The troubadours

We now come to the third division of the subject—the formation of the stanza. The consideration of metre and rhyme has shown the skill of the troubadours, but also the danger into which their marvellous finish led them, viz., to forget the real poetical value of their work over the beauty of form. This danger was not equally imminent in forming the stanza, seeing that its rules were too securely founded on the national sense of harmony, and too difficult to comply with, to afford an occasion for easy trifling. The stanza, accordingly, we have to consider as the highest development of art reached by the troubadours. It is therefore surprising that the ‘Leys d’Amors’ says nothing of any importance regarding its composition. The author of the work had evidently a very vague idea of the real essence of the stanza. The only guide, therefore, in our investigation must be Dante’s work, ‘De Vulgari Eloquentia,’ in which he draws a full and logical account from his own deep knowledge and experience. But many of his expressions will be obscure until the musical system of the middle ages is better understood. The rules of musical rhythm are indeed inseparably connected with all forms of Provençal poetry. But the way at least may be partially cleared, and the words of the great Italian poet illustrated by examples from the works of the best troubadours.

The definition which the ‘Leys d’Amors’ gives of a stanza, or, as it calls it, cobla, is very unsatisfactory, or rather is no definition at all. It is expressed in a long poem of the author’s own manufacturing, the meaning of which is simply that a cobla may consist at least of five, and at most of sixteen verses, not including the shorter lines known under the name of bordos biocatz. Nothing more is said on this important subject. Of a division of stanzas according to their metrical and musical composition the author seemingly knows nothing. Dante, on the other hand, begins his long and careful investigation by stating first that ‘omnis stantia ad quandam odam recipiendam armonizata est.’ The word ‘odam’ in this connection must be understood in a double sense—a musical and a metrical: in the former it means simply melody, in the latter the metrical scheme of the stanza. But this oda is very different in different cases: ‘quia quædam [stantiæ] sunt sub una oda continua, usque ad ultimum progressive, hoc est sine iteratione modulationis cujusquam et sine dieresi;⁠[41] et dieresim dicimus deductionem vergentem de una oda in aliam; hanc voltam vocamus cum vulgus alloquimur.’ In the cases here referred to, therefore, the flow of melody or verse must not be interrupted by a marked rest or pause, but must go on in an equal strain to the end of the stanza. This kind of stanza, Dante continues, was used chiefly by the great Arnaut Daniel, and especially in the sestina invented by that troubadour, and imitated by Dante himself. A stanza of one of Arnaut’s sestine will at once make the meaning of Dante’s words clear:

It would be impossible to find a point where to divide this stanza on any principle. There are no groups of verses marked by rhyme, seeing that there is no rhyme; there is no change between accen agut and accen greu; there is not even a strong grammatical pause. Accordingly it may be concluded that the musical accompaniment of the words was not interrupted by any striking harmonious modulation such as would have made a rest necessary. Exactly the same may be said of Dante’s own sestina,

which is constructed on the same principle. In many other cases also where there are rhymes a division of the stanza according to Dante’s system is utterly impossible, because the different parts allow of no forming into groups by the recurrence of the same order of rhymes. For instance, the following stanza of Jaufre Rudel must have been sung to a continued oda without any interruption:

Stanzas, however, ‘sub una oda continua,’ are not the rule. ‘Quædam vero sunt,’ Dante continues, ‘dieresim patientes, et dieresis esse non potest, secundum quod eam appellamus, nisi reiteratio unius odæ fiat vel ante dieresim vel post vel utrimque.’ The criterion, therefore, of the possibility of a dieresis or volta is, first of all, that in the poem there should be certain groups defined musically by the repetition of the same melody, and metrically by the recurrence of the same rhymes and of verses of the same length. The volta can, as has been seen, be either before or after such a group, or between two different groups if both parts of a stanza are divided in this way. Dante gives the terms for all these combinations in the following words: ‘Si ante dieresim repetitio fiat, stantiam dicimus habere pedes, et duos habere decet, licet quandoque tres fiant, rarissime tamen. Si repetitio fiat post dieresim, tunc dicimus stantiam habere versus; si ante non fiat repetitio, stantiam dicimus habere frontem; si post non fiat, dicimus habere syrma sive caudam.’ These few words contain in a nutshell the whole theory of Italian, and, with some slight changes, also of Provençal, stanzas. It remains to enter into the special cases referred to by this rule. The first alternative Dante mentions is that of a division effected by the repetition of certain melodic and rhythmic phrases in the first part of a stanza. After these groups, which in this case are called pedes, a rest or volta becomes necessary; and after this a new melody begins, which lasts to the end of the stanza, and is called a cauda.⁠[43] The following is one of the very numerous examples of a stanza consisting of two pedes and a cauda:

In this case the cauda is as long as one pes, consisting, as it does, of four verses. Very seldom, says Dante, are there more than two pedes to a cauda. This, however, applies only to the Italian literature of his time. In Provençal poetry there are many instances of three pedes in a stanza; and the favourite form of the Italian poets of the cinquecento, the ottava rima, must also be defined as a stanza consisting of three pedes and a cauda. The most important form of lyrical Italian poetry, the sonnet, consists of pedes and cauda. The two quatrains show the required repetitio unius odæ, and the two terzine form the cauda. If the repetition of a melodic and metrical phrase takes place after the volta, and only there, the two groups in the second part of the stanza are called versus, while the first undivided part assumes the name of frons. The number of versus scarcely ever exceeds two. This form is also very common in Provençal poetry. In the following stanza of Guillem IX. of Poitiers, the first three lines form the frons, and the last four are divided into two versus of two lines each:

These two principles of division in a stanza, viz., pedes and cauda, or frons and versus, Dante seems to consider as the most important. In both cases the stanza is actually divided into three parts; and this, indeed, was the fundamental principle of the Italian lyrical stanza, which in this respect, in conformity with the middle-high-German strophe, differs from the langue d’oc. Into old Italian poetry this tripartite division was perhaps introduced from the leonine hexameter, which, as has been stated before, sometimes took a similar form. In the Teutonic languages it seems much older; and indeed it is to be found in the old Icelandic ljôđahâttr, where the first two lines are of equal length and belong to one another, while the third one, longer than each, stands by itself. In the German popular epic this principle is not visible; but it appears again unmistakeably in the mediæval ‘minneliet.’ The usual form in the latter is pedes and cauda, which here are called Stollen and Abgesang. It is impossible here to consider the interesting phenomena arising from the conflict of this principle with the Provençal bias in those cases where the German minnesinger tried to imitate the stanzas of the troubadours. The prevalent principle in Provençal poetry seems to have been the division of a stanza into two corresponding parts; and, accordingly, to the above-mentioned combinations two more of great importance must be added. The first of these, which Dante also is acquainted with, is the division of a stanza into pedes and versus. In this case the stanza is divided into four parts; but each pair of these is so closely connected that the Provençal principle of a division into two halves is fully borne out—the more so, as there were evidently only two different melodies, each of them being repeated. The instances are again very numerous. In the following stanza of Peire d’Alvernhe’s, the pedes and versus consist of two lines each:

The fourth and last combination occurs when the stanza consists of a frons and a cauda, that is to say, when the two parts are undivided in themselves, but when a new melodic and metrical period begins after a certain number of verses. This form shows the Provençal principle of a division into two parts more clearly than any of the others; but Dante, from his point of view, is also right in not approving of it, or rather in not acknowledging it as a division at all, seeing that there is no ‘repetitio unius odæ.’ One out of many examples of this phenomenon is a stanza of Bertrand de Born’s, where the frons and cauda consist of three lines each:

The end of the first and the beginning of the second musical and metrical phrase, as marked by the diesis or volta, was generally further strengthened by the conclusion of the grammatical sentence. In most of the numerous stanzas already quoted, the volta contains either a full stop or a semicolon, or at least a comma. The sentence is seldom carried on through the volta, though even the best troubadours were not always careful in applying this rule.

By these various methods, the stanza was strictly divided into different parts. But, on the other hand, the feeling of the troubadours for unity and harmony was too keen not to make it desirable to bridge over somehow the gap made by the volta, and to preserve the connection between the two sides. This was done by means of the rhyme, which, as has been seen, was used for a similar purpose between the different stanzas of a poem. This process is called by Dante concatenatio; and this concatenatio might be effected in two different ways. The first and simpler mode consists in the cauda or versus adopting one or several rhymes of the frons or pedes. This is the course usually followed; and almost all the stanzas above quoted may serve as examples. So in Bertrand de Born’s poem the cauda repeats both the rhymes of the frons. In Guillem IX.’s stanza the versus take up the only rhyme of the frons, adding a new one of their own. This concatenatio, however, was not considered absolutely necessary; and Peire d’Alvernhe, for instance, one of the most finished troubadours, introduces into the versus of his stanza rhymes entirely different from those found in the pedes. Another kind of concatenatio is effected by adding, either before or after the volta, a line which contains the rhyme of the other part of the stanza. How the troubadours contrived not to disturb the flow of their melody by this new and seemingly inharmonious element it is difficult to say. Perhaps it was sung to a sort of recitative or arioso of its own, which served as a prelude to the new melody of the second part. The meaning of this will be clearly shown by the following stanza by Raimbaut de Vaqueiras:

This stanza consists of a frons and two versus of three lines each. The first four lines are a whole in themselves; and after them the frons ought to be concluded by the volta, as is indicated by the punctuation after ‘pojar.’ But the poet adds a fifth line, in order to introduce a rhyme from the versus, and in this way effects the concatenatio he needs. There was yet another way of connecting the two parts of the stanza which might in a certain sense also be called concatenatio, but which was seemingly unknown to Dante. The following stanza of Cercalmont’s,

consists evidently of a frons and a cauda, of three lines each. The fourth verse does not belong to either, and its rhyme is not to be found in the same stanza; therefore it is to be called a clavis. But nevertheless it is of use for the purpose of connecting the frons with the cauda; for, by considering it as a kind of centre, and going from it to the beginning and to the end of the stanza, the reader will see that the two parts exactly agree as regards the length of the verses, and even their feminine rhyme.

In the eleventh chapter of his ‘De Vulgari Eloquentia’ Dante speaks of the relations between the parts of a stanza so far as the number and length of the lines are concerned. He enumerates and explains no less than nine different cases. For the present purpose, however, these are scarcely of any importance; for the rules given by him cannot be traced to the poems of the troubadours. In this respect the Provençal poets seem to have exercised great liberty, being protected against choosing bad proportions by their refined sense of harmony.

It still remains to mention a form of Provençal poetry, of which the essence cannot be explained nor the rules defined without the aid of Dante’s statements. This is the tornada, a kind of postlude or envoi to the stanza, in which the poem is dedicated to the lady-love or the protector of the troubadour, who is generally introduced under a senhal or pseudonym to keep the real name secret. Peire Vidal always calls the beautiful Azalais ‘Vierna,’ while the senhal ‘Bels Castiatz’ designates his noble protector Sir Aimeric de Monrial. The ‘Leys d’Amors’ says that, as a rule, there ought to be two tornadas, the first of them containing the senhal. This, however, is not confirmed by the majority of Provençal poems, which contain many examples of a single tornada. In another respect also the ‘Leys d’Amors’ is very inaccurate. It says that the tornada must be identical in form with the latter half of the stanza, if this consists of an even number of lines, adding or leaving out one line where the number of the verses cannot be divided by two. But this applies only to those stanzas where no division is to be found. Where there is a diesis the rule is quite different, and can be learned only from Dante, who speaks of the tornada in the ‘Convito,’ where he derives the word from tornar, owing to a part of the oda returning in it. Accordingly the rule in the divided stanza is that the first tornada repeats the metrical form and rhyme of that part of the stanza—cauda or versus—which stands after the volta. Where there is a second tornada it generally agrees with the first, being, however, always the shorter of the two. All this, of course, the author of the ‘Leys d’Amors’ could not know, because he was ignorant of the metrical and musical formation of the stanza; but it is of the highest importance for the study of Provençal versification, and shows again the great value of Dante’s work in that respect. The above-stated rule is confirmed by so many examples from the canzos of the troubadours that it is scarcely necessary to bring new evidence for it. It will be more useful to mention some of the more important exceptions, which in this, as in other cases, ‘firmant regulam.’ If the last stanza of a poem ends with two versus, the tornada sometimes repeats only one of them. Sometimes also part of the cauda remains unrepeated. In other cases the tornada repeats exactly the metre of the cauda, but differs slightly from its rhymes. In a sirventes of Marti de Mons, which was written in the fifteenth century (1436), and by which the poet gained the ‘englantina’ in the competition of the Academy of Toulouse, the cauda of the last stanza consists of the following four verses:

The first of these lines serves as concatenatio; and for that reason its rhyme agrees with the first part of the poem. In the tornada this reason of course did not exist; and therefore the poet very skilfully rhymes the first line with the last line of the tornada instead of making it like the first verse of the cauda. The tornada, therefore, is this:

In many cases also there is no tornada at all, or it may be said to consist of the last stanza of the poem, if in this the senhal and dedication are introduced.

The principles insisted upon in the foregoing remarks may perhaps best be illustrated by an accurate metrical analysis of the subjoined canzo of Bernard de Ventadorn. For the purpose in question this poem has the double advantage of presenting a great complication of metrical rules, and of showing at the same time how the troubadours succeeded in combining such a complicated structure with the beauty of genuine poetry.

This poem consists of six stanzas and a tornada. The length of each stanza is seven verses, that of the tornada three. In each stanza there is, according to Dante’s expression, a diesis or volta, for there is the required reiteratio unius odæ. This reiteratio takes place before the volta, while after the volta no division is possible. The stanza therefore must be divided into two pedes of two lines each and a cauda of three lines. According to rule, the metrical division is marked by a strong grammatical break (at least a semicolon), the only exception being stanza b, where a punctuation in the volta is not possible. The tornada repeats as usual the form and rhymes of the cauda; and in it the poem is dedicated to the poet’s lady-love, who is addressed by a senhal. Bel Vezer was in this case Agnes de Montluçon, wife of the troubadour’s lord and protector, who raised him from the state of a common servant and gave him the first lessons ‘del gay saber.’ The verse of the stanza is decasyllabic; it occurs with masculine and feminine rhyme. The stanza may be formulated metrically by using capital letters for the decasyllabic line, and adding to them the sign ~ for the accen greu; the volta may be marked by a semicolon, and the division of the pedes from each other by a colon:

A B~: A B~; A A B~.

Hence it appears that in each stanza there are only two different rhymes, the cauda repeating those of the pedes, which is the simplest form of concatenatio. Moreover, each couple of stanzas have the same rhyme, or are coblas doblas; the poet in consequence had to find, three different times, eight masculine and six feminine rhyming words, which, though not a very difficult task in the langue d’oc, required a certain amount of skill. In the last group of stanzas this number was increased by the tornada to ten and seven respectively. Nevertheless there are only two cases of the same words with the same meaning occurring in the rhymes, or of motz tornatz en rim as the ‘Leys d’Amors’ calls them. Both these cases, f 3 = g 1 and e 4 = g 3, occur in the tornada, where they were not as strictly forbidden as in other positions. On the other hand there are many examples of ‘rime riche’ in its masculine as well as feminine form. The former or rims consonans lejals are a 1 = b 1 = b 6, c 6 = d 1, d 5 = d 6, e 3 = e 5 = f 5. The latter or rims leonismes parfaitz are e 2 = f 2. But in all these cases it is very doubtful whether these rhymes were intentional, since they exhibit no system or order. A remarkable sense of the effects of sound is shown in the alliterative use of the letter f in b 4 and 5, by means of which the pains of the unhappy lover are onomatopoetically expressed.

In the verse of ten syllables the cæsura is always of importance: it therefore remains to take note of it. The cæsura, where it appears feminine, has been divided into the epical and the lyrical, the difference being that in the epical pauza del bordo the accent always remains on the fourth syllable, after which another unaccentuated syllable is added to the first hemistich, while in lyrical poetry the accent itself is removed from the fourth to the third syllable. In the present poem the pauza con accen agut is by far the more common; where it occurs with accen greu it always takes the lyrical form. These cases are d 5 and 6, e 1, and f 1.

It may be worth while to notice that once, d 5, the word domna is placed in the lyrical pauza. The troubadours, in addressing their lady-loves, seem to have liked this particular position of the word, by means of which it received a certain emphasis. In many canzos of different troubadours there are instances of this device; in one of Guillem de Cabestanh’s songs it occurs twice, or, according to a Parisian manuscript in which the poem is also preserved, even three times.