musical experience.

low-ceiling prolongation that does not make sense. And the antecedent *does* bring an interesting harmonic excursion, but it is replaced by an overall glass-ofwater tonic-to-tonic prolongation.

Salzer's analysis is musically irrelevant to the point of paying utter disrespect to what Beethoven wrote. His reading is indeed quite resistant to surface salience, a fact that qualifies it as a "non-trivial" description according to adherents of Schenkerian theory. 5 But the price for this non-triviality is as high as it is paradoxical: standard configurations are brought out at the expense of vital processes. When dealing with unique musical designs, as musicians are bound to do, an analytic theory that according to its very nature assigns itself the task of demonstrating generalities is of little avail.

Paradoxical is also the fact that a theory, claiming that it provides exclusive access to the essence of Masterpieces, gives rise to analyses that present the commonplace as primary and treat the unusual as secondary. Generally speaking, and assuming that you need to understand what is particular in a piece of music in order to play it well, it would be more illuminating to study reductions assigning structural status to traits that disclose what is unique.

### **Drabkin's amended reduction**

Adopting Schenker's concept of '*Konsonanter Durchgang*', and using it in the same way as Schenker did in his analysis of the first-movement theme from Op. 26, William Drabkin proposes an amendment of Salzer's reading of the antecedent; cf. Ex. 3 showing mm. 5–8 only.

In order to avert the (not very impending) threat of consecutive middleground fifths (c2 /f-to-b♭<sup>1</sup> /e♭) in mm. 7–8 he relieves the tonic chord in m. 7 of its structural duties. His reading also allows space for the "unfolding" of the subdominant by giving d♭<sup>2</sup> status as a structural neighbour-note and by accepting a structural IV. The c2 in m. 7 is analysed as a passing-note supported and made consonant by a root-position A♭-major chord. This sonority is preceded by a first-inversion dominant seventh-chord effecting a re-harmonization of d♭<sup>2</sup> ; the g in the bass is understood as a passing-note between f and a♭.

Drabkin's truly Schenkerian description of his analysis runs: "… in which the return to the tonic – at the fourth quaver of bar 7 – is the result of a passing note

<sup>5</sup> Cf. Nicholas Cook, "Music Theory and Good Comparison: A Viennese Perspective", *Journal of Music Theory* 33(1989), 117–141, and Bengt Edlund, "Schenkerian theory and better comparison: An out-of-the-way perspective", ch. 1 in *Questioning Schenkerism*, Frankfurt 2015, Peter Lang Verlag.

harmonised by a tonic to avoid consecutive fifths in the progression IV6-5 –V5 ". (p. 154)

To begin with, the origin of the consecutive fifths – the sixth d♭<sup>2</sup> /f – is a fabrication. These notes are not simultaneous: if we turn to Beethoven's m. 6, it is perfectly clear that d♭ supports f2 and that a♭<sup>2</sup> is supported by f. For rhythmic reasons the structural neighbour-note status of d♭<sup>2</sup> is seriously undermined. As to the consecutive fifths, the first one (c2 -over-f) involves structurally ill-matching notes that are not present at the same time. The non-structural f has already been left for (the not merely passing) g when c2 -over-a♭, being the resolution of the immediately preceding first-inversion dominant, arrives. The second fifth (b♭<sup>1</sup> over-e♭) brings the delayed dominant of the fundamental structure.

Conforming to the demands of Schenkerian theory, Drabkin has made a mountain out of a molehill: there are no consecutive fifths in Beethoven's music or even in Drabkin's own middleground. And it is most unlikely that any impending, sub-surface voice-leading flaw disturbed the composer, making him "harmonise" the (resolving) c2 by a root-position tonic chord. The preceding firstinversion dominant seventh-chord over g has already done the fifth-averting job, which of course does not imply that Beethoven inserted this chord to "avoid consecutive fifths"; his inspiration presumably derived from other, better but unknown, sources. 6

It is a laudable idea to diminish the structural significance that Salzer accords to the tonic in m. 7, i.e. to do away with the most improbable seven-bar prolongation of the tonic, but Drabkin provides his analysis with a questionable motivation, not likely to convince any musician. His reading of this tonic chord, and of the entire antecedent, is predicated on the necessity to explain away a negligible problem due to a self-imposed principle within Schenkerian theory, the rule that deeper (background and middleground) structures must conform to strict counterpoint.

Drabkin's reading of mm. 5–8 may seem as a precarious construction in comparison with the patent tonal stability of Salzer's reduction, and yet it makes up an improvement since it implicitly takes some account of the gradual

<sup>6</sup> If an alternative speculation as to what was going on in a dead composer's mind is allowed, Beethoven rather thought of the rising sequence of consecutive tenths, starting already in m. 3 and being unmistakably present at all accented positions in mm. 5–6 – a too simple, straightforward observation to be of Schenkerian interest, obviously, but also a fact indicating the presence of a theoretically undesirable overall tonal structure; cf. below.

destabilization characterizing Beethoven's tonal design. His analysis is also to be preferred because the crucial subdominant is acknowledged, and because some of Salzer's fuss over the rising motion in the treble is cleared away. But the subdominant is still slighted – Beethoven offers more than IV – and the peaking a♭<sup>2</sup> is still not displayed as the main top-line event, as the event that the antecedent aims at and eventually achieves – and that the consequent will fail to match.

Salzer's mm. 5–7 sub-surface copy of the initial neighbour-note motif is dispensed with – which is fine. Instead there are two similar-looking triad motions supported by I and IV root-position chords, respectively. But the two motions are in fact different since for top/down reasons – a structural neighbournote is needed for the subdominant – the note d♭<sup>2</sup> is brought out in m. 6. Unlike the starting c2 in m. 5, the short, unaccented d♭<sup>2</sup> is hardly a very important note. Whether intentionally or due to a misprint, the notes in the treble and the bass are not correctly aligned in Ex. 3. 7 In m. 5 it is in fact only the second-beat notes e♭<sup>2</sup> and c2 that have first-inversion chord support; in m. 6 it is in fact the peak note a♭<sup>2</sup> , not the d♭<sup>2</sup> , that enjoys first-inversion support. There is nothing in the reduction that discloses the rhetoric impact of the unstable peak sonority a♭<sup>2</sup> over-f. Musicians are interested in learning about growth and culminations, and they prefer events having an inherent tonal momentum to events like d♭<sup>2</sup> -over d♭, weighed down by stability.

Although Drabkin's reading is an improvement, most pianists are likely to consider it to be beside the point. The melodic toy (it is a giraffe) inside the *Kinder-Egg* has still not been properly assembled. As long as you believe that a very short, upbeat upper-neighbour-note d♭<sup>2</sup> dominates mm. 6–7, you are not encouraged to engage in a dedicated right-hand rising expansion. For what else than a static keep-to-the-same-register approach does the "long-line" neighbournote motion, does this anodyne standard voice-leading configuration, suggest? If you pay attention to the fact that the tonal ceiling is in fact raised to a♭<sup>2</sup> , it would from a Schenkerian point of view be tantamount to letting in an element of surface salience, an element of illusion (*Schein*), into your performance. No reduction is really helpful for a musician (or for anybody else wanting to understand a piece of music) if it renders what is obviously bold and essential as

<sup>7</sup> Intentionally? Yes, it does happen in the Schenkerian trade that notes are moved to places where they are not in the score, and where they should not be in a fair analysis, and sometimes these changes apparently serve manipulative purposes.

subordinate, if it gives precedence to what is pedestrian, however sub-surface it is supposed to be.

#### **An alternative reduction**

Beethoven's theme deserves better, and in Ex. 4, presenting a non-Schenkerian analysis, the harmonic structure of the antecedent is I–IV6 –V, and an arch encompassing an entire octave emerges in the treble: first 1–2–3–5–6–8, then 7–6–5–4–3–2. From m. 5 the bass follows a similar path in terms of parallel tenths and it reaches its octave as well, but the a♭ arrives too late for joining the right-hand a♭<sup>2</sup> . The eighth-degree peak note of the melody is instead supported by the first-inversion D♭-major chord. This is the event that Beethoven actually and quite emphatically offers as the culminating turning point in the antecedent; it has a quality of remoteness worth striving for as well as a sense of instability explaining the quick retreat from it. The rise from the tonic note to its octave and the concurrent harmonic redefinition of the pitch-class A♭ make up the main structural events of this reduction, according to which mm. 1–8 are not read as merely a routine antecedent eventually arriving at the dominant, but is understood as an expansive voice-leading structure having the exposed subdominant as its core.

In mm. 1 and 3 there are no bland upper neighbour-note motions, as the short resolution notes bids and Salzer's reduction affords, but rising thirds that right from the start of the melody announce that they are the constructive interval of the theme. It might be objected that the ascending thirds in mm. 1 and 3 involve falling appoggiaturas, and hence that these bars bring rising seconds. But it may be argued that the c2 in m. 1 and then the d♭<sup>2</sup> in m. 3 count for more than their resolutions; Beethoven was not entirely mistaken when inscribing turn ornaments to make the listeners pay attention to the rising thirds producing the dissonant notes. And musicians play melodies, not minute exercises in counterpoint: in m. 1 and m. 3 the appoggiatura/resolution motions, however expressive, are subordinate; in mm. 1–2 and 3–4 the chromatic ascents count for more than the neighbour-note motions. Later on in mm. 5–6 rising thirds, brought out by swift upbeats instead of turns, urge the melody to proceed upwards.

The sub-surface structure in Ex. 4 is no less "tonal" than the ones exhibited in Salzer's and Drabkin's readings, and certainly no less tonal than Beethoven's theme. Quite to the contrary, it reflects the tonal growth, the treble/bass parallelism, and the accelerated motivic pace producing the antecedent. Indeed, it illustrates a basic idea in Schenkerian theory, the notion that the life of musical "organisms" is a matter of first acquiring, then releasing tonal potential. The melody of the antecedent – slowly rising to a state of high tension then quickly falling – is (no offence) like a mechanical toy, which you first wind up then let go.

To ultimately conceive of the antecedent as simply a specimen of an interrupted 1/I–2/V "*Ursatz*", would be to misrepresent what happens. (To miss the point is a grave misrepresentation.) The focus on the high-tension 8/IV6 is an indispensable aspect of the antecedent, and therefore 1/I–8/IV6 –2/V emerges as its fundamental structure.

Is this alternative account of any help to a musician? Admittedly, the pianists are able to do quite well in this particular theme without any reductive assistance; musicians are good at winding up toys. But it may be favourable for the true long line up to a♭<sup>2</sup> to have a graph indicating that the c2 in m. 5 is neither a start, nor an arrival, but a mid-point from which the melody takes on a more active and directed, aspiring character. And musicians may appreciate that this reduction brings out the unstable 8/IV6 as the core event; if they take a look at Ex. 4, they will get a confirmation that they are on the right track. Furthermore, the fact that it is shown that the bass shadows the rise of the melody suggests that it might be a good idea to play so as to make it clear that the bass motion is pursued all the way up to the post-culmination a♭. 8

It seems, then, that this reduction, the outcome of which is very far from being an admissible *Ursatz*, is quite possible to render (or at least suggest) at the keyboard; it is also quite worthwhile to listen to since it embodies a voiceleading structure that may have some significance in terms of human expression. This is not to say that a pianist steeped in Schenkerian theory would not in practice be able to play the antecedent in a meaningful way; when it comes to the crunch ill-matching analyses cannot resist the music printed on the page. (Probably such analyses are simply disregarded.) The point is that Salzer's and Drabkin's Schenkerian analyses of the antecedent are not helpful when it comes to discovering its vital peculiarities: the focal but unstable subdominant, the long *and* tall upper line prompted by rising thirds, and the imitative, eventually

<sup>8</sup> This bass motion does not cook down to a prolongation of the tonic – there is a dominant and then a subdominant in the way. That there is a quasi-imitative relationship between the treble and the bass is corroborated by the corresponding passage in the (slightly varied) restatement of the theme' cf. particularly mm. 51–52. The left-hand figuration is bifurcated so as to let an imitating middle-register line emerge; being at first ahead of the right hand when injecting its impulses, the lefthand finally has to rise precipitately in order to catch up with the right and to reach its belated peak at a♭<sup>1</sup> .

out-of-phase relationship between melody and bass. According to Exs. 2 and 3 mm. 1–6 never leave the ground, whereas in Ex. 4 there is both a starting run and a moment of take-off.

One of the things that Salzer's and Drabkin's readings have in common is that they dwarf what Beethoven wrote. Even in small formats comparable to musical clauses, a reductive analysis predicated on process and rhetoric may evidently be more productive than Schenkerian parsing according to a tonal syntax taking its rules (and its rule) for given.

By now, readers of the right stamp will have protested for two reasons. I have not at all understood the point of Schenkerian analysis, and my analysis of the theme is trivial since it just brings out what is patently there. But I do think that I know the point of "tonal" analysis; the very motivation for my alternative account (and this essay) is that I am not convinced that this point is worth pursuing, particularly when it leads to unmusical readings of defenceless scores, and particularly if you are about making analyses to serve musicians. As regards triviality, Beethoven's noble and well-constructed theme/melody does not keep any deeply embedded tonal secrets, and this fact should be acknowledged and brought out, rather than be hidden away by means of convoluted and far-fetched reductions – by reductions that fathom the boldness of a musical design by marvelling at the discrepancy between the given text and the inadequate, theorydriven analysis supposed to explain it. It is better to be trivial and stay in touch with the musical realities than to pat the cat against the fur.

## **The consequent**

Turning to the consequent of the theme, cf. Ex. 1, it can safely be assumed that Salzer and Drabkin would have analysed it in adherence to their readings of the antecedent: the last-moment d♭<sup>2</sup> in m. 14 would probably again have been singled out as a crucial upper neighbour note. Why should the preceding, out-of-the-*Ursatz* peak note f2 , however prominent it is in Beethoven's melody, be selected when it, no matter its root support, was merely regarded as a covering note in the antecedent, and when the peak note a♭<sup>2</sup> in m. 6 was virtually left out of account?

The most remarkable trait in the consequent is that the rising melody makes a halt already at the sixth-degree f2 , comma at a note enjoying subdominant root support, and being the final note of a rising line marked for attention by three leaps/upbeats; then the melody falls towards the tonic without much ado; cf. Ex 5. Thus, contrary to ingrained convention and hence to what one expects, the consequent offers less melodic expansion and brings less harmonic tension than the antecedent: the octave is exchanged for just a sixth, and the first-inversion subdominant chord for just a root-position one. 9 The harmonic momentum associated with the melodic turning point is decreased in the consequent, and considering the entire theme, a falling a♭<sup>2</sup> –f 2 top-note contour hovers over the period.

This crucial, overall aspect of Beethoven's theme would not have presented itself readily in Schenkerian analyses of the theme. The subdominant is tucked away (Salzer) and misrepresented (Drabkin) already in the antecedent, and due to the hierarchical fragmentation supposed to explain the melodic process, the crucial top notes occupy a far too low structural rank in both the antecedent and the consequent. Indeed, had the reductions been pursued beyond what is shown in Exs. 2 and 3, the peaking a♭<sup>2</sup> 's would have been altogether reduced out of sight. 10 The two Schenkerian analyses fail to account properly for vital aspects of the theme – aspects that are crucial for its interpretation and for the understanding of the entire movement, cf. below – and this failure is not due to unfortunate oversights: quite to the contrary, the premises of the theory conspire to block analytical understanding.

# **Formal overview; the transitions**

Before proceeding, an overview of the movement is due. Its form may be summarized as follows:


<sup>9</sup> The peculiarity of this reversal of the current rhetoric pattern within periods can be appreciated if one exchanges mm. 13–16 for mm. 5–8, and *vice versa* – providing of course the necessary adjustments to close in the dominant and tonic, respectively.

<sup>10</sup> This is what happens at Salzer's next reductive level, not shown in Ex. 2.

The first transition consists of three exclamatory melodic phrases followed by a connecting right-hand bridge. The falling sequence of the first two phrases, f 3 –e♭<sup>1</sup> –d♭<sup>1</sup> and e♭<sup>3</sup> –d♭<sup>1</sup> –c1 , may perhaps be understood as making up the first two units of a broader, more emphatic descent faintly alluding to the very quick one issuing from the top-note f2 of the consequent; cf. Exs. 6a and 6b. But there is also a more robust complementary association, which is clarified by the three leaps marking the c2 –e♭<sup>2</sup> –f 2 ascent in mm. 13–14 of the consequent and then each exclamatory phrase in the transition, f3 –e♭<sup>2</sup> –c♭<sup>1</sup> . This backward reference and the concomitant sense of retroversion in mm. 17–20 make the final, low-register c♭<sup>1</sup> –C♭–B♭<sup>1</sup> phrase emerge as the third unit of the descending sequence, bringing the music down to the dominant of the E♭-major second theme to come.

A third exclamatory phrase akin to the preceding ones is due in m. 21, but this expectation is thwarted when c♭<sup>1</sup> unexpectedly turns up. This entry is certainly in the wrong register, but however surprising it may be, the pitch is not entirely wrong considering that the model started from c2 in m. 13. 11 Making up for the discontinuities there is a middle-register connection: the final bass motion C♭– B♭1 brings a suitable end to a three-member sequence starting with e♭<sup>1</sup> –d♭<sup>1</sup> and d♭<sup>1</sup> –c1 . And the falling tendency of the transition, taking place in three registers, is firmly held together by parallel sixths and tenths.

The complex sense of resumption involved in this passage is a crucial aspect of the music and its interpretation. Beethoven apparently attached great importance to the retrograde association making for a connection between the consequent of the first theme and the transition –why else did he plant such unmistakable hints as the three quick rising leaps? Why else did he compose a transition that retrospectively confirms that f2 was a structurally essential note in the consequent?

What would a "tonal" reduction of the transition be like? The long-term survival of the drabbest, i.e. the *Urlinie* once started from c2 in m. 5, can be secured if one takes a Schenkerian look at mm. 17–20: the root-supported c1 in m. 20, preceded by its upper neighbour-note d♭<sup>1</sup> in m. 18, is bound to be selected as the primary note: it is no doubt "pre-covered" by e♭<sup>3</sup> , just as the d♭<sup>1</sup> was "precovered" by f3 . This reminiscence of the would-be structural d♭<sup>2</sup> –c2 apex of the antecedent is as dead as mutton, but it would probably count for more than the obvious resumption relationship referring back the consequent. But the "tonal" logic is as impeccable as it is inescapable: if the sixth-degree f2 is not conceived

<sup>11</sup> For "diatonic" listeners expecting and missing a d♭<sup>3</sup> in m. 21, there is a delayed d♮<sup>2</sup> starting the second theme.

of as a structurally significant note in the consequent – as it is according to Ex. 5 – why should there be any structurally important descent starting from f3 in the transition? It is a pity that Schenkerian analysis is so immune to "surface" rhetoric.

Turning to the corresponding transition in the second part of the movement, the third exclamation involves three deceptions: to the already known deviations as to register and pitch-class is added a harmonic surprise in relation to the transition in the first part: F♭ major occurs instead of A♭ minor. The net effect is to prepare for the second theme, now set in the tonic.

# **The second theme**

The two presentations of the second theme are essentially identical; in order to keep to the tonic the discussion will primarily deal with the second, A♭-major statement; cf. Ex. 7a.

The theme is bisected, and each part is immediately repeated and varied so as to give rise to an expansive twenty-bar period. Bar 78 obviously serves as both an ending and a start, and m. 82 corresponds to mm. 87–88, which are then promptly duplicated one octave below. The repeat of the first part merely offers rapid standard figurations. 12

Since there are first a c3 and then two c2 's supported by root-position tonic chords in mm. 74, 78, and 83, it is a fair guess that the structural upper line of a Schenkerian reading of the second theme would once again feature a descent from the third degree. If you have once chosen a primary note, tonal unity demands that you stick to it. But this "fact" (or rather artefact) of tonal analysis emerges as immaterial when it comes to interpretation: an *Urlinie* from the third degree would again be at odds with the musical surface, and also quite impossible to express when playing.

Taking humble account of Beethoven's text, the music makes straightforward sense, however; cf. Ex. 7b. Bars 71–74 bring a bi-partite rising motion from g2 to c3 – in m. 72 the two units are connected in a way that recalls mm. 2 and 4 – and a corresponding c2 turns up in m. 78. But neither of these notes has the quality of a *Kopfton* starting a structural descent. The c3 in m. 74 clearly closes of a four-bar phrase bringing an ascending fourth, and this goes for the c2 in m. 78 as well, but

<sup>12</sup> This passage makes up an embarrassing concession to virtuoso mannerisms, and in the best of worlds Beethoven would have come up with something better.

the latter note makes up the node of two elided phrases, and its obvious function is to launch a further ascent.

Bars 78–81 and 83–86 feature more or less chromatic, rising motions in the treble, supported by parallel tenths in the bass; indeed, the parallel tenths begin already in mm. 73 and 77. Disregarding the excessive top notes b♭<sup>2</sup> and c♭<sup>3</sup> , respectively, both ascents lead to a♭<sup>2</sup> 's, but these notes are associated with quite unstable and remote harmonies: a diminished seventh-chord on d♮<sup>1</sup> , and an applied Neapolitan F♭-major seventh-chord, respectively.

The first descent from a♭<sup>2</sup> occupies just one-and-a-half bar; in mm. 82 the melody clings to e♭<sup>2</sup> while pursuing the descent beneath this note in a way suggesting that the melody is split into two melodic strands. In the varied repeat, the descent takes on a more abrupt quality: e♭<sup>2</sup> occurs in m. 87 as the end-point of a triad motion, and after the swift fall to b♭<sup>2</sup> at the beginning of m. 88 e♭<sup>3</sup> turns up again, now followed by a precipitate drop to the tonic note, a manoeuvre that is repeated in mm. 89–90. Thus, e♭<sup>2</sup> /e♭<sup>3</sup> is repeatedly marked for attention and then quickly left; features that may be taken to imply that the music will return to the fifth degree and treat the descent towards the first degree with more dignity as becomes the serious quality of the movement.

On this description, the over-all similarities between the second theme and particularly the antecedent of the first theme emerge as striking. Extended rising motions, recharged at the third-degree c♭<sup>2</sup> 's in mm. 78 and 83, and propped up by parallel tenths, lead to precariously harmonized eighth-degree a♭<sup>2</sup> 's followed by hasty retreats downwards bringing out the fifth degree. The "diminutions" in mm. 1–4 and 71–74 are also quite similar (cf. Exs. 4 and 7b). The fact that the second theme starts with a rising fourth issuing from a dominant-supported g2 in m. 71 emerges as a smart way of throwing the listeners off the track.

Whatever its specific implications for interpretation, this thematic metamorphosis is a crucial aspect of Beethoven's design that should be brought out in a worthwhile analysis. But having one's mind clogged with reductions of the first-theme antecedent such as Salzer's and Drabkin's – the bold rise to the octave torn to rags and being replaced by an insipid neighbour-note appended to the third degree, and the urging parallel tenths being neglected – means that the chance of discovering the similarity between the two themes comes close to nil.

Before turning to the final part of the movement, the first statement of the second theme, set in E♭ major, should be commented upon from the point of view of tonal form.

Schenkerian theory insists that modulations are to be suppressed in favour of the main key, that they should be understood as matters of large-scale prolongation. To secure the rule of the tonic and the survival of the primary note initially chosen, passages in other keys – be it entire sections – must be encapsulated and hence (so it seems) be robbed of their independence and significance. But listeners are not prone to experience music in this way, nor is such an attitude adequate for musicians, and it is unlikely that the listeners are wrong and the musicians mistaken.

After all, a modulation by definition implies that a certain tonic, and the diatonic set going with it, is exchanged for another one. For this reason (and analytic efforts to demonstrate tonal unity notwithstanding) modulations tend to be very important musical events with a phenomenal impact second to none in the musical process as a whole. Turning to interpretation, it is not possible to subdue modulations, nor is it wise to even try to do so. Quite to the contrary, changes of key should more often than not be highlighted so as to make the listeners appreciate the new tonal perspective.

In this specific case, there are both a "first-time" subtlety and an interpretational challenge involved; cf. Ex. 6a. Whereas the new key of E♭ major is implied already by the B♭-major seventh-chord in m. 22, the tonic-to-be is not an established fact until in m. 27; the first three bars of the second theme bear a sense of taking place in the territory of a prolonged dominant, in a harmonic no man's land. 13 This passage of extended suspense cannot but be slighted if one entertains the view that it merely involves a transition to an encapsulated, *cul-de-sac* E♭-major episode within a tonal form ruled by an encompassing A♭-major *Ursatz*. For the pianist: in order to make someone believe you must believe.

# **The varied repeat of the first theme; the coda**

That mm. 91–102 make up a variation of the first theme is quite evident from the first four bars, featuring an identical melodic rise from a♭<sup>1</sup> ; cf. Ex. 8a. But in m. 95 something entirely unsuspected happens: the third-degree c2 , the would-be *Kopfton*, does *not* turn up; instead a descent featuring parallel tenths issues from a harmonically unstable e♭<sup>2</sup> – don't understate this note! This falling line, sliding down from a first-position fifth degree and passing a veiled root-position third degree, heads for a root-supported tonic note due, but not delivered, in m. 98.

The rising skip from b♭<sup>1</sup> to e♭<sup>2</sup> across the phrase demarcation in mm. 94–95 can be understood as opening up an implicational gap that demands, and gets, a descending realization; cf. Ex. 8b. This gap also suggests that the fifth-degree

<sup>13</sup> When listening to the A♭-major restatement of the second theme, we know what is going to happen in its first four bars.

e♭<sup>2</sup> will perhaps be reached in a more continuous, stepwise manner, and this is borne out as well. 14 Bidding defiance to the ingrained, closing tendency of the six-four chord, mm. 97–98 bring an emphatic rise from a♭<sup>1</sup> to e♭<sup>2</sup> – a chromatic rise, partly doubled in octaves and strongly supported by parallel tenths, and bypassing the root-position third-degree c2 without any ado. Bars 99–101 vary mm. 95–97, and finally the tonic note arrives.

But the ending of the melodic motion rather belongs to an inner voice; the top voice insists on e♭<sup>2</sup> until it quickly skips down to a♭<sup>2</sup> in a way recalling the close of the second theme in mm. 82 and 87–88 (89–90). Hence, a coda pursuing a slow, decisive motion towards the tonic is required in order to achieve full closure, and mm. 101–106 do bring a broad stepwise descent from the fifth degree, a motion that is then echoed one octave below as a falling triad.

Again, it is hard to tell with certainty what a Schenkerian account of the varied main theme would be. But since the third degree was chosen as the primary note already in m. 5 and was presumably still valid in the transition and the second theme as well, and since c2 , however veiled and by-passed it is, after all is present in mm. 91–102 with root support, the third degree is most likely to be selected as *Kopfton* in the varied version of the main theme as well. But this analysis militates strongly against the musically essential observation that the fifth-degree e♭<sup>2</sup> is demonstratively seized and then twice regained, and that e♭<sup>2</sup> is the uncontested top note starting no less than three ever broader and ever more decisive descents towards the first degree.

However, according to the reductive habits within the Schenkerian trade, the fifth degree in m. 95 cannot qualify as structural in the final, varied main theme because, no matter how conspicuous it is, the e♭<sup>2</sup> by default, as it were, merely amounts to a "covering" note over an unstable first-inversion tonic chord, just as it was in m. 5 and m. 13. A musician, on the other hand, searching for the event that actually dominates the formal unit under consideration, and being more interested in momentum than in stability, would certainly opt for e♭<sup>2</sup> as the essential and expressible, and hence structural, note in the last section of the movement. The c2 's always have a passing-note character, and it would be most detrimental to the real "long line" to treat them as anything else.

<sup>14</sup> Readers of the right stamp may react adversely against the use of concepts deriving from Leonard B. Meyer's theory of melodic implications within a reductive account. But why not? This is not a "tonal" reduction, and being rooted in our habit to envisage the course of linear connections, Meyer's ideas may contribute significantly to musical understanding (including reduction) and not least to interpretation; when playing, it might be a good idea to take into consideration what your listeners are likely to expect.

### **The entire movement**

Finally, what is there to say about the movement as a whole? In a Schenkerian analysis the entire musical "organism" must be subsumed under an *Ursatz* umbrella, showing that tonal unity prevails in the music and has been successfully established by the reduction. In the present case, the mission-achieved, toplevel "structure" lending tonal coherence to the movement will feature an overall descending *Urlinie*, issuing from the root-supported, third-degree c2 -over-a♭ in m. 5 and eventually arriving at the tonic note via a belated dominant-supported second degree presumably occurring somewhere in the last section of the movement.

Leaving this vacuous scheme for more productive observations highlighting traits disclosing something important about the musical substance of the slow movement of Op. 10, No. 1, two relationships obtaining between the various themes will be put forth. Taking account of the order of presentation, two tendencies emerge that appear to have both expressive significance and a potential of being brought out in performance. 15

Within both the first and the second theme, the skew balance between extended rising and short falling upper-line motions emerges as a conspicuous feature – the abrupt finishing motions/skips from a persisting fifth degree down to the first are especially notable. In the varied main theme closing the movement, however, an even balance between ascent and descent is achieved. Indeed, the coda starting in m. 102 is entirely devoted to exposing the motion down to the first degree, as if making up for the previous haste.

The other tendency involves the peak notes in the various statements of the main theme, and like the change in temporal balance just described, it embodies a sense of resolution; cf. Ex. 9. Presupposing a reductive analysis that accepts the peak notes as the most important events in the various passages, and that accepts them not in spite of, but due to their harmonic instability, the tendency leads from boldness in terms of register and harmony to compliance. In the antecedents the musical focus is a♭<sup>2</sup> -over-f (mm. 6 and 51) whereas in the consequents the peak is withdrawn to f2 -over-d♭ (mm. 14 and 59). And in the closing variant of the main theme, the tonic replaces the subdominant as the harmony supporting the melodic culmination, and the melody does not reach further than e♭<sup>2</sup> -overc. That this final, less charged structural focus may represent a kind of overall resolution is apparent from the way Beethoven marks it for attention: in m. 95 e♭<sup>2</sup> is exposed since it is introduced as a substitute for the expected c♭<sup>2</sup> , and in

<sup>15</sup> It should be noted that these tendencies are not matters of voice leading but connections based on association and comparison.

m. 99 it is emphatically regained by means of a demonstratively added and quite sonorous chromatic ascent.

# **Conclusions**

It appears that Schenkerian tonal reduction (presumably) fails to do justice to this movement: already Salzer's and Drabkin's readings of the antecedent of the first theme are strikingly inadequate to the point of paying utter disregard for Beethoven's tonal design. As a result of this failure due to theoretical prejudice, producing a foreseeable and narrow-minded result, it may be assumed that crucial relationships in the movement as a whole are likely to be missed – the relationship between the antecedent and the less bold consequent of the first theme; the sense of resumption obtaining between the consequent and the transition; the close structural kinship between the first and the second theme; the sense of resolution inherent in the final, varied statement of the main theme; and the fact that the movement suggests an overall tendency in terms of the withdrawal of the peak notes and their harmonic support.

Turning to interpretation, and taking for granted that the way you conceive of a piece of music will somehow affect how you will play it, the Schenkerian analysis of this movement has very little to offer. Contrary to what is often held, "tonal" reductions may inhibit rather than encourage the expression of "long lines". The connections emerging in Salzer's and Drabkin's readings of the antecedent – pedestrian neighbour-note motions, effete covering notes, slack fundamental descents – are too modest to be able to fuel an interpretation. This should come as no surprise: every good speaker (Beethoven is one of them) knows that eloquence is not a matter of subordinating oneself to the syntax, but of creatively expanding its resources.

This opposition between syntax and rhetoric, between the necessary and the interesting, was (perhaps) very well put by Nietzsche in the obscure "parable" starting this essay. There is a dialectic relationship between "endings" (closure) and "goals" (purpose) in music, but it is a pity that tonal reduction has been so obsessed with endings, and cared so little about goals. Understanding Nietzsche's parable literally, he (perhaps) meant that whatever you throw will come down, but when throwing a stone – or when embarking on a musical unit whatever its size – you should imagine the height of the curve, not the final bounce when the stone hits the ground. By the same token, a Schenkerian primary note, announcing the predetermined route to the final cadence, must not block our access to what we may call the "focal note". What can poison enjoyment and inspiration more than being assured in advance that nothing really important is going to happen, than being told that the ending is always-already present?

# **Chapter 9 Dissentient views on a minuet**

# **Introduction**

When attending a music theory conference,1 two presentations renewed my doubts as to the merits of Schenkerian analysis in general and as a guide for interpretation in particular.

In order to draw attention to a difference between two recordings, Joel Lester referred to Heinrich Schenker's middleground graph of the main part of the Minuet movement of Mozart's Piano Sonata in A major K. 331. 2 He was worried about the distrust among musicians of analysis as an aid for interpretation, and doubted that the analyses given to them were always helpful and to the point. Later on in the same session Eric Clarke, when accounting for his research on musical performance, used a reduction by Carl Schachter to give an idea of the structure of Chopin's E-minor Prelude Op. 28, No. 4. 3

To follow up Lester's worries and doubts, two protagonists will appear in the discussion making up the first part of the present text: the analyst and the pianist. This way of presentation cannot but recall an earlier dialogue on the relationship between analysis and interpretation, Janet Schmalfeldt's fictional exchange of views, casting the Schenkerian analyst as the wise teacher and the pianist as the humble (and not very bright) student – arguably a biased and all too conflict-free picture. 4

Since it will turn out that Schenker's reduction of Mozart's minuet is in various ways unsatisfactory as an account of the musical process, it is necessary to propose alternative readings agreeing better with how listeners and musicians deal with the music, and to argue for a less dogmatic approach to reduction.

<sup>1</sup> Southampton Music Analysis Conference 26–28th March, 1993.

<sup>2</sup> The title of Lester's paper was "Interactions between Analysis and Performance"; the graph in question is fig. 35, 1 in Heinrich Schenker, *Der freie Satz* II, Wien 1935.

<sup>3</sup> Cf. Bengt Edlund, "Music at the analyst's couch and at the musician's stand", ch. 3 in *Chopin. The Preludes and Beyond*, Frankfurt 2013, Peter Lang Verlag.

<sup>4</sup> Janet Schmalfeldt, "On the Relation of Analysis to Performance: Beethoven's Bagatelles Op. 126, Nos. 2 and 5", *Journal of Music Theory* 29(1985), 1–31. For a critical discussion of her paper, cf. Bengt Edlund, "Interpreting a bagatelle", Chapter 10 in the present volume.

Suppose that a pianist studying the second movement of Mozart's A-major Sonata (Ex. 1) is presented with Schenker's middleground reading of the Minuet (Ex. 2). What would he or she make out of the analysis? Are there any conclusions pertinent for interpretation to be derived from it? Or are there any odd things in the analysis that are likely to put the pianist off? Although quite a few analysts claim that Schenkerian readings demonstrate the tonal unity underlying the music and hence should lead to coherent and penetrating interpretations, these incredulous questions suggest that there may be people who are not prepared to accept Schenker's reading as valid and valuable. Let's furthermore assume that the pianist is not stupid, that his/her sceptic attitude is that of an outsider, of a person not already convinced of the merits of Schenkerian theory and not subscribing to the agenda of tonal analysis.

#### **The first four bars and beyond**

The pianist starts by saying that the first thing that anyone faced with an analysis should do is to check with the actual music. She has to confess that she was at first quite bewildered when trying to match Schenker's analysis with the score. While she accepted that the graph must be a kind of synoptic picture of the music, the notes picked out to represent Mozart's music were often displaced both in time and register in ways that militated against her musical intuitions. 5

Starting her scrutiny of the reduction from the beginning, the c♯<sup>2</sup> and c♯<sup>3</sup> shown to go with the root-position I chord in "T 2" actually appear in m. 2 and m. 3, respectively, while the tonic chord is introduced in m. 1. But even worse is the fact that the c♯<sup>3</sup> put within parentheses actually tops a VI6 chord in m. 3, or rather an applied II6 chord since it clearly belongs to a local cadence tonicizing the e2 -over-the-dominant in m. 4. This is, she says, as bad a support as a note can get if it is to assume the role of the initial tonal anchor of an A-major piece.

Due to the obvious mm. 2/3 phrase demarcation, the c♯<sup>3</sup> cannot very well be thought of as "popping up" from the c♯<sup>2</sup> in m. 2, and if it pops up at all, it does so from the immediately preceding c♯<sup>2</sup> in m. 3. This fact discloses the true nature of the c♯<sup>3</sup> : it is, as anyone should be able to hear and as anyone should respect, merely a melodic diversion on the unmistakable rising melodic route from c♯<sup>2</sup> to e2 , and hence its structural significance comes close to nil. The pianist cannot

<sup>5</sup> The pianist will henceforth be referred to as "she", which is politically correct, whereas the analysts will be called "he", which is empirically correct – by and large, Schenkerian analysis seems to be a masculine project. (Perhaps it is even a technique of male domination?).

but suspect that the importance of the c♯<sup>3</sup> is trumped up for two reasons: to hide away the motion to the stable E-major e2 in m. 4 and to provide a same-register note that the b2 in m. 11 can attach to. Do you really believe, she asks, that the c♯<sup>3</sup> in "T 2" (i.e. m. 3) really survives all the way to the b2 in m. 11 as shown by the beam in Ex. 2? Why not instead accept the straightforward a1 –b2 association between m. 1 and m. 11, a connection that Mozart urges the listener to notice by means of similar grace-note arpeggios?

She furthermore thinks that the analysis fails to take account of the rising Amajor triad a1 –c♯ 2 –e2 announcing itself at accented positions in mm. 1–3. Due to the imitative falling inflections, this motion comes to the fore most clearly in spite of the discontinuity caused by the rest, the dynamic shift from (implicit) *forte* to *piano*, and the lack of doubling left-hand octaves in m. 3. Within this three-member, sequential opening the third-degree c♯ 2 in m. 2, having no (simultaneous) root support, shrinks considerably in importance, whereas the ascending triad motion lends structural weight to its fifth-degree peak note e2 , enjoying some root support since the preceding left-hand octave may still linger on in the listener's ears.

The pianist also wants to draw attention to another pattern inherent in the music: the two initial, *forte* bars describe a quite stable, triple-octave circular motion, issuing from and returning to a root-supported a1 . In other words, the minuet starts with a demarcated, somewhat pompous, fanfare-like introduction lending structural emphasis to the tonic note.

It is a pity, she says, that the analysis ignores these two concurrent patterns, amounting to a sense of formal ambiguity that opens up for interesting interpretative options. A circular, self-contained two-bar introduction may be brought out along with the triadic rise to e2 , but the passage may also be rendered so as to hide away the initial fanfare with its emphasis on a1 . But if the fanfare is allowed to dominate the impression, mm. 3–4 will seem to start the music with a two-bar cadence! The ambiguity and the way it is dealt with cannot but influence the harmonic interpretation of m. 3: does it start in the tonic, or does its first note perhaps already belong to the forthcoming dominant?

The pianist knows Mozart's sonatas quite well, and she plays the beginning of the first movement of the C-major Sonata K. 309; cf. Ex. 3. The same concurrent aspects of a three-bar rising triad and/or a circular two-bar fanfare are patently present here as well. Considering the fact that the post-fanfare melody in K. 309 obviously issues from the fifth-degree g2 in m. 3, is it really a good idea to hold that the following descent starts only from the third-degree e2 in m. 5 (however root-supported it is), or indeed from the un-supported fanfare e2 back in m. 2?

Returning to K. 331, is it really enlightening to select the c♯<sup>2</sup> /c♯<sup>3</sup> in "T 2" as the point of departure for mm. 1–10? As to herself, the pianist cannot but conclude that the theme of the A-major minuet has even less to do with the third degree than the theme of the C-major first movement. The c♯<sup>2</sup> is merely a mediating note in a rising-triad sequence – and even more insignificant it is in mm. 5–10 – and it lacks stable root support.

If you want an anchor note for the initial period, she advices the analyst, why don't you choose the e2 in m. 3, still enjoying some A-major root support? And no matter whether you hear a rising triad or an emphatic motion around the tonic note, the fifth-degree e2 emerges as a significant note, a fact that will be amply corroborated. In m. 4 the E-major e2 concurrently emerges as the goal of the antecedent and the starting-point of the consequent. And when playing ahead, e2 and its upper neighbour-note f♯<sup>2</sup> (then f♯<sup>1</sup> ) attract attention in mm. 6–9 whereas the third-degree c♯<sup>2</sup> is less prominent. Why is the evidently quite important e2 not even shown in the graph?

Summarizing her objections this far, she is sceptical of the very start of the two high-level descents of the minuet, the initial one as well as the one starting from c♯<sup>2</sup> in m. 31 after the interruption marked in Ex. 2. This means that she questions the overall descending thirds accounting for the main sections of the minuet – indeed for the entire minuet since its D-major Trio is represented by a huge upper neighbour-note. The seemingly solid content of the crucial "T 2" in Ex. 2 turns out to be a package of untrue observations involving mm. 1–3, and this fact cannot but make her suspicious about what the rest of the analysis will bring.

The analyst listens patiently and rejects the pianist's objections: she may be excused for being ignorant and naïve, but not only does she call into question the structural unity of a piece by no less a composer than Mozart, questioned is also Schenker's gift to humanity. He points at Ex. 2 and claims that mm. 1–10 prolong the tonic by means of a local "structural descent", and that it is in virtue of this higher-level fact that the third-degree c♯<sup>2</sup> is taken to have root support and remains in force until m. 8. In local terms the c♯<sup>3</sup> is a "covering" duplication of c♯<sup>2</sup> , but if the perspective is widened, it is "structural" on the highest level and connects to the b2 in m. 11. Thus, c♯<sup>2</sup> and c♯<sup>3</sup> serve as third-degree "primary" notes for the smaller and larger structural descents, respectively. As to the "covering" fifth-degree e2 in m. 4, it cannot close the antecedent properly since any midway, "dividing" dominant that counts must be introduced by a falling motion from its second degree; the seventh-degree d♯<sup>2</sup> is not "structural".

The pianist is quite baffled, but says that she used to think that analysis should as a matter of principle proceed from the details to the whole, that analysis, being after all a kind of empirical undertaking, should never allow high-level hypotheses to suppress low-level evidence. In other words: she is still curious to know how the insignificant c♯ 3 in m. 3 (31) lacking satisfactory tonic support can work as the primary note for the entire minuet. And as a musician she wants to know whether or not the initial ten-bar period features an antecedent at all, considering the rule that there must be a falling, tonally stable second-degree treble note in m. 4 or later on.

The analyst curtly retorts that it is a beginner's mistake to confuse surface salience with structural importance, and that fundamental structures (always having descending upper lines) are not hypotheses to be empirically tested every now and then by any beginner, but make up the very law of tonal music, a law verified by innumerable reductive analyses undertaken by Heinrich Schenker and his many followers.

#### **The first period and beyond**

Not analyses like this one, I hope, the pianist murmurs, but she nevertheless goes on scrutinizing Schenker's analysis.

Ignoring the cadence to the dominant in m. 4, the next notes considered to be important according to Ex. 2 are a bass-note d1 and a treble-note b1 to be found in "T 8". But m. 8, being the third, harmonically unstable member of a rising melodic sequence to be suddenly abandoned, is hardly the optimal place for the continuation of any local fundamental structure, she thinks. The "T 8" notes also turn up in m. 9, introducing an abrupt contrast in dynamics and register as well as a patent root-position subdominant that obviously forms part of an A-major cadence. Isn't the penultimate bar of the period a better place to pick out the notes wanted for the antepenultimate event of the reduction?

Furthermore, she strongly feels that the upper-line b1 in Ex. 2 belongs to the subdominant and not to the ensuing dominant, to which the slur between the eighth-note d(1) and quarter-note e(1) in the reduction unconvincingly refers the quite prominent subdominant chord as if it had some kind of appoggiaturalike status. The fact that both harmonies make up adjacent members of the same cadence does not entail that the subdominant and the dominant share the treble note b1 , and this is all the more evident since Mozart gives the dominant a top note of its own in m. 10, the seventh-degree resolution note g♯<sup>1</sup> proceeding upwards to the eighth degree, as leading-notes use to do and as exemplified already in m. 4.

In addition she points out that Mozart also provides a releasing, falling motion towards the tonic note in this two-bar cadence, a swift descent from the fifth-degree e2 , a descent in which the second-degree b1 has merely a subordinate passing-note function.

Turning to Schenker's graph, she wants to know whether the unlikely "T 8" is preferred to the patently cadence-like m. 9 because the first-beat f♯<sup>1</sup> in the latter bar opens up for a theoretically unwanted ascent via g♯<sup>1</sup> towards the tonic note, or because the falling-fifth motion from e2 (preceded by its out-of-register neighbour-note f♯<sup>1</sup> ) makes up an unwanted complication, calling into question the prior decision to select the third-degree c♯<sup>2</sup> in m. 2, not the fifth-degree e2 in m. 3, as the primary note?

The analyst asks the pianist to reconsider Ex. 2. As shown by the slurs, there is a local *Ursatz* in mm. 1–10, and, "as we all know", fundamental structures always feature a stepwise, descending *Urlinie* to go with the *Baßbrechung*. Furthermore, the structural dominant always comes with the second degree in the treble. So just forget about the g♯<sup>1</sup> in m. 10 (and the f♯<sup>1</sup> in m. 9) – a seventh degree cannot be the penultimate member of an *Urlinie*. Structural descents simply do not behave like that – if they did, they were ascents! And can't you hear that the e2 's in m. 3 and m. 9 are merely "covering" notes? So just forget about the fifth degree – it is useless as a *Kopfton* since it is obviously the c♯<sup>3</sup> in "T 2" that makes for the largescale falling connection to the b2 in m. 11.

After being fed with some German terms and been asked to accept dogmas instead of explanations, the pianist is still not convinced. 6 Why are fundamental upper lines always descending? In this case there is a quite obvious, root-supported rising f♯<sup>1</sup> –g♯<sup>1</sup> –a1 connection in mm. 9–10; just as there was a similar ascending c♯<sup>2</sup> –d♯<sup>2</sup> –e2 motion in mm. 3–4. Can't you hear through simple high-register diversions? Are you incapable of appreciating the subtle sense of structural rhyme? And why is anything else than a third-degree primary note out of the question when it comes to this minuet? After all, there is a quite demonstrative

<sup>6</sup> By now, the reader may have noticed a peculiarity in the analyst's way of rejecting the pianist's objections: instead of discussing the relationship between his reduction and Mozart's minuet, he prefers to talk *ex cathedra* about what is theoretically given. Either (as Nietzsche once put it) he only hears the questions that he can answer, or he follows Schenker's authoritarian example. The attitude and arguments of our fictive analyst are not meant as a parody of a wrong-headed schoolteacher, but modelled after those of Schenker in the discussion between him and a recalcitrant student of his, Felix Eberhard von Cube; cf. William Drabkin, "Schenker, the Consonant Passing Note, and the First-Movement Theme of Beethoven's Sonata Op. 26", *Music Analysis* 15(1996), 149–189, and Bengt Edlund, "Disciplining reduction and tonalizing interpretation", ch. 2 in *Questioning Schenkerism*, Frankfurt 2015, Peter Lang Verlag.

initial ascent along the triad up to e2 , this note is then kept alive for four bars, and finally there is an obvious, quick descent from e2 to the tonic in mm. 9–10. And is it reasonable to argue that what happens later on – the b2 in m. 11 – can retroactively determine what you make out mm. 1–4? So far, she thinks (but doesn't say) that the defence of Schenker's reading is a miserable attempt to produce evidence for a preconceived idea having very poor support in the music.

Looking at the parallel passage mm. 31–41 for comparison, she notices that Ex. 2 does not tell when the structural b1 supposed to go with the e(1) dominant root actually occurs. If it is located to m. 38 (being the third member of the rising sequence, as was m. 8), the initial appoggiatura-note b1 must be selected instead of its resolution (as in m. 8). And the bass-note e of the six-four A-major chord in m. 38 cannot very well represent the root-position dominant chord needed for the fundamental harmonic structure shown in Ex. 2. For this bar does not precede the dominant, as six-four chords usually do, but forms the left-hand part of a second-inversion applied-dominant seventh-chord transiently tonicizing the patent subdominant in m. 39. There is a last-moment b1 in this bar, but even more than in m. 9 it must be associated with the subdominant root d in the bass; the left-hand e in m. 40, eventually bringing the root of the dominant, occurs only after a deviation down to c♯. There is a second-degree note that clearly belongs to the penultimate dominant chord, but the b in m. 40 appears in the tenor voice and hence in the wrong octave; in the relevant treble register there is again a seventh-degree g♯<sup>1</sup> .

The pianist thinks that it is regrettable that the unexpected left-hand duplication of the A-major six-four chord m. 38, the tonicized D-major subdominant in m. 39, and the following sudden twist back to an A-major cadence in m. 40 – i.e. deviations raising the interest of the A1 section of the minuet and being pertinent for interpretation – have been flattened out in the analytic mangling-house. She cannot but conclude that the succinct way of taking down mm. 31–41 in Ex. 2 hides the fact that Schenker's reading of this passage has even less support in the music than his analysis of mm. 1–10.

#### **The second period and beyond**

The pianist begins by observing that the second period of the outer sections, mm. 11–18 and 41–48, are not treated in the same way in Ex. 2. In m. 41 the initial, non-top grace-note a1 is selected as a closing note at the highest structural level, whereas the actually closing, descending-fifth progression starting from the quite prominent top note e2 is disposed of within parentheses. In the corresponding E-major passage of the first section, on the other hand, the top note b2 in m. 11 (connected backwards over the formal demarcation in m. 10 to the c♯<sup>3</sup> in "T 2") and the ensuing descending fifth are brought out as structurally important, while the non-top grace-note e2 is not even allowed to show up in the graph. Being fond of consistency, she turns to the analyst for an explanation.

When asked why the reduction suggests an elision rather than a new start in m. 41, as the formal parallelism with the first section bids, the analyst tells the pianist that the completion of the *Urlinie* must be given priority at the expense of the sense of a new start. Furthermore, according to the principles of Schenkerian analysis, a fundamental descent from the third degree – launched in this case from the c♯<sup>3</sup> in "T 2" and then starting again in m. 31 after having been interrupted by the last-moment dividing b1 in m. 30 – cannot be followed by further structural descent issuing from another note, i.e. from the fifth-degree e2 . It is of crucial importance for the tonal unity of the minuet that the latter descent is shown as structurally subordinate. In terms of the minuet's "tonal form" the final eight bars are in fact redundant: from a tonal point of view the piece closes with the grace-note a1 , and mm. 41–48, never really leaving A major, merely bring an appended cadence prolonging the tonic.

But the pianist cannot accept these arguments. Is it really true or necessary that tonal unity is that strongly predicated on a single *Ursatz*? How can a theory rule out compositional possibilities? What is the use of an analytic method preventing you from taking account of what you hear and see in the works you study? Besides, and turning to this specific case, don't the concluding eight bars – certainly making up an essential, balancing part of the minuet's form in current sense, and patently featuring a structural descending fifth just as did mm. 11– 18 – amount to a strong indication that it was a mistake to select the third degree as structural back in "T 2"?

And why, she carries on, cannot the very emphatic arrival at the tonic note in m. 31 just as well bring the tonal close of the piece? After all, the graph brings out that an upper-line descent (a descent from e2 !) makes up the tonal content of the minuet's middle section. According to the score this motion lands, not on the insignificant b1 in m. 30, but on the top-note a1 starting m. 31, a most emphatic, bi-functional elided note that is suppressed, simply taken away in the graph. The falling-fifth progression of the middle section seems quite fit to end the piece, which (it might be argued) in virtue of being a kind of recapitulation does not really leave A major after m. 31. Since what follows can be understood as prolonging the tonic, why cannot the alleged c♯<sup>2</sup> –b1 –a1 structural descent in mm. 31–41, being merely an appended motion, be put within parentheses just as the falling fifth starting from e2 in mm. 41–48?7

No, the theorist answers, it is a mistake to think that the middle section connects to the repeat section: as the analytic slurs seeing through the *Schein* indicate, it closes on the dividing b1 -over-the-dominant in m. 30. Quite to the contrary, the middle section belongs to the first section: structurally speaking and as shown in the graph, the B-section of the minuet forms the second part of a bold b2 -to-b1 "prolongation" of the second degree introduced by the top note back in m. 11. By a stroke of genius to be appreciated only by the happy few, Mozart has extended the dominant of the *Ursatz*-to-be-interrupted for no less than twenty bars! A reduction must reflect the "inner", *tonal* form of the minuet, and in accordance with one of the basic procedures of tonal music the fundamental structure of this minuet is at long last interrupted at a dominant-supported second degree, the b1 in m. 30; then the *Ursatz*-to-be-completed turns up again along with the A1 section of the "outer" form.

The middle section ends in m. 30, the pianist objects, but it does so on e2 -over-e, not on b1 (-over-e), the latter is evidently just the penultimate note of a swift and structurally subordinate, low-level mediating motion. And m. 31 no doubt starts with an elided a1 , just as there is an elision in m. 41, involving two notes, a1 and e2 . How can Schenker's high-level descending motion b2 –b1 , issuing from the first, starting emanation of the dividing-dominant-to-be, end with the penultimate note of a local connection patently leading to a1 ?

The analyst now informs her that the selection of b2 in m. 11 is a consequence of the prior decision to select the c♯<sup>3</sup> in "T 2" as the *Kopfton* of the minuet: the initial third-degree must have a high-level dominant-supported second degree to descend to, how else could there be a prolongation of the dominant lasting from m. 11 to m. 30, followed by a resumed *Ursatz*? Or reversing the matter, he adds, the forthcoming, dividing duty of the b2 in m. 11 made it necessary to choose an *Urlinie* starting from the third degree. In other words, in order to

<sup>7</sup> The pianist does not really mean what she is saying, she just wants to make a point by exaggerating. She is in fact suspicious about Schenker's reduction of the middle section, however straightforward and persuasive it looks at first sight, and she does not really think that the music after m. 31 (or 41) is superfluous, tonally or otherwise. Since she insists that the final eight bars should be included in the analytic account, she is of course not willing to disregard the entire A1 section.

match the b2 in m. 11 the relative-minor, out-of-register c♯<sup>3</sup> in m. 3 had to be selected as the large-scale primary note rather than the c♯<sup>2</sup> in m. 2.

But this circular argument deriving from, and disclosing, the heart and art of tonal reduction fails to convince the pianist. She lacks the analytic sophistication needed to understand music backwards and flatly rejects the end-towardsbeginning approach in analysis.

As regards mm. 11–18, the falling-fifth descent from b2 is certainly highlighted in Ex. 2, the pianist observes – it is even provided with a shadowing, parallel-third motion. But the descent does not show up properly in "T 17", and it is hard to tell whether this failure is due to cross-eared or wall-eared listening, she says. If you want the thirds to turn up in the correct register, they are present already in Mozart's m. 16, but only if you are willing to crumple the descending scale; cf. Ex. 1. And, whereas a cadence motion a–b in the left hand does occur in m. 17, it does not support a treble motion from a2 to g♯ 2 . The content of Schenker's "m. 17" is present in Mozart's m. 17 only if you accept that it shows up higgledy-piggledy in various voices and registers, and if you accept that the bass note a supports itself as a treble note (a2 ). Isn't the strange "m. 17", she wonders, actually an attempt to hide away the fact that the left-hand cadence motion a–b–e1 in mm. 17–18 actually supports a c♯ 2 –d♯ 2 –e2 ascent from the sixth to the eighth degree, thus bringing a structural parallel to the rising motions in mm. 3–4 and 9–10, including the capricious outof-register top notes?

In his reply, the analyst gives a long lecture on the concept of voice exchange, on the idea that strict counterpoint underlies and explains all wilful windings of the musical surface, and on the desirability of supporting important upper-voice descents with stable harmonic functions. And he stresses once again that tonally decisive, structural, upper-line motions are by definition always descending, and that, consequently, rising upper-line connections (should they occur and however important they may seem) are never structural, but just *Schein*.

The pianist doesn't listen since she is intrigued by what the analyst refuses to acknowledge – the fact that both periods of the outer sections of the minuet feature a similar, unity-making tonal structure. Whether being in A major or E major, the periods hold on to the fifth degree for a long time, and when the final, theoretically prescribed descent towards the first degree is due, it is outdone by an ascending motion to the eighth degree. Indeed, particularly mm. 47–48 sounds very familiar since virtually the same cadence turns up in mm. 17–18 in the theme of the first movement of the sonata. She plays Ex. 4 and asks whether this observation does not indicate that the first-movement theme also closes with a rising structural line to the eighth degree?

Of course it doesn't, the analyst immediately replies. It has repeatedly been demonstrated beyond any reasonable doubt by several authorities (among them Schenker himself!) that the theme of the variation movement of the K. 331 Sonata ends with a structural descent to the first degree. 8

# **The middle section**

Turning finally to the developmental middle section mm. 19–30, the pianist feels that there is something more to this passage, something interesting that Ex. 2 fails to illuminate. At first she finds the description of mm. 20–23 and 24–27 fair enough, but then she must disagree since the reduction obviously misrepresents the essence of mm. 27–30.

Mozart's conspicuous (♮II)–V cadence has disappeared altogether, and the upper voice in the graph, featuring b1 instead of e2 , is as incorrect as the bass, showing the rising second d♯<sup>1</sup> –e1 instead of the falling second f♮–e, making up the final step of a prominent fourth, descending stepwise from the a in m. 27. She cannot but conclude that the bass motion d♯<sup>1</sup> –e1 of the graph in fact belongs to the soprano, reading d♯<sup>2</sup> –e2 in mm. 29–30, whereas Schenker's b1 in the treble must be the very last sixteenth note of the connecting bridge in m. 30.

Why this exercise in inverted counterpoint, why not keep to Mozart's voice leading in mm. 27–30? This passage emerges as quite meaningful and very determined: contrary motions – the falling a–g♮–f♮–e fourth in the bass against the broken but unmistakably rising a1 /a2 –b2 –c♮<sup>3</sup> –d♯<sup>2</sup> –e2 fifth in the treble – lend closing emphasis to the final e2 -over-e dominant at the downbeat of m. 30. But this event has disappeared in Ex. 2 although it means that the middle section of the minuet is brought back to its e1 /e2 point of departure in m. 19.

To the pianist Schenker's manoeuvre, completely doing away with Mozart's design, stands out as entirely unwarranted: it amounts to a grave falsification of the text that must be rejected. What makes her particularly upset is the element of deceit involved. In Ex. 2 the shift in mm. 29/30 looks rather like the two preceding ones, which suggests that there is an orderly descent engaging three parallel voices. Having by now learnt the tricks of the Schenkerian trade, she suspects that the suppression of the actual harmonic bass progression f♮–e and the relegation of

<sup>8</sup> Cf. Bengt Edlund, "Analytical variations on a theme by Mozart", ch. 1 in *Analytical Variations*, Frankfurt 2020, Peter Lang Verlag.

the treble d♯<sup>2</sup> –e2 motion to the bass line serve the same purpose: to exalt the very last, transitional sixteenth-note b1 in m. 30 to structural top-voice prominence and accented position in order to supply a suitable end-point to the preceding e2 –d♮<sup>2</sup> –c♮<sup>2</sup> motion in spite of the fact that there is no connection between c♮<sup>2</sup> and b1 . In other words, the manipulation is needed to complete the descending-fourth progression spanning the middle section according to Ex. 2, and to bring home the even more extended b2 –b1 connection starting back in m. 11.

She frankly tells the analyst that it is embarrassing to be the victim of such cheap eyewash. She is also upset since she has now become aware of the analytical motivation of the series of "thirds" brought out before m. 19: they evidently serve to make the contrived sequence of "thirds" after this bar more convincing. To the extent that there are at all any descending thirds on each side of the double-bar sign, they are musically incommensurable.

But the analyst ardently defends Schenker's reading of the middle section, and far from being embarrassed he is inspired. There is a *Koppelung* between the b2 in m. 11 and the b1 in m. 30, and this octave, overriding the "outer" formal demarcation at the repeat sign so as to mediate between the two framing sections, effects a huge dominant prolongation, incorporating the second period of the first section (otherwise expressing E major) as well as the various harmonic and voice-leading events making up the middle section. It just takes a final "voice exchange" in mm. 29–30 to make this wonderful tonal scheme come true. Besides, he adds, the b1 in m. 30 is structurally very important since it finally brings the interruption of the initial *Urlinie* of the minuet's tonal form. As we all know, dividing dominants must be topped by second-degree notes.

The pianist, until now absorbed by and vexed at the details of the graph, has not yet paid attention to the dotted slurs and the extended V symbol, details that now turn her even more distrustful. Will anybody hear this descending octave, made up of a very strained fifth and a bogus fourth, and how can it be expressed? Aren't several other keys visited during this allegedly all-dominant middle section, aren't there two auxiliary tonics (B minor and A minor) introduced along the way by means of clearly audible root-position cadences involving three-bar applied dominants? What is the musical point of connecting the conventional right-hand passages of the patently closing period before the double-bar with the bold and entirely different, harmonically vagrant middle section? Hasn't Mozart taken every measure to separate the E-major cadence in m. 18 from the E-minor (the subdominant-to-be of B minor) start in m. 19? While the tightly mediated formal shift in mm. 30–31 is rendered in a gravely misrepresenting way, the unmediated formal articulation in mm. 18–19 is not respected: what's wrong with letting reductions correctly reflect major demarcations of the "outer" form? Schenkerian orthodoxy aside, doesn't the emphatic downbeat on e2 in m. 30, preceded by its seventh-degree d♯<sup>2</sup> , serve much better than the last-moment b1 as the top note of the dividing dominant? After all, the A-major tonic has not yet turned up; we are still in E-major territory.

The man who came up with this reduction, she concludes, must either be deeply unmusical or an unfair demagogue using analytic notation to argue against his own better knowledge. To the extent that the point of "tonal" analysis is first and foremost to extract cadences with falling upper lines, Mozart's minuet suffers badly from its third-degree primary note; it would be better off it were allowed to start with a fifth-degree e2 . Granting that Ex. 2 is a quite condensed representation of the music, how can anyone have so little left in the bucket after having been to the well?

#### **The value of reduction when it comes to interpretation**

A theorist worth his salary is never at a loss for an answer, and now he generously initiates the pianist into the deepest secrets of tonal analysis. Schenkerian reduction shows how every detail of a (first-rate) compositional design grows organically out of the fundamental *Ursatz* cadence expressing the tonic. Hence there are no "modulations" – he cannot help pronouncing this word with a ring of contempt because the very idea of modulations (impeding the free flight of prolongations) smells of low analytic competence and under-developed listening capacity. The unique advantage of Schenkerian analysis is that it penetrates far beneath the music's surface and transcends the circumstances of the "outer" form. The most common mistake in tonal analysis – typical of untutored amateurs and musicians – is to think that surface prominence has anything to do with structural importance.

"Discrepancies" (now a tone of weariness and condescension can be heard) *visà-vis* the score are not signs of analytic inadequacy but indicate that the analysis is non-trivial, that it discloses a wealth of wide-ranging tonal connections deeply hidden and yet present in the musical design. 9 Incidentally, he closes his sermon, this is why musicians may benefit so greatly from Schenkerian reductions.

<sup>9</sup> Nicholas Cook has defended Schenkerian analysis along this line of thought: the "discrepancies" make for interesting "comparisons" with the actual music; cf. "Music Theory and 'Good Comparison': A Viennese Perspective", *Journal of Music Theory* 33(1989), 117–141, and Bengt Edlund, "Schenkerian Theory and Better Comparison: An Out-of-the-Way Perspective", ch. 1 in *Questioning Schenkerism*, Frankfurt 2015, Peter Lang Verlag.

Yes, what are the consequences of Ex. 2 for the interpretation of the minuet, the pianist wants to know since she does not like wasting her time. Schenkerian deep layers and long lines cannot simply be brought out, can they? If the allegedly structural notes lack surface prominence, if they are insignificant or simply not there, bringing out the tonal "structure" would either sound quite strange or be impossible. Besides, if the structural notes are really important, i.e. structural, shouldn't the inherent tonal connections come to the fore without any interventions? So, she concludes, if tonal reductions are to be of any use at all, the connections highlighted in the graphs must somehow influence the playing indirectly, perhaps by regulating how the details of the music are related to the underlying connections.

Hence and assuming that Ex. 2 is valid, the rendering of the minuet must in some way reflect this particular analysis – at least to the extent that the interpretation does not just as well (or rather) fit in with some other reading of the music. Indeed, if tonal analysis is really such an asset for musicians, having a Schenkerian representation of the music in one's mind should make for unique differences in performance, differences to the better. But, she claims, the musical consequences of Ex. 2 are few, difficult or impossible to express, and unattractive.

For example, there are apparently two non-Schenkerian readings of the first four bars – competing readings, and yet both of them are quite convincingly supported by the musical events. You may conceive of the first two bars as an emphatic quasi-introduction circling around a1 , or you may give priority to the triad inherent in the rising melodic sequence so as to arrive at e2 . But how can you show your listeners that the passage actually brings the third-degree primary note, the c♯<sup>2</sup> and c♯<sup>3</sup> in "T 2", decisive for the initial period and as it turns out for the entire minuet, respectively?

What tonal momentum can the local primary note c♯<sup>2</sup> have, can it really assert itself against e2 all the way to the cadence in mm. 9–10? Or turning to Schenker's second-degree b1 in "T 8", how can it be played so as to suggest a connection back to the third-degree c♯<sup>2</sup> in m. 2 when the fifth-degree e2 (and its upper neighbour-note) is so prominent in mm. 3–9? And how can the listeners be made to appreciate a sense of closure in terms of a structural b1 –a1 descent when what they actually hear in mm. 9–10 is a swift fall from the fifth degree and/or a patently root-supported rise from the sixth-degree?

Turning to the encompassing *Ursatz*, crucial for the tonal unity of the minuet according to Ex. 2, it seems to depend on the fact that the b2 in m. 11 appears in the same register as the diverting, non-root-supported, non-tonic c♯<sup>3</sup> in m. 3. But how can you avoid bringing out that the swift *arpeggio* from b1 to b2 in m. 11 invites to be understood as an emphatic local motion to the upper neighbour-note of the preceding a1 ? Or for that matter, how can you avoid bringing out that the start in m. 11, due to this *arpeggio*, invites to be associated with the *arpeggio* at the very beginning of the minuet, a plausible connection suggesting a long-term a1 –b2 rising connection that makes for a link between the two periods?

Later on, is it possible or favourable to counteract the impression that the second period brings a quite patent close at e2 in m. 18, a cadence that, like the one closing the first period, involves an ascending rather than a descending motion? And no colleague of mine, she says, is prepared to understate the fresh start in m. 19 in order to somehow suggest that the passage mm. 11–30 essentially makes up a no-matter-the-double-bar falling b2 –b1 octave over a prolonged dominant. Nor would anyone want to somehow suppress the falling fourth of the bass in mm. 27–30, or try to tuck away the fact that the melody lands firmly on e2 in m. 30 in order to prevent this note from closing the middle section. Is it at all possible to play the b1 in this bar as anything else than the last note of a local transition down to a1 , can it at all be rendered as the ultimate offshoot of the dominant introduced back in m. 11?

Finally, in as far as Ex. 2 implies a commendation to the effect that the a1 in m. 41 is to be rendered as the tonally closing note of the minuet, it would put the last period, still and strongly insisting on e2 , in the shadow. And if you somehow managed to play in this way, what would happen to the formal symmetry of the minuet? The passage mm. 11–18 starting from b2 must have a counterpart.

The analyst refrains from giving the pianist any specific advice, but assures her that there is no need for worry. For just as the tonal connections highlighted in Ex. 2 assert themselves against various recalcitrant details of Mozart's compositional design – shifts in register, lack of harmonic support, actual voice-leading strata, formal demarcations, and the like – the tonal structure of the minuet will in virtue of being true emerge as manifest for any listener – for any true listener – no matter what the pianist chooses to do or not to do. For instance, the c♯<sup>3</sup> in m. 3 attaches quite well to the b2 in m. 11 in spite of the cadence in m. 10, and in spite of the fact that the pianist cannot do anything to clarify the connection. By the same token, the b2 –b1 *Koppelung* made up of a fifth *Zug* and a fourth *Zug* is present irrespective of Mozart's formal demarcation in mm. 18/19 and the merely local connective function of the final b1 in m. 30, and irrespective of how one plays. And yet, of course, it is essential that musicians are aware of such underlying tonal progressions colon how else could they express any "long lines" in their interpretations, how else could they show that the minuet is tonally unified?

The pianist is not convinced that a performance, which (somehow) manages to suggest a linear c♯<sup>3</sup> –b2 connection between the first two periods of the minuet, is necessarily better than a performance bringing them out as separate A-major and E-major units. Nor does she think that performances that (somehow) amalgamate the second period and the middle section are better, more structurally informed, than performances that do not.

In general, she does give some credit to the idea that connections emerging at deeper layers may provide an interesting framework for surface events, and hence that they may be of some use when it comes to interpretation; the notion of a mutual dependence between detail and framework makes sense. Underlying structures do lend meaning and direction to the details attaching to them, but without supporting details the structures emerge as analytical figments – Ex. 2 may serve as a warning example of the latter deficiency. In other words: if there is to be any artistically valuable influence running from the tonal structure to the compositional details, it is vitally important that the reductions are grounded in, and do not distort the musical design. And she adds that Schenkerian tonal analysis does not hold a monopoly of reduction or of discovering "long lines".

But the pianist also points out that it is important to keep in mind that the musician's primary and inescapable responsibility is towards the events appearing at the work's surface, including configurations residing not very far beneath it. This does certainly not imply that interpretation is a matter of details only, or that all insights gained from reductive thinking are useless. Largescale considerations are vital, but they must be fair, must never be imposed on the music.

This implies that music awaiting interpretation should be exempted from Schenkerian analysis – the analyst's attitude and answers have made her suspect that the reading of the minuet from K. 331 is not the only distorting reduction in the Schenkerian tradition. After all, when studying (say) a piece by Mozart, a musician wants to find out, not what is Schenkerian about it when its details have been removed or adjusted so as to fit in with a preconceived idea of tonal unity, but what is Mozartian in it, and hence worthy of and accessible for expression.

#### **Starting from scratch**

Having now learnt why Schenkerian analysis considers itself to be a valuable (indeed invaluable) guide to informed interpretation, and why musicians tend to be sceptical of Schenkerian theory, it is time to leave the analyst, and grant the pianist the privilege to help herself. 10 It is time to release the minuet from the analyst's couch, and place it where it belongs: at the musician's stand. 11

If one wants to understand what really happens in the minuet and to get some hints that may perhaps be useful for interpretation, the first thing to do is to dispose of the idea of a fundamental descent from the third degree; it has very scant support. Concurrently with the two-bar introduction, leaving and returning to a1 , the music starts with an ascent along the A-major triad up to e2 in m. 3; cf. Ex. 5 showing the "foreground" as well as indicating the deeper layers of the minuet. This note is promptly tonicized by the cadence to E major in m. 4. The top note g♯<sup>2</sup> in m. 5 is a falling appoggiatura, preventing a further rise up to a2 and introducing the upper neighbour-note f♯<sup>2</sup> , which leads the melody down to the appoggiatura a1 in m. 6, i.e. to the note where the melody started in m. 1. Later on and considering the f♯<sup>1</sup> in m. 9, the former upper neighbour-note proceeds up to a1 .

Thus, along with bringing a "primary" note e2 , supported by the dominant rather than by the tonic, the minuet's first period features three excursions from and back to the tonic note; it is very much about a1 . A further conclusion, relevant for the interpretation of this extended, ten-bar period is that m. 5 emerges as an added bar, as a mediating *cadenza*.

The consequent passage issuing into the cadence in m. 10 can be understood in three ways, and it is crucial that a reductive analysis purporting to be of any use for musicians takes them into account – to a large extent interpretation is a matter of choosing between alternatives.

The fifth degree (assisted by its upper neighbour-note) evidently holds its position until the rapid fall along the scale in m. 9. From a Schenkerian point of view, it may be objected that this descent lacks harmonic support, but the minuet has left the couch, and the very swiftness of the motion, offering a contrast to the long stay around e2 , makes it musically rewarding: it bears Mozart's stamp, not Schenker's. But mm. 6–8 may also be heard as a new and frustrated attempt to attain the upper tonic: there is a rising sequence of three appoggiatura motifs that is eventually blocked by g♮<sup>2</sup> -instead-of g♯<sup>2</sup> ; the ascending motion is then diverted by the unexpected low-register f♯<sup>1</sup> , and finally replaced by the falling scale. The

<sup>10</sup> Actually, she fires him, as anyone would do having hired a person bent to forgery of documents.

<sup>11</sup> When dealing with someone lying on the couch, the analyst has the prerogative of formulating the problem. From a Viennese perspective, whatever the client has composed, it amounts to an *Ursatz* with a descending upper line.

swift descent in m. 9 cannot hide the fact that mm. 9–10 also or primarily brings a rising line from the sixth-degree f♯<sup>1</sup> via the seventh up to the lower tonic, a motion that is supported by a complete and patent root-position cadence – a "fundamental ascent", perhaps pursuing the tonal ambitions of some distant fifth degree. 12

All three readings make sense of the abrupt shift in mm. 8/9 and locate the decisive events to the unexpected *subito-piano* cadence. Yet and paradoxically, the final two bars bring a sense of being appended. In retrospect the shift is/ was not entirely abrupt, however – m. 8 anticipates the subdominant in m. 9 by tonicizing it. The elements of discontinuity offer artistic opportunities that might be exploited; releasing a structural descent from the third degree already in "T 8" as suggested in Ex. 2 is not only quite unwarranted, it is bad timing as well. Schenker's reading, missing a number of striking features of the first period, emerges as utterly unmusical.

The second, E-major period issues from the fifth-degree b2 , and it eventually brings the treble down to a firmly tonicized e2 . Again the final descent is quick and harmonically unsupported – it takes place in m. 16, of course – and again the closing eighth-rather-than-first degree is reached from below by a motion from the sixth degree in mm. 17–18, a harmonically well-supported rising line that even more than the one in mm. 9–10 outdoes the less significant descent from the fifth degree.

Thus, there are interesting similarities between the two periods. However much this witty association contributes to the coherence of the minuet (as well as to its wilful character), and however much Schenkerian theory boasts of its unique capacity to disclose *Verborgene Wiederholungen*, there is no hint of this relationship in Schenker's analysis. 13 Both periods concurrently bring swift final descents from the fifth degree to the first *and* patently root-supported structural ascents from sixth to the eighth degree. But Schenker, occupied with his entirely mistaken *Urlinie* from the third degree, misses both of them. In addition to this sub-surface similarity between the two periods, a close-to-the-surface local resemblance with mm. 3–4 is suggested in mm. 17–18: a c♯<sup>2</sup> –c♯<sup>3</sup> –(e2 )–d♯<sup>2</sup> –e2 wink between the two cadences to the dominant.

<sup>12</sup> There is no such thing as a rising fundamental line in Schenkerian theory, but this is a problem for the shrink behind the couch, not for the minuet at the stand.

<sup>13</sup> For a discussion of these issues, cf. Bengt Edlund, "Hidden Repetitions and Uncovered Parallels", ch. 4 in *Analytical Variations*, Frankfurt 2020, Peter Lang Verlag.

In the two corresponding periods of the repeat section, some important differences should be accounted for in an analysis that is to be of avail to a pianist. The ascending sequence in mm. 36–38 does (provisionally) reach a2 , but this means that one step is skipped in the rising scale of top notes. As a result, the subdominant g♮<sup>2</sup> –f♯<sup>2</sup> appoggiatura, seamlessly attaching to the quick falling motion from e2 in m. 39, emerges as strongly implied. But instead of the expected arrival at a1 , this note is exchanged for the fifth-degree e2 . Just as in m. 17, there is a short "wrong-register" sixth-degree note in m. 40; then the rising structural line to the eighth degree is pursued in the lower register. But the *arpeggio*, making the last period start from the fifth-degree e2 , restores the upper register at the cost of the closing a1 of the lower line.

Hence, the first cadence of the repeat section is more complex than that of the initial section: mm. 39–41 bring more register shifts and the lines mutually deflect each other. Yet, although the swift falling-fifth line is temporarily interrupted just before its goal, and although the rising-fourth connection up to the eighth degree is broken in terms of register, there is a sense of continuity – due to the *arpeggio* the fifth-degree e2 is regained across the demarcation, thus forging the two periods together.

This is certainly not what we are made to see in Schenker's reduction, closing off the *Urlinie* and the tonal process at a1 in m. 41, and turning the rest of the music into an appendix. According to the reading advanced here, the final period emerges as a necessary constituent, discharging tonal tensions that are still present. The descending fifth put within parentheses in Ex. 2 not only suggests that an inadequate *Urlinie* issuing from the third degree has been imposed on the minuet from its very start, it also reveals that an important element of tonal coherence in terms of the fifth degree has been suppressed, that Schenker's miserable analysis has missed the insistent prolongation of the fifth-degree e2 , a crucial aspect of the minuet's "tonal" form.

As to the cadence in mm. 47–48, it first very quickly shows that the tonic is reached from below, a detail suggesting that the final structural motion reads f♯2 –g♯<sup>1</sup> –a1 , rather than b1 –a1 . As already mentioned, a virtually identical shift of register finishes off the theme of the first, variation movement of the K. 331 sonata. An intertextual association, or perhaps rising fundamental lines are more common than they are supposed to be?

Turning to the middle section, a sense of long-term sub-surface descent may indeed be present, but this intuition must be analytically established in a way that does not distort Mozart's music. Taking account of the unmistakable target notes of the bass, a falling progression b–a–e comes to the fore, a large-scale motion whose last stage is gravely misrepresented in Schenker's graph. According to Mozart, its first two stages (b and a) are introduced by three-bar applied dominants making for rising fourths in the bass, whereas its final stage (e) is rendered prominent by a stepwise descent, finished by the falling minor second of a (♮II)–V cadence. The emphatic quality of the outlet in m. 30 is substantially increased by the fact that it is preceded by two "Neapolitan" bars.

As to the treble, there is, in concurrence with the slow descent e2 –d2 –c♮<sup>2</sup> to be found in Ex. 2, a sense of overall ascent in mm. 19–27. The sequences of appoggiatura/resolution motifs make up two rising-third progressions (mm. 20–22 and 24–26), motions that are introduced and linked together, respectively, by the fast-pace rising thirds in m. 19 and m. 23. It is crucial to observe how these rising motions jointly form an extended connection, imparting a sense of ascending continuity to the treble. This long linear ascent, broken in terms of register after mm. 19 and 23, connects the e2 in m. 19 with the c♮<sup>2</sup> in m. 27, and each note in the bass enters a receding step below the immediately preceding member of the right-hand sequence.

Then, along with the activating falling motion in the bass starting in m. 27, Mozart changes tactics in the treble as well – but the long-term goal e2 is retained. Instead of three appoggiatura motifs forming a rising sequence as in m. 23, two such motifs now form a four-note descending scale. After resuming this motion in the upper octave and extending it to six notes, the leading-note d♯<sup>2</sup> is eventually reached at the end of 28, but the e2 target is achieved only after a deflection upwards to the lowered sixth-degree c♮<sup>3</sup> .

Thus, far from making up the second, falling-fourth part of Schenker's impossible b2 –b1 *Koppelung* straddling the double-bar, the middle section features a long and complex line connecting e2 with e2 like the stairs in an etching by Escher – a motion that does not amount to a "prolongation" any more than leaving the house for a walk and eventually returning home. It is also important to notice that the tonal procedure finishing off the middle section is a familiar one: the swift, melodically conspicuous but harmonically unsupported falling motion in m. 28, obviously heading for e2 , is interrupted and eventually outdone by a structurally prominent, cadence-supported ascent from the lowered sixth degree, appearing in the upper register, to the dominant. The affinity between the middle section and the surrounding sections of the minuet is further underscored by the rising sequences of three appoggiatura motifs in mm. 20–22 and 24–26, reminiscent of the ascending sequences in mm. 6–8 and 36–38.

# **Conclusions**

The reading of the minuet just accounted for is not intended to show the workings of an overall cadence lending tonal unity to all middleground voiceleading activities and by extension to all surface motions. No encompassing *Ursatz* (not even a flawed one) is to be seen in Ex. 5. Instead, and this perhaps belongs to the unique, Mozartian quality of the music, there are no less than five separate "tonal" progressions, one for each period/section. 14 These progressions represent in turn their A-major, E-major, towards-E-major, A-major, and Amajor tonalities; tonalities that are finally confirmed not by descending, but rather by ascending upper lines supported by local cadences featuring rootposition chords. But the minuet seems to have another unifying agent: the fifthdegree e2 , held as a structural drone to be eventually released in the first and fourth/fifth periods, eventually established as the local tonic note in the second period, and left-and-eventually-returned-to in the middle section.

In other words, the minuet makes up a coherent and reasonably unified Amajor piece of music. It is a pity that it cost Schenker a mistaken third-degree *Urlinie* to demonstrate its tonal unity. His reading leaves the pervading presence of the fifth degree out of account; and his imposed fundamental structure destroys the fresh start of the second period, degrades the fifth period that actually brings the tonal conclusion, and gives rise to a most questionable b2 –b1 coupling/prolongation expressing the dominant – an idea that obliterates the harmonic expansion of the middle section and obscures the form of the piece. In order to show that the music unfolds "organically", i.e. that it can be pressed into one of his standardized *Ursatz* corsets, it was necessary for Schenker to resort to unlikely readings of various details and to sheer manipulations. Why else did he enforce a misleading and confusing reduction that the music obviously resists, an analysis that leaves the listeners out of account and the musicians without assistance?15

<sup>14</sup> If you refrain from forcing it, the theme of the first movement of K. 331 exhibits a similar, non-unifying tonal design; cf. Bengt Edlund, "Analytical variations on a theme by Mozart"; cf. also "Syntactic vs. rhetoric structure in music", ch. 7 in *Questioning Schenkerism*, Frankfurt 2015, Peter Lang Verlag.

<sup>15</sup> Schenker was prone to condemn works that defied his analytic efforts, persistent and cunning as they were. It seems that he should rather have regarded the amount of analytic violence called for in the "successful" cases as indications to the effect that his analytic method and his concept of tonality were inadequate.

Considering the alternative analysis proposed in Ex. 5 from the musician's point of view, it appears that the connections it brings to the fore are possible and also rewarding to convey because they have not lost contact with the musical surface. There *are* "long lines" in this minuet – trust Mozart and respect the integrity of his text, and you will find them.

Giving the last word to our pianist, she is still not quite satisfied with the account of the middle section; it is still too Schenkerian for her purposes. It would have been better if Ex. 5 had brought out the appoggiaturas, not their resolutions, in mm. 20–26. She is certainly aware of the rule that "tonal structures" are not supposed to be built on dissonances, that consonances are to be retained when construing reductive connections, but she does not care since in music wine counts for more than water.

Understanding the right-hand melody in this way brings out the similarity between m. 19 and m. 23; cf. Ex. 6. The latter bar will now emerge as a transformed, more urgent copy of the former, and as an extra bonus this conception of the middle section makes for a continuity that underscores its sense of forward direction. Starting from e2 in m. 19, getting momentum in m. 23, and accelerating in mm. 27–28, a series of falling seconds make up a longrange descending melodic implication heading for e2 , a note that is concurrently delayed and precipitated by the c♮<sup>3</sup> –d♯<sup>2</sup> gap in m. 29.

# **Chapter 10 Interpreting a bagatelle**

In her "split-personality" essay on two Beethoven Bagatelles, Janet Schmalfeldt addresses a very important issue: what can a music analyst offer that is helpful to a performer?1 I found her study representative, thought-provoking, and – after some provoked thinking – requiring a thorough discussion paving the way for another and quite different understanding of the issues. 2 Apparently there are quite a few people who believe that Schenkerian analysis is a most valuable tool for musicians; hence the present text by an observer who is not a member of that church.

# **Is the form of Beethoven's Op. 126, No. 5 binary or ternary?**

In her analysis of the Bagatelle Op. 126, No. 5, JS (the analyst) uses tonal reduction in order to determine whether the form of the piece is binary or ternary, i.e. whether mm. 35–42 constitute a coda or a recapitulation – a decision that crucially depends on the status given to the cadence in m. 32; cf. Ex. 1. The investigation was occasioned by JS (the performer) who complained about the very short, G-major chord in m. 32 that both serves to lead back to start the repeat of the C-major middle section and to reinstate the main, G-major tonic for the final section. 3

In her graphs, cf. Ex. 2 A/B, JS shows that mm. 1–8, mm. 9–16, and mm. 35–42 all feature initial ascents from the first to the fifth degree, followed by the sixth degree. The "antecedent" touches c♯<sup>2</sup> as a lower neighbour-note of the fifth degree, whereas the "consequent", the "cadential phrase" mm. 29–32, as well as the (unnamed) passage mm. 35–42 use d♯<sup>2</sup> /d♯<sup>3</sup> as a passing-note between the fifth and the sixth degree. The "presentation" mm. 17–24 hovers around e2 , which in the "continuation" mm. 25–28 rises to e3 so as to introduce the "cadential phrase",

<sup>1</sup> Janet Schmalfeldt, "On the Relation of Analysis to Performance: Beethoven's Bagatelles Op. 126, Nos. 2 and 5", *Journal of Music Theory* 29(1985), 1–31. According to footnote 18, her graphs have been sanctioned by John Rothgeb. Schmalfeldt's analysis of Op. 126, No. 2 is discussed in "Interpreting another bagatelle" in Varia.

<sup>2</sup> Some years after this study, I wrote a further one on the relationship between analysis and interpretation, a text that employs Schmalfeldt's dialogue approach; cf. Bengt Edlund, "Dissentient views on a minuet", Chapter 9 in this volume.

<sup>3</sup> The shortness of the final eighth-note G-major chord in m. 32, cannot reasonably cause any serious formal problems; anyone at home in Beethoven's late music has encountered such precipitate cadences, for instance in the Diabelli Variations Op. 120.

bringing at long last the bagatelle's structural descent down to the first degree in m. 32. In Ex. 2 A/B the fifth degree is shown as being preserved all the way from m. 9 to m. 29, holding together the second part of the tonal form despite the intervening C-major "presentation".

This reading is supported by a Schenkerian *tour de force* connecting the "presentation" back to the "consequent" by means of a hidden repetition that emerges as crucial for the tonal form of the music. 4 The small-scale G-major d2 –d♯<sup>2</sup> –e2 passing-note motion in mm. 9–11 "foreshadows" the high-level d2 – "d♯<sup>2</sup> "–e2 motion to be seen if you join the G-major (I) d2 in m. 9, the B-major (V of the never-realized relative minor) tenor-voice d♯<sup>1</sup> in m. 16, and the C-major (IV) e2 in m. 24. As already mentioned, the latter note will eventually be raised to climactic prominence as e3 , which is then released by the descending fifth of the *Urlinie* closing the piece already in m. 32. JS (the analyst) concludes that the bagatelle makes up a binary form with an enormously expanded consequent and an appended coda, mm. 33–42.

Space does not allow of a detailed scrutiny of the reduction offered in Ex. 2 A/B. Suffice it to say that many of the voice-leading exercises displayed in the graphs emerge as valuable and valid only if you have a Schenkerian frame of mind. The main and intimately related issues in the present context are the large-scale "hidden repetition" and the location of the encompassing *Ursatz*, and in these respects this writer finds JS's analysis far-fetched and the conclusion based on it questionable.

The crucial hidden repetition sets in after the first interruption of the fundamental structure at the second-degree a1 in m. 8, and as a result the restarted *Ursatz* encompasses the music all the way from m. 9 to m. 32, engulfing the C-major episode as well as obliterating the much more obvious and tonally remote second interruption at the applied-dominant "third-degree" b1 in m. 16. 5 There is very little continuity between the half-cadence to B major and the harmonically deceptive start of the C-major middle section, and yet they are supposed to be connected by means of the "d♯ 2 " in m. 16 and the e2 in m. 24. To the extent that it depends on this questionable "hidden repetition", the survival of the fifth-degree *Kopfton* from m. 9 to m. 29 cannot but emerge as a most precarious analytic construct.

<sup>4</sup> For a critical discussion of "verborgene Wiederholungen" cf. Bengt Edlund, "Hidden repetitions and uncovered parallelisms", ch. 4 in *Analytical Variations*, Frankfurt 2020, Peter Lang Verlag.

<sup>5</sup> If you free yourself from preconceptions, interruptions can of course happen more than once, at other degrees than the second, and at other chords than the dominant.

Schenkerian analysis, unhampered by current listening (which is habitually disqualified as uninformed and obsessed with details), takes a pride in linking together remote events in order to demonstrate large-scale connections, but – and this applies especially when addressing musicians – it is a minimal requirement that the proposed readings work from a musical point of view. Considering the unexpected shifts often occurring in Beethoven's late music, there is nothing to explain or explain away in mm. 16–17: after the modulating close in B major, a Trio-like section starts in C major, i.e. in the subdominant, which is quite normal. The net (and wonderful) effect at the juncture is that the music is suddenly raised by a semitone.

A similar disregard of the obvious applies to JS's reading of mm. 35–42. Notwithstanding the retreat into the high register, the final eight bars, first modelled after the "antecedent", then owing crucial traits from the "consequent", clearly make up a shortened, all-in-one A-section of a tripartite form.

Turning particularly to the "coupling" e2 –e3 and the following descending *Urlinie* in mm. 24–32, some critical comments are warranted. The rising octave expends itself inconspicuously and unexpectedly: the e3 at the very end of m. 28 is root-supported, but it is metrically weak, "covered" by g3 , and introduced by a rising skip. And it turns up before the quite exposed, after-d♯<sup>3</sup> e3 , having subdominant root support and arriving at the main downbeat of m. 30. The d3 launching the structural descent, in relation to which the latter e3 makes up an upper neighbour-note, occurs already in m. 29. In other words, the e2 –e3 *Koppelung* and the *Urlinie* from d3 intersect in a way that seems problematic. Since the important e3 , i.e. the one in m. 30, clearly comes from d3 via d♯<sup>3</sup> , it should by rights be derived from the g2 in m. 26; cf. mm. 9–11 and 35–37. It may therefore be argued that there is a crucially important ascent starting from the root-supported tonic-note g2 in m. 26 rather than an e2 –e3 "coupling" issuing from the quasi-closing e2 in m. 24, i.e. from a note before the double bar-line.

As to the *Urlinie* of the "cadential phrase", its *Kopfton* d2 in 29 lacks root support. And unsupported is also the fourth-degree c3 at the end of m. 31 – the "voice-exchange" and "unfolding" symbols do not even add up to an oblique root support, and the C beginning m. 30 has already served as a fundament for the vital upper neighbour-note e3 . Considering what happens in between, do we really hear that e3 -over-C in m. 30 has somehow been exchanged for c3 -over E in m. 31, or that two voices are somehow involved in the bass motion C–C♯–D–E?

The most serious problem with JS's problematic analysis is that it blocks an alternative reading suggesting another form of the Bagatelle; cf. Ex. 3 showing the top voice and important bass notes. If we pay attention to the re-modulation in mm. 25–26 – it is unmistakably indicated by the changed key signature, the *cresc*. mark, and the fact that the long organ-point on c is abandoned – we will also hear that the g2 over the root-position G-major chord in m. 26 brings a new start, followed by a quite emphatic and active rising motion up to the fifth and sixth degrees, to d3 –d♯<sup>3</sup> –e3 in mm. 29–30. In m. 28/29 this most obvious line is for a moment eclipsed by two g3 's, but retroactively these seemingly excessive top notes become important: the *rinforzando* culmination on g3 may be taken to forebode what will eventually happen after the double-bar.

It is immaterial whether or not the rising motion in mm. 26–29 qualifies as "structural" in a Schenkerian sense: the main requirement is that it is conspicuous. And when it comes to interpretation, generally speaking, it is immaterial whether or not sub-surface structural connections comply with Schenkerian standards, as long as they are possible to bring out and have a meaningful function within a larger context – criteria that are satisfied by the ascent in mm. 26–29. The rise from g2 to d3 and then to e3 is clearly exposed on prominent metric positions, and both the starting g2 and the cresting neighbour-note e3 enjoy root-position support by tonic and subdominant chords, respectively.

It makes very good musical sense to regard the rising gesture starting from g2 in m. 26, not as a "continuation", but as a determined, and yet unsuccessful third attempt at the same endeavour that engaged the antecedent and the consequent, namely to reach beyond the sixth degree up to the eighth in a convincing way. The impatient gesture in mm. 28/29 brings the two g3 's prematurely: only the fourth attempt started in m. 35 may be said to succeed. Very quietly and after four insisting e3 's the melody comes to rest at a root-supported g2 in m. 42; cf. below.

It seems, then, that JS's Schenkerian voice-leading analysis obliterates a patent and essential rising motion from g2 by letting it be swallowed by the e2 –e3 "coupling". Considering the fact that she takes account of ascents from the tonic note up to the fifth and sixth degree elsewhere in the bagatelle, it is all the more remarkable that she misses or suppresses the most emphatic manifestation of this motion. Apparently, it did not fit in with the antecedent-expanded-consequentplus-coda reading that she wanted to establish as the "tonal" form of the piece.

If the reading proposed in Ex. 3 is accepted, we have to rename the parts of the bagatelle. The C-major "presentation" emerges as a middle section – as already pointed out, it undeniably sounds very much like a Trio set in the subdominant – and the "continuation" and the "cadential phrase" combine to form a passage displaying obvious affinities with the initial antecedent and consequent. Bars 35– 42, being also clearly related to the antecedent and consequent, are upgraded from the status of a coda to make up an essential formal constituent that eventually overcomes the downward gravity of the sixth-degree upper neighbour-note, allowing it to rise.

And if this re-functioning makes sense, we have a ternary, quasi-ABA', form, an insight that JS merely suggests in passing when mentioning the similar cadences in mm. 7–8 and 31–32, and in mm. 16 and 42, respectively. Otherwise put, mm. 35–42 make up a second, varied recurrence of mm. 26–32, and the bagatelle closes with two formal units balancing the initial antecedent-consequent pair in an asymmetric, late-Beethovenian way, an A' pair being separated – and joined – by a two-bar interlude, mm. 33–34.

The first unit, mm. 26–32, of this varied A' period grows seamlessly out of the Trio-like middle section. Its apex recalls mm. 10–11 in the consequent whereas the smooth closing motions are reminiscent of mm. 7–8 in the antecedent. Bar 32 is open rather than closed: D prevails in the bass, taking the sting out of the last-moment root-position tonic sonority. Up to m. 40, the second unit of the A' period is virtually identical with the antecedent, but it features a d3 –d♯<sup>3</sup> – e3 motion; then it closes with a variant of the final semitone inflection of the consequent.

#### **A closing structural rise?**

Perhaps one reason for JS's degradation of the passage mm. 35–42 – according to her description it merely brings an addition to the binary (tonal) form – is the fact that it does not give sufficient support for a structural descent; cf. Ex. 2A. Starting already from the d3 in m. 35, a preliminary descent runs via the root-supported c3 in m. 39 and the root-supported b2 in m. 40, but it comes to a premature end since no suitable dominant-supported a2 turns up. The a2 starting m. 41 is a slightly dissonant resolution note over (say) the subdominant, and the dominant-supported a2 just before the bar-line is a non-existent virtual note; the a2 in m. 42 is an appoggiatura over the final tonic. Nevertheless, JS maintains that a 5–4–3–2–1 descent somehow takes place in mm. 40–42, a descent enjoying patent cadential support; cf. Exs. 2A/B. The dormant fifth-degree d3 from m. 35 is activated by the absent d3 in m. 40, and as the following parentheses show the next three notes of the descent are not present either.

If this extremely precarious structural descent makes up one of JS's arguments for taking the bagatelle to have a binary form closing already in m. 32 – the other argument is apparently the extremely questionable d2 –d♯<sup>1</sup> –e2 *verborgene Wiederholung* incorporating most of the middle section into the consequent – it is tantamount to saying that the piece (in spite of everything that speaks for a ternary form) cannot be ternary since Beethoven's way of finishing off the music does not attain Schenkerian standards. But we must recall that the "cadential phrase" did not fare much better in Schenkerian terms, and yet it was accepted by the theorist JS as providing the true structural descent closing the piece, although it made up a stumbling block to the pianist JS. It should also be kept in mind that no complaints as to poor closing qualities were (or can reasonably be) raised against the would-be subordinate and structurally deficient final descent of the appended coda.

According to JS, then, Beethoven's failure is twofold: not only did he compose a tonally decisive falling cadence in mm. 31–32 that was musically too precipitate to be properly understood, he also committed the mistake of writing a theoretically most problematic subordinate falling cadence in mm. 40–42, which otherwise, due to its excellent closing qualities, could be taken for the ultimate cadence.

Joking (not entirely) apart: what Beethoven apparently did not accomplish was to sufficiently destabilize the cadence before the repeat sign. In spite of all his efforts to do so by harmonic, melodic, and rhythmic means, a smart analytic theory eventually turned up that, when so required by a particular analyst's agenda, made it possible to ignore all surface evidence for the obvious conclusion that the Bagatelle is *not* finished already in m. 32. Furthermore, Beethoven did not manage to sufficiently stabilize the final cadence; despite all its closing rhetoric, its ample root-position support, and its reference back to m. 16, a shrewd analytic theory eventually turned up that made it possible to mobilize arguments for claiming that the final cadence after all lacks the structural qualities required to turn it into a tonally true ending.

But if the Schenkerian dogmatic is ignored, a convincing *rising* upper line can be quite readily found, a line eventually overcoming the sixth degree and reaching the eighth degree. 6 After twice exposing e3 with subdominant root support (mm. 37 and 39) and after twice touching g3 (mm. 39 and 40), the upper structural connection is at long last allowed to proceed upwards in mm. 41–42, where both the e3 and the seventh-degree f♯<sup>2</sup> enjoy root support; cf. Ex. 3. The structural rise is obviously deflected downwards in terms of register; see below.

This closing ascent is attractive not only for reductive reasons. It gratifies both local and distant melodic implications; the bagatelle may be understood as being about this finally achieved rise. The tonic note above the sixth degree is so far in piece never reached in a harmonically and rhythmically convincing way. The

<sup>6</sup> Whether this structurally important final rise from the sixth degree is accepted as an *Urlinie* or not, is of course immaterial.

bagatelle turns out to be unified by a series of similar, more or less frustrated upper-line attempts to reach beyond the sixth degree – it emerges a post-termpregnancy piece with a quite compelling "implicational" kind of unity that is both rewarding and possible to express.

Some readers may object that there is not actually any final rise to an eighthdegree g3 . But, especially when strongly implied, a rising line can of course be heard as being pursued and completed in a lower octave. 7 It should be recalled that corresponding twists are often to be seen in Schenkerian analyses whenever structural descents ought to be continued or finished in the proper, original register; such transpositions are also most suitable when doing away with unwanted rising *Urlinien*. (Generally, picking needed notes from other registers is quite common in Schenkerian analyses; cf. for instance the d♯<sup>1</sup> -read-as-d♯<sup>2</sup> in m. 16.) In this light, what Beethoven does in mm. 41–42 may simply but inaccurately be described as returning to the *obligatorische Lage*, set already in m. 1. But theoretical sophistry aside, the final g2 's in m. 42 are g3 's, musically speaking.

Alternatively, and turning to interpretation, it may be argued that the final outcome of the bagatelle's accumulating set of implicative rising gestures heading for the eighth degree should not be allowed to just disappear into the wrong lower register. It seems that everything depends on the pianist's ability to render mm. 39–42 in a way that is crucially different from how the similar passage in mm. 5– 8 was played. In the last part of the antecedent, the melodic rise from c2 up to g2 and a2 should be treated as a subordinate, diverting motion delaying the descent towards g1 , the elided starting-point of the consequent. In mm. 39 and 41, on the other hand, the motion and the skip, respectively, up to e3 should be played so as to suggest insistent attempts to let the high register outdo the concurrent lower line, represented by the intervening, deflecting b2 –a2 motion. And the final motion e3 –(d3 )–f♯<sup>2</sup> –f♯<sup>2</sup> /(a2 )–g2 –g2 –g2 will emerge as an ascent in spite of the register shift and the covering a2 if it is played so as to remind the listener of the a1 –(e2 )–a♯<sup>1</sup> –a♯<sup>1</sup> –b1 –b1 –b1 cadence in mm. 15–16. Thus, it is after all possible to give an impression of a final ascending gesture; although the bagatelle does close with a motion to an eighth degree that is suddenly transferred downwards, this does not affect the overall rising upper-line tendency of the coda.

<sup>7</sup> Another rising line, obviously reaching a deflected eighth degree is to be found in the variation theme of Mozart's A-major Piano Sonata K. 331; cf. Bengt Edlund, "Analytical variations on a theme by Mozart", ch. 1 in *Analytical Variations,* Frankfurt 2018, Peter Lang Verlag; cf. also "Dissentient views on a minuet" in this book

It may be questioned whether mm. 26–29 really makes up an expanded counterpart to the swift ascent to d2 in m. 1 (or m. 9). But the affinity with the two units of the initial period is supported by the following emphatic rise to e3 via d♯<sup>3</sup> in mm. 29–30, by the ensuing eighth-note descent featuring some identical or closely similar surface traits, and (retrospectively) by the way this associative relationship fits in with the attractive idea of the bagatelle as characterized, as driven by a recurring generative gesture being repeatedly denied its ultimate rising realization.

## **Form or tonal form?**

A categorical mistake seems to be involved when JS uses her location of the fundamental descent to mm. 29–32 as a decisive criterion when establishing the form of the piece. One of the main points of Schenkerian theory is that formal divisions – i.e. emanations of "form" understood in the current sense of the word – are independent of the "inner", "tonal" form deriving from the distribution of the *Ursatz*. One and the same "tonal form" may give rise to various "outer forms"; the inner form is essential while the outer is merely apparent. It is difficult to understand how this assumed general state of affairs can be reversed, how it is possible to conclusively establish the outer formal layout of a specific piece from the analytically derived distribution of its inner fundamental progression – unless, of course, "tonal form" is granted absolute primacy over form in current sense, unless the tonal form *is* the form.

The analyst JS holds that the true structural descent takes place in mm. 29– 32, and hence that whatever follows upon it must be understood as appended. A musician, on the other hand, cannot very well disregard the various cues indicating the formal divisions within the bagatelle. The "outer form" of a piece of music is very hard to escape and, generally speaking, it is not desirable to try to do so. In this specific case, it is as obvious that mm. 35–42 (no matter the transposition to a higher register) represent much more than a coda, as it is that the C-major "presentation" amounts to a Trio-like section, a middle section eventually transforming itself into a passage strongly reminding the listener of the antecedent/consequent.

It may be argued that the recurring and ever-more insistent ascending gestures from the first degree up to the fifth and sixth degree also belong to the tonal design of the Bagatelle. And in their capacity *not* as Schenkerian initial ascents, but as generative events gradually disclosing an ever-stronger urge for a continuation beyond the sixth degree, they suggest another form. There are four such generative ascents, and the last of them, displaying unmistakable similarities with the initial antecedent and consequent, starts in m. 35, i.e. after JS's fundamental descent closing her binary form, which is most asymmetrically divided already in m. 8.

The last and final ascent is crucial since – in addition to eventually offering a rhetorically quite convincing end of the piece and satisfying the implied tendency of a rise beyond the sixth degree – it strongly indicates a ternary form. After the middle section the piece is rounded off by two varied statements of the initial antecedent/consequent material, statements separated and concurrently joined by the two mediating bars 33–34. These statements are quite different from each other as well as from the initial antecedent and consequent; the first of them brings a culmination in terms of dynamics and widely separated registers, while the second, shimmering replica in its own non-straightforward and yet insisting way reaches the eighth-degree goal.

JS's idea of initial ascents up to fifth-degree primary notes is unfortunate since it suggests that the pitch-class E brings upper neighbour-notes bound to descend instead of making up sixth degrees aspiring to ascend. She makes us see, hear, and play rising motions taking us to the fifth-degree point of departure for four standard structural descents, of which the last one does not count when it comes to the tonal form, instead of inciting us to discover and convey four implicative gestures, of which three are frustrated and only the last one convincingly arrives at the eighth-degree goal.

There are several problematic aspects of JS's bi-partite, analytically derived (tonal) form of the Bagatelle. Since the second, more notable B-major interruption of the *Urlinie* in m. 16 is disregarded, the form is characterized by a very short initial section, and a quite extended closing one, starting already in m. 9 with the consequent ending in B major, featuring then a Trio-like episode set in the C-major subdominant, and concluding with a very expansive ascending-thendescending G-major motion reminiscent of the antecedent/consequent – i.e. a section made up of three constituents, disparate as to tonal centre, function, and musical content. It is very difficult for a listener to understand, and for a pianist to render, such a disparate construct.

The binary reading also severs the antecedent from its consequent in a most counterintuitive way; JS's analysis prevents them from forming a pair, which they obviously do – indeed, this is unmistakable since mm. 1–16 are to be repeated.

Furthermore, repeating mm. 17–32 means that the first eight-bar unit of the long and heterogeneous second part of JS's binary form is removed, and since we now start with the C-major episode, it turns out that the second B-major interruption is after all respected, and that we after all have a Trio. Finally, however tonally and hence formally decisive JS considers her fundamental descent located to mm. 29–32 to be, the bagatelle badly needs a further section to attain satisfactory balance and a sense of closure. Since it is impossible to stop after m. 32, the form is not completed when JS's *Ursatz* has expended itself.

Beethoven did his very best to sabotage JS's binary form; for a pianist to express it must be like swimming uphill. But "tonal" analysts are better off. Being in command of a theory that is always stronger than the music under study, they have learned to be insensitive to the stream of the water. Beethoven's sophistication comes to nil when compared to that of Schenker.

Turning to the ternary, quasi-ABA1 reading advocated here, it might seem strange that the second, varied "antecedent/consequent" pair, mm. 26–42, is divided by a repeat sign. But this turns out to be a problem that can be settled analytically, and that a clever pianist can deal with and turn into an asset.

JS points out that an analyst may propose different readings of a piece of music – which is true, but rarely happens – whereas a musician has to settle on and make perfect only one of them. 8 But this bagatelle (and a host of other pieces featuring repeats or restatements) offers opportunities to render more or less identical sections of the music in different ways, indicating different readings of its structure.

Consider a performance displaying first an expansive middle section keeping to C major all the way to m. 32, no matter the changed key signature, a performance expressing the e2 –e3 rising octave as well as the withdrawing falling line, whose very last note, the root-supported g2 , is rendered so as to connect back to the C-major subdominant in m. 17. Imagine then a repeat, featuring a shortened Trio section ending in m. 25, followed by a varied G-major antecedent, bringing a determined ascending fifth/sixth in mm. 26–30 and then a receding motion down to a non-conclusive root-supported a2 in m. 32, i.e. to a dominant issuing into two bars of suspense in the G-major tonic (mm. 33–34) before the serene consequent.

Such an interpretation of the ternary form would make for interesting listening; there is a special delight in discovering redefinitions of formal functions. In order to clarify the difference, the G-major chord in m. 26, the C-major chord in m. 30, and the last-moment G-major sonority in m. 32 have to be played so as suggest the dominant, tonic, and dominant of C major the first time, and so as to suggest the tonic, subdominant, and tonic of G major the second time.

<sup>8</sup> Cf. Bengt Edlund, "A Defence of Musical Ambiguity", ch. 2 in *Analytical Variations*, Frankfurt 2020, Peter Lang Verlag.

The transition 203

To subdue the modulation back to G major when playing the middle section the first time is a very delicate, but not impossible task. Later on it appears that the two cresting g3 's, unexpectedly turning up above e3 –d3 in mm. 28/29, have a crucial function when distinguishing C major from G major. If the g3 's are subdued when the second repeat is played the first time, the goal of the C-major rising octave e2 –e3 , started in m. 25, emerges as uncontested. As a result and bonus, the middle section will seem to end with an antecedent that demands but is denied its consequent, a situation giving rise to a sense of formal suspense. On the other hand, if the top g3 's are brought out the second time, they will complete the rising octave g2 –g3 , started in m. 26 and expressing G major. In this case, the antecedent eventually gets its consequent: after two delaying bars the formal suspense is relieved in mm. 35–42.

When returning to m. 17, the root-position G-major chord transiently turning up as the last event of m. 32 may be slightly brought to the fore, allowing the listener to re-function the resolving g2 into a dominantic upbeat to the subdominant. Perhaps, and unlike in m. 16b, it is preferable not to tie the eighthnote g2 in m. 32 into the dotted quarter-note starting m. 17, as suggested in the original edition, but not in the autograph. The second time m. 32 has to be rendered in a way suggesting an ending in the D-major dominant of G major, and so as to highlight the b2 –a2 –g2 motion inherent in the treble line. In other words, when proceeding to the concluding part of the piece, the six-four-chordfollowed-by-resolution aspect of m. 32 should be brought out.

# **The transition**

The transition in mm. 33–34 has been mentioned a few times, and this peculiar and seemingly inactive passage, which seems to set in so abruptly, should be shortly described; cf. Ex. 1. The sense of immobility is due to fact that the G-major tonic is (pre-)prolonged for two bars by means of similar mid-bar excursions to the dominant. Whereas the full sonority of the preceding bars is suddenly reduced to a single melody, continuity is well provided for: m. 33 unmistakably brings a transposed imitation of the upper right-hand melody while preserving the register of the lower right-hand line. The harmonic immobility is combined with a sense of accumulation, however. A lower voice, repeating the solo melody, is added in m. 34 while the upper voice, raised by a third, imitates the preceding bar. (In other words, an inverted-counterpoint relationship obtains between the sixths in m. 32 and the thirds in m. 34.) As to the link with m. 35, it displays both continuity and a sense of arrival: the last of the parallel thirds in m. 34 is retained as a left-hand double-stop whereas the starting notes of mm. 33–35, b1 –d2 –g2 , describe a rising G-major triad up to the tonic note. It is a challenge for the pianist to strike the balance between immobility, accumulation, and arrival, between abruptness and continuity in mm. 32–35.

It is an unsophisticated but quite valid observation that mm. 33–34 separate the final varied statement of the initial material from the rest of the piece. This fact may make for an alternative, asymmetric ternary form: a repeated Asection featuring a modulating antecedent/consequent pair; an eventually quite expansive and repeated B-section, held together by the left-hand figuration and finished off by obvious associations back to the A-section; and finally, delayed by two bars of tonic preparation, a shortened, shimmering restatement of the A section, in which the initial antecedent and consequent are brought together to form a single eight-bar unit. Thus, the bagatelle has a concurrent, straightforward ternary form upon which a quite convincing interpretation might also be based.

# **Conclusions**

There are two general lessons to be learnt from this discussion of form.

Firstly, whereas it might be true that there is a dialectic relationship between "inner" (tonal) form and musical form in current sense, it seems that the "outer" form must ultimately be given the upper hand. And when it comes to performance, it is after all the outer form that really counts – you cannot very well play along as if it did not exist or in ways that run contrary to it. Furthermore, and whatever analysts say, taking primary account of the outer form does not amount to superficiality.

Secondly, one should not try to establish what the "tonal" form of a musical work is without considering how, and if, it can be expressed. Interpretation may certainly benefit from thorough – and unprejudiced – analytic study, but analysis should in the first place be predicated on interpretation.

Ultimately, to choose one or the other of two (or several) available formal options may be to destroy an ambiguity that is there to be suggested and enjoyed. Being positive is not the best attitude when dealing with musical *koans*.

I have mainly been discussing the analyst JS's use of Schenkerian long-range tonal connections, supposed to give the pianist JS clues as to the form of the bagatelle. The gist of my objections is that if reductive analyses are to be useful when it comes to guiding interpretation, they must not be far-fetched. After all, we can (if we want) see much more in a score than we can hear in the music, and musicians have no very powerful means at disposal to invite, let alone force, listeners to notice or even to unconsciously grasp extended and analytically

#### Conclusions 205

sophisticated, perhaps convoluted, sub-surface connections rather than shortrange, obvious surface phenomena. As to hidden parallelisms in particular, they must display a reasonable degree of similarity and structural comparability in order to be meaningful for a musician or, for that matter, to be used as analytical evidence when it comes to formal considerations.

Some analysts have claimed that Schenkerian reduction is a valuable, indeed indispensable, tool for musicians when groping for a structure to guide their interpretations. But is it really the deep-layer structures, allegedly representing necessarily inherent, inevitable tonal forces that should be looked for in the first place? Wouldn't it be more productive to pay attention to connections and associations that complement or even defy the unifying workings of tonality as conceived within Schenkerian theory? Isn't it more important to identify motions pointing towards less remote events, actually forthcoming or imagined, or to discover connections that, unlike encompassing tonal structures, may be readily expressible and more directly capable of informing the musical process? And perhaps there are alternative, less standardized and less hierarchical ways of conceiving tonal structure than the *Ursatz*, ultimately predicated on the authentic cadence?9

Musical notation is open in many important respects – for instance, it seldom specifies matters of form, and it does not even hint at content – and this indeterminacy ranges from details to the total conception. Thus, when playing a piece of music, it is necessary to make up one's mind about its form and to find its content. In music, "form" and "content" are closely interdependent, and therefore analysis is not only a matter of eventually deriving content from a careful study of the form, but also the art of letting content elucidate the form. 10 Both intellect and intuition are tools of understanding, and without understanding there cannot be any interpretation of a work of music, perhaps not even a performance of it. 11

<sup>9</sup> Cf. for instance Bengt Edlund**,** "Disciplining reduction and tonalizing interpretation", ch. 2 in *Questioning Schenkerism*, Frankfurt 2015, Peter Lang Verlag, and "How could analysis be deconstructed by Chopin's A-major Prelude?",Chapter 5 in *Chopin, The Preludes and Beyond*, Frankfurt 2013, Peter Lang Verlag, as well as "Reduction and interpretation", and "Dissentient views on a minuet" in this volume.

<sup>10</sup> This view is advocated in an essay by Donald M. Callen, "Making Music Live", *Theoria* 48(1982), 139–168.

<sup>11</sup> Thomas Carson Mark cogently argues for a qualifying sense of "musical performance" in "Philosophy of Piano Playing. Reflections on the Concept of Performance", *Philosophy and Phenomenological Research* 41(1980/81), 299–324.

# **Chapter 11 Tonal structure vs. modes of continuation**

### **Introduction**

According to Thomas Carson Mark, performing a piece of music that someone else has composed is like asserting something by means of a citation. 1 The intention to assert something with the music is a necessary condition since it lends a particular kind of significance to the playing; without this intention the performance is aesthetically invalid. But since it is impossible to assert anything without understanding it, performance in this emphatic sense implies that you must understand the music. If you have no intention to assert anything, or if you fail to understand the music, you just play it through in the same way as you do when you merely quote what someone has said or written without meaning it, or without even understanding the meaning of the cited words.

Hence, performing a piece of music presupposes interpretation, but it is important to notice that interpretation is not a perspective that the musician adds to or imposes upon the music. The core of the kind of assertion that Mark has in mind is a presentation of how the musician has understood the music, a presentation of its structure and character. Whether this presentation is identical with what the composer intended the music to be and mean is immaterial for the notion interpretation, although people with normative minds are prone to demand a close agreement between the musician's understanding of the music and the composer's intentions. 2

Leaving Mark's views, what is it that the musician must discover in the score and must understand in order to truly perform a work of music? The standard answer is that you have to grasp the structure as well as the content of the music. But, ignoring a host of complex questions, it is obvious that the current dichotomy of form and content cannot be applied without qualifications to music. Given its poor resources when it comes to semantic reference, whatever "extramusical" meanings music may have must reside within its formal substance – if the content

<sup>1</sup> Thomas Carson Mark, "Philosophy of Piano Playing. Reflections on the Concept of Performance", *Philosophy and Phenomenological Research* 41(1980/81), 299–324.

<sup>2</sup> For a further discussion of these issues, cf. Bengt Edlund, "Sonate, que te fais-je? Towards a Theory of Interpretation", Chapter 2 in this volume; also published in *The Journal of Aesthetic Education,* 31(1997), 23–40.

cannot find a hold there, it does not count. It seems, then, that musicians must understand music in two ways, clearly distinct as a matter of principle and yet closely related. They have to penetrate deeply into the music in order to discover its structural properties, and they have to attach broader associative intuitions to the music in order to enrich and individuate the structure as conceived of in purely musical terms.

To find out what this dual approach might yield in terms of interpretational options, it will be applied to *Träumerei*, the seventh piece of Schumann's *Kinderszenen* Op. 15; cf. Ex. 1. But before doing so it is mandatory to find out how the structure of this piece turns out when it is subjected to Schenkerian analysis. If we are to believe quite a few adherents of Schenker's theory, this particular brand of analysis ("tonal reduction" as it is often called) is by far the most productive way to understand the structure of tonal music as well as an indispensable tool when it comes to the interpretation of such music.

# **Schenker's analysis**

Schenker's analysis of *Träumerei* was published in *Der Tonwille* in 1924,3 and it represents his thinking prior to *Der freie Satz* (1935), where his theory of tonal analysis is to be found in its mature form. While it is possible that he might later on have wanted to dissociate himself from some details in his *Tonwille* reading, it is most likely that he would somehow have provided Schumann's piece with an encompassing *Urlinie*.

Anyway, the route from "background to foreground" runs as shown in Exs. 2 a/d; the "*Urlinie-Tafel*" 2d brings a more detailed and recognizable account of the music. Since Schenker's order of presentation insidiously begs the question, one should preferably study his analysis in reversed sequence. Starting from Schumann's score, proceeding with the foreground 2d, and then studying the middleground 2b, one should check whether the background 2a is an acceptable representation of the music. If you want to maintain your critical distance when encountering a layered analysis of this kind, you have to study it as a reduction, not as a set of recursive prolongations. And if you wish to find out whether there are any alternative readings, you must behave as an analyst, i.e. start from the musical surface.

<sup>3</sup> Heinrich Schenker, "Schumann: Kinderszenen op. 15 No. 9 Träumerei", *Der Tonwille* 4(1924/4), 36–39.

Before starting the scrutiny of Schenker's analysis, a preliminary opinion is due. The background 2a with its three ascents/descents, one for each part of the ABA1 form, cannot but evoke suspicions. *Träumerei* is obviously made up of six separate but similar melodic excursions forming pairs, but this does not imply that the six episodes can be combined to form three tonal ascents/descents in the way shown in 2a. Starting like the first episode, the second episode brings the excursion further and in another direction, and corresponding observations apply to the other two pairs. The f1 in m. 1 or the g1 in m. 4 simply do not lead to any a1 /a2 in m. 5/6 as shown in 2a/2b, and again similar observations hold for the other pairs. These misrepresentations are particularly unfortunate when it comes to the second pair, involving a modulation issuing into a raised point of departure for the fourth excursion.

Schenker's reading is far from a satisfactory account of the tonal process, and hence it is questionable whether it can contribute to the kind of understanding that is a prerequisite for an informed interpretation. Indeed, having this analysis in mind amounts to an obstacle when playing *Träumerei*. Detailed reasons for this dismissal will follow, but it is regrettable that so many words must be wasted.

#### **The outer sections**

According to Ex. 2a the two A-sections bring ascents from the first to the third degree, followed by descents to the second and first degree, respectively. The first thing to strike an observer as odd when checking 2a against 2b is that, notwithstanding the obvious parallelism, the first (fifth) episode is not treated in the same way as the second (sixth) episode: the apex note a2 in m. 6 (22) is taken to be highly structural but not the corresponding top note f2 in m. 2 (18). Can a second-episode a2 really "cover" a first-episode f2 ? The explanation of Schenker's reading might perhaps be that accepting f2 (enjoying a very precarious last-moment tonic support) as structural would have made for an undesirable, competing local descent issuing from the eighth degree and leading down to a second-degree g1 in m. 4 (20). This suppression of an obvious and important tonal connection paves the way for 2b, registering the first (fifth) episode as entirely taking place in a low register and as merely bringing a rising second, f 1 –g1 , made up of the first and last note of the episode(s).

The rising/falling third shown in 2a is compressed in terms of register – the a2 in m. 6 is shown as "a1 " in "m. 5". But 2b discloses that the rising structural third f1 –"a1 " in 2a takes place in two registers, and that it is divided between two separate formal units. The "a1 " does not issue from the f1 in m. 1, but from the suppressed, restarting f1 in m. 5. As to the g1 in m. 4, it does not have any ascending relationship to the "a1 " – according to 2b it is in fact the final, dividing note of a local descent, not a passing-note, and the a2 belongs to the next formal unit (where it enjoys a very precarious last-moment tonic support).

Hence, the background ascending third shown to take place in mm. 1–"5" is not a musical reality: this connection emerges as highly questionable already in the middleground 2b – to say nothing of the "surface" that Schumann actually composed – and it does not become true by referring to the background 2a. A falling note signalling a patently dividing half-cadence (g1 ) does not turn into a rising, mediating event with passing-note qualities just because you look at the music a farther step away, as it were. The initial ascent up to the third-degree *Kopfton* shown in 2a is just an optical illusion due to the fact that Schenker allowed himself the dual liberty of writing "a1 " instead of a2 and of moving this note to "m. 5" so as to enjoy patent support from a root-position tonic chord (cf. 2c), a support that it does not have in m. 6, just as the f2 was not actually supported by the tonic in m. 2. Anticipated notes belong to the following chord.

The would-be tonic quality of a2 *alias* "a1 " in "m.5" is boosted in 2b by a non-existent f2 placed right below the top note a2 in m. 6. And to further erase the A7 -quality of the arrival at a2 , the preceding a2 is taken away – whereas the corresponding f2 in the first episode is preserved. In virtue of the added f2 the melodic descent in mm. 6–8 is represented so as to emerge as a series of parallel thirds, just as in mm. 2–4. But this strained move concurrently exaggerates and diminishes the similarity between the two episodes. In mm. 2–4 the first three thirds are actually soprano/tenor tenths while the following three thirds are to be heard in the right hand. Bar 6 starts with a solo melody in the soprano, followed by accented soprano/alto sixths in mm. 7–8. In the fragmentary foreground shown in 1c the annihilation of the A-major a2 is completed: the made-up third a2 /f 2 is moved so as to become the first beat of m. 6, where it enjoys F-major root support from m. 5, while A-major "c♯<sup>1</sup> " (c♯) is delayed so as to accompany the following "third".

The manipulation is apparent also in the *Urlinie-Tafel* 2d where the rising motif c2 –a2 actually starting m. 6 is omitted: it would have disclosed that the F-major a2 wanted by Schenker to represent the primary note of mm. 1–8 is a forgery; gone is the conspicuous dotted quarter-note on the second beat, the surprising A7 -note giving the second episode its tonal momentum. Hence, Schenker's would-be tonic-supported top note of mm. 1–8 owes its musical significance to an applied dominant. The difference can be tested at the keyboard: play so as to arrive at an F-major a2 at the main downbeat of m. 6 – this is very different from what Schumann wrote and it sounds quite bland, but it is the music that Schenker's analysis deals with.

This observation can be generalized. In mm. 2, 18, 6, and 22 Schenker uses the last-moment, anticipatory top notes of the preceding F-major tonics, not the prominent notes introducing the B♭-major subdominants, and the A-major and G-major applied-dominant chords. This normalizing, castrating analysis is achieved at the cost of constantly preferring water to wine, by suppressing the harmonic shifts that make up the essence of *Träumerei*. Indeed, his reading is apparently allowed to contradict his better musical judgement as a pianist: he advises his readers to play the left-hand broken chords before the second beats, thus tainting the anticipating notes with the subdominant and applied-dominant harmonies to come. (p. 39)

Turning to the second degree on the structural line's way downwards in the second episode, the "g1 -over-V" shown in 2a corresponds to the non-existent g2 over-V-of-V at the beginning of m. 8 in 2b; as to Schumann, he ventures to have but is not allowed to keep a d2 , cf. the consecutive fifths marked in in 2c. Since a second-degree "g2 /g1 " is needed for the structural descent, hence, it simply has to be there. Schenker explains why c2 -over-V, i.e. the final, actually dividing note in m. 8, cannot serve as a structural note (as was g1 allowed to do in m. 4): it is simply a fact that "the *Urlinie* moves by means of passing-notes within thirds or fourths".

In the sixth episode there is in fact a g2 in m. 22, a fairly inconspicuous note sharing its root-position V7 -of-V-chord support with the top note a2 . In order to get a stepwise structural descent closing the piece, Schenker (apparently for want of a better second-degree candidate) allows this non-dominant g2 , a resolutionnote with passing-note quality, to suppress the patently dominant-supported penultimate seventh-degree e1 (not "e2 ") appearing in m. 24 so as to arrive at the wrong-register tonic note "f2 "; cf. 2b and 2d. The seventh-degree was not desirable since structural descents never rise.

This reading is not convincing, nor is the fact that the account of the descent of this episode deviates so much from that of the similar descent in the fifth episode, cf. 2b and 2d. What Schenker sweeps under the carpet in the sixth episode is the musically and tonally crucial, next-to-final G-minor (II) unit within the descent, corresponding to the final F-major unit of the preceding episode – an exquisite substitution that apparently interfered with the final cadence he had in mind.

According to Schenker G-minor is virtually swallowed by its C7 context. In 2d the G-minor unit, actually issuing into a root-position II chord, is merely represented as a secondary sonority built on a neighbour-note B♭ in the bass. This means that the accented root-position G-minor chord in m. 24 is suppressed in favour of its weak-beat first-inversion predecessor, misleadingly displaced so as to look like a downbeat. And in Schenker's harmonic analysis, G-minor is resolutely annihilated: there is a G-major II♮ representing the applied dominant in m. 22, but no G-minor II symbol although this relative-minor chord unmistakably belongs to the broad cadence. 4 In 2b the G-minor root is present, one might think, but it is embedded into a progression of interior-voice falling thirds wrapped into the structural dominant erroneously supposed to be already present.

The G-minor deviation from the expected route has certainly far more structural and aesthetic weight than Schenker's stone-deaf analysis grants it – structurally it is essential for the retroactively understood sense of a final complete cadence. And quite important are also the corresponding, rising Fmajor motifs in the first/fifth episodes, likewise obliterated among the falling thirds in 2b.

To sum up the critical discussion of the A sections, Schenker's reading as it finally emerges in 2a suggests that the first/fifth episodes are merely launching pads for the would-be tonic-supported third-degree primary notes occurring only in the second/sixth episodes. This structural ascent-then-descent scheme is an entirely misleading account of the relationship binding the episodes together. The first/second and fifth/sixth excursions make up pairs because both members are engaged in the same endeavour, and because the second member of the pairs obviously, in terms of both pitch and harmony, exceeds the first. Anybody can and will compare these excursions, and anyone can and will hear that the rising sixth to a2 over A-major and G-major, respectively, is "more" than the preceding rising fourth to f2 over B♭ major. It is a pity that Schenker apparently did not (did not want to?) understand that much, that his analysis goes to such lengths to suppress the vital element of harmonic change that destabilizes the top notes, whether f2 or a2 – notes that inject tonal tension and bring forth the ensuing melodic descents.

<sup>4</sup> Schenker mentions the disturbing G-minor element within the final cadence in his comments (p. 38). The "composing-out" of the first-inversion G-minor chord prolonging the resolution of the suspended fourth within the dominant complex "seems to fall out of the *Satz*"; as an "end in itself ", it "lets us fear that the voice-leading has somehow gone wrong". What a strange way to describe what is indeed an "end in itself ", what is a musical stroke of genius! What really "goes wrong" in this passage is Schenker's voice leading, not Schumann's. How lucky we are to have a responsible German gardener who does not hesitate to remove with his pruning shears every poetic branch blocking the structural view! Perhaps we should allow music to remain an English garden, rather than be forced to become a French one?

# **The middle section**

If the net effect of the first (fifth) and second (sixth) episodes is like raising the ceiling of the tonal space, the relationship between the third and fourth episodes brings a sense of lowering the floor as well. Is it then illuminating to present the outlet into a fresh B♭-major tonic in m. 13, an outlet making up the *raison d´être* of the entire B-section, indeed of the whole piece, as a IV-supported "b♭<sup>1</sup> " within yet another eight-bar I–V progression, as is eventually shown in Ex. 2a?

Unlike the same-start-different-outcome pairs of the A-sections, the two similar middle episodes, both eventually exhibiting a kinship with the second episode, are forged together somewhat like a cause and its effect. The third excursion leads urgently into the fourth, and the compressed rising seventh in mm. 9–10 is exchanged for the rising octave in mm. 13–14, bringing a sense of relief. None of these properties emerge from the background 2a, representing the entire B-section as a rising/falling fourth.

To begin with it must be pointed out that already in 2d the top notes of the two episodes are again misrepresented by being moved to the first beat where they seem to enjoy a root support that they do not have. The manipulation is particularly flagrant in m. 9, where the e♭<sup>2</sup> is provided with a virtual, unwarranted F♯ filling in a bass motion that is as long-range as it is far-fetched.

Turning to 2b and the treble, the f1 –f♯<sup>1</sup> –g1 –a1 (–b♭<sup>1</sup> ) motion makes up the essence of the third episode in terms of inherent voice leading. But the Dmajor-supported "f♯<sup>1</sup> ", preceded by a subordinate a1 –g1 progression taking place under what is shown as an F-major top-voice seventh e♭<sup>2</sup> , is not easy to locate in Schumann's music. And this goes for the root-supported upper-line g1 , preceded by a falling-fifth motion, as well as for the following "a1 "-above-"f♮<sup>1</sup> ". But a careful study of the *Urlinie-Tafel* 2d sheds light on these middleground mysteries.

Already 2d hides away the fact that the melodic turning point e♭<sup>2</sup> is actually supported by a root-position C-minor applied-subdominant seventh-chord. This is a strategic suppression because it helps to obscure the fact that there is a complete, root-supported cadence to the G-minor chord topped by b♭<sup>1</sup> starting m. 11, an undesirable harmonic progression that Schenker could not allow to stand in the way for his four-bar rising *Quartzug* in the bass taking us from "f " to "b♭" (i.e. from F to B♭<sup>1</sup> ).

The *Urlinie-Tafel* also reveals that the "a1 –g1 –f♯<sup>1</sup> " motion in 2b is displaced in terms of both time and register. If we consult the *Träumerei* that Schumann composed, we can see that f♯ *alias* "f1 ♯" coincides with the second e♭<sup>2</sup> , and that the motion takes place in the tenor voice where it reads a–g–f♯, or rather g–f♯

if we respect the rest. Anyway, from its most inconspicuous would-be start in m. 9 Schenker's inner line in fact issues into the accented g in m. 11, bringing additional emphasis to the root of the internal and most patent cadence to G minor. But this g in fact coincides with and gives structural weight to b♭<sup>1</sup> (not g1 ) in the soprano, a note that Schenker subdues as just a member of the descending upper-line fifth progression in 2b, a progression that is shown to precede the G-minor root.

Evidently, Schenker regards the g1 beginning m. 12 as structurally important although this note – unlike the first-beat b♭<sup>1</sup> in m. 11 – is not simultaneous with the G-minor root (as incorrectly shown in Ex. 2b) but only weakly supported, destabilized by a second-inversion d. Apparently, he wanted a descending-fifth motion and a root position to boost the g1 that he was interested in, the g1 that he needed for the stepwise structural ascent laid bare in 2a. As the g–G slur in 2d makes clear, the root g in m. 11, i.e. the unmistakable first-beat, auxiliarytonic end-point of the hidden-away cadence, is to be understood as a "preprolongation" of the inconspicuous G on the fourth beat of m. 12, a note that in turn makes up a member of his ascending *Quartzug* in the bass, and that in fact enters only after the treble has in fact left g1 for d1 . Unlike Schenker, Schumann avoids providing a root support for g1 . Why not let the composer have it his way?

Schenker's two long-range rising fourths in Ex. 2b, the f1 –b♭<sup>1</sup> one in the middle register and the dotted-slur "f–b♭" one in the bass, are eventually coordinated in mm. 12–13 by a motion that in Ex. 1 reads g–G–f 1 /A–b♭<sup>1</sup> /B♭<sup>1</sup> . But in 2b the bass note A turns up as the leading-note "a1 " in the upper line, whereas the melodic upbeat f1 is relegated to the bass ("f♮") so as to provide the nonexistent root of an applied dominant leading into B♭ major. This is an entirely unwarranted exercise in inverted counterpoint committing two mistakes in just one manipulation, and concealing the undesirable fact that Schumann does not afford any stepwise continuity for Schenker's rising upper-line fourth, as 2a and 2b want us to believe.

The seemingly smooth, inner-then-upper-voice rising fourth shown in 2b actually reads f1 –f♯–g1 –A–b♭<sup>1</sup> . This is a tortuous line indeed, and Schenker regards especially the bass dip down to A and the hasty melodic upbeat f1 (shown in 2b as the root of the applied dominant) as emergency solutions by a secondrate composer, not quite able to meet the demands of tonal "synthesis". (p. 38) Evidently, the great Schoolmaster was unable to let the idea dawn upon his arrogant mind that his own analysis, forcibly imposing an *Anstieg* across a major formal boundary, amounted to a falsification. As is the habit of authoritative teachers, Schenker blames the pupil (i.e. the Master that should have been his master) for his own shortcomings.

As already mentioned Schenker adds an unwarranted F♯ in m. 10 of 2d, so as to complete the questionable, no-matter-the-intervening-cadence-to-G-minor *Quartzug* in the bass, a motion suggesting a chromatic progression to go with the chromatic inner-voice fourth emerging in 2b. This F♯ clumsily brings out a sub-surface element of consecutive octaves in the third episode. But why does the inner-voice connection feature a structurally important f♯<sup>1</sup> in 2b? Because the non-existent motion "a1 "–b♭<sup>1</sup> in mm. 12–13, demanded by the background 2a, "requires" a preceding stepwise motion up to g1 , which in turn must be mediated by f♯<sup>1</sup> due to the "roots D–G" (p. 37). And yet, its important role in Schenker's ascending fourth notwithstanding, the cadence to G minor is only "apparent". How can you both eat the cake and decline it?

Commenting on 2b, Schenker (p. 37) observes that the impending series of consecutive octaves between the two *Quartzüge* is avoided thanks to the "inserted thirds", "f♯<sup>1</sup> /d1 " and "a1 /f♮<sup>1</sup> ". However, since the d ("d1 ") in m. 10 is not a member of the lower *Quartzug*, but belongs to the unwanted, merely "apparent" G-minor cadence, and since the f♯ ("f♯<sup>1</sup> ") is merely a chromatic passing-note within the upper rising fourth, it might still be argued that the tandem rising-fourth progressions in the middleground 2b still begin with the consecutive octaves "f1 / f "–"g1 /g". And with the consecutive octaves "g1 /g"–"b♭<sup>1</sup> /b♭" it also ends because (needless to say) neither the non-existent bass note "f♮<sup>1</sup> ", nor the non-existent treble note "a1 " can be accorded structural status.

But since there are no *Quartzüge* in the third episode, there are no consecutive octaves to worry about. Due to the complete G-minor cadence Schumann did not compose any middleground consecutive octaves within the third episode; they are fabrications by Schenker, who is then at pains to explain them away. On the other hand, the truly structural, top-level event of the entire piece obviously involves deep-layer consecutive octaves: the f1 /F–b♭<sup>1</sup> /B♭<sup>1</sup> relationship between the starts of the third and fourth episodes, a fact that clearly emerges and is acknowledged in the background 2a.

The episodes mm. 9–12 and mm. 13–16 run in closely parallel paths apart from the fact that the harmonic relationship between the point of departure and the goal of the internal cadence is different (F major to G minor and B♭ major to D minor, respectively) – a subtle difference that most listeners are likely to miss. But the harmonic analysis as well as the voice leading in 2d and 2b turn out differently since Schenker insists that there is an underlying *Quartzug* in the fourth episode as well. In other words, and this amounts to an even greater mistake, he presumably wanted to demonstrate that, no matter the intervening cadence to D minor, from m. 14 on the music gravitates towards the last-beat dominant in m. 16. It takes a *clairécoutant* to know that much.

What the episodes certainly have in common is their internal iv(7)–V9 –i cadences. In the third episode Schenker takes account of the applied dominant – the D-major root is used in 2b only because it serves as the consecutive-octave repelling anchor for the unnecessary f♯<sup>1</sup> of the would-be upper *Quartzug*, but it is not allowed as a structural step in the bass since this would have destroyed the would-be *Quartzug* in the bass starting from the tonic root F in m. 9. Turning to the fourth episode, both the applied subdominant and dominant roots are necessary for the bass-voice *Quartzug* in view, but not the root of the preceding B♭-major tonic. The root-position D-minor chord in m. 14, an auxiliary tonic that would have given unwanted structural emphasis to f2 in the treble, is suppressed in 2d and entirely removed in 2b. The rising-fourth progression allegedly present in the fourth episode starts later than the corresponding would-be motion in the third episode, but the listener has no reason to hear the two passages differently.

The consecutive fifths in Ex. 2c are entirely a result of this unwarranted, noauxiliary-tonic *Quartzug* in the bass. Only if the D-minor auxiliary tonic in m. 15 is disregarded, are there any adjacent A-major and B♭-major root-position triads. It is not the top-down "duty of the next diminution unfolding to avoid" forbidden consecutives (p. 37), but the first-place, bottom-up job of analysts not to invent reductions with voice-leading flaws. And needless to say, when such flaws (if flaws they necessarily are) are actually present in the music, they must be unconditionally respected.

The harmonic analysis of the B-section given in 2d is questionable. Without top-down prescience tapped from the score there is no harmonic continuity between mm. 9 and 13 to support even a dotted connection between the I and IV chords starting the two episodes. And even less can, as shown in 2d, the embedded, applied minor subdominant chord in m. 14 establish a solid-line II–V relationship to the closing/mediating dominant in m. 16 as if the internal cadence to D minor did not exist. Rear-view harmonic parsing tends to be devoid of musical meaning.

#### **Motivic content and interpretation**

Exs. 2 a/d virtually lack motifs in current sense – only in his remarks on 2d does Schenker mention this aspect of the music (p. 38) – and yet it may be argued that the motivic content is a crucially important agent of continuity in *Träumerei*. (In order to fully realize this, a thorough bottom-up approach is required; cf. below.)

He observes that the initial iambic quarter-note motif is replicated in the higher octave as an eighth-note trochee, and in the second episode he identifies a four-note motif, a "new motif " forming a chain of imitations. As will be shown later on, this composite four-note motif is but a part of a larger motif with greater explanatory significance. The kindred third and fourth episodes are described differently as well as inadequately.

Schenker pays much attention to the fact that the first note of the pervading rising-fourth upbeat motif is shortened to an eighth-note in mm. 8 and 12, and to just a grace-note in m. 16. (pp. 37–38) Indeed, he regards this rhythmic contraction as an expedient that a true master of tonal diminution like Beethoven would never have resorted to. But it may be argued that Schumann's change of note values, introducing an element of variability and urgency, is favourable, and that it is possible (for those who want) to use the *ritardandi* in mm. 8 and 16 so as to approximately restore the "proper" duration of the upbeat. 5 As will be argued below, some lengthening of the eighth-note upbeat is commendable also in m. 12.

There are just a few suggestions for performance in Schenker's commentary, and they are not derived from his analysis. Indeed, in one case his advice makes up a kind of correction: the somehow-gone-wrong, G-minor rising gesture in mm. 23/24 should be "played in a tone of its own". (p. 39) This motif is gravely misrepresented, indeed tucked away, in Schenker's graphs, but when it comes to musical practice the G-minor deviation is evidently so important that it demands to be brought out by extraordinary means. This contradiction amounts to a concession to the effect that his analysis fails as a guide for musicians. Particularly if something "seems to fall out of the *Satz*", it must – since it is there and no matter the "*Satz*" – be correctly represented in any musically relevant analysis.

# **Preliminary observations**

It may be argued that it is both possible and legitimate to undertake layered analyses without adopting the Schenkerian agenda. Indeed, a reductive analysis carefully implemented according to a bottom/up approach is likely to yield better insights into the musical process, and hence to provide a more informed basis for interpretative decisions, than an enforced top/down application of preordained schemes. The truth-value of confessions obtained under torture is slight.

A non-dogmatic multilayered representation of *Träumerei* – an all-in-one graph that may be useful when it comes to interpretation – will be advanced in due time. But first a number of important input observations should be

<sup>5</sup> Alice would not have balked at a four-and-a-half bar in a common-time piece whereas Heinrich apparently had problems when visiting Wonderland.

presented. Some of them are quite basic, and some have already been mentioned; occasionally we will touch upon matters of "extra-musical" content.

Disregarding the repeat of the first eight bars, *Träumerei* is made up of six closely related four-bar episodes, each presenting a rising-then-falling melodic arch. Since the (relatively) contrasting third and fourth episodes follow parallel paths and are closely forged together, since there are clearly articulated cadences to the dominant in mm. 8 and 16, since the first and fifth episodes are identical, and since the sixth episode peaks in G major instead of A major as did the second episode, the piece primarily exhibits a symmetric 8+8+8 bar ABA1 form.

But lack of symmetry is more interesting than obvious balance, and interpretation lives by whatever elements of indeterminacy and tension that you can find. In *Träumerei* the basic formal layout is undermined – or rendered ambiguous – by competing, coexisting formal options.

The six episodes differ significantly with respect to their harmonic point of departure. Thus, all episodes, all rising melodic excursions, but one issue from F major, and the B♭-major point of departure of the fourth episode cannot but give rise to a sense of expansion and formal ambiguity. In terms of starts, the form of the piece emerges as an asymmetric 12+4+8 bar construct. On the other hand, if you pay attention to the way in which the melodic arches descend, rather than to their similar ascents, episodes two, three, and four exhibit a kinship suggesting another asymmetric organization: 4+12+8 bars. (This formal pattern comes better to the fore if one disregards the repeat.)

The top notes of the excursions give a meaningful over-all contour: the f2 of the first/fifth episode is outdone by the a2 of the second/sixth episode, and the b♭<sup>2</sup> of the fourth episode reaches above the apex of the third episode merely rising to e♭<sup>2</sup> . Thus, more modest upward excursions are followed by more expansive ones.

The long, exposed top notes e♭<sup>2</sup> in m. 10 and b♭<sup>2</sup> in m. 14 are dissonances demanding resolution, which implies that they are "non-structural" according to current tonal analysis. Nevertheless, their urgent quality cannot but make them vitally important from the performer's point of view. Being a major ninth, the a2 in m. 22 is also a (mild) dissonance, but the situation is likely to be understood in the same way as m. 6, where the underlying A7 chord is dissonant rather than the top note, and therefore the consonant g2 in mm. 22 has a concurrent sense of being a quasi-dissonant passing-note. In m. 6 as well as in m. 22 an extra eighthnote occurs in the upper line in comparison with the preceding prototype in mm. 2 and 18. Depending on whether the first or the third of the descending eighth-notes appears to be (is played as) added, the motion may seem to be either urged or delayed.

The chords supporting the top notes of the melodic excursions bring the most expansive and most remote harmonic event in each episode. All pianists are aware of these target chords and the tension they infuse, and analysts should consider this aspect as a constitutive property of the piece; the more or less distant chords provide the impetus driving the music forwards.

The upper-register rising fourths of the first and fifth excursions land on stable B♭-major subdominant chords, whereas the rising sixths in the second and sixth excursions open up a wider harmonic space. The poignant first-inversion A-major seventh-chord in m. 6 is resolved to D minor at the following primary downbeat. The soft G-major major-ninth-chord in m. 22, on the other hand, is not resolved to a C-major triad but, introducing a sense of delay, to a six-four chord, and it owes much of its remoteness to the fact that the listener is likely to expect an A-major seventh-chord after the rising sixth, just as happened in m. 6 – no pianist is likely to neglect the sense of sweet deception caused by this substitution. The sharp dissonances of the D-major and A-major minor-ninthchords in the third and fourth episodes, being root-position applied dominants in local cadences, push strongly towards their G-minor and D-minor resolutions, respectively. These minor-ninth chords are preceded by "anticipating" minorsubdominant seventh-chords bringing a sense of additional off-beat stress and directing the listener's attention towards the following dominants.

The third and fourth episodes feature left-hand replications of the initial steep melodic ascents in the right hand; then follow passages made up of a falling sequence of soprano/alto/tenor imitations, cf. Ex. 3a. These six-note motifs, beginning with an eighth-note upbeat, overlap each other. The second member of the sequences has a falling conclusion, which means that it ends on the same pitch-class as the others, namely on B♭ and F, respectively. If one accepts that the motif may be augmented in terms of duration and pitch, there is in the third episode an additional fourth imitative entry in the bass voice, starting from e♭ in m. 11 and issuing into the deep root of the B♭-major chord starting the next episode. In concurrence with these imitative descents, the other voices form iambic rhythms, local dominant-to-tonic repercussions deriving from the subdominant-to-dominant shifts in mm. 10 and 14, rhythms that seem to check the falling series of imitations.

A corresponding sequence of imitations and similar iambic rhythms appear in the second episode as well, a trait that links this episode to those of the B-section.

In mm. 10 and 14, Schumann indicates a bifurcation of the melody into separate soprano and alto strands, diverging motions that are brought back to unison when these lines seem to converge at the accented b♭<sup>1</sup> and f2 in mm. 11 and 15, respectively. The fourth-beat notes a1 and e2 in mm. 10 and 14 may seem to be shared by the alto and the steeply rising tenor; if so, the accented g1 and d2 in mm. 11 and 15 are preferably to be understood as belonging to the tenor.

Turning to the first (fifth) episode, the arrival of the top note, and then the descent from it, is underscored by the tenor a tenth below. The falling melody in mm. 2–4 is divided into three similar four-note motifs, the third of which is a transposed replica of the second. The second of these gestures ends openly, rising like a question – and yet its last two notes derive from the affirmative rising fourth ending the preceding motif. The third gesture is extended by one note so as to include a descending skip bringing a sense of arrival. As to the first motif within this sequence, its initial notes e2 –d2 have a double function: while forming an upbeat to the following c2 , they also make up an afterbeat to the preceding peak note f2 . The tenor voice seems to agree with the first and third soprano motifs since it moves in parallel tenths and octaves, respectively. On the other hand, since it runs in contrary motion to the second motif, the tenor may be taken to bring a sense of gentle opposition to the soprano melody; the second unit in the sequence may suggest a touch of hesitation.

This account is valid also for the sixth episode until the fourth beat of m. 23 comes up with a remarkable harmonic deviation. Instead of the first-inversion F-major chord heard in m. 3 and then in m. 19 – and thus likely to turn up once more – a first-inversion G-minor chord appears; the third motif of the sequence is simply elevated by one step as compared to the model episodes. But the effect is paradoxical since the substitution means that the melody of the second motif is in fact repeated at the same pitch; due to the re-harmonization the question seems to be restated in a mood that is both poignant and pensive, that gives a glimpse of another world, as it were. The third, G-minor motif is extended to seven notes so as to include the final dominant-to-tonic cadence. But it should be observed that the last three notes of this prolonged gesture concurrently bring the rising beginning of a fourth, unfinished motif; within the final seven-note slur there is a sense of a 4+3 note interior organization.

The bass strand of this (essentially) four-voice piece interferes significantly in the process at some points. In mm. 8 and 12, and particularly in mm. 4 and 20, where it enters already at the weak second beat, the activity of the bass gives rise to anacrustic motions linking the episodes together. The re-modulation brought about by the veiled motion a–b♭–c♮<sup>1</sup> in m. 16 is of course important, and so is the soprano a1 –g1 inflection. Complementary mediating strands at this very tight juncture are the alto's f1 –e1 , preferably thought of as issuing from the third imitation motif, and the tenor motion d1 –c1 .

# **An alternative bottom/up reduction**

The alternative reductive analysis held in prospect is highly due; cf. Ex. 3a. This representation of the structure of *Träumerei* is non-Schenkerian both as to intent and result, but not necessarily less "tonal". The "surface" is readily recognizable, and yet the graph is reductive in the sense that different layers are distinguished. Few notes have been reduced out of sight; all notes of motivic or expressive interest are retained.

If hierarchic top-down access is replaced by sequential understanding, *Träumerei* obviously consists of six melodic episodes, six rising excursions crested by various, more or less tension-producing target chords followed by variously formed, relaxing descents. The reduction to be proposed takes a pride in faithfully observing Schumann's musical design.

The interpretation of music is a matter of details as well as of extended connections, and with all respect that Schenker deserves, the musical process in a well-composed piece like this one is fed by its details rather than born out of its deep structure. There may be a tonal ghost inside the *Träumerei* machine, but the chances of finding it are bettered if we realize that God, not the Devil, is in the details.

In the first (fifth) episode, a local descent from the fifth to the second degree presents itself – due to the recurrence of the rising-fourth motif in m. 3, we are reminded of c2 after the intervening subdominant chord. But the music also suggests that there is a complementary descent issuing from the eighth-degree peak note f2 of the B♭-major target chord, a descent carried by three rising motifs, and eventually reaching the fifth-degree c2 .

The sixth episode features several disturbances, and it is not the task of a reductive analysis to putty up these ruptures in order to demonstrate a seamless continuity that the music does not exhibit. The G-major ninth-chord turns up instead of the A-major seventh-chord, at the bar-line mm. 22/23 the melody brings an unexpected skip from f2 to d2 , and the G97 chord is not resolved into a C-major sonority as one might expect. But the six-four chord that does turn up is familiar, and for this reason the lower structural descent from c2 (found also in mm. 3 and 19) seems to be resumed.

Then, most surprisingly, the four-note motif is restated at the same pitch but in G minor, a deviation from the F-major harmonization in the first and fifth episode. As a result the lower structural descent is arrested at b♭<sup>1</sup> , and it will not reach any further. As to the temporarily arrested upper line, the re-harmonization reduces the tonal charge of the top note d2 from that of a C-major ninth to that of a G-minor fifth. Due to the final three rising notes of the melody and to the obvious II–V–I cadence in the bass, this d2 may retroactively be understood as representing the sixth degree in F major. The upper line, brought to a standstill due to the suspending G-minor motif, has taken over and concludes the piece by leading up to the eighth degree, a motion pursued in the lower register. To be perceived and enjoyed, the complex tonal transactions in the sixth episode require that you slow down the tempo, which is what Schumann prescribes.

Turning to the second episode, both c2 and a2 will (after having heard the first episode) seem to start falling lines, but the lower strand is temporarily blocked by the A-major seventh-chord. Both connections can be retrieved in m. 7, pursuing their route downwards to c2 and e1 in m. 8; iambic cadences with prominent falling minor seconds bring out the crucial descending steps. The two falling lines are introduced one at the time by imitative entries in the soprano and alto voices.

Apart from the different relationships between the harmonic point of departure and the peaking target chord, and from the fact that m. 16 eventually brings a modulation back to F major, the third and fourth episodes are virtually identical. The tension, and forward urge, introduced by the poignant minor-ninth D-major and A-major target chords, preceded by minor-subdominant seventh-chords, is considerable, and it is further increased by the bifurcation giving rise to two melodic strands and to diminished fifths between the upper voices. The cadences to G minor and D minor in mm. 11 and 15 coincide with the converging motions to b♭<sup>1</sup> and f2 , respectively. In concurrence, there are important rising motions in the middle register, gestures that emerge as imitations of the initial melodic ascents.

After the cadences, the auxiliary tonics are repeated three times as secondinversion chords at accented positions, repercussions marked by iambic harmonic progressions. Since the six-note imitation motifs always issue into the pitch-classes B♭ and F, respectively, the starting downbeats b♭<sup>1</sup> /B♭<sup>1</sup> and f1 /F in mm. 13 and 17 emerge as anticipated: the modulation and the re-modulation are achieved by means of reinterpretations of retained pitch-classes. The immediate agent of modulation in m. 12 is the eighth-note upbeat f1 whereas in m. 16 a four-voice converging motion produces a C-major dominant seventh-chord out of a D-minor six-four sonority. Since the descending parts of the third and fourth episodes are characterized by harmonic stasis, a slight *accelerando* might seem appropriate. At the end of m. 12 the tempo must be slowed down to give weight to the short applied dominant, whereas a broader *ritardando* is required to convey the complex voice-leading in m. 16.

It seems that the analysis given in Ex. 3a does justice to Schumann's music – his voice leading is certainly much better than Schenker's *Stimmführung* mess in Exs. 2b and 2d – and that it may be useful for a pianist when dealing with the large-scale connections within the piece as well as with some of its details.

But shouldn't a reduction establish an encompassing deep structure, comparable to a Schenkerian *Ursatz*? Not necessarily, but in this case there is a comprehensive pattern inherent in the music, but it does not emerge explicitly from Ex. 3a, nor of course from Exs. 2 a/d. The main event in *Träumerei* is that the fourth excursion sets out from B♭ major, and an interpretation that does not make this change of tonal departure stand out as the very focus of the piece is bound to appear as defective in a vital respect – but it must be pointed out that a "negative", withdrawing emphasis in m. 13 works as fine as a "positive" affirmative one.

In contradistinction to a "tonal" analysis in current, Schenkerian sense, the "focal" reduction sketched in Ex. 3b accounts for the relationships between the starts of the episodes rather than for the more or less conventional ways in which they close. Hence, there are no deep-layer "structural" dominants in this reading; being merely routine penultimate members of authentic cadences, they simply do not score high enough as musical events. On the other hand, the background in 3b features a structural 4/IV subdominant (a very becoming tonal core in a romantic piece) receding by means of a stepwise descent, supported by a complete harmonic cadence effecting the retreat from B major back to F major. The piece closes with a rising upper line from the sixth to the eighth degree.

#### **Modes of continuation**

Interpretation involves, indeed requires, that the musician tries to find out,consciously or not, how the music proceeds, that he/she identifies and decides upon what we may call "modes of continuation". 6 It often appears that an evolving musical passage is ambiguous in the sense that it embodies several options as to how it will proceed, and a crucial aspect of interpretation is to choose among

<sup>6</sup> For a more detailed introduction to this concept, cf. Bengt Edlund, "Prelude to the Art of continuation", Chapter 12 in this volume.

these alternatives. It furthermore seems that it may be productive to characterize these modes of musical continuation, inviting to be associated with domains beyond music, in ways that transcend current technical description.

Unfortunately, music theory provides only a few, crudely generalizing concepts like repetition, development, variation, and contrast to capture the finely differentiated varieties of continuation met with in music. This lack of terms suitable for describing the ways in which music might proceed from one event or unit to the next is a serious drawback since it means that the musician has little or no access to categories that could sharpen his/her faculty of musical intuition. Sensitive analysis, interpretation, and appreciation of musical continuation require a more developed terminology, words bringing richer connotations: addition, confirmation, answer, echo, alternation, polarity, parallelism, opposition, complement, expansion, contraction, etc.

Some of these words bear affinity to physical changes or events whereas others rather refer to human actions or changes of mind. Indeed, some shifts of continuation may aptly be associated with reporting verbs, indicating that we are prone to apprehend music as if it were a kind of speech, a way of listening with a long tradition. Music is sometimes understood in terms of impersonation, and if it is impersonated, it may emerge both as a monologue and a dialogue (or even multilogue). The latter alternative is bound to influence the categorization, expression, and perception of musical continuation: it makes a difference if an answer is understood as a reply, or heard as a response coming from the same musical "persona". 7

The metaphoric character of these characterizations when applied to music is obvious and also essential since it is their transcending aspect that opens up for associative content and stimulates the artistic imagination. Most of these finely individuated modes of continuation are likely to emerge as variants of the four basic alternatives first mentioned, but the point is that access to a diversified terminology, to words having slight differences in connotation, will lead to a more discerning analysis and a deeper musical involvement, and eventually to a richer and more compelling interpretation. When keenly analysed with regard to the character of its specific ways of continuation, the music will yield more interpretative options than when understood in purely structural terms. There

<sup>7</sup> The relationship between personification of musical structure and interpretation is dealt with in Bengt Edlund, "Musical dialogue in a Romantic violin sonata", Chapter 14 in this volume; it is also to be found as "Forming a Musical Dialogue", pp. 144–170 in Beckman, Sven et al. (eds.), *Rytm och Dialog*, Göteborg 2003.

is, for instance, a musically vital distinction between a contrast that is also an answer and a contrast bringing a complement, and when you have made up your mind, the music will sound differently.

# **Options of interpretation**

Having established a number of important structural properties of *Träumerei*, what are its options when it comes to interpretation?8

The initial, rising thematic idea in *Träumerei* is not uniformly slurred, a fact that cannot but encourage pianists to consider various alternatives. 9 The notes f 1 –e1 –f 1 in m. 1 can be rendered as a passive, dragging neighbour-note motion, in which case the rising melody starts from a1 . But the notes e1 –f 1 may also be linked with the following a1 so as to form a three-note upbeat leading into the high register. The final rising motion c2 –f 2 –f 2 incorporates an anticipation of the weak-beat, i.e. the quasi-syncopated top note of the B♭-major target chord, a fact that cannot but also lend an upbeat quality to the first f2 , no matter the articulation slur suggesting that this note is an afterbeat. The c2 –f 2 trochee, indicated by the slur, may be understood as an urging replica of the slow initial c1 –f 1 iamb, and therefore it invites to be separated from the preceding rise and to be brought out. 10 Finally, the ascending melody can also, unimpeded by interior articulations, be allowed to soar from e1 all the way up to the second f2 , which is what the slurring in episodes 3–5 suggests.

<sup>8</sup> Needless to say, a number of important interpretational considerations will not be discussed here: matters of tempo, sonority, general character as well as the crucial question of how different modes of local continuation should be combined in order to make up a convincing interpretation of the piece as a whole. Mistaking dreaming for drowsiness, *Träumerei* is often played excessively slow, cf. Schumann's (very) fast metronome indication.

<sup>9</sup> Considering the aim of the present investigation, it is more productive to let the musical structure suggest its inherent options of continuation than to let the composer's instructions circumscribe the possibilities. In principle, it may be argued that interpretation should be granted a certain freedom *vis-à-vis* the interpretational marks in the score, cf. Bengt Edlund, *"Sonate, que te fais-je?* Towards a Theory of Interpretation", Chapter 2 in the present volume, also published in *Journal of Aesthetic Education* 31(1997), 23–40.

<sup>10</sup> Taking account of *Kinderszenen* at large, it may be a good idea to highlight the melody in m. 2 since the following piece of the set, *Am Kamin*, starts in a closely similar way.

Two general remarks are due before proceeding. Musical interpretation is to an appreciable extent a matter of identifying and expressing aspects that coexist in the composed structure, like the rabbit/duck in visual perception. Disregarding some cases where a sense of ambiguity seems to be intended, it is not satisfactory to offer the listener passages that are both-rabbit-and-duck (or neither-rabbitnor-duck). It is rather the duty of the musician to make up his/her mind and to show the listener either the rabbit or the duck by underscoring distinctive traits of the two options.

As to the start of *Träumerei*, it seems that most pianists prefer to let the episodes start with a long gesture up to the peak note. 11 Evidently, unprejudiced analysis is seldom used to outwit tradition, to find other options inherent in Schumann's puzzle picture.

Turning to the second remark, interpretational ideas are not only a product of analysis or musical intuition, they are also intimately and reciprocally related to the playing motions, to the musician's proprioceptive sensations and to technical matters such as fingerings. Indeed, interpretational decisions and manners of execution often emerge as two sides of the same coin. 12

The unimpeded-rise idea of the theme strongly suggests the continuouslegato fingerings 13–231245 or 13–212355; conversely, in the mind of the "proprioceptive" pianist these fingerings are expressive of the unimpeded-rise idea of the passage. As to the other options of continuation, they are just as likely

<sup>11</sup> Cf. Bruno Repp, "A Constraint on the Expressive Timing of a Melodic Gesture. Evidence from Performance and Aesthetic Judgment", *Music Perception* 10(1992), 221–242, and "Diversity and commonality in music performance: An analysis of timing microstructure in Schumann's *'Träumerei'"*, *Journal of the Acoustical Society of America* 92(1992), 2546–2568. Repp found that the timing profile of the inter-onset durations in the gesture e1 –f 1 –a1 –c2 –f 2 –f 2 as measured in 28 recordings was generally "parabolic", i.e. suggestive of physical motions having such a character, and that this kind of timing was also preferred by a group of listeners. It seems likely, however, that if this passage were taken to embody any of the alternative modes of continuation, other patterns of timing might have turned up – other durational inflections matching other meanings of the passage, other timing profiles that could have been appreciated by the listeners, given that they had understood the underlying musical meaning of these, less usual interpretations. Mainstream and immediately pleasing, "natural", choices are not a prerequisite for artistically valid interpretations.

<sup>12</sup> For a further discussion of these issues, cf. Bengt Edlund, "A Comprehensive Approach to musical idiomatic", Chapter 6 in this volume, and "The phenomenology of fingering. Structure and ontology in Chopin's F-minor Etude from Méthode des Méthodes", ch. 7 in *Chopin. The Preludes and Beyond*, Frankfurt 2013, Peter Lang Verlag.

to be discovered as by-products when trying out fingerings as to be found by means of analytical reflection, and to a pianist they will not emerge as musically satisfactory until the matching fingerings have been devised. The neighbournote interpretation seems fully natural only if you play 25–45/2355, whereas the quasi-imitative relationship between the initial iamb and the high-register trochee demands the fingerings 13–235/255 or, even better since it brings a manual sense of imitation, 25–235/255.

Turning to the culminations in mm. 2, 6, 10, 14, 18, and 22, several options present themselves.

Should the melodic expansion, the rising sixth, in m. 6 be prepared by infusing some extra energy already during the e1 –f 1 –a1 motion in m. 5, by giving an impulse that was not appropriate when playing the first episode merely issuing into a rising fourth? Or should the A-major seventh-chord, transforming the tonal quality of the peak note, be announced by bringing out the event where the first sign of the mutation actually occurs, i.e. the enlarged melodic skip c2 –a2 ? The fingering 13–212355, involving a slightly uncomfortable stretch to reach a2 , supports the former interpretation, whereas the latter is best served by 25–235/255.

The situation in m. 22 is altogether different: in order to distinguish this final melodic excursion from the A-major one in m. 6, the mild G-major ninth-chord should preferably be approached, perhaps only from the rising sixth on, in a way suggesting withdrawal. This is most welcome, even necessary, since you may need some extra time to stretch your hands to cover the widely spaced sonorities

Bars 10 and 14 must be distinguished from the corresponding bars of the preceding episodes as well as from each other. The upbeat entries of the left-hand imitations lend weight to the iambic subdominant-to-dominant progressions, and the rising middle-register gestures underscore the fact that the dissonant D-major and A-major minor-ninth-chords head for their tonics. Indeed, if the rising tenor voice is allowed to be prominent, the soprano/tenor imitations make for a considerable joint impetus directed towards mm. 11 and 15. As a result of this, the culmination points of these episodes may seem to shift from the applieddominant peaks to their G-minor and D-minor auxiliary tonics, respectively, to the chords of resolution which are then reiterated and gradually undermined by the receding sequence of iambic rhythms issuing into ever-more faint secondinversion chords.

Bar 10 brings a deviant; the final interval of the rising upper line is reduced to the minor third c2 –e♭<sup>2</sup> , a contraction that heightens the tension and drives the music forwards. This is also due to the dissonant minor subdominant seventh-chord, of course, but to bring out its poignant quality it seems that the preceding c2 , the unison point of departure for the right-hand soprano/alto bifurcation should be made clear. The unexpected occurrence of an alto voice to be pursued downwards may be prepared if the a1 in m. 9 is rendered as an upbeat announcing the forthcoming c2 –b♭<sup>1</sup> –a1 alto motion. The fingering that seems most idiomatic for the concurrent alto strand is 13–231321 since it puts the continued soprano rise c2 –e♭<sup>2</sup> –e♭<sup>2</sup> ♯somewhat out of focus. The diverging motions in the right hand should be followed up by giving some emphasis to the fourth-beat alto a1 in order to show that it leads to the unison note b♭<sup>1</sup> ; the strong-beat g1 beneath the latter note should be rendered as the final note of the rising tenor strand.

The first descending notes in m. 2 may be attached to the preceding top note as a passive gesture, or they may (preferably) be played so as to introduce the four-note motif, so as to clarify the imitative structure of the ensuing passage. In m. 6 the g2 starts the descent "too early", a fact that may be used either to evoke a sense of eagerness or to give an impression of a three-note lagging hesitation. A slight *tenuto* on the top note a2 will suggest a touch of transient rhythmic ambiguity – did perhaps the g2 enter on the fourth beat?

Such a "premature" start appears in the sixth episode as well, but the situation is different in three respects. A "too-early" g2 in a similar context has already been heard in m. 6. From a harmonic point of view the g2 in m. 22 is in fact a resolution, but it may nevertheless – if you clarify its weak metric position – sound as a passing-note. The third difference is highly relevant for interpretation: the unexpected gap between f2 and d2 can be played in three ways. The absent e2 can be used to express a transient state of inhibition immediately followed by resumption, and the three eighth-notes may either be rendered so as to belong to the prolonged ninth-chord or so as to make for a connection after the fermata. In the first case, these notes must be played slowly and quietly; in the second, they should be rendered *in tempo* with a sense of relief as becomes a return to familiar ground.

In mm. 2–4 the crucial aspects are the dreaming lack of closure inherent in the first two of the four-note motifs – the sense of irresolution may be underscored by giving some emphasis to the soprano/tenor contrary motion in m. 3 – and the arrival brought by the note added to the third motif. A hint of forthcoming closure can be suggested already from the start of the third motif if it is rendered in a slightly firmer manner than the preceding one, or if one slightly brings out the parallel motion in the tenor, now supporting the rise of the melody. But the fifth, added right-hand note might also enter inconspicuously without any prior announcement, a way of playing that fits well with a prominent bass entry, introducing a sense of dialogue.

Considering the beginning of this passage, the falling inflection e2 –d2 in m. 2 conceals the resemblance between the first motif and the all-rising second motif. Exploiting this fact, the latter motif may be rendered so as to gently rectify the former. As already mentioned, the contrary motion in the tenor voice suggests a sense of simultaneous opposition, but the tenor actually moves as it did when accompanying the first soprano motif, and it gives in to a parallel rising inflection only when the third motif occurs. The left-hand slur in m. 2 marks a falling fournote motif d1 –c1 –b♭–a, reminding us of the fact that the right-hand melody is ambiguous: the e2 /c1 –d2 /b♭ concurrently make up an afterbeat and an upbeat, and it is up to the pianist to strike the proper balance.

Turning to the sixth episode, the exact replica of the second motif, coloured in mm. 23/24 by the elevated G-minor harmonization, is a most charged moment. The qualitative change may be rendered either as a renewed, decisive effort or as a radical withdrawal; both interpretations are suitable for bringing the piece to an end. The more emphatic option provides the energy needed to expand the four-note motif into a concluding seven-note phrase and to bring a full harmonic cadence. A withdrawing quality gives rise to a third four-note motif with an un-worldly quality, not present in the first and fifth episodes, a motif with a three-note closing unit appended to it – a final gesture that may either be rendered so as to quietly suggest an incomplete four-note motif or be highlighted as a rising melodic cadence.

The descending passages in mm. 10–12, 14–16, and (as far as the parallelism goes) in mm. 6–8 allow of several modes of continuation. The upper line is certainly in focus when the descent from the peak note of, say, the third episode begins, but after the receding motion d2 –c2 –b♭<sup>1</sup> it gradually subsides until it merely contributes to the repeated iambic cadences that in vain attempt to hold back the steep downward motion eventually issuing into B♭<sup>1</sup> . The overlapping soprano/ alto/tenor/bass imitation chain c2 –b♭<sup>1</sup> –d2 –g1 (–a1 –b♭<sup>1</sup> )/e♭<sup>1</sup> –d1 –g1 –b♭(–c1 –b♭)/c1 – b♭–d1 –g(–a–b♭)/(e♭–d)–d–g-G-A–B♭<sup>1</sup> should be brought to prominence for both expressive and structural reasons, and because the frequent pitch-class B♭ serves as a link preparing for the fourth episode.

But in order to articulate this series of imitations clearly from its very start, the soprano motif must be played so as to intrude upon the withdrawing d2 – c2 –b♭<sup>1</sup> motion after the peak note. Also in the interest of clarity, one may either abstain from bringing out the alto motif, giving instead priority to the complete six-note motifs in the soprano and the tenor (starting an octave lower), or settle on curtailed four-note motifs throughout, letting the final two notes of the sixnote motifs be obscured by the first two notes of the next link in the chain. It should be pointed out that the falling d2 –c2 –b♭<sup>1</sup> soprano motion in m. 10 may make sense even if it is entirely disconnected from the melodic peak at e♭<sup>2</sup> since it can be played either as a soothing or a culminating imitation of the preceding c2 –b♭<sup>1</sup> –a1 alto motion.

In the fourth episode, the four-voice cadence issuing into the dominant in m. 16 requires a substantial *ritardando* and careful attention. The final third g1 /e1 is not included in the slur, but it reasonably makes up a resolution of a1 /f 1 , only the grace-note c1 provides the link to the fifth episode.

Finally, a reading will be presented that even more transforms the structure as it appears in the score, and that should therefore be used as a variant in the fourth episode rather than be introduced in the third episode. In m. 14 the rising left-hand imitation of the initial melodic ascent may, more or less blurring the receding descent from the peak note b♭<sup>2</sup> , be connected to the descending series of imitations, giving rise to a sweeping gesture d1 –c♯<sup>1</sup> –e1 –a1 –c♯<sup>2</sup> –e2 –g2 –f 2 –a2 –d2 – etc. If this way of playing were applied already in the third episode – causing some confusion since the model has not yet been heard – it would produce a truly grand gesture starting with g in m. 10, peaking at d2 and issuing into the B♭<sup>1</sup> starting the fourth episode in m. 13.

Some people are likely to consider the above reading of the third and fourth episodes as unwarranted or objectionable since it transgresses the strands given in the score. But devices that analysts should be reluctant to use may yield rewarding insights for musicians. The tradition of piano playing provides many examples of voice-leading re-arrangements uncovering latent polyphony in the music, and there is no reason to assume that the composers were less fascinated by, and less creative in discovering, such possibilities. As to this particular reading, it emerges as musically meaningful, and it may be argued that some variety might be favourable at the episode-four stage of the piece.

The four-member imitation chain in mm. 11–12 has a vital musical function: to link together the third and fourth episodes in a compelling way, and it serves a most important purpose since it subtly undermines the ABA1 regularity and the sense of neatly paired episodes: the B♭-major start of the fourth episode becomes the climax of the piece, and the third episode emerges as a preparation leading into the new tonic; cf. Ex. 3b. And yet, even if the start of the fourth episode is conceived of as the tonal focus and culmination in *Träumerei*, there is more than one way of rendering it: taking account of its more transparent register, it could also, as already suggested, be played as a less emphatic, shimmering complement to its forerunner.

Leaving the notion of climax as a matter of peaking tonal tension and formal outlet, *Träumerei* also features a poetic highpoint: the final episode with its unexpected G-major ninth-chord substitution and its equally unexpected Gminor re-harmonization of a repeated motif.

# **Conclusions**

This much about how various modes of continuation inherent in a well-known piece of music can be distinguished by means of a non-reductive analysis. But in practice it is often left to the player's musical intuition and discernment to decide how the music is to continue after events that seem to offer interpretative options involving forthcoming events.

Turning to the implementation of interpretational ideas, we have insufficient knowledge of what musicians actually do in order to clarify their intentions as to how they want the music to continue. Sometimes the effects can be brought about by minute, almost imperceptible modifications; sometimes the situations call for quite drastic interventions challenging the respect for the notation and its directions (or proposals) for interpretation. It would be of great interest to study in detail the means used by musicians when expressing various options of continuation, and especially to find out how they differentiate between modes of continuation that are quite close to each other in terms of content – provided that such fine distinctions can be conveyed with any precision at all.

As to the listener, the situation is quite complex since the apprehension of musical continuation is based on the musical structure, including the modes of continuation it may seem to embody, as well as on how these options of continuation are selected, understood, and expressed by a certain musician. Are the performances successful in conveying the very mode of continuation that the musician has identified in the music and decided to bring out?

There is a need to complement analytic studies like the present one, which should be regarded as a contribution to the understanding of interpretational decision-making, with further investigations dealing with the actual expression and communication of musical continuation.

# **Chapter 12 Prelude to the art of continuation**

# **Introduction**

There is probably a consensus within the musical community that interpretation may benefit from a thorough analysis of the score. A good sense of style and a keen artistic intuition are no doubt important assets for musicians, but traditions must sometimes be outwitted, and there may be insights that can only be gained by using our intellect. It seems, then, that excellence in interpretation presupposes analytic engagement.

On the other hand, it cannot be denied that there is also a distrust of music analysis among musicians and music students, and this holds for the stock varieties of musical description as well as for the sophisticated methods developed by music theorists. Analysis, sometimes couched in a discouragingly cerebral language, is considered by these sceptics to deal with matters of little relevance for interpretation.

This combination of need for and distrust of analysis amounts to a challenge that must be accepted. A better understanding of the relationship between analysis and interpretation holds out the prospect of two gains for music as a performing art. Musicians may be spurred to test new analytic approaches or to use the traditional methods more penetratingly. And if analytic strategies for informing interpretation can be found, the art of interpreting music might at least to some extent become something that can be learnt.

The problem of relating analysis to interpretation appears to be twofold. A framework must be developed that gives analysis its proper place in interpretation, a framework that accounts for its diverse functions when preparing a performance. And taking the sceptic attitude among musicians seriously, the analytic approaches must be modified so as to be productive. 1

It seems that musicians have quite often been fed with stones rather than bread, but there are methods that can be used if carefully adapted to artistic purposes and to the music in question. Needless to say, it must be legitimate in this context to apply the analytic methods with some discretion. A musical text is not likely to be optimally accessible for interpretation when analysed in a rigorous and systematic way. Quite to the contrary, the music may turn out to be

<sup>1</sup> Cf. Bengt Edlund, "Dissentient Views on a Minuet" and "Interpreting a bagatelle", Chapters 9 and 10 in the present volume.

richer if approached with less respect for the demands of strict theory, if studied with an attitude allowing of unorthodox applications, free interplay between analytic perspectives, and fragmentary but multifarious structural observations.

In the first section to follow some ideas will be presented that emerge as fundamental to musical interpretation and particularly to musical 'continuation', a core concept of interpretation. Since a comprehensive theory of interpretation is advanced elsewhere,2 this presentation, taking a few significant quotations as points of departure, can be rather brief. The bulk of the text will be devoted to a study of the Prelude in F minor from J. S. Bach's *Das wohltemperierte Klavier II*.

# **Interpretation and modes of continuation**

A performance of a work of music is distinguished from mere quotation [i.e. an instantiation of the sound structure prescribed in the score] in that the performance, though it includes quotation, is also the *assertion* of the material quoted. […] In music, the performer intends that the sounds he produces will be taken as having cogency, as articulating how things musically are. […] Granted the similarity between assertion and performance, we can say that one cannot perform a work without attributing to it some meaning [i.e. without some interpretation, without being understood]. 3

By describing musical performance in analogy with the speech act of asserting by means of quotation, Thomas Carson Mark convincingly brings out that interpretation, which in turn presupposes understanding, is a necessary condition for performance in an emphatic, qualified sense.

Reflection along purely musical lines leads to the same conclusion. In spite of many musicians' contention that "everything is in the score" – an exaggerated, normative statement directed against performances that are either licentious or unimaginative – there are lots of things, ranging from details to comprehensive matters of form, that are left open by the notation. This indeterminacy demands clarification, and whether they think of it or not, musicians spend some of their time making decisions that specify their understanding of the music printed on the pages. Excepting cases where vagueness or ambiguity may be called for, a passive or neutral, "objective", approach is likely to produce inferior results, to

<sup>2</sup> Cf. Bengt Edlund, "*Sonate que te fais-je?* Towards a Theory of Interpretation", Chapter 2 in this volume, and also in *The Journal of Aesthetic Education*, 31(1997), 23–40.

<sup>3</sup> Thomas Carson Mark, "Philosophy of Piano Playing. Reflections on the Concept of Performance", *Philosophy and Phenomenological Research* 41(1980/81), 299–324; pp. 312 and 317

give rise to performances lacking convincing shape and/or expression. Hence, interpretation entails choosing, consciously or not, between various options inherent in the score, options suggested by the score to observers who take a closer look.

Given the primary purposes of musical composition and performance, a musician must be concerned with discovering whatever expressive content a music work may have. […] A music work is necessarily ambiguous with respect to its expressive features. So while there is much for a performer to discover about how a work *should be played* in order to exhibit its expressive content, the discovery of such ambiguities requires the performer to make choices among a number of ways of filling in that content in performance. 4

Donald M. Callen rightly calls attention to the importance of expressive content in the process of interpretation. The expressive structure of a piece of music is no less ambiguous than its formal structure, and just as the structural configurations, the expressive content must be subjected to careful and discerning analysis to discover the inherent options of interpretation.

It is customary to start with a thorough study of the musical structure, and then to proceed to aspects of expressive meaning, but there is also a complementary possibility, that of letting expression elucidate structure. Whereas most varieties of structural analysis enjoy a high prestige as applications of (more or less) scholarly theories, and have a fairly strict terminology, descriptions of expressive content tend to be dismissed as subjective and irrelevant fancy talk put in an imprecise and inadequate vocabulary. And it cannot be denied that there is a need to develop a sensitive language doing justice to our intuitions of the subtle and intricate expressive processes met with in music.

But the idea that music exemplifies – indeed, exploits and glories in – aspects of change that are among the most fundamental and pervasive characteristics of living seems to me true. Music, we might say, is in essence *continuation*: the question is always where it will take us next, and every happening is marked by the essence that possibilities are opening or closing, that there is development or retrogression, that there is continuity or abruptness, doubt or decisiveness, hesitancy or determination, building or disintegration. 5

Monroe C. Beardsley draws attention to a core phenomenon in music appreciation. And continuation is of paramount importance in interpretation as

<sup>4</sup> Donald M. Callen, "Making Music Alive", *Theoria* 48(1982), 139–168, pp. 150 and 146.

<sup>5</sup> Monroe C. Beardsley, "Understanding Music", pp. 55–73 in Kingsley Price (ed.), *On Criticizing Music. Five Philosophical Perspectives*, Baltimore 1981; the quotation is from p. 70

well, for it seems that much of what musicians actually do when interpreting a piece of music can be described as selecting, intentionally or unknowingly, the manner in which the music is to be continued.

In a trivial sense a work of music necessarily continues once it has started, but the kind of continuation that Beardsley brings out, and that will be at the core of this study, is associated with perceptible changes in the mode of continuation. At some points, occurring densely or sparsely as the case may be, the musical structure suggests shifts, barely noticeable or quite conspicuous, as to the manner of continuation, shifts that demand or invite to a change of musical attitude from the musician in order to be effectively conveyed to the listener. And conveyed they must be for such shifts of continuation are aesthetically vital.

It is important to observe that in a good performance these changes in attitude, with their concomitant adjustments of playing characteristics, cannot be arbitrary. Normally such shifts are occasioned by some aspect of the musical structure, by cues indicating a certain change in the mode of continuation. 6 But it must be pointed out that these changes are seldom prescribed: as a rule, the scores do not explicitly signal shifts as to mode of continuation, nor do they specify what character these shifts should have as musical phenomena or in terms of human import.

To train our thinking towards this generalizing capacity and to control the myriad of potentialities we should bear in mind that at any point of articulation a composer has four basic options: recurrence, development, response and contrast. 7

What is required for the present purpose is not generalization, but specification, and the training proposed by Jan la Rue, writing about style analysis, must be reversed. To distinguish in terms of human import between finely individuated changes in continuation does not belong to the domain of music analysis as currently conceived, but since it makes up an extension of music analysis that is indispensable when it comes to interpretation, it can and must be brought closer to the intellectual discourse on music.

Music theory has so far been content with just a few, grossly generalizing concepts for changes in continuation, but the faculty of formulating adequate

<sup>6</sup> Interpretation includes other kinds of change of musical attitude than those that can be analytically motivated. Musicians may – and sometimes they do so with irresistible effect – interfere with the music in more unforeseeable and even highly idiosyncratic ways, and this is fine as long as their initiatives are compatible with the work's structure and seem to be in line with its content.

<sup>7</sup> Jan la Rue, *Guidelines for Style Analysis*' New York 1970, p. 80.

distinctions should be cultivated by actualizing and sensitively applying useful metaphors. In order to be successfully rendered, the finely differentiated modes of continuation suggested in a piece of music must be captured by categories of continuation that are finely differentiated as well.

The lack of a good descriptive terminology for musical continuation is a serious deficiency since the musicians are reduced to use just a few widely applicable but crude concepts when trying to do justice to their analytic insights or musical intuitions. To have words for finely individuated modes of continuation means to have access to connotations leading to further discoveries and to more informed interpretative choices. There is a subtle distinction between, say, a contrast that is also a response and a contrast that brings complementation or alternation, and when the musician has made up his/her mind, the music will sound differently.

Many of these categories of continuation transcend the music, imbuing its structure with content. Continuations may be associated with physical events or emotional changes, they may be thought of in terms of intentions ascribed to the music itself or to interfering outer forces, and if the music invites to be conceived of as a kind of dialogue, it seems apt to resort to reporting verbs to describe changing attitudes.

Changes as to mode of continuation are inferred from the score, and it is the task of the musician to locate these shifts, to determine what musical and human character they seem to embody, and to find the adequate means to render them convincingly. Since they are somehow suggested in the score, a substantial part of the analysis should be directed at identifying the (sometimes quite subtle) cues indicating shifts in continuation. Further analytic reflection may then help to uncover their musical meaning.

But the decision as to when and how often conceivable changes of continuation are to result in shifts of performance attitude belongs to the things that must be left to the interpreter's artistic discrimination. Possible changes as to mode of continuation can, and should sometimes, be understated or entirely suppressed. The very best interpretations seem to maintain a delicate balance between clarifying interferences and restraint.

#### **General premises for the analysis**

The analysis of the prelude as well as the discussion of its various options of interpretation, belongs to our own time. No attempt will be made to describe the music as Bach might have described it – the analysis will be modern in terms and application – and no arguments deriving from Baroque musical rhetoric will turn up in the discussion. The syncretism of our present ideas as to musical expression may be taken for granted; the once valid system of rhetoric signification may have fallen into neglect, but vital fragments of it have no doubt been transmitted to later epochs. Musical rhetoric, in a general sense and as an intuitive background pertinent to all speech-like music, emerges as one among many sources of understanding.

The options of interpretation to be proposed will hopefully emerge as reasonable by present-day standards, but they are not necessarily in keeping with the findings of "historically informed" practice, nor do they add up to an "authentic", stylish interpretation. While we now know much about 18th-century performance practice, we have much less information on what the interpretations of that day were like. This essay is not an exercise in performance practice, but an attempt to present and discuss a number of artistic options emerging from a careful study of a specific piece.

The distinction between performance practice and interpretation is a fundamental one. The former is a body of generalized knowledge, gained from compositions, from writings about music, from the properties of period instruments, etc.; the latter is made up of the artistic conclusions drawn from the text of an individual work. When in conflict, the demands of the work always transcend those of the style. God is not religious, and Bach never played in style – he created a style by transgressing what was given.

This does neither imply that interpretative decisions cannot form traditions (apparently they often do), nor that the discussion to follow cannot to some extent be generalized. However much the options of continuation to be presented are grounded in the particular structural configurations of the F-minor Prelude, the insights gained should not be unduly restricted. There is, from case to case, something to learn that may be applied to similar situations in other works.

In order not to diminish the scope for the interpretative options, the prelude is not thought of as being played on the clavichord or harpsichord, but on the piano, an instrument having access to gradual dynamics within a wide range. Indeed, since many points of the discussion are valid for a variety of instruments – and for a variety of compositions within as well as beyond the Late Baroque – it is not even necessary to think of the prelude as a keyboard piece. As it reads, or with minor modifications, it might just as well be conceived of as an ensemble piece.

#### **The F-minor Prelude: general observations**

Before discussing the options of continuation, some general traits of pertinence for the interpretation of the F-minor Prelude will be presented.

The score of the prelude is reproduced in Ex. 1; various signs to be explained later on have been added. Since all motifs, phrases, and other formal units begin in the middle of the bars or at the last eighth-note, all references to specific spots or passages in the prelude will for the sake of convenience be displaced. Starting half a measure ahead of the actual bar, "m. 21", for instance, refers to the measure-size unit beginning in the middle of m. 20.

Although virtually all motifs/units have a sighing quality, there are five distinct kinds of material in the prelude, marked with the letters A–E in Ex. 1.


In addition, there are some passages with cadential function (mm. 47–48, 59–60, 70), a transition vaguely reminiscent of both the A- and the B-material (mm. 61–62), and a *fermata*-like passage built on a diminished seventh-chord and effecting a drastic change of register (mm. 53–54).

Counting the number of bars and using the designations for the five different materials, the form of the piece turns out like this:


From a rhythmic point of view, the A-passages stand out from the rest of the prelude since they feature eighth-notes in contrast to the sixteenth-notes to be found elsewhere, and since the melodic motifs start with eighth-note upbeats. As already mentioned, the prelude is pervaded by one-bar sighs, and the core of the corresponding rhythmic groups is made up of an upbeat/downbeat configuration around the bar-line. The two-bar units of the E-material are exceptions of course, and so are, more notably, the two-bar cadences mm. 47–48 and 59–60, where a rhythmic reversal seems to occur: the *nominal* mm. 47 and 59 suggest strong/ weak patterns.

In order to identify where changes as to mode of continuation may occur, the formal demarcations at various levels must be localized. These spots, delimiting independent formal units from each other must be graded according to their weight/importance.

The prelude turns out to be predominantly regular at the low level, but especially in its second part there are some interesting high-level ambiguities. They will be discussed in due time taking the structural details into careful account, but they are disregarded in the formal synopsis above, showing the most straightforward reading.

A tripartite form vaguely emerges within the double-repeat bisection framework of the prelude. But it is hard to determine where the formal return actually begins (m. 45 or 57), or indeed whether the form is tripartite at all. When it comes to the crunch this depends on how the prelude is played – if the performer so wishes he/ she can clarify the form by bringing out the moment of return, if any.

Turning to the melodic design, it is readily observed that the prelude is replete with falling seconds and extended stepwise sequences.

The main motif, virtually always forming sequences, is typically a falling second arranged so as to give rise to a sighing pattern made up of an anticipating note and a descending appoggiatura; cf. the brackets in Ex. 1. In a few cases, and notably in mm. 11–12, the sigh motif is inverted so as to feature rising seconds. (For other transformations, see below.)

Extended parallel descending motions are to be found in mm. 21–24, 25–27, 33–34, 37–40, 49–52, and 63–66. The sigh motif is engaged in rising motions in mm. 1–2, 9–11, 29–32, and (on a larger scale) in mm. 9–16, whereas mm. 3–4, 11–12, 13–14, 15–16, 49–52 feature falling sequences of such motifs. Contrary motions turn up in mm. 61–62, 67–69, and also in mm. 35–36, where the soprano ascent reverses the previous descent, which in turn emerges as a reversal of the rise in mm. 29–32. Considering units of two-bar format, there is a slowly falling sequence in mm. 41–44, complemented by a faster, rising one in mm. 45–47.

The melodic motions in the prelude are predominantly stepwise, and the main exceptions stand out quite clearly – the leap up to a♭<sup>2</sup> /c2 in m. 3, and the falling sixths in mm. 37 and 38. The leap up to g♭<sup>2</sup> /e♭<sup>2</sup> in m. 15 is expected since it starts the last of four two-bar units setting in a third apart.

The prelude calls for analysis in order to uncover long-term connections. It must be pointed out, however, that the "long lines" to be presented are not tonal connections in a Schenkerian sense. 8 Based on salient notes – dissonances are often preferred to consonances – these lines rather make up long-range "implications" of immediate relevance for the phenomenal form and content of the piece. 9 Provided that they are truly inherent in the structure, and not analytic illusions, connections of this sort are likely to be appreciated by the listener, and to identify them is important since they should somehow imprint the performance.

The long-range implications in the first part of the prelude are self-evident. It is both obvious and meaningful that the initial rising implication that was deflected downwards in mm. 3–4 is resumed and pursued in mm. 9–16, although the extended rise does not turn out quite as expected. These connections are not based on the consonant resolutions, but on the dissonant anticipation/ appoggiatura notes since they are likely to be in the focus of the player's as well as the listener's attention. The passage mm. 21–24 features descending parallels in both hands, proceeding as chains of suspensions; in the left hand, a falling motion of this kind is then pursued until the cadence.

The extended connections in the second part are more complex. Singling again out the exposed starting notes of the phrases, there is a descending line (supported by parallel tenths) from g♭<sup>2</sup> in m. 37 (or 33) over f2 to e♭<sup>2</sup> in m. 43. Following the top notes of the ensuing units, an ascending line (supported by tenths) emerges, reaching from f2 in m. 45 over g2 to the repeated a♭<sup>2</sup> 's in mm. 47–48. If the sigh motifs in the lower right-hand register are selected as important, the same change in direction comes to the fore, and a complementary rising line will emerge, running a sixth below the upper voice and reaching from a♭<sup>1</sup> in m. 45 to c2 in m. 47.

<sup>8</sup> The value of Schenkerian reduction when it comes to interpretation can be contested; cf. Bengt Edlund, cf. Chapters 8–11 in this volume.

<sup>9</sup> The word "implications" has been chosen to refer to these connections since, having once started, they give the reader, player, or listener some more or less exact idea as to the future course of the music. This does not mean that all of them necessarily count as implications in the sense proposed by Leonard B. Meyer; cf. *Explaining Music*, Chicago 1973.

Both these strands are left in the air at the half cadence in m. 48 in a way that strongly implies a descending continuation. And this expectation is realized in the following emphatic passage featuring falling parallel tenths between the top voice and the sigh motifs in the bass together with a lagging right-hand descent in parallel sixths. The upper line is interrupted at d♭<sup>2</sup> in m. 53, but the progression may be taken to continue in the left hand by means of the insistent quasi-motivic parallel thirds bringing b♭ /d♭<sup>1</sup> down to a♭/f in m. 54. Alternatively, the right-hand sequence of sixths is transferred downwards within the diminished seventhharmony. In any case, the suspension/resolution chain of sixths is resumed in m. 55, and the descent finally ends on e♮<sup>1</sup> /g in m. 56, completing a huge fall from a♭<sup>2</sup> /c2 in m. 49. An extended falling motion in the bass from f in m. 49 to c in m. 56, using the bridge passage in mm. 53–54 as a transfer to a higher register, is also quite obvious.

Due to the unexpected melodic twist in m. 58, c2 is exposed as a note demanding a descent, and a swift relief of tension does follow. But again the descent overshoots the target note f1 : the cadence in m. 60 is deceptive, and the note e♮<sup>1</sup> , being exposed as both an anticipation and an appoggiatura, is quite prominent. While the dissonant e♮<sup>1</sup> is prolonged in the alto voice, c2 is again approached in the soprano. In m. 63, a most emphatic descent is launched, featuring a chain of suspended sixths in each hand, but after having inconclusively touched f1 in m. 67, the hands part company. The bass proceeds down to the dominant C while the soprano returns up to b♭<sup>1</sup> and then makes a halt at a♭<sup>1</sup> before it is allowed to establish the tonic note in a definitive way.

The extended and unresolved implicative lines suggest that, notwithstanding the possible points of formal return in m. 45 or 57, the prelude might also be understood as bi-partite, and they also indicate that mm. 49–52 make up the climax of the piece. In this perspective, mm. 53–54, far from being just a cadenzalike passage on a diminished seventh-chord, emerge as crucially important: this passage both brings a transformation of the main motif in the left hand and a rhythmically compressed, vital link within an encompassing linear descent. The function of the deceptive cadence in m. 60 is also clarified: it preserves tonal tension and necessitates a renewed descent from c2 .

Since they may be highly pertinent for interpretation, a number of passages involving variation and imitation must be presented.

The sigh motif is present just beneath the surface in mm. 37–40, 42, 44, and 45–46; the upbeats are changed rhythmically as well as melodically. This motif is also present within the right-hand figuration in mm. 55–56, and it appears quite disguised in mm. 53–54 as well. (Imagine the pairs of three-note left-hand groups as double-stop thirds; the appoggiatura core is omitted.) Again with the appoggiatura suppressed, it may also shine through in the righthand part in mm. 61–62. Furthermore, if the left/right hand configurations in mm. 5–8 and 17–20 are also thought of as incomplete motions (e♮<sup>1</sup> /g1 –[e♮<sup>1</sup> / g1 ]–f 1 /a♭<sup>1</sup> etc.), a series of sighs, reminiscent of the beginning of the prelude, will emerge, replacing contrast by affinity; this applies also to the left-hand part in mm. 33–36.

Turning to the section mm. 29–40, there is more continuity than one might think. The descending alto c♭<sup>2</sup> –b♭<sup>1</sup> –a♭<sup>1</sup> –g♭<sup>1</sup> and/or tenor a1 ♭–g♭<sup>1</sup> –f 1 –e♭<sup>1</sup> motions in mm. 35–36 may be taken to imitate the soprano descent g♭<sup>2</sup> –f 2 –e♭<sup>2</sup> –d♭<sup>2</sup> in mm. 33–34; another feature lending continuity to this passage is the falling-thenrising parallel motion in the soprano and the bass, lagging somewhat behind. The beginning of the phrase mm. 37–40 appears to be a free variation of the descent in mm. 33–34, and being made up of disguised sighing motifs the entire phrase subtly refers back to mm. 3–4. The tenor a♮<sup>1</sup> –b♭<sup>1</sup> in m. 33 may be heard as pursuing the series of leading-note minor seconds in the left hand in mm. 31–32, and the juxtaposition of materials in mm. 32–33 discloses the affinity between the sequence of sigh motifs and the following units featuring sixteenth-notes; cf. mm. 4–5 for the same shift of rhythm. Indeed, the left-hand accompaniment in mm. 33–36 might be converted into the eighth-note amphibrach rhythm of the sigh motif, and so might the right-hand melody in mm. 37–40, its falling sixths notwithstanding.

In the middle of the piece, a three-note particle, consisting of a falling/rising second, gains importance. Appearing first in m. 36 as the most compressed final stage of the convergent motion of the tenor and the bass, it seems to prompt the right-hand entry beginning m. 37. Later on, this motif turns up as the initial germ of the left-hand descents in mm. 41–42 and 43–44, suggesting a descending link (e♭<sup>1</sup> –d♭<sup>1</sup> –c1 ) back to the start of the left-hand accompaniment in m. 36. Thus, the tiny figure e♭<sup>1</sup> –d♮<sup>1</sup> –e♭<sup>1</sup> is crucial in many ways: it puts an end to the preceding left-hand converging motion by launching a left-hand descent towards the cadence in m. 40, and it anticipates both the immediately following right-hand entry as well as the left-hand initiatives after the cadence.

The last two-bar unit (starting from c2 ) of the rising sequence begun in m. 45 may be described an expanded variant of the preceding one-bar phrases that in turn represent the sigh motif. Alternatively, one may regard the whole passage mm. 45–48 as a varied recurrence of the first four bars of the prelude. From yet another point of view, the rising sequence in mm. 45–46 emerges as a reversal of the slowly falling sequence in mm. 41–44, the two-bar phrases being shortened to one-bar motifs. The sixths down to the accent in mm. 45 and 46 (being inversions of the rising thirds in the model in mm. 42 and 44) are also reminiscent of the five-note falling-sixth motifs introduced in m. 37.

There are many sequences in the prelude, and some of them may also be thought of as chains of imitations. Whether a recurrent musical motif or unit is presented by the same or by a different voice is unclear when you hear the music, and emerges as a matter of interpretation that cannot be solved by simply referring to the notation. Hence, the left-hand part in mm. 25–28 is notated as two voices imitating each other, and so it is in mm. 63–66, whereas the closely similar passage in mm. 67–69 is written as a sequence within one voice. 10 This observation holds for some of the main-motif passages as well, for instance mm. 1–4. Being notated as sequences in the score, such passages might in various ways be interpreted as chains of imitations, as made up of interjections in a dialogue.

We will finally turn to some observations on harmony and elements of chromaticism.

The first part of the prelude ends in, and the second part sets out from, A♭ major. This use of the relative major is certainly a conventional trait, but one should notice that A♭ major is established, although in a rather in-conspicuous, passing manner, already in m. 12, and that it functions as a latent secondary tonic from there on.

This observation gives rise to two alternatives. If the cadence in m. 12 is thought of as defining a secondary tonic, the applied-dominant E♭-major passage in mm. 17–20 emerges as corresponding to the C-major passage mm. 4–8. On the other hand, if no dividing articulation is introduced in m. 12, the rise and the harmonic progression along with it will start already from F minor in m. 9 and reach all the way to the foreign key of E♭ major, where the music is suspended for a while in a state of high tension. In any case, E♭ major is left in an unexpected manner that must be rendered convincingly: the figuration is abruptly changed in m. 21 by the A♭-major seventh-chord, functioning as a local applied dominant to D♭ major.

In addition to the final cadence and the quite transient one to E♭ minor in m. 32, the second part of the piece brings four cadences that must be carefully

<sup>10</sup> Why should, as a matter of principle, the option of a one-voice sequence be ruled out in mm. 25–28, and that of a two-voice imitation in 67–69? This is not to say that these disagreements in Bach's notation may not be intentional, and that the possible difference in execution is without meaning. Quite to the contrary: the former passage can take, and will profit from, the increase in information that imitative playing entails, while the latter passage features an important rising line in the right hand that should perhaps not have to compete with too much additional activity in the left hand.

weighed in order to balance closure and continuity. The cadence to the B♭-minor subdominant in m. 40 is harmonically and rhythmically fully developed, and might therefore interfere with the linear continuity of the music.

The cadences to the dominant in m. 48 and 56, on the other hand, are weak – in both cases, the F-minor tonic turns up at the following beat, but these chords are divorced from the cadences and function as points of departure. How much closure these two half-cadences should bring crucially depends on the formal function you have in mind for m. 49 and 57, i.e. on whether or not you think of the prelude as having a tripartite form. If you want to suggest a trisection of the music at any of these two points, one of the half-cadences has to be rendered prominent.

The cadence in m. 60 deceptively veers off into D♭ major. This suppression of the tonic can either be treated with discretion or be brought out, and the choice again depends on formal considerations – if the music from m. 61 on is to be played as a coda, D♭ major should be introduced with some emphasis.

A diminished seventh-chord makes up the harmonic essence of mm. 53–54 and 57–58; another such high-tension chord brings the motion to a halt just before the final cadence.

Since chromaticism usually means heightened emotional tension, the shifts between diatonic and chromatic writing make up an important feature in the prelude. Even single altered tones may have a poignant quality that should not be overlooked and underplayed.

Three passages in the prelude are predominantly chromatic: the falling sequence in mm. 25–27 preparing for the A♭-major cadence, the two descents departing from g♭<sup>2</sup> in mm. 33–40, and the highly dissonant combination of right-hand arpeggios and left-hand sigh motifs in mm. 49–52.

The chromatically raised bass notes in mm. 4, 16, and 31, leading-notes signalling forthcoming local half-cadences, invite to expressive treatment. Considering the cadence-like quality of m. 31 and its models in the first part of the prelude, the true cadence in m. 32, completing the four-bar metric unit in due time and establishing E♭ minor (instead of B♭ major as did m. 31), at first seems to make up a sequential addition apparently stretching the phrase. The note e♮<sup>1</sup> is introduced in m. 60, and it persists as a dissonance for two more bars until it yields to e♭<sup>1</sup> . This resolution coincides with the quite bold augmentedoctave motion from A♭ to a♮ in the left hand.

Three conspicuous chromatic clashes should also be mentioned: the alto f♭ in m. 27 giving rise to a transient but most poignant dissonance with the other voices, the e♭<sup>2</sup> /d♮<sup>1</sup> clash in m. 36, and the conflict between a♮<sup>1</sup> and a♭ in m. 63.

There might be readers who fail to be convinced by some of these analytic observations. Preferring the sense of contrast, they might want to deny that m. 5 has anything to do with the previous sigh motifs, that there is any imitative relationship between the soprano in mm. 33–34 and the alto or tenor in mm. 35–36, that mm. 45–48 display any significant similarity with mm. 1–4, or that the left-hand entries in mm. 53–54 vary the sigh motif while forming a link in an encompassing descending line. They might think that these relationships are coincidental, strained, trivial, beyond Bach's intentions, or whatever.

Analytic findings can rarely be strictly verified. Rather, they gain credibility and seem justified to the extent that they fit in with other observations and contribute to musical understanding, to the kind of insight one wants or needs. Analyses serving various purposes or issuing from various perspectives often intersect, but they do not cover the same ground, and there may be other things to a piece of music than those that are likely to turn up in, say, a Schenkerian tonal analysis, or those that may have entered the mind of the composer.

When analysis is used in order to inform interpretation, as it will be here, the value of the observations depends on whether they give rise to worthwhile options when it comes to playing the music. For example, the rising parallelthird aspect of m. 5, associating to the preceding sigh motifs, is a valuable analytic insight if it can be conveyed when playing the prelude, and if it makes sense when you listen to the passage. This is not to say that anything goes, but it appears that analytic ideas manifested in performances should not, indeed can not, be assessed by just referring to the structure as notated, but must be evaluated as aspects of the structure as heard. Only when played by a great artist do some analytic ideas work, do some, otherwise less convincing, relationships or connections appear real and compelling.

#### **Options of continuation; the first part of the prelude**

Having so far merely brought together analytical observations that may be pertinent to interpretation, these insights (and further ones to be presented) will now be applied.

For each point of demarcation a number of options of continuation will be proposed and (when feasible) be described with respect to content. Altogether 36 such demarcation points, associated with possible changes in the mode of continuation, will be discussed. (In addition, there are some further, more or less similar passages that do not need to be commented upon.)

Since you cannot shift mode of continuation arbitrarily or too often, and since you must always make sure that the shifts contribute to a meaningful rendering of the work as a whole, that they do not ruin the coherence of the music, the options of continuation must be carefully selected so as to match each other. These comprehensive aspects will be considered after the presentation of the shifts.

Before embarking on the account of the various options of continuation in order of appearance, before finding out what to do (or not to do) at points inviting to or allowing of a changed attitude, the pace of the music must be settled. A certain option of continuation may work only, or work best, within a certain tempo range; indeed, having chosen a certain tempo, some options might not even occur to you. Conversely, the modes of continuation envisaged by the musician may make up an important set of determinants when settling on the tempo.

As we all know, the tempo influences the character and emotional content of the music. If a very slow tempo is chosen, the F-minor prelude will take on a tragic and solemn character, which is fine *per se*, but in the long run (there are double repeats!) the music may seem tedious. On the other hand, if played quickly, the prelude will get an elegant, slightly elegiac quality. This is also quite acceptable but involves the risk that some exquisite details in the music may be passed over too hastily. The following account presupposes a moderately slow tempo, a pace suitable for expressing sadness. This implies a tempo characterized by two beats of different accentual weight in each bar, rather than inviting to additional, secondary weak beats between two equally strong beats. The tempo range aimed at corresponds to a quarter-note pace of, say, M.M. 46–56.

0 We do not reflect on our stylistic knowledge when it has turned into secondnature intuition, but already understanding the main motif as a sighing gesture involves interpretation. The sigh motif exemplifies interpretation in another sense than the one at the core of this essay, a sense of interpretation leaving no choices: the initial, three-note melodic configuration *is* a sigh motif – there is no musically defendable, alternative interpretation. Fundamental for its inescapable sighing quality is our understanding of what bar-lines mean. We know and accept that metric notation is normative – just move the first bar-line one eighth-note forwards or backwards, and this most human meaning of the initial motif will disappear. It is virtually impossible to think of another position of the bar-line than the one it has in the score, and when starting to play the prelude, an emphasis on the second eighthnote as well as a connecting articulation slur binding it to the third note will simply be there – musical details that Bach could safely assume that the musicians would understand.

The first bar, "m. 1", of the prelude can be conceived of either as beginning with the quarter-note in the bass or, relegating the left hand to the background, as starting with the right-hand third. This is perhaps not a choice that you are likely to consider when beginning to play the prelude, but since it involves a basic option pertaining to all later passages of this kind, it will be discussed here in general terms.

The bass note on the second beat of the notated, nominal bars may either be played as an afterbeat, allowing the following right-hand eighth-note to function as an upbeat – this seems to be the current way of playing – or be rendered as a quarter-note upbeat announcing the forthcoming entry of the right-hand motif. No matter whether the left-hand part in these passages is played as a series of within-the-bar trochees or as a sequence of crossingthe-barline iambs, the rhythmic patterning brings subtle consequences for the musical character of the prelude. If the bass brings passive trochaic afterbeats, the music will emerge as an ongoing sequence of short melodic motifs. On the other hand, if you opt for active iambic upbeats in the bass, this cannot but rob the right hand of some of its rhythmic initiative.

But however interesting the left-hand-afterbeat-then-right-hand-upbeat alternative may be, it should be pointed that solo keyboard players are not likely to consciously entertain this complex rhythmic constellation. But if the prelude were performed as (say) a trio sonata, the bass part would be free to insist on maintaining its own steady trochaic course in contrast to the sighing amphibrachs of the treble instruments.

1 The second sigh motif offers three different options of continuation that to some extent will determine later choices. Since it starts a rising sequence, it may be rendered so as to invite the listener to anticipate a further ascent along the scale. The fact that there will not be any further stepwise ascent (at least not in the way the listener is likely to envisage) is no argument against this option. You may very well, and sometimes with good effect, hold out the prospect of continuations that will not turn up.

On the other hand, if the local harmonic motion back to F minor and the return to the initial third a♭<sup>1</sup> /f 1 are put in focus, a transient sense of circularity and closure will come to the fore, a closure charged with the expectation that a complementary two-bar unit will follow – which it does. Playing the bass notes f and c as upbeats will help to bring out this option. But the second motif may also be rendered so as to suggest that it makes up an immediate replication, an option promising further one-bar units. Indeed, it introduces a constraint to the effect that the remainder of mm. 1–4 (and also mm. 9–16) are to be played as a quasi-dialogue in terms of one-bar interjections. 11

2 One way of doing justice to the unexpected raised entry in m. 3, which associates back to m. 1 since (speaking of the soprano) it brings an octave transposition of the initial motif, is to play it as an imitation of the beginning of the prelude, thus suggesting that the high-register melody represents a new voice. This option, frustrating the expectation of a further stepwise rise, is supported if m. 2 was rendered as a local cadence requiring a complementary two-bar phrase, and if the F in the bass is played so as to announce a nonroutine entry in the treble.

Alternatively, this way of playing may suggest that the actual alto in m. 3 represents the "true" soprano, and that the actual soprano is to be understood as an outer-voice replica of a virtual, below-the-"true"-soprano alto voice; in other words, m. 3 may be taken as an exercise in inverted counterpoint. 12 But the soprano leap may also be relegated out of focus by simply bringing out the c2 –c2 –b♭<sup>1</sup> motion of the alto voice so as to connect with and pursue the soprano's rising sequence started in mm. 1–2.

3 Whatever the previous choice, it seems necessary to respect it in m. 4. The soprano line, apparently belonging to a newly introduced top voice, may be rendered so as to complete an exhausted, descending two-bar answer, reversing the ascending motion in mm. 1–2. If, on the other hand, the alto line is given continued precedence in m. 4, it brings the melodic curve started by the soprano in m. 1 back to its point of departure, as if preparing for a new excursion of the same sort. This option obviously makes for a four-bar formal unit.

<sup>11</sup> This option, involving quite close imitations within what is basically a sequence, illustrates the importance of the tempo. The passage will not get the time needed to be understood as imitative in terms of one-bar interjections if the prelude is played in a fast tempo; conversely, the idea of a close interchange between two musical protagonists will gain in plausibility (and keep up the interest) if the tempo is slow. The imitative option can of course be patently brought out by means of instrumentation.

<sup>12</sup> This idea may seem far-fetched, but considering again a trio-sonata rendering of the prelude, the passage might very well be scored so as to have a continuous upper-thenlower voice. Generally, the parallel thirds and sixths in the F-minor Prelude do not call for any upper-voice dominance in dynamic terms – an interpretative cliché that may be inimical to the overall sonority, and that pianists had better save for other pieces/ passages where it is appropriate. Besides, being a top voice the actual soprano tends to be heard as prominent anyway.

The fresh interpretational choice actualized by m. 4 concerns the amount of closure, and this seems to depend both on prior decisions and on decisions as regards the future course of the music. If a local cadence was suggested in m. 2, a corresponding, but more important one is expected in m. 4. The fact that there is no high-register continuation in m. 5 means that top-line prominence in mm. 3–4 lends more weight to the half cadence and prepares for a forthcoming contrast. But considering the alto line, m. 5 offers a continuation in the same register, a fact that makes for understating the cadence.

Planning ahead, the closure wanted or needed in m. 4 depends on how the passage mm. 5–8 is conceived. If it is taken to be unrelated to the initial four bars, the half cadence may be rather pronounced. On the other hand, if the following four bars are thought of as a series of varied restatements of the sigh motif, they should not be demarcated from the preceding bars. The bass voice is also crucial. If the B♮ is given some expressive emphasis, releasing its upbeat and leading-note potential, the cadence to C major will gain in importance. Somewhat lengthened left-hand upbeats on F and B♮ will also check the musical flow and prepare for a cadence.

4 If m. 5 is to be heard as the start of a contrasting episode, the left-hand notes should be highlighted so as to mark each beat. But mm. 5–8 may also be taken to hide four sigh motifs, and if this idea is to be conveyed, the top notes of each right-hand figuration must also be given some emphasis, suggesting the presence of inherent thirds and reminding the listener of the manifest thirds of the initial sigh motif. 13

But there is also a general decision as regards the course of the piece involved at this point. The prelude is made up of quite a few, seemingly disparate sections that appear to be more or less abruptly juxtaposed. Whether the player wants to give the prelude a quasi-collage character by means of clear-cut differences, or prefers to bring out its unity by making smooth connections and by clarifying motivic affinities, this is the first opportunity to announce his/her intention.

<sup>13</sup> It is taken for granted that interpretative interferences of this kind are legitimate, i.e. that Bach's notation in mm. 5–8 does not preclude that the left/right-hand thirds e♮<sup>1</sup> /g1 and f1 /a♭<sup>1</sup> are brought out in m. 5. Generally, since notation (to the extent that it indicates performance details at all) can only specify one of several possible interpretative options, it seems to be an uncreative point of departure to hold that other ways of playing than the one that is (or seems to be) prescribed are excluded or unwarranted.


start in m. 11. This association is clarified if the bass in mm. 9–11 is played as trochees with somewhat stressed downbeats rather than as connecting iambs with prominent upbeats.


Whereas the right-hand entry in m. 13 is what might be expected in terms of the long-range, rising-thirds implication, the rising octave and then the rising second a♭–b♭ in the bass as well as the shift back to falling sighs in the treble make up deviations. Giving emphasis to the a♭♯means directing attention to the rising octave and announcing the following b♭, effects that suppress the sense of a local A♭-major cadence in m. 12 and make mm. 13– 14 more ongoing despite the fact that they bring falling motifs forming a descending sequence.

Later on, in m. 15, the entry on g♭<sup>2</sup> /e♭<sup>2</sup> , completing the long-range rising sequence but reached by a quite large skip, and being again introduced by a connecting rising octave in the bass, increases the sense of tension.

12 The entry of the final sigh motif in m. 16, featuring a raised top-line pitch and sixths instead of thirds, deviates from its immediate precursor in m. 15. As a result, the last two sigh motifs do not seem to form a pair. The final motif may either be rendered as an added, somewhat retarding effort to reach a new tonal level, E♭ major – in which case one should play the passage mm. 13–16 in a way that urgently leads up to this goal – or as a final, vague gesture, attenuating the musical flow to the utmost.


Later on, there is some additional interest in the alto voice which eventually via a chromatic alteration (f♭<sup>1</sup> ), making for a poignant dissonance, issues into a falling sigh motif along with the rising one in the soprano. Highlighting this chromatic detail does not compete with the dialogue involving the tenor and bass, landing on a prolonged D♮ contributing to the strange discord.

# **The second part of the prelude**


The most obvious way to render the series of right-hand entries in mm. 29–32 is to gradually underscore the rising melodic implication (eventually issuing into g♭<sup>2</sup> ) and to use the "additional" cadence in m. 32 as a means for increasing the tension. But mm. 31–32 may also be played as a precariously balanced pair corresponding to the closed unit mm. 29–30, as an added two-bar unit made up of two juxtaposed cadences, seemingly delaying the motion up to g♭<sup>2</sup> by suggesting that e♭<sup>2</sup> is a preliminary, deflecting goal. There is also, it seems, a third option: one may let mm. 29–30 form a restrained separate A♭-major unit, which is interrupted by mm. 31–32, intruding as a sudden, agitated two-bar upbeat to m. 33, an upbeat prompted by rising semitones in the bass.


a sense of opposition due to their descending direction, both these voices may also be taken to imitate the soprano's preceding descent from g♭<sup>2</sup> . If the alto voice is highlighted, it links in with the soprano so as to suggest a falling octave from g♭<sup>2</sup> to g♭<sup>1</sup> ; if the rhythmically delayed notes of the tenor are brought out, they will together with the notes of the rising bass (running in parallel tenths with the soprano) eventually converge to form the compressed motif e♭<sup>1</sup> –d♮<sup>1</sup> –e♭<sup>1</sup> , crucial for the continuity of the passages to follow.


strengthened. Those who want to underscore this similarity might skip the ornaments.


them as the ultimate outlet that the more and more densely paced preceding activity was heading for – you can use the right-hand leaps in the preceding bars to suggest a sense of accumulating, checking preparation. The cadence to the dominant may be given different degrees of closure, depending on what you think of its function within the overall form of the prelude.


attaching instead to what follows and giving rise to a sense of a stretched, too-early half-note upbeat. 14 The latter way of playing also introduces the e♮<sup>1</sup> –f 1 conflict as an insistent and dissonant drone in the alto register, over which the soprano resumes its activity with falling inflections subtly alluding to the sigh motif.

34 A similar, and potentially even more consequential, source of metric ambiguity is to be found at the beginning of the next passage. Whereas mm. 63–66 may be played in a iambic way exposing the soprano descent and expressing the right-hand harmonic units, the precipitately rising sixteenth-note motion in the left hand brings an out-of-phase, quarter-note upbeat that starts a competing, bar-lines-in-the-middle-of-the-notated-bars organization persisting all along to m. 66 and giving precedence to the tenor rather than the bass. It should be noticed that the harmonic differences in the left hand as compared with the quasi-parallel section mm. 21–24 support this reading: now the consonances (preceded by upbeat sevenths) making for perceived accents turn up on the second beats of the notated bars.

If the left hand is allowed to determine the meter – the tenor-then-bass dialogue, growing out of the A♭-versus-a♮ shock, is expressive and merits attention – the alto descent, located to what now appears to be primary downbeats, will emerge as more important than the seemingly weak-beat soprano descent. Whether a soprano-bass vs. alto/tenor asynchrony between the right and left hand as regards the position of the main accents can be effectively conveyed is uncertain, but it might perhaps be suggested by giving equal emphasis to the soprano and alto lines.

In any case, it is necessary to bring out the long-due e♮<sup>1</sup> –e♭<sup>1</sup> resolution of the e♮<sup>1</sup> /f 1 conflict, the motion launching the sonorous alto descent. It is a fascinating aspect of the tonal design that the activation of the alto line concurs with the startling rise in the left hand, issuing from A♭ but overshooting its target by reaching a♮.

35 Particularly if the previous passage has been rendered in an ambiguous way, m. 67 has to restore metric order and reintroduce an unchallenged iambic grouping across the bar-lines by means of an extended three-note upbeat in

<sup>14</sup> "Emphasis" refers to the phenomenal effect and does not say anything as to how it is achieved. Dynamic stress is only one of several means to suggest emphasis, and the listeners' attention may very well be aroused by a reduced dynamic level.

the left hand, A♭–c–e♭. 15 Supporting the notated meter at this point is all the more necessary since any impression that f1 has been prematurely reached in a conclusive manner, i.e. within a strong beat, must be counteracted.

36 The a♭<sup>1</sup> in the soprano deflects and arrests an ascent that, in virtue of being a counterpoint in contrary motion to the bass, is expected to reach c2 and eventually d♮<sup>2</sup> . Nowadays it is possible (and also quite acceptable) to include a firm B♮<sup>1</sup> in the diminished seventh-chord.

When played with a return back to m. 29 in mind, m. 69 must be linked to m. 70 in a continuous way. It appears to be a good idea (and reasonably not in conflict with Bach's intentions) to dissolve the left/right-hand diminishedtriad chords into a further sixteenth-note figuration so as to get rid of the impeding caesura caused by the block chords; cf. Ex. 2. On the other hand, when heading for the conclusion of the prelude, the sense of arrest brought by these chords is quite appropriate. Bar 70 should now unequivocally signal final closure, and this is best achieved by playing it as a separate concluding unit. 16

This much, and no doubt more, can be said about discovering and selecting modes of continuation in Bach's F-minor Prelude. In passing, some suggestions have been advanced as to what one might actually do at the keyboard in order to bring out a certain option of continuation in contradistinction to other possible ones. But there is no reason to suppose that there are generally one-to-one correspondences between modes of continuation and ways of execution; the relationships between means and ends tend to be more complex. Disregarding minute matters of shade and nuance, the means appear to be comparatively few whereas the modes of continuation to be expressed are many and finely individuated. But the cues of expressive performance are likely to be both ambiguous and interchangeable – and they may presumably be combined to form musical gestures having definite, and yet ineffable, meanings.

<sup>15</sup> The corresponding three-note upbeat in mm. 24–25 is less charged with structural importance: no restoring of the correct location of the main accent is needed.

<sup>16</sup> Confident that the musicians' competence and taste enabled them to amend the text when necessary, Baroque composers like Bach did not always bother to write down the variant to be used when turning back to play repeats (or when heading for closure). What Bach chose to write down in m. 69 is obviously the deviation from the current figuration that was to be used when finishing off this *perpetuum mobile* prelude.

# **Interdependence and constraints; consistency**

This presentation of local options of continuation has amply illustrated the fact that the choices you make depend on past decisions and introduce constraints on future ones. If, for instance, the first sigh-motif passage mm. 1–4 has been played in a way suggesting imitations, consistency demands that similar passages are to be rendered correspondingly, unless there are valid reasons for exceptions. Or, having chosen to allude to the sigh motif in mm. 5–8, you cannot very well play mm. 16–19 so as to evoke a sense of strong contrast. But there are also choices entailing less strict constraints: if certain sections of the prelude are abruptly juxtaposed in the manner of a musical collage, it remains possible to link other sections together in a more continuous manner.

Still other interpretational decisions do not involve any future constraints, strictly speaking. Certain ways of playing a passage generate expectations as to the course of the music and its interpretation., expectations that may or may not be satisfied; other passages are more or less conditioned by expectations aroused by preceding events and the manner in which they were rendered. For instance, the rise in mm. 29–32 may lead directly up to the implied g♭<sup>2</sup> , or it may be played in a receding way, closing on e♭<sup>2</sup> , letting then the g♭<sup>2</sup> occur as a sudden exclamation. And the choice to bring out the resolution e♮<sup>1</sup> –e♭<sup>1</sup> in mm. 62/63 depends on a prior decision, namely the choice to bring out the deceptive cadence in m. 60 that introduced the persistent e♮<sup>1</sup> .

Consistency may be required when repeats are involved, and sometimes it is indeed preferable to refrain from changes when playing repeats. But in general it seems that musicians nowadays all too seldom make use of the opportunity – or all too often fail to fulfil their duty – to introduce interpretational variety when playing repeats (and other recurring passages), and this applies especially in pieces like this prelude. It appears as if the laudable ambition to devise an optimal series of interpretational decisions discourages the musicians from presenting alternative ways of continuation. This reluctance might perhaps be explained by the discomforting belief that presenting different interpretations of the same (or a closely similar) passage somehow calls in question the validity of the various readings, makes them seem arbitrary.

But it may be argued that interpretational choices made the first time are not necessarily binding when playing repeats. For instance, it seems possible not to allude to the sigh motif when playing mm. 5–8 the first time, but to do so in the repeat. While differences as to interpretation must not emerge as gratuitous or strained, repeats present a challenge to render some passages of the music in other, and yet equally convincing, ways.

Interpretational changes in repeats are in fact often commendable. When the listener has already heard a certain passage, its progressions are known, and therefore some additional information may be needed. More specifically, some interpretational choices crucially depend on the uncertainties that arise as the music unfolds, and whereas such options for continuation work well the first time, they are likely to lose much of their effect, or even turn meaningless, if used again when the same passage recurs. Conversely, certain changes as to the mode of continuation introduced only in the repeat may be rewarding, or indeed possible, just because the music has already been heard; such changes reveal new perspectives on the musical structure or make up interesting deviations from already presented interpretational decisions. Consider, for example, the passage mm. 9–16, and suppose that you have first chosen to suggest a rising line up to m. 11, then yielding to a recess in m. 12. When playing the repeat, you can with good effect embark on a bold unbroken eight-bar ascent starting from m. 9 and reaching all the way to m. 16.

Interpretation is not just a matter of local analysis of musical structure and piecemeal decisions as to modes of continuation, or even of finding out how such decisions should be combined with and adjusted to each other within larger sections in order to avoid arbitrary or counteracting options of continuation. Interpretation is also a question of working top-down, of establishing the overall form that you want to convey and of finding a convincing quasi-narrative content in the music as a whole. Hence, interpretation also involves large-scale decisions to be effected by means of carefully selected and balanced local and sectional options of continuation. In addition to introducing and then conforming to bottom-up constraints, and besides offering variety, the modes of continuation have to serve high-level musical ideas.

## **The influence of interpretation on form**

So far this essay has dealt with local or sectional decisions, and the perspective has been bottom-up. To make up for this, some top-down considerations are highly due.

You may entertain various high-level musical ideas that you want to suggest when playing this prelude, but fundamental to the interpretation to be proposed is the assumption that it may be preferable to undermine as much as possible the rigidity of the prelude's double-repeat form. To achieve this, the repeats can be played differently in ways letting a tripartite form emerge along with the obvious bipartition at the double-bar, and it seems that particularly the second part of the prelude opens up for several formal options. 17

Whether Bach as a musician would have used interpretation to bring about differences in order to suggest a less clear-cut form is impossible to know. But varying repeats (and parallel passages in general) in order to create diversity, and no doubt also to show off a prolific musical mind, was both expected and considered to be in good taste.

In the first part of the prelude, mm. 9–20 are crucial. In order to let this section emerge as more bold, as having more far-reaching consequences when playing the repeat, you may do as follows. The first time it is favourable to render mm. 9–10 (confirming F minor) and mm. 11–12 (establishing A♭ major) as balanced cadences, and then to play mm. 17–20 (set in E♭-major but forming a subordinate episode within the A♭-major context) in the same way as you played mm. 5–8. But when repeating the first part you can by-pass the cadences in m. 10 and 12 so as to suggest an extended rising progression up to m. 16, letting the following passage temporarily establish the more remote, glorious key of E♭ major. When closing the first part of the prelude, the secondary A♭-major tonic should emerge as a key that the music without any effort hovers down into from its D♭-major subdominant.

When playing the second part for the first time, it seems advisable to give the E♭-minor cadence in m. 32 a sense of closure that is sufficient to make the g♭<sup>2</sup> in m. 33 emerge as a firm start to which the following two descents can relate. In mm. 45–46, the rising sequence in the bass and the ascending top-note line, preparing for the top-note a♭<sup>2</sup> to be left only in m. 49, should be given prominence. In order to clarify the long-range falling progression down to e♮<sup>1</sup> in m. 56, the left-hand interjections in mm. 53–54 can be played so as to allude to the sigh motif, and there must be no loss of intensity in mm. 55–56, leading directly into the distorted theme and its peak notes d♭<sup>2</sup> –c2 in m. 58. A rather weighty cadence may then prepare for the emotional relief in the two-bar sighing transition to the arpeggio passage, to be played in a light manner with soprano dominance.

The point of an interpretation of this kind is to hide all demarcations that might have helped the listener to identify the moments when the sigh-motif

<sup>17</sup> For a discussion of bipartite vs. tripartite form in another piece, Beethoven's Bagatelle Op. 126, No. 5; cf. Bengt Edlund, "Interpreting a bagatelle", Chapter 10 in this volume

theme returns and hence to discover the tripartite ABA' aspect of the prelude's bi-partite double-repeat form.

When playing the repeat, on the other hand, this priority should be reversed by supplying cues indicating the two possible points of thematic return. To begin with you may fully exploit the rising sequence up to g♭<sup>2</sup> in m. 33, the starting point of a clearly bisected and expressive 4+4 bar section. The following passages up to the half cadence in m. 48 can then be rendered in a relaxed way, taking care that mm. 45–46 recall the sigh motifs from the beginning of the prelude by giving prominence to the alto-register. The left-hand entries in mm. 49–52 should be clearly articulated as sigh motifs, leading *crescendo* to the high-tension rupture in mm. 52/53. From this point a *diminuendo* transition follows, issuing into the dominant in m. 56, after which, in spite of the distortion, a counterpart to m. 1 should be clearly recognizable. The following deceptive cadence may be played in a resigned manner, but the e♮<sup>1</sup> , interfering with the deceptive D♭-major chord and supplying a link to the remaining ten bars of the prelude, must be prominent enough to launch a weighty coda, featuring alto prominence and a sense of metric conflict between alto and bass.

This interpretation, devised to express the inherent ABA' form, brings out two passages as possible *loci* of thematic return, but only one of these options should of course be selected.

It may be argued that the interferences just proposed make the doublerepeat layout of the F-minor Prelude more interesting, and also that rendering the second repeat differently in order to hide and then bring out the prelude's inherent tripartite design is a legitimate idea. The current view is that musicians are obliged to respect the integrity of the work by fully submitting themselves to the specified pitch-time structure. And this duty has been observed; disregarding the necessary adjustment in m. 69 when turning back to repeat the second part, all "structural" inscriptions in the score have been respected. 18 And yet the joint effect of these interpretational decisions, making the prelude's evolving form ambiguous, emerge as substantial almost to the point of altering the formal identity of the prelude.

<sup>18</sup> According to Nelson Goodman, ontologically valid exemplifications of music works cook down to no less, and no more, than this; cf. *Languages of Art*, Indianapolis 1968. A critical discussion of this idea, that seems to both overshoot and undershoot the target, is to be found in Bengt Edlund, "Scores and works of music. On interpretation and identity", Chapter 1 in the present volume; also in *British Journal Of Aesthetics*, 16(1996), 367–380.

### **Elements of variation**

Bold (or just historically informed) musicians might consider it proper to use further, more far-reaching interferences to undermine the double-repeat form of the prelude.

The fact that the A- and B-materials are closely related may be exploited to vary the form by structural means: the B-material can be transformed into sigh motifs (Ex. 3a) and the sighing quality of the A-material can be ironed out (Ex. 3b). Thus, when turning back to repeat the first and the second parts of the prelude, you may substitute right-hand wagging motions for the sighing double-stops in mm. 1–4 and 29–32, so as to achieve smooth connections and to prevent immediate recognition, effects that will at least postpone the moment of manifest return. Another option is to transform the last four/two bars of the sighing passages mm. 9–16 and mm. 29–32 into B-material. Conversely, you can play sighing motifs all the way from m. 29 to m. 36 (Ex. 3c). Likewise, the inherent sighs in the left hand in mm. 53–54 may be rendered manifest (Ex. 3d), and so may the right-hand ones in mm. 55–56 (Ex. 3e).

Using devices of variation already to be found in the prelude, the inverted counterpoint in mm. 3–4 can be undone –playing thirds instead of sixths will change the melodic contour as well as the musical content of mm. 1–4; the dialogue option is gone (Ex. 3 f). By keeping to simultaneous sixths and by adopting the rhythm of the sigh motif, the passage mm. 45–48 may be altered so as to clearly present itself as a thematic/formal return (Ex. 3g). Conversely, in order to conceal the motivic recurrence in m. 57 the right-hand might be played with the melody of the following bar as a model (Ex. 3h).

Still other variants may be introduced in order to bring out voice-leading features. You may, for instance, want to highlight the e♮<sup>1</sup> eventually giving in to e♭<sup>1</sup> in mm. 60–63 but feel that this delayed resolution should be complemented by a descending counterpoint from d♭ down to A♭. This motion is already inherent in the left hand, but it will emerge more clearly if you play octaves throughout; a further advantage is that this way of playing issues into the deviating A♭–a♮ motion (Ex. 3i).

Some of these paraphrasing ideas may be good ones, others are perhaps less so, but the crucial question is whether they are legitimate. (Some readers are likely to at least file a protest against changing m. 57 in the way proposed in Ex. 3h.) Whereas adding and/or deleting conventional ornaments are usually considered to be matters at the performer's discretion, deviations of the kind shown in Exs. 3 a/i – arguably also belonging to improvisation – may emerge as controversial since they change the given pitch-time structure of the music. But it cannot be maintained that such interferences really make the prelude hard to recognize, that they really change its identity. What they do is to temporarily suspend recognition when the parts of its bisected form are repeated, thus enhancing the interest of the music by displaying the ambiguity of some of its passages.

One should not forget that the F-minor Prelude predates the era of the emphatic work concept, as well as the times when the score was regarded as the source of non-negotiable prescriptions with regard to structure as well as interpretation: the prelude is by Bach, not Stravinsky. Yet, musicians venturing to undertake such alterations of the text run the risk of being censured by the Authorities for having transgressed their competence as musicians, for having meddled with the music.

A touch of prohibitive musical ontology has slipped in, and we should therefore close the discussion of this Prelude to the Art of Continuation.

# **Chapter 13 Interpretation as continuation**

# **Introduction**

In his book from 1989, *Musical Structure and Performance*, Wallace Berry discusses the principles and means of musical interpretation, and in particular the relationship between analysis and interpretation. 1 Berry presents his analyses as "exhaustive", but even if almost every note in three selected pieces is studied and tapped as to its consequences for interpretation, it seems that there are still important aspects left out of consideration, aspects offering further insights into the music as well as further conclusions of relevance for interpretation.

Barry's account is biased towards tonal reduction – Schenkerian analysis is often, indeed all to often, resorted to as the ultimate source and touchstone of structural insight. For this reason it is difficult to simply accept Berry's approach as a model for musicians wanting to refine their interpretations by means of keen analytic study. If an analysis is to be used for devising or supporting interpretations, it must also pay attention to transient phenomena and trace the causes and effects of the tonal events as they emerge when the music is unfolding. If the reflection of human intentions and emotions makes up music's core value, the great challenge for any method of analysis purporting to be useful when it comes to interpretation is to penetrate into the finely individuated modes of musical continuation. 2

Turning to another matter, there is a problem involved in reading and performing musical texts that Berry says very little about. Judging from his frequent references to the composers' performance indications, and from his loyal, almost submissive attitude towards them, it appears that Berry subscribes to the view, quite common among musicians, that virtually all signs in the scores are strictly prescriptive. But this view must be qualified. 3 It may be argued that

<sup>1</sup> New Haven 1989, Yale University Press. Concurrently, another work, quite different in outlook and approach, was published in Germany: Jürgen Uhde & Renate Wieland, *Denken und Spielen, Studien zu einer Theorie der musikalischen Darstellung*, Kassel 1988, Bärenreiter.

<sup>2</sup> The idea that 'continuation' is a crucial concept in interpretation is developed in Bengt Edlund, "Prelude to the Art of Continuation", Chapter 12 in this volume.

<sup>3</sup> Cf. Bengt Edlund, *"Sonate, que te fais-je?* Towards a theory of tnterpretation", Chapter 2 in the present volume; also published in *The Journal of Aesthetic Education*, 31(1997), 23–40.

all signs should be understood and observed according to their nature, i.e. with some discretion when the signs obviously refer to interpretation. Generally speaking, the problem with the composers' prescriptions (or perhaps merely hints) as to interpretation is that, while they specify one possibility, they seem to rule out other, quite conceivable options of interpretation. But fundamental for a creative approach to musical interpretation is the idea that the notation, just as the musical structure itself, is open in many respects, that the notation allows of various ways of rendering the music.

But the present investigation is not a belated review of Berry's book, and certainly not a wholesale dismissal of it; using one of his main examples, it should rather be regarded as a complement. Our goal is the same: to enhance the understanding of a delicate masterpiece of Romantic piano music, and to further the intellectual element within the art of musical interpretation.

Instead of drawing a reductive map of the work, I will use traditional and theoretically less committed methods and concepts to find out how its details combine to produce subtle turns of musical meaning, how it again and again opens up for alternative ways for the pianist to continue the musical process. The core issue is to demonstrate how interpretation as the art of continuation may benefit from a thorough study of rhythmic/metric properties and harmonic features, and from paying close attention to the motivic make-up and the melodic diction.

#### **Some preliminary observations on the Brahms** *Intermezzo*

Brahms's *Intermezzo* Op. 76, No. 4 is shown in Ex. 1. The melodic structure is indicated by brackets above the staff; the signs under it clarify the accentual relationships and the rhythmic grouping. 4 It can readily be seen that the music is complex and quite ambiguous, but before delving into its details, some general observations are due.

The *Intermezzo* is basically a melody/accompaniment piece, and the listener's attention is drawn to the upper line. A main concern must therefore be to render the melodic inflections and turning points in a telling way, and in order to do so the pianist has to take account of the melodic gestures as they emerge when the music unfolds, and to identify the quasi-psychological states of mind that these gestures and the shifts between them seem to reflect.

<sup>4</sup> Cf. Grosvenor Cooper and Leonard B. Meyer, *The Rhythmic Structure of Music*, Chicago 1960, Chicago University Press.

Counting "bars" (i.e. measure-size units starting from the second beats), the irregular construction of the thematic periods making up the bulk of the outer sections (mm. 1–13 and 33?–45) is [2+3(+1)] + [2+2+3] and [(2?)+3(+1)] + [2+ 2+1+1+1], respectively. 5 The bars within parentheses are bass interjections, and the question mark refers to the fact that the moment of formal return is veiled.

The tonality of the thematic periods is vague. The antecedents issue from and return to F7 chords, whereas the consequents dwell on first-inversion A♭-major chords until they settle in G minor and B♭♯ major, respectively.

Turning to the motivic constituents, a three-note motif, or rather family of motifs, is characterized by its dotted rhythm. Excepting some ambiguous cases, the various occurrences of this motif have either starting (x) or closing (y) function. Starting motifs are always made up of a rising third plus a falling second (a), whereas closing motifs originally and typically feature a stepwise descent (b) – but they may also consist of a downward leap followed by a rising second, or of a stepwise ascent, or even assume the appearance of the starting motif. An intriguing detail of the melodic design is the fact that motif (a), otherwise associated with starts, is first heard in closing position (y). The two initial bars also present a five-note phrase (c), which will return in free inversion (ci) in the middle section, providing its melodic substance and making for thematic integration within the piece.

Scanning the periods of the outer sections, a falling melodic tendency emerges if the high-register starts in m. 7 (39) and m. 9 (41) are disregarded. Selecting the final notes of the phrases, orderly descents (c2 –b♭<sup>1</sup> –a♭<sup>1</sup> –f♯<sup>1</sup> –g1 and c2 –b♭<sup>1</sup> –a♭<sup>1</sup> –g♭<sup>1</sup> – f 1 , respectively) come to the fore. Picking out b♭<sup>1</sup> in mm. 5 (36) instead of the last note f1 of the finishing (b) motif makes sense for several reasons. Although it is harmonically unstable, the b♭<sup>1</sup> is a rhythmically and metrically salient note, and the metric parallelism with the initial phrase turns the second (a) motif issuing into b♭<sup>1</sup> ambiguous – it may also function as a closing motif. 6 Listening ahead and considering the high-level rhythmic/metric structure, the (b) motif actually ending the three-bar phrase emerges as an added afterbeat; moreover, the final note f1 tends to be obscured by the melodic activity of the bass.

If you study the two-phrase antecedent and the three-phrase consequent, you will notice that both of them bring three starting (a) motifs, and that only the

<sup>5</sup> Henceforth, bar numbers refer to the "bars" straddling the bar-lines, not to the actual bars in the score. To find the passages, count the bar-lines, not the bars.

<sup>6</sup> Imagine a B♭-major start of m. 4 as indicated by the added tiny notes in Ex. 1, and the latent closing potential of the second (a) motif becomes obvious.

last of these efforts, entering one step lower, comes up with a releasing change, a subtle application of the third-time-lucky melodic archetype. The antecedent reinstates uncertainty (mm. 5 and 37) while the consequent eventually leads to tonal stability (mm. 13 and 45). One may understand the melody as a twolayered structure: each phrase but the first is made up of an initial high-pitched starting utterance followed by a low-pitched and less emphatic closing answer; cf. the oblique lines in Ex. 1.

Taking account of these and other observations yet to be made, how can the melody be animated, what options for continuation does it offer? Since analytic findings cannot always be expected to converge, and since the *Intermezzo* is inherently ambiguous, it is likely that several readings will present themselves. The next two sections give an idea of the scope and diversity of the interpretational options that a sensitive analytic reading of the outer periods may suggest. Then we will in turn consider the complex middle section and the two *stringendo* episodes.

#### **The first thematic period**

The searching and open-ended character of the initial two-bar phrase is unmistakable. To achieve a sense of hovering, the pianist must be cautious not to stress the left-hand F, inviting to be heard as an all-too unequivocal cue for a primary metric accent. This note is certainly the deep root of the F7 chord, but you should not make a big deal out of it. If you want to signal a main downbeat in this first tentative phrase, it can be postponed until c2 -over-f in m. 2; it would simply be blunt to emphasize the a1 -over-F in m. 1. In order to underscore the yearning quality of the phrase, the initial e♭<sup>2</sup> may be slightly lengthened. This note is a weak beat according to the notation, but if it is imagined as a strong beat, a virtual triple bar will emerge, making for a sensually dragged-out start of the music. 7 Indeed, it is possible to suppress also the accent on c2 in m. 2, so as to bring out a quite extended, initial anacrucis, blurring the phrase demarcation and leading first to an understated c2 -over-F in m. 3, and from there on to the deceptive b♭<sup>1</sup> -over-G in m. 4.

(When going back to play the repeat, the three *sostenuto* beats in mm. 19–20 may be rendered so as to make up a connecting triple bar, which means that the initial e♭<sup>2</sup> will again emerge as a virtual downbeat. Otherwise, if played in duple

<sup>7</sup> Think of starting a car having a somewhat worn-out clutch.

time, the strong beat of m. 20 would be followed by an all-too-prominent a1 over-F in m. 1.)

Turning to the (a) motif finishing the first phrase, it may either be played so as to disclose its quite vague closing position, or so as to forebode its use in the next phrase as a starting motif. There is a subtle distinction of meaning and affect between an inconclusive ending and a tentative beginning being abandoned for a renewed, more determined attempt. The way the first phrase is finished will influence the rendering of the next (a) motif. Whether played as a fresh start – as a fresh start transforming a closing motif into a starting one – or as a second, perhaps slightly more insistent and eventually successful attempt to proceed, a delicate relationship is established between the two phrases.

In any case, the resuming (a) motif must have enough energy to imprint its dotted rhythm on the rest of the phrase and to reach the focal note b♭<sup>1</sup> . A slight emphasis on the left-hand F may contribute to this energy, but utmost care must be taken so as not to give an impression of a premature entry of the complementary left-hand melody, which is to start only at G in the next bar. Alternatively and preferably, you can play so as to suggest that the b♭<sup>1</sup> is produced by the accumulated momentum of three dotted rhythms, interlocked to form an extended upbeat; in this case there is no reason to take the risk of misleading the listener by emphasizing the F.

The arrival at b♭<sup>1</sup> , underscored *and* undermined by the harmonically deceptive G in the bass, makes for an unmistakable downbeat in m. 4 and relegates the concluding (b) motif to serve as an afterbeat in the encompassing rhythmic group. 8 A prerequisite for this to work is that the c2 in m. 3 is outdone as a high-level downbeat by the b♭<sup>1</sup> in m. 4; otherwise the tendency towards metric regularity might turn the closing f1 in m. 5 into a primary accent, which would sound quite heavy-handed. Another option is to conceive of, and play, the (b) motif as if it belonged to a lower register, i.e. to put it in the shadow so as to suggest a sense of dialogue within the melody, a dialogue to emerge more clearly in the consequent.

The second phrase has so far been regarded as consisting of 2+1 bars, a reading implying that its second (a) motif has a closing (y) function. But, and perhaps somewhat against the grain of the music, the phrase may also be conceived of as made up of 1+2 bars with this motif in a starting (x) position. This reading means that the transposition of the (a) motif is presented as the third, decisive

<sup>8</sup> This reading is not compatible with the hairpin sign in the score, but it seems patently *non-grazioso* to play this falling motif with its closing function *crescendo*.

and triggering gesture, as the first (a) motif unequivocally rendered so as to have a starting quality. In other words, if a configuration of 1+2 bars is to emerge, the transposed (a) motif must be played so as to serve as the final member of a cumulative succession of three (a) motifs. In effect, the two phrases will emerge as joined into one. The 1+2 configuration also brings out the soprano connection to the consequent: the two last motifs of the antecedent are transposed to yield the first phrase of the consequent.

What are the interpretational options of the bass voice, intruding between the antecedent and the consequent? Already the G is problematic. Being of crucial harmonic significance – the sonority introduced in m. 4 amounts to something like a deceptive relative-minor root supporting a deceptive subdominant chord9 – this note would be brought out even if the extra half-note and the slur were absent. It is also melodically significant, but due to the fact that it is preceded by the root F, it is difficult to clarify that the melodic phrase starts from G.

If the pianist is anxious to avoid the impression of a too-late start of the bass motif from the A in m. 5, he/she runs the risk of over-emphasizing the G, which would highlight the unexpected harmonic shift rather than the overlapping start of the five-note counterpoint phrase in the bass. The outcome after the bar-line *is* deceptive, but it is necessary not to make too much out of the unexpected harmonic turn of events occurring within the phrase. On the other hand, it is perhaps not altogether inappropriate if the listener does hear a bass line starting only from A since this would actualize the similarity with the soprano in mm. 1–2. But bringing out this association has a considerable price: if the overlapping start on G is obscured, the intrusive character of the bass interjection will be lost.

Turning to the next rhythmic layer, the half-note G functions as a prolonged upbeat, corresponding to the initial quarter-note e♭<sup>2</sup> . The similarity with the first soprano phrase may be taken to imply that the (a) motif in m. 6 is again used in a closing (y) function. But this time – due to the overall rising direction of the left-hand phrase and to the following motif in the bass – the situation is more inconclusive, transition-like.

There are two motivic links between the intruding phrase in the bass and the consequent. The (a) motif of the bass is immediately answered by the top voice,

<sup>9</sup> Try again the normalized B♭-major start of m. 4. It is not only the left-hand G that is unexpected, but also the persisting drone on e♭<sup>1</sup> in the alto voice. To simply say that the chord after the bar-line is a first-inversion E♭-major chord does not capture its dual meaning.

but it is also, deprived of its dotted rhythm, at once replicated in the bass voice itself.

If the (a) motif in the bass is infused with a sense of initiative, we might hear that the bass urges the soprano to continue, that the low voice prompts the high voice how to go on. This relationship between bass and treble suggests that the beginning of the consequent may be rendered with a sense of relief setting it off from the ambiguous and irresolute start of the antecedent. Or, turning to the harmonic aspect, the consequent can be launched in a way fitting its transformed, heightened A♭-major setting.

If, on the other hand, the (a) motif in the bass is rendered in a way suggesting relative closure, its inherent B♮–c motion will prepare for its later replicas in the bass. The mood stays dark, and an element of continuity is introduced, connecting the two parts of the period and spanning the consequent until m. 11 where the bass introduces a lowering transposition. 10

The third and fourth phrases, deriving from mm. 4–5, may either be conceived of as coherent melodies falling towards a♭<sup>1</sup> and f♯<sup>1</sup> , respectively, or (the slurs notwithstanding) as quasi-dialogues, featuring the (a) and (b) motifs in separate registers. A slight delay of the (b) motifs, or playing them somewhat softer, might convey a sense of a dialogue. Another option is to bring out the perceptible difference between the effort inherent in the rising (a) motifs and the relief associated with the falling (b) motifs – the sixteenth-note in the former might be slightly lengthened and emphasized in contrast to the swift and light sixteenthnote in the latter. In any case, it is essential that the first motif bears the high-level metric accent; it would sound most ugly to play so as to give an impression of the opposite, weak-to-strong pattern.

In the second of these phrases, the (b) motif is transposed down by a third and appears in a chromatically compressed form; important melodic mutations bringing a decisive harmonic change. Notwithstanding the *piano* indication, one way of making the unexpected falling fifth within the second phrase stand out is to announce the leap by playing the preceding (a) motif louder, suggesting an additional effort explaining the lower pitch of the (b) motif. Another option, consistent with the notion that the two motivic constituents may be taken to belong to different pitch layers, involves an increase of energy at the unexpected

<sup>10</sup> Both the bass/soprano imitation and the motivic allusion in the bass may of course be brought out by employing a cliché of Romantic piano playing in m. 7, by letting the start of the motivic contour of the (a) motif in the left hand be heard slightly before the start of the right-hand imitation.

a♭<sup>1</sup> in m. 10. As a result, the low-register (b) motif will appear to have seized the initiative, which means that a still greater, resuming effort – or conversely, an altogether different, more relaxed, attitude – is demanded from the following, transposed (a) motif, accompanied by the lowered bass motion A♮–B♭.

The last phrase of the consequent consists of three dotted motifs, of which the second is divided by the slurring. This motif is also divided in terms of its internal falling fifth, and by the fact that its first note d2 , a rising resolution of the c♯<sup>2</sup> left hanging in the air by the preceding motif, connects strongly backwards. But it is not until the alto voice at long last has left its e♭<sup>1</sup> for d1 – it is forced to move by the d2 of the soprano – that the c♯<sup>2</sup> is (retrospectively) heard as a dissonance, and that d2 turns out to be its resolution. (Indeed, for a very short moment the d2 is dissonant.)

The final motif has, at a cursory glance, the same features: the last note is again divorced from the first two notes by a slur; indeed, the slurred a1 –b♭<sup>1</sup> ascent seems to imitate the preceding rising suspension/resolution motion. But on a closer inspection important differences emerge. The soprano a1 is at first consonant – it is only the continued downward motion in the alto voice to the dissonant c♯<sup>1</sup> that turns the a1 into a soft discord – whereas its "resolution" up to b♭<sup>1</sup> is certainly dissonant. This note, occurring on the relatively weak second beat of the notated bar, has a suspension-like quality which is but partly relieved by the descent to c♮<sup>1</sup> in the alto, a note that itself turns into a dissonance after the bar-line.

The net result of these bewildering complexities seems to be that the two slurs conflicting with the three dotted motifs invite to be accommodated so as to allow of a melodic organization featuring two subtly irregular sub-phrases, an organization that agrees with the idea of separate pitch registers within the melody. As a consequence, the second dotted motif will be suppressed; turning to practice, this means that an afterbeat (d2 ) becomes appended after the first motif, and that a two-note initial motion (g1 –a1 ) is added before the third motif. A further effect is a latent stretching of the high-level meter: not only are two expanded sub-phrases made to coexist along with the three regular dotted motifs, but the main accent due at a1 seems to be postponed until g1 , the last note of the period.

The final phrase of the consequent is characterized by its ambiguity and sense of continuity. The latter quality is particularly apparent in the chromatic alto descent from the drone e♭<sup>1</sup> ♯ to the b♭ starting the *stretto* section, a connecting motion that should be gently brought out. The need to establish the last motif as a closing formula, the activity of the alto, and the final tenor rise e♮–f♯–g, conspire to slow down the tempo in mm. 11–13; if you play this passage too fast, its harmonic complexity cannot be appreciated.

# **The second thematic period**

Turning to the second thematic period of the *Intermezzo*, some important differences must be mentioned. (The ambiguities of its beginning will be discussed later on.)

Bar 41 brings a crucial event in the bass, decisive for the further tonal course of the music and drastically deviating from what the listener expects. The quarternote essence of the (a) motif is inverted when B♮–c is exchanged for c♭–B♭, and such an event cannot very well be rendered in an unobtrusive way, as something that just happens. The falling inflection in the bass, the B♭-instead-of-c, must have a cause matching its significance: the falling leading-note c♭, and its sense of enharmonic contrast to the rising leading-note B♮ of the preceding (a) motif, must be highlighted by giving it an urgent upbeat quality.

Due to the second-inversion E♭-major chord in m. 41, the a♭<sup>1</sup> starting the (b) motif becomes poignantly dissonant. In order to do justice to the next harmonic shift, introduced by the dark E♭-minor g♭<sup>1</sup> in m. 42, it may be a good idea to change the established high-level accentual relationship, letting the two motifs of the phrase form a iambic rather than a trochaic group.

In mm. 43–45 the slurring does not obscure the three dotted motifs; there are separate slurs for each motif, leaving their internal relationships undetermined and opening up for various modes of continuation. The sonorous chromatic bass descent from E♭ down to B♭<sup>1</sup> provides a strong sense of direction, and like the corresponding alto descent in mm. 11–13, it should be brought out.

It appears that this passage offers three quite different options of interpretation. In keeping with the bass motion down to the tonic, the melody may be played so as to end with two slightly checking closing motifs. Indeed, the passage can even be rendered so as to let three progressively more halting finishing motifs emerge; already the (a) motif, ending on b♭<sup>2</sup> over a first-inversion B♭-major chord, may be given a closing function. But the second motif can also be joined with the first one, giving rise to a pair of motifs as in the preceding phrases; then a further closing motif is added to finish off the phrase. The period is stretched by means of the rising suspension reaching f1 , the elided start of the *stringendo* section.

### **The middle section**

The next part of the *Intermezzo* to be discussed is its middle part; some of the points to be brought up deal with interpretative options in situations where the notation does not give any clear guidance, and where the pianist must assume a vital role in determining the structure.

Bars 19–20 serve both as a transition back to the beginning and forwards to the developmental middle section. This difference in function may be brought out by using shared pitch-classes as a device for smooth connection. Thus, when returning to m. 1, the lower right-hand motion c♯–d–e♭ can be slightly emphasized so as to attach to the e♭<sup>2</sup> starting the repeat. The bridge to the middle section is made up of a pitch-class dyad. In order to make this clear, the b♭ in m. 20 may be gently struck again, laying bare the double stop b♭/e♭ to be resumed as the falling fifth b♭<sup>1</sup> –e♭<sup>1</sup> heard in m. 21.

The latter notes must stand out for other reasons as well. Being a melodic upbeat, b♭<sup>1</sup> cannot do without some extra emphasis, and the alto e♭<sup>1</sup> is crucial if you want to do justice to an exquisite point of the harmonic design: the forthcoming shift of the alto drone from e♭<sup>1</sup> to f♭<sup>1</sup> . It would be quite out of place to stress the f♭<sup>1</sup> , but if its origin e♭<sup>1</sup> is brought out, it will be noticed. This is also what the local situation demands since the f♭<sup>1</sup> turns out to be the (slightly dissonant) resolution of a transient suspension – the G♭ in the bass has turned E♭ minor into G♭ major, making the e♭<sup>1</sup> somewhat dissonant; being the consonant preparation of a suspension, the e♭<sup>1</sup> has to be brought out. In order to further underscore this moment of relief and minor-to-major shift, one may halt somewhat at the downbeat of m. 21. A slight *tenuto* will mark the root of the G♭-major harmony, highlight the fact that e♭<sup>1</sup> is now dissonant, and delay the mellow rising resolution to the seventh f♭<sup>1</sup> in the alto, a note that completes the harmonic change and concurrently discloses the mild but poignant suspension quality of the lingering e♭<sup>2</sup> which deserves to be heard before it falls to d♭<sup>2</sup> .

A similar description applies to the next phrase as well; although itself consonant, the top note g♭<sup>2</sup> in m. 23 should also be rendered prominent so as to prepare for f♭<sup>2</sup> , a note giving in to the actual dissonance f♭1.

Bars 21–26 bring three soprano phrases, free inversions of the initial fivenote phrase (c) of the piece, and two complementary tenor motions indicated by Brahms in mm. 22–23 (b♭–c♭<sup>1</sup> –b♭) by means of extra stems, but not in mm. 24–25 [b♭–a♭–d♭<sup>1</sup> ]. The overlapping of the soprano phrases, involving voicecrossing at the junctures, can be clarified if the accented dissonances a♭<sup>1</sup> and c♭<sup>2</sup> in mm. 22 and 24, respectively, are somewhat lengthened and slightly stressed at the expense of their falling inner-voice resolutions; there is no need to bring out what is expected, what will be understood anyway. This leaves room for announcing the new phrases by giving suitable emphasis to the upper-voice upbeats d♭<sup>2</sup> and f♭<sup>2</sup> .

Turning to the tenor, should the second, non-indicated complementary motion be brought out? If not, mm. 24–25 will sound thin and incomplete in comparison with the richer contrapuntal texture of mm. 22–23. Indeed, leaving out this motion seems to require that the first tenor motif should be suppressed as well. On the other hand, the rising inflection of the second tenor gesture [b♭– a♭–d♭<sup>1</sup> ] may be heard as leading to f♭<sup>1</sup> , a left hand-note actually belonging to the alto-voice drone that so far has been played by the right hand. Bringing out the second tenor motion therefore entails the risk of a momentary confusion as to the voice leading of the passage. But considering the future course of the middle section and especially the widely spaced registers of mm. 25–26, some alto prominence may be needed to boost the overall sonority; a false tenor/altovoice connection may prepare for the desirable fullness of sound.

Before discussing some options for the interpretation of mm. 25–32, the voice leading and the rhythmic structure of this quite complex passage must be studied.

In m. 27 the texture is suddenly enriched: the soprano is doubled in the lower octave, and a new and quite active voice is added in between. Since it starts at the same pitch class as the alto and eventually proceeds downwards in octave parallels with it, this "upper" alto voice may be regarded as an expressive intensification of the inner motion that eventually brings the alto voice back to its original drone position at e♭<sup>1</sup> in m. 31. The bass is also significantly active in this passage. After having ascended G♭–A♭–B♭, it descends B♭–A♭–G♮, adding a further parallel motion to the falling sequences in the duplicated soprano and alto voices.

Bars 31–32 bring further changes. The "lower" soprano voice disappears – its e♭<sup>1</sup> is (apparently) taken over by the lower alto voice, now transferred from the left to the right hand – and the "upper" soprano comes to a standstill at e♭<sup>2</sup> . The octave parallelism between the two alto voices is transformed: the "upper" alto, bringing a further descent c♭<sup>2</sup> –b♭<sup>1</sup> –a♮<sup>1</sup> , will eventually emerge as the soprano melody to come, whereas the "lower" alto is pursued as a tenor voice. The motion of the "upper" alto/forthcoming soprano is paralleled by the harmonically decisive bass descent A♭–G♭–F, a progression that due to its faster pace and syncopated quality may be heard as correcting the preceding bass descent B♭–A♭–G♮.

The metric/rhythmic properties of this passage are also of great interest. Bars 25–26, making up the third, culminating link of the melodic sequence started in m. 21, bring two statements of the inverted (c) phrase; the second statement curtails the first, robbing it of its falling resolution. This construction makes for two downbeats in succession, of which the second outdoes the first. The iteration of motivic material in mm. 27–30 confirms and pursues this shift to short metric units: the units start at the bar-lines rather than in the middle of the bars, and they are not grouped by any slurs in the score but form a continuous rhythmic sequence due to the overlapping of the falling-second motifs of the soprano strands and the motions of the upper alto.

The fact that (the notated) m. 27 is exactly replicated as m. 28 makes up a most conspicuous and unusual feature. 11 Since the first a♭<sup>2</sup> –g♭<sup>2</sup> motion brings the resolution of the last (c) motif in the top voice, it tends to be associated with the closely linked bars 25–26 as an afterbeat. Consequently and although it repeats the preceding bar, (the notated) m. 28 will emerge as metrically strong; mm. 29 and 30 then follow as afterbeats. Considering the low-level rhythmic grouping, the falling soprano octaves make for trochees in (the notated) mm. 27–30, whereas the upper-alto motions just as patently give rise to iambs in the "bars" straddling the bar-lines. The passage ebbs away with a prolonged metric unit: due to the sense of a slow syncopation in (the notated) mm. 31–32, the gradually suspended activity, and finally the extended dominant harmony, a hyper-measure equalling 2 1/2 or 3 1/2 notated bars seems to begin in the (notated) m. 31. The rhythmic structure at the end of the middle section is vague – the sense of accentuation is attenuated and the events do not form stable groups.

Another crucial aspect of this multi-layered passage is the non-coinciding culminations. The peak of the soprano is reached already at c♭<sup>3</sup> in m. 26 – notwithstanding the octave doubling, the next bar makes up a withdrawal, and (as it will turn out) it represents the first stage of the descent of the topmost voice. The drone on f♭<sup>1</sup> is suddenly transformed into a peak when the passionate upper alto voice enters on f♭<sup>2</sup> in m. 27 – this note will disclose its origin if the pianist lets it grow out of ever more intense middle-register f♭<sup>1</sup> 's in the preceding bars. The culmination of the bass occurs only in m. 28, whose B♭, apparently triggering the descent of the other voices, emerges as a more important event than the arrival up to this note in the preceding bar.

<sup>11</sup> There is a duplicated bar also in the *Intermezzo* Op. 118, No. 1 that also offers an interpretational challenge; cf. Bengt Edlund, "Loyal disobedience", Chapter 4 in this volume.

It is necessary to come to terms with the dispersion of culmination points and the literal repetition in (the notated) mm. 27–28, i.e. to find the metric and melodic configuration to be expressed. Whereas the sudden introduction of the upper alto and the concurrent doubling of the soprano seemingly make for a new start in m. 27, it appears that this sense of a beginning is overruled by the joint effect in m. 28 of the iteration of the soprano and alto material, and of the turningpoint B♭ in the bass. Thus, the metric structure of (the notated) mm. 25–30 turns out to be 3+3 rather than 2+4 bars.

The 3+3 configuration is actually strongly preferable since it is very hard to find a convincing musical meaning of the soprano and bass in (the notated) m. 28 if this bar is attached to its identical forerunner – it would amount to a kind of formal stuttering. Furthermore, if these bars were joined, they would almost inevitably seem to form a closed pair in which the last member is relegated to subordinate status, a configuration that in a most unfortunate way would dissipate the tension needed for the rest of the descent. But when divorced from its predecessor, (the notated) m. 28 becomes readily understandable as a repeated and more emphatic quarter-note appoggiatura in the soprano and as the starting point of the ensuing withdrawal heralded by the second B♭ in the bass.

Yet it is possible to use the active upper alto voice introduced in m. 27 to supply a link; particularly if it has been rendered so as to arise from the middleregister f♭<sup>1</sup> 's, it may (opposing the receding gesture of the soprano resolution) be understood as leading to its immediate repeat in m. 28, thus contributing to a joint structural downbeat in this bar. Granted that the upper alto voice at first enters on a downbeat, its connecting iambic character is unmistakable, and this quality may with good effect be maintained by local *crescendos* counterbalancing the *diminuendos* expressing the trochaic suspensions of the soprano voices. 12

It remains to study the transition back to the repeat of the thematic section, involving a subtle ambiguity as to the actual moment of return.

The recapitulation may be taken to start on the second beat of (the notated) m. 32 since at this moment everything is restored: the first two notes of the initial melody are simultaneously present in the upper voices, the alto drone has already returned to its original position on e♭<sup>1</sup> , the tenor figuration murmurs like it did when the piece started, and the bass has reached the F of the dominant seventhchord. But it is difficult convey that the recapitulation starts here. The e♭<sup>2</sup> , already

<sup>12</sup> Obviously, the (redundant) hairpins in mm. 27–30 refer to the patently beginningaccented falling motions of the soprano octaves.

present, cannot begin the melody in a convincing way – nor is this what Brahms's two-bar slur asks for – and it is impossible to make the listeners understand that there is a falling e♭<sup>2</sup> –a♮<sup>1</sup> motion inherent in the double stop. Furthermore, it seems to go against the grain of the music to let the F, being the (delayed) final note of a local bass descent from A♭, signal a new section. The only thing you can do to suggest a beginning at this point is to release the pedal, creating a clean slate exposing e♭<sup>2</sup> /a♮<sup>1</sup> .

Consequently, the recapitulation has to start after the fact, i.e. either with the first or the second of the ensuing (a) motifs. Judging from the second and third e♭<sup>2</sup> -to-e♭<sup>2</sup> ties, Brahms may have preferred the most postponed moment in m. 35, which suggests that the first (a) motif, still shadowed by the top note, should be rendered so as to express utmost dissipation, as expressing a state of neither-starting-nor-closing-hesitation to be followed by a clearly articulated, starting (a) motif in restored tempo. The earlier moment of return in m. 34 might be rendered with essentially similar means – recession followed by resumed activity – a start that has to be repeated with more confidence since the first (a) motif is veiled by the lingering e♭<sup>2</sup> .

But it seems that there are still further options when it comes to this transition, interpretations that take notice of the fact that the b♮<sup>1</sup> starting the first (a) motif issues from a♮<sup>1</sup> . This a♮<sup>1</sup> is the last note of the descent c♭<sup>2</sup> –b♭<sup>1</sup> –a♮<sup>1</sup> , a descent whose importance is underscored by the fact that it is doubled by the lagging tenor and firmly supported by tenths in the bass. A quite convincing bridge to the recapitulation period may emerge if this falling motion, announced by four active, upper-alto sixteenth notes, is brought out and attached to the first (a) motif.

This connection might in turn be understood in several ways. Since, according to this reading, the first of the (a) motifs is incorporated in, and has a closing function within, a phrase belonging to the middle section, the recapitulation appears to be postponed until the second (a) motif in m. 35. Indeed, a very bold interpretation, suggestive of a truly long-term, overlapping continuity, involves playing the whole upper-alto/renewed-soprano strand suddenly beginning in (the notated) m. 27 as a huge anacrucis, a motion that eventually and very inconspicuously reaches its downbeat on the c2 in m. 34. On the other hand, since a melodic strand has been identified that apparently issues into a♮<sup>1</sup> , a quarter-note prolonged by a tie to make up a half-note, there is a strong sense of an inconclusive ending, making for a fresh start already with the first (a) motif. But the prolonged a♮<sup>1</sup> may itself be taken as the beginning of the main theme – a hesitating start making for a subtle elision between the development and the recapitulation.

## **The** *stringendo* **episodes**

To make this account complete, some words should also be said about the two *stringendo* sections.

The essentially similar passages "bars" 14–20 and 46–52 finish off the outer, thematic sections of the piece, and each of them consists of a pair of virtually identical two-bar phrases featuring parallel thirds, phrases that are followed by a pair of identical, low-register one-bar additions in parallel sixths. While in the first *stretto* passage a leap separates the one-bar additions from the preceding two-bar phrases, the "too-early" start of the sixths in the second passage makes for a smooth connection. The passages end with a one-and-a-half-bar transition and a four-bar plagal cadence, respectively.

Basically, the interpretation is quite straightforward: the second two-bar phrase invites to be played softer than the first one; then comes two even more subdued one-bar repercussions.

Brahms's slurring summarily divides the flow into units comprising two notated bars, which means that a trochaic metric order is imposed, clearly forming a contrast to the so far prevailing iambic organization. The initial righthand fifth in the first passage as well as the corresponding chord in m. 45 are long appoggiaturas belonging to and extending the consequents, and the following melodic units set out from elided appoggiatura-resolutions in the same way. What Brahms evidently wanted to indicate by his slurs is that the events after the bar-lines are to be rendered so as to have a double function, so as to suggest starts in concurrence with the obviously closing dissonance-to-consonance. Without the slurs, one would be tempted to unequivocally locate the phrase shifts to the event after the resolution in the middle of the notated bars, i.e. one would continue to play unequivocal iambs.

The slurs may also be understood as asking for *quasi legato* playing throughout the passages in spite of the fact that from a technical point of view some doublestops invite to be executed *legato* while others, involving repeated notes, necessarily require *non legato* articulation.

Whether the *poco stringendo* is meant to refer to a slightly raised tempo during the first two-bar phrase or is valid for the next one as well is not altogether clear. In the former case, this direction may perhaps just indicate a gradual return to the main tempo after the natural, non-prescribed *ritardando* in the preceding bar. Furthermore, whether the *stringendo* should release anything more than a local and very slight agogic wave in the first two-bar phrase, perhaps followed by another one in the next phrase, depends on the main tempo chosen for the *Intermezzo*. In terms of melodic and harmonic change, these double-stop passages are the most dense in the entire piece, and if you want to convey this wealth of musical information, it is not advisable to play very fast.

The *stringendo* and *pianissimo* directions are difficult to reconcile. Due to the low register and the speed of the sixteenth-notes, it takes utmost care to avoid being too loud. Ideally, these passages should sound as faint, fluffy veils of sound.

As to the *stringendo* closing the piece, it may either be taken to mean just a gradual return to the main tempo after the *sostenuto* or be understood as asking for a progressively faster tempo, lending a *scherzando* touch to the very end of the music.

# **Concluding remarks**

Reviewing the analysis put forth above as well as the proposed options of interpretation, i.e. the choices opening up for various ways of continuation, some of the readings entail that certain directions in the score are disregarded or violated. The very beginning of the piece was tentatively understood as a virtual triple bar, and in m. 5 the *crescendo* hairpin was overruled; in m. 4 the feasibility of playing so as to unequivocally let the bass line issue from G was questioned. The *piano* mark in m. 9 was left out of account, and in mm. 11–13 Brahms's slurring as well as the motivic make-up were disregarded in favour of another structure within the phrase. In mm. 24–25 was proposed a tenor complementary voice not shown in the score, and in mm. 26–33 the interpretative options were not altogether compatible with the composer's dynamic signs. As to the later part of the middle section, the voice leading, the high-level metric as well as the location of the formal return emerged as under-determined in the score, and some of the proposed solutions involved quite radical interferences that might be contrary to the composer's intentions.

It is evident that deviations from notation such as these cannot simply be condemned as licentious; it may rather be argued that conflicts between notation and interpretation make up an aspect of normal interpretative work. 13 Musical structures are often ambiguous, and musicians are keen not only to explore ambiguities, but also to clarify structure, bring out significant emotional content, and enhance appreciation by offering variety. Furthermore, the gravity of such transgressions must reasonably be judged according to the nature and intention of the signs; arguably, some signs are to be understood as non-imperative.

<sup>13</sup> Cf. Bengt Edlund, cf. "Directions and compliance" for further examples, and "*Sonate, que te fais-je?* Towards a theory of interpretation", Chapters 3 and 2 in this volume

It is also necessary to keep in mind that there are several kinds of transgressions. Some involve violations or substitutions whereas others are matters of disregard or omission; a third kind includes various sorts of additions, and a fourth category is made up amendments of passages whose notation is incomplete, deficient, or contradictory. Indeed, there are passages that appear to be musically unsatisfactory to the point of requiring interventions from the musician. 14

Just as interpretations (whether involving any deviations from the text or not) are generally evaluated in terms of structural understanding, richness of content, or sheer beauty, we should judge questionable deviations from what is written in the score – the only deviations we are likely to notice in ordinary circumstances – not as crimes, but as artistic mistakes. What should be condemned are dull interpretations that trivialize the music by disrespecting its notation, or harm it by clumsy and self-indulgent ideas having no support in its structure.

Some scores have more marks that explicitly (or implicitly) refer to matters of performance than others. Sometimes these signs are redundant, telling you little or nothing more than what you could have read out of the musical structure itself. It also happens that musicians are faced with scores that appropriate vital parts of the musicians' job by unduly restricting their musical perspective and creative freedom. But if the essence of musical interpretation is to bring out what there is potentially to be heard in a piece of music, it makes sense to assume that scores with interpretational directions are as rich in artistic options as those without, and to deal with them accordingly.

Consider, for instance, the opening four bars of Chopin's F-minor Etude from *Méthode des Méthodes*, a sequence of notes that can be, and is, rendered by pianists in infinitely many ways; cf. Ex. 2. 15 We must appreciate that the composer refrained from specifying the interpretation, that he did not add any signs discouraging his fellow pianists from finding their own way of rendering the melody. Chopin's reticence when writing down this etude contributes to its cumulative richness.

<sup>14</sup> For examples, cf. Bengt Edlund,"Loyal disobedience. When is it OK not to play as written?", Chapter 4 in the present volume.

<sup>15</sup> This passage is analysed from an idiomatic and musical point of view in Bengt Edlund, "The phenomenology of fingering. Structure and ontology in Chopin's F-minor Etude from *Méthode des Méthodes*", Chapter 7 in *Chopin. The Preludes and Beyond*, Frankfurt 2013, Peter Lang Verlag; it also appears on pp. 88–105 in Poniatowska, Irena (ed.) *Chopin and His Work in the Context of Culture*, (Warszawa) 2003, Polska Akademia Chopinowska, Narodowy Instytut Fryderyka Chopina, Musica Iagellonica.

It has, hopefully, been demonstrated how analysis, if keenly taking account of all musical events that produce and specify musical continuation, may offer insights that cannot but inform the interpretation of the music. Close reading of this sort provides a wider scope for artistic freedom by disclosing interpretational options and ambiguities that may otherwise have gone unnoticed.

The various options of continuation have only sparingly been characterized in terms of emotions, states of mind, intentions quasi-inherent in the music, or extra-musical associations. This is not because such elements of content are absent in Brahms's *Intermezzo* – quite to the contrary – or are unwarranted or of no value. The inclination to associate music with aspects of human life is deeply ingrained in us and, if allowed to influence our artistic decisions, it may contribute to interpretation by lending an intuitively intelligible and compelling meaning to our structural insights. Indeed, many musicians are prone to first discover the potential for human expression, and then to turn to the score in order to have their ideas confirmed, refined, and perhaps modified by analytic observations.

# **Chapter 14 Musical dialogue in a Romantic violin sonata**

#### **Impersonation and dialogue in music**

It is a commonplace that music matters because it emerges as animate; music is of profound human interest because it sometimes seems to behave like we do. No wonder, then, that we are prone to impersonate music when listening to it or playing it. Depending on the properties of the music and on the disposition of the individual listener or musician, this impersonation may take on various forms.

Some music lends itself to be understood in terms of a quasi-dramatic sequence of events in which various structural entities take part as protagonists. Sometimes the music rather invites to be thought of as an evolving subject, as representing a fictive musical *persona*, whose character and inner development we get to know; the music is taken to depict states of mind passing through someone's consciousness. 1

The latter kind of impersonation can also assume another form: the musical structure itself may be heard and conceived of as an organism, abstract and yet having a quasi-human mind and sense of purpose. The various musical events do not only fit in with each other, they seem to be actively generated from within the music. The music emerges as a being with its own intentions as well as the consistency required to bring them about, it appears to be a living substance providing the necessary and sufficient causes for its own process of change, its sense of continuation.

This kind of personification is in fact common to the point of being almost inevitable. When describing music, we frequently use expressions like "and then the music proceeds to the dominant, accumulates tension, recedes to the point from where it started", etc.. This may seem to be just examples of figurative speech, but on second thoughts most of us are willing to concede that "music" in such cases amounts to more than a convenient formal subject – it does refer to a substance experienced as being active, animate.

Finally, and approaching the subject of this study, we sometimes have a strong impression that the music speaks, or that it consists of a sequence of utterances, and when we apprehend music in this way, we adopt a long tradition. Musical phrases may often be aptly described as having an understandable, characteristic,

<sup>1</sup> The latter view is, it seems, taken by Donald Callen in "The Sentiment in Musical Sensitivity", *The Journal of Aestetics and Art Criticism* 40 (1981/82), 381–393.

and emotionally meaningful diction. And in many cases it does not seem overly metaphorical to capture the essence of a musical passage with words like "talk" or "conversation", and to resort to reporting verbs when trying to catch the content and attitudes of such human intercourse. Music can be conceived of as a monologue of one "speaker"; in other cases the listener is bent to imagine different people engaged in a musical dialogue (or multilogue).

It is evident that the musical structure to an appreciable extent determines the mode of personification to which it may lend itself. If the music lacks sufficient consistency, for instance, it will hardly emerge as a conscious intra-musical subject taking form before the listener's ears. 2 And if speech-like properties are not very prominent or entirely absent in the music, nobody is likely to understand it in terms of human utterances. Turning to the distinction between monologue and dialogue, there are a number of properties making us identify more than one "speaker": structural interplay involving voices, motifs, or formal units, and of course the use of different instruments. Needless to say, there are many multivoice works that are predominantly heard as monologues, just as there may be elements of dialogue even in mono-linear solo works. 3

It is important to realize that musical passages are often indeterminate or ambiguous with respect to personification. More than one mode of impersonation may present itself to the musician or the listener, and whether we apprehend a piece of music as, say, an evolving *persona* or a speech-like dialogue, may ultimately depend on the properties of the performance.

To a considerable extent musical interpretation is a matter of selecting options for and then expressing musical continuation; and if this is not to be accomplished gratuitously, the musical structure must be carefully studied. 4 Among the things to be considered when devising an interpretation, the indeterminacy as to personification and hence the freedom to choose mode of impersonation is a quite important one. It makes a difference whether the structure is conceived of in terms of a sentient musical subject or as a sequence of utterances in a dialogue – if the music is thought of as a dialogue, the scope for interpretational contrasts may be quite wide.

<sup>2</sup> Cf. Jerrold Levinson, "Truth in Music", *The Journal of Aesthetics and Art Criticism* 40(1981/82), 131–144.

<sup>3</sup> For an attempt to understand the Allemande of Bach's Suite for solo flute BWV 1013 as a dialogue, cf. Bengt Edlund, "Monologue as conversation" in Varia 1.

<sup>4</sup> Cf. Bengt Edlund**,** "Prelude to the Art of Continuation" and "Interpretation as continuation", Chapters 12 and 13 in the present volume.

This brings us to the aesthetic aspects of apprehending music as a non-verbal dialogue. Obvious musical dialogues have a number of structural properties – elements of imitation, antithetic phrases, and the like – predisposing the music to be heard in this way. Furthermore, and adapting one of Jerrold Levinson's notions of musical truth for the present purpose, it seems that the utterances in a genuine musical dialogue must be characterized by human attitudes that, when joined to form a sequence, make up a meaningful, psychologically credible, conversation. Sometimes these attitudes are unmistakably inherent in the music whereas in other cases they are potential qualities that must be underscored and conveyed by the musicians in order to emerge.

Using the notion of 'dialogue' as a guiding idea for an interpretation is an artistic strategy that can be misused. It must be kept in mind that suggesting a sense of dialogue may sometimes require distinct cues in the performance, and this means that an extravagant or inappropriate use of the dialogue mode might be inimical to long-term musical growth. Some pieces can take many and obvious interpretative interventions suggestive of a dialogue while others are quite vulnerable. It seems that the dialogue mode of conceiving musical structure depends on the "structural pace" of the music. Assigning different phrases (and the like) to different "speakers" means increasing the musical information. It is reasonable to assume that pieces in slow/moderate tempos or being characterized by fairly sparse structural shifts can take more exchanges than fast pieces or music with dense shifts; excessive fragmentation should be avoided.

The artistic gains from bringing out exchanges of musical utterances must therefore be judged from case to case, and there is for each and any piece a limit beyond which efforts to express a sense of dialogue turn excessive and mannered. On the other hand, some music may benefit from touches of informality and elements of surprise, from the kind of vitality that we associate with a spirited dialogue.

#### **Structural dialogue in Brahms's Violin Sonata Op. 100**

We will study the interpretative options inherent in a musical structure that abounds in the confirmations and oppositions characterizing a musical (or any) dialogue; exchanges that to an imaginative mind suggest a multitude of finely differentiated human intentions, emotions, and situations, and that demand finely adjusted shifts of attitude from the performers if the exchanges are to be rendered in a way that makes the musical conversation transparent and moving. 5

<sup>5</sup> This study was initially published as "Forming a Musical Dialogue", pp. 144–170 in Sven Bäckman et al. (eds.) *Rytm och Dialog*, Studier framlagda vid Åttonde nordiska

That the first movement of Brahms's Violin Sonata in A major Op. 100 is characterized by an intimate relationship between the two instruments is hardly controversial – it is after all both a duo and an "amiable" piece. This music can with great profit be conceived of as a dialogue, extremely rich in terms of dramatic design, subtle references, and affective meanings. 6 The score will be probed in order to identify and describe – structurally as well as with respect to human content and options for interpretation – shifts in the music that may embody a sense of dialogue.

#### **The first theme – initial statement**

Several important motifs are presented in the first 4+1 bar unit of the sonata's main-theme section; cf. Ex. 1a. The germinal motif *a* is immediately heard in the upper voice, and it is complemented by an important counterpoint motif *ac* in the bass. Just as the *a* motif is doubled an octave below the soprano, *ac* is also present a tenth above the bass. The latter middle voice reaches f♯<sup>1</sup> , the starting point for motif *b*, a seemingly insignificant connecting particle that will emerge as prominent later on in the movement. The falling-fourth gesture *d* within the larger, arch-formed five-note motif *c*, is immediately repeated and filled in by the violin's *d'* motif, adding an extra bar to the regular four-bar half-period of the piano. The *d–d'* imitation is unmistakable, and the quality of the violin answer seems to be one of consent and perhaps quiet intensification. 7

metrikkonferensen i Umeå 4–7 oktober 2001, Skrifter utgivna av Centrum för Metriska Studier 14, Göteborg 2003. For a companion study of a piece that does not immediately invite to be played as dialogue, the first movement of Mozart's Piano Sonata K. 333, cf. Bengt Edlund, "Musical Dialogue in a Classical Piano Sonata" in Varia 1.


But this is not the only function of the violin interjection. The last note of this answer supplies the leading-note to the entry of the following transposed *a* motif in the piano part. If the violin's final c♯<sup>2</sup> ♯ is rendered as a tonal cue, and if the piano perceptibly attaches to it, accepting the invitation rather than starting anew on its own, the fifth bar will suggest a delicate balance between obvious afterbeat status and furtive upbeat demeanour. As to the *b* motif, it certainly has a connecting function within the theme but it may amount to more than that: it initiates the broad gesture of the *c* motif in a way that might be understood as an element of dialogue within the piano part itself. If the *b* motif is clearly articulated as an upbeat, the melody seems to be resumed after the *a* motif, possibly by a new (intra-piano) musical protagonist.

Turning to the similar consequent phrase, it brings some unexpected harmonic shifts but lands on an E7 chord in its fourth bar, a moment of relative stability that is subtly undermined by the violin when the answering *d'* motif is extended to a descending fifth suggesting A major. Harmonically, the violin utterance in m. 10 is associated with what is to come in the piano part, and it can hardly be understood as expressing consent; it rather seems to exceed and controvert the preceding falling fourth of the piano. A gently opposing quality of the violin answer might be announced already on the accented e2 , suggesting that it is the violin that induces the accompanying A7 chord in the piano part. But since the listener's attention is likely to be directed to the expected entry of the violin, an anticipating harmonic effect in the piano part may without any acute sense of paradox precede its forthcoming melodic cause, i.e. the expansion of the violin's *d1* motif.

But on second thoughts another twist of meaning presents itself. It may after all be the A7 chord, introduced by the piano after a short initial rest, that causes the ensuing *d'* motif to encompass a falling fifth. 8 If this aspect of the structure informs the interpretation, the violin is robbed of its opposing initiative; it is the piano that finds an unexpected way to continue the conversation by coming up with another, dominant-seventh "up-bar" that transiently tonicizes the thirdbeat D-major chord in m. 11. In this perspective, it is the piano that makes the violin, prepared for a confirming imitation of the *d* motif, change its mind and

would not have been a gesture of consent, but of dissidence, since the violin changes the subject (as it were) by shifting the key and by introducing a poignant contraction of the motif.

<sup>8</sup> Unlike in m. 5, there is no *decrescendo* sign in the piano part in m. 10, a difference that may perhaps be taken to support this alternative interpretation.

yield to the new harmony introduced by its partner by playing a descending fifth instead of a fourth.

The next unit also features five bars, but the two altered and curtailed variants of the *a* motif make for a hemiola configuration in mm. 13–14; finally a one-bar violin commentary is added, repeating the last three notes of the piano part. (These three notes, being also an inversion of the *ac* motif (*aci*), turn up in significant ways later on.) Reinstating regular metre by repeating the piano's falling inflection, the violin may be heard as correcting the piano – provided that the pianist has pursued the hastened motivic pace and the metric compression by rendering the e♮ 2 in m. 14 as an afterbeat and the following d♮ 2 as a downbeat, and that the violinist clearly reverses the duple pattern by playing the e2 in m. 15 as a strong beat. But the violinist may also suggest consent by imitating the way the pianist has just played the preceding bar, i.e. by treating e2 as weak and d2 as strong.

Another possibility – less interesting and probably against Brahms's intention – is that the pianist maintains the triple metre throughout the passage by subduing the metric conflict introduced by the compressed *a* motifs, i.e. by playing the e♮<sup>2</sup> in m. 14 as a downbeat. In this case, the violin, confirming what has already been achieved, simply plays ahead as if nothing very much had happened.

Anyway, after a rest of hesitation the piano conforms to the violin's interjection, echoing itself one octave lower and shadowing the violin interjection. Looking ahead, the appended fifth-bar violin descent makes up a subtle bridge over to the piano entry on b1 in m. 16. But the pianist may ignore this falling connection and signal a fresh start for the following five bars, suggesting that from its very beginning the chromatic upper-line ascent in mm. 16–20 has a large-scale anacrustic function.

The following unit also comprises five bars but this time there is no fifth-bar violin addition: the expansive piano phrase does not allow of any comment from the partner. The metric irregularity is more pronounced than in the preceding unit: beginning with two soprano-to-bass derivatives of motif *a*, the lengths of the sub-units are 3+3+2+2+3+2 beats, and the disorder is not cleared away within the passage. The final gesture of this piano effusion invites the violin to take over as the principal part – the leading-note b♯<sup>1</sup> is offered as an upbeat to the violin's c♯<sup>2</sup> ♯ while there is a gap down to the c♯<sup>1</sup> ♯ of the delayed right-hand *ac* entry of the piano. But this sense of invitation may be embedded into two different motivic contexts.

The piano/violin bridge e2 –b♯<sup>1</sup> –c♯<sup>2</sup> ♯ makes up an incomplete and chromatically altered *a* motif overlapping with the regular *a* motif starting the main theme in the violin; *a* motifs overlapping in this curtailing way will occur frequently in the development section. This reading turns the last two quarter-notes in the piano part into a local upbeat, but these two notes may also be severed from the c♯<sup>2</sup> , suggesting a sense of retreat from, or interruption of, the long-term rise in the piano's treble voice (the violin's entry is lower than the preceding top note of the piano).

A less obvious, less tight and less urging, but quite moving, option is to bring out the metrically displaced allusion to the *c* motif beginning in m. 19. It is first hidden in the middle voices, d1 –f 1 –a1 –, then pursued in the top voice as –e2 –b♯<sup>1</sup> , representing a diminished *d* motif. This reading might come to the fore if the pianist lingers somewhat on e2 in order to give the listener time and reason to associate back to the half-note of the *d* motifs proper, and it takes on an interesting twist of meaning since a closing gesture is re-functioned into an introductory one – the first two notes of the *a* motif in the violin may be understood as a transposed and augmented answer to the diminished *d* motif in the piano.

#### **The first theme – second statement and transition**

Bars 21–30 make up an intensified and eventually quite expansive variation of mm. 1–10. In mm. 21 and 26, the violin's *a* motifs are combined with *ac* motifs in both hands of the piano part. It seems that the pianist for reasons of clarity and rhythmic independence should give priority to the right-hand line, a way of playing that also makes the rising inflection within the violin's *a* motifs seem urged by the delayed off-beat entry of the upper *ac* motif of the piano. (The *ac* motif is in fact contained in the *a* motif.) The rising *c* motif is now provided with a falling counterpoint (*cc*), and again it is preferable to bring out the upper righthand descent starting and proceeding off the beats.

Within these 5+5 bars of simultaneous dialogue, the passages in parallel motion bring a sense of encouraging agreement, whereas those featuring contrary motion give the impression that the *cc* motif in the piano part represents a force that holds back the upsurge of the violin. It should be noted that the *d'* motifs, now played by the piano, are delayed, which makes these imitative answers even more tender.

At this point few remarks should be made on the musical process in mm. 158–171, making up the drastically shortened first-theme section of the recapitulation; cf. Ex. 1b. The *d* motifs are again followed by *d'* answers expressing consent and slight disagreement, respectively, but this time the element of dialogue is changed since the imitative violin and piano entries (mm. 162 and 167) lead directly to melodic statements in the *same* instrument. As a consequence of this new distribution of the material, another high-level metric pattern tends to dominate, a pattern that conceals the demarcations and reverses the internal grouping within the units. Thus, instead of four-bar phrases *followed* by appended one-bar imitations as in the exposition, the recapitulation rather seems to bring first a regular unit of four bars and then two five-bar units, i.e. four-bar units *preceded* by one-bar additions.

The passage mm. 31–38 consists of two four-bar units, of which the second replicates the first; cf. Ex. 1a. High-level metric regularity is established, but the motivic content makes for hemiola configurations (3+3+2+2+2 beats). The two units, bringing varied statements of mm. 11–14, are made up of *a* motifs, adopting the shifting-register model presented in mm. 16–17. But the motifs are shared between the instruments so as to suggest a quite agitated conversation; the violin resolutely maintains the downbeats while the piano vehemently breaks in with the afterbeats.

This fragmentation is counterbalanced and supplemented in mm. 32–34 and 36–38, where four-note motions can be heard in the piano part in spite of the skips in register. As a result of this fusion of afterbeat fragments, a new idea, independent of the clash between the instruments, works its way along. If these motions are made prominent, a superimposed quadruple metre temporarily comes to the fore, and since the four-note melodic gestures in the piano part straddle the bar-lines, a conflicting accent seems to fall on the second beats of mm. 34 and 38, bars that otherwise bring the *aci* motif, cf. m. 14.

In the following four bars, the low-pitched, chromatically falling violin notes bring an initiative that eventually will put an end to this developmental portion of the varied exposition of the theme. The very point of release, however, comes only in m. 43 at the final downbeat of a four-bar upbeat in the piano part: after two separate afterbeat particles, re-functioned to form subordinate impulses within an anacrustic sequence rising by octaves in the right hand and by semitones in the left, the violin's chromatic descent is arrested and turned into a sequence of rising octave skips when the third afterbeat fragment in the piano part is allowed to multiply into a falling chain of dissonance/resolution chords.

The seemingly free eight-bar transition is related to the main theme: the violin descent brings three altered and metrically displaced allusions to the *a* motif. Since the violin is the principal instrument here, it is most natural to let it announce this turning point of the evolving form, but the first sign of change is introduced by the start of the piano's rising *arpeggio*. The violin's prolonged leading-note a♯<sup>1</sup> in m. 50 strongly implies, invites, the upbeat b1 starting the second theme introduced by the piano.

A sophisticated thematic transformation connects the violin transition and the second theme. Beginning with a1 in m. 47, omitting the raised a1 in m. 50, and ending with e2 in m. 52, all motifs of the main theme can be found in due order. This amounts to an extended and quite convincing hidden allusion – although the *b* motif forming the very bridge is obscured due the a♯<sup>1</sup> , and although it is divided between the instruments.

# **The second-theme episodes, the piano interlude, and the transition**

The second-theme section begins with an eight-bar period featuring the piano as the leading instrument. Whereas the first two bars of each half-period clearly derive from the main theme, the following two bars bring a vivid interplay between the instruments. It may be described as a swift, animated talk between persons knowing each other well, a conversation in which one of the participants, the violin, tries to get a word in edgeways.

In the first half-period the violin is first incited to take part in the conversation by an inconspicuous middle-voice accompaniment figure in the piano part, then it decides to pick up the piano's clearly exposed and more urgent rising-triad idea. If this second answer is played with slightly more emphasis than becomes an imitation, the violin may appear to have got the upper hand. The next triad from the piano, serving also as the upbeat to the second half-period, accepts the violin's triad by echoing it – the piano seems to have paid some attention to its partner prompting the way back to the start of the second theme.

In the second half-period, the violin, at a higher and more insistent pitch than before, first asserts and then, perhaps in a slightly resigning vein, re-asserts the rising-triad idea in the minor mode. But this time the piano has greater things in mind, as it were, and negates the violin's rising-triad fragments by falling triads; and yet it is the violin that dictates which triad that is to be played.

The next period (mm. 59–66), characterized by a most powerful melodic and rhythmic growth, is a piano solo that may be understood as a kind of internal dialogue, or perhaps rather as a rhetoric monologue. 9 The determined,

<sup>9</sup> The possibility of a dialogue between the instruments is saved for the coda of the movement, where this eight-bar period is expanded to the utmost by a series of internal reiterations.

double-dotted *e* motif is intensified by means of intervallic augmentation and rhythmic diminution, then (without its dotted rhythm) it recedes in a final, resigned wave; the peak notes are c♯<sup>2</sup> , e2 , g♯<sup>2</sup> , and e2 . The solo is connected to the return of the second theme by a very refined shift. Joining and taking over the final b1 of the piano in mm. 66, the violin adds a further, rhythmically augmented sighing motif to those of the piano. Concurrently, and lending a subtle sense of ambiguity to the transition, this gesture makes up a lengthened upbeat to the second theme.

In the following varied repeat of the second theme the violin plays the melody while the piano has a secondary but by no means insignificant role. The passage mm. 67–74 exemplifies still another kind of musical dialogue. The piano part is melodically and rhythmically entirely independent of the violin part; its figurations may be understood as making up a free paraphrase of the motivic accompaniment in mm. 50–52. And yet, since these contrasting melodic fragments – the effective pace of which is dragging quarter-note triplets – set in during the long notes in the violin part, the sense of a dialogue between the instruments is unmistakable. But it is a dialogue of a peculiar musical sort: the exchanges fit in perfectly, but (as in some opera ensembles) the protagonists apparently do not hear, or do not pay attention, to each other.

A four-bar passage, mm. 75–78, divided into two similar but cumulative stages, leads up to the third-theme section. The *e* motif, already used to generate the culmination of the piano-solo episode, sets in simultaneously in both instruments, but the violin plays the most contracted form and appears to have the initiative, and it also brings the raised top pitches in mm. 76 and 78. The seemingly delayed rising intervals of the piano part bring a kind of third-beat repercussions that tend to overshadow the final descending skips of the violin. i.e. the inversed *e* motifs (*ei*). And it is the piano that brings the important leading-notes, a♯<sup>1</sup> and d♯<sup>2</sup> – the latter being played twice, prematurely in m. 77 and insistently restated in m. 78.

#### **The third theme and the transition to the development**

The third and concluding theme (mm. 79–87) is made up of a determined one-bar rhythmic idea *f* that may feature both ascending (*g*) and descending *(gi)* triplets on the third beat. The *f* motif is first played twice by the piano – it is introduced in concurrence with two further statements of the *e* motif in the violin – and then taken up by the violin playing it no less than five times. Finally, augmented so as to cover two bars and imitating the preceding one-bar phrase of the violin, it reappears in the piano part, closing the exposition. The dialogue between the instruments is made denser by a rising four-note counterpoint motif *gc*, confined to the piano part and always entering on the first beat.

In mm. 81–82 the rising first-beat *gc* triplets give strong urging impulses to the rising third-beat *g* triplets in the violin part. Then, in mm. 83–85 the *gc* and *gi* motifs counteract each other: the ascending piano triplets on the first beat of each bar strive against the receding tendency of the violin's descending triplets on the third beat. But in spite of the differences in timbre and register between the instruments, these three bars may also suggest another sense of dialogue. Since the piano ascends to a1 (then a) on the second beat, the violin may seem to link in by beginning its descent from a2 (then a1 ); the *gc* and *gi* motifs may be heard as complementing each other, jointly producing arch-like melodic shapes in concurrence with the *f* motifs of the violin, motifs that from m. 83 on are reminiscent of the d*'* motif.

Turning finally to the transition to the development, the descending quarternotes of the augmented *gi* motif in m. 87, ending at a third with a leadingnote potential hanging in the air, are unmistakably answered in m. 89 by the ascending counterpoint motif *ac*, a juxtaposition suggesting another motivic link between the beginning of the movement and the triplet particle of its thirdtheme section. 10 The *gi*–*ac* shift makes for an element of dialogue within the piano part connecting the exposition to the development across the whole-bar rest in m. 88. After a moment of suspense the conversation, starting with motif *a* and an independent motif *b* in the violin, is resumed.

# **Expressing a sense of dialogue**

The cues suggesting exchanges in a dialogue may often be the same as those used when expressing phrase structure. Reflecting the musician's understanding of the musical structure, cues suggestive of shifts within, say, an evolving musical *persona* will emerge as natural and predictable to musically informed and sensitive listeners; they are not likely to stand out. Bringing out a sense of dialogue, on the other hand, will in many cases rather appear as a series of deliberate interferences,

<sup>10</sup> When augmented to quarter-notes in m. 87, the *gi* motif also betrays an affinity with the calm descent in m. 15, which in turn may have its root in the falling eighth-notes in m. 5. As already pointed out there is also a resemblance between the *f* motif and the d*'* motif. When dealing with short fragments, similarities tend to crop up in meaningful or inordinate ways.

as a manner of playing that demonstratively renders a certain passage according to a specific idea, that of distributing "lines" to participants in a fictive dialogue. Yet and needless to say, identifying cues indicating a sense of dialogue, and distinguishing them from cues reflecting an interpretation in terms of a musical *persona*, is a very delicate task.

In other words, cues indicative of a dialogue are matters of degree that must be identified against the background of the cues that would be appropriate when playing without any intention to suggest a musical dialogue. Searching for cues of the former kind is therefore associated with a dual risk: that of exaggerating the importance of what you hear, and that of missing what there is to be heard. The core of the problem is that what you are bound to hear is structure *and* performance, and that you must be wary of confusing them. The more convincing a reading in terms of a dialogue emerges from an analytic point of view, the more difficult it is to establish whether the dialogue is actually expressed in a certain performance, and to identify the means used. If the inherent sense of a dialogue is obvious, you may "discover" cues for dialogue in virtually all performances of the music in question although no such cues are present, although the musicians had no intentions to render the music as a dialogue.

Aural detection of cues suggestive of a dialogue is to an appreciable extent guided by expectations as to what cues that are likely to turn up, by your experience of making and listening to music, and probably also by your knowledge of cues used in linguistic conversations. When searching for cues expressive of a musical dialogue, you are likely to notice temporal and dynamic inflections similar to speech acts such as breaking in, giving way, or resuming.

#### **Cues for a sense of dialogue in performances of the sonata**

The exposition of Brahms's Violin Sonata Op. 100 features a number of structural shifts that emerge as pertinent when it comes to mediating a sense of dialogue; cf. the boxed numerals 1–18 in Ex. 1a/b. In what follows various ways of rendering these sites of possible exchange will be described. 17 recordings have been studied – for each entry below is given the quarter-note tempos as measured in mm. 1–4, 31–34, 51–54, and 75–78. 11

Ferras/Barbizet 1971 (DGG 2538105) 107, 126, 113, 109 Grumiaux/Sebök 1976 (Philips 9500108) 97, 124, 100, 101

<sup>11</sup> I am most grateful to the Swedish Broadcasting Corporation for making these recordings available for study.

Heifetz/Bay 1936 (RCA ARM 4-0944-4) 114, 133, 114, 126 Kogan/Mijtnik 1956 (Columbia 33 CX 1381) 114, 133, 107, 129 Kremer/Afanassiev 1987 (DGG 423620-2) 76, 95, 75, 76 (!) Kulenkampff/Solti 1965 (Decca ACL 250) 113, 128, 107, 109 Mutter/Weissenberg 1983 (HMV 1C 157-43443) 122, 141, 116, 124 David Oistrakh/Richter 1968 (Eurodisc 87954) 103, 138, 104, 111 Igor Oistrakh/Ginzburg 1969 (Eurodisc XK 80570) 92, 122, 114, 94 Olof/Wayenberg 1965 (Iramac 6510) 116, 122, 113, 109 Perlman/Ashkenazy 1983 (EMI CDC 7-47403-2) 106, 122, 109, 104 Schneiderhan/Seeman 1962 (DGG LPM 18633) 113, 129, 122, 109 Stern/Zakin 1973 (Columbia M 32228) 92, 116, 100, 100 Suk/Katchen 1968 (Decca SXL 6321) 107, 116, 106, 101 Szeryng/Rubinstein 1962 (RCA LSC 2619 B) 113, 120, 95, 104 Varga/Bennette 1971 (Musical Heritage Society MHS 923) 124, 131, 106, 96 Zukerman/Barenboim 1974 (DGG 415989-2) 104, 122, 101, 94


Apparently, a close imitation of the gestural quality is essential if you want to establish an intimate relationship between these motifs, but it is a fact that pianos, having percussive, clear-cut tone attacks, and violins, capable of treating each note with dynamic flexibility, are not very good at imitating each other. But they may try to do so. Thus, the violinists can refrain from starting the first note of their *d'* motifs very softly since this is something that the piano cannot achieve. When they play the *d* motifs, on the other hand, a slightly delayed emphasis on the first note might be appropriate since the piano will enter one eighth-note late in the following *d'* bar. In such exchanges a slight *portamento* when playing the *d* motif would further increase the similarity with the ensuing filled-in motif of the piano.

Generally, the filling-in eighth-notes of the *d'* motif should not be rendered so as to emerge as important; they are absent in the preceding *d* statement of the motif. The two variants of the motif will become more closely associated if they exhibit an affinity as to dynamic and temporal inflection. This means that the final quarter-note of the *d* motif may be somewhat delayed, approaching the value of an eighth-note, whereas the answering *d'* motif may be played so as to reach its final eighth-note somewhat early; when forming the filledin motif the violinist/pianist must be cautious not to play *ritardando*. It also seems that one should avoid separating the two motives from each other; while it is of course possible to suggest an answer even if the *d'* motif is perceptibly delayed, this would lend a certain emphasis to the answer and diminish the sense of confirming agreement. Separation might therefore be suitable in mm. 10 and 167.

Turning to the recordings, Perlman/Ashkenazy and Suk/Katchen somehow manage to convey an intimate relationship between the motifs, and so do Grumiaux/Sebök and Mutter/Weissenberg, by playing motif *d ritardando* and motif *d1* in tempo. In m. 5 Igor Oistrakh plays his answer with a conspicuous *ritardando* and also with an initial *crescendo*; later on in m. 25, however, Ginzburg imitates the *diminuendo* of his partner with success. Szeryng/Rubinstein and to some extent also Ferras/Barbizet establish similarity by means of initial accents. In Kogan/Mijtnik's recording, the *non-diminuendo* answer in m. 5 is perceptibly louder than the pianist's rendering of the preceding *d* motif, which ruins the sense of consent; in m. 10, however, this trait is more appropriate since it suggests a sense of opposition. In general, the recordings exhibit very few, if any, cues distinguishing the situation in mm. 9–10 from that in mm. 4–5.

In m. 25 Mijtnik and some other pianists audibly observe the rest, which is detrimental; Richter plays mm. 24–25 in one pedal achieving a very good sense of imitation. Breaking the chord is unfavourable, but Zakin gets away with it. Heifetz/Bay exemplify a strict, in-tempo approach that is fairly successful. Schneiderhan/Seeman separate the *d* and *d'* motifs by a short silence, and Kulenkampff 's *portamento* in m. 24 prepares for Solti's filled-in gesture. In most recordings the *d–d'* shift in mm. 29–30 sounds quite convincing, perhaps an effect of the increasing loudness.

As pointed out in the analysis, there is also a potential sense of an intimate leading-note contact in mm. 5/6, 25/26, and 162/163. The take-over in mm. 5/6 is very tight in most recordings, but this effect depends on how much the pianists stress their first note to mark the B-minor entry of the *a* motif.

Separating the motifs, Stern makes his last note very short. David Oistrakh's and Varga's phrasing is suggestive of eighth-note afterbeats, whereas Kogan clearly re-functions his descent into an upbeat to m. 6.

In mm. 25/26 very few recordings exhibit close contact. But Sebök plays his eighth-notes *crescendo* supplying an unmistakable upbeat to the entry of the violin; this recording is also the only one in which m. 162 is rendered so as to unequivocally emerge as an answer to m. 161: a short silence separates Grumiaux's last eighth-note from his *a* motif in m. 163. Excepting Kremer, who clearly treats the entire *d'* motif as an upbeat to his B-minor statement of the theme, most violinists play the last eighth-note as a local upbeat – Heifetz makes it even more proximate to the *a* motif by virtually turning it into a sixteenth-note. Three violinists (Kogan, Szeryng, and Kulenkampff) render m. 162 in a genuinely ambiguous manner by shifting from afterbeat to upbeat character when playing the three eighth-notes.

The formal/metric position of m. 167 is quite precarious since it is difficult for the pianists to bring out its answering and perhaps opposing character in concurrence with the fact that the preceding E7 chord tends to turn this A7 bar into a downbeat. And yet it is unmistakably attached to mm. 168–171 by Barbizet, Barenboim, Bay, Bennette (attacking it quite loudly), Weissenberg, and Seemann, while Ashkenazy and Katchen render it ambiguously. In seven recordings m. 167 seems appended to the preceding *d* motif of the violin, most clearly so when played by Sebök and Rubinstein.

3 Turning to the imitative relationship between mm. 14 and 15, most recordings successfully convey a sense of imitation, but it is achieved by different means. Most often the pianists have re-established triple time by emphasising the first beat of m. 14, a fact that is then confirmed by the violinists. The association is often underscored by the same *diminuendo* (or *crescendo-diminuendo*) inflection. Igor Oistrakh and Ginzburg resort to the same amount of *ritardando* as well, but it seems that the sense of a dialogue is strengthened if there is tempo continuity over the bar-line. Suk and Katchen both play *portato*; Katchen extends the duple metre into m. 14 by slightly stressing the second chord, and Suk, refraining from correcting the piano, then carefully imitates this trait in m.15. A similar effect is achieved by Kulenkampff and Solti, who group the last two chords of these bars so as to form separate units. A clear sense of correction characterizes Szeryng/Rubinstein's recording: the pianist maintains duple metre in m. 14, and the violinist then supplants it by playing m. 15 in triple time. Hard to understand, on the other hand, is the sequence triple-followed-by-duple-time exhibited in Varga/Bennette's interpretation.

The continuity from the violin's c♯<sup>2</sup> down to the piano's b1 in mm. 15–16 is quite weak in most recordings. Turning to mm. 171–172, Wayenberg brings out a connecting relationship by playing the last two chords of m. 171 in a *portato* manner that prepares for the thick chords of the following section.

4 As pointed out in the analysis, the transition in mm. 20–21 offers several options. Beginning with two general traits shared by most recordings, the pianists tend to bring out the long rise in the soprano (b1 –c♮<sup>2</sup> –c♮<sup>2</sup> –c♯<sup>2</sup> –d2 –e2 ), and they also mark the chromatic motion B♭<sup>1</sup> –B♮<sup>1</sup> in the bass. The option involving two overlapping *a* motifs seems to shine through only in Kogan/ Mijtnik's recording, in which a very short demarcating silence is inserted after the first beat of m. 20, and in which there is no *ritardando*. Approaching this effect in another way, Bay and Weissenberg play a prominent leading-note upbeat to the violin's entry.

None of the interpretations clearly expresses the reminiscence of the *c* motif hidden in the piano part, but nine pianists come rather close to an allusion. The note d1 starting the motif is not sufficiently stressed, but then these pianists suggest a rising middle-voice strand connecting to the final descending e2 –b♯<sup>1</sup> motion, played either *ritardando* (Barbizet, Katchen, Richter) or with a stressed e2 (Rubinstein, Seeman). In some recordings (for instance those of Richter and Zakin) the listener is uncertain of whether the a1 on the first beat of m. 20 skips up to e2 or falls to the leading-note g♯<sup>1</sup> , indicating that the motion is merely a supplementary inner-voice.


Ashkenazy, and Richter play them *legato*. The chords are generally grouped in pairs, although the final ones are fused into a four-chord unit in some performances. Barbizet plays the first two attacks in each four-bar section *staccatissimo*.


metric articulation; the pianists, on the other hand, are no doubt tempted to bring out the first note of their triads and hence to render mm. 53–54 and especially mm. 57–58 in a way that more or less suggests 6/8 time – a kind of metric ambiguity (or indeed metric shift) often met with in Brahms's music. Thus, whereas the violinists are likely to play so as to conceal the association in terms of rising triads, the pianists are likely to expose it.

Turning to the recordings, most pianists bring out the accompaniment motion b1 –d♯<sup>2</sup> –e2 in m. 52, later to appear as e1 –g♯<sup>1</sup> –a1 in the violin. This embedded answer is perceptible in the recordings of Grumiaux, Mutter, Perlman, Stern, and Suk. While a majority of the pianists, and certainly Bay, seems to leave 3/4 in favour of 6/8 when finishing their half-periods, most violinists take care to clarify the motivic imitations by stressing the first note of their rising triads (or by playing their interjections *crescendodiminuendo* to the same effect). Others emphasize the note after the bar-line (Ferras, Grumiaux, Igor Oistrakh, and Olof), give weight to the very first note (Kremer and Stern), or play the interjections *crescendo* (Zukerman). A few pianists, especially Bennette and Mijtnik, apparently adopt 6/8 time throughout the half-periods, which does not fit with the "off-beat" profile of the rising triads as played by their partners Varga and Kogan. To sum up, three recordings clearly exhibit a dialogue in terms of triadic fragments (Heifetz/Bay, Perlman/Ashkenazy, and Suk/Katchen), and three others do it to some degree (Stern/Zakin, Szeryng/Rubinstein, and Zukerman/ Barenboim).


such a relationship, it seems that the second statement of the four-note motif d♯<sup>2</sup> –c♯<sup>2</sup> –c♯<sup>2</sup> –b1 in the piano may be slightly separated from the first one, and also be played *ritardando* in a way that gives some relief to the final motion c♯<sup>2</sup> –b1 –b1 . Turning to the violin, its lingering b1 –a1 –a1 entry may then be played *à tempo*, so as to approach as much as possible the motivic pace of the piano, and also not too emphatically which would betray that it rather (only) functions as a prolonged upbeat to the theme.

Some violinists (Ferras, Heifetz, Kogan, and Zukerman) clearly show that they are engaged in starting a theme; Igor Oistrakh stresses the first a1 . Afanassiev and Ashkenazy make a *ritardando*, and the latter also separates the two statements of the four-note motif from each other – features that help to suggest the affinity with the entry of the violin. Sebök and Grumiaux play their motifs with a similar, connecting *ritardando/diminuendo* gesture. In Schneiderhan/ Seemann's recording, the pianist's *ritardando* is followed by a slightly hurried violin entry, a way of playing that may make the listener think that a further, varied imitation (rather than an upbeat) is added to the descending sequence.


But imitation of the rhythmic gesture of the *e* motifs seems to be more important for achieving a sense of dialogue between the instruments. The duration of the sixteenth-note is important, but even more crucial for an immediate association between the parts is a close agreement as to the articulation of the first note. The pianists generally pedal once in each bar, and this works fine if the violinists tie the first two notes of their *e* motifs tightly together as do Ferras, Kulenkampff, Schneiderhan, Suk, Szeryng, and Zukerman. But some violinists prefer to separate the two notes, and this requires that their pianist partners also insert a short moment of silence between their chords, as Ashkenazy and Mijtnik do, making for a clear affinity with the impulse just given by the violin and suggesting a sense of confirmation or support. Separation between the notes in the violin part followed by tight connection between the chords in the piano part, as exemplified in some recordings, considerably obscures the sense of imitation and hence of dialogue. Stern/Zakin achieves a close relationship by inserting short silences before the bar-lines.


# **Conclusions**

The first-movement exposition of Brahms's Violin Sonata Op. 100 provides ample scope for expressing a dialogue. The investigation just accounted for has shown that these opportunities were sometimes, and in various ways, used by the musicians, sometimes not. Indeed, a performance being inexpressive in this respect would come close to an artistic failure, it would seem to involve a fundamental lack of understanding of the very essence of the movement's design. But whereas it may be possible to guess how a musician would play a passage in order to convey a sense of dialogue, one cannot with any certainty make inferences from a certain performance to the underlying interpretational idea, to the musician's intentions. 12

Elements of dialogue (or of impersonation in general) are no doubt important facets of music appreciation and interpretation, but as all kinds of understanding,

<sup>12</sup> Jerrold Levinson, "Performative vs. Critical Interpretation of Music", pp. 33–60 in Michael Krausz (ed.), *The Interpretation of Music. Philosophical Essays*, Oxford 1993, Clarendon Press

the dialogue mode of conceiving music must be applied with discernment. It appears that a study of the element of dialogue in music may improve our critical understanding of music, as well as contribute to our growth as musicians, helping us to discover and convey musical meaning.

# **Chapter 15 Chopin themes**

# **Introduction**

Stravinsky once held that rendering music is just a matter of playing as prescribed in the score – presumably he was fed up with licentious musicians. But on second thoughts it is evident that even meticulously detailed scores require contributions from the those who play or sing. The truly musical properties are not to be found on the printed pages; even if every single sign is painstakingly respected, it takes a discerning musician to turn the prescribed music into worthwhile listening.

Some musicians are blessed with a steady stream of creative ideas and praise their intuition. Others maintain that their interpretations gradually emerge while practising – you understand the music in the way you play it, and then you play it in the way you have understood it, and so on. Some musicians are unfavourably disposed towards "theorizing", an attitude that may be explained by the fact that during many years of training they have been forced to participate in dreary journeys through analytic deserts.

Yet there is some scope for a keen study of the musical text and for creative applications of theoretic knowledge when working on a piece of music. A manyfaceted and flexible analysis may help musicians to arrive at good interpretations – to avoid mistakes, to find a consistent and expressive whole, to discern and then choose among options that would perhaps not have been discovered.

That Chopin's music should be subjected to interpretation is a generally shared opinion. In what follows, I will by means of some examples try to show how current analytic thinking may lead to better, more informed interpretations. Readers who are allergic to influences from others will have to excuse me if I occasionally turn somewhat normative. But as far as I have read the scores in defensible ways and drawn reasonable conclusions, it is the music itself (not even Chopin) that tells me how it is to be played. The text will be devoted to the themes of three important works: what is the nature of these ideas, what happens to them, and how should we deal with them?

#### **A theme starting seven times**

The beginning of Chopin's G-minor Ballade Op. 23 is extraordinary for two reasons. In the introduction the tonic is approached from "outer space" by a sweeping four-octave Neapolitan gesture, and the main theme, repetitious in a way bordering to monotony, has a most unusual construction that (apparently) defies current periodicity.

This theme offers a challenge to all pianists, and (using the computer keyboard) I will take up the gauntlet. 1 Which conclusions for interpretation can be drawn from a close study of the music? How can you turn the series of starting phrases into a compelling whole?

The theme consists of seven "attempts" to get somewhere; cf. Ex. 1. All these phrases, except the fourth, start from D7 dominants implying G-minor tonics and, excepting again the fourth attempt, they are all clearly subdivided into two units. Each phrase first features an immutable rising gesture, then follows a variable melodic utterance made up of two long notes, bringing falling or rising seconds – the fourth attempt starting from F7 issues into a three-note combination of these intervals. The rising gestures involve a paradox: they are used to start the attempts, but they have an unmistakable closing quality.

The harmonic relationship between the start of the attempts and the preceding chord varies: in some cases the harmony is retained, in others it is changed. There is also a sense of harmonic uncertainty between the rising gesture and the following melodic fragment: sometimes a new chord turns up, sometimes not. The third and seventh gestures are followed by dominant-seventh sonorities implying forthcoming auxiliary tonics, cf. the delayed entries of B♭ major and C minor in m. 15 and m. 22, respectively.

Preceded by just a hint of a modulation, the fourth attempt deviates from the other ones in several crucial respects. Attaching immediately after the inconclusive end of the third attempt, the F7 mutation of the rising gesture starts one eighth-note "too early" with an accented note. This brings the effect that the first F of the bass seemingly comes too late, and that the second F occurs after the right-hand top note. The rising gesture is stretched to comprise an octave, and it includes an extra note.

But which note is the added one? The answer that first springs to mind is that it is the chromatically altered anticipation-note before the accented c2 -raised-toc♯<sup>2</sup> , which then turns out to be a long appoggiatura delaying the rising resolution to the B♭-major d2 . Another, not very satisfactory, answer is that the added note must be the apparently inserted, too-early e♭<sup>1</sup> . But the fourth gesture starts in a similar way to the preceding ones. Considering the D7 models, it is more to the

<sup>1</sup> I was incited to deal with this theme by some interesting remarks made long ago by Prof. Eugene Narmour during a conference in Sydney.

point to think of the e♭<sup>2</sup> as the extra note; it is after all the e♭<sup>2</sup> that expands the gesture to an octave. (You can test this idea by starting the gesture in due time and playing the e♭<sup>2</sup> as a quick (consonant) appoggiatura before the d2 .) Another aspect of this complex state of affairs is that the rising gesture seems to have been robbed of its given final note b♭<sup>1</sup> . The trivial, normal course of events would have been e♭<sup>1</sup> –f 1 –a1 –d2 –c2 –b♭<sup>1</sup> , starting and closing the forth gesture according to the preceding models.

The final descending motion d2 –c2 overlaps the delayed rising resolution c♯<sup>2</sup> – d2 , and as a result the fourth attempt issues into a three-note melodic utterance seamlessly growing out of its rising gesture. The ascending second c♯<sup>2</sup> –d2 , prompted by the F7 applied-dominant chord, already having B♭ in the bass and then being supported by the auxiliary B♭-major tonic, brings a subtle association back to the rising D-major cadence in mm. 11–12, although in m. 15 the metric relationship is strong-to-weak. But due to the anticipated c♯<sup>2</sup> and the fact that the harmonic outcome is already present in the bass, m. 15 seems to involve a harmonic suspension rather than a metrically displaced cadence. On the other hand, to the extent that the delayed melodic arrival at d2 seems to occur at a strong beat, the actually accented final note c2 of the fourth phrase emerges as an added weak beat.

Yet the last note of the extended, but tonally contracted, three-note motion c♯<sup>2</sup> –d2 –c♮<sup>2</sup> is likely to remind the listener of the strong-beat final note of the twonote d2 –c2 motion of the first attempt. We are likely to think that we are suddenly brought back from the F-major/B♭-major harmonic excursion, and that we are back to the end of the first attempt. But the latter conclusion turns out to be a mistake since the last two notes of the fourth attempt are then replicated by the two-note melodic utterance attached to the fifth rising gesture. But the harmonization is different: while alluding to the end of the fourth attempt, the fifth phrase turns out to be an exact copy of the first one – now we are back to where the theme started. The return to the point of departure has been disguised as a local replication.

Indeed, the main theme of Op. 23 amounts to breathtaking, extraordinary composing. We may take it for granted that the dedicatee, *Monsieur le Baron de Stockhausen*, did not fully appreciate it.

Considering the fact that the rising gesture keeps on recurring and that monotony may impend, a most natural idea when facing this theme would be to play the seven attempts differently. But before discussing what you might do in order to create diversity, the option of staying as constant as possible must be acknowledged as a quite valid one. Monotony is a good point of departure for future grand-scale development and may sometimes be a rewarding experience – as it is here. 2

Indeed, refraining from varying the rising gestures may in fact be quite faithful to Chopin's intentions. Indeed, a uniform approach is perhaps what the stress signs under the initial notes of each rising gesture indicate. And the following melodic utterances have separate slurs, a fact that (among other things) perhaps suggests that they should not be influenced by the rising gestures – at least not in ways that make the two components of the attempts form a virtual *legato* phrase. But again the fourth attempt, featuring only one encompassing slur, brings an exception.

Chopin's indications do make good sense, but it is hard to maintain that the composers' directions always specify the only permissible or worthwhile interpretation. Thus, it seems quite legitimate to render the rising gestures in ways that match their various melodic outcomes, i.e. to play these gestures so as to suggest that they bring forth their continuations.

The point of departure for what follows is the fact that the pianist, as opposed to the listener, knows what is going to happen and may prepare the listener for it. Rather than just thinking ahead from a certain point, the pianist can take care of forthcoming notes by dealing sensitively with their predecessors. In other words, thinking prospectively sometimes entails being retrospective.

The fact that musicians guide their listeners does not imply that it is without interest to take account of what listeners are likely to infer about future events. Quite to the contrary, music should often be rendered with this in mind. Sometimes listeners cannot guess what will happen, or tend to be mistaken about it. Such situations – and they abound in the main theme of the G-minor Ballade – are often crucially important and call for discerning musicians.

Let's assume that there is just one encompassing slur in the *first* attempt. This would mean that the otherwise stable g1 ending the rising gesture should have a sense of mobility making the ensuing rising-fifth skip credible. But taking the two actual slurs into account, the d2 has to produce itself, as it were, and it must be stressed in a way that suits the following sighing descent to c2 , a releasing note in spite of its accented position and slightly dissonant harmonization. It seems, then, that even if you respect Chopin's two separating slurs, you have to supply causes for future effects.

<sup>2</sup> There is, as Eugene Narmour showed, an excellent recording in which Arthur Rubinstein comes very close to a uniform approach to the rising gestures.

The outcome of the *second* attempt is both less and more than that of the preceding one. The deeper register of the melodic utterance means a more sonorous sound, but there is no effort involved when leaving g1 for e♮<sup>1</sup> . The tension of the A7 chord is immediately released when D major turns up, but the melody rises to f♯<sup>1</sup> , which either suggests a loss of energy or implies that you have to infuse the energy needed to make the e♮<sup>1</sup> move upwards. The e♮<sup>1</sup> –f♯<sup>1</sup> inflection runs in opposite direction to the preceding d2 –c2 sigh, falling to an open, slightly dissonant C-minor-sounding chord, and yet the rising motion in m. 11–12 towards harmonic stability in D major may also seem like a sigh, a sigh of relief.

Should the fact that the second attempt will deviate from the first one be prepared, and how can this be done if you have decided to keep the rising gesture constant? How loud should the e♮<sup>1</sup> be, and what about the rising resolution to f♯<sup>1</sup> ? If the f♯<sup>1</sup> is played softly, the kinship with the preceding falling sigh will come to the fore; if it is made prominent, it will provide a contrast that demands some extra emphasis on the e♮<sup>1</sup> , suggesting the effort required to ascend. A rendering of the latter kind introduces a sense of punctuation within the series of attempts; the first two attempts will seem to form a pair.

A receding or prominent D-major f♯<sup>1</sup> provides the immediate context for the *third* rising gesture, which (since it does not come up with a new chord but just with a D7 sonority) might emerge as reinforcing or confirming. For two reasons the third phrase is somewhat problematical: like all rising gestures, this one has a sense of resumption, but D-major is already there, and the first "utterance" note sets in after a rising octave, a leap that is melodically vacuous. But g2 is the highest note so far in the theme, which means that it should be brought out, positively or negatively. When it recedes to f2 , the slightly dissonant harmonization points forwards; although it has not yet been heard, the influence from the fourth attempt begins to be felt. Since the third phrase ends by suggesting a forthcoming modulation, the descending sigh cannot be treated in the same manner as the one in mm. 9–10: it can be played softer or louder, but not slower.

For no matter how you prefer to render the *fourth* attempt, it is imperative that it starts in due time with an e♭<sup>1</sup> clearly understood as accented and resuming, and that the rising gesture is played with rhythmic precision; the listeners must not loose their metric orientation. If this happens, they will hear a gesture that, like the preceding ones, starts with a stressed unaccented note, and that eventually comes up with a surplus note – the seemingly appended, anticipating c♯<sup>2</sup> , too late and too short at the same time. The fourth rising gesture, set in a higher register, may invite to be played in a dreaming way, but only a firm, no-delay start can bring out the urgency inherent in the "too-early" entry of its first note. The (potentially misleading) stress sign under e♭<sup>1</sup> is crucial, provided that it is understood as a reminder of the fact that the start of the rising gesture has been moved by an eighth-note to occupy a strong metric position.

Adopting a bird's-eye view, the e♭<sup>1</sup> is not too early at all: it serves to connect the third and fourth attempts, joining them into a four-bar phrase, suggesting that we are presented with the final, double-size unit of an eight-bar antecedent to be closed in the relative major. And there is a sense of linear continuity in the treble: a stepwise connection from the g2 in m. 13 to the c♮<sup>2</sup> in m. 16. Chopin's slurring does not indicate the close relationship between the two attempts, but the pianist should nevertheless try to bring out the fact that they are concurrently separate and united.

There are, it seems, two ways to render the three long notes in mm. 15–16. The association back to the second attempt with its authentic, rising cadence may be suppressed by letting the motion c♯<sup>2</sup> –d2 head for c♮<sup>2</sup> . Alternatively, you can bring out the affinity to the D-major cadence by suggesting that the B♭-major d2 arrives as an accented event. This option means that the metre will be transiently veiled, and that the final, apparently unaccented c♮<sup>2</sup> will seem to be appended to the phrase like a furtive second thought taking back what was achieved, i.e. the resolving cadence to B♭ major.

Turning to the *fifth* attempt, the straightforward option is to play it exactly as the first one so as to signal the start of the consequent. But the fact that it echoes the close of the preceding three-note melodic utterance may also justify a rendering that is sensitive to how the fourth attempt was finished. But one should not aim at imitation since a too obvious similarity between the final inflections of the two attempts would introduce a sense of immediate duplication that seems foreign to the theme.

If the fifth attempt's identity with the first attempt has been subdued, if the eventual kinship between the fourth and fifth phrases has been brought out, the formal return will seem to be delayed until the *sixth* attempt.

The *seventh* rising gesture issues into a quite unexpected harmony, a turn of events that it is difficult and perhaps not desirable to announce. The first, lowregister G7 chords may be somewhat delayed so as to connect with the highregister G7 chords after the octave leap, a way of playing that hides away the return of the octave leap from m. 13. The double-size phrase mm. 20–23, completing the consequent of the theme, is preferably to be played with a *crescendo* fuelled by the G7 chords and leading to an emphatic C-minor downbeat, followed by an equally natural *diminuendo* as the melody returns downwards.

But the interpretation must be adjusted to what follows – a varied repeat of mm. 22–23 – and to how one wants to deal with it. Since a second and similar peak at g2 again followed by a *diminuendo* descent might sound repetitious, a different way of rendering mm. 24–25 should be devised. It is possible to reverse the dynamics and start softly in m. 24; another option is to subdue the g2 in m. 21, allowing for a long *crescendo* leading to the second g2 in 24 and beyond. This will merge the two phrases so as to form an escape out of the small metric formats prevailing so far.

It appears that the main options for interpreting the theme of the G-minor Ballade are associated with what we may call a Romantic and a Classical attitude, respectively. The alternatives involve a more or less varied series of seven attempts or a sense of periodicity integrating the attempts into a balanced unity, respectively.

Nothing has been said about the regularly appearing pairs of accompanying chords. And there is perhaps not more to say than that they should be played in a "neutral" way. It would be as wrong to try to load them with emotional content as it would be to (intentionally or unwittingly) let them bring associations to a waltz. 3

# **An introductory theme and its culminating return**

The start of the Ballade Op. 47 may sound as an improvisatory introduction, but the four phrases are in fact tightly held together and they bring the main theme. Those who are familiar with the ballade know that this theme turns up again in full splendour towards the end of the work. This model/copy relationship deserves a close study since it keeps a crucial difference indicating how the two passages should be played.

Let's begin with the initial main theme; cf. Ex. 2. A♭-major is prevented from settling as the tonic all the way up to m. 8 – mm. 2, 4, and 6 issue into secondinversion A♭-major chords vaguely promising a cadence that is withheld until the last phrase. It is important that the pianist carefully observes the sense of tonal irresolution of the first three phrases; in particular, this means that the root-position A♭-major chord starting m. 2 should be subdued.

The four two-bar phrases make up a regular period, but there is no half-way cadence to the dominant. But mm. 1–2 correspond to mm. 5–6 since in the latter phrase the initial right-hand melody turns up in the left hand and *vice versa* – a most poetic specimen of inverted counterpoint.

<sup>3</sup> Much later on, in the *scherzando* passage starting in m. 138, a waltz-like touch may be apposite.

It is often possible to tell within pairs of bars which bar that is strong and which is weak. From a harmonic point of view, and assuming that we already know about the key-defining cadence eventually turning up in m. 8, m. 1 makes up an upbeat in the dominant issuing into m. 2, representing the tonic. But in order to counteract the sense of harmonic arrival in m. 2, you should try to play the first two bars not as a weak-to-strong configuration, but as a strong-to-weak one. In other words, you should play so as to let the listener, being ignorant of m. 8, have a glimpse of the harmonic ambiguity of the first two bars. The sense of a closing weak-strong progression from dominant to tonic should be challenged by that of an opening strong-weak progression from tonic to subdominant. Apart from matters of harmony and metre, it is hard to think of a passage that is more opening than these two bar: the theme, and the entire ballade, grows out of a single note.

As it eventually turns out in m. 8, A♭ major in m. 2 was not a subdominant, but the point of a strong-weak, opening way of playing mm. 1–2 is that it makes it difficult to hear the root-position A♭-major chord starting m. 2 as a tonic. Thus, the first bar should preferably be rendered against the grain: retrospectively, m. 1 turned out to be an upbeat in the dominant, but it should be played *as if* it represented a downbeat tonic. In this case it is desirable that the listeners get hold of the wrong end of the stick, that they initially think that E♭-major is the tonic of the ballade.

This balance in favour of a sense of harmonic opening is then confirmed in mm. 3–4, where it makes itself felt without any help from the pianist since the upbeat-like activity in m. 3 issues into a first-then-second-inversion A♭-major chord that cannot be heard as closing. To test this, play m. 4 with a c1 –b♭–c1 –a♭ motion in the left hand: everything is destroyed by this premature belly landing in A♭-major – there is no need for mm. 5–8.

This much about the main theme when presenting itself as an introduction; we will now turn to its properties when it returns just before the coda as the apotheosis of the work.

In mm. 205–208 the root of the E♭-major dominant rumbles in the left hand; then, in mm. 209–212, its root emerges clearly at accented positions; cf. Ex. 3. Throughout both passages a continuous *crescendo* supports the eventually rising right-hand line. According to the slur, the melodic culmination in mm. 213–216 is to be held together, and it is quite evident that its first two bars make up a variation of mm. 1–2, and that the next two bars bring a variant of the left-hand melody of mm. 3–4. Phrases that were initially parts of a calm dialogue in the introduction are now joined so as to form a triumphant melodic arch.

From a harmonic point view the passage mm. 213–216 consists of two dominant-to-tonic progressions. Now the A♭-major chords are certainly in root position, and there can be no doubt that m. 213 issues into m. 214, a fact that is then confirmed in mm. 217–218. Bars 219–220 bring the dominant and the tonic as well, but the situation is changed due to a shift in the phrasing. Bar 220 does correspond to mm. 214 and 218, both having a closing function, but according to the slur the function of the A♭-major tonic in m. 220 is also to start a transition section bringing a chromatically falling bass line issuing from E♭/e♭ and reaching B♮<sup>1</sup> /B♮. The imposed double function of m. 220 brings important consequences that must be observed when playing the transition to the coda.

In contrast to mm. 1–2, there are no unstable second-inversion tonic chords when the same thematic substance recurs in the climactic passage mm. 213– 220. Up to mm. 217–218 every pair of bars exhibits the accentual relationship weak-strong (i.e. tense-relaxed). Thus, when playing this culminating passage, it is crucial to release what was subdued in the timid beginning of the thematic introduction. In mm. 1–2 it was desirable to suggest a sense of metric balance and harmonic ambiguity by giving suitable weight to the first bar so as to make the listeners hear A♭-major in m. 2 as a subdominant. But particularly in mm. 213–214 it is essential to bring out the opposite relationship – the relative metric weights now have to support the sense that the dominant reaches out for and emphatically arrives at the tonic, a grand event taking place as the melodic gesture opens up towards an even higher register. There is nothing ambiguous in the relationship between mm. 213–214: a massive dominant leads irresistibly up to the decisive final tonic of the work, and hence it is in m. 214, the very culmination, that you have to spend all your force – if you have any left.

But there is a complication to consider and cope with. Whereas mm. 213–214 are unequivocal with regard to metre and harmony, these bars are involved in a large-scale formal ambiguity that you must take account of and do justice to. Since the main theme recurs in m. 213, this bar makes up the starting point of a new and most important formal constituent, but it is also the last bar of an extended, cumulative dominant. Hence, m. 213 is the locus of a high-level elision: a new formal unit begins while the dominant preparation started back in m. 205 is still on its way; the formal and the metric/harmonic downbeats do not coincide.

The sense of release in m. 214 is considerably increased if it seems to be delayed by one bar, if you manage to convey that the last stage of the dominant preparation is expanded from four bars to five, that the dominant is finally boosted by a local one-bar upbeat in the right hand, an upbeat that concurrently starts the return of the main theme. A new formal constituent certainly begins in m. 213, and this must be conveyed, but this bar must be heard as metrically weak (i.e. tense); the point of gravity occurs only in m. 214 when the tonic arrives.

But there are several things that may entice the pianist into giving so much emphasis to m. 213 that the true locus of culmination in m. 214 becomes eclipsed. Whereas the sixteenth-note motion is discontinued in m. 213, this bar is crowded with thick chords; the *fortissimo* mark seems to indicate the endpoint of the long *crescendo*; the *ottava* sign prevents you from seeing that the melody keeps on rising until m. 214.

What can you do in order not to take out the culmination in advance, in order to avoid excessive emphasis in m. 213, which would arguably amount to a misinterpretation? Invent a sixteenth-note left-hand motion suitable for this bar, and play it in context a few times so as to get used to the idea of a further, one-bar extension of the dominant preparation. Play just the right hand of mm. 211–214 until you feel an unbroken melody line up to the peak. Imagine that there is a *crescendo* pin in m. 213 and a triple *forte* in m. 214, indications making you withhold some of the force at the beginning of m. 213. And last but not least: do not mark the downbeat of m. 213 by staying on it or delaying it – this is where the theme starts, but it is not its culmination. 4

It is a meaningful delight for attentive listeners to notice the metric/harmonic difference between mm. 1–2 and mm. 213–214, but this is an effect that the pianist must prepare for. You must have played the first two bars of the ballade so as to suggest at least a sense of harmonic and metric balance, so as to supply the background against which you can excel in withholding the final climax by one bar.

Talking about misinterpretations, I can just as well disclose how I use to think in order to achieve the desirable vague equilibrium in the first two bars of the ballade. Hopefully nobody can hear what I imagine – it is so wrong!

My metric remodelling serves three aims. I want the first note to be charged with some of the mobility that goes with a syncopation. I also wish to move the root-position A♭-major chord starting m. 2 away from its accented position so as not to emerge as a harmonic goal. Finally, I want to promote the following

<sup>4</sup> I should confess that long ago I belonged to those who whole-heartedly offered a premature climax in m. 213. But Alfred Brendel gave me something to think about when he asked a young pianist to save her energy until the next bar. He did not explain his advice – however culminating this rising melody may be, start it with a one-bar upbeat! – but when he showed what he wanted, it sounded most convincing.

dissonant chord by suggesting that it occupies a more prominent metric position than the actually accented second-inversion A♭-major chord ending the phrase.

When furtively playing according to Ex. 4, the initial syncopation lends a sense of a slight push into the music. Admittedly, it also works fine when the first note is played as a downbeat quarter-note, i.e. as written by Chopin, but I want to experience an effect akin to that felt by a French-horn player starting a note with a soft attack and then immediately subduing it. 5 As to the dissonant chord, I like the rich sound it gets when played as an appoggiatura at an accented, but quickly left beat. When rendered as an eighth-note upbeat (as actually written in m. 2) it just slips in – but if that is what Chopin wanted, I cannot but apologize.

At this point a further digression may be allowed. The harmonic, metric, and formal situation at the return of the main theme in Op. 47 recalls – and deviates from – the passage in the Ballade Op. 23 when its second theme turns up to bring a midway culmination.

The second theme begins with a pair of two-bar phrases, of which the second makes up a transposed replica of the first. (Ex. 5) Melodically as well as harmonically, both phrases have a closing, dominant-to-tonic quality, and metrically they make up weak-strong configurations. The beginning of first phrase has a closing function in virtue of being introduced at the very end of a formal unit, but it demands to be heard as an elision in terms of shared material, as a closing formulation retroactively understood as the start of a new theme.

But in mm. 106–107 this double function is virtually lost: what we tend to hear is the first, intruding and unequivocally starting, strong-weak phrase of the second theme. (Ex. 6) To an appreciable extent the strong-weak impression is due to the fact that the massive sound of the E-major dominant in m. 106 outweighs that of the A-major tonic in m. 107. The first bar of the theme presents itself as a downbeat at the expense of its second-bar outlet, which by (overridden) harmonic rights should make up the metric accent. It seldom happens that pianists want, or manage to play, these bars as a weak-strong pair. 6

<sup>5</sup> I play the e♭<sup>1</sup> with both index fingers in order to feel that two melodic strands issue from the same note. To prevent it from getting too strong, I put the left index finger on the key slightly after it is has been touched by the right index finger.

<sup>6</sup> This cursory discussion does not take into account the different slurring to be found in various sources/editions.

# **A bass theme and its possible sequel**

More often than not when listening to the F-minor Fantasy Op. 49, I have a feeling that the first quarter-note f/F sounds too short. To find out why, we have to consult the score – and to question our ingrained habit of being unthinkingly faithful to its inscriptions – as well as to study the musical particularities of mm. 1–2.

Since there is no extant autograph, the *Stichvorlage* for the first French edition is as close to Chopin as we can come, and this source has a *staccato* dot at the initial quarter-note and then two three-note slurs starting from the sixteenthnotes. (Ex. 7) Subsequent editions feature *staccato* dots also at the eighth-notes although they are followed by sixteenth-note rests. (These dots were presumably added in order to indicate that the eighth-notes were to be be played in the same way as the initial quarter-note). There are even some editions starting the theme with a two-note slur suggesting that the first falling fourth is on a par with the following ones and should be treated accordingly.

The added *staccato* dots at the eighth-notes emerge as redundant, and yet they might be harmful, possibly misleading pianists into adopting a jerky way of playing the theme, and especially into playing the first note too short. Otherwise put, one might get the wrong idea to adjust the *staccato* of the first note so as to agree with that of the eighth-notes, and not the other way around. (Ex. 7a) And although the falling fourths undeniably correspond to each other, the added initial slur runs against Chopin's intention, demanding detached articulation. (Ex. 7b) But we know that Chopin sometimes changed his mind in such matters.

Leaving issues of authenticity out of account, we must admit that this two-bar introductory bass theme allows of several quite meaningful articulations. (Exs. 7 b/f) An initial slur does not amount to a musical disaster, nor do two-note slurs starting from the sixteenth-notes or from the quarter-notes. Indeed, it is quite possible to play the theme *legato* throughout, and so it is to render it with nothing but detached notes, even quite short ones – imagine that the passage prepares for the entry of the soloist in an otherwise lost Chopin concerto, and that it is to be played *pizzicato* by the double-basses.

Turning to the problem to be solved: what is the appropriate length of the initial f/ F? If a tempo is slow or moderately slow and/or if the *staccato* indication attaches to a note with a long or fairly long note value – conditions satisfied here – it makes manual and musical sense to release the key at a time-point determined by the metre. This seems to be the rationale for the rule of thumb stating that a *staccato* dot halves the duration of the note. Conventions aside, in this particular case we have three options: either the key is to be left at the moment when you can insert an eighth-note rest, or one or three sixteenth-note rest(s). In other words, the duration of the initial *staccato* quarter note might correspond to an eighth note, to a dotted eighth-note, or to a sixteenth-note.

The first, conventional option is preferable – as can be tested by playing the passage according to Ex. 7g. The eighth-note moment of releasing the f/F keys feels just right, and the duration agrees with that of the ensuing eighth-notes, un-dotted but followed by rests. Assuming some reverberation in the room this is likely to make for an effective duration of somewhat less than a dotted eighthnote. 7 Staying for more than an eighth-note at the initial quarter-note will hardly give rise to any perceptible silence at all, while the leave-quickly *staccato* option, so often heard, will disconnect the first sound from what follows. Chopin's dot, then, does not require a pronounced *staccato* effect, it just indicates an eighthnote left in due time.

In addition there is a dual and quite important musical reason for not playing f/F insignificantly short: it is a downbeat and it represents the tonic. If the first event of the work lacks sufficient durational emphasis, the listener might misunderstand the theme, metrically as well as tonally.

To prepare the ground for the presentation of a possible sequel to the bass theme, we must get a fair idea of its tempo. The character of the music, the common-time signature, and the indication *Tempo di marcia* disclose that the fantasy starts with a funeral march. The additional qualification *Grave* to be found in some editions is potentially misleading if it is understood as referring to the quarter-notes of the time-signature; they must not be slower than you can march. But the uneven rhythm of the theme suggests a sense of *alla-breve*, implying that the second and fourth beats carry less metric weight than the first and third, and that it is the half-note pace that should be slow.

More often than not, the 24-bar *Lento sostenuto* section in B major, interrupting the agitated main section of the work, emerges as a *longueur*. (Ex. 8a) This impression seems to be due to an overly slow tempo and to the metric quality within the bars; all beats tend to be played so as to have the same metric weight. Can a reason be advanced in favour of a somewhat raised tempo in this contrasting triple-time theme and of adopting a lighter "*alla-breve*" way of playing it?

Since the interpretational idea to be proposed derives from a perhaps only subjectively valid discovery of a similarity, it might be dismissed as an analytic

<sup>7</sup> The influence of reverberation on articulation is difficult to appreciate; cf. Bengt Edlund, "Distant Listening", Chapter 7 in the present book.

fancy lacking sufficient structural support. The affinity between the initial bass theme and the *Lento* theme is admittedly somewhat farfetched, and yet this association may be defensible if it emerges as productive, if it makes for an integrating reminiscence and leads to a more telling, somewhat faster rendition of the B-major section.

What makes the themes similar? Both of them start with two long notes forming a falling interval followed by a rest; then the melody returns with a shorter note to the initial pitch. The left-hand accompaniment features the same rhythm as the right-hand melody. Turning to the disagreements, the difference as to metre is undeniable, the second event in the *Lento* is longer than it "should" be, and the falling interval is a third, not a fourth. After the first three notes the two themes exhibit various differences but a falling fourth appears in both of them.

If a pianist wants to exploit this association, he/she has to somehow bring out the inherent affinity. The duration of the quarter-notes of the *Lento* can be adjusted so as to agree with the quarter-note pace of the *Tempo-di-marcia* theme (or the other way around), and the second beat in the *Lento* may be played with less weight than the first in order to recall the *alla-breve* quality of the introductory theme. The eighth-note rest must then be strictly observed so as to remind the listener of the strict treatment of the sixteenth-note rests in the bass theme. In addition, the first two chords of the *Lento* theme may be slightly detached from each other so as to recall the articulation of the initial notes in m. 1. (Ex. 8b)

The latter interference may appear controversial since it means that the twonote slur in m. 199, to be seen in many editions, is disrespected. But the original French edition merely features a long overall slur from m. 199 to m. 205, which may be taken to imply that the details of the articulation are left at the pianist's discretion as long as the phrase hangs together. It might be added that the initial mismatch as to articulation in the *Lento* and bass themes might also be resolved in another, less satisfactory way. If you want to bring out the affinity, if you want to integrate the work by suggesting that the *Lento* represents a furtive return of the introduction, the two-note slur in m. 199 may be taken as a pretext for rendering the first falling fourth of the bass theme *legato*, just as the following ones.

This essay on three Chopin themes lacks conclusions but wants a conclusion, hence

#### **The three D1 's**

To play the three D1 's ending the D-minor Prelude in some violent, nonstandard manner is no doubt a post-Pleyel idea. But in our grand times it might nevertheless be defensible – if you have played the prelude very dramatically, consistency demands that you finish it off in an emphatic, brutal way.

But there is a risk that you get some unwanted notes into the bargain, and this applies even if you use your left hand to gently press down the neighbour keys, which is difficult to do properly/silently while plunging down along the keyboard with the right hand.

Why don't we, piano-playing mammals, solve the problem, not by inactivating the surrounding keys, but by pre-selecting the key we want?

So, while your right hand is on its way, put your left-hand thumb at the surface of the D1 key; support the hand by placing the other fingers at the wooden block at the end of the keyboard. Use your right-hand fist for the first D1 , hammering down on your left-hand thumb. For the second and third D1 's, let the key and the thumb return to their initial positions and strike again with the fist.

This is how it looks. If it hurts, you are too loud.

# **Chapter 16 Keyboard commentaries on K. 282**

*"a fundamental human quality: the ability to see with closed eyes, to make colours and forms emerge out of black letters on a white paper"*

*(Italo Calvino, Six Memos for the Next Millennium)*

*"It is to the silence that you should listen. […] What I have written is written between the lines."*

*(Gunnar Ekelöf)*

#### **Introduction**

First of all I owe the readers an explanation of why this text about the first movement of Mozart's Piano Sonata K. 282 was written, and of why my commentaries turned out as they did.

When, after almost seventy years, I returned to this sonata, I was once again simply overwhelmed by the sheer beauty of its first three bars, and in addition to enjoying the music by just playing it, as I did in my youth, I now wanted to understand why its beginning is so magic. Trying to formulate the ineffable is a challenge and a typically grown-up thing to do.

When analysing the main theme of the first movement, I took the position of the interpreter, a perspective coming quite natural since this passage – centuries ago it was picked up and miraculously transformed by a teenager – urges the pianist to discover and take care of its secrets. Over and over again it requires to be improvised into existence, as it perhaps was in 1774.

After completing some four pages of tentative remarks, I sent my musings to a friend who might take an interest in them. And Eugene Narmour did, but he also asked me whether I was aware of the fact that this very movement had been the object of an entire issue of *Music Perception*. Much to my embarrassment I had to confess that I didn't know about these texts, written by seven renowned analysts and music cognition researchers. For an unfortunate period of some years my department ceased to subscribe to *Music Perception*: I never saw vol. 13, 1996.

After more than a year of hesitation I turned back to Mozart's theme and to my sketch, penetrating much deeper into the music, tightening up my analytic argumentation, and extending my observations so as to cover aspects of the entire movement. In order to increase the prospects of arriving at some fresh, independent findings in my belated reflections, I decided to finish my own endeavours before I turned to the reading homework occasioned by professor Narmour's question. Thus, only after completing the account of my observations, did I read the *Music Perception* studies, taking grateful notice of the additional insights brought by the widely diverging approaches of the authors.

The first part of the present essay is about analysis and interpretation. The core issue, then, is to describe what seems to go on musically in the movement and especially in its initial theme, to find things in Mozart's music that demand to be expressed and are possible to convey. But I will have less to say about how one should actually play the music in order to bring these things out. The entire movement is to be found in Ex. 1.

The second part is devoted to the analytic forum in *Music Perception*. Since dissent is more interesting and ultimately also more productive than assent, some critical observations will be advanced. And since the present essay is about interpretation, I will again assume the role of the pianist and focus on issues that stirred up my inclination to contradict. It goes without saying that the value or *raison d´être* of a theoretic investigation or an analysis does not stand or fall on whether it informs interpretation or pleases musicians. But it may be argued that musical applicability may serve as one kind of standard for theories and analytical methods – if they are somehow useful, they are likely to be about something essential.

But first, and as an introduction to the analysis, the auto-biographical track should be pursued.

#### **Youthful mistakes**

After having got the complete Henle edition of Mozart's piano sonatas as a Christmas gift from my father in 1955, I eagerly started playing the sonatas on my own and (sort of) learned some movements that I was particularly attracted to. The first thing I discovered, and had to correct when returning to K. 282, was that my earlier self apparently made a couple of mistakes in the first three bars, misreadings (or rather adjustments) disclosing interesting properties of the theme.

I did not understand that the second beat of m. 2 involves a subtle sense of elision, and felt that the long note made up a *longueur* hampering the flow of the melody. Differently put, I did not understand that making a start out of an ending may require some time. On the other hand, I was responsive to the potential sense of upbeat inherent at the start of the third, strong beat. My sight-reading intuition solved both problems by dislocating the d2 –c2 sixteenth-note motion over e♮<sup>1</sup> so as to form a weak event appended after the second beat, shortened to an eighth-note. As a consequence, the following c2 became accented; how I cleared up the mess, I don't quite remember.

Nor did I understand the exquisitely grown-together harmonies sharing the tied a♭ in the bass. Adopting the preceding B♭-major root-position chord as a temporary tonic, I played a♮ instead of a♭, and so I unwittingly replaced Mozart's sweet sense of longing by the sadness of a shift from F major (turning D-minor-like) to G minor – a never-realized G-minor-like sonority, a discord to be deceptively clarified by the E♭-major first-inversion chord.

The one-beat-too-short theme with uniform upbeats and misread harmonies that I repeated over and over again as a ten-year old probably ended as shown in Ex. 2. When young people play wrong note-values or wrong notes, they sometimes have interesting reasons for doing so; unawares they are solving problems that disturb them. As to myself, I am happy that the peculiarities in Mozart's theme, and the slight pains of discomfort they once gave rise to, are still alive within me and may inform the way I now play the passage. We will return to my misreadings later on.

### **The main theme**

The first three bars obviously make up the main theme of the movement, but they do not add up to any conventional, readily understandable formal structure; the passage fails to exhibit any clear-cut motivic constituents and lacks obvious demarcations. Bars 1–3 neither emerge as a "theme" in current sense, nor as a periodic construct, although two parts, two "phrases", may be distinguished. Perhaps the theme is not even "unified"? (Whatever that means.)

The evasive quality of the initial bars presents a paradox that cannot but increase the value of the music. Along with their exposed formal position, and for all their beauty, these bars also emerge as an introduction to the following section. The next, five-bar constituent with its firm start is more overtly expressive, eventually pathetic, but it turns out that it is likewise devoid of periodic rigour, likewise lacking a sense of being a theme, and eventually it escapes from the tonic. Thus, the initial three-bar section of the movement, its "theme" and emotional point of departure, is disguised as an introduction to a non-theme. Therefore it is left out at the beginning of the recapitulation. (This is a hasty, all-to-square conclusion that must and will be qualified later on.)

At this point we must say something about the movement at large. Its form might be thought of as quasi-binary with a short Coda, and if this description were accepted, it would save us idle discussions of whether the main theme "should" have turned up in m. 22. The principal reasons for settling on a ternary, "sonata form" is that the movement gets a development section, and that the sonata-form category, involving a set of specifications which may or may not be satisfied, makes the music more interesting. The sonata form offers more opportunities to deceive expectations.

The "theme" is metrically unusual since it seemingly comprises only three bars. Analytically, it does "in fact" consist of four bars since the g1 and perhaps also the b♭<sup>1</sup> in m. 4 belong to it. Concurrently the left-hand accompaniment signals that the theme is over, that a new section has intruded on the territory of the initial one. In other words, there is a sense of layered elision in m. 4, and g1 –b♭<sup>1</sup> may be understood and played either as an odd closing formulation or as an odd start – or in some other, non-excluded third way.

The melody of the theme is very much about b♭<sup>1</sup> ; this note recurs again and again, and even after the melody has eventually slid down to g1 – with a trill or rather with a shudder of pure delight – the b♭<sup>1</sup> immediately returns. This final b♭<sup>1</sup> , redundant as it may seem, is enigmatic since it lacks obvious connections, and it makes up a challenge when it comes to interpretation.

That the insistent presence of b♭<sup>1</sup> is a crucial characteristic of the theme can be demonstrated negatively by simply taking away the second-beat b♭<sup>1</sup> in m. 4 or by playing a second g1 in its place – or positively by exchanging the trill on a♭<sup>1</sup> for an eighth-note c2 followed by one or two b♭<sup>1</sup> 's. (Ex. 3) Apparently, the theme does not "want" to leave its anchor note. One can learn much about subtle music by subjecting it to Ditters-von-Dittersdorfian changes.

As we will eventually see, the theme is arguably also about e♭<sup>2</sup> .

#### **The first phrase**

What happens during the initial five/six beats? Well, the outer voices release themselves from the narrow compass b♭<sup>1</sup> -over-e♭<sup>1</sup> and reach e♭<sup>2</sup> -over-b♭. Not very much, one might think, but it feels as if the world opens up. I cannot fully explain the sense of expansion involved, but it may to some extent be due to the fact that the left hand moves ahead of the right and seems to induce it to move beyond c2 . Ditters might have opted for still another, trivializing left-hand e♭<sup>1</sup> /g1 in the middle of the bar, or for a d1 /f 1 already on the second beat (Ex. 4), whereas Amadeus's third-beat d1 /f 1 offers the pianist the privilege of disclosing a prospective potential in the melody. As to the left hand, what the pianist has to do is to feel a furtive upbeat quality in the second-beat e♭<sup>1</sup> /g1 and to play it accordingly – the message to the listener being that the tonic prepares to leave, although it is still merely confirmed.

Meanwhile the anchor note b♭<sup>1</sup> prepares for its excursion up to e♭<sup>2</sup> . The effect of the short visit to the neighbouring c2 is that b♭<sup>1</sup> gets more insistent, becomes charged with an intent to break the initial equilibrium. But the initiative is taken by the left-hand third d1 /f 1 , urged by its virtual e♭<sup>1</sup> /g1 upbeat. Only then the last eighth-note b♭<sup>1</sup> and its pianist get enough confidence to follow the articulation slur showing the way up to e♭<sup>2</sup> , a note that would have been extremely dissonant, had not the left-hand proceeded to c1 /g1 . In order to feel the graceful effort needed for producing the rising fourth, you can imagine a quick, mediating c2 – d2 motion, an idea that may be realized when playing the repeat. (Ex. 5) But d1 /f 1 does not have to issue into the C-minor fifth c1 /g1 ; von Dittersdorf might have returned to e♭<sup>1</sup> /g1 , annihilating the prospective meaning of the second e♭<sup>2</sup> , robbing it of its aspiration to eventually attain full, downbeat stability as an upper tonic note. (Ex. 6)

But how can we know (or rather suspect) that the upper tonic note is the goal that the first phrase strives for? A retrospective answer is that the target is disclosed by the second phrase, in which the melody starts to rise again, eventually reaching an off-beat e♭<sup>2</sup> -over-g in m. 3, another non-satisfactory arrival.

For the first attempt at e♭<sup>2</sup> apparently failed as well; hence, there is also an immediate answer within the first phrase. Already when d1 /f 1 is followed by c1 / g1 , we can guess that b♭ will follow, and when a♮<sup>1</sup> turns up, the diverging voice leading is a fact, making b♭<sup>1</sup> strongly expected. The applied dominant transiently suggested by the middle-voice a♮<sup>1</sup> cannot but issue into a thin, six-without-four chord. The soprano melody is the last one to know. The e♭<sup>2</sup> before the bar-line is still full of confidence, and it is still an anticipation, but (as it will turn out) not of the eighth-degree tonic, but of the fourth-degree over the dominant. And the six-non-four sonority crowning the phrase lasts for just a flickering moment before d2 starts the quick retreat from the peak.

There is a sting of disappointment and sweet melancholy in the long preparation for the expansion upwards versus the swift descent, in the understatement of the dissonant note of rhetoric arrival and its quick, ignominious resolution; after all, the world did not open up.

Dittersdorfian variants of the first phrase run as shown in Ex. 7.

Mozart makes the pianist co-responsible for how this phrase grows and recedes. All phrases demand one thing or the other from the musician, but the beginning of this sonata is demanding to a breathtaking degree.

#### **A sense of elision?**

It has already been suggested that the first phrase lasts five or six beats. This indeterminacy is due to the fact that the second beat of m. 2 involves a sense of elision, a complication that I was blissfully ignorant of in 1955. Since this is a crucial and subtle feature of the theme that cannot but have consequences for its interpretation, a thorough discussion is needed.

It appears that there are three alternatives. The second-beat b♭<sup>1</sup> may belong to first phrase, thus postponing the start of the second phrase until the third beat. The two phrases may also be thought of as sharing the second-beat b♭<sup>1</sup> , which implies that this long note has both closing and starting functions. And the first phrase may be taken to close already at the inconspicuous sixteenth-note b♭<sup>1</sup> , in which case the second phrase starts immediately at the quarter-note b♭<sup>1</sup> .

The middle-voice e♮<sup>1</sup> , an accented passing-note introducing a dissonant sonority, does not seem compatible with a third-beat start of the second melodic phrase. On the other hand, and owing to that very dissonance, the accented melodic particle d2 –c2 on the third beat does make up an interesting and potentially quite rewarding flying start for a new phrase. But the fact that the bass brings a half-note and a root-position dominant on the second beat supports the idea that this is where the new phrase should begin. Again on the other hand, the second-beat third b♭/d1 in m. 2 has a furtive upbeat potential similar to that of the second-beat e♭<sup>1</sup> /g1 in m. 1; hence, the closing quality of the quarter-note b♭<sup>1</sup> may emerge as undermined.

The quarter-note b♭<sup>1</sup> -over-b♭ evidently represents the dominant, and preceded by a short, anticipating b♭<sup>1</sup> it emerges as a stable note – although one might feel that it turns up somewhat too precipitately to close the first phrase. But this second-beat b♭<sup>1</sup> also affords a stable point of departure for the second phrase, which eventually grows into varied version of the b♭<sup>1</sup> –e♭<sup>2</sup> –b♭<sup>1</sup> essence of the first phrase. These observations support the possibility that the second-beat b♭<sup>1</sup> starts the second phrase.

Opening up for yet another perspective, the mid-bar d2 –c2 motif may also be understood as a curtailed imitation of the similar first-beat particle e♭<sup>2</sup> –d2 . (Play b♭<sup>1</sup> –a♭<sup>1</sup> –a♭<sup>1</sup> after d2 –c2 as shown in Ex. 8.) Indeed, taking the second beat in m. 2 to be a closing/starting, subordinate connecting point, there is perhaps but one long, four-bar melodic arch taking us from b♭<sup>1</sup> to e♭<sup>2</sup> and eventually to g1 in m. 4. Playing the second beat in m. 2 so as to suggest tightly overlapping phrases requires that you render the quarter-note b♭<sup>1</sup> as a shared note; it should somehow have a non-demarcating quality. Or perhaps it can be virtually re-functioned from a closing note to a starting one while it sounds?

Closing the first phrase at an anticipated, second-beat b♭<sup>1</sup> makes for a stress that may seem too blunt, that may disperse the melancholy of the quick retreat from the non-satisfactory, dissonant top-note e♭<sup>2</sup> . It makes for a fine effect to let the quickly falling notes, and the first phrase, breathe out with a sensual sigh barely reaching the sixteenth-note b♭<sup>1</sup> , leaving the quarter-note b♭<sup>1</sup> to the second phrase. Yet, since there are structural factors making it "natural" that the first phrase closes on the second beat rather than before it, we must ask whether it is feasible to convey that a new phrase begins at the quarter-note. What can the pianist do? One way of tipping the balance is to bring out a quasi-upbeat b♭/d1 third in the left hand at the expense of the closing quality of the right-hand b♭<sup>1</sup> . Another option is to suppress the melodic anticipation: don't play the two postappoggiatura sixteenth-notes c2 –b♭<sup>1</sup> as another appoggiatura since this would make the following consonant b♭<sup>1</sup> seem emphasized – i.e. play them lightly and *non-legato*.

# **The second phrase; metric peculiarities and virtual delays**

The following discussion presupposes that, no matter the *Adagio* prescription and the common-time signature, the theme has an *alla-breve* touch. The reason is simply that, at least when it comes to the theme, it is more productive to deal with metrically differentiated beats. Furthermore, thinking of the theme in *alla breve* terms is a way of achieving an *Adagio* tempo without having to play very slow. 1

Due to the passing-note quality of the left-hand e♮<sup>1</sup> the nominally strong, applieddominant third beat in m. 2 invites to be understood as a weak beat, which implies that the second beat will retroactively assume the quality of being accented, no matter its nominally weak metric position. This makes for a problem if the second beat is meant to finish the first phrase; in such a case this beat should preferably emerge as weak. The re-functioning of the third beat also implies that the nominally weak fourth beat will either (in accordance with its transient,

<sup>1</sup> As will be discussed later on the tempo of this movement seems to change as the music proceeds.

auxiliary-tonic F-minor appearance) take on an accented quality, or emerge as the second of two consecutive weak beats, suggesting an ongoing quality.

Although the (nominally) strong third beat makes up an applied-dominant seventh-chord with passing qualities, and although the three-note melodic gesture that goes with it involves an upbeat, we cannot exclude the possibility that it brings the accented start of the second phrase. Perhaps Mozart plays a wildcard here, making a move that retroactively does away with both the sense of elision on the second beat and the metric perturbation just accounted for.

As to von Dittersdorf, he might have disenchanted the whole situation by simply repeating the left-hand b♭/d1 so as to provide a regular, middle-of-the-bar dominant start of the second phrase. (Ex. 9) Turning to myself, in 1955 I evaded the problem when relocating the third beat by reducing the length of the second beat to just an eighth-note. Laudably and yet most reprehensibly, I achieved an ongoing quality by refusing to choose when unwittingly being faced with the contradictory qualities of the third beat.

Presumably and broadening the perspective, I might also have felt a need to adjust the time-point of the nominally accented right-hand d2 –c2 entry, having the quality of an upbeat, so as to fit in with the following, true upbeat entries after the fourth beat in m. 2 and after the second beat in m. 3. Here the lefthand double-stops are introduced during the quarter-notes in the melody; in comparison, the "empty", second-beat quarter-note b♭<sup>1</sup> in m. 2 might have seemed too long, too closing. Leaving aside my mistake, statute-barred long ago, the sense of a delay inherent in the closing and/or starting b♭<sup>1</sup> contributes immensely to the peculiar appeal of the theme.

As already confessed, I am happy that I can still appreciate the disturbance felt in 1955: the "too-long" second beat remains a challenge. If I had played the violin, and assuming that a shared-note elision is involved, I might have let the tone increase in intensity so as to show that the closing afterbeat b♭<sup>1</sup> is gradually charged with the prospect that something new will grow out of it. At the piano such a shift of function must take place in the player's imagination, from where it may perhaps perceptibly influence the execution of the preceding and following events as well as (somehow) the crucial note itself.

But, as already mentioned, it also possible to render the second-beat B♭-major third in the left hand as a virtual starting upbeat, and as a consequence the e♮<sup>1</sup> , a dissonant passing-note, will seem to be located to a strong beat as the notation bids. Alternatively, b♭/d1 can be played as an accented event, turning the third beat into a weak event, no matter its nominally strong position. Finally, you can disregard the sense of elision altogether and force the third beat to start the second phrase of the theme – a weak third beat, preferably, since this is what the passing left-hand e♮<sup>1</sup> "wants". If the third beat is played as accented, the two-unit left-hand harmonic progressions (if any, see below) will move "uphill"; i.e. with the C-major and B♭-major third-position dominant seventh-chords located to strong positions, a perhaps less likely, but quite expressive option.

There is another note in the theme that also seemed, and still seems, "too long" in a wonderful way: the left-hand e♮<sup>1</sup> . Being just a mediating leading-note, it "should" merely be an eighth-note, and since the middle-voice d1 –e♮<sup>1</sup> –f 1 quarternote motion invites to being understood as a slow replica of the g1 –a♮<sup>1</sup> –b♭<sup>1</sup> motion in m. 1 (cf. Ex. 1), it asks for being hastened.

It might have been my shortened e♮<sup>1</sup> that in 1955 enticed me into playing a wrong F-major first-inversion chord on the fourth beat. Mozart's a♭/f 1 sixth in the left hand is not only correct but also better since it introduces still another event that seems to be lengthened. The transient, or just apparent, F-minor firstinversion chord is gradually and furtively transformed into a B♭-major thirdinversion dominant seventh-chord, but due to the tied bass-note the net effect is that of an extended harmonic complex. In retrospect, a two-quarter-noteslong, weak-beat B♭-major dominant-seventh chord seems to be hanging over the bar-line. Allow von Dittersdorf to destroy this exquisite detail by bringing the resolving E♭-major left-hand sixth already at the start of m. 3. (Ex. 10)

According to this extended-dominant reading the two "uphill" harmonic progressions from strong (applied) dominants to weak (auxiliary) tonics turn into two overlapping three-unit progressions supported by the slow, quasisyncopated bass motion b♭–a♭–g underlying much of the second phrase. To grasp this harmonic pattern – suggesting that there is a prolonged dominant comprising four beats – a middle-voice imitation, a syncopated c1 –d1 motion can be played instead of the d1 at the first beat of m. 3, a variant that may be used when repeating the theme. (Ex. 11)

The impression of a delay (or rather harmonic standstill) in the left hand coincides with the melodic peak at f2 and the subsequent falling motion, introducing an extension from three to five notes and hence suggesting an additional delay effect in the right hand. (The distance between the syncopated quarter-note c2 in m. 2 and its metric companion b♭<sup>1</sup> in m. 3 is longer than that between the closing/starting b♭<sup>1</sup> and the syncopated c2 in m. 2, and also longer than that between the syncopated b♭<sup>1</sup> and its non-syncopated follow-up b♭<sup>1</sup> in m. 3.) As a result, the melody of the second phrase emerges as perceptibly stretched.

Throughout the second phrase there is a persisting vagueness in the rhythmic domain due to the syncopated quarter-notes in the melody, to the metrically non-identical sixteenth-note entries in the right hand, to the possible elision on the second beat of m. 2 (being perhaps both weak and strong), to the sense of an extended dominant complex, and to the fact that no unequivocal strong beat turns up until the third beat in m. 3; perhaps there are as many as four or even five perceptually weak beats in succession. Bars 2–3 abound with rhythmic subtleties: the irregularly introduced melodic motifs are of different length, the underlying right- and left-hand descending strands are out of phase with each other, and the two-unit authentic harmonic progressions run "uphill" in opposition to the nominal accents. These peculiarities cause virtual delays and dislocations within the metric hierarchy, as well as a loosened rhythmic grouping, qualities that in turn may foster interpretational ideas. Some of these effects can be captured by means of the current symbols for rhythmic analysis. Leaving out some of the ambiguities, the passage might turn out as shown in Ex. 12.

Due to the rapid descent from e♭<sup>2</sup> in m. 2 and the anticipation of b♭<sup>1</sup> , and due to the dominant third b♭/d1 in the left hand, the second beat is likely to emerge as a too-early strong beat, intruding on the territory of the preceding main accent. Since the left-hand e♮<sup>1</sup> is a dissonant passing-note making for an applied third-inversion C-major dominant, the nominally accented third beat may take on a "down-hill" weak quality. Its metrically dislocated resolution, the apparent first-inversion F-minor chord on the fourth beat, does not bring much of an accent, nor does the third-inversion B♭-major dominant starting m. 3, a chord that rather seems to prolong or re-function the would-be F-minor sonority introduced before the bar-line. An accented tonic chord should follow on the second beat, but the first-inversion E♭-major chord is insufficient. Only the dominant suspension occurring on the third beat brings an unambiguous sense of accent, the first solid strong beat since the second (or first) beat of m. 2.

But if the melody rather than the harmony is allowed to determine the metre, the picture changes. No matter whether the quarter-note b♭<sup>1</sup> closes the first phrase on a weak beat or starts the second phrase on a displaced, too-early strong beat, we are likely to hear a slow sequence of strong and weak events, respectively.

#### **The second phrase; linear connections**

The pitch contour of the second phrase will turn out differently depending on whether you begin the second part of the theme in the eliding way, i.e. with a (perhaps shared) starting b♭<sup>1</sup> on a seemingly strong second beat, or with the d2 –c2 impulse on the third beat, understood as weak rather than strong. In the first case, the final long notes of the motifs are likely to attract your attention, producing a slowly descending line, c2 –b♭<sup>1</sup> –a♭<sup>1</sup> . (The last note is delayed, but the third beat of m. 3 can be Dittersdorfed, i.e. normalized and contracted, as shown in Ex. 13.) In the second case the short notes starting each melodic gesture will come to the fore, and a rising contour will emerge: d2 –f 2 –e♭<sup>2</sup> ; cf. Ex. 14, showing both readings.

According to the second reading, the theme once again attains e♭<sup>2</sup> – this note is withheld, and hence perceptibly strived for, by being avoided on the way upwards. The sense of a second culmination is eclipsed by the fact that the target note has been overshot by f2 as well as by the fact that the e♭<sup>2</sup> merely appears as a short note in upbeat position over a first-inversion tonic chord. The upper line is precipitately closed by two b♭<sup>1</sup> 's, recalling the b♭<sup>1</sup> –b♭<sup>1</sup> anticipation in m 2. Considering the likewise precarious nature of the arrival of the first phrase, i.e. the short, appoggiatura six-non-four e♭<sup>2</sup> starting m. 2, the second phrase emerges as an extended but less confident, convoluted replica of its predecessor.

But taking account of the fact that a lower-register falling tendency is also present in the melody of the second phrase, it might be questioned whether the point of mm. 2–3 is to reach the e♭<sup>2</sup> summit once more. To find out, play an imitating e♭<sup>2</sup> –c2 –b♭<sup>1</sup> –a♭<sup>1</sup> –a♭<sup>1</sup> five-note motion instead of the putting-off three-note e2 –b1 –b1 gesture in m. 3, a regularizing but perhaps not altogether Dittersdorfian substitution lending more emphasis to the lower descending line heading for g1 , and that may be used when varying the repeat. (Ex. 15)

Excepting the rising fourth b♭<sup>1</sup> –e♭<sup>2</sup> in m. 1, all melodic gestures in the theme, as well as the overall tendency in the second phrase, are falling. If there is any general rule when it comes to interpretation, it amounts to counteracting (or understating) what is too obvious, what the listeners will understand anyway. There is a sense of sadness in the theme that may partly derive from the fact that virtually every motion is pulled downwards. To render this even more moving – and to prepare for the next section turning more and more pathetic – the pianist is called upon to make the most of the rising melodic aspirations, of the two eventually futile attempts to defy gravity by striving for a stable e♭<sup>2</sup> . It might, as already mentioned, help to imagine two quick, mediating sixteenth-notes c2 –d2 in m. 1 (cf. Ex. 5), and in mm. 2–3 you may bring out the top notes d2 , f2 , and e♭<sup>2</sup> , suggestive of a thin strand opposing the falling order of things.

If you pay attention to the middle voice, there are two rising motions that can be used to counter-balance the overall falling tendency of the second phrase. As already pointed out, the eighth-note motion g1 –a♮<sup>1</sup> –b♭<sup>1</sup> in m. 1 may seem to be imitated first in quarter-notes, d1 –e♮<sup>1</sup> –f 1 , then (if you choose to play so in the repeat) with a syncopated rise, c1 –d1 –e♭<sup>1</sup> as shown in Ex. 11.

Interpretation also feeds from encompassing, more or less sub-surface connections that may serve to support an otherwise perhaps precarious sense of unity.

It is an obvious aspect of mm. 1–4 that the bass proceeds stepwise from e♭<sup>1</sup> down to e♭, a long-term motion that is clearly divided into two stages: the first-beat dominant chord in m. 2 ends a descending fourth, the second-beat dominant starts a descending fifth, respectively. In addition mm. 2–3 bring a parallel c2 –b♭<sup>1</sup> –a♭<sup>1</sup> descending connection to g1 , and the melody of mm. 1–4 is all about b♭<sup>1</sup> , an observation that due to the "added" final b♭<sup>1</sup> is compatible with the fall to g1 in m. 4. (Ex. 16) One way of unifying the theme might be to gently bring out motifs involving the notes c2 and b♭<sup>1</sup> so as to make the recurring pitch b♭<sup>1</sup> conspicuous throughout the theme.

As already pointed out, the second phrase seems to replicate the melodic contour of the first phrase. Less obvious is an encompassing similarity that perhaps qualifies as a Schenkerian "hidden repetition". 2 The asymmetric risingthen-falling fourth motion b♭<sup>1</sup> –e♭<sup>2</sup> –b♭<sup>1</sup> straddling the first bar-line may seem to turn up at a grander scale in mm. 1–4. Since the crucial notes appear at accented positions, this expanded sub-surface recurrence may be expressible and serve to unify the two parts of the theme. (Ex. 17) Perhaps this observation even explains the seemingly "surplus" b♭<sup>1</sup> in m. 4?

It has been proposed that what the two phrases within the theme may have in common is a frustrated aspiration to attain a stable e♭<sup>2</sup> . Perhaps there is a third attempt, even more disguised than the second one in mm. 2–3, an attempt that makes for an association between the theme and the following, contrasting section? Starting from the "surplus" b♭<sup>1</sup> , a reminiscence of the essence of the first phrase can be traced: two preparatory e♭<sup>2</sup> 's to be played *forte* are followed by a prolonged *piano* appoggiatura e♭<sup>2</sup> –d2 . (Ex. 18)

#### **The Coda – the main theme revisited**

Although it does not recur to begin the recapitulation, the theme does reappear in the movement; cf. Ex. 1.

<sup>2</sup> Cf. the critical discussion of this concept in Bengt Edlund, "Hidden repetitions and uncovered parallels", ch. 4 in *Analytical Variations*, Frankfurt 2020 Peter Lang Verlag

Acute ears will discover and appreciate that the development starts with a rhythmically diminished allusion to the first three notes of the exposition, but this swift reference is accompanied by a diminished-seventh sonority. 3 As a result the motif does not just issue into a skip upwards to e♭<sup>2</sup> ; it is now charged enough to make a leap up to b♭<sup>2</sup> . (For another, possible reminiscence in the development of the main theme, see below)

Mozart must have realized how excellent the beginning of his sonata was – why else did he compose a Coda, starting with this very idea? But there might be a further reason for the fact that the theme turns up to close the movement: in mm. 1–3 the aspirations to reach and stabilize the upper tonic note failed. In other words, there is an unsettled business in the theme.

In m. 34 the rising fourth is fleshed out with a sixteenth-note passage, releasing the potential for growth and incorporating the preparatory e♭<sup>2</sup> 's. But the crowning e♭<sup>2</sup> , again over b♭/b♭<sup>1</sup> , is both brought out and hidden away. It is rendered prominent by being the resolution of a falling appoggiatura, but concurrently it is put out of the rhythmic focus, and it immediately turns into a suspension demanding d2 for resolution. Engaged in a dragging asynchrony with the accompaniment, the melody then returns to b♭<sup>1</sup> .

And just as in the initial main theme, there is a second attempt to reach e♭<sup>2</sup> . This time the upper tonic note is quickly reached without any effort, and it is supported by a root-position tonic third in the left hand. But the tonal goal of the theme does not seem to be important any more; it is merely a transient detail in a sequence of swift, descending triplets suggesting the subdominant on the route to the six-four e♭<sup>1</sup> of the final cadence.

#### **Formal variety and matters of transition**

The 36-bar movement may be described as a collage of abruptly juxtaposed episodes. The serious atmosphere in mm. 1–8 suddenly gives way for easy-going warbling in mm. 9–12, and the exposition ends with three bars of brilliant-style closing formulations. The main theme has little to do with mm. 4–8, nor is the substance of mm. 9–10 very closely related to the one in mm. 11–12. Bar 8, seemingly bringing in new material, may sound as a fairly odd ending of the second section, but, starting from the upbeat b♭<sup>1</sup> , it perhaps hides an elaborate, checking variant of the preceding motif. (Ex. 19)

<sup>3</sup> A similar, even more shocking effect is to be found at the start of the development in the second movement of the Sonata K. 333.

If persistently subjected to a Schenkerian gaze, the movement cannot but somehow exhibit a unifying *Ursatz*; otherwise it is far from unified. But Mozart gets away with it, of course, and on top of it he seems to have made a point of avoiding or disguising episodes that make up multiples of two bars.

The movement's sonata form is irregular – the main theme does not recur to head the recapitulation – and it is also ambiguous since it cannot be positively established where the recapitulation begins. The trill and the transition passage in m. 21 do signal that the development is completed after just five bars, but this is not unequivocally confirmed by m. 22, whose first half is identical with the first half of m. 4. For a short moment we may think that we are listening to the second episode of a truncated recapitulation, but soon the music deviates from its expected course in a way that makes us suspect that the development was in fact not finished. Later on, when m. 26 presents an exact, transposed replica of m. 9, we are likely to change our minds once again: the five bars just heard must after all have been the modulating second episode of the recapitulation.

This state of affairs opens up for two options when it comes to interpretation. If you bring out the sense of a composed-out fermata functioning as a transition in m. 21, and then begin m. 22 just as you played m. 4, the listener will hear the start of the second episode of a shortened recapitulation. If, on the other hand, you play or improvise in a way suggesting a sense of continuity between the two bars, an extended development will come to the fore, which later on (perhaps only in m. 26) turns out to be a recapitulation in progress. (Ex. 20)

When playing the second repeat, which of these formal configurations is to be played the first time? Being most straightforward, the demarcating, truncateddevelopment option should preferably be suggested the first time. The connecting alternative, making for a longer development, may be more appropriate, more effective the second time since only then will it emerge as a deviation.

When proceeding to the development from the exposition, the right-hand transition may, somewhat against the grain of the rising motion, be played *diminuendo* so as to attach to the *piano* start of the development. This will also make for a contrast to the *crescendo* that seems appropriate when returning to the main theme, which may be played louder the second time – a four-note trill on the initial b♭<sup>1</sup> can be added to give more emphasis to the entry of the theme. (Ex. 21) But when starting the development again after m. 33, it seems to be a good idea to render m. 16 in a more overtly passionate way – this applies especially if you are going to connect m. 21 with m. 22 – and if this more dramatic interpretation is chosen, the descending transition in m. 33 should be played *crescendo*.

Turning to the start of the Coda, the self-evident option is to close the circle by starting m. 34 as an exact replica of m. 1, which means that the falling transition now has to be rendered *diminuendo*. But it is also possible to argue that the beginning of the Coda should not sound as the start of the exposition. The listeners (we leave out of account those who know this sonata inside out) and the pianist do not deal with the same music. Whereas the listeners do not know what will happen after the second presentation of m. 33, the pianist knows, and being in command of the future he/she is responsible for how it happens.

Thus, one may begin m. 34 exactly as m. 1 was started and wait for the event where the difference between the two bars becomes manifest, i.e. the second, syncopated b♭<sup>1</sup> of the second beat, which can be charged in a way that brings about the following melodic expansion. But from the pianist's synoptic vantage point m. 34 may also from its very beginning be rendered as being qualitatively different from its model in m. 1, as being pregnant with its forthcoming expansion. Perhaps a slight *crescendo* is due when playing the first notes of the Coda?

#### **Inherent tempo shifts**

The tempo indication reads *Adagio*, which is unusual for a first movement in sonata form. In the previous discussion of the main theme, a slow tempo was taken for granted, but not a tempo being slow to the point of ironing out the metric difference between the first/third and the second/fourth beats: the second beat should not be on a par with the first beat in the accentual hierarchy. Thus, the tempo should be moderately slow, or as slow as it becomes if you think of the movement as an *Adagio* with a touch of *alla breve*.

There is no reason to change the speed when playing the rest of the movement. But this does not imply that the same tempo necessarily prevails throughout the piece. Tempo is not primarily a matter of speed, but of perceived pulse density, which in turn is determined by the rhythmic properties of the musical substance as well as influenced by how the music is performed.

As a pianist you can adopt two quite different approaches to this movement. In order to promote unity, you may choose to maintain a sense of constant tempo by subduing any shifts of accentual density that seem to be suggested by the various musical ideas that turn up. On the other hand, if you are interested in furthering a sense of variety, you may give in to the shifting pulse rates that appear to be inherent in the music.

Considering the fact that this movement is a collage-like assembly of quite disparate episodes, and that this contributes substantially to the charm of the music, it seems that the second approach might be quite rewarding. 4 It would be a pity not to bring out the various characters of the episodes, to abstain from letting their inherent rhythmic properties set your tempo, and presumably also the tempo felt by the listeners. If the movement is analysed with respect to possibly occurring shifts in accentual density, it appears that the music suggests no less than three, or even four, different pulse rates, and hence three/four widely different tempos; cf. Ex. 1.

The *alla-breve* pace of the initial *Molto adagio* tempo lasts until the "commontime" accompaniment in m. 4 brings in the *Adagio* four-beats-per-bar pulse. The dense, checking sequence of harmonic events in m. 8 closes the second episode, and the pathetic mood of this bar is underscored if the music is played so as to suggest accents on each eighth-note. This accentuation, introducing a *Moderato* tempo, is then confirmed by the right-hand articulation of the first idea of the third episode, and an eighth-note pulse also suits the complementary idea, starting on the second beat of m. 11. Concurrently, a sixteenth-note pulse corresponding to a brisk *Allegro* tempo seems to be appropriate for the lively 32nd-note motifs in mm. 11–12 as well as for mm. 13–14 of the closing episode. The sequence of four sixteenth-notes in m. 15 may then be used to hold back the fast pulse rate, leaving it to the following right-hand transition to restore the initial, *Molto adagio* tempo.

The rocking idea in mm. 11–12 is likely to attract the interest of both the pianist and his/her listeners since it betrays a funny similarity with the behaviour of physical objects – imagine a coin, thrown on a table so as to produce an accelerating series of sounds. Repeated shifts between *Moderato* and *Allegro* pulse rates may help to make this association vivid, may help the listeners to appreciate the similarity between the sixteenth-note motion and its quicker follow-up.

If a sequence of increasing pulse rates/tempos is allowed to imprint the sequence of decreasing note-values within the development, it will get a strong momentum as well as a strong sense of coherence: two *Adagio* bars – not *Molto adagio* since this would be detrimental to the syncopated rhythms – followed by two bars of *Moderato*, and one *Allegro* bar.

<sup>4</sup> Pieces consisting of disparate episodes suggesting different tempos are not rare in 18th Century music. The first movement of Haydn's C#-minor Sonata Hob. XVI:36 is but one example; another one is Mozart's Sonata K. 332 with its peculiar tempo contrast already between mm. 1–4 and 5–9 in the main theme of the first movement.

# **Ornamentation**

The magic of the main theme slowly evaporates as the music proceeds, but the witty and clever composing affords a highly rewarding listening experience as well as opportunities for pianists who want to adopt a habit of Mozart's days, namely to vary the repeats. But before discussing improvisation and other kinds of possible alterations, it should be stressed that for present-day ears there is nothing compulsory about varying repeats. There is an aesthetic choice involved in the decision to introduce changes in repeats (or other recurring passages) or to abstain from it, and presenting new perspectives is not intrinsically better than offering confirmation.

As to melodic improvisation in current sense, you might hesitate to interfere with the delicate state of things in Mozart's score. But if you want to embellish the melody, moderation and an unfailing sense of style are required. 5

In addition to the suggestions already offered in Exs. 5, 11, 15, 20, and 21 only one further embellishment, having structural implications, will be proposed. The melodic endpoint at g1 in mm. 4 can be avoided in favour of a connection to the following b♭<sup>1</sup> if the bar is started with the appoggiatura c2 –b♭<sup>1</sup> or, if you want to do something more, with the figuration a♭<sup>1</sup> –g1 –c2 –b♭<sup>1</sup> –b♭<sup>1</sup> . The preceding trill on a♭<sup>1</sup> must of course be taken away. (Ex. 22)

Leaving matters of improvisation in current, melodic sense, there is another aspect of the movement that offers some scope for interpretational variety. It seems that some of Mozart's dynamic marks in mm. 4–8 and 22–26 are negotiable proposals for expression rather than defining traits of the musical structure. In the interest of variety, the *forte* indications in mm. 24 and 25 may, for instance, be replaced by *piano* marks, and the other way around, when playing the repeat.

In mm. 4 and 5 the left-hand *forte* marks precede those in the right hand, which seems redundant since the left hand does not bring any change; one can do quite well without the left-hand *fortes*, leaving the dynamic contrasts to the right hand. Indeed, the left-hand *forte* entries in mm. 4 and 5 may be subdued in order to bring the corresponding ones in mm. 22 and 23 into relief. The former do not herald anything very remarkable whereas the latter bring important harmonic shifts and signal formally crucial deviations from the "normal" course of events heard in the exposition. When repeating the second part of the movement,

<sup>5</sup> The recording by Friedrich Gulda is highly recommended for its sparse and exquisite ornaments/alterations as well as for the casual way in which he slips them in.

the *piano*/*forte* contrasts might be skipped altogether in favour of a *crescendo* pursued all the way to the cadence in m. 26.

The *piano* indications in the second part of m. 6 opens up for a subtle difference. If you play *pianissimo*, you will introduce a minor-tonic shadow; if you disregard the *piano* marks and play *forte*, the listeners will rather hear a minor subdominant holding out the prospect of the F-major applied dominant, eventually turning up in m. 8, whether hushed or not. Irrespective of these options, there is a chromatic bass descent in mm. 6–7 that deserves to be brought out.

# **Matters of performance**

The subject of Carol L. Krumhansl's paper appears from its title. 6 In the present context the most interesting thing is her account of an additional experiment, aiming at the relationship between performance and perceived tension.

A performance by a professional pianist (Philippe Entremont) was contrasted with three "performances" derived from it: one with same-level dynamics, one with constant tempo (speed), and one with same-level dynamics as well as constant tempo. Krumhansl's conclusion runs: "the manipulation of performed tempo and dynamics had remarkably little effect on the tension judgments". (p. 426) In other words, most of the information needed to *identify* tension is encoded already in the musical substance. This outcome may at first seem to be bad news for those trying to play music in an expressive way; but how a musician plays might still be important when it comes to actually *conveying* tension (and musical content in general).

Caroline Palmer has studied the performance characteristics of the first movement of K. 282 as played by Philippe Entremont. 7 The recording was made on a Bösendorfer SE grand piano, an instrument equipped so as to register exact data on timing (when the keys are depressed and released), dynamics (i.e. hammer velocity), and the use of the right and left pedal. Palmer is quite aware of the fact that the results of her case study cannot be generalized in any simple manner; other pianists are likely to play differently in a number of respects. Furthermore, it should be added, it is also in the nature of things that her findings cannot be interpreted with any certainty – Entremont's intentions

<sup>6</sup> "A Perceptual Analysis of Mozart's Piano Sonata K. 282: Segmentation, Tension, and Musical Ideas", *Music Perception* 13(1996), 401–432

<sup>7</sup> "Anatomy of a Performance: Sources of Musical Expression", *Music Perception* 13(1996), 433–453

are not known, and all Palmer (or anybody else) can do is to advance reasonable, systematic explanations or propose plausible musical motivations. A number of observations from similar earlier studies are corroborated by Palmer's investigation, and what follows is therefore only brief remarks on some details of particular interest in the present context.

Onset asynchrony – the melody being slightly ahead of the accompaniment – is a frequent trait in Entremont's recording. But in m. 7 the bass note e♮ is played before the top note e♭<sup>2</sup> . The non-simultaneity was arguably a means to reduce [and concurrently expose] the sharp dissonance. (p. 441)

As to the moment when the keys were released, Palmer presents a close-up study of the initial five beats. (p. 438) The right pedal is depressed just before the first chord [which produces a richer sound], and the keys of the initial lefthand third are quickly left. [Perhaps Entremont conceived of the second third, held much longer, as an internal upbeat.] The eighth-notes of the fourth beat are played *non- legato*, particularly in the left hand, but lagging pedal shifts blend the sounds. "Hyper"-*legato* playing – i.e. not releasing a key until after the next one has been struck – can be observed already in connection with the first sixteenthnote, and it turns quite extreme at the very beginning of m. 2, where e♭<sup>2</sup> is released at virtually the same moment as d2 . [One may assume that Entremont considered the dissonant note to be important, perhaps even structurally crucial.] This appoggiatura as well as the following c2 –b♭<sup>1</sup> one, played *legato*, are kept together by means of the pedal.

The timing profile of mm. 7–8 (p. 440) may reveal an interesting pattern. Although there are no slurs, the eighth-notes c2 and b♭<sup>1</sup> arguably have upbeat function, and Entremont plays them accordingly. Palmer does not report how long the keys are depressed, but these notes (marked by wedges in the score) involve conspicuously long inter-note durations.

The last, anticipating sixteenth-note d♭<sup>2</sup> in m. 9 is also quite prolonged although it is not (officially, as it were) an upbeat. Since this long inter-note duration does not belong to any final retard, Palmer explains it as a means "to draw the listeners' attention to the upcoming cadence". (p. 442) [Moreover, if you have played the first part of m. 9 *crescendo*, some time is needed to let the sound dissipate. But there is perhaps something more to the situation. This very long sixteenth-note may be a way of alluding to the preceding, "wedged" eighth-note upbeats, perhaps a way of playing that suggests the presence of a sub-surface b♭<sup>1</sup> –b♭<sup>1</sup> –a♮<sup>1</sup> , d♭<sup>2</sup> –d♭<sup>2</sup> –c2 imitative essence in m. 8; cf. Ex. 19.]

# **The theme as a tonal hierarchy**

Using the first eight bars of K. 282 as a testing-ground, and bringing in his later research on "tonal pitch space", Fred Lerdahl updates the bottom-up-and-yeteventually-top-down hierarchical method advanced in *A Generative Theory of Tonal Music*. 8

Lerdahl arrives at a quite Solomonic verdict when saying that "the inadequacy of the sequential model at surface levels has a reverse counterpart: the hierarchical model's predictions are perceptually attenuated at global levels". (p. 334) Later on, and in another context, he returns to this problem: "my feeling is that the [attractional] field does not reach very far over time, certainly less far than the larger levels of prolongational connection […] In other words, global prolongations depend not on attractions but on memory of structural features". (p. 358) But which features do we, as listeners, have reasons to remember? And why should we?

The latter question is partly answered by a surmise put forth in yet another context: "Probably naive listeners stay close to the surface while experienced listeners tend to hierarchize it". (p. 326) Experienced listeners are privileged because they have access to "global prolongations", to a dimension of musical appreciation that is closed to the naive ones.

These are complex matters that cannot be fully explained, let alone be exhaustively discussed, in the present context. Without detracting from the theoretical merits of Lerdahl's contribution – and there may be a good deal of sense in his musical conclusions, too – a few of his examples will (somewhat unjustly) be evaluated from the musician's perspective.

Let's first study the "time-span reduction" of a passage that makes up a small, but very important, part of the input for the "prolongational reduction" of the theme, i.e. for the global hierarchical representation characterizing an "experienced" recollection of the music.

Ex. 23 is supposed to show that it is "fruitless to do a sequential tension analysis of the musical surface". (p. 334) But just as in war the first victim is the truth, the first thing to disappear in the initial phrase is its goal and *raison d´être*, the e♭<sup>2</sup> giving a poignant glimpse of an accented six-non-four sonority. Not even von Dittersdorf would have considered Ex. 24 as equivalent to what

<sup>8</sup> "Calculating Tonal Tension", *Music Perception* 13(1996), 319–363; cf. Fred Lerdahl & Ray Jackendoff, *A Generative Theory of Tonal Music*, Cambridge, Mass 1983, and Fred Lerdahl, *Tonal Pitch Space*, Oxford University Press 2001

Mozart wrote. But obviously and unfortunately, the sweet/painful, short appoggiatura discord is not a feature worthy of being stored in an "experienced", i.e. hierarchic, recollection of the music. This in turn means that the urge for once more reaching e♭<sup>2</sup> (the baby) is an inaccessible value for the "experienced" listener (whose amnesia makes him/her content with the bathing water).

Within and far beyond m. 2 the small-print sixteenth-note e♭<sup>2</sup> is certainly more remarkable than the following d2 – arguably, it may even be more important than the quarter-note b♭<sup>1</sup> – and far from demonstrating that sequential listening is "fruitless", the barren, ever-more "experienced" structural essence of the passage shown in Ex. 23 casts doubts on the representativity of top/down hierarchical musical engagement.

The "naive" listener, with his/her myopic foresight and well-adapted memory handles the situation perfectly; he/she expects at least a Dittersdorfian resolution of e♭<sup>2</sup> to turn up at the next eighth- or quarter-note slot (cf. Ex. 25), and long after this resolution has expired, he/she will remember the precarious hot-spot e♭<sup>2</sup> . Sequential listening is not necessarily "inadequate", and the analytical principle that "an unstable event is assimilated to the goal that attracts it and is reduced out" (p. 357) is bound to have many exceptions when it comes to actual listening, let alone playing.

Recalling the way Entremont clings to the e♭<sup>2</sup> as long as possible, it seems likely that he does not whole-heartedly subscribe to the "experienced" hierarchical reduction of the first beat shown in Ex. 23. 9 Like so many other excellent musicians (and quite a few good listeners) he has preserved his capacity of being "naive", i.e. he is still able to understand music sequentially – an approach that does not necessarily yield but trivial, immediate gratifications or preclude longterm objects of contemplation. Considering what essentially *happens* in the first phrase, Entremont's playing may be explained by the fact that the "small-note" appoggiatura e♭<sup>2</sup> is irreducible. Without it, the preceding e♭<sup>2</sup> loses its sense of being an anticipation; without it, the emphatic rise from b♭<sup>1</sup> to e♭<sup>2</sup> is robbed of its meaning. Besides, we must not forget that musicians, unlike analysts, are bound to deal with music sequentially.

But there may be some listeners, perhaps trained rather than just "experienced", who do keep track of certain, supposedly essential features that may be used to construct overall tonal packages as the one shown in Ex. 26, a combination

<sup>9</sup> Perhaps there are some "naive" – and yet quite experienced – pianists out there who use the pedal to hold on to the e♭<sup>2</sup> until the first b1 ?

of Lerdahl's figures 8 and 34. Pointedly put, this "idealized" thirteen-event prolongational tree is a representation of how the theme emerges to superficial, impoverished listeners, who with or without clever top-down tension calculations adjust their memories.

To some listeners the passing leading-note in the left hand may perhaps in hindsight turn event 4 (i.e. the last beat of m. 1) into a wholesale applied dominant, but what happens to the clearly exposed, accented root-position Cminor chord produced by the stepwise motion in the bass and noticed in real time even by minimally attentive ears?10 The defoliated event-5 sonority has already been discussed. Another misrepresentation involves the second half of m. 2 and the first half of m. 3. The harmony is bluntly disambiguated by showing events 7–8 and 9–10 as dissonance-to-resolution units; the close, connecting similarity between the sonorities on both sides of the bar-line is disregarded, which means that the other option, involving a floating dominant-complex overlapping the units, is absent. 11 And the (possible) closing/starting elision at event 6 is not accounted for, nor is the fact that the theme is (presumably) made up of two phrases.

Although the hatched lines, allowing of alternative connections, mean that the strictly hierarchical approach is loosened, it is for various reasons questionable whether the prolongational tree in Ex. 26 really models an "experienced" experience of the theme, let alone what a musician has in mind. For one thing, all asynchronies are obliterated and the resolving-chord blobs are moved to accented positions, thus destroying the sense of delay and syncopation in Mozart's theme.

The top, prolongation-of-the-tonic node indicates that the representation of the theme is predicated on a commonplace. It is trivially true that formal units often return to their harmonic point of departure, but it is hard to believe that taking notice of this routine fact makes up a very important element in experienced listening or artistic music-making – or for that matter that such turn-of-the-mill recurrences determine how people understand what happens in between. A subtle and yet "structural" feature of the theme is that it is somehow made up of two phrases, and it may be assumed that even "experienced" listeners

<sup>10</sup> If understood as a C-minor chord, event 4 would not attach to the passing event 3 but directly to the stem issuing from event 1.

<sup>11</sup> Or perhaps it is taken into account, but event 9 should attach to event 8, rather than the other way around.

are able to pay attention to, remember, and enjoy this fundamental, good-making fact. 12 But in Ex. 26 event 10 attaches at a higher level than event 6.

Turning to the next-highest node in the tree, the first-inversion tonic chord in m. 3 owes its structural exaltation to the fact that it enjoys a prolongational relationship to the initial tonic. But it is most unlikely that anyone, however "experienced" he/she is, cares about this structural connection, predicated on stability-to-death. When this I6 chord occurs, the starting E♭-major chord has long ago lost its musical actuality; the description of the first-inversion chord as a prolongation presupposes the implausible activation of a "perceptually attenuated" memory. The I6 chord in m. 3 is busy with its own low-level duties, and so should those who play or listen to it be: it provides the unstable resolution of the preceding dominant seventh-chord (complex), it comes up with a slightly syncopated note for the ensuing suspension, and it brings a further note in the stepwise descent of the bass line.

The last observation makes for an additional remark. A "naive" listener, having a short but vivid and adequate memory-span, is likely to think that the penultimate dominant features the second-inversion f rather than the following root b♭ as its bass note. The f still resounds in the ear since pursuing the route of the by now quite obvious bass motion down to e♭ counts for more than the deflecting skip up to the dominant root, and since the b♭/d1 third appears to be a middle-register affair. If the naive listener is also a pianist, this perspective gains in strength; the third is played by the "middle-register" fingers of the left hand. Musicians are also listeners and, generally speaking, they are prone to listen in their own peculiar, bodily way. In other words, for listeners and pianists alike a performance that puts the fourth-beat root-position dominant in the shadow is better than a rendering turning it into a high-level structural event. There may be pianists who use the pedal or the little finger to preserve the f into the fourth beat.

The musically unfortunate thing with the tree shown in Ex. 26 is that it shows all that happens up to the second beat of m. 3 as being encapsulated within the E♭-major tonic. ("At last a tonic sonority again"; is this really how a good listener understands, or a good pianist treats, the I6 chord?) A truly experienced listener, i.e. a listener who is still fairly naive, is perhaps rather inclined to suspect that a cadence may be forthcoming and connect the I6 chord to the final tonic.

But let's hope that no listener is so clever at rear-view listening that, say, everything from the third beat of m. 2 on is also included into this forthcoming

<sup>12</sup> It is true that a sausage is held together by an overall, skin-like strong prolongation, but this fact does not enjoin us to eat the enclosed meat in one big bite.

cadence. As listeners we do not know, and do not want to know, that much about the musical future. And if we, due to many encounters with K. 282, do understand this C-major applied-dominant chord as already belonging to the theme's cadence, we are very experienced listeners indeed, but hardly "ideal" ones since we have ceased to appreciate the uncertainties, the passages of free flight, that good music affords. Nor would of course a performance, robbing m. 2 of its sense of an opening towards the unknown, be an ideal presentation of the music.

The lesson to be learnt is that just as we must be on guard against structural beginnings engulfing more than they can digest, we have to resist the insatiate appetite of structural conclusions. The best things in music tend to happen in between, and when playing (and analysing) it is wise to leave some space open between Scylla and Charybdis.

The next node in Ex. 26 involves the progression from the initial tonic to the would-be closing second-beat dominant in m. 2. But if there are two phrases in the theme, and if a sense of elision is involved, the second-beat closing dominant (topped by b♭<sup>1</sup> ) and the second-beat starting dominant (also featuring b♭<sup>1</sup> as its top note) should by rights have separate stems. The former dominant should be attached to the initial tonic, whereas the latter only retrospectively (if at all) emerges as belonging to the final tonic.

In short, Ex. 26 is too Schenkerian to be of any use for a musician. Whereas "experienced" listeners are supposed to excel in the art of remembering (and forgetting!) – no matter whether the "attractional field" is likely to be operative – an experienced musician listens ahead, taking full advantage of the fact that he/ she knows what is going to happen. And from his/her perspective the devil is not in the details. Quite to the contrary – a good musician knows that the only way to influence the future course of a musical process, and hence to be in command of the whole, is through its details.

It might be of some interest to adapt Lerdahl's tree representation so as to show an alternative picture that corresponds better to what a pianist wants to convey, and that perhaps also models what a truly experienced listener gets out of the theme, a listener who is still "naive" enough to respond to what a passage of music may mean beyond its hierarchic "tonal structure". The decomposed tree shown in Ex. 27 suggests that there is some kind of "sequential" adventure in mm. 1–4; new events emerge, forming aggregates that, as the case may be, attach to past events or hold out the prospect of future ones. Unlike Lerdahl's hierarchic structure, Ex. 27 is an attempt to represent what musicians and listeners are interested in: the "structure of discovery".

Lerdahl also discusses the dual harmonic interpretation of mm. 6–7, and holds that the source of the minor-tonic versus minor-subdominant ambiguity of the first-position E♭-minor chord is not a matter of "prolongational level but of prospective versus retrospective hearing". (p. 341) This is true, but as previously pointed out it is the pianist's prerogative to tip the balance in favour of one or the other of the two alternatives. Judging from the dynamic marks, Mozart seems to have preferred the immediate and retrospective, *forte*-then-*piano*, major-tominor darkening shift, "naive" as he was.

#### **Tensions in tonal space; attractions and yearning**

The many-faceted algorithms proposed in Lerdahl's paper clarify important issues in our understanding of tonal tension/attraction both when it comes to melody and harmony. It is particularly enlightening that he makes calculations accounting for the music both when it is heard as a sequence and when it is understood as a hierarchy.

Lerdahl's Fig. 32, shown here as Ex. 28, offers a comparison between his own, careful quantifications of melodic attraction and the numbers signifying tonal position. 13 Tonal position numbers, determined in accordance with the local harmonic context, are highly relevant for tension and hence for interpretation. For instance, the 4 over the second e♭<sup>2</sup> in m. 1 indicates that this note should be conceived of and played as an anticipation, that it is to be understood in relation to the forthcoming dominant, established only after the bar-line.

That interpretation is involved emerges particularly from situations where alternatives are conceivable. The number 6 above the second d2 in m. 2 might be contested. Does the listener really understand that he/she is already in the territory of the ii-chord? The pianist knows that an F-minor-like sonority will turn up, but it may be better to disregard this precognition in order to feel and perhaps express the uncertainty that in 1955 enticed me into playing an F-major sixth on the fourth beat. Turning to the f2 in m. 2, it is difficult to hear it as just a bland 8-over-ii since the descending gesture is rather understood as belonging

<sup>13</sup> According to (experimentally corroborated) musical wisdom, the following numbers bring information as to tonal attraction: 2 is more drawn towards 1 than towards 3; 4 and 7 are strongly attracted to 3 and 8, respectively; 6 pulls downwards to 5, etc. When assessing melodic attractions Lerdahl laudably takes account also of melodic direction; when a rising or falling motion has been established, it may reinforce or counteract (perhaps even outweigh) the "raw" tonal attractions.

to the forthcoming dominant-seventh chord, or as embedded into the dominant complex straddling the bar-line; hence, the position number should rather be 5 as given between the staves.

But as Ex. 29 shows, the latter passage opens up for still another, and quite meaningful interpretation. Here Lerdahl combines tonal-position numbers deriving from the overall E♭-major tonality with arrows accounting for the sense of tension/attraction; the hierarchical account is fuelled by relationships that are local enough to establish "attractional fields". Don't be too slow, and two strands will come to the fore, clearly separated in terms register and time, and inviting to be expressed. If the first quasi-appoggiatura note of the pairs are suitably brought out, the passage as a whole will invite to be played *diminuendo*.

Jamshed J. Bharucha proposes a mathematical model for how unstable notes are attracted upwards or downwards within their tonal neighbourhood so as to find resolution in stable notes. 14 He postulates that the force of the attraction depends on two factors: the proximity of the notes involved, and their order of appearance – attraction is regarded as a matter of expectation, and therefore the implied, "anchor" note must follow after the implying, anchored note. Bharucha tests his "yearning-vector" model by studying the behaviour of non-chord vs. chord notes in the exposition of K. 282, and the actual motions agree closely with the predicted ones.

All this appears most plausible, and yet there is something to add. The anchoring is arguably not quite as "asymmetric" as Bharucha takes for granted. It is certainly true that non-chord notes yearn for future resolution, but there must also be some scope for what we may call "pre-anchoring". This applies especially if we adopt the perspective of the musician, if we bring into the model the person being responsible for how the music proceeds from one note to the next.

Passing-notes and neighbour-notes are not just drawn to the following stable note, they must also free themselves from the stable note of departure, and to a musician it is more important to clarify this active backward relationship than to waste expression to demonstrate the attraction emanating from a forthcoming note that tends to be patently expected. As a musician you have to prepare for such non-chord notes; they must not just happen.

Turning back to Lerdahl, he describes harmonic and melodic tensions/ attractions in terms borrowed from Newtonian physics, an analogy that makes

<sup>14</sup> "Melodic Anchoring", *Music Perception* 13(1996), 383–400

good sense. Bharucha and Lerdahl cover the same ground, and both of them describe attraction/anchoring as an asymmetric phenomenon: high-tension events emerge as prominent because they seem attracted to some following stable event. But Lerdahl also opens up for the possibility that the preceding event should be considered when assessing the forces determining attraction. He remarks that "it is unclear what force could be invoked for the pushing of stable to unstable tones" (p. 342), and goes on by proposing an "attraction algorithm", making it possible to "formalize multiple attractions by summing all the attractional vectors".

But if we turn from listening to music making, it becomes clear that the "force" responsible for the "pushing of stable to unstable tones" is the musician, and it also appears that the concept of 'attraction' becomes inverted. Using dynamic and/or durational means, high-tension events must be brought out because they *resist* the attraction emanating from the *preceding* stable event. (Generally speaking, understated high-tension events are a characteristic of dull performances.)

When it comes to neighbour-notes and passing-notes this principle – amounting to the necessity to free the unstable note from the tonal gravity inherent in the preceding note – is quite evident, but what about appoggiaturas? Since such dissonances are also rendered prominent, it appears that it is the attraction of the *following* event that must be resisted. Consider, for instance, the appoggiatura beginning m. 2 in Mozart's theme. When you hear the e♭<sup>2</sup> , you cannot but feel the attraction to the following d2 , but when you play it, when you bring it out, you are expressing resistance.

#### **A bottom/up implicational analysis**

Dealing primarily with the theme and its extension up to m. 8, Eugene Narmour shows how a bottom-up analysis – pursued in four musical dimensions (melody, harmony, duration, and metre), working according to a limited number of shared principles, and being constantly involved in negotiations with learned stylistic conventions – can "account for a plethora of relations" and "an extreme amount of artistic diversity". 15 (p. 308) It is impossible to give a fair idea of Narmour's

<sup>15</sup> "Analyzing Form and Measuring Perceptual Content in Mozart's Sonata K. 282: A New Theory of Parametric Analogues", *Music Perception* 13(1996), 265–318. Narmour's point of departure is his two-volume work *The Analysis and Cognition of Basic Melodic Structures* and *The Analysis and Cognition of Melodic Complexity*, Chicago University Press 1990 and 1992, respectively.

work here, suffice it to say that he, unlike Charlot in *The Pawn Shop*, after having studied the parts carefully puts them back into the watch.

The bottom-up approach is advantageous when it comes to interpretation. Implications, allowing us to envisage the event(s) to follow, are analyzed on a note-by-note basis, and they indicate at what points the musician may prepare the listener for future events – or, when suitable' keep them secret or even suggest never-to-be-realized outcomes. Another bonus is that Narmour's idea of metre allows of flexibility as to the phenomenal location of accents – strong events may occur at nominally weak beats, and *vice versa*.

Narmour calls attention to a number of motivic associations in the theme (and beyond). There is an obvious and rewarding association between the regularly occurring accented appoggiaturas e♭<sup>2</sup> –d2 , d2 –c2 , and c2 –b♭<sup>1</sup> , suggesting a falling line overlapping the vague phrase demarcation. [But if you want to launch a second phrase beginning with the third-beat appoggiatura d2 –c2 , it is better to think of it as a fresh idea.] The closing five-note gesture starting from e♭<sup>2</sup> in m. 2 and the non-closing five-note motion beginning on f2 do have traits in common, and the association gains in strength if one assumes that there is a dominant complex in mm. 2–3.

It is a bit puzzling that the g1 –rest–b1 fragment starting m. 4 is described as making up a rising-third "dyad". The g1 emerges as patently implied in virtue of being the strongly expected endpoint of the preceding b♭<sup>1</sup> –a♭<sup>1</sup> – descent; the final g1 is conspicuously short, but nothing unsaid is left in the air. One might have thought that the "additional" eighth-note b♭<sup>1</sup> would be analysed as a "monad", i.e. as becomes its strange isolation.

But perhaps a g1 –b♭<sup>1</sup> dyad can be rendered so as to suggest a denied implication starting from a shared g1 ? The "natural" next note would have been an accented e♭<sup>2</sup> , but the e♭<sup>2</sup> that does turn up arrives too late as the first short note of a weakbeat ornament lending emphasis to g2 . Consider Dittersdorf 's version of m. 4; cf. Ex. 30.

Turning to the monad option, and considering the fact that non-connected melody notes are rare and often problematical, what musical meaning can the b♭<sup>1</sup> have? It might perhaps be played so as to bring the final (and additional) b♭<sup>1</sup> of the b♭<sup>1</sup> 's permeating the theme, a reading that connects backwards beyond the preceding g1 and that, notwithstanding the new accompaniment starting on the first beat of m. 4, postpones the formal shift to the *forte* third beat.

Ex. 31 shows Narmour's hierarchical analysis of the theme's melody. (p. 273) Reading from the bottom to the top, the music gradually turns less inspired and inspiring, but there is no reason to complain; there is a skeleton in all of us, and in Mozart's theme there is evidently an "all-about-b♭<sup>1</sup> " backbone. But if you want to fuel your artistic imagination, you should study the implications as you proceed on the route upwards; again, it is not necessarily true that the devil is in the details. The sad thing is that the growing lack of ambiguities threatens to kill off some of your interpretational darlings.

For instance, the bracket from the first e♭<sup>2</sup> in m. 1 to the second b♭<sup>1</sup> in m. 2 turns the latter note into a closing event, which means that the (possible) sense of an elision has disappeared. This is the necessary outcome of the fact that Narmour's implicational analysis – the joint result of perceptually given *gestalts* and learned schemata – is predicated on the most probable structure, the structure that we are (more or less) bound to hear. But the outcome of some situations may perhaps be re-negotiated, i.e. the incoming perceptual data and/or the stylistic environment may sometimes be understood differently.

It seems that a crucial aspect of musical interpretation is to transgress what *must* be heard in order to convey what *might* be heard. Or otherwise put: you do not always have to play melodies as *they* want – sometimes it may be warranted and productive to bend them according to *your* will. This childish, or rather this adolescent, approach means that neither Narmour's bottom/up implicational hierarchy, nor Lerdahl's top/down reductive tree, should be regarded as imposing binding restrictions as to what is possible and perhaps rewarding to express. For instance, if you want the first phrase to end as an inconspicuous sigh following after an exposed dissonance, you are free to do so by holding on to e♭<sup>2</sup> as long as possible, by not bringing out c2 –b♭<sup>1</sup> as a second appoggiatura, and by avoiding as best you can to give the impression that there is a b♭<sup>1</sup> –b♭<sup>1</sup> anticipation closing the phrase on the second beat with a quasi-accented bump.

#### **Remarks on Meyer's commentary**

As his role bids, Leonard B. Meyer covers a great variety of topics. 16 Some of them address methodological issues and are therefore of less concern in the present context. The emphasis here will be on observations of relevance for interpretation, and of particular interest are his discussions of subjects touched upon previously in this essay.

<sup>16</sup> "Commentary", *Music Perception* 13(1996), 455–483.

As regards the first two beats of m. 2 Meyer points out that they "are experienced not as a surface embellishment but as vital parts of a psychological process", i.e. "the realization that fills the gap generated by the preceding rising fourth". (p. 461) This observation seems to bear on the question of the (possible) elision on the second beat.

It seems that two layers of understanding are involved. The local gap, starting from the last b♭<sup>1</sup> in m. 1, can be heard as expending rather closing itself with the first, sixteenth-note b♭<sup>1</sup> in m. 2. The high-level gap, or perhaps rather the superordinate b♭<sup>1</sup> –e♭<sup>2</sup> –b♭<sup>1</sup> motion, issues from the first b♭<sup>1</sup> in m. 1 and is completed only by the second b♭<sup>1</sup> in m. 2, where another connection eventually leading to g1 in m. 4 apparently starts. Alternatively, both descents from e♭<sup>2</sup> close on the accented, quarter-note b♭<sup>1</sup> , in which case there is a sense of elision, of a shared event – or else the second phrase is postponed to have a flying start on the third beat.

(In this context it may be mentioned that Meyer regards the enigmatic b♭<sup>1</sup> in m. 4 as a further attempt to open up a rising gap, this time to g2 -beyond-e♭<sup>2</sup> .)

Turning to what may be called a large-scale formal elision, Meyer suggests that mm. 19–21, otherwise understood as making up the last half of the development, perhaps bring a disguised return of the theme; cf. Ex. 1. The left hand in m. 19 resembles the bass motion of m. 1 (c) while the right hand hints at its rising middle voice (b); the treble passages in m. 20 twice hurry through the melody of the first phrase (a); the final c2 –b♭<sup>1</sup> step turns up in m. 21. If this relationship is somehow clarified (perhaps when playing the repeat), the listener might hear these bars as a stand-in for the main theme and feel that m. 22 (corresponding to m. 4) brings in the contrasting idea in due time. These allusions to the theme pave the way for understanding the movement as a binary construct.

Like the present author, Meyer argues that dissonant notes are often "anchored" to the preceding note, rather than (just) to the following one, and that the former relationship may overrule temporal proximity – as it does in the neighbour-note motion starting the first movement. Here the relationship between the initial b♭<sup>1</sup> and the dissonant c2 strengthens the sense of implication inherent in the first two notes of the three-note neighbour-note formula, and the tension is enhanced by the temporal distance between the anchored note and its anchoring predecessor.

This brings us to Meyer's remarks on the timing between the anchored note and the ensuing anchor note. In composing and playing there is often an inverse relationship between pitch proximity and temporal proximity: "when pitch proximity creates a strong implication, temporal delay (distance) enhances expectation". (p. 464) Succinctly put, "yearning" takes some time. This observation is borne out twice in m. 8. (cf. Ex. 32) Entremont devotes himself to "pre-yearning" when prolonging the duration of the last, anticipating sixteenthnote chord, while Mozart, when he chose a quarter-note instead of an eighthnote value on the third beat, prolonged the appoggiatura chord to make the most out of its yearning quality.

# **A gambit and the Gambit**

The theme certainly makes up a complex *Vexierbild*, and if there is any word that captures the essence of its melodic, rhythmic, harmonic, and formal course it is – with all due respect – obliquity. How do you play such music? Perhaps you have to close one of your eyes to be able to choose among the options? Or perhaps we should turn to Robert O. Gjerdingen for advice?

In his analysis of the movement and its three-bar "theme" in particular, he adopts ideas to be found in 18th-Century compositional practice, ways of thinking that were perhaps also relevant for how musically cognizant people once listened. 17 Simply put, compositions of taste were to be started according to a certain protocol, stipulating a number of "gambits" to be followed by certain "ripostes", or as Gjerdingen puts it in one of his headings: "What do you say after you say hello?"

The first movement of K. 282 begins with a 5–8–5 melodic idea combined with a stepwise falling I–V bass motion (perhaps derived from the *Romanesca* pattern), and after this gambit comes a *Prinner* riposte, a configuration made up of a falling fourth in the melody proceeding in parallel-tenth tandem with a descending fourth in the bass. But this is not all: the top voice of the *Prinner* is overlaid with a *Fonte*, i.e. with two falling motions, of which the first ends in the minor mode and the second, set one step lower, closes in major. (cf. Ex. 33 and Ex. 1)

Already at the age of ten, I was enough of an 18th-Century musician to identify the *Prinner*. Although Mozart's treble/bass co-ordination is far from pedantic (fortunately), I was quite able to appreciate the presence of two principal falling strands in mm. 2–3. On the other hand, I did not realize that the *Prinner* "riposte" phrase was the courtly consequence of the preceding, *Romanesca*-based "gambit" phrase. To me the juxtaposition of the two items as Mozart wrote them down emerged (and still emerges) as a unified stroke of genius.

As to the superimposed *Fonte* layer, it is arguably less convincing. Bars 16–19 may perhaps, as Gjerdingen holds, make up a *Fonte* construction, but due to

<sup>17</sup> "Courtly Behavior", *Music Perception* 13(1996), 365–382.

the elision there is no double-bar situation in m. 2, and the two members of the would-be *Fonte* figure are quite rudimentary.

But if Gjerdingen is right, von Dittersdorf was also right when he (in general) criticized Mozart for piling up figures in complex ways, instead of presenting them as neatly separated subjects of a conversation. 18 (pp. 277–278) On the other hand and leaving courtly behaviours, both of them may be wrong. The further course of analytic thinking (and composing) was not dictated by Dittersdorfs, but inspired by Haydn, Mozart, and Beethoven at their best, and – if *I* am not entirely wrong – there are in the first three bars of K. 282 seeds of the future amalgamated with conventional traces of the past. These seeds once disturbed people like von Dittersdorf whereas our syncretistic ears are able to appreciate them.

Gjerdingen seems to advocate listening to Mozart in terms of "courtly behavior", but this mode of listening (musical understanding) cannot very well be the only one that "rewards experience, attention, and active engagement". Nor is it – all the better – quite fair to say that "listeners of later ages" are "accustomed to a more passive mode of listening to broad harmonic progressions". 19 (p. 281) But it is not necessary to accept these Dr. Jekyll and Mr. Hyde alternatives of musical appreciation. As my description of the first three bars of K. 282 has hopefully shown, there are (and presumably were) many quite worthwhile and engaging facets of the music that exceed (and exceeded) courtly listening in terms of keeping track of well-mannered procedures, and that transcend "structural" enjoyment of contrapuntally prolonged harmonies.

If composing is reduced to a matter of exchanges in a courtly conversation, courtly listening runs the risk of becoming a matter of disengaged identification and decoding. But when listening out of the court, is the very beginning of K. 282 merely a "Hello", doesn't the *persona* hiding in the music rather utter something startlingly original while looking you straight in your eyes? Has this extraordinary opening really a "Hello" function? No, it may be claimed that the human content of the first phrase in K. 282 goes (and presumably went) far beyond a "Hello".

<sup>18</sup> Hello again, Ditters! I owe you an apology since your composing once qualified as being fully in accord with the state of the art. I chose your very name because of its funny, quasi-tautological ring, and because of the courtly correspondence on the part of its bearer between would-be hereditary nobility and acquired mediocrity.

<sup>19</sup> Evidently, Gjerdingen thinks of Schenkerian theory: "the late-Romantic belief that all great music grew out of a governing harmony extended through time on a framework of 'pure' counterpoint" (p. 367).

Adopting for the sake of argument the courtly perspective, *if* the first idea of the sonata merely amounts to a "Hello", the ensuing *Prinner* idea is merely a "How-are-you?" phrase, complemented by a *Fonte* remark about the weather. But Richard Rosenberg said in 1972 that the *Prinner/Fonte* phrase in mm. 2–3 is like a "melodic bud". This may or may not be an anachronistic observation, and it may even be an "organic, Romantic metaphor of growth". (p. 278) Nevertheless, his wording suggests that, at least to him, this particular "riposte" offers a bonus far beyond what you get out of a polite "How are you?". Instead of being derided, Rosenberg is to be held in high regard since he has not given us a "nothing-but" explanation.

It seems that a basic mistake is involved. There is no necessary relationship between on the one hand how a work was (perhaps) composed, and on the other how one should listen to it. This "replication" principle obviously fails in dodecaphonic music, and it may be too fool-proof to be useful even when it comes to K. 282. As a ballet viewer, forget about the *jetés*, *pliés*, and *relevés* – the vocables of choreographic conversation – and enjoy the dancing, try to catch what it means!

Turning to the counterpart of courtly listening, courtly playing does not seem very attractive. In an 18th-Century courtly conversation it was presumably a minimal requirement that you talked very clearly so as to let everyone get the message (if any). But wouldn't it be disappointing to listen to a pianist, who first and foremost clearly brought out the "How-are-you" essence of the *Prinner* sequence of falling tenths? Any pianist with ambitions surely want to make mm. 2–3 disclose its more existential "who-are-you?" aspect. Gjerdingen has Dittersdorfed Mozart's theme, but musicians are not pleased at being reminded of things that they prefer to remain ignorant of, even if the observations are, or once were, true.

At this point a short, beyond-*Music-Perception* digression is due. It seems that another historically inspired reading of the movement may be more productive when it comes to interpretation. At least if you are looking for variety, Gjerdingen's compositional devices can be exchanged for the musical/cultural topics that the music may be taken to represent or allude to. The composers of the 18th Century had access to a thesaurus of musical clichés – melodic, rhythmic, harmonic, and structural patterns associating to various, more or less extra-musical phenomena. A piece of instrumental music like a piano sonata might have used these clichés so as to offer a witty sequence of musical references that the listeners were likely to recognize.

At least to present-day, harder-to-amuse ears there is a potential risk involved in this, of course. If virtually every distinct passage of, say, the first movement of K. 282 (heterogeneous as it is) were demonstratively played and understood as reminiscent of some other kind of music and its cultural context, the score would turn into a picture-book, and the music would run the risk of becoming a parade lacking coherence. Hence, if a musician adopts this attitude, spurring him/her to indulge in contrasts and characterizations, he/she has to rely on the composer's – and the musician's own – capacity to create continuity.

Leonard G. Ratner is a good guide to this world of references. 20 As to the first bars of K. 282 he suggests that they "evoke the style and texture of a wind serenade". (1991, p. 616) But this remark seems more to the point when it comes to the second *Menuetto* movement of the sonata.

The term "gambit" bears unfortunate associations to chess. Google has it that "a gambit (from ancient Italian *gambetto*, meaning 'to trip') is a chess opening in which a player, more often White, sacrifices material, usually a pawn". In books on chess playing various gambits are described together with the appropriate countering "ripostes". It is common knowledge that the first moves in games of chess are standardized and not very exciting to watch; nor are we keen on listening to musical openings of that sort. But the "game" of K. 282 is extremely interesting from its very start – indeed, the initial three bars (and the Coda) make up the best part of the first movement. Generally speaking' musicians and listeners have little or nothing to gain by tabulating musical procedures, and even if there are conventions afloat since the days of the *partimento* practice, it is more rewarding to pretend that they do not exist.

But a "gambit" is also a fictional superhero in American comic books published by Marvel Comics. Relying again on Google: "Gambit has the mutant ability to tap into the potential energy contained within an object and transform it into kinetic energy upon touching it". This is, if I am not entirely mistaken, what happens in the "conversation" overheard in mm. 1–3.

So, after all this has been said, what can I – and what can you – do to make the theme levitate? Don't let its first phrase end straightforwardly at the quarternote b♭<sup>1</sup> , but try to turn the preceding b♭<sup>1</sup> into a gently closing non-anticipation. Use the long b♭<sup>1</sup> , which is too expected to be interesting, to start the second phrase, or (even better) let the melody's own urge to continue issue from the "false-upbeat" motif d2 –c2 . Since the falling fourth of the *Prinner* takes care of

<sup>20</sup> Cf. *Classic Music. Expression, Form, and Style*, New York 1980, and specifically "Topical Contents in Mozart's Keyboard Sonatas", *Early Music* 19(1991), 615–619.

itself, give attention to the second, futile attempt to reach e♭<sup>2</sup> , and hide away the dividing *Fonte* by bringing out the sense of a long, yearning dominant, turning the accented leading-note e♮<sup>1</sup> into a passing-note. And while the melody eventually expends itself at g1 , don't forget that it is still about b♭<sup>1</sup> .

The Gambit in the K. 282 theme is Mozart, but he needs another, assistant gambit – the pianist.

# **Music examples**

# **Chapter 1 On scores and works of music**

Ex. 1a

Ex. 1b

# **Chapter 2 Sonate, que te fais-je?**

Ex. 1

Ex. 2 a/b

# **Chapter 3 Directions and compliance**

### **Chapter 4 Loyal disobedience**

Ex. 1 a/b

Ex. 7 a/c

Ex. 8a Exposition

Ex. 8b Recapitulation

Ex. 9 a/b

Ex. 13 a/b

Ex. 19

Ex. 20 a/c

Ex. 21 a/b

Ex. 23

Ex. 26 a/b

Ex. 30 a/c

# **Chapter 6 A comprehensive approach to musical idiomatic**

Ex. 2 a/c

Ex. 3 a/b

Ex. 3 c/d

# **Chapter 8 Reduction and interpretation**

Ex. 1

Ex. 6 a/b

# **Chapter 9 Dissentient views on a minuet**

Ex. 3

Ex. 6

# **Chapter 10 Interpreting a bagatelle**

# **Chapter 11 Tonal structure vs. modes of continuation**

Ex. 2 a/c

Ex. 3b

# **Chapter 12 Prelude to the art of continuation**

Ex. 1 (cont.)

Ex. 3 a/i

# **Chapter 13 Interpretation as continuation**

Ex. 1

# **Chapter 14 Musical dialogue in a Romantic violin sonata**

Ex. 1a

Ex. 1a (cont.)

Ex. 1b

# **Chapter 15 Chopin themes**

Ex. 7

Ex. 7 a/g

Ex. 7 a/g (cont.)

Ex. 8 a/b

# **Chapter 16 Keyboard commentaries on K. 282**

Ex. 1 (cont.)

Ex. 1 (cont.)

Ex. 3

Ex. 7

Ex. 13

Ex. 14

Ex. 17

Ex. 18

Ex. 22

Ex. 23

Ex. 25

Ex. 28

Ex. 31

# **References**


metrikkonferensen i Umeå 4–7 oktober 2001, Skrifter utgivna av Centrum för Metriska Studier 14, Göteborg 2003a


Bengt Edlund, "Musical Dialogue in a Classical Piano Sonata" in Varia 1


Bengt Edlund, Varia 1 https://doi.org/10.37852/oblu.156


# **Bibliography**

	- Allusions and affinities. Tracing an ominous motif
	- Evidence and counter-evidence. Making sense of the A-minor Prelude
	- Music at the analyst's coach and at the musicians stand. The tonal structure of the E-minor Prelude
	- Left-hand melody and tonal structure. Towards a non-Schenkerian account of the B-minor Prelude
	- How could analysis be deconstructed by the A-major Prelude?
	- Reconsidering the C-minor Prelude
	- The phenomenology of fingering. Structure and ontology in the F minor Etude from Méthode des méthodes
	- From structure to content. Ominous allusions and the programme of the Second Ballade
	- Schenkerian theory and better comparison. An out-of-the-way perspective
	- Disciplining reduction and tonalizing interpretation
	- Is tonal music hierarchic? An impenitent sermon
	- Prolongation vs. implication
	- A hitch-hiker's guide to the repeat
	- Schubert, Schumann, and Schenkerism. Tonal vs. focal reduction
	- Syntactic vs. rhetoric structure. Language, music, and tonal reduction
	- Tonics and returns. A modest investigation
	- Shaving Schenker
	- Analytical variations on a theme by Mozart
	- In defence of musical ambiguity
	- Mozart out of proportion. Searching for the golden section in music
	- Hidden repetitions and uncovered parallelisms
	- An das ferne Verwandte. Common ideas and ideas in common
	- Warum Grillen?
	- Suing a sound-alike
	- Schubert's promising note. Further exercises in musical hermeneutics
	- On scores and works of music. Interpretation and identity
	- Sonate, que te fais-je? Towards a theory of interpretation
	- Representation of metre in performance. A study of Bach melodies
	- Communicating musical metre. An expanded restudy
	- Categories and types of anticipation in music
	- Symmetry in music
	- Interpreting another bagatelle
	- Navigating in moonlight
	- Interpreting Syrinx
	- Approaching the irresistible and preparing the scaring
	- Musical dialogue in a Classical piano sonata
	- Monologue as conversation
	- Directions and compliance. The development
	- Proprioceptive patterns in music
	- Musical conception of abstract film. The case of Viking Eggeling's Diagonal Symphony
	- Additional examples:
		- Representation of metre in performance
		- Hidden repetitions and uncovered parallelisms
		- *An das ferne Verwandte*

Chopinowska, Narodowy Instytut Fryderyka Chopina & Musica Iagellonica, 2003; cf. 1.7


(In Swedish; selection)

	- Interpretationsforskning med hjälp av fonogram
	- Musikvetenskap en vetenskap om artefakter
	- Att sjunga efter sin näbb
	- Redogörelse framlagd för en akademi
	- Reduktiv analys och reducerad insikt
	- Taktstreck och musicerande
	- Teatralitet i absolut music eller bara dialog?
	- Tecken och tolkning