From Phrase to Phrasing

A Classical Perspective

Jan Willem Nelleke

De Muzyk zy er toe bestemd, ons gestadig, op eene streelende wyze, orde voor te houden, in te prenten en tot derzelver liefde meer en meer te ontvonken. (Jacob Wilhelm Lustig – Inleiding tot de Muzykkunde, 1751)

Music is destined to continually, and mellifluously, present us with order, impress it unto us and more and more kindle the love for said order. (translation Jan Willem Nelleke)

Introduction

This paper aims to put phrases at the centre of performance practise, an approach that is, in my opinion, long overdue. The research question on which it is based, ‘What can we learn about musical phrases from period sources?’ is an open question, and therefore this paper will be more like an exploration. I have been studying many treatises, mainly from the Classical period, about composition (concerning topics related to the structure of a phrase) and performance (on how to give shape to a phrase as a performer). By exploring this information in relation to scores, I hope to bring theory and practice together, and raise interesting and maybe unexpected aspects of the subject.

Why is this research necessary at all? Probably no-one will disagree that good phrasing is essential. One could even argue that phrasing almost seems to define musicality: someone may play a score perfectly and precisely but a lack of phrasing will demonstrate without fail that he or she has not ‘understood’ or ‘felt’ the music. Phrasing thus encompasses understanding and feeling, and is therefore not an embellishment but fundamentally at the heart of music.

But where is the problem? We all know how the phrases go, don't we? It is my impression that we often take phrasing for granted; we know, or think we know, where a phrase leads because we feel it, but is that really all there is to it? Surely if phrasing makes music ‘understandable’ it implies that it is an agent in clarifying how music is structured; it is likely that a better understanding of phrase structure will enhance our playing. This is the main reason I chose this subject.

The focus will be on music from the Classical period which I define, rather arbitrarily, as from 1750 to 1800. The dates are not crucial but the style is, with its emphasis on melody and well-balanced structure. These structures were prominent within this time frame but not limited to it; we find them sometimes earlier but definitely later like in present day commercial music. In that sense they are truly ‘classic’.

The Classical period was chosen because this well-organized music seemed a good starting point for looking at fundamental principles in phrases and phrasing; but also because the period is interesting as a meeting point of different performing traditions, on modern and period instruments. Though trained on the modern piano, I have experience on period instruments as well and I have found the difference in approach intriguing. It has forced me to reconsider musical certainties and has triggered an interest in basic concepts in a historical context, hence the choice for period sources.

Some further clarifications: I am not a music theorist. I have had the basic training of any professional musician but I have no special qualification. I am a performer doing research to inspire and inform performance and teaching; for music theory specialists my discussion of the theoretical aspects might be too basic.

The show cases and examples other than from treatises, are chosen from compositions I encounter in my capacity as pianist, coach and accompanist. The gamut is therefore limited but the principals behind it not.

Conventions

Words like phrase, period, sentence, etc. have both a general and a musical meaning. But even in a specifically musical context these terms are not unequivocally defined. I will use these words according to eighteenth-century practice with a strong relation to punctuation in language. I want to emphasize that I do not use these terms as defined in later times by for example Arnold Schoenberg; therefore a period does not necessarily have an antecedent-consequent form, and to prevent confusion I will completely avoid the use of sentence as a musical term.

The following terms and definitions will be used, see also Fig. 1 (after an example by A. Reicha: (Anton Reicha, Traité de haute composition musicale, Czerny edition, 4 vols (Vienna, 1824), II, 351.)

Phrase - The shortest passage of music that expresses a more or less complete thought.
Phrase-member - A part of a phrase, an idea that needs to be elaborated.
Period - A group of phrases that reaches a conclusion.
Section - A main part of a movement, usually formed by a group of periods except in shorter pieces.
Phrase-length - The length of a phrase as related to the number of bars (4 in the example).
Phrase-rhythm - A higher level of rhythm created by alternation of phrase-lengths (4 + 4 in the example).
Cadence-point - The beat where the phrase wants to land (not necessarily the last note of a phrase).

Figure 1: Terminology

These terms and definitions have been chosen in an attempt to translate and correlate treatises from different authors. There is an abundance of words for identical concepts and, confusingly, the same word can mean different things with different authors. Appendix 1 contains a list of terms I have encountered during my research, categorised according to the terminology used here.

Unless stated otherwise, all translations are mine.

Unless indicated otherwise, all modernised type-setting of musical examples is mine, with occasional additions or clarifications in square brackets [ ].

From Phrase to Phrasing

Chapter 1 – Prelude: Playing with style

Rather than making a doomed attempt at defining ‘Classical style’, I would like to explore the style from within. When asked to characterize Classical, most of us would probably come up with something akin to ‘melodious’ and ‘well-proportioned’. But what would that mean to a musician of the period? Or, to turn it around, what was considered to be not a proper melody, or a not well-balanced composition? Anton Reicha (1770-1836) provides an example of a sequence of tones without melodic sense: (Anton Reicha, Traité de haute composition musicale, Czerny edition, 4 vols (Vienna, 1824), II, 590.)

Figure 2: A nonsensical melody by Anton Reicha

He is right of course, but why? Heinrich Christoph Koch (1749-1816), one of the most important music theorists of his time, defines a proper melody as a sequence of notes that: (Heinrich Christoph Koch, Musikalisches Lexikon (Frankfurt am Main, 1802), s.v. Melodie.)

(a) is taken from alternating steps of the key it is based on,
(b) is arranged according to a time signature,
(c) contains certain repose-moments (Koch's beautiful description is Ruhepunkte des Geistes meaning ‘moments where the mind can come to rest’), by which
(d) it can be subdivided into individual parts.3

We can check:

(a) The key is C major, only the F# is not from the scale but it is the most common one to be added. (Resolving the F# upwards would have been more elegant but as it stands, it is strangely reminiscent of the alpenhorn theme in Brahms’s First Symphony)
(b) It neatly fits in c or ¢.
(c) There are some long notes, surely your mind could repose there.
(d) Here is a problem. At a pinch it would be possible to make divisions but none feel obvious or satisfying. This means a long note is not necessarily a repose-moment; the repose needs to create a sense of completeness so we can perceive the previous notes as belonging together.

Before this repose is reached there is only expectation of what a string of tones might want to say; we need some sort of decline in the music (Abfall as Johann Philipp Kirnberger (1721-1783) calls it (Johann Philipp Kirnberger, Die Kunst des reinen Satzes in der Musik aus sicheren Grundsätzen hergeleitet und mit deutlichen Beyspielen erläutert, 2 vols (Berlin, 1776), II–I, 138.) that allows the ear enough rest to unite the previous notes in meaning–in other words: they create phrases.

The repose-moment is therefore a key-concept. Basically it is a feeling but at the same time it is a describable musical phenomenon that allows us to be more specific and, for lack of a better word, objective.

Next an example of an unbalanced composition. Joseph Riepel (1709-1782) criticises this minuet by his fictional student (‘not even worth a sniff of good pipe tobacco’): (Joseph Riepel, Anfangsgründe zur musicalischen Setzkunst: De Rhythmopoeïa oder von der Tactordnung, 7 vols (Regensburg and Vienna, 1752), I, 1.)

Figure 3: A unbalanced minuet, according to Joseph Riepel

A summary of his main objections:

(e) Only bars in even numbers are pleasant to the ear (there are 21).
(f) A minuet usually has 8 + 8 bars (instead of 8 + 13).
(g) The beginning, or theme, should have recognizable divisions in two or four bars (not clear enough).
(h) There should be a clear separation between bar 4 and 5; both need to be rhythmically differentiated.
(i) Do not use ‘dead’ notes (i.e. without movement as in bar 5), except at cadences.
(j) Use only a limited amount of faster scale-like passages.
(k) A minuet should mainly have alternating bars with ‘perfectly- and imperfectly’ moving notes (vollkommen- und unvolkommen erhebende Noten). This means bars with notes on (at least) every beat, should alternate with bars containing a minim (stretching two beats). (Riepel, Anfangsgründe zur musicalischen Setzkunst - Tactordnung, I, 5.)

Figure 4: Unvollkommen- and Vollkommen erhebende Noten, as illustrated by Joseph Riepel.

What emerges is an attempt at describing a sophisticated sense of proportion and balance. Though Riepel’s starting point is not primarily melodic, there is much in common with Koch. For example in the focus on clearly recognizable divisions as in (f) (g). But also in (j) where continuous fast notes could obscure these divisions, and (k) where the alternation of perfect and imperfect movement creates divisions of two bars. Riepel does not mention repose-moments but since ‘dead’ notes are only allowed at cadences (i), they appear at end of sections where you would expect something to come to rest. Likewise (m) where one could conjecture that if ascending means gaining energy, and descending losing energy, this would enforce the greatest repose to be neatly created at the end of the piece.

Having started the chapter with the assumption that bad examples could clarify Classical style, we can try to change Reicha’s nonsensical melody into something more Classical. Here is a my attempt:

Figure 5: (A) Reicha's non-melody, (B) my attempt at improvement

Bar 1-2 has been made rhythmically symmetrical to bar 3-4 (g); the symmetry makes the bars easily recognizable as 2 + 2, emphasised by a feeling of release (or slight repose) at the end of bar 2 and 4. This is why I based the symmetry on bar 3-4 rather than on bar 1-2.
The F# was changed to F to strengthen the key of C (a), alternatively bar 4 could be changed to start with a G, resolving the F# as a neighbour note, improving its melodic function.
I removed the first note of bar 5 to emphasize the division (h), and simplified the rhythm to not stand out from the rest.
The ‘dead’ notes in bar 6 and 7 have been removed, condensing the original nine bars into eight (e). Conversely a dead note was created in the final bar (i), now ending on the note C to better emphasize the key (a).

Most of the pitches have been kept in place but the structure is much clearer, and paradoxically the rhythmical adaptations have made it sound more melodious. It is not a great melody yet (that requires more than rules, as Reicha, Koch, and Riepel would acknowledge) but that is not the point, the process is important. I set out to get into a more Classical mindset and by playing with the music I think I have done that.

It is remarkable that both Koch and Riepel focus primarily on rhythmical aspects rather than pitch organisation, and they are far from unique; almost all authors, when discussing phrases, give examples on one stave, thus explicitly not involving harmony. As Reicha states, melody almost always indicates the harmony it requires, but harmony by no means indicates a true melody; if you start with harmony you will usually remain a médiocre mélodist. (Anton Reicha, Traité de haute composition musicale, Czerny edition, 4 vols (Vienna, 1824), II, 585.)

Mission statement

I will summarize and explore the theory relating to the repose-moments as I understand it from eighteenth-century treatises. This should enable us to create better grounding for choosing phrases. Furthermore, the length and rhythm of phrases, and the cadences will prove to be agents in appreciating the beauty in balance and order in Classical style.

But balance is not static: you can not create balance where there is no movement. Therefore the theory can also inspire us to discover different ways of shaping these phrases (i.e. phrasing). Punctuation, breathing, language and rhetorical constructions are among the agents to transmit life into this beautiful order. Phrases are there to be creative with.

I have sought to integrate performing practice with case studies that illustrate how phrase theory sits at the heart of every aspect of music making. Though the subject matter might at first suggest a quest to quantify what is ‘correct and historically justifiable’, my ultimate goal is exactly the opposite: I want to inspire musical performance through deep affinity with the source material of music itself.

Chapter 2 – Cadences and Punctuation

The musical phenomenon associated with the repose-moment is the cadence. Nowadays we tend to think of cadences as chord sequences but the concept as it emerges from the treatises is much broader: a cadence is any tone formula that creates the feeling of a complete rest point. (Koch, Musikalisches Lexikon, s.v. Cadenz, see Tonschluss.)

A cadence can therefore be just a melodic closing formula and the earliest description in English (according to the Oxford English Dictionary: ‘Cadence, N.’, OED Online <http://www.oed.com.access.authkb.kb.nl/view/Entry/25947> [accessed 19 March 2016]) by Thomas Morley in 1597 is essentially that: ‘A Cadence wee call that, when coming to a close, two notes are bound togither, and the following note descendeth.’ (Thomas Morley, Plaine and Easie Introduction on to Practicall Musicke (London, 1597), 73.) This is of course from an earlier musical style, nowadays we would probably call this a clausula rather than a cadence, but still the description is of a melodic- rather than an interval-based pattern.

Figure 6: (left) A cadence as illustrated by Thomas Morley, (right) a clausula

More than two hundred years later, some of Reicha’s examples of melodic cadences still look remarkably similar: (Reicha, Traité de haute composition musicale, II, 361.)

Figure 7: Melodic cadences by Anton Reicha

We could argue that cadential harmonies are implied in the melody but that does not diminish the point: a melodic formula, recognized as such, can signal the end (or coming to a rest) of the phrase. Compare for instance some of the trills traditionally added in Baroque music; in unaccompanied solo sonatas they literally signal the end of a phrase. (In older French treatises a trill is habitually called a ‘cadence’.)

Figure 8: Trills signalling cadences: G. P. Telemann, Fantasy for flute solo, TWV 40:9

Reicha makes the interesting observation that rhythm alone can create cadences. Hearing a drum play the following would not cause any problems in understanding the structure by hearing moments of repose. (Reicha, Traité de haute composition musicale, II, 363)

Figure 9: Cadences created by rhythm only, example by Anton Reicha

Structural function and punctuation

Almost all treatises compare music to language when structuring a composition: in language we also need repose-moments to understand a discourse. The most noticeable reposes would be reserved for the main paragraphs/sections, whereas the lesser ones will divide these further into sentences and clauses. Basically we are talking about punctuation.

Likewise in a good composition one would reserve the greater degrees of repose for bigger sections, and the lesser to divide these further. Musical phrases are then separated by colons or semicolons, phrase-parts by commas, and a group of associated phrases (i.e. a musical period) is terminated by a period (dot). (See chapter 6 for problems in comparing language and music.)

It follows that repose-moments need to be variable in strength with weaker cadences acting like commas, and stronger cadences marking important endings.

Hierarchies of strength

The amount of repose can be changed by varying the melodic, rhythmic, and harmonic proportions of the cadence. For each we could set up a sort of hierarchy of cadence-strength, though obviously a ‘feeling’ of repose can never be reduced to a rigid system.

Harmonic aspects

- The perfect cadence (Fig. 10) is strongest, it creates so much repose that it closes the piece or at least a very big section. For some, like Augustus Kollmann (1756-1829) (Augustus Frederic Christopher Kollmann, An Essay on Musical Harmony according to the Nature of That Science and the Principles of the Greatest Musical Authors (London, 1796), 57.) the added seventh is required to be called perfect, others are not so strict.

Figure 10: A perfect cadence by A.F.C.Kollmann

The full cadence is any other V-I that is not strong enough to close the piece, usually for melodic reasons.
A half cadence (ending on V) is one degree weaker and always needs to be followed by at least one other phrase.
Plagal (IV-I) and mediant (chords a third apart) relations are weaker still. In Classical style they cannot close a phrase, though they can punctuate phrase-members.
Deceptive (or interrupted) cadences create minimal repose, or rather deny repose, forcing a phrase to continue.
Inversion of chords makes cadences weaker than in root position.
Generally a leap in the bass makes the cadence feel stronger than when reached stepwise. Koch even differentiates between a leap downwards (stronger) and a leap upwards (weaker). (Heinrich Christoph Koch, Musikalisches Lexikon (Frankfurt am Main, 1802), s.v. Tonschluss.)

Melodic aspects

Melodic cadences are basically melodic formulas that end on the first, second, third, fifth, or seventh note of the scale–relating to an (implied) full- or half cadence.

The cadence is strongest if the melody ends in the octave of the chord, (a) and (b) in Fig. 11. Only this so-called octave-cadence can create a perfect cadence. A third (c) or fifth (d) in the melody makes the cadence progressively weaker and unfit for the perfect cadence.

Figure 11: Identical harmonic cadences in decreasing order of strength, example by Kollmann

Some consider the melodic cadence stronger when reached descending (b), others claims the opposite (a). Riepel wisely states that context should decide. (Riepel, Anfangsgründe zur musicalischen Setzkunst - Tactordnung, I, 14.) It is interesting to know the word ‘cadence’ is derived from the Italian cadere (to fall, to drop, to decline); also the German word for cadence, Schlussfall (literally ‘enddrop’), relates to this.
If the melody jumps into the endnote it is considered less strong than when reached stepwise (i.e. the opposite effect as with the bass).
Adding an appoggiatura increases the strength because the added tension leads to more repose in the subsequent release.

Rhythmical aspects

The cadence-point usually has a longer note value than the preceding notes as to create a feeling of landing somewhere. Longer values will create more repose.
There is often a pause after a cadence, notated or implied, to increase the repose of the cadence before. Without it, or immediately followed by moving notes, the amount of repose is reduced.
Ending in a single ‘dead’ note, what Riepel calls unbeweglich (motionless) makes for a stronger cadence than having a subdivision in several notes (beweglich). This clashes with the rule that appoggiaturas strengthen the cadence. For Riepel an appoggiatura is not really part of the melody: written as an ornament it qualifies as unbeweglich and therefore stronger than when written out in normal values. (Riepel, Anfangsgründe zur musicalischen Setzkunst - Tactordnung, I, 19.) This feels rather contrived, unless he implied a chord change in the beweglich bar.

Figure 12: Unbeweglich and beweglich, as illustrated by Joseph Riepel

When subdividing on the cadence note, a long-short division is usually considered stronger than short-long.
Rhythmical movement in the accompaniment takes away from the strength of the melodic cadence.
Context also plays a part. If cadences appear at regular intervals, it becomes predictable where you are supposed to feel repose. (See also phrase-rhythm in chapter 3.)

On the other hand, Beethoven's First Symphony opens with a perfect cadence, but since we are only at the beginning of a substantial piece it is not perceived as an ending. Or the other way around: Haydn’s ‘Joke’ quartet ends with a perfect cadence in bar 166, but the listener, teased by the many G.P.’s that Haydn added, does not perceive it as an ending any more–allowing for the punchline four bars later (see Fig. 13).

Figure 13: Joseph Haydn's 'Joke' quartet, op.33,2, final bars of the last movement. Edition Eulenburg, [Fine] added by me

Case study 1: W.A. Mozart, Minuet K.315a, 2

Figure 14: W.A.Mozart, Minuet K.315a, 2

I chose a minuet not only for practical reasons (short and simple) but also because many authors recommend the study of minuets. According to Riepel composing a minuet is no different from composing a concerto, aria or symphony. (Riepel, Anfangsgründe zur musicalischen Setzkunst - Tactordnung, I, 1)

We might not immediately consider bar 2 as having a cadence but melodically it conveys the right formula, helped by the left hand coming to rest, followed by a pause. It confirms with Riepel's rule for having a clear division at the beginning in either 4 or 2 + 2 bars. It is not a very strong cadence though; using the hierarchies we can increase the feeling of repose by, for instance, creating pauses in the melody (Fig. 15, I added the left hand notes to prevent an empty moment.)

Figure 15: Increasing repose by adding pauses

Conversely the cadence could be weakened by, for instance, creating more rhythmic ‘unrest’ (displacing the expected bass note in bar 2 to the second beat, and delaying the appoggiatura in the melody to an unimportant moment between beats), and by adding harmonic instability (see Fig. 16). Incidentally, the cadence in bar 4 is relatively stronger now by leaping into the final bass note.

Figure 16: Decreasing repose by adding rhythmic 'unrest' and harmonic instability

The cadence in bar 4 is stronger than in bar 2 anyway by virtue of being on the tonic, the slight harmonic instability created by the G# in bar 1 comes to rest here in the keynote, and the expectation of a cadence here (created by the symmetry of bar 1-2 and 3-4) increase the feeling of repose.

A cadence at bar 6 is carefully avoided: there is harmonically a IV, no rhythmic variation, and melodically no specific formula—the imitation between top and bass undermines the possibility of coming to rest in the same spot even further. In bar 8 follows the strongest cadence till now with an unbeweglich octave-cadence and a bass ‘falls’ on the tonic with a leap.

However, the cadence of bar 8 cannot be the strongest in the piece, that should rightfully be reserved for the end of the piece. Compared to bar 16, the cadence in bar 8 is weakened by the avoidance of a proper V-I; an F# in the melody (bar 7) would probably have caused too definite an ending.

Figure 17: Making the cadence too strong

Also context makes this bar 16 more powerful: by the absence of a strong cadence in the previous 8 bars (this could be the reason why the rhythm remains so active in bar 12), and not to forget the expected release after 8 + 8 bars (symmetry).

----

Taking the relatively simple cadence theory seriously, illustrates the beauty and ingenious structure of a simple piece like this. In tweaking cadences we are actually playing with musical punctuation. For Mozart’s minuet we could chart the punctuation as follows:

Illustration 18: Possible punctuation of Mozart's minuet

More interesting than the exact names and hierarchy is the fact that it is not a rigid system and leaves room for interpretation. Georg Simon Löhlein (1725-1781) gives an example of a punctuated minuet where he, as the composer(!), leaves you a choice to treat bar 1-4 as either a short period or a phrase (Fig. 19 at (a)). (Georg Simon Löhlein, Clavier-Schule oder kurze und gründliche Anweisung zur Melodie und Harmonie, 2nd edition (Leipzig and Züllichau, 1773), 181.)

Figure 19: Löhlein's punctuated minuet

The ambiguity is of course caused by the relatively strong cadence; had it been a weaker cadence on the dominant there would be no doubt it was a phrase instead of a short period:

Figure 20: Cadence in bar 4 weakened by going to the dominant

Likewise in Mozart’s minuet I perceive bar 14 as a weak cadence, but strong enough to create a comma, helped by the thematic similarity to bar 1-2. In performance I would support the comma coming across by making the most of the absence of a slur in bar 15, maybe exaggerating a staccato touch to set it apart from the previous bar. You could as well argue the cadence is so weak (or even non-existent) that the last 4 bars have to be taken together, the absence of a slur just being short-hand notation for an implied slur similar to bar 14–or one could even reason the slur in bar 14 to be mainly structural, indicating the third beat is not an upbeat like in bar 2.

This freedom does not imply that punctuation is any less essential or even arbitrary. Daniel Gottlob Türk (1750-1813) gives an example in his piano tutor to prove the point: Er verlor das Leben, nicht nur sein Vermögen as opposed to Er verlor das Leben nicht, nur sein Vermögen (‘He lost his life, not only his wealth’ as opposed to ‘He lost his life not, only his wealth’.). In this case the comma is literally a matter of life and death. Less life-threatening but ‘equally unclear, or rather false, will the performance of a musical thought become by incorrect punctuation.’ (Daniel Gottlob Türk, Klavierschule, oder Anweisung zum Klavierspielen für Lehrer und Lernende (Leipzig und Halle, 1789), 340.)

Löhlein brings up another interesting point. The phrase-member at Fig. 19 *** gets a special mention requiring continuation like a double dot (duo puncta), so basically it is a comma which functions like a colon. Therefore a comma can separate (like Fig. 19 *) or connect. It is tempting to think of punctuation as only creating separations but it also serves the other side of the coin: linking. As defined by McArthur, punctuation is ‘the practice in writing and print of using a set of marks to regulate texts and clarify their meanings, principally by separating or linking words, phrases, and clauses, and by indicating parentheses and asides’ (my italics). (McArthur, Tom, Concise Oxford Companion to the English Language (Oxford, 1998), s.v. punctuation.)

Three examples from his book:

Period - ‘There was no truth in the accusation. The other problem was why they had not been warned.’ (greatest separation, creating strong and distinctive statements.)
Colon - ‘There was no truth in the accusation: they rejected it utterly.’ (more connected, one thought leads to a conclusion.)
Semicolon - ‘There was no truth in the accusation; it was totally false.’ (continuity of thought by linking the two related statements.)

But punctuation not only clarifies structure (like phrases in music), it is also an agent in the delivery of these phrases (like phrasing in music), marking especially the pauses where breath should be taken. In one of the first textbooks on the English language, the Orthoepia Anglicana (1640), Simon Dains specifies ‘a small pause for the necessity of breathing’ for a comma, a double pause ‘yet very small’ for a semicolon, and double again for a colon. A period after which a new sentence starts gets double the pause of a colon, but a period that ends an ‘integrall part’ gets even four times the pause of a colon. He teaches his pupils to count in their reading, just as he was once taught to count by his ‘singing-Master’. ( Simon Daines, Margarete Rösler and Rudolf Brotanek, Simon Daines’Orthoepia anglicana (1640), Neudrucke Frühneuenglisher Grammatiken (Halle a.S., 1908), III, 70.)

Additionally punctuation can also be responsible for the more subdued delivery of a subordinate clause as opposed to a principal clause, the raising of the voice towards an exclamation mark, etc.

Chapter 3 – Phrase-length and Phrase-rhythm

Before continuing it is important to point out some exceptions where the theory does not apply. In random order:

Characteristic dances can have cadence-points that are not on the first beat (e.g. Polonaise), or rhythms that prevent complete repose (e.g. Scottish dances).
Recitatives have a form that is completely determined by speech rhythm, therefore they are irregular in respect of cadences, phrases and phrase-rhythm. Vocal forms tend to be less regular anyway because the meaning and grammar of the lyrics can not be interfered with.
Fugues and pieces in similar style are constructed by different principles where imitations and the continuous weaving of independent voices overrule the need for regular phrases.
Fantasies and other improvised music do not necessarily have bar lines, let alone regular phrases. But they need to be judged carefully: some fantasies are regularly structured although notated without bar lines, and free fantasies often contain perfectly regular melodic and thematic sections for contrast.

Phrase-length

Most authors agree that four bars is by far the most common, useful and agreeable length for a phrase. Counting bars seems easy but there are some issues.

Upbeats can cause confusion. In Figure 21 the upbeat takes more beats than left over in bar 4, but upbeats may be neglected when counting bars. (Heinrich Christoph Koch, Versuch einer Anleitung zur Komposition, 3 vols (Leipzig, 1787), II, 368. )

Figure 21: An upbeat makes the phrase look too long, example by Koch

In Figure 22 the opposite case: the bass proves the phrase starts before the melody. Since it is therefore not a real upbeat but rather a late entry, the first bar needs to be included in counting. (Heinrich Christoph Koch, Versuch einer Anleitung zur Komposition, 3 vols (Leipzig, 1787), II, 370.)

Figure 22: A late entry instead of an upbeat, example by Koch

A more surprising issue is the relation between metre and bar count. Theory dictates the cadence-point to fall on the first beat of the last bar of the phrase, counted in simple metre. This is a metre with either 2 or 3 for numerator. A 4/4 is therefore really two 2/4 measures notated in one bar, making Fig. 23 four bars long(!), as indicated by Koch. The proof is in the fact that the cadence-point is then on the first beat of the fourth (virtual) bar.

Figure 23: Cadence-point halfway the bar, example by Koch (vol.II, 372)

According to John Wall Callcott (1766-1821) ‘the Caesure [cadence-point], in ancient Music, most frequently occurs in the middle of the compound Maesure, and thus appears to a modern view irregular and incorrect’–this was written in 1806. (Callcott, A Musical Grammar, in Four Parts, First American edition, from the last London edition (Boston, 1810), para. 509.)
This old-fashioned way of notating we often find with Haydn who frequently has the cadence-point on the third beat.

Early Classical style tended to move away from the compound 4/4 towards simple 2/4 and 2/2, as stated by Johann Adolf Scheibe (1708-1776). (Johann Adolph Scheibe, Über die musikalischen Composition, im Schwickertschen Verlage (Leipzig, 1773), I, 205.) Later in the Classical period 4/4 became more popular again, but with a difference.

Figure 24: The opening bars of Mozart's Violin Concerto K.217 in 'new' 4/4

Fig. 24 is a ‘new’ type simple 4/4. The proof is the cadence-point at the beginning of bar 4. Koch would have argued that it should rightly have been written in (simple) 2/2 and blames sloppiness and even ignorance of composers and publishers. (Heinrich Christoph Koch, Versuch einer Anleitung zur Komposition, 3 vols (Leipzig, 1787), II, 295. ) But a quick survey of scores by Mozart and Beethoven shows the new 4/4 became standard notation. Almost all concertos by Mozart are in simple 4/4 and could just as well have been written in 2/2.

This may seem like a dry theoretical discussion but there are practical consequences. For instance in Mozart’s Quintet for piano and wind, K.452:

Figure 25: Mozart, Quintet for piano and wind, K.452, Bärenreiter NMA edition. The metre indication in bar 1 is C

Recognizing the cadence-point halfway the bar, we realize that there are two phrases of 4 (virtual) bars in 2/4 instead of one 4-bar phrase. Seeing it as two phrases will have you experience two moments of repose, probably leading to a more relaxed tempo, clearer punctuation and maybe a more explicit delivery than when you would try to pace it as one phrase, arching over the break halfway towards the last bar. Dividing it into two phrases does of course not deny that the whole passage belongs together, punctuation also serves to link after all, it only gives a different scope and direction.

It is important to recognize that Fig. 23 and 24 are two different types of Allegro even though the tempo marking and metre are identical. The old type (like K.452) is characterised by the cadence-point halfway through the bar, often with quicker harmony changes and a more active bass line, and is usually played about twice as slow as the new type. It means we can not take metre and tempo at face value without relating it to phrase-length.

We can only speculate why Mozart did not write it in 2/4. Maybe he felt the piece referenced an older style of music that warranted the more traditional notation. Maybe he felt that the many bar lines in 2/4 would visually contradict the broader scope of the movement, making it visually too busy and active. It is interesting that in his 18 piano sonatas Mozart uses a 2/4 only one time in a first movement but 8 times for a ‘finale’ movement. Maybe he wanted to indicate that the first and third (virtual 2/4) bar were the important bars, which could be melodically true but is harmonically less convincing. Though the question can not be answered it definitely has implications for the way each of us phrases and plays the music.

Phrase-rhythm

Having cadences at regular intervals creates a higher level of rhythm, the phrase-rhythm, that is especially important in Classical music. Phrase-rhythm does not have to be constant but it does need to be in balance.

The usual Classical phrase is a 4-bar phrase, or Vierer in German, which I will abbreviate to 4-er. Other lengths are not impossible but they are typically used as a variation rather than the basic unit for a piece. (Antonio Borghese (fl.1777) is the only exception I found, explaining phrases by means of 3-ers.) (Borghese, A New and General System of Music, 11 ) Even as late as 1885, Saint-Saëns still felt the need to clarify the use of 5-ers in the Scherzo from his Violin Sonata op.75.

Figure 26: 5-bar phrases indicated by Camille Saint-Saëns, Violin Sonata op.75, mov.2 (Manuscript)

Cadences appearing at regular intervals can become all too predictable (e.g. in Galant style) and Reicha argues strongly in favour of experimenting with different lengths because audiences have become bored by the current predictability of melodies. (Anton Reicha, Traité de haute composition musicale, Czerny edition, 4 vols (Vienna, 1824), II, 554.)

Johann Bernard Logier (1777-1846) illustrates how the ‘spirited (geistvoll) and powerful expression’ created in Mozart’s Serenade K.388 by phrases of 5 + 4 + 3 bars, would be sacrificed if kept in a regular 4 + 4 + 4. (Johann Bernard Logier, System der Musikwissenschaft und der praktischen Composition (Berlin, 1827), 296.)

Figure 27: Mozart's original irregular phrases

Figure 28: Phrases made regular by Logier.

Indeed, it influences expression in a powerful way, but what about Classical balance? When changing phrase-rhythm general advice seems to be:

Phrases with even numbered bars are still preferred over odd numbered bars.
Use odd numbered bars in pairs to balance them into an even number again.

The phrase-rhythm in the aria ‘Plaire au coeur de ce que j’aime’ by Giovanni Paisiello (1740-1816) is praised by Reicha for being ‘varied in an ingenious way’. (Reicha, Traité de haute composition musicale, II, 426) According to his analysis the first period is 26 bars (5 + 5, 2 + 2, 2 + 2, 4 + 2 + 2), the second 18 (5 + 5, 2 + 2, 4). A final phrase has been added to close the aria with more character (3 + 3 + 2 + 2). The phrase-rhythm is definitely not regular but still highly symmetrical, every odd number paired with a compagnion (Reicha) to maintain balance.

He opposes this by the aria ‘Ah! Que je fus bien inspirée’ from Dido by Niccolò Piccinni (1728-1800) with a first period of 29 bars (3 + 2 + 3 + 4 + 5 + 7 + 5) and a second of 40 bars (6 + 6 + 3 + 2 + 3 + 4 + 5 + 6 + 5). Reicha claims the, otherwise well-composed aria, has a shaky, undecided melody that is hard to remember, and he blames lack of symmetry in phrase-rhythm for making the individual phrases appear unconnected. (Reicha, Traité de haute composition musicale, II, 465.)

Phrase-rhythm is not just a theoretical concept. Especially Haydn is a master of irregular phrases and the phrase-rhythm adds spice and originality, but also more than that: the composer apparently needed to deviate from the norm to express something that could not be expressed within it. In order to bring that across in our playing we need to recognize the three techniques to vary the length of a phrase:

Invent a melody that sounds completely logical and complete at this different length (Koch calls this a ‘narrow’ phrase, eng).
Add more material to a standard 4-er (an ‘extended’ phrase, erweitert).
Combine 2 phrases into a longer one (a ‘compounded’ phrase, zusammengesetzt).

Extended phrases

Repetition

The most common way to extend a phrase is by repeating a part of it. Looking at Paisiello’s aria from Reicha’s analysis, we see the first 5-er was created by repeating two halve bars from the original 4-er:

Figure 29: Paisiello's phrase of 4 bars extended into 5 by repetition.

There are many possibilities: you can repeat 1 or 2 bars literally, or embellish the repeat, or even repeat them at another step of the key. Two examples by Koch:

Figure 31: Schubert Impromptu D.899,1 - a narrow 5-er or an extended 4-er?

Whatever you call it, this 5-er (which follows eleven regular 4-ers) is one of the most beautiful changes in phrase-rhythm in the literature.

Importantly, repeating part of a phrase is more than a mechanical process. Koch states the repeated motif must be ‘worthy’ of repeating: it needs to represent the expression of the phrase in a high degree, or highlight a new aspect of it. (Heinrich Christoph Koch, Versuch einer Anleitung zur Komposition, 3 vols (Leipzig, 1793), III, 155.)
The Paisiello is an excellent example: the repeated passage has the text de ce que j’aime (‘of the one I love’); a worthy subject to dwell upon. Musically a repetition can be highlighted by changing dynamics, adding embellishments, creating a new twist in the accompanying voices (harmony, texture), etc. Though Koch is writing for aspiring composers it is also inspiring advice for performers, really giving us a chance to bridge the gap between theory and practice.

Repetition does not fundamentally change the proportions of a phrase because the repetition can be omitted. With added repetition, Reicha claims, a phrase usually still sounds even numbered instead of the odd bar count it looks. The inserted bar ‘has something appealing because it adds originality and attraction’. (Anton Reicha, Traité de haute composition musicale, Czerny edition, 4 vols (Vienna, 1824), II, 457)

Expansion

By expansion one bar gets stretched into two like Koch illustrates:

Figure 32: (left) A narrow phrase, which is expanded into a 5-er (right). Example by Koch (III,171)

Expansions lend more weight to a phrase. Expansions at the beginning of phrases are typically found at the grand finale of a bigger composition, particularly oratorios. Expansions at the end of a phrase usually broaden the cadence to finish a section with more grandeur, often used in solo concertos. Reicha considers this retard de la cadence a pleasant way to break symmetry. In chamber music it is often notated with fermatas as in Fig. 33 (a), in concertos it is usually written out like (b). (Anton Reicha, Traité de haute composition musicale, Czerny edition, 4 vols (Vienna, 1824), II, 399.)

Figure 33: Retard de la cadence, examples by Reicha

The difference in notation demonstrates that in essence the phrase-rhythm has not been disrupted even though the bar count has. If the cadence is embellished we get a proper cadenza; this type of free cadenza again does not change the bar count.

Figure 34: Cadenza by Reicha over the previous example

It is important to realize that the cadenza is not a ‘stop’, the phrase must still finish and Reicha warns that a cadenza should never be so long as to make the audience forget the resolution of the cadence ‘that it rightfully so eagerly awaits.’

Other techniques

Repetition and expansion add melodic material ‘from the inside’. Koch lists several other techniques that add on (usually) less substantial material. Their less distinctive features make it unlikely to upset the balance in phrase-rhythm.

inserting/repeating a characteristic rhythmic pattern.
using a progression, which is a sequence of the same material on different steps of the scale.
using a passage which is basically adding unsubstantial passage-work. In this way you can easily create very long phrases.
with an interpolation you can insert non-related phrase-parts into the phrase. They must however add meaning to the existing phrase.
with an appendix (Koch: Anhang, Reicha: addition) you add material at the end of the phrase. An appendix can give more decisiveness to a cadence or, the opposite, move towards a cadence on a different chord. Often the appendix in its turn gets an appendix, which leads to another appendix, etc.; this can extend the material substantially. Or the appendix can be just an added bar in another instrument. as in this example by Reicha:

Figure 30: (A) Extended by 2-bar repeat on different step of the scale, (B) extended by repeating 1 bar embellished. Examples by Koch (II, 431 en III, 161)

Whereas the Paisiello example is an obvious repeat, the Koch examples are less easy to recognize. Sometimes it is hardly possible to decide whether a 4-er was extended or simply conceived as a narrow 5-er, like in Schubert's C minor Impromptu D.899,1. Is bar 3 a repetition? It can be left out after all.

Figure 35: Appendix by added bars in orchestra, example by Reicha.

Reicha: ‘If used well, the addition has such an appealing effect: it is like a hesitant lingering, almost taking breath to better comprehend what follows. Without the addition the melody in the example would be worn out and less appealing.’ (Reicha, Traité de haute composition musicale, II, 456.)

A prefix usually acts like in introductory statement and is therefore not really a part of the melody and unlikely to upset the balance in phrase-rhythm. Importantly, Koch acknowledges that some extensions cannot be described by a certain technique or reduced to a shorter original; they are simply created by the genius of the composer. (Heinrich Christoph Koch, Versuch einer Anleitung zur Komposition, 3 vols (Leipzig, 1793), III, 205.)

The Mozart game

Despite the previous remark, Koch treats extensions under the heading ‘About the art of connecting a melody according to mechanical rules’. This gave me the idea for an ‘app’ to play around with extensions. It is based on the first 2 phrases of Mozart's Musikalisches Würfelspiel (musical dice game) K.294d where a different minuet is ‘composed’ depending on how the dice roll.

I have taken the game one step further by having the dice add extensions according to Koch's methodology. The first aim was to test how these extensions can work within phrases. But it also illustrates Koch’s remarks about different genres requiring different kinds and amounts of extensions. Consequently by having the app add extensions, the minuet is taken away from its dance foundation and turned into something more akin to the opening of a sonatina. Besides clicking ‘auto extension’ it is also possible to experiment manually by clicking the check boxes. Of course the musical outcome of this mechanical experiment will necessarily be contrived but sometimes interesting examples appear!

Click on the picture below to go to the game

Figure 36: Illustration from talkclassical.com

Compounded phrases

Where extensions vary the phrase-rhythm by changing the length of a single phrase, compounded phrases create a different phrase-length by connecting 2 (or more) phrases in such a way they seem to be only one. Basically there are two ways: bar suppression and denying completion.

Bar suppression

When the chord on which the cadence lands is identical to the opening chord of the following phrase you can often leave out one bar. The German word Takterstickung (bar suffocation), and the term ‘interwoven’, (Callcott, A Musical Grammar, in Four Parts, First American edition, from the last London edition (Boston, 1810), para. 567.) are more graphic descriptions of the technique. Below two examples from Mozart’s violin concertos, each with a completely different result.

The orchestral introduction of the G major concerto, K.216 shows two phrases clearly set apart:

Figure 37: Mozart Violin Concerto no 3 in G, K.216, bar 17.

Similarly we would expect at bar 72:

Figure 38: Mozart Violin Concerto 3 in G, K.216, bar 72 without bar suppression (my reconstruction).

but instead the fourth bar is suppressed, creating a smooth connection.

FIgure 39: Mozart Violin Concerto no 3 in G, K.216, bar 72 with bar suppression (original)

This is made possible by the identical chord and leaving out the upbeats. Riepel advises to omit upbeats in bar suppression anyway. (Joseph Riepel, Anfangsgründe zur musicalischen Setzkunst: De Rhythmopoeïa oder von der Tactordnung, 7 vols (1752), I, 58.) According to theory, you are supposed to count the bar where the suppression happens twice, so what could look like a 7-er still counts as eight (4 + 4), therefore balance is maintained and phrase-rhythm does not feel disturbed. (Heinrich Christoph Koch, Versuch einer Anleitung zur Komposition, 3 vols (Leipzig, 1787), II, 454)

In the A major concerto, K.219 the orchestral introduction ends with the following appendix after which the solo violin enters:

Figure 40: Mozart Violin Concerto no 5 in A, K.219, bar 37 appendix

Mozart could have used a similar construction before the violin entrance in the development section:

Figure 41: Mozart Violin Concerto no 5 in A, K.219, bar 116 without bar suppression (my reconstruction)

instead he suppresses the bar before the entrance,

Figure 42: Mozart Violin Concerto no 5 in A, K.219, bar 116 with bar suppression (original)

creating almost a shock effect—no smooth transition but rather a plunging into. In this case the chords are different, taking the conventional bar suppression to a different level, but the technique is essentially the same.

Denying completion

Another way to compound phrases is to deny completion by taking away the feeling of repose of the cadence, hence forcing another phrase to be added on. The most familiar and powerful way is by harmony with (the surprisingly apt named) deceptive and interrupted cadences.

But it can also be done by melodic means. In Fig. 43 the top stave has two regular phrases of 4 bars, the first divided into two phrase-parts. On the bottom stave the melodic cadence in bar 4 has been weakened: loss of completion now turns it into an 8-bar phrase with three phrase-parts.

Figure 43: Top stave has two phrases of 4 bars, bottom stave one phrase of 8 bars by denying completion. Example by Koch, II, 459

Chapter 4 – Case study 2: Haydn Hob.XVI:16

Though many Classical compositions are in fact quite clear-cut structurally I also want to tackle a more challenging piece. The music of Beethoven and late Mozart can be very complex but generally has a more regular foundation than the music of early Haydn and C. P. E. Bach. As a test case I decided on Haydn’s Piano Sonata Hob.XVI:16. It is a quirky piece, maybe that is why its authorship is nowadays debated, and even the title is disputed, Sonata or Divertimento. Haydn or not, the piece seems to defy regularity. For a complete score of the first movement, see appendix 2.

Complete first movement by Jan Willem Nelleke - fortepiano (McNulty)

My original point of departure was to apply the discussed theory but I found that to be insufficient for an in-depth view. I will bring in elements that are not mentioned in treatises but are based on studying scores and noticing recurring patterns. Of the books used, William Caplin’s Classical Form deserves special mention. (William E. Caplin, Classical Form: A Theory of Formal Functions for the Instrumental Music of Haydn, Mozart, and Beethoven, New Ed edition (New York; Oxford, 2000).)

Bar 1-10

Figure 44: Haydn, Ho.XIV:16 - opening statement

The opening statement has two repose-moments (bar 5 and 10), therefore there are two phrases but are they really two 5-ers? Bar 5 is both the end of the first and the beginning of the second phrase, suggesting bar suppression from something like:

Figure 47: First phrase without extension

The phrase-rhythm is thus theoretically an even 4 + 6 though this feels contrived to me, and probably to the composer as well because bar 5-10 is so obviously in even bar numbers as if to compensate for the irregular beginning.

There is also another way of looking at this. Johann Friedrich Daube (1733-1797) uniquely does not even mention cadences in relation to phrases. (Johann Friedrich Daube, Anleitung zur Erfindung der Melodie und ihrer Fortsetzung, 2 vols (Vienna, 1797), I.) For him the melodic content shapes music into phrases (only 4-ers or 8-ers are allowed). In his examples he makes a fairly clear change every fourth bar by changing melody or texture without necessarily having the same regularity in cadences. I have found no specific source for using texture in structuring music but it often plays an important role. One of Mozart’s favourite tricks is to seemingly break phrase-rhythm by changing texture too early.

Figure 48: Mozart, Violin Concert in D, K.218, mov.1. Texture changes one bar early

We see something similar here in the Haydn where texture structures the music into 4 + 2 + 4 bars even though the cadences are at different places.

There is another reason the first phrase feels slightly aimless. Harmonically it is set up to lead to the dominant:

Figure 49: First phrase turned into a dominant-phrase

This is a perfectly regular phrase and it would be easy to add a complimentary phrase leading back to the tonic—a conventional bicolon again. Instead of letting the phrase go to its obvious destination, he is now forced to add material because going to the tonic takes one step extra. This longer road explains that we tend to find extensions more often in tonic phrases than in dominant phrases. Another solution would be to distribute this longer road differently by making the harmonies move earlier. Thus we often see, with parallel tonic and dominant phrases, that the tonic one moves towards the cadence earlier than the dominant one, possibly like:

Figure 51: Bar 5-6 can be cut out

But this would make it sound like an appendix rather than a new phrase, the similarity of the left hand rhythm emphasizes that even more. We could indeed read the 6 bars as 2 appendices (2 + 4 bars) to the first phrase but the strong cadential move into the sub-dominant seems to imply it is a phrase though. The interpolation of (the original) bar 5-6 is therefore functional: it favours a phrase-feeling by having the character of something new opening up, helped by the move away from triplets and left hand rhythm. The 2 phrases compliment each other: searching versus stabilizing—except for the unusual last cadence. The I6 undermines the expected perfect cadence and associated close of a period: a significant change in punctuation! It announces that what follows is not a repeat of the beginning as bar 11 might suggest, but rather a continuation like a semicolon, confirmed by a modulation to the dominant.

Bar 11-17

This third phrase gets extended significantly in bar 15-17, leading to the dominant of the new key using the repetition of a rhythmical motif (trill). Two bars would have been enough though, making it an even 4 + 2 bars:

Figure 52: Bar 14 with minimal extension

Instead he expands the cadence, adding weight to it by adding a bar. The use of chromatics contributes further to a feeling of weightiness, the quick succession almost causing an implied ritardando. From here he could have cut straight to bar 18 but he adds another extension in the form of a cadenza, further raising the level of expectation. The cadence looks free but is essentially structured and could have been written out as a normal 4-er with an expansion again at the cadence.

Figure 50: First phrase reshaped by change in harmonic rhythm

Again, he did not choose this. Personally I like the quirkiness of the extension, it has a more recognizable character than the bland 4-er, and that is the point of course. If Koch says a repetition needs to be ‘worthy’ and represent the expression of the phrase in a high degree, the other side of the coin is that there must be a need to enhance the character: as if the composer felt he had not been clear enough–in that sense extensions can be viewed as rhetorical devices.

The second phrase has been extended as well, the first 2 bars can be left out without any problem:

Suddenly we are in the realm of 3-ers, three in a row, the third one being a variation of the previous two with the melody in the bass and added figuration.

But if these are 3-ers, where are the cadences to indicate the end of a phrase? Regular 3-ers would have been like the example below,

The changes in phrase-rhythm are remarkable, maybe even a step too far, and need to be stabilized. In the next bars even numbers are back in abundance: a 4-er in the form of a progression, followed by a 2-er to confirm the key, and 2 added chords to confirm the end of the section. Since a progression by nature can keep continuing, the choice for 4 bars is probably deliberate to restore balance; it could just as easily have been 3 or 5 bars (see below).

Figure 45: Opening without bar suppression

The bar-suppression prevents an empty moment, but why actually? An empty moment can be of great effect, keeping tension and raising expectation. But here the rather bland material that follows cannot fulfil that expectation. We would either expect a new theme (but that would be too early in the piece) or a repeat. Repeating would create a conventional open-close, or question-answer structure. (I will refer to this common construction of two even parts as a bicolon, the background to this is explained in chapter 6).

Figure 46: Opening turned into a bicolon

The composer did not choose this obvious solution: maybe due to the ‘repetitive’ character of theme itself he did not want to repeat it immediately, maybe he felt that the stable and conventional construction of the bicolon contradicted the exploratory character of the theme, or the composition would develop too quickly in relation to the length he had in mind. But in essence this bicolon is the foundation of the first section.

The bar suppression he actually wrote allows for less significant material to follow. According to Koch we count the suppressed bar double, but that would make the irregular but symmetrical 5 + 5 into an unbalanced 5 + 6. However, the first phrase feels almost too long and could well be considered an extended 4-er, repeating bar 2 at an other step of the scale (see Fig. 47). This repetition also explains aforementioned repetitive character.

Figure 53: Cadenza written out as a phrase

After so much delay it would have been a damp squib if the cadence was simply resolved. Instead, he suppresses the resolution bar and plunges into completely different material. Indeed, after a build up like that we need such a big contrast that even a tempo change is warranted–a sort of alla breve.

Bar 18-31

Figure 54: Regular 3-ers with cadences

but as it stands there are no cadences, hence no repose-moments. Theoretically you can presuppose bar suppressions but that seems an artificial way of making the theory fit.

Figure 55: Bar 18-23 with hypothetical bar suppression.

Though, without going so far as bar suppression, it is an interesting point that some of the dominant tension is probably released into the next bar.

We can also decide that texture determines these to be 3-ers, or we could even declare these bars to be a string of appendices. Koch’s examples show series of appendices that are far longer than the phrases they prolong but here they are far too important thematically to be considered as such.

I would like to look at this section in a different way. Having introduced the bicolon before, I now want to add the tricolon (more about this in chapter 6). This is the almost archetypal Classical construction of the tripartite Bar form, or short-short-long. I find it peculiar that none of the treatises specifically recommends this construction or gives it a name, though many if not most of their examples have this shape. Bar 18-31 is fundamentally a tricolon which is clearer when simplified.

Figure 56: The tricolon at the heart of bar 18-31

The advantage of looking at it this way is that the length of the individual 3-ers is now defined in relation to a bigger shape rather than by their individual identity: essentially one phrase with three phrase-members. Simplifying it one step further, incorporating the supposed tempo change in notation, shows the perfectly regular foundation (Fig. 57). Two beats have been cut out to create 3-ers, with great musical impact, and the penultimate bar has again been expanded.

Figure 57: Tricolon in hypothetical regular bar structure

Bar 31-38

Figure 58: With possible 3-bar progression

Figure 59: With possible 5-bar progression

Bar 39-end

The next section is surprisingly regular. It starts with two times 2 bars, a sort of modulating sequence in the form of a bicolon. From bar 42 he could have cut to bar 47, instead he attached a 4-er: the descending sequence balancing out the previous ascending sequence. If we would embellish the fermata that would provide even an extra extension. Next (bar 47) a 5-er that includes an extension by a repeat (bar 49), and a series of appendices identical to the first section. The last section is identical to the second, only in the tonic this time.

With all these extensions I was not surprised that the first section counted a very odd 41 bars (I count the cadenza as written out) but the almost identical number of bars in each section (20 + 21 + 19 + 21) seems remarkable to me–with a little cadenza in bar 48 it could even be made into a symmetrical 20 + 21 + 21 + 20.

continue to Chapter 5 - Notation and Performance