The most fundamental question of all in playing early music today is: how can the music be played to reflect historical practice? This is the third of three articles on this topic for medieval music, aiming to be practical guides with plenty of musical examples and illustrations, and a bibliography for those who wish to delve further.
The first article discussed historical instrument combinations and the second how to create polyphonic accompaniments for music written monophonically. This third and last article discusses a wide variety of questions of style: the performance of the non-mensural (non-rhythmic) notation of the troubadours; the role of the voice and instruments; ornamentation; questions of intelligibility, language and sung translations; musical preludes and postludes; and the effect of the various functions of music on the way it is performed.
This article features a video of Martin Carthy singing a traditional English song on the basis that his free style, with the voice leading and guitar following, each verse phrased differently, so free that it is mensurally unwritable, may have something important to tell us about the historical performance of troubadour songs.
Non-mensural music, free rhythm, and its implications
The troubadours were the lyric poets and songwriters of Occitania, what is now southern France. Their influence upon medieval song is difficult to exaggerate, and their work is explored is in the first and second articles of the series about the 13th century Iberian Cantigas de Santa Maria. Modern singers of troubadour material have a fundamental problem: their songs were written down without rhythm. Medieval notation was without rhythm until Leonin. Leonin was a member of the Notre Dame school of polyphony, composers at or around the Parisian cathedral of that name from 1160 to around 1250, and he established six underlying pulses or rhythmic modes to indicate note values in a given piece of music. The underlying pulse was read from the way notes were grouped together on the page. Whether or not this can be applied to music outside of the Notre Dame circle is a question of great contention, and it doesn’t necessarily help us with music not laid out in Leonin’s fashion. Fully mensural music arrived by degrees in the 13th century and was collated into a comprehensive system by Franco of Cologne, who revolutionised written notation to the point where what is to be sung could be read directly from the page, formalised in his highly influential Ars Cantus Mensurabilis (The Art of Mensurable Music), 1250–1280. Troubadour songs were written down, on the whole, long after the composers had died, in an era of mensural notation. Still, for some reason, scribes of troubadour music notated in the non-mensural fashion that was current during troubadours’ lifetimes.
This has led to heated debates among modern musicologists about how troubadour music is to be sung. Into the historical silence, theorists have given their ideas: every note is of equal value; or the music is isosyllabic, meaning every syllable is of equal value, with the result that a melisma – a syllable sung over several notes – is quickened; or the notes are in natural spoken rhythm; or the songs are in modal metre; or the music is unmeasured and sung to a rhythm of the performer’s discretion. All these theories have something in common: they cannot be supported by evidence from any historical source. These ideas suppose that lack of rhythm in written troubadour music was a choice with a method behind it, but in that sense there was no choice to be made: in the music notation the scribes of troubadour music used, there was no method for writing rhythm.
However, this is not the whole story, as there is a remaining mystery: why did troubadour music continue to be written in non-mensural (non-rhythmic) notation after rhythmic notation was available? There is a startling example in the 13th century manuscript, BnF MS fr. 844, folio 78v, seen on the right. On the left column is an anonymous dança, a poetic form that was sung to instruments or an instrument, usually the vielle, while dancing. It begins “Tant es gay es avinentz” and is written in mensural notation. On the right, third stave down, is a variant of the same melody used for the troubadour song by Blacasset (Blacassetz, Blacssetz, or Blachessetz), beginning “Ben volgra que.m vengues mer[c]es”, but this song is written on the same page in non-mensural notation.
We don’t know why lack of rhythm persisted in notation of troubadour music. It may simply have been to honour their history, a tradition of notation that was seen as integral to troubadour song, with no musical significance and which therefore cannot tell us anything about the performance of troubadour compositions. However, the lack of mensuration after written rhythm was available – and even on the same page as mensural music – may tell us something important about troubadour performance practice. A definitive answer is impossible, but we have to have some method for ascertaining rhythm if we are to sing such songs at all.
Sometimes a song in non-mensural notation makes complete musical sense by putting it in one of the six rhythmic modes, the underlying pulses used by the church, and likely outside the church, too. (Rhythmic modes are explained and explored in the articles about Kalenda maya and Foweles in þe frith.) This is not, however, a solution for all non-mensural music, some of which resists this modal interpretation. For these songs, we need to find another way if they are to be interpreted and sung.
I am struck by how much English folk singer and guitarist Martin Carthy’s rendering of traditional English and Scottish song, Georgie (other versions are called Geordie), meets one of the modern solutions for a way of singing non-mensural troubadour music. Martin heard Georgie from gypsy singer Levi Smith, in a style that was unlike anything he had heard before, which he describes as “speak-singing”. Levi’s and Martin’s notes follow natural spoken rhythm; but Martin adds emphasis to words with discretionary rubato. Martin’s performance is an excellent example of the impossibility of accurately notating a song sung in free rhythm, with a pulse which changes as particular words are emphasised for effect. Taken as an example, this may possibly tell us something about the troubadour style of singing and its continuance in non-mensural notation, if this is the reason for the differential notation in BnF MS fr. 844. At the very least, “speak-singing” gives us a method of singing non-mensural songs where historical methods such as rhythmic modes cannot apply.
An exercise in trying to notate Martin Carthy’s free singing of Georgie may be instructive. Below is the melody as Martin Carthy sings it, notated without his rhythm, as troubadour song was written. In this case, every note is written as a nominal crotchet, indicating nothing about its actual value in time, and vertical lines are not bar lines but represent a caesura at the end of a phrase.
Taken straight from the page, as it were, the effect is completely unmusical. In this case, we have the advantage of a recording, so here is Martin Carthy singing the song very freely, using the natural rhythm of speech but pushing and pulling the tempo for effect, to emphasise particular words or add narrative tension, adding or subtracting notes of the melody to suit a particular phrase within a verse.
If we did not have Martin Carthy’s performance, and only the non-mensural notation above, and we then followed the modern theory of isosyllabic performance, every syllable of equal value, then the result would be as follows:
We can hear immediately that the result is quite unlike Martin’s performance, and there is the fatal flaw in this theory that different words match the notes differently in every verse: two adjacent notes in one verse may be two words, and in another verse a single word with two syllables. On the isosyllabic model, where a two syllable word halves the note value of each syllable, this has the effect of changing the melody in different verses. This result explains in practical terms why the isosyllabic idea is unconvincing.
You’ll hear the difficulty of writing Martin Carthy’s performance down: the rhythm is free and, because of this, the phrasing and rhythm is slightly different in each verse. Nevertheless, below is an attempt in write down what Martin sings. In the sense that the written note values are more accurate, it takes us one step closer to what he does than the entirely non-mensural version. But the fact that the rhythm is nailed down also takes us a step away from the freedom-in-the-moment of his performance and the variety between verses.
Finally, below is a fully mensural version, with a time signature and bar lines, conforming to all the prescriptions necessary for a coherent piece of written music. In order to fit 4 beats per bar, some note values have had to be changed, taking us away from what Martin actually sings, so much so that I’ve had to make one bar 5/4 to get back to his performance. Though close to what he sings in the first verse, the effect of formalising this music is to fit it into a straightjacket that ties up and restricts the freedom of his real-life playing, pushing and pulling the tempo, in later verses.
Some historical clues about secular medieval singing style are instructive and relevant here. In his Scienta artis musicae, written in the 1270s, Elias Salomon is keen to distinguish between godly and ungodly music, and complains that church singers are straying: “they scarcely deign at times to perform plainchant at its proper pace when they sing by anticipating, accelerating, retarding, and improperly phrasing the notes.” This may suggest that church singing was characterised by singing at an even pace with every note in its predictable place, whereas secular singers felt free to make tempo shifts for effect within a piece, and not necessarily sing every note in its allotted place on the page – like Martin Carthy.
In his Ars musice, written 1270s–1300, Johannes de Grocheio made a distinction between two ways of performing music that may offer further help: “Others divide music into plain or unmeasured, and measured, understanding by plain or unmeasured the ecclesiastical [chant] … if by unmeasured they understand music not measured in any way but rather uttered completely at will, they are wrong, since any operation of music and of any art ought to be measured by the rules of that art. If by unmeasured they understand not so precisely measured, it is possible that this division seems to remain.” Grocheio is saying that measured music is ordered by what he calls “rectas percussiones”, correct measured beats, a predictable underlying rhythm; whereas unmeasured music is still ordered, otherwise there would be chaos among singers, but it has no underlying rhythm. Certainty is impossible, but the continued writing of non-mensural notation for troubadour music long after mensural notation was available may be due to it being performed not only in what Grocheio calls a “not so precisely measured” way, but also with the secular pushing and pulling of tempo and note values criticised by Elias Salomon. The fact that chant was sung in groups and troubadour songs most often by individuals gives the performer of the latter complete rhythmic freedom, in the manner of the complaint by Elias Salomon, so it may have been that tying the rhythm down mensurally was against the spirit of free rhythm for troubadour song.
So if troubadour songs were sung freely, and if this was the reason they continued to be written non-mensurally when rhythmic notation had become common, then the impossible exercise of trying to standardise notation for Martin Carthy’s performance may be a parallel example to illustrate the liberty with which troubadour songs were performed, and may possibly be the reason they continued to be written in non-mensural notation.
Regardless of the truth of this conjecture, the question raises an important point about performance practice generally. Music is dead as ink and paper, and only has life in performance, so how far does written music, however detailed in its notation, represent what was then or is now actually played and heard? Modern examples of the impossibility of capturing a performance on the page would include the sliding blue notes of a jazz musician, the extempore solo of a blues guitarist, pianist, or trumpeter, the singing ahead or behind the beat of Billie Holiday or Stevie Wonder, and so on. We know what all of this sounds like or, if we don’t, we can seek it out. With medieval music we cannot.
This raises the further question of expression of sung words. If one is telling a story in a song, which is the case for all medieval secular texted music, then there is a real argument for having some freedom in the pulse of rhythmic expression, in order to make dramatic pauses at key points or quicken the pace for momentum at points of action. This will not work with every song or every song type – a sung dance, for example, needs a regular rhythm – but it is one possible option that fits with Elias Salomon’s complaint about secular singing style and opens a way for singers today to perform non-mensural songs which do not fit a rhythmic mode.
Playing and/or singing
Christopher Page (1987) points out that the evidence for singing and playing instruments at the same time is usually lacking in medieval accounts. A text may describe a person singing and describe them playing an instrument, without stating that they did the two together. If, at first, this seems like pedantry, he gives convincing evidence that this is a real question. In the Anglo-Norman Roman de Horn (Romance of Horn) by Thomas, c. 1170, Horn sings a cappella and then plays a melody on the harp. Gerald of Wales, c. 1146–c. 1223, historian and Cambro-Norman archdeacon of Brecon, describes a band of horsemen who travelled with a fiddler and a singer: the singer sang, and then the fiddler made a musical reply, back and forth. Similarly, in Claris et Laris, a 13th century Old French romance, knights and ladies listen to a minstrel “who sang them a song and performed the refrains on his fiddle”.
There are other accounts and there is medieval iconography that certainly does show an instrumentalist singing while playing, so this is a not a question of universal practice, but it does reveal another option in the repertoire for performing medieval music.
The medieval voice: rhetoric as performance
It should go without saying that there is not one ‘medieval singing voice’ any more than there is one ‘modern singing voice’. Today we can hear distinct stylistic differences between, say, singers of opera, jazz, rock and folk music, and within each genre there is also a range of different vocal styles. So it is with medieval music: we would not expect a minstrel, jongleur or troubadour to sing in the same manner as a singer of ecclesiastical chant.
There are surviving descriptions of singing styles. In Roman de Horn, c. 1170, the fighter-hero sings to the harp with “a loud and clear voice, just as the Bretons are versed in such performances”. In Thomas’ Tristan, also c. 1170, “Iseut sings very sweetly, attuning her voice to the [harp]”. Similar words are used to describe the singing voices of vielle-playing minstrels. In the romance by Gautier de Tournai, Gille de Chyn, 1230–40, a knight named Gerars Malfillastres travels with six companions, including two fiddlers who sing with “loud and clear voices, accompanying themselves with their fiddles”. In the anonymous epic, Hervis de Metz, c. 1250, a minstrel plays fiddle after a banquet and sings “in a beautiful and sweet way”.
Such descriptions are helpful to an extent, since they tell us there was colour and tone in the performance, but terms such as “loud” and “sweet” are subjective and contextual: “loud and clear” suggests skilled voice projection, but the tonal quality is not described, and may be smooth or harsh; and what is considered a “sweet” sound in one time and place may not be so in another. So the quality of the medieval voice is still a mystery in many respects, but one thing is clear from the recent debate known by some as the vibrato wars. I will not rehearse the arguments here, except to state that the modern use of indiscriminate vibrato on the voice or an instrument is a recent and modern development, unknown in historical music. It is not that vibrato was entirely absent, but it was an ornament, a condiment to add spice, to be used sparingly, rather than a thick sauce to be poured over everything.
As noted in the second of these three articles, words of ecclesiastical approbation tell us not only of the singing style conservative clergy were aiming to maintain, but also the influence of secular singing styles that were taking hold in the church and being fought against. English cardinal of the Roman Catholic Church, Robert of Courçon or Robert Curzon, complained in his Summa, c. 1208, of “minstrelish little notes”, fast virtuoso passages played or sung by minstrels which were becoming a feature of ecclesiastical organum. And, as noted above, French cleric Elias Salomon complained in the 1270s of church singers who sped some phrases, slowed others, and did not sing each note in its alloted place. These tempo shifts may be connected with a medieval cultural and artistic preoccupation: rhetoric.
One aspect of rhetoric is individual expression, the heart of the performer reaching the heart of the audience. The theory and practice of rhetoric, the art of moving and persuading listeners, was a favourite topic of medieval writers, and this has major implications for vocal and instrumental music. Medieval and renaissance accounts of musical performances stress their impact on the audience, the impressiveness of the singer’s or musician’s ability to touch emotions and move the soul. It may be that such artistic power was expressed in part by the very features Elias Salomon complained of, displayed by 13th century minstrels, the church singers he disliked, and modern singers like Martin Carthy. The more conservative ecclesiastical writers considered any personal expression, any deviation of the individual from the whole body of singers, to be ungodly hubris; to the secular singer, such personal expression and individuation was fundamental to their performance.
Another aspect of rhetoric is written and formal. The significance of poetic form in medieval song is difficult to underestimate: the number of lines per strophe (stanza or verse) and the number of syllables in each line was rigidly fixed in order to create a particular poetic and aural effect. This was clearly a particular challenge for the less-skilled medieval writer, such as the Iberian King Alfonso X in his Cantigas de Santa Maria, 1257–83, who mangled meaning and syntax in order to make his lines conform to the challenging demands of the syllable count and rhyme scheme. It can be equally challenging for the singer, too: many of the syllable counts in each line of the Cantigas are long enough for the rhyme scheme to be lost by the time the melody reaches the next line’s end, and this was deliberate: the effort required of both performer and audience was integral to the experience. Both the lyricist and the singer were expected to be cognisant of form and to have considerable rhetorical power, overcoming the technical challenges in bringing the art of rhetoric to performance, using the form to convey its story-telling function to a listening audience.
Medieval songs are, first and foremost, stories. This is particularly so in the case of a collection such as the Cantigas de Santa Maria, where almost all the music is contrafacta, pre-existing music to which new words were written, and therefore the music is essentially a vehicle for the story. The central question, then, is about using the form in the service of the story, and to do this we must not only have skill as a singer but also have a personal and rhetorical attitude towards the words and the music. As such, fundamental questions of motivation and performance need to be answered. What is my reason for singing this? Who am I when I am singing it – am I an impartial or an interested narrator, or a character in the story? What is the rhythm and sonority of the lyric? How do I convey and convince the audience of my character’s sadness, joy, wistfulness, anger, indifference, etc.? What effect should this have on the tonal quality of the voice and of the instrumentation? What indications does this give for musical pace? Should the underlying rhythm be regular, as in a dance, or have quickening passages and rubato as in speech when giving drama to a story?
A convincing performance is what gives a song life. I would suggest, then, that rhetorical skill is the key to singing a medieval story-in-song. Whether one’s performance is “loud and clear” or “beautiful and sweet” will therefore depend on the message being conveyed and the character being played as the drama unfolds.
Ornamentation in renaissance music was encoded on the page, with signs for an extra note falling from above, or rising from below, and so on. Medieval music is more of a mystery in this respect. Ornamentation of the medieval singing voice seems to have carried two meanings: the first pertaining to timbre, voice quality, in the aid of rhetoric; the second pertaining to grace notes in the modern sense, though this is not certain.
In the 9th to 11th centuries, written music was arguably no more than signs acting as aide-mémoires for singers who already knew the melody. The daseian (or dasian) notation of the anonymous 9th century Musica enchiriadis and its companion volume Schola enchiriadis, for example, shows precise pitch relationships but does not notate rhythm. Another of the earliest known forms of European music notation originated in the 9th and 10th centuries in the Benedictine monastery of Saint Gall (Sankt Gallen) near Lake Constance, in (what is now) Switzerland. These early neumes (figures of movement) give the pitch direction of the melody but, since this notation lacks ledger lines, there is no precise pitch, and it does not indicate rhythm. This means that music-making from such sources today can only ever be conjectural.
Nevertheless, the Saint Gall notation gives information that is lost in the more specific pitch-relation system developed by 11th century Italian Benedictine monk, Guido of Arezzo. The Saint Gall system includes signs which suggest that medieval decoration at this point, at least for the ecclesiastical voice, was thought of in terms of the quality of the note rather than the addition of ornamental notes, and that this voice quality was linked to the mood of the modes (see the section on medieval modes in the second of these articles). One Saint Gall notation sign was a quilisma, a dot followed by repeated curls and an upward sweep. It isn’t clear precisely what this was intended to convey, but we have two clues: the shape of the quilisma is suggestive; and quilisma means ‘authoritative shout’ in Greek (possibly – this is a subject of continued debate). A further suggestion about voice quality comes from the anonymous author of Summa musice, c. 1200, who wrote that chants that are sad and lamenting should be set to “low-lying and ponderous modes”, and for this reason the Tracts of the Mass, part of the liturgical celebration of the Eucharist sung during penitential seasons such as Lent, should be sung to the second “dark and profound” mode (hypodorian) or to the eighth mode (hypomyxolydian). Lotario de Segni (1160/61–1216), who became Pope Innocent III, was particularly interested in the hermeneutical function of chant, its interpretive use in conveying theological truths by moving and influencing human emotion, and thus Lotario stated that the Tract should be sung with a “bitter voice”. Thus we see again the importance of rhetoric in the performance of medieval music, expressed through the ornamentation of vocal tones.
For ornamentation as we understand it in modern music, the adding of grace notes, a comparison of the way the same troubadour song is notated in different manuscripts may be helpful. Written down in the 13th and 14th centuries (long after their composition), they were notated with clear pitch but without rhythm. Ron Cook (2013) has compared musical phrases within the same melodies of troubadour songs in various manuscripts, and attempted from this to ascertain types of ornamentation within the troubadour repertoire. Whether he has achieved this depends entirely on whether one believes the extra variant notes in some manuscripts are changes to the melody itself or the same melody with added decoration. The information that would be decisive in this matter, written note values on the page, is the very thing that is missing. The performer and the scribe both presumably knew some norm of performance practice in these cases, but any such understanding is not conveyed in writing.
Such figures in the versions of songs with extra notes can be incorporated into medieval performance as melodic variations or as ornamental decorations. The principle of the variation in each case is a note of the melody, divided into units that begin one, two, three or four steps above, coming down to the note of the tune, or beginning on the melody note with additional rising or falling notes, or starting on the note below and rising to the melody note. Some of Ron Cook’s examples are shown in pairs below, first the simpler figure as it appears in one or more manuscript, then the more complex version of the same figure as it appears in another manuscript. The notation being non-mensural, the crotchets are nominal.
Here are the same musical figures as I imagine they may have sounded as ornaments. For each example, first the simpler figure as it appears in one manuscript with conjectural note values, then the more complex figure as it appears at the same point in the same song in another manuscript, with the additional notes as ornamentation, also with conjectural note values. These examples are played on gittern by Ian Pittaway.
We cannot know if this is correct, but it seems to me a good working model. To take an arguably parallel example from English traditional music, Percy Grainger recorded traditional singer Joseph Taylor on wax cylinder in 1908, singing Brigg Fair. This is the plain melody.
This is how Joseph Taylor sang it.
This is a better representation of what Joseph sang, with the ornamentation written in.
Below is the first full bar of Brigg Fair written non-mensurally with nominal crotchets, without ornamentation; followed by the same bar with Joseph Taylor’s sung ornamentation, written as non-mensural nominal crotchets. The visual difference between the two is arguably the same as that discovered by Ron Cook in comparative troubadour songs, and it happens to look very much like his example 6 above.
As an experiment in applying this imputed ornamentation, here is the first part of the melody for Cantiga de Santa Maria 10 (used as the basis for musical illustrations in the second article), played on recorder by Kathryn Wheeler.
Here is the same melody, with some of Ron Cook’s putative troubadour ornamentation added, played on vielle by Kathryn Wheeler: bar 1 and 3, ornamentation example 2; bar 5, example 5; bar 6, example 7; bar 7, example 5; bar 8, example 2.
Since the troubadours wrote and performed songs, here is the same example sung by Kathryn Wheeler.
The medieval idea of musical rhetoric, of convincing a listener through the communication of ideas and emotions, leads to the fundamental question of intelligibility. Sung poetry can only be rhetorical if it is comprehensible. There are three performance options for communicating to an audience:
1. Sing in the original language and explain the meaning verbally. A skilled singer with skilled instrumentalists can add enough colour, expression and meaning to a song to hold an audience’s attention in a language the audience does not understand. To hear the cadences of the original language, which requires research into original pronunciation, can be a delightful experience. But it does have major and insurmountable drawbacks. Though a singer may take the audience along musically and emotionally, it cannot be story-telling if the audience cannot understand the words. For example, if the audience hear ‘Chanaghanadivawei’, is this ‘Cha na gha na di va wei’ in the language sung, or ‘Chana ghana divawei’? It hardly matters, since it is meaningless to the audience in either case, and communicates no meaning.
2. Sing in the original language and give out translation sheets. A printed programme can be devised with the original words side by side with a translation in the vernacular language of the audience, and this can work well, a helpful aid for the audience to follow the story. However, attention is then divided, and the immediacy of the performance is lost.
3. Have a repertoire of medieval songs translated into modern English for English-speaking audiences. The original, historical singers sang songs and listeners heard them in a language they understood, and so it is arguably a more historically informed performance to do the same, to tell the stories in as direct a way as possible, fulfilling the rules of rhetoric for a modern audience. Indeed, in translating a song into the listeners’ vernacular, one is following evidenced medieval practice. Ar ne kuth ich sorghe non – Formerly I knew no sorrow – is one of the first surviving songs in the English language, dated to the first half of the 13th century. It is French rather than English in origin. The manuscript page has the French lyric below the music, words and notes aligned, and the English words below the French. The English lyric is a contrafactum (substituting one text for another), with the same meaning as the French. The same is true, for example, of Angelus in virginem in the late 13th century English manuscript, Arundel 248. Underneath the Latin words are English words, Gabriel fram evene king. It is not strictly a direct translation, since as such it would neither scan nor rhyme, but each verse in English has the same thematic content as the Latin original. This happened several times to the sequence (a chant or hymn, sung or recited during the liturgy of the Eucharist) Letabundus exultet fidelis chorus, the first evidence of which is c. 1100 (British Museum MS Add. 18, 302). All the translated versions of Letabundus are 15th century and macaronic, in two languages, mostly Middle English but retaining some lines of Latin, and they are the anonymous Regem regum a mayde hath borne (Bodleian Library MS Ashmole 189, f. 104v–105r, 15th century); The incorrupt wombe virginall, written by James Ryman in c. 1492 (Cambridge University Library MS Ee. I. 12, f. 35v); Now wel may we merthis make (Selden Carol Book, Bodleian Library, Oxford, MS. Archive Selden B. 26, f. 10r); and Glad and blithe mote ye be (Selden Carol Book, Bodleian Library, Oxford, MS. Archive Selden B. 26, f. 19v). The common medieval pratice of singing in a vernacular translation the intended audience will understand, translated from a language the audience wouldn’t understand, drives home the same point about medieval song: the rhetorical purpose for the singer is to communicate, and that requires direct comprehension by the audience.
Since the purpose of language is to communicate meaning and an unknown language communicates no meaning, my preference is the versified vernacular. In just the same way, the words of a novel are meant to communicate, to tell a story, so I would read the translated novel The House of the Spirits by Chilean author Isabel Allende, rather than the original La casa de los espíritus in a language I don’t understand; and I would read Fyodor Dostoevsky’s The Brothers Karamazov, rather than the original Бра́тья Карама́зовы, as reading the latter would mean wading aimlessly and pointlessly through 900 pages of script that is meaningless to me. Songs are stories, and if it isn’t worthwhile for me to ‘read’ a story in a language I can’t understand, then I also gain no meaning from the words of a song I don’t understand.
Since no two languages are the same, with different idioms and rules of syntax, one cannot simply sing a straight translation, since the act of translation removes the syllable count and the rhyme scheme. My own method has been to reconstruct the syllable count and rhyme scheme of every stanza of a given song. With the Cantigas, the syllable count and rhyme scheme is given you by Andrew Casson on his excellent website, Cantigas de Santa Maria for singers, which also includes, for every song, the words in Galician and the original notation. Then, with an English translation of the words, write English verses using the original syllable count, rhyme scheme and stresses, trying, as far as possible, to reproduce the melismata (groups of notes sung over a single syllable) of the original. The source for all my versions in English of the Cantigas, performed in the series of six articles on the subject, was Kathleen Kulp-Hill’s Songs of Holy Mary of Alfonso X, The Wise, a non-versified translation into English of all 420 Cantigas and both Prologues. In writing an English lyric – i.e. fulfilling the same role as the 13th century authors of Ar ne kuth ich sorghe non and Gabriel fram evene king – it helps greatly to remember the principles of rhetoric: Who am I when I am singing this? What is my emotional state? What do I wish to convey to an audience, to convince them of? In other words, it is vital before starting to have an attitude, as that will influence the choice of words.
If possible, read the music from the original page. Many holding institutions have scanned their manuscripts and made them freely available on their websites. The advantage of reading the original music is that medieval notation often indicates information that modern notation cannot convey, such as syllable emphases or stresses.
In my experience, one problem in this process keeps recurring: languages use different numbers of syllables to convey the same meaning. I regularly find myself able to convey two lines of meaning in the Cantigas using only one line in English. It is at these times that I fall back on some of the idiomatic usage of English traditional ballads to fill the syllable gap. For example, in CSM 173: The kidney stone Cantiga (performed here), Kulp-Hill’s “He woke up” becomes the traditional English stock phrase, “And when he woke on his bed of rest”. Like many song traditions, English folk music uses lyrical repetition, and I find utilising this method invaluable in both filling the syllable gap and in giving the story momentum. In CSM 159: The pilgrims and the stolen chop (performed here), for example, for Kulp-Hill’s “They had ordered nine chops to be put into the pot, as God is my witness, for they were of that number”, I revisit an earlier line that they “ordered meat, bread and wine, bought for their supper”, to arrive at the repetition, “The meat they ordered with their bread, / the meat they ordered with their wine / was, as God he is my witness, / chops in the pot which numbered nine. / For the pilgrims were that number …”
The disadvantage of singing in English is that the original language has disappeared, but I think that is a sacrifice worth making on the basis that the original story cannot be coherently communicated to an audience in a language they do not understand, even less so if it is a language the singer does not understand. Indeed, if the singer does not understand the language, it is difficult to comprehend how anything meaningful can be communicated. Since some medieval songs, and some Cantigas in particular, are filled with hateful celebration of the deaths of Jews and Moors (explored in the fifth article of the series on the Cantigas), it is absolutely vital that the singer has an understanding of the song before beginning the creative journey that leads to its performance, or the avoidance of its performance in the case of songs promoting hatred.
There is one final issue of language for both performer and listener with a great many of the Cantigas, which was raised for me by an instrumental tutee after seeing me perform some Cantigas: for the modern listener, the rhyme scheme can be difficult to hear in the original Galician, as well as in any versified translation. As opposed to the easy rhymes of moon/June/spoon type songs with a short syllable count in each line, the Cantigas are mostly what we might call art songs, composed in such a way that they were intended to be exacting to write, to fit the syllable count, with long lines so that the listener has to be attentive and listen out for the rhymes. They are mostly in the virelai form, beloved of the troubadours, difficult to write because typically there are only two rhymes across the chorus and the verse. The rhyme scheme of CSM 363: The troubadour in chains, for example, is AA bbba but, though this rhyme scheme is simple, 13 syllables in every line creates a significant distance between each rhyming word. One either has to trust the modern audience to attune to the unfamiliar, or explain the mechanics of the song so they can hear the rhyme structure more readily.
Preludes, the length of songs, and postludes
Johannes de Grocheio, in his Ars Musica, late 13th century, stated that good musicians play preludes to every song and to every piece of music played on the vielle, and that much music ends with a coda or postlude. This raises several important questions.
The first is the musical content of the prelude. Preludes were never written down in the medieval period, but Roman de Horn, c. 1170, is instructive. “Then [Horn] took the harp to tune it. God! Whoever saw how well he handled it, touching the strings and making them vibrate, sometimes causing them to sing and at other times join in harmonies, he would have been reminded of the heavenly harmony.”
This is the prelude to a song, and the description is very much like pieces called Taster de corde (testing the strings) in Italian lutenist Joanambrosio Dalza’s print of 1508, Intabulatura de Lauto Libro Quarto. Dalza includes 5 pieces called Taster de corde, consisting of brief, fast-moving passages up and down to check the melodic tuning, with numerous spread chords to check harmonic tuning, in readiness for the main piece. The pieces called Taster de corde appear to be intabulations of improvised passages. Though more than 300 years apart, the function of Dalza’s testing the strings is precisely what Horn is doing, and this is the content of the prelude: “sometimes causing them to sing”, checking the melodic tuning, “and at other times join in harmonies”, checking the harmonic tuning.
Much medieval iconography shows an instrumentalist tuning an instrument. Above, we see King David with his harp from the Tickhill Psalter, 1310. This is an account of the biblical story in 1 Samuel 16:23: “So it came about whenever the evil spirit from God came to Saul, David would take the harp and play it with his hand; and Saul would be refreshed and be well, and the evil spirit would depart from him.” The first scene on the bottom margin of folio 9v shows King Saul with a devil at his back as David arrives with his harp in its bag. The next scene shows David tuning the harp and Saul becoming free of the devil. In the carefully coded visual representations of the period, tuning was a symbol of heavenly harmony: the harp strings represent members of the church, so harp-tuning represents keeping the body of the church in harmony, and the restoration of harmony between body and soul. The description of Horn’s harping in Roman de Horn, c. 1170, is strikingly reminiscent of this idea: “Whoever saw how well he handled it, touching the strings and making them vibrate, sometimes causing them to sing and at other times join in harmonies, he would have been reminded of the heavenly harmony.” This description, and the fact that harp-tuning in particular is such a common image in medieval iconography, suggests that the symbolic connection of tuning with the restoration of heavenly soul and earthly body was also integral to the idea of the musical prelude, bringing heavenly order to music, uniting body and soul.
The second question raised by Johannes de Grocheio’s statement, that good musicians play preludes and postludes, is song length. With a song sandwiched between a prelude and postlude, the implication is that a medieval song is not a throw-away ditty like a 3 minute pop song, but a significant and dramatic cultural event. How long would such a three part performance take: 10 minutes? 20 minutes? We don’t know the answer, and clearly it would have varied from song to song. But the general point is that today we take our drama largely from novels, films, TV series and theatre: in the medieval world it was enacted through song and story-telling. Horn, the text says, “performed the whole lai [lyrical, romantic verse] for he wished to omit nothing”, implying that this was a lengthy performance that a lesser performer may have edited down. Similarly, some of the Cantigas de Santa Maria are very long by modern standards, but if we wouldn’t rip out every other page of a novel, or watch only the first and last 10 minutes of a film, then there is an argument for singing all the verses of a long medieval song to enact the drama of the story as it was originally conceived. This raises the fear that a modern audience wouldn’t sit through it, and the question of the very particular skills of story-telling, which itself raises the issue of language and intelligibility discussed above in order to retain the audience’s attention. Remember that Horn’s audience were enraptured: “All those present marvel that he could play thus”.
The third question is the content of the postlude, and on this Roman de Horn is very clear: “afterwards he makes the strings of the harp play exactly the same melody as he has just sung.” All in all, then, the performance consisted of an improvised prelude to test tuning, the entirety of an epic song, and then the postlude, an instrumental version of the song.
I have played for dancers in three morris sides and a ceilidh band, and this has taught me a great deal: watch the dancers’ feet, play to their rhythm, stay at their pace, follow their every step. It had an effect on the play I played morris music whether performing for dancers or playing it for listeners in a folk club: a sprightly dance needs sprightly music, a lolloping dance needs to be played with a lolloping rhythm, and so on. The same is true for all music: its function and its style are one.
Herein lies a problem for the players of medieval dance music: we know next to nothing about dances until the 15th century. For earlier choreographies, we can only speculate, since no clear, detailed or unequivocal descriptions have survived. Dancers need a distinct rhythm, and the rhythm of a dance sets its character: since, in modern terms, a hornpipe is distinct from a jig in its rhythm and steps, this should be clearly conveyed by the musicians. In medieval dance music we have to be creative where we lack evidence, but at least one thing is clear: it must be rhythmic, in relation to the underlying pulse detectable in the music.
The question of rhythm and style has a wider bearing on medieval music. In modern pop music there is not a clear distinction between a song and a dance: songs are often danced to, and in dance genres such as techno, house, drum and bass and so on, pieces are by no means exclusively textless instrumentals. In the middle ages, the situation was no different. In c. 1300, Jofre de Foixàm wrote his Doctrina de Compondre Dictats, a treatise in Catalan on the writing of lyric poetry in Occitan. Jofre describes poetic genres according to their stanza forms, subject, length and musical settings. The dança, he writes, “should always have a new tune … [and] is so-called because it is normally sung while dancing, hence it should have a pleasing tune; and it is sung with instruments”.
The carol, evidenced from the 12th century on, was a round dance sung by a ring-leader as the carollers danced in a circle. For commentators of the 12th and 13th century, the carol was simply synonymous with northern French secular music. In Alfonso X’s Iberia of the 13th century, it may be that some or all of the Cantigas de Santa Maria were conceived as carols: they are all songs, they begin with the refrain, as carols do, and the illustration for CSM 120 (below) shows Alfonso instructing not only his musicians to praise the Virgin, but three dancers in a ring.
The link between the Cantigas and carols is uncertain, as often is the question of which medieval songs were sung dances. In dulci jubilo (video here) is a tantalising example. The first evidence for In dulci jubilo is in the life of Heinrich Suso, German Christian mystic. In his The Life of the Servant, he describes an event in 1328: “Now this same angel came up to the Servant [Suso] brightly, and said that God had sent him down to him, to bring him heavenly joys amid his sufferings; adding that he must cast off all his sorrows from his mind and bear them company, and that he must also dance with them [angels] in heavenly fashion. Then they drew the Servant by the hand into the dance, and the youth [angel] began a joyous song about the infant Jesus, which runs thus: ‘In dulci jubilo’, etc.” In dulci jubilo, in the first rhythmic mode, certainly has the feel of a dance, and this passage states that it was a dance, but any evidence of its choreography has not survived. The subject raises the unanswerable question of how many surviving medieval songs were meant to be danced to, and what this may imply for the style in which they are played.
Medieval music was part of a recycling culture: not only would melodies be reused with new lyrics – as troubadours often did, and is almost exclusively the case in the Cantigas – but music would be repurposed, a song tune becoming a dance tune, a dance tune becoming a song tune. The 13th century Razos of the troubadours, accounts of the circumstances of their compositions, states for example that Raimbaut heard an estampie played by two French viellists and used the melody for his song, Kalenda maya. We have seen above that two 13th pieces shared a melody, the dança, Tant es gay es avinentz, and the song by troubadour Blacasset, Ben volgra que.m vengues merces, and in this case it is impossible to be sure which way the traffic was travelling, from song to dance or vice versa. If they felt so free with the material, then arguably so can we, and doing so places emphasis on attending to the music’s purpose in new ways: a melody for a dance cannot be performed in the same way if it becomes a more freely-phrased exercise in dramatic story-telling.
In summary, we have learned that:
Troubadour music was written non-mensurally, even after rhythmic notation was available, and this may have been because it was sung too freely to accurately notate.
Medieval musicians played an instrument and sang, but not necessarily at the same time: sometimes a song was sung a cappella and then played on an instrument; or the verses sung and only the refrains played on the instrument.
The medieval singing voice is predicated upon the art of rhetoric, using the voice to tell a story to convince, emotionally involve and move an audience.
Medieval vocal ornamentation means not only singing notes additional to the melody, but also changing the quality of the voice to suit the music as a rhetorical device.
Medieval audiences heard songs in a language they understood. I advocate the same, to make the material as accessible to an audience as it originally was. This involves devising verses in English which reconstruct the original syllable count and rhyme scheme. At the very least, if performing in the original language, singers have a duty to know what they are singing and convey this to the audience.
Some performances of the late 13th century, and possibly a much wider timeframe, were in 3 parts: an improvised prelude to test tuning, the song, and an instrumental version of the song.
When performing a piece, be aware of its function: dance music should be rhythmic and feel like a dance; a dramatic song should have drama, with rhetorical light and shade to tell the story. It was common in the medieval period to repurpose material – a song became a dance, or a dance tune became a song – and often a piece was already both, a sung dance, in which case it needs a dance rhythm and rhetorical vocal skills.
Final thoughts: the known and the unknown
These three articles about secular medieval music have been an attempt to present some of the available evidence for historically informed performance, in practically applicable ways: instrumentation; extemporised or composed polyphonic lines; non-mensural music; a cappella and instrumentally accompanied singing; the singing voice; ornamentation; rhetoric and the importance of intelligible language; preludes and postludes; and musical function. While some information is clear and based on solid evidence, I hope I have shown that to present all of this as firm fact would be to overstate the case, since the existence of conjecture, supposition and limited sources means that some knowledge exists somewhere on a sliding scale between certainty and best guesses. A single significant discovery can change what we thought we knew, such as Giovanni Varelli’s accidental discovery in 2014 of English polyphony from c. 900, which put back in time a style of European music previously assumed to have belonged to a later period (as described in the second article).
While there is much we don’t know, and the picture is patchy, there are certainly enough clues to inform our medieval playing and singing in productive and creative ways, to help us make musically viable choices. For me, immersing myself in the research is an exciting journey of discovery, expanding my repertoire of performance options. I hope, dear reader, that here you have found something of use for your music-making, too.
Bibliography (for parts 1, 2 and 3)
Bell, Nicolas (2014) ‘Earliest’ polyphonic music discovered in British Library. [Online – click here to go to website.]
Biblioteca Nacional de España (2016) Cantigas de Santa María, digital facsimile of the Toledo (To) manuscript. [Online – click here to go to website.]
Casson, Andrew (2015) Cantigas de Santa Maria for singers. [Online – click here to go to website.]
Colton, Lisa (2016) Angel Song: Medieval English Music in History. Abingdon: Taylor & Francis.
Cook, Ron (2013) The Early Medieval Harp: A Practical Guide. Columbus, Ohio: Dlanor Publications.
Cross, Lucy E. (2000) Musica ficta. In: Duffin, Ross W. (ed.), A Performer’s Guide to Medieval Music. Bloomington: Indiana University Press.
Dick, Alastair (1984) The Earlier History of the Shawm in India. In: The Galpin Society Journal, Vol. 37 (March 1984), pp. 80-98. [Available online by clicking here.]
Dyer, Joseph (1980). A Thirteenth-Century Choirmaster: The “Scientia Artis Musicae” of Elias Salomon. In: The Musical Quarterly, Volume 66, Number 1. [Available online by clicking here.]
Ferreira, Manuel Pedro (2013) Editing the Cantigas de Santa Maria: Notational Decisions. In: Portuguese Journal of Musicology, 1/1 (2014), pp. 33-52. [Available online by clicking here.]
Green, Robert (2000) Symphonia. In: Duffin, Ross W. (ed.), A Performer’s Guide to Medieval Music. Bloomington: Indiana University Press.
Haynes, Bruce (2007) The End of Early Music. A Period Performer’s History of Music for the Twenty-First Century. Oxford: Oxford University Press.
Holladay, Richard (1977) The Musica enchiriadis and Schola enchiriadis: a translation and commentary [Online – click here to go to website.]
Johannes de Grocheio (1270s–1300, modern publication 2011) Ars musice. Edited and translated by Constant J. Mews, John N. Crossley, Catherine Jeffreys, Leigh McKinnon, and Carol J. Williams. Kalamazoo: Medieval Institute Publications.
Kelly, Thomas Forrest (2015) Capturing Music: the story of notation. New York: W. W. Norton & Company.
Kulp-Hill, Kathleen (2000) Songs of Holy Mary of Alfonso X, The Wise. A translation of the Cantigas de Santa Maria. Tempe, Arizona: Arizona Center for Medieval and Renaissance Studies.
Lindahl, Greg (undated) The Cantigas de Santa Maria: Facsimiles. [Online – click here to go to website.]
Loewen, Peter (2013) Music in Early Franciscan Thought. Leiden: Brill.
Mahrt, William P. (2000) The Gamut, Solmisation & Modes. In: Duffin, Ross W. (ed.), A Performer’s Guide to Medieval Music. Bloomington: Indiana University Press.
Mahrt, William P. (2000) Proportion. In: Duffin, Ross W. (ed.), A Performer’s Guide to Medieval Music. Bloomington: Indiana University Press.
McGee, Timothy J. (1990) Medieval Instrumental Dances. Bloomington: Indiana University Press.
Monks of Solesmes (1953) Chants of the Church. Selected Gregorian Chants, Edited and Compiled by the Monks of Solesmes. Interlinear Translations by Rt. Rev. Msgr. Charles E. Spence. Ohio: Gregorian Institute of America. [Available online by clicking here.]
Myers, Herbert W. (2000) Reeds & Brass. In: Duffin, Ross W. (ed.), A Performer’s Guide to Medieval Music. Bloomington: Indiana University Press.
Page, Christopher (1980) Fourteenth-Century Instruments and Tunings: A Treatise by Jean Vaillant? (Berkeley, MS 744). In: The Galpin Society Journal 33, March 1980. [Available online by clicking here.]
Page, Christopher (1987) Voices & Instruments of the Middle Ages. Instrumental practice and songs in France 1100-1330. London: J. M. Dent & Sons.
Page, Christopher (1991) Summa Musice. A thirteenth-century manual for singers. Cambridge: Cambridge University Press.
Parrish, Carl (1957) The Notation of Medieval Music. New York: W. W. Norton & Company.
Planchart, Alejandro Enrique (2000) Sacred Music: Organum. In: Duffin, Ross W. (ed.), A Performer’s Guide to Medieval Music. Bloomington: Indiana University Press.
Sherman, Bernard D. (2003) Inside Early Music: Conversations with Performers. Oxford: Oxford University Press.
Simó, Ester Peretó (2017) A vós, Dona Verge Santa Maria. [Online – click here to go to website.]
Smith, Douglas Alton (2002) A History of the Lute from Antiquity to the Renaissance. Massachusetts: Lute Society of America.
Solez, Kevin (2002) Lyrecraft: The Origins and Adoption of the Greek Word Kitharis. [Online – click here to go to website.]
Thornton, Barbara (2000) The Voice in the Middle Ages. In: Duffin, Ross W. (ed.), A Performer’s Guide to Medieval Music. Bloomington: Indiana University Press.
© Ian Pittaway. Not to be reproduced in any form without permission. All rights reserved.