CHAPTER TWO

APPROACHES TO THE STUDY OF FORM IN THE TROUBADOUR SONGS

Twentieth‑century Studies

The main principles underlying our understanding of musical form in the troubadour songs are fairly straightforward and can be traced back to Dante. The musical phrase corresponds to the line of verse, and the repetition (or lack of repetition) of musical phrases determines the melodic form of the song. It is conventional to compare this with the metric form, determined by the rhyme scheme and number of syllables per line. This could be criticized as a linking of unlike elements, since the rhyme sound occurs only at the end of the line of verse, while the melodic phrase fills the entire line, but there is little doubt that the troubadours themselves established the convention, as will become clear not only from Dante's discussion, but from a study of the sources themselves.

The whole issue of musical form has become problematic in relation to the troubadour and trouvère repertoires, however, for various reasons. Foremost among these has been the negative influence of Friedrich Gennrich's formal studies of the 1930s.^[1] The distortions and misconceptions that compromise Gennrich's treatment of form in medieval song seem to have cast a pall over the topic, and scholars have preferred to address other issues instead. Furthermore, these other issues and questions have sometimes been taken as implicit critiques of the legitimacy or value of the study of form, and this in turn has only contributed to the neglect of the subject. For example, questions concerning the variability of the repertoire, orality, and compositional context are often contrasted with the supposed fixity that has been associated with the notion of form as it was understood in earlier generations.

Even one's opinion on the intrinsic value of the melodies can have a bearing on the relative importance assigned to the study of their form; if the troubadours were amateur composers with minimal skill, relying on highly conventional formulas and simple, primitive melodies, one would not expect them to display much craft in their musical structures. One of the premises of the present study is that the troubadours in reality possessed a relatively sophisticated sensitivity to musical construction, on a par with their interest in poetic structure, and that musical design parallels and complements poetic design in the aesthetic of the canso.  Before examining the evidence of the sources in support of this view, however, it will be useful to clarify the relevant problems and issues as they have developed from the earlier studies of Gennrich to the present.

The problematic aspects of Gennrich's approach to form in medieval song and in the troubadour repertoire concern both the broader theoretical premises involved and the more practical details of constructing analytical graphs for the individual songs.^[2] Gennrich felt that he could detect a kind of evolutionary will towards form or "Formwille" in all the music from Latin hymns to the fixed forms of the fourteenth century, the latter standing as ideal types towards which earlier forms were groping.^[3] The effect of Gennrich's historical outlook can be seen most easily in his catalogue of troubadour song forms which accompanies his edition of the songs.^[4] Along with the graph of the musical form, each song is labelled according to one of the categories developed by Gennrich to cover all of medieval song. Least problematic of these is the form he calls oda continua; the term is taken from Dante, who used it to designate melodic forms with no repetition of entire musical phrases.^[5] (Gennrich makes no reference to the three other kinds of form discussed by Dante, however.)  This is one of the most common and simple designs used by the troubadours, and is generally not difficult to recognize. In Gennrich's system, however, the oda continua belongs to the broader category of the hymn type, and is thus supposedly descended from the Ambrosian hymns of the fifth and sixth centuries, a derivation that is entirely hypothetical. He also qualifies a number of the songs given this designation by calling them "oda continua with repeats," which would seem to be a contradiction in terms.

Two other of Gennrich's forms allegedly originating with the hymn are those labelled canso and "rounded canso," and he applies the terms to the forms often referred to as Bar forms. In the "canso" form the first two phrases are immediately repeated and the rest of the song is through-composed, while in the "rounded canso" one of the first phrases is repeated at the end. Here again the derivation from the hymn is somewhat dubious, given the universality of the form; it has even been compared with the Greek epode with its pattern of strophe/strophe/antistrophe. Nor can one justify canso as a valid term for this musical form; it was used by the troubadours as a genre designation based mainly on the subject matter of the poetry, and any high‑style song with courtly love as its theme was known as a canso. The vast majority of the songs in the troubadour repertoire is composed of cansos, but the term says nothing about any song's musical form; a canso could have an oda continua musical form just as readily as an ABAB form,^[6] or indeed a different form altogether.

  Much more problematic are Gennrich's sequence‑related categories such as the "lai‑segments," "strophic lais," and "reduced strophic lais," which he applies to a fair proportion of the songs. Leaving aside the thorny question of possible links between the sequence and the lai, the basis for all of these categories is the immediate repetition of one or more phrases, which is a typical feature in lais, often referred to as "lesser responsion." Of course, such sequential repetitions do occur in the troubadour melodies, but again, the principle is such a common one, and its employment so varied, that the implied fixity and reality of categories such as "lai‑segment, 3rd group" soon evaporates when one confronts the sources more directly. It is not too difficult to find confusing and contradictory classifications even within the terms of Gennrich's system, as can be seen in his classification of the song "Fis e verais e plus ferms que no soill" by Guiraut Riquier (P‑C 248,29), which he graphs as follows:^[7]  

The only difference between the form of this song and those labelled "rounded canso" by Gennrich is that in the former, both of the initial phrases are repeated at the end instead of only one of them. If one groups together all the songs that demonstrate the ABAB form in their first half, one will find that there is a wide variety of ways that material can be repeated in the second half, should this option be chosen. What Gennrich calls a "reduced strophic lai," is no more than another variation within the broad category of ABAB forms. By adding qualifiers such as "reduced" and "segment," Gennrich has created formal types that correspond only to the efforts of his imagination, adding another hypothesis to the underlying one linking the courtly love‑song to the sequence or lai.

By elevating his formal categories to the status of ideal essences, comparable to geometric or mathematical forms, Gennrich ignores both the variability of the medieval sources, and the role of other elements in the overall structure of a song. Almost a quarter of the surviving troubadour songs with melodies are transmitted in more than one version, yet Gennrich gives only one formal graph for each, as he gives only one transcription for each, confident that he has ascertained the best version. While the formal outline of a song is generally more stable than melodic details, there are cases where the different versions present significant differences in form as well.^[8] Nor can the music be considered the ultimate formal determinant in this repertoire. The poetic form is at the very least equally important, and when other aspects such as genre, register, thematic content, and stylistic features of the melodies are considered together, Gennrich's categories, based on purely musical liturgical predecessors, appear less and less defensible.

The flaws that compromise Gennrich's work have long been recognized, and some of the older prejudices, deriving largely from nineteenth‑century ideologies, have been corrected. Since the 1960s and 70s, both musical and literary scholars have re‑examined issues such as orality, performance context, transmission, and editorial procedures, with particular emphasis on the variability of the repertoire. Gone is the confidence of an earlier generation of scholars, who believed they could establish an author's original version of a text through painstaking comparison of manuscript sources and construction of stemmata. Instead, the uncertainties surrounding the genesis, transmission, and purpose of the surviving sources have tended to undermine modern assumptions about the objective nature of a text. The earliest sources that contain troubadour melodies were produced in northern France in the middle of the thirteenth century, a good century and a half later than the period of activity of the earliest known troubadour, William VII, Count of Poitiers (1071‑1127). There is little doubt that the songs were created without the use of notation, and the exact nature of the interaction between written and oral transmission remains uncertain. Some scholars go so far as to deny the validity of the idea of a first version of a tune as the product of individual invention, preferring a model based on the adaptation of common formulas; the variants found in multiple versions are seen as a possible reflection of performers' and scribes' modifications or re‑interpretations that are all equally valid.^[9]

The changing perspectives of the past decades have also raised issues that challenge the relevance of formal studies, and these must be addressed if such studies are to be justified. One of the most fundamental questions concerns the role of the music in the troubadour aesthetic of song‑making. While Gennrich had implicitly assigned a higher role to the music, a different opinion is now finding its way into standard reference works, as the following quotation from the Dictionary of the Middle Ages makes clear: "The poet was expected to compose a new melody for each song, but the melodies are not sophisticated and are even quite conventional, presenting less interest to musicologists than the texts do to literary scholars."^[10]

As with Hendrick van der Werf's comments in his influential study The Chansons and various articles, one derives an image of the troubadour musical tradition as a domain of rank amateurs whose simple melodies all sound alike. For van der Werf, and others who stress the oral nature of the compositional context for troubadour song, there is little sign of art or craft in the surviving notations; the melodies functioned in a limited utilitarian manner as a vehicle for the words. In The Chansons, van der Werf gives a balanced and reasonable assessment of what information there is pertaining to the transmission of the troubadour melodies.^[11] He suggests they were composed and passed on orally until about the middle of the thirteenth century, when collectors and scribes began to preserve the songs in manuscripts. The combined effects of performers' and scribes' alterations can for the most part be understood as legitimate variants, attributable to medieval attitudes and conditions, rather than errors, as older editors believed, and give witness to an inherent fluidity in the tradition. 

In his chapter on the form of the melodies and poems,^[12] van der Werf does not explicitly connect his opinions there to his account of the transmission process, yet there is a logical progression from the earlier chapter, through the immediately preceding one on melodic characteristics, to the discussion of form and versification. When discussing melodic structures, van der Werf refers only to the most elementary tonal formations such as interval chains or recitative‑type melodies which are common to a large number of melodies. In the concluding section of the final chapter, "The Chansons as Creations of a Notationless Culture,"^[13] van der Werf attributes the simple, formulaic and conventional nature of the melodies to their having been composed in an oral tradition, without the assistance of notation. He suggests that the use of writing in the creation of the poems may have contributed to the much higher level of craft found in the texts as compared with the melodies, which he says, "sound to us like remembered improvisations in a very traditional and simple fashion."^[14]

Because of the role of improvisation in the creation and performance of the songs, "it was difficult for the performers to retain it [i.e., the melody] precisely as made up by the composer and, because of the lack of design, it was impossible for the notators to reconstruct the original."^[15] "Real composition," for van der Werf, only begins with "written composition," and he likens the songs of the troubadours and trouvères to "folk music," recalling his earlier references to the primitive melodic types discussed by Curt Sachs.^[16] With reference to melodic form, van der Werf finds "convention and lack of sophistication" to be typical for three main reasons. First, he blames the troubadours and trouvères for employing only two stereotyped musical forms, the ABAB form with a repeat of the first pair of phrases and varied cauda,^[17] and the oda continua form without repeats; this is in contrast to the wide variety of versification schemes devised by the troubadours and trouvères. Secondly, "the lack of attention to melodic form," says van der Werf, "is especially attested by the varying relationship between the melodic lines for the pedes. One would expect the melody for the second pes to be a literal repeat of that for the first pes. Yet in reality the relation of the second melodic sentence to the first may be anything from a literal repeat to an elusive echo."^[18] Furthermore, he finds that in multiple versions the treatment of these repeats may differ between manuscripts, and that it can be hard to distinguish varied repeats from new material. The third area in which the troubadours' and trouvères' "lack of interest" in musical form shows itself is the relation between poetic and musical forms. For van der Werf, there is almost no relation between the two in this repertoire, since the rhyme schemes and musical forms do not parallel each other throughout the entire course of any individual song except in rare instances; he cites an example of such an "ideal agreement between musical and textual form" in a song by the trouvère Conon de Béthune to emphasize its exceptional nature.^[19]

It is clear that what van der Werf views as the stark contrast between the troubadours' skill and refinement in poetic composition and their musical composition is seen by him as a natural outcome of the oral nature of the musical tradition; the possibility that a more developed sense of melodic construction could occur within such a context is practically ruled out by him.

Van der Werf's references to improvisation recall the compositional model proposed for chant by Leo Treitler, which was inspired by Parry and Lord's investigations into the performance and composition of the epic.^[20] Essentially, this model explains both the family resemblances and variants among chants as resulting from a process whereby an underlying schematic structure is "actualized" anew in different performances which maintain the basic elements of the structure while consciously or unconsciously altering the specific "surface" details. A singer would remember the most salient features of the chant's scheme or plan, much the way a reciter of epic would have in his mind the plot of his story. He would then realize this scheme by drawing upon his repertoire of stock melodic formulas, according to the specific textual and liturgical exigencies of the case at hand, the way an epic poet would draw upon his stock of phrases and epithets already designed to fit the metrical requirements of the poem. One of the main purposes of such a model was to provide an explanation for the transmission of such a large body of chant (and of very long epic poems) without the use of writing; memory is combined with active re‑creation within a set of given constraints.

It is sometimes forgotten, however, that oral traditions may differ considerably from one another, and all do not necessarily conform to the models proposed for the epic by Parry and Lord, and adapted for chant by Treitler. More specifically, in a lyric tradition such as that of the troubadours, neither textual nor musical aspects pose any of the memorization problems associated with chant or epic. The average troubadour song has about seven stanzas of seven to twelve lines of verse.^[21] The texts were certainly composed and transmitted orally, since the ability to read and write was possessed by very few of the troubadours;^[22] the variants that occur in the texts may be attributed in part to the freedom that both performers and scribes felt to change or "improve" the poem as they received it, but in no way does the compositional process involved in producing such a body of lyric poems resemble that of epic.^[23] As the many references to this point in the poems themselves make clear, a great deal of care was lavished on the "polishing" and "refining" of  the troubadour canso to produce a highly wrought artifact of technical complexity and poetic subtlety to match any comparable poetry from other periods.

As every detail of the poetic text was normally created orally, every detail would also have been transmitted orally and memorized word for word. If this was so for the texts, there is no reason to think it would necessarily have been otherwise with the music. It has been estimated that the average rate of production for a troubadour poet was no more than two or three songs per year; if he was concerned about the quality of his melody, he would certainly have had plenty of time to work on every detail of it and teach it accurately to his jongleur. And the literary evidence suggests the melody was just as important as the words, not less, as van der Werf believes; the frequent couplets like los motz e.l son testify to this, as well as the judgments of particular troubadours' composing talents contained in the vidas and razos.^[24] Van der Werf compares the supposed simplicity of the troubadour and trouvère melodies with the thirteenth century motet as an example of genuine composition dependent on notation. A fairer comparison would be with the monophonic music of other genres such as the pastorela, where the contrast between the so‑called "high style" and the "low style" is evident in the music as well as the text and versification.^[25]

It is true that one finds typical melodic formulas for phrase initia and cadences, but then there are plenty of stereotyped conventions in the poetry as well, and in musical traditions that rely on notation. The use of conventional formulas is a common feature of the medieval aesthetic that values the type above the individual; each troubadour song is marked by features that make it a partial reflection or echo of all the others of the same genre. This feature in itself need imply nothing about the relative degree of melodic complexity or interest to be found in a piece.

Returning to van der Werf's more specific complaints about the troubadours' and trouvères' lack of interest in their musical forms, the charge of conventionality and a reliance on no more than two stereotyped forms is simply not true if applied to the troubadour repertoire. (The remarks may have more validity in regard to the trouvères, but van der Werf generally fails to distinguish the two repertoires, which are by no means alike in all respects.)  As the following chapter will show, the troubadours were as inventive in the realm of musical form as they were in the technical aspects of poetic form. Basic types are clearly recognizable, as are basic rhyme schemes, but within the spectrum ranging from songs with no repetition of musical phrases, to those in which all phrases are repeated, there is an enormous range of possibilities for devising unique musical forms, and the troubadours exploited these to the full. Van der Werf's interpretation of the varied repetition of phrases in the pedes as a sign of carelessness seems unwarranted; indeed, such modified repeats can be viewed quite differently as attempts to avoid the very conventionality he found so objectionable, by shunning the obviousness of an exact repetition. With regard to the supposed lack of agreement between the poetic and musical forms, van der Werf is here demonstrating a misunderstanding that some other writers have also been prone to. For it is van der Werf's "ideal agreement," with every rhyme repetition matched by a musical phrase repetition, that would have produced the most trite, mechanical, and tedious musical results; rhyme sounds, being limited to the end of a line only, unlike an entire musical phrase, can tolerate much more repetition than the latter. In this case it is the troubadours as composers who have shown the greater sensitivity to the aesthetics of song composition, by opting for a less predictable solution to the problem of text/music relations. Rather than a lack of any relation between poetic and musical forms, one finds a range of ways in which the two can be conjoined in a meaningful way, and one of the intentions of the present study is to demonstrate these in a systematic manner for the whole body of surviving songs with music.

Secondary Medieval Sources

Turning now to the medieval sources, it must be acknowledged that the writings of the troubadours contain almost no reference to musical form per se. The lyric poems, as well as the vidas and razos, do contain numerous references to music, but these are of a general nature; melodies are characterized as gais, joios, and plazens, for example, and different troubadours' composing and singing skills are singled out for praise or blame. There are also a number of suggestive remarks about the adequate joining of music and text, such as the following quotation from Peire Vidal (P‑C 364,2):

Ajostar e lassar

sai tan gent motz e so,

que del car ric trobar

no.m ven hom al talo,

quant n'ai bona razo.^[26]

This has been translated, "I know how to couple and lace words and music together so gracefully that no one can compete with me in the precious, rich style, when I have a good theme for it."^[27] Such comments, along with the frequent couplets such as el motz e.l son, indicate at least that both words and music were equally products of careful craftsmanship and that their adequate conjoining in the canso was also a matter of skill rather than chance.

For more specific discussions of the technical aspects of versification and music our only sources from the period are the Leys d'amors^[28] and Dante's De vulgari eloquentia.^[29] Even these are somewhat removed from the heart of the tradition, and so must be used with caution. The Leys provide an exhaustive compendium on poetic composition, but say very little about music. They represent one side of an effort to uphold and continue a tradition that was no longer alive. References to poets from the classic period of the twelfth century are few, and the whole organization and pedantic tenor of the treatise aligns it with the Latin clerical tradition of grammatical and rhetorical treatises; the living art of the troubadours is tirelessly codified into an endless list of rules and prescriptions. The scattered references to music that do occur in the Leys give the impression that it was no longer integral to the poetry by the fourteenth century. In so far as musical and poetic structures are interdependent, however, the Leys provide a wealth of information on the relevant poetic aspects such as the division of the stanza, metrical structures, rhyming patterns and all the technical aspects of versification.

It should also be borne in mind that Dante's main concern in the De vulgari eloquentia was with the poetry of his immediate predecessors and contemporaries in Italy, and that his moral and spiritual universe is very different from that of the troubadours, if not alien to it. In spite of the weight of his authority, and his citations of troubadour poets in the De vulgari and the Divina commedia, scholars have shown that his familiarity with the poets of southern France was in fact rather limited.^[30] Dante's treatise nevertheless remains our most important source of information on melodic construction in the troubadour canso, and his terms are still the ones most commonly used today. The relevant sections of the treatise will therefore be examined now to see what light they may shed on the role of musical form in the aesthetic of the canso.^[31]

Music is discussed in conjunction with Dante's treatment of structural features of the canso in the De vulgari Book II, sections viii‑xi. In section iv of the same book, however, Dante gives the following definition of poetry in its essence:  

[Poesis . . .] nichil aliud est quam fictio rethorica musicaque poita.^[32]

This can be translated, "Poetry is none other than an invention poetically expressed according to rhetoric and music."^[33] As the commentator P. V. Mengaldo notes, poesis here is used in a collective and abstract sense to refer to imaginative constructs in general, not only the making of poems. Likewise, "music" is employed here not in the sense of "melody," but with the Boethian signification of "numerical structure or arrangement of parts."^[34] A poetic composition would thus have a musical component whether or not it had a melody. It is well to note this important distinction, even though it becomes difficult to separate the two senses in parts of Dante's later discussion.

While poesis need contain no element of "practical" music, a cantio in Dante's time could still include poetic types normally sung to a melody. The term, however, cannot be taken as a Latin equivalent for the Provençal canso, which was limited thematically to the singing of fin'amors, for Dante has extended the term's range to include other subject matter treated in an elevated style, be it moral, philosophical, spiritual, or political. A more accurate parallel term might therefore be vers as it was employed by the later generation of troubadours and in the Leys d'amors, namely as an inclusive term covering the canso as well as other genres such as the sirventes, planh, etc. In section viii Dante asks whether the term cantio should refer to the composition of harmoniously disposed words, or to the melody ("Preterea disserendum est utrum cantio dicatur fabricatio verborum armonizatorum, vel ipsa modulatio."^[35]) He answers that the "modulatio" is never called  "cantio" but either "sonus, vel thonus, vel nota, vel melos," and that the term cantio is used for the verbal creations in question even when they are in written form and no one sings them.^[36] The following is the definition Dante then proposes for the cantio, which is for him the highest and noblest of all poetic forms:

Et ideo cantio nichil aliud esse videtur quam actio completa dicentis verba modulationi armonizata.^[37]

The cantio is thus "an action completed by one who composes words harmoniously in accordance with the (numerical structure of the) melody." It is further characterized by its tragic (that is, "lofty") style, by a unity of thought, by stanzas of analogous structure, and by the absence of a refrain. The melody is not absolutely essential, but its structure is, and this may be abstracted from the sounding melody per se as a harmonious arrangement of parts, a numerical structure. It is this structure (along with the melody) that remains invariable from one stanza to the next in the cantio, and Dante next examines the stanza as an organism "in which all the technical craft of the cantio is contained."^[38] The term, in Latin stantia, had only recently been applied to this poetic unit, and Dante exploits the architectural metaphor it contains to convey the importance of the underlying structural framework in a poem.^[39] The stanza is a privileged locus for the elaboration of a harmonious structure which must be carefully designed and proportioned to receive the other elements, the way the skeletal bone-structure of a living organism is the foundation for its external form.

According to Dante, the technical craft of the cantio ("ars cantionis") consists of three main elements: the "cantus divisionem" or melodic division; the "partium habitudinem" or disposition of parts; and the "numerum carminum et sillabarum" or number of verses and syllables.^[40] There is a certain amount of overlap in the presentation of these three areas, making the distinction between them unclear. The third element, comprising the length of the stanza or sections of the stanza, and the meter, is not treated because the treatise was left unfinished. In the section devoted to the "partium habitudinem," however, the relative excellence of various meters is considered, as well as the number of lines in each part of the stanza, according to the options envisioned by Dante. The second element in fact also includes the first, and is seen to be the most important of all, as Dante states at the beginning of chapter xi:

 

Videtur nobis hec quam habitudinem dicimus maxima pars eius quod artis est. Hec etenim circa cantus divisionem atque contextum carminum et rithimorum relationem consistit.^[41]

These are all elements pertaining to the cantio as a manifestation of formal design: the melodic structure, the choice and arrangement of meters, and the rhyming-pattern. That the melodic structure is understood by Dante as an abstract pattern independent of the melody itself becomes apparent in chapter x where he outlines the possible types of melodic structure available.

The first type is the oda continua, in which there is no repetition of any of the melodic phrases, and no diesis or division of the stanza. Dante says that this form was used in all the songs of Arnaut Daniel, and that he himself had used it in his "Al poco giorno e al gran cerchio d'ombra."^[42] Of the 18 cansos of Arnaut Daniel whose texts have survived, two are preserved with music, "Chanzon do.l moz son plan e prim" (P‑C 29,6), and "Lo ferm voler q'el cor m'intra" (P‑C 29,14). Both can be considered to conform to the oda continua melodic type, but as for Dante's canzone, no one would pretend that Dante had ever composed a melody for it, or intended it to be sung. In fact, from this and subsequent statements, it is apparent that Dante assumed that the pattern of melodic phrase repetition was strictly congruent with that of rhyme repetition within the stanza, and that he bases his whole discussion of melodic form on rhyme-schemes, not music. It is entirely possible, and perhaps probable, given the ambiguities of his references to music, that Dante's knowledge of troubadour poetry was derived entirely from manuscripts, and that he did not hear a canso sung with its melody. When he says, therefore, that Arnaut Daniel used the oda continua for preference, he is reflecting this poet's predilection for stanza forms in which there is no repetition of rhymes within the stanza, as in his famous sestina "Lo ferm voler" (P‑C 29,14). (The other song of Arnaut with music does not have a similar rhyme-scheme, however.)  This type of rhyme scheme is used only rarely by other poets, though, for it goes against the principles of versification to have no rhyme sound answered within the stanza. And yet, the oda continua as a musical form was very common among the troubadours; whether intentionally or not, Dante has singled out one of the most important melodic designs in the repertoire.

The other three types of melodic form described by Dante have phrase repetition and divide the stanza into two large sections. The possibilities are succinctly described in the following paragraph:

Quedam vero sunt diesim patientes: et diesis esse non potest, secundum quod eam appellamus, nisi unius ode fiat, vel ante diesim, vel post, vel undique. Si ante diesim repetitio fiat, stantiam dicimus habere pedes; et duos habere decet, licet quandoque tres fiant, rarissime tamen. Si repetitio fiat post diesim, tunc dicimus stantiam habere versus. Si ante non fiat repetitio, stantiam dicimus habere frontem. Si post non fiat, dicimus habere sirma, sive caudam.^[43]

 

Using Dante's terminology, the forms described may be graphed as follows, X standing for any number of unrepeated phrases:

           1) pedes cum cauda (or sirma)   —    AB AB   /   X   or

                                                                    ABC ABC /   X

           2) frons cum versibus           —        ABCD... / YZ YZ

           3) pedes cum versibus         —        AB AB   /  YZ YZ

Dante's categories have the quality of simplicity and symmetry that one finds in the syllogisms of logic; they encompass all the possible combinations within a very limited set of options, namely, repetition or non‑repetition before or after the diesis. As such, they are fascinating as a kind of distorting lens through which the concrete reality of the troubadour canso is both illuminated and obscured at the same time. Since it is likely that Dante had in mind the textual stanzaic forms rather than the musical, we may first consider the degree to which he is reflecting the actual state of affairs with regard to forms of versification in the troubadour repertoire. Considering that Frank's Répertoire métrique^[44] lists a total of 884 different rhyme schemes, it is not surprising that Dante's patterns correspond to only a handful of those that were used. The symmetrical pedes in the form ab ab is without a doubt one of the more common rhyme schemes for the first half of the stanza, and the use of three or more pedes, or pedes consisting of more than two lines, is also readily attested in practice, and conforms to Dante's types. One may also find versus in the same form, which we may represent as cd cd or ef ef, but equally common choices if not more common, when the second half is constructed symmetrically of repeated rhymes, are the patterns cc dd and cddc. The latter pattern, abba, is also one of the most popular for the first half of the stanza. As for the frons and cauda without rhyme repetition, these are extremely rare. At the same time, there are several hundred schemes that simply fall outside Dante's categories.

If we now look at these categories as melodic structures, once again we find both a striking correspondence and lack of correspondence with the facts. The pedes cum cauda type is an accurate reflection of a genuine and common form in the repertoire, and is an exact equivalent of what Gennrich labelled a canso. In fact exactly half of the songs with an ABAB melodic structure in their first half have no repetition of whole phrases in their second half, just as the pedes cum cauda stipulates. The other half of this group of songs, however, conforms much less closely to Dante's pedes cum versibus category. Only three or four songs in the whole repertoire have the form which may be diagrammed AB AB / CD CD;^[45] all the others with an AB AB beginning have different forms and degrees of repetition in their second half. As for the frons cum versibus type, it would have to be considered virtually unrepresented in the troubadour songs, since every song that has any repeated phrases in its second half also repeats one or more phrases in its first half.

It must be admitted that even if the musical component was for Dante more virtual than actual, he nevertheless has identified two of the most standardized and popular musical forms in the troubadour repertoire. It is possible that his recognition of the oda continua comes from his admiration of Arnaut Daniel's sestina, in which rhyme scheme and musical form are identical; if the music was unknown to Dante, then his recognition of this form would have been based on the poetic form, and thus more or less fortuitous in regard to its importance as a musical form. The pedes cum cauda form is more common as a musical than a poetic form, however, so it remains very difficult to gauge the nature and degree of Dante's awareness of troubadour musical practice.

From the more general point of view of the role of musical form in the troubadour aesthetic, Dante's discussion is invaluable as the only one of its kind that comes even close to being contemporary with the main period of troubadour activity. One must exercise caution in this area as well though, for it is uncertain how much of Dante's perspective on the canso would have been shared by the troubadour poets who were his models for a high‑style poetry in the vernacular. There is a strong Boethian strand in his conception of poetry and music, which I suggest would have been alien to the majority of the Provençal poets.^[46] However, Dante's assimilation of the poetic and melodic structures could be taken as a sign that the two were viewed as functionally comparable in his time, and that just as the invention of unique and elaborate stanzaic structures was an important element in the troubadours' aesthetic of form, and appreciated for its own sake, the invention of such a variety of musical forms can be understood in terms of the same aesthetic.

The abstractness of a formal design may seem far removed from the physical and emotional presence of the human voice in the immediacy of performance, whether one is considering a musical or poetic structure. As Frank's Répertoire amply documents, however, this concern with form was a real one in terms of the technical side of the poetry, and there is no prima facie reason to think that the music did not form part of the same preoccupation. While relatively abstract when taken alone, the musical and poetic formal designs together serve, to paraphrase Dante, as the underlying framework for the conjoining of the other components in the song, and thus provide an excellent paradigm for the analysis and understanding of text/music relations in the canso. As will be seen, the graph of the musical form can also act as an eminently useful framework for the study of other levels of compositional practice, such as melodic relationships at the sub‑phrase level, and stylistic features in the melodies. The following section considers these and other aspects of musical form as seen in the troubadour songs transmitted with their melodies.