Accentuation and Duration in the Music of the
Cantigas de Santa Maria
Hendrik van der Werf
University of Rochester, Eastman School of Music

In Studies on the Cantigas de Santa Maria: Art, Music, and Poetry. Proceedings of the International Symposium on the Cantigas de Santa Maria of Alfonso X, el Sabio (1221-1284) in Commemoration of Its 700th Anniversary Year - 1981. Madison: Hispanic Seminary of Medieval Studies,1987, 223-234.

In the three Cantigas de Santa Maria manuscripts, which contain musical notations, melodic intervals are specified with great precision and, as far as we can determine, with admirable accuracy.1 The notations do not contain indications for melodic accents; however, and with regard to duration of individual pitches, the scribes' notational practices may well have presented us with more problems than solutions. Higinio Anglés, in his pioneering edition, (1943­1964), evaluates the indications for duration in an ambiguous, if not contradictory manner. He considers the notation of the Toledo manuscript as the least perfect of the three, in that only its single notes, not its ligatures (groups of notes over one syllable) indicate duration. He further states that the notation of the two Escorial manuscripts is "mensural in all its details"; he even uses the term "notación mensural perfectísima" (1943:11, 48). His transcriptions, however, display an unmistakable deviation from his own pronouncements, and this ambivalence, by:and large, seems to have resulted from two causes. He seems to have taken for granted that the duration of individual notes was precisely measurable and could have been represented adequately in the mensural notation of the late thirteenth century. In addition, .the medieval mensural notation itself is not without its ambiguities. When the manuscripts for the Cantigas were compiled, the notational symbols, or neumes, which were used in them, already had been in existence for some time and, more importantly, they had acquired rather diverse meanings. It is widely accepted that the neumes had no mensural significance in the notation of Gregorian chant; in certain motet collections, however, they indicated duration precisely.2The mensural significance of a neume is determined in part by its shape (most notably the presence or absence of a stem) and in part by the duration of surrounding notes. With few exceptions, the symbols of mensural and non­mensural notation look alike. In some manuscripts, therefore, distinction between mensural and non­mensural notation depends exclusively on the notational context. A few shapes, however, were designed specifically for mensural notation, most notable among them is the ligatura cum opposita proprietate, i.e., a grouping of two notes, the first of which bears an upward stem on its left side indicating that both notes are to be read as semibreves. This ligature--either by itself or as a part of a larger cluster of notes--appears fairly often in both Escorial manuscripts, lending to their notation a convincingly mensural mien. A close examination of almost any of the melodies in these sources, however, reveals some instances in which the order of the neumes does not conform to the requirements of mensural notation. In addition, a few of the neumes from the Escorial manuscripts appear frequently in plainchant, but rarely, if ever, are they found in motets. Most notable among these is the double note, 3 which, in non­mensural notation, indicates that the pitch concerned is to be held twice as long as a single note; in mensural notation, the doubling of a note­value was indicated in a different way. Thus, the question arises whether the notation in the Escorial manuscripts is as mensural as Anglés claims. Because the style of the individual Cantigas melodies varies rather widely, the degree to which Anglés transcribed them according to the medieval rules of mensuration is variable. For the syllabic ones, he adhered fairly strictly to the rules, but for the melodies with many ligatures he frequently ignored them. 4 The latter case may be shown best by examining a melody (Cantiga 340) as notated in its manuscript source (the lines marked E1 in Ex. 1), and by comparing it with Anglés's transcription (1943:II, 371­72); given on the staves marked A), and a transcription made strictly according to the medieval rules of mensuration (given on the staves marked M).5 See.Example 1 and 1b (cant340a.jpg; cant 340b.jpg)

Since the first two verses of text are sung to the same melody, only one melodic line is notated. The melodic lines 4 and 7 are identical, but were left in their proper order.

On the staves marked M, the ligatures, which, because of their form, are non-mensural, are transliterated in stemless black notes. For the present discussion, it seems of minor relevance to determine how they could be made to fit a mensural transcription.

In Anglé's edition, the second pitch of the two liquescent neumes (over the first syllable of figura in line 8 and, at the very end, over alva) is printed with small note heads. Because the mensuration of this neume is unclear, I refrained from transcribing it mensurally,. Instead, in both cases, I transcribed ti as one note with a curved stem.

The juxtaposition of the respective transcriptions provides some insights into Anglés's methods. Two features stand out: 1) he reduced the duration of many long notes, and 2) he placed bar lines in a rather unusual manner. In almost every line of the mensural transcription, long and short notes occur side by side in a manner which seems disproportionate. Equally disturbing are the strings of long notes in lines 4, 7, and 9. By reducing the value of many long notes, Anglés achieved a better and even an attractive balance between long and short notes. It is possible, and even likely, that he was correct in rejecting the disproportionate differences in duration, but, in so doing, he relied upon subjective judgment.

Medieval notation does not contain bar lines; there are, however, some practical reasons for inserting them in modern editions. In most motets of the thirteenth century, excepting the very late ones, long and short syllables alternate according to ordered units, called "rhythmic modes". A syllable may be sung to one or more pitches, but, in general, its total time equals one unit of the modal rhythm, that is, in modern notation, a quarter note, a half note, or a dotted­half note. In transcriptions of thirteenth­century motets, it is fairly standard practice to notate a complete pattern of a rhythmic mode in one measure. It appears that Anglés started out with the basic idea that the occurrence of long and short time values in the Cantigas corresponded vaguely, but not precisely to modal rhythm. In the literal transcription of Cantiga 340, "Virgen Madre" (Ex. 1), the alternation of long and short syllables does not even come close to resembling any rhythmic mode. Furthermore, it contains many instances in which a syllable takes up more than one measure, the longest duration normally allotted to any syllable except the very last one. Perhaps, in order to achieve some uniformity between cantigas and motets, Anglés not only reduced the value of many notes, but he also placed the bar lines, so that, with some exceptions, no syllable took up more than one measure. Toward the latter goal, Anglés's introduced not only binary meter, in the form of four­quarter time, but he also introduced five­quarter time, disguised by dotted bar lines. (See line 3, second measure; the end of line 5, over podia; and again at the end of the last line, over alva.)

All in all, Anglés's transcriptions adhere as often to the medieval rules of mensural notation as they depart from them. In his critical study of the notation (1958:III, 141­87), he defends many of the departures. However, in this process he considers each neume in isolation and actually ignores the durational context, which may well have been the most important feature of early mensural notation. In a very curious way, there is an element of truth in Anglés's evaluation of the notation in the Escorial manuscripts: many of the details appear to be mensural, but the notation in its entirety is decidedly non­mensural. At best, his transcriptions support the idea that the notation is partially mensural and allows the transcriber to decide which neumes indicate duration and which do not. As far as I can determine, such decisions can be made only on the basis of conjecture and personal preference.

If the notation of the Cantigas is partially mensural, it has no parallel in the motet collections of the time, yet, except for the frequency with which the ligatura cum opposita proprietate occurs, it is not entirely different from that which is found in a couple of manuscripts containing trouvere chansons, and in one of the sources for Gautier de Coincy's Miracles de Nostre Dame. In these collections, the notation varies rather widely in its apparent indications for duration. In a few melodies, stemmed and unstemmed notes alternate in a manner which may indicate the first, second, or third rhythmic mode, if one assumes that the ligatures do not have any mensural meaning and if one, instead, considers each neume to represent a durational unit of the rhythmic mode. (Since in this notation only the single notes-not the ligatures-have mensural meaning, some authors have called it "early mensural"; others have preferred the term "semi­mensural".) In a number of other melodies, the alternation of stemmed and unstemmed notes, although not completely consistent, seems to suggest, nevertheless, one of these same modes. On the opposite end of the spectrum, we find many melodies in which the alternation of stemmed and unstemmed notes is too erratic to be considered as having mensural significance. Understandably, it is impossible to distinguish clearly between partially consistent and generally inconsistent alternation of stemmed and unstemmed notes; thus proper transcription of a given melody may be subject to dispute and its implied modal rhythms led some scholars to the conclusion that modal rhythm, as found in motets, also prevailed in trouvere chansons. Accordingly, many editors transcribed trouvere and troubadour melodies in modal rhythm.6 For several reasons, this practice became quite controversial. From a purely scholarly point of view, reservations were made about the subjective manner inwhich a rhythmic mode was chosen for a given melody. Other objections were based upon aesthetic considerations: most melodies were far from attractive in their modal rhythm and, more importantly, the modal straitjacket seemed ill­suited to the meter of the poems. Anglés may not have disapproved of the subjective factor involved in choosing the proper mode for a given melody, but he seems to have shared the objections made on aesthetic grounds. In his edition of the Cantigas, he tried to solve the problem by applying modal rhythm in a more flexible way.

Not long ago, I had the opportunity to investigate the chanson of the troubadours and trouvères.7 Many of them have been preserved in multiple versions, and a close examination of the similarities and differences among them disclosed, in a fascinating manner, the stable and variable components of their melodic style. In my estimation, the melodic variants make sense only if one assumes that the songs were disseminated primarily in oral tradition and that the vast majority of them were performed in a free rhythm, that is, the melodies were performed so that all notes-not the syllables-were more or less equal in duration. It seems impossible to determine how much variance existed in the duration of individual notes. Multiple versions, however, display a fair, although incomplete, uniformity when differentiating between single and double notes, which suggests that, to a large extent, it was possible for scribes to distinguish between notes of average and double length. It is likely that the duration of individual notes varied somewhat from one performer to another and, especially, from one song to another. Since all strophes were sung to the same tune, the duration of a given note is also likely to have varied from one strophe to another. We may no longer be able to reconstruct the precise manner in which a given chanson was performed seven or eight centuries ago, but it is fairly clear that duration could not be adequately expressed in the mensural notation of the Middle Ages, nor can it be represented by our eighth notes, quarter notes, half notes, etc. These conclusions appear to hold true even for many of the songs which, in one source or another, were notated semi­mensurally. In many cases, modal rhythm seems to have been superimposed, probably by a scribe, upon an existing melody. Needless to say, in some instances, including those melodies preserved in a single source, it is difficult to determine whether the semi­mensural notation reflects the composer's or some one else's intentions. All in all, there is not much basis to support the notion that the durational patterns created for motets were imitated in all other compositions of the thirteenth century.

The differences and similarities among multiple versions, which provide invaluable information about stable and variable elements in the French repertory, are of limited use for the study of the Cantigas, because the few multiple versions for the latter were clearly copied-directly or indirectly-from a single set of exemplars. Nevertheless, they are not entirely without value, because one may assume that the more essential a given element, the more attentive a scribe would be to its proper notation. It may not be of great significance that, in the Cantigas manuscripts, the stems of the notes are so faint and thin that many of them are hardly visible, whereas they are very clear in mensurally notated motets. It is important, however, that discrepancies among the three Cantigas sources often concern duration, rarely pitch. My knowledge of these differences is based almost entirely upon the information supplied by Anglés in his footnotes and in his representations of the neumes as they appear in the original notation. Placing a note correctly on the staff appears to have been mandatory, while it was of secondary importance whether a stem was present or not. The Cantiga "Virgen Madre" may again serve as an example, because it occurs twice in manuscript E1 (Escorial B.1.2) (on f. 4 a­b and f. 304 c­d; Anglés's transcriptions [1943:II, 439 and 371­72]). As far as the pitches are concerned, the two readings are identical, but where duration is concerned, we find nine discrepancies between them. In seven instances, the versions differ as to whether a given single note should have a stem; thus they disagree as to whether the pertinent notes should be long or short. In the other two instances, the versions differ as to whether a given ligature should be with or without opposita proprietate.8 The melody of "Virgen Madre" also happens to have been preserved with a poem by the troubadour Cadenet.9 The existence of one demonstrable contrafact does not constitute evidence that the meter of all cantigas resembled that of troubadour songs. More importantly, however, the diflferences and similarities between the two melodies do not support Anglés's transcriptions nor the durational indications in either reading of this cantiga.

The determination of melodic accentuation forms an even more elusive problem that that of duration. As remarked above, the manuscripts do not contain indications for melodic accents. The position of the textual accent in a polysyllabic word, on the other hand, is well known. As far as the text is concerned, the situation is fairly clear: a line has a predetermined number of syllables, but, excepting the rhyme syllable(s), there is no predetermined pattern for the placement of word accents and otherwise stressed syllables. Consequently, the placement of stressed syllables within a line varies not only from one line to another, but also from strophe to strophe. It is often assumed that in mensurally notated motets, the first note of a modal pattern (i.e., the first note after the bar line in a transcription) receives a melodic accent. If this is indeed the case-in my estimation, it has never been proven satisfactorily-motets contain many instances in which melodic accents fail to coincide with textual ones. This feature may not be objectionable in motets, because their polyphonic texture and durational patterns seem to have been of primary concern. In the Cantigas, however, we may assume that conveying the text to an audience was of paramount importance, and conflicts between melodic and textual accents might well have defeated that purpose. Thus many of the bar lines in Anglés's transcriptions seem to have been placed without consideration for proper alignment of textual and melodic accents. In addition, there is a curious lack of relation between melodic duration and textual accentuation. In Anglés's transcription of "Virgen Madre", many syllables are lengthened because they are sung either to a long note or to a group of notes. Thus, in this particular cantiga and in the many others resembling it, a performer may be able to treat the stressed syllables appropriately. In the more syllabic cantigas, however, the relation between melodic duration and textual importance of a given syllable is often awkward. This problem increases when all strophes are considered.

Returning once more to the songs of the troubadours and trouvères, one may recall that the principles of versification for the chansons are the same as for the cantigas. The similarities and differences among multiple versions of chansons warrant the conclusion-in addition to those mentioned above-that their melodies do not have accents of their own. Thus diflferences in the distribution of stressed syllables from one line to another and, more importantly, from one strophe to another, could be treated as the performer saw fit. A rendition of cantigas in such a declamatory rhythm would not contain any conflicts between melodic and textual accents, but it would still contain passages in which an unimportant syllable, such as a mute e, is sung to several pitches, while a crucial word receives only one pitch. While this may sound awkward to us, the medieval listener may not have perceived it; thus, since such disproportionately distributed ligatures and melismas occur frequently in Gregorian chant and even in some French chansons de toile, which, because of their narrative character, are close relatives of the Cantigas.

All in all, little or no unequivocal evidence may be found to solve the problems concerning accentuation and duration in the Cantigas. Even a large­scale comparison of musical and textual characteristics in motets, chansons, and cantigas is not likely to reveal reliable information on the topics at hand, but it is obvious that the Cantigas are more closely related to the monophonic songs of the troubadours and trouveres than to the polyphonic and polytextual motets.

There may be an ironic aspect to the above conclusions. Limited experimentation with performance of troubadour and trouvère chansons has convinced me that a rendition of medieval songs in a free and declamatory rhythm does not have great appeal to present­day audiences, yet the public and recorded performances of Anglés's transcriptions have become remarkably popular. This does not necessarily contradict my conclusions; instead, it probably reflects a profound change in taste over the past seven centuries. It appears that, in the twelfth and thirteenth centuries, most, if not all genres of poetry were performed to some kind of melody rather than being recited in a speech­like manner. In addition, it is quite possible that, regardless of language, such performances were rendered in a free and declamatory rhythm. Nowadays, poems are recited, or declaimed, in a free rhythm without melody. In songs, on the other hand, accentuation and duration in the melody (as well as the accompaniment) are at least as important as the proper prosody of the words. Thus, while Anglés may not have answered the question as to how the Cantigas were performed seven centuries ago, he, nonetheless, contributed much to enable performers to bring these melodies back to life in our times.


1. Many of the melodies occur in two, some even in all three sources. In the footnotes to such songs, Higinio Anglés records their pitch discrepancies; the small number of such notes suggests that the scribes were, indeed, admirably accurate in copying the thousands of notes.

2. Motets are compositions for two, three, and four voices, in which each voice carries its own text. Thus, depending upon the number of voices, two, three, or four texts are sung simultaneously. These texts may be religious or secular, in Latin or French. Accordingly, motets are polyphonic, polytextual and, often, also multilingual.

3. See, for instance, the double F over nobre in line 3; the double b over podia in line 5; the double a over figura in line 8; and the double C and E over the final word alva in Example 1.

4. For descriptions of mensural notations by medieval authors, see Carl Parrish (1957) and Willi Apel (1953).

5. Despite their obvious weaknesses, Anglés's theories and transcriptions are widely accepted. In what, thus far, may well be the best text book on medieval music, Richard H. Hoppin wrote:

As a result of their unexpected use of mensural notation, the Cantigas are unique in being the only large repertory of monophonic song that can be transcribed with assurance in modern meters and note values. Many songs are in triple mete r, sometimes transcribed as 6/4 or 6/8. Others are clearly duple. Songs in triple meter do not always adhere to the rhythmic modes, and some do not maintain a regular meter throughout but change from triple to duple meter or vice versa (1978:320).

Clearly, this evaluation is based on the theories and transcriptions of Anglés, not on the original notation.

6. For an extensive discussion of the application of modal rhythm to monophonic songs of the Middle Ages, see Burkhard Kippenberg (1962)

7. For findings based on this research, see Hendrik van der Werf (1972).

8. Perhaps one should not attach great significance to the fact that Anglés failed to indicate two of these differences above his transcription of Cantiga 340.

9. For a transcription of the latter, see Ismael Fernandez de la Cuesta (1979:548). This Cantiga may also be a distant relative of Giraut de Bornelh's famous alba, "Reis glorios" (Idem. :169).

References Cited

Anglés, Higinio
1943-1964 La música de las Cantigas de Santa Maria del Rey Aifonso el Sabio. Facsimil, transcripción y estudio crítico. Barcelona: Diputación Provincial, Biblioteca Central. Vol. 1 (1964), Vol. 11 (1943), and Vol. 111 (1958).

Apel, Willi
1953The Notation of Polyphonic Music 900-1600. Fourth edition, revised and with commentary. Cambridge, Mass.: The Mediaeval Academy of America.
Fernandez de la Cuesta, Ismael
1979Las cançons dels trobadors. Melodias publicadas per Ismael Fernandez de la Cuesta. Textes establits per Robert Lafont, amb una revirada alemanda, Anglésa, castelhana e francesa. Tolosa: Institut d'Estudis Occitans/Madrid: Editorial Alpuerto.
Hoppin, Richard H.
1978 Medieval Music. New York: W.W. Norton
Kippenberg, Burkhard
1962Der Rhythmus im Minnesang: eine Kritik der literatur­ und musithistorischen Forschung mit einer Ubersicht die musikalischen Quellen. Munich: C. H. Beck.
Parrish, Carl
1957The Notation of Medieval Music. New York: W.W. Norton.
Van der Werf, Hendrik
1972 The Chansons of the Troubadours and Trouvères. A Study of the Melodies and their Relation to the Poems. Utrecht: A. Oosthock.