From Phrase to Phrasing

A Classical Perspective

Jan Willem Nelleke

Chapter 5 – Notation and Performance

Whereas in language there is an age-old system of punctuation marks, in music we only have implied notation through cadences. But there have been instances where composers felt the need to indicate phrases explicitly, inventing their own symbols.

Symbols

François Couperin (1668-1733) kept a direct link with language, using a comma to mark the end of melodies or harmonic phrases. He explains ‘the separation should be almost imperceptible, but its absence would make persons with taste feel that something in the performance is missing—it is like the difference between those that read everything through and those who make stops at commas and periods.’ (Couperin, Troisième livre de pièces de clavecin, first edition (Paris), preface)

From Phrase to Phrasing (cont.)

Figure 60: Couperin: Le Carillon de Cithére from the Third book of harpsichord pieces, first edition (1724)

Michel Blavet (1700-1768), in his Flute Sonatas op.2 (1732), uses a small h (for haleine, breath) to indicate places to breathe. A breath mark is not the same as a phrase indication but as he writes in his preface: ‘I have often noticed with students that as a consequence of difficulties in catching one's breath, they mix up one phrase with another, or interrupt a melody that should pass in one breath. To prevent this confusion I have put the letter h on the spots where one should take a breath; above all in melodious pieces ... where all charm depends on the arrangement of phrases.’

Figure 61: Blavet: from La Vibray, op.2,2 (1732)

In works with an educational purpose I have found various symbols for breath marks like ꞌ or * or = or ; or ^.

In some composition treatises we find symbols indicating phrase structures. Riepel uses ■ for a tonic phrase-ending (Grundabsatz), and □ for all other phrase-endings (Änderungsabsatz). (Joseph Riepel, Anfangsgründe zur musicalischen Setzkunst: Grundregeln zur Tonordnung insgemein, 7 vols (1755), II, 41.) Koch copied Riepel’s □ as a general symbol for phrase endings, and added Δ to indicate phrase-parts. (Heinrich Christoph Koch, Versuch einer Anleitung zur Komposition, 3 vols (Leipzig, 1787), II, 361)

Türk is, as far as I know, the first to propose a dedicated new symbol to mark repose-moments in a composition. (Türk, Klavierschule, oder Anweisung zum Klavierspielen für Lehere und Lehrende, 342)

Figure 62: Türk: example from his Klavierschule

In works with an educational purpose I have found various symbols for breath marks like ꞌ or * or = or ; or ^.

He calls this // an Einschnitt (incision), adding to the confusion whether a term relates to the phrase-part itself (in German often called an Einschnitt) or to the point of separation. Türk places his symbol on the last note of the phrase and not between phrases, thus giving it the appearance of an articulation mark. He invites other composers to follow his lead and counters expected criticism: ‘to everyone who claims it is unnecessary to indicate phrases in music, I ask why we have introduced it in writing and still use it even in books aimed at learned people.’

But even if the composer goes as far as to indicate the phrases, there is still the performer who, as a unique musical individual, has to give shape to his or her understanding of the music. This personal aspect it too specific for the treatises; usually they only mention that good phrasing is essential and can be developed by listening to good performers, especially singers. Good advice as this may be, it is impossible to bring those performances back to life, and we have too little knowledge of the actual use of the voice to make present-day singing a reference.

On the other hand, authors have written about what you could call ‘skills’ related to phrasing. Türk, for instance, describes separating two phrases by lifting the finger off the key, causing ‘a little pause which should come in the time of the last note of a phrase. The end of the phrase will be even more noticeable if you lift the finger softly from the key, and mark the first tone of the next phrase a bit stronger again’. (Daniel Gottlob Türk, Klavierschule, oder Anweisung zum Klavierspielen für Lehrer und Lernende (Leipzig und Halle, 1789), 341) In Fig. 63 an example in notation:

Figure 63: Examples by Türk: (A) as notated, (B) as played, (C) added by me according to Türk’s text

I would like to classify these notations as:

(a) implicit - a notation where we have to deduce the phrase-ending from the strength of the cadence.
(b) implied - a notation where the phrase-ending is implied from how the end of a phrase is supposed to sound—softer and shortened.
(c) explicit - a notation that indicates where the phrase-ending is.

In essence the three notations are identical, except for the degree that the composer draws your attention to the phrase-ending. I want to stress that notation (a) does not prohibit any pause or tapering off of the phrase. The fact that the phrase-ending is not explicitly written out, or implied by notation, does not change the fact that it is needed there: the phrase structure demands it.

Türk further explains the amount of finger lifting is determined by context: separation between periods needs to be bigger than between phrases, and those require more again than between phrase-members. Phrases in the same character need less separation than phrases of a completely different character. (Daniel Gottlob Türk, Klavierschule, oder Anweisung zum Klavierspielen für Lehrer und Lernende (Leipzig und Halle, 1789), 342) Therefore implied notation is not to be taken literally; the crucial thing is that the final note is shortened and softer, not that it is exactly half the value, nor that it is only one written degree softer. In other words, punctuation is the guide, not the notation.

Since smaller repose-moments need a more delicate sensitivity to detect, the more ‘meticulous’ (Türk) composers show phrase-members in their beaming patterns:

Figure 64: Examples by Türk of phrase-members indicated by beaming patterns

For unbeamed note values like crotchets, minims, etc., he recommends a notation like (b) or (c):

Figure 65: Türk’s examples of implied notation (B and C) when beaming patterns are not possible (A)

As an implied notation (b) is clear, (c) is more confusing to modern eyes. Since Türk (like many Classical authors) does not differentiate between a dot and a wedge (‘both have the same meaning; though some use wedges to indicate a shorter Absetzen “setting apart” than by dots’) (Türk, Klavierschule, oder Anweisung zum Klavierspielen für Lehrer und Lernende, 353), he achieves the same as in (b)–namely, the dot/wedge shortens the note by approximately half its value. Using an articulation mark this way explains probably why his // sign is also placed on the note instead of in between. The notation definitely does not mean the notes with wedges should be accented as this would be counter to the idea of reaching a (however small) repose-moment. Türk explicitly warns against it: ‘many players have the false idea that a staccato note should always be abgestossen “pushed off” with a certain vehemence.’

Again we may conclude that music is phrased by the intrinsic nature of music, not by the outward appearance of notation; notation can clarify but does not create the music itself.

Emphasis

Apart from tapering off, Türk also mentioned starting a phrase with some emphasis. He specifies stronger emphasis at the beginning of periods, and progressively smaller emphasis for phrases and phrase-members. (Daniel Gottlob Türk, Klavierschule, oder Anweisung zum Klavierspielen für Lehere und Lehrende (Leipzig und Halle, 1789), 336) Fig. 66 demonstrates the degree of emphasis by the amount of pluses. (One ‘+’ would be the normal bar accent.)

Figure 66: Türk indicating different degrees of emphasis

The example shows emphasis should be made regardless of dynamics; it does of course not mean that similar emphasis in a different dynamic should be equally loud. The one important exception is the upbeat (marked ° ) of which the emphasis must be transferred to the following first beat. Despite the forte marking, the upbeat must not yet be played as strong as the first beat–the passage as a whole should be stronger than the previous one though.

In his discussion of flute tonguing, Johann George Tromlitz (1725-1805) hints at a similar tapering off and a more marked beginning. In a nutshell, a tone is articulated with ‘ta’ for a stronger and clearer attack, or ‘da’ and ‘ra’ for a softer and lighter one. To emphasize the separation of phrases he recommends ‘ra’ at the end of phrases, and ‘ta’ at the beginning. (Tromlitz, Ausführlicher und gründlicher Unterricht die Flöte zu spielen, 170)

Figure 67: At NB the end of a phrase is articulated with a soft 'ra', the beginning of the next with a clear 'ta'. Example by Tromlitz

With regard to upbeats it is interesting that Tromlitz still preferes the strong articulation ‘ta’, even if a weaker ‘ra’ is used on the first beat. (Johann George Tromlitz, Ausführlicher und gründlicher Unterricht die Flöte zu spielen (Leipzig, 1791), 167)

Shaping

Domenico Corri (1746-1825) makes the remarkable statement that ‘a phrase in music is like a sentence in language, with this difference, that one word will not form a sentence, but one note can form a phrase in music.’ (Corri, The Singer’s Preceptor, I, 65)

Figure 68: Examples with cresc.-dim. are marked 'Thus a Phrase', the ones with cresc. only are marked 'Thus not a Phrase' (from Corri)

For Corri a phrase is any passage were the voice ‘falls’ by a diminuendo, which, I suppose, could qualify as an other interpretation of cadence or cadere. Contrary to all the discussed theory where cadences are decisive, Corri, as a singer, focuses purely on the melodic shape and for him completion can be created by way of performing.

We don’t have to take his advice too literally; what counts is the idea of a diminuendo, rather the feeling of a decline than a dynamic effect. Looking at his second Solfeggio, Corri would have wanted a diminuendo at the end of the first phrase (bar 4) which makes sense with the appoggiatura there; but the second phrase looks contrary to his advice, with a crescendo towards the end of that phrase (bar 8). (Corri, The Singer’s Preceptor, I, 54)

Figure 69: Opening bars of the second Solfeggio by Corri

Corri does not offer an explanation but some tapering off is musically almost unavoidable in bar 8, and the typical minimalistic dynamic indications of the period do not contradict it: a diminuendo hairpin in bar 8 would have meant going back to piano which probably is too extreme an effect, and a diminuendo hairpin with a mf sign, or something similar, is simply contrary to Classical notation practice. A similar reasoning holds true for the melody. From a phrasing point of view, I think it is more important to create a convincing punctuation (connecting or separating phrase 2 and 3 in such a way that it fits the narrative of the music) than creating a dynamic contrast for its own effect.

Corri continues with another difference between language and music, ‘A sentence [in speech] is seldom or ever broke in the midst by taking breath, whereas, in a musical Phrase you are frequently compelled to do so’. And when taking breath ‘[one] should always contrive to do so by a dying or diminuendo of the Voice, because the Break will then be less perceived.’ (Domenico Corri, The Singer’s Preceptor, Or Corri’s Treatise on Vocal Music, 2 vols (London, 1810), I, 65) In the previous Solfeggio we can probably conclude bar 6 to have a small diminuendo before the breath, and bar 2 to have a diminuendo even if it would not have been indicated. (The diminuendo is probably explicitly notated here to give more effect to the deceptive cadence.)

Legato

Fig. 68 is also interesting because Corri uses an arched slur as symbol for a phrase; this at a time when slurs were not yet used to indicate phrases. It suggests a basic legato attitude, confirmed by the singing treatises of Johann Adam Hiller (1728-1804) (Johann Adam Hiller, Anweisung zum musikalisch-zierlichen Gesange (Leipzig, 1780) and Corri, but how that actually sounded will always be speculative by lack of recordings.

Probably legato in those days would have been more nuanced than nowadays where it seems to be almost on a par with vibrato: it is either ‘on’ or ‘off’. An example by Johann Peter Milchmeyer (1750-1813): (Johann Peter Milchmeyer, Die Wahre Art Das Pianoforte Zu Spielen (Dresden, 1797), 5)

Figure 70: Slurs within phrases, example by J. P. Milchmeyer

Three phrases are marked (1,2,3), each one subdivided by slurs. Milchmeyer demands absolute legato within these slurred subdivisions: breaking the slurs would be like ‘taking a breath in the middle of a word’. But after each subdivision (slur) the finger should leave the key.

Türk approaches the subject from the opposite side, in Fig. 71 it would be wrong to play (a) like (b). (Türk, Klavierschule, oder Anweisung zum Klavierspielen für Lehrer und Lernende, 340) Curiously he uses the same metaphor as Milchmeyer for a different phenomenon – breaking between the slurs would be like taking a breath in the middle of a word.

Figure 71: (a - top stave) Slurs and (b - bottom stave) how not to play them. Example from Türk's Clavierschule

But this does not mean we should sacrifice the individual slurs to an overall legato. Türk and Milchmeyer only seemingly contradict each other: Milchmeyer breaks the legato between slurs but he does so to create shaping within the phrase. It is a form of articulation rather than breaking the line; whereas Türk’s notation in Fig. 71(b) suggests breaking the line and therefore he warns against it.

As a consequence of this articulated way of playing, authentic keyboard fingerings give little conclusive evidence of phrasing. If a fingering allows for perfect legato, it does not mean there couldn’t have been a comma anyway; if a fingering does not allow legato, this could indicate articulation just as well as phrasing. We must not confuse phrasing with articulation.

Breathing

Singing, like wind playing, requires breathing which divides music into units and therefore has a direct relation to phrasing. Ideally phrase structure and breathing overlap but difficulties arise where one feels phrases are too long and without any obvious space to breathe. Wind players and singers often feel compelled to stretch their breath beyond normal human capacity in order not to break the line, but would composers really not have left enough time to breathe? That would be a beginner’s mistake with which they would be confronted every time they rehearsed their compositions.

It is my impression that historically musicians took breaths more often and aimed for a shorter way of phrasing. Consequently, it might be that nowadays in our fear of ‘breaking the line’ we are sacrificing clarity and expression.

I particularly like this quote from the ‘Méthode de Flûte du Conservatoire’ (1804) by Antoine Hugot and Johann Georg Wunderlich, ‘Generally you don’t acquire the art of good phrasing by playing in one breath for a long time, but by knowing how to breathe in time, and at the places indicated by the harmony in the phrases.’ (Antoine Hugot and Wunderlich, Méthode de Flûte du Conservatoire (Paris, 1804), 15) Also Tromlitz warns never to play until the end of your breath; lack of power will make your performance sound frightened whereas if you keep adding breath at suitable places, it will not impair your delivery. (Johann George Tromlitz, Ausführlicher und gründlicher Unterricht die Flöte zu spielen (Leipzig, 1791), 330)

So, breathing within a phrase is allowed, sometimes recommended even. But it still creates divisions in a phrase, so what are the most logical places to breathe? Hiller (Johann Adam Hiller, Anweisung zum musikalisch-zierlichen Gesange (Leipzig, 1780), 14–22) and Tromlitz (Tromlitz, Ausführlicher und gründlicher Unterricht die Flöte zu spielen, chap. 13) are the ones who discuss the issue most methodically and I summarize their advice:

Regarding lyrics: you cannot breathe in the middle of words or at grammatically illogical places. You are always allowed a breath at a punctuation mark or at the end of a (poetic) verse line. In long melismas a breath is also allowed though, even if this means breaking a word.
An appoggiatura, either abbreviated or written in full, is never separated from its main note, nor are two slurred notes ever separated.
You can always take a breath at a rest, or at an implied rest like before a syncope.

Figure 72: Possible breaths (after note with ') at syncopes, example by Hiller

You can breathe after a longer note with a tie. In faster tempo you can even leave the tied note out as Tromlitz demonstrates (Fig. 73 lower stave).

Figure 73: Breathing after tied notes in slow (top) and fast tempo (bottom), a (slightly adapted) example by Tromlitz

Since a dotted note is essentially shorthand notation for a tied note, you can also breathe after dotted notes, provided only part of the dot is cut out.

Figure 74: Dotted notes (A) notated as ties (B) and possible breathing at (C), example by Hiller.

In extreme cases Hiller suggests the following adaptation:

Figure 75: Splitting a tied note for breathing, example by Hiller

As a rule of thumb you don’t take a breath before the bar line, except before a very long note or before a long passage. In both cases one should not hesitate to breathe in the middle of a word. (The breath in the next examples is again indicated by a stroke on the note.)

Figure 76: Breath (after note with ') at the barline before a long note (a), and before a long passage (b). Examples by Hiller

Even more interesting is the next example where a word is broken twice.

Figure 77: Breathing before barline and fermata, interrupting a word twice. Example by Hiller. (V - I indication added by me)

According to Hiller one should have no qualms breathing before the long note, thereby breaking the word and splitting two notes that ‘should not be broken for harmonic reasons’ (i.e. cadence). He recommends taking another breath before the fermata since it requires embellishment.

In extreme long passages neither Hiller, nor Tromlitz are against leaving out notes.

Figure 78: Leaving out notes for breathing, example by Hiller

For the example in Fig. 78 Hiller suggests leaving out either the -notes (version 1), or the ^-notes (version 2), though preferably not all the omissions of a version. Tromlitz has a similar example that I have abridged in Fig. 79 to show taking quick breaths (at comma), leaving out a note (at * ), or adjusting rhythm to create breathing space (in ossia).

Figure 79: Leaving out notes for breathing, (abridged) example by Tromlitz

Hugot und Wunderlich specifially mention the possibility to suppress a repeated note for breathing. (Hugot and Wunderlich, Méthode de Flûte du Conservatoire, 18)

Figure 80: Suppressing a repeated note to breath, example by Hugot-Wunderlich.

In these virtuoso passages, most examples appear to favour breathing after the first note in the bar which seems a safe bet, not inadvertently breaking a cadence.

I want to stress that neither Hiller nor Tromlitz qualify these adaptations ‘for beginners only’; if your breath is not long enough you simply create places to breathe. I am not denying the need for good breath control but if I hear recordings where performers somehow manage these endless runs without breathing I, as a listener, feel out of breath. Maybe the aim was to be impressive, but I certainly don’t think it is expressive. Every human being can, from birth, relate to the natural span of breath; to strain that relationship somehow makes me feel uncomfortable. It is also in my opinion alien to the Classical period with its emphasis on a natural and language based style.

Case study 3: ‘An thou were mine ain thing’

This Scottish song appears in both Corri’s Perceptor and Gunn’s cello tutor, and may serve as an example for shorter phrases than we are used to.

Figure 81: From Corri, The Singer’s Preceptor, vol. III, 98 (National Airs).

Figure 82: From Gunn, 40 Favorite Scotch Airs for Violin, Flute, or Cello

Corri uses two symbols: a * for a breath equal to a comma, and an asterisk with a dash on top for an imperceptible breath (i.e. not related to punctuation). (Domenico Corri, The Singer’s Preceptor, Or Corri’s Treatise on Vocal Music, 2 vols (London, 1810), I) There is no textual reason to breath in the first bar (ain means ‘own’), nor would anyone run out of breath there, even in slow tempo. Also with Gunn there is no technical reason for the cello (sounding an octave lower) to break the phrase here. It does add to the expression though, specially in combination with the slurred notes before. It could be that this was a recognized, traditional phrasing for this song (both authors had a relation to Scotland). But the Austrian composer Joseph Haydn would not have know that. He set the song twice, for Thompson and Smythe, and in both settings the broken beams in the instrumental parts suggest a similar phrasing (the breaking of the beam in the voice line is conventional notation to clarify the placing of syllables).

Figure 83: An thou wert mine ain thing, as set by Haydn for Smythe, Henle edition, vol.XXXII,5 no 413. Top to bottom: violin, voice, piano, and cello. Tempo: Larghetto amoroso

In Thompson the violin does not have a similar beam break but the string crossing would cause a little interruption any way, moreover the sudden unison between violin and voice creates a new colour, setting it apart from the previous phrase-part. It all adds extra expression but in the recordings I found, no singer took a breath. (Jamie MacDougall with the Haydn Trio Eisenstadt, (2) Laura Skuce with the Chamber Music Society of Lincoln Center)

Figure 84: An thou wert mine ain thing, as set by Haydn for Thompson. Henle edition, vol.XXXII,3 no 168. Top to bottom: violin, voice, piano, and cello. Tempo: Larghetto

Chapter 6 – A not so Classical view

The previous chapters contain the research as I envisaged it at the beginning of this project. But gradually I have come to recognize the need to address topics not mentioned in the sources. With the help of modern authors I would like to add some observations that made me look at phrases in a different way.

Language in music, music in language

Initially I had thought to use the parallel between language and music as a stepping stone for this paper, till someone alerted me to T. W. Adorno’s essay ‘Music and Language: A Fragment’. (Theodor Adorno, ‘Music and Language: A Fragment’, in Quasi Una Fantasia, Essays on Modern Music, trans. by Rodney Livingstone (London, New York, 1956)) The first paragraph already states ‘Anyone who takes music for a language will be misled’ and it made me give up my original design. I could of course sidestep the issue by arguing that the sources for my research take the relation between music and language as axiomatic, but that feels too easy. (Koch, Versuch einer Anleitung zur Komposition, II, 342)

Music and language are not identical but resemble each other. Both are ‘temporal sequences of articulated sounds which are more than just sounds’ (Adorno). These sounds refer to something, they express, they communicate. But language references things outside the language itself. Words are coded information for something else. Language is (after Helmuth Plessner) double-layered: the layer of sound can be separated from the layer of meaning, that is why one language can be translated into an other. Music only references itself, meaningful units are created anew in every composition. It is therefore singe-layered; it still has meaning of course but the meaning is of a different order–‘you could not even order a pizza with it’. (Andreas Luckner, ‘Andreas Luckner: Musik - Sprache - Rhythmus’, 2007, 42 <http://www.links.linse.uni-due.de/sprache-und-weitere-zeichenmodalitaeten/andreas-lucknermusik-sprache-rhythmus.html> [accessed 20 January 2017])

This interferes with the traditional concept of punctuation. Simplified: punctuation in language uses meaning to create units, punctuation in music uses units to create meaning. Cadences operate on the level of music to indicate punctuation, punctuation in a text acts on a higher level than the words.

But this fundamental difference does not necessarily prohibit further discussion. Language conveys more than just information. In poetry, for instance, we find an emphasis on the ‘musical’ qualities of language: sound patterns, rhythm and a highly formalised structure in verse lines. These can rightly be called musical because they do not carry ‘coded’ meaning and their frame of reference stays within the single layer of sound. Rhyme creates a structure independent from grammar or meaning. Admittedly this is only a minor part of punctuation (we cannot determine the amount of closure) but when Riepel states that cadences of phrase-ends ‘rhyme’ with each other, he has an interesting point. (Joseph Riepel, Anfangsgründe zur musicalischen Setzkunst: Grundregeln zur Tonordnung insgemein, 7 vols (1755), II, 43)

Also syntax differs fundamentally between language and music. According to Manfred Bierwisch: musical syntax is time dependent, and language syntax is dependent on conventions. (Manfred Bierwisch, ‘Musik und Sprache : überlegungen zu ihrer Struktur und Funktionsweise’, 74 <http://publikationen.ub.uni-frankfurt.de/frontdoor/index/index/docId/12535> [accessed 20 January 2017]) Therefore repeating something in speech does not really impart more information; in music repeating a motif creates the musical form itself. Again this only holds true if language would convey only information. In poetry rhythm creates patterns in the metrical feet, length of phrases, caesuras and enjambments; like in music these patterns give shape to time and create their own ‘meaning’. Many rhetorical figures use repetition to make the message come across more strongly; in that sense repetition does create more information about character, build-up, tension, release, etc.

Moreover, in a highly stylised tonal music like the Classical, chords start acting like recurring ciphers, reappearing in identical functions, sequences, and stock melodic figures. (Adorno, ‘Music and Language: A Fragment’, 2) Tonality can at least claim a form of conventional syntax.

A recent study by Kunert et al. (2016) demonstrates that the grammatical structure of sentences (as opposed to their meaning) and harmonic sequences (as a determinant of tonality and musical phrase) are processed by the same part of the brain–thus proving that, specifically, syntax processing in language and music originate in the same part of the brain.. (Richard Kunert, Roel M. Willems and Peter Hagoort, ‘Language Influences Music Harmony Perception: Effects of Shared Syntactic Integration Resources beyond Attention’, Royal Society Open Science, 3/2 (2016), 150685)

Bicolon, tricolon

Without denying the fundamental difference between language and music, there are still good reasons to make and use parallels between language and music, especially in Classical style. Rhetorical figures are a case in point, like the bicolon and tricolon.

The bi- and tricolon are constructed from two or three parts that are more or less equal in structure, length, and rhythm. In short form they are popular because they are catchy:
‘I came; I saw; I conquered’ (Julius Ceasar, tricolon)

In more elaborate form they are successful by emphasizing a point in a memorable and pithy way:
‘Ask not what your country can do for you; ask what you can do for your country.’ (John F. Kennedy, bicolon)

These powerful devices can work in language and music alike because they have to do with style, order and delivery rather than content; they are mainly based on correspondence in length, sound/rhyme, and rhythm.

A musical equivalent of the bicolon would be a bipartite structure like question-answer, or open-close; of the tricolon it would be a tripartite structure like the bar-form. I will use bi- and tricolon for these musical structures as well.

Structuring in twos and threes is of course nothing new, but why are these figures so powerful? As rhetorical figures they are designed to have an effect. The symmetrical structure of the bicolon makes the second part easier to process by the created expectation. As Thomas S. Kane puts it, ‘Balance and parallelism do not communicate meaning by themselves, but balanced and parallel constructions do reinforce and enrich meaning’. (Thomas S. Kane, The New Oxford Guide to Writing (1994)) Dividing things into two is a powerful device because the brain will try to find a connection, a simile, an opposite (basically any pattern) because two suggests a relationship.

If two sets up a pattern, then a third element can confirm or break that direction. The tricolon is basically a list, a simplification of complexity; opposed to a list of four or more elements, three has a sense of completeness. Churchill said he had ‘nothing to offer but blood, toil, tears and sweat’ but four doesn’t work and everybody remembers the line as in the tricolon ‘blood, sweat and tears’. (Mark Forsyth, The Elements of Eloquence, Icon Books Ltd, 2013, 87) As Mark Forsyth demonstrates, tricolons often sound especially good when the third element is longer than the preceding two, like in ‘Life, Liberty, and the Pursuit of Happiness’. (Mark Forsyth, The Elements of Eloquence, Icon Books Ltd, 2013, 85) You could argue the specific order in meaning makes the last element more noticeable (from general to specific (Danyal Freeman, ‘Tricolon’, Vernacular Discourse <http://www.vernaculardiscourse.com/tricolon.html> [accessed 11 December 2016])) but for the musical ear the third element stands out because it provides a place of landing for the direction the previous elements have taken.

Bi- and tricolons create shapes for phrasing as well. If you speak a bicolon aloud it has a natural up-down intonation pattern. The tricolon has a natural up-up-down pattern with the ups raising the expectation that the down confirms. We are back where we started: bicolons, which through parallelism confirm expectation; and tricolons, by reaching completion of a list, create repose-moments.

In a way the bicolon stands at the basis of the antecedent-consequent form. Adding a contrasting idea to each ‘colon’ brings us close to a modern definition of period, as found with Schoenberg and others. (I write modern definition because like the 1 + 1 + 2 construction it is not really described as such in the sources though obviously present in a lot of Classical music.)

Figure 85: Period after Caplin: Mozart, Eine kleine Nachtmusik, K.525, mov. 2

This example by William E. Caplin. (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), 12) is like a bicolon in which each part has been qualified like in the famous:
Morning has broken, like the first morning.
Blackbird has spoken, like the first bird.
(From the song ‘Morning is broken’, lyrics by Eleanor Farjeon, made popular by Cat Stevens)

Balance and parallelism give bicolons a more static nature; in music they are often also perceived like that.

Similarly the tricolon can be said to be at heart of the modern sentence. The archetypical example is by Beethoven, Piano Sonata op.2,1, mov.1: (Caplin, Classical Form, 10)

Figure 86: Sentence after Caplin: Beethoven, Piano Sonata op.2,1, mov.1

Two shorter statements culminating in a longer (and stronger) third one make it a tricolon, in this case the third part is a tricolon in itself (added indication in Caplin’s example by me). Tricolons have a more developing nature since they are based on a list reaching completion, and in music they are perceived likewise.

Caplin makes many more interesting observations and goes on delving into hybrid types and exceptions. Personally I find the bi- and tricolon sometimes more helpful in their fundamental simplicity because they feel related to the way I hear and play music; as a string of thoughts that leads to something bigger (additive), rather than a bigger construction that gets divided in smaller parts (divisive). Although there is of course the bigger idea we want to get across, the art of rhetoric is in shaping our phrases in such a way that, step-by-step, we convince our audience. In that sense it adds meaning with every phrase and that idea appeals to me as a performer.

Development in a phrase

If repose-moments determine the end of a phrase, it implies that, when listening to music we only construct in hindsight how a phrase goes. Or as A. Luckner puts it, ‘Musical form constitutes itself in essence only in the fading away of sound, it can only exist where and if, the anticipation of “Gestalt” finishes.’ (Andreas Luckner, ‘Zur Philosophie des Rhythmus (Abstract)’, Lebenswelt und Wissenschaft <http://www.dgphil2008.de/programm/sektionen/abstract/luckner.html> [accessed 20 January 2017]) Anticipation is the keyword: if we can shape anticipation, we can convey a sense of order even before a phrase actually finishes. Phrase rhythm, bicolon, and tricolon are clear examples of that, but we also find it in the development in and over phrases.

I am stating the obvious when I say that Classical phrases tend to develop towards their ending, but why is that? We can often recognize a pattern where a phrase starts by an exposition of thematic material, followed by an intensification that comes to rest in the cadence.

Figure 87: Schematic development of a phrase

In it simplest form this would be 2 bars of melodic character, a change in bar 3 with rhythmical- and harmonic drive that releases itself in the cadence of bar 4. But we also find this outline over groups of phrases. Through this intensification we not only feel a bigger release afterwards, it also signals the approach of an ending and gives us a sense of where we are in the development of a phrase or group of phrases. Recognizing what contributes to this development will probably make for better phrasing and an increased sensitivity for relations between phrases.

One powerful agent is fragmentation (after Caplin). He defines it as the ‘process of shortening the units’. (Caplin, Classical Form, 41) It is very obviously present in the previous Beethoven sentence:

Figure 88: Fragmentation in Beethoven's sentence

It creates a feeling of added movement, development, as if increasing the drive towards the end—and it can consequently be expressed in phrasing. The tricolon within a tricolon is therefore an often found construction: intensification within expectation.

Fragmentation does not necessarily require taking a fragment of the preceding melody, as demonstrated in the previous period by Mozart:

Figure 89: Fragmentation in Mozart's period

Often slurs are carefully indicated to confirm this principle. For instance, without the first slur (marked *) there would be an upbeat feeling for the G. I think Mozart wanted to prevent precisely that, making the first phrase an undeniable long unit so that the fragmentation afterwards could work stronger.

Similarly the notation of rests is often more precise than expected. In the previous example the notated semiquaver rests in the final bar are superfluous because two slurred notes would be played like that anyway; but it does force attention to the shortening of units.

Usually harmonic activity increases during the intensification. An example from Mozart’s Horn Concerto, K.447, mov.2.

Figure 90: Opening of the second movement of Mozart's Horn Concerto, K.447

It is a bicolon in two phrases with an obvious increase in the number of chord changes towards the end of each phrase. But in the bigger picture the second phrase uses stronger harmonic connections than the first, recognizable by the jumps in the bass. This helps to create the expectation of the stronger (perfect) cadence in the tonic phrase as opposed to the half cadence in the preceding dominant phrase. In Mozart’s keyboard parts the left hand is usually carefully worked out to support this principle, where at the beginning of phrases bass notes are often notated short and with rests (i.e. lighter), changing into full length notes (i.e. heavier) at the stronger harmonic changes at the end of the phrase. Unfortunately many pianists blur this with the pedal.

And what about the beginning of a phrase? Koch uses again a parallel with language that I find intriguing. He compares a phrase to a subject (bars 1-2) and predicate (bar 3-4), in the sense that the predicate gives direction and destination to the subject. (Heinrich Christoph Koch, Versuch einer Anleitung zur Komposition, 3 vols (Leipzig, 1787), II, 352) A subject can thus have many different predicates, and if an extension is added it further defines the predicate.

Figure 91: A subject with 3 different predicates, the top one extended to further define the predicate. Examples by Koch

This is of course not literally true but it indicates a flexible interaction within phrases that gives life to the musical structure. It confirms the general observation that, in Classical phrases, melody and character/expression often determine the beginning, to be taken over by a more rhythmic and harmonic drive towards the end. Maybe it can even be argued that a phrase is initially lead by melodic interest (content) and towards the end more by harmonic interest (syntax).

Without wanting to turn this into a template it is worth noticing that this notion is often supported by the sub-dominant which frequently only appears at the end of a phrase—almost acting as a pivot point after which the phrase is forced to tip over towards a cadence.

Awareness of above mentioned principles have made an important difference in phrasing for me.

Chapter 7 – The Performer as Composer

Apart from influencing phrasing, phrase theory can also be helpful in situations where the performer gets close to being a composer himself. Some real-life examples.

Figured bass

Figured bass practise was still very much alive in Classical repertoire (specially in songs) and phrase theory can be helpful. For instance, top notes are influential in determining the strength of a cadence and hence punctuation; this is particularly important when playing with lower instruments and voices where the right hand accompaniment is forced to play the highest note of the chord, on top of the melody. More in general, awareness of punctuation can be a guide towards filling in ‘gaps’ in the melody, and awareness of the development in a phrase can be supported by pacing the activity of the right hand.

Ornamentation

Koch and Reicha recommend never to have two identical cadences following each other, to prevent uniformity and lack of clarity in punctuation. If forced to repeat the same harmonic cadence it must be at least melodically and/or rhythmically different, as demonstrated by Koch in Fig. 92. (Heinrich Christoph Koch, Versuch einer Anleitung zur Komposition, 3 vols (Leipzig, 1793), III, 67)

Figure 92: Two different phrase endings on identical harmony, example by Koch (abridged)

The point I want to make is that, when embellishing, we should take care not to accidentally make endings too similar, as I did on purpose in Fig. 93.

Figure 93: Koch's previous example, badly ornamented by me, creating too similar phrase endings

Also Reicha warns that in embellishing, phrase-rhythm and cadences should never be changed. (Anton Reicha, Traité de haute composition musicale, Czerny edition, 4 vols (Vienna, 1824), II, 497) In his examples of lavishly decorated arias (copied down from famous singers) the end of a phrase is almost never changed. Occasionally a conduit (lead in) is added to the next phrase but it always leaves a pause for punctuation.

Figure 94: Original version (top), and ornamented with a lead in (bottom), example by Reicha

The only place where a phrase ending gets significantly embellished and changed is at a point d’orgue (fermata) but by then we have crossed into the realm of the cadenza.

Case study 4: Creating the introduction to a song

Many Classical songs have no piano introduction, forcing the singer to pick the first note out of the air. Since this was an age where performers routinely improvised preludes to introduce a composition (Hummel (Johann Nepomuk Hummel, Ausführliche theoretisch-practisch Anweisung zum Piano-Forte-Spiel, second edition (Vienna, 1827), 465)), it seems highly unlikely they would allow for such an abrupt start of a song. In church services this is still a regular practice.

Nowadays the postlude is often used as a prelude in order to stay as close as possible to the original composition. That doesn’t always work well. In the early Schubert song ‘An den Mond’, D.259 (1815) it would create:

Figure 95: Schubert's An den Mond, D.259, using the postlude as introduction

Though I’ve heard it performed like this, some things don’t sit well. (1) The first chord is dissonant, which is not impossible but usually reserved for a special effect. (2) The dense harmonic rhythm suggests an ending rather than an opening, melodically and harmonically it comes to a close. Also it would be unlikely for a composer to give away the most special chords at the beginning of a piece. (3) The construction of the phrase with its one bar phrase-members and tricolon suggests fragmentation (and therefore a continuation) though there is nothing yet ‘to fragment’ or to continue; if we look at the song itself it shows the postlude is indeed the continuation of a phrase (bar 9-12) rather than an appendix.

The usual written-out preludes in Classical songs are either original inventions, something musically neutral, or a combination of the opening and final measures of the song. Original inventions fall outside the scope of this paper but a neutral opening could be something like:

Figure 96: Schubert's An den Mond, two possible neutral introductions

Both are possible but quite boring. The second example needs two bars to keep an even bar count, in the first example the chord doesn’t really participate in the song yet and therefore one bar can be enough.

Combining the opening and end measures of the song would produce something like this:

Figure 98: Schubert's An den Mond, combining opening and final bars - version 2

Case study 5: Writing cadenzas

To cover this topic extensively would fall outside the scope of this paper. My aim here is to to demonstrate the usefulness of phrase theory by means of a simple template I made. It is definitely not a historical model or a manual to create cadenzas, though it can serve as a foundation to elaborate on.

Rather than trying to string melodic ideas from a concerto together, I would suggest to start with a basic lay-out of phrases: an introductory phrase, a melodic phrase and a concluding phrase. Though historically you are not obliged to use themes from the concerto, it is often easier than inventing something new. But I would advise using the material in a very free way, just to trigger ideas and stay in style. Being literal normally does not work because the original material is made to fit the development of the piece, whereas the cadenza has its own development.

It makes sense to start at the end because it is fixed: always I64-V-I. A basic phrase would suffice, four bars with some harmonic acceleration towards the end:

Figure 97: Schubert's An den Mond, combining opening and final bars - version 1

Bar 3-4 are melodically rather wild compared to 1-2. This often happens: by the nature of things opening measures are usually simple whereas the end measures conclude a certain development. I would therefore reduce the energy of the second part by limiting the compass of the melody line and simplifying the rhythm:

Figure 99: Outline for a basic final phrase of a cadenza

We can add some chords in between to increase harmonic rhythm, thereby making clear we reach the conclusion of several phrases. Expanding the cadence would also be appropriate:

Figure 100: Outline for a more elaborate final phrase of a cadenza

You would usually fill this with scales and passage-work for a virtuoso finish. To illustrate I will use Mozart’s Flute Concerto no 1 in G, K.313 as an example:

Figure 101: Final phrase filled in

We can easily lengthen the cadenza by putting a melodic phrase before it, pinching some melodic intervals from a theme and turning them into a bicolon. Harmonically I-V-V-I or I-V is enough, with a bridge to the (already composed) next phrase.

Figure 102: Melodic phrase filled in

Now we still need an introductory phrase. Starting with an upbeat is always a good way to get going; using scales and chords on alternating I and V would be enough to ‘start exploring’. Finishing on the dominant can properly introduce our melodic phrase:

Figure 103: Introductory phrase filled in

The complete cadenza as played by Philippa Davies:

Of course this needs to be polished, and more phrases can be added or extended, but at least a proper framework is in place that will yield results quicker than stringing one’s favourite melodic bits together and hope for the best.

Appendix 1 – Terminology

Appendix 2 – Sonata (Divertimento?) in Eb Major, Hob XVI:16

My typesetting of the score is a composite based on several editions. All editorial suggestions have been removed and any minor contradictions solved by my judgement, none which change the structure though.

Appendix 3 – Bibliography

Aarts, Bas, The Oxford Dictionary of English Grammar, 2nd edition (Oxford, 2014)
Adorno, Theodor, ‘Music and Language: A Fragment’, in Quasi Una Fantasia, Essays on Modern Music, trans. by Rodney Livingstone (London, New York, 1956)
Bierwisch, Manfred, ‘Musik und Sprache : überlegungen zu ihrer Struktur und Funktionsweise’ <http://publikationen.ub.uni-frankfurt.de/frontdoor/index/index/docId/12535> [accessed 20 January 2017]
Borghese, Antonio, A New and General System of Music, John Gunn (translator) (London, 1790)
Callcott, A Musical Grammar, in Four Parts, First American edition, from the last London edition (Boston, 1810)
Caplin, William E., Classical Form: A Theory of Formal Functions for the Instrumental Music of Haydn, Mozart, and Beethoven, New Ed edition (New York; Oxford, 2000)
Corri, Domenico, The Singer’s Preceptor, Or Corri’s Treatise on Vocal Music, 2 vols (London, 1810), i
Couperin, François, Troisième livre de pièces de clavecin, first edition (Paris)
Daines, Simon, Margarete Rösler, and Rudolf Brotanek, Simon Daines’ Orthoepia anglicana (1640), Neudrucke Frühneuenglisher Grammatiken (Halle a.S., 1908), iii
Daube, Johann Friedrich, Anleitung zur Erfindung der Melodie und ihrer Fortsetzung, 2 vols (Vienna, 1797), i>
Forsyth, Mark, The Elements of Eloquence, Icon Books Ltd, 2013
Gunn, John, The Theory and Practice of Fingering the Violoncello, second edition (London, 1789)
Hiller, Johann Adam, Anweisung zum musikalisch-zierlichen Gesange (Leipzig, 1780)
Hugot, Antoine, and Johann Georg Wunderlich, Méthode de Flûte du Conservatoire (Paris, 1804)
Kirnberger, Johann Philipp, Die Kunst des reinen Satzes in der Musik aus sicheren Grundsätzen hergeleitet und mit deutlichen Beyspielen erläutert, 2 vols (Berlin, 1776), ii–i
Koch, Heinrich Christoph, Musikalisches Lexikon (Frankfurt am Main, 1802)
———, Versuch einer Anleitung zur Komposition, 3 vols (Leipzig, 1787), ii
———, Versuch einer Anleitung zur Komposition, 3 vols (Leipzig, 1793), iii
Kollmann, Augustus Frederic Christopher, An Essay on Musical Harmony (London, 1796)
———, An Essay on Practical Musical Composition (London, 1799)
Kunert, Richard, Roel M. Willems, and Peter Hagoort, ‘Language Influences Music Harmony Perception: Effects of Shared Syntactic Integration Resources beyond Attention’, Royal Society Open Science, 3/2 (2016), 150685
Logier, Johann Bernard, System der Musikwissenschaft und der praktischen Composition (Berlin, 1827)
Löhlein, Georg Simon, Clavier-Schule oder kurze und gründliche Anweisung zur Melodie und Harmonie, 2nd edition (Leipzig and Züllichau, 1773)
Milchmeyer, Johann Peter, Die Wahre Art Das Pianoforte Zu Spielen (Dresden, 1797)
Morley, Thomas, Plaine and Easie Introduction on to Practicall Musicke (London, 1597)
Reicha, Anton, Traité de haute composition musicale, Czerny edition, 4 vols (Vienna, 1824), ii
Riepel, Joseph, Anfangsgründe zur musicalischen Setzkunst: De Rhythmopoeïa oder von der Tactordnung, 7 vols (Regensburg and Vienna, 1752), i
Scheibe, Johann Adolph, Über die musikalischen Composition, im Schwickertschen Verlage (Leipzig, 1773), i
Tromlitz, Johann George, Ausführlicher und gründlicher Unterricht die Flöte zu spielen (Leipzig, 1791)
Türk, Daniel Gottlob, Klavierschule, oder Anweisung zum Klavierspielen für Lehrer und Lernende (Leipzig und Halle, 1789)