First and second pages of
CSM 377: Sempr' a Virgen Groriosa
Cantiga 377 folio 1 from [E] Click image for larger version from Greg Lindahl's
Cantigas de Santa Maria site (external link)
Cantiga 377 folio 2 from [E] Click image for larger version from Greg Lindahl's
Cantigas de Santa Maria site (external link)
(Just six lines remain on the following page. Note the missing decorated capital Q from the last stanza.)

This project is based on the principle that there is one, and only one, correct way to fit the words of the Cantigas de Santa Maria to the music, and that it is worth searching for even though it can at times be rather elusive. In this Introduction I hope to explain to the potential users of this site just why I believe this to be so, and how I've set about achieving the goal.

Cantiga words and music

The images on the right show the first two pages of a typical Cantiga, number 377 Sempr' a Virgen Groriosa, as it appears in the [E] manuscript (the Escorial "codice princeps"). You can see the refrain and the first stanza set to music, followed by the remaining stanzas as text only, with reprises of the refrain written between them. The large decorative S is preceded by an epigraph summarising what the Cantiga is about—here it is missing the initial letter C from the word Como.

Clearly, the musical notation is not modern: it is medieval syllabic square notation, in which pitches are conveyed by shapes that represent either a single musical note, in the case of .on and .o and .e, or a ligature of multiple notes in all other cases, here including .owo and .ood and .oron, though there are many others.¹ When transcribing the music into modern notation, the convention is to show the ligatures as slurred note groups. Knowing this, we can state the absolute number-one rule for fitting the Cantiga words to the music as follows:

Rule 1: One syllable per note or ligature, and one note or ligature per syllable.

or, referring to modern transcriptions, one syllable per single note or slurred group, and vice versa. (The musical notation in the Cantigas is mensural, meaning that the different note and ligature shapes represent not only the pitch of the component notes, but also their duration. The correct interpretation of the latter aspect has been the subject of a great deal of scholarly debate, and is naturally of fundamental importance to performance, but it has no bearing on the basic fit between words and music.)

The second rule and third rules are equally straightforward:

Rule 2: All stanzas should have the same poetic metre as first stanza.

Rule 3: All stanzas should fit the same music as the first stanza.

Unfortunately, things are not as simple as these three rules, because if you apply the correct, natural syllabation for medieval Galician-Portuguese, and then try to match the syllables to the musical notation, you will find that, across the whole collection of Cantigas, around 2% of lines don't actually fit. There are either too many syllables for the notes and ligatures, or too few. This figure is based on the 1986–1989 edition of the text of the Cantigas de Santa Maria by Walter Mettmann, which is a synthesis of all four manuscripts (although biased towards [E]), and which I took as my original starting point for this project. Were it based directly on any individual manuscript, then the percentage would certainly be higher—though how much higher is impossible to know without going back and comparing the manuscript with Mettmann's version line by line.

So what can be done about these mismatches? I will come back to that below, after a small detour entitled...

How not to sing the Cantigas de Santa Maria

The naïve approach probably goes something like this:

  1. get a basic idea of the pronunciation of the language;
  2. learn the melody;
  3. write or print out the lyrics of the refrain and every stanza on a piece of paper; and
  4. improvise a fit between the words and the music for each line, being content with any outcome where both end at the same time.

The problem with this method is that even when 1. and 2. are done well, it's still easy to go astray. Taking the first step: the performer may well have a good grasp of modern Portuguese, or even, if they're lucky, Galician (in which the pronunciation has changed much less over the centuries than Portuguese). This will undoubtedly get them a long way, but it won't be perfect. For example, when Mettmann writes oya, how many syllables should that have? It's three, in fact. And oyu? Two, normally, but sometimes three, as diaeresis is common in 3rd personal singular preterite verb forms. What about saya? Two or three, depending on whether it's a present subjunctive or an imperfect indicative. And mi ajude? Three, because mi always forms a diphthong with a following vowel. Reinna? Always three. Reino? Always two. And just what are the syllables in tĩya? The natural pronunciation is actually tiinna [ti.i.ɲa] with three syllables, but in a dozen or so cases it is compressed to tiinna [ti.ɲa] by synaeresis. You get the idea.

As for the melody, one of the main problems is that once it is off the page and in the head, it is way too easy to forget which notes are separate and which are ligated, and to end up singing a syllable to half of one ligature and half of the next. I've done that myself more than I like to admit.

One might counter, "Well, it doesn't matter, does it? The audience will never know". And the fact is that the audience probably won't know, unless Don Alfonso X el Sabio and his Court turn up in a time machine and get tickets on the door. But to my mind this attitude is wilfully negligent, and shows immense disrespect for the art and craft of the poets and musicians who composed the cantigas. (Remember that up to 98% of lines do fit the music without a problem, and metrical accountancy was only a small part of their skills.) If being true to the proper form of the cantigas is not important for performance or recording, then when is it important? I think we have to accept authenticity not just as an academic ideal, but as something that all performers should be striving for in practice.

How to deal with metrical mismatches

So, just what is the solution when a line of the Cantigas doesn't fit the music because there are too many syllables, or too few? Let's assume from now on that we have those syllables completely correct. We have four broad options, in roughly descending order of attractiveness:

  1. Increase or decrease the number of metrical syllables without changing the natural syllables.
  2. Change the number of natural syllables.
  3. Override Rule 1 stated at the top of the page, and relax the fit between syllables and music.
  4. Change the music itself, to increase or decrease the number of individual notes and ligatures.

It is perhaps paradoxical that Options 1 and 2 do not directly involve the music. That's to say, rather than trying to fit the wrong number of syllables to the melody, the preferred approach is to correct the number of syllables to match what is required. The important principle behind this is that the Cantigas should work well as recited poems, not just as songs. And it is abundantly clear that, whatever their individual faults, the ideal for which they all strive is a completely regular metre, in which all stanzas have the same syllable counts.

Note that the repertoire of fixes listed below under Option 1 might be considered generally preferable to those under Option 2, but only marginally. In fact, when I have had to choose between them, the deciding factors have almost always been:

  • preferring a specific fix that occurs many times over one that is unique; and
  • going with what works best musically.
(So the music is involved after all, albeit indirectly.) On the other hand, I have treated Option 3 as a last resort, and it has been necessary in only eight Cantigas out of the whole collection. Option 4 is also to be avoided except in cases where the original music scribes have made obvious mistakes or oversights—very often in cases where the music was penned to fit a metrically wayward first stanza, but then didn't fit any of the remaining ones. Up to August 2012, this website only included lyrics, without musical transcriptions, so I wasn't really able to formalise changes to the music, and just suggested them in my lyric footnotes by reference to Elmes' edition where available, or directly to the music in [E] otherwise. Now that I've introduced the [E] transcriptions, however, it has become possible to annotate and reference musical fixes more directly.

Option 1: Changing the number of metrical syllables

Here we have to distinguish natural (or "grammatical") syllables from metrical ones: normally there is a one-to-one correspondence between the two, but there is a range of completely respectable processes that break the link. (You can tell just how respectable they are from their classical Greek names.)

Click on the names to see a longer description of each process, along with a full list of cases where I have employed it.

  • Synalepha: Two independent vowels merge into a diphthong or short vowel across a word boundary.
  • Synaeresis: Two independent vowels merge into a diphthong or short vowel within a word.
  • Diaeresis: A diphthong in a word divides into two independent vowels.

Option 2: Changing the number of natural syllables

These fixes are listed more or less in order of preference. Elision is an extremely common feature of the language of the Cantigas de Santa Maria, and adding more cases is quite acceptable if done with due care and attention—Mettmann certainly does this himself. At the other end of the spectrum, introducing entire new words must be done with the utmost caution, and only when all other options fail.

As above, you can click on the names for a longer description and the full list of cases.

  • Elision: A vowel is lost completely, when a word that ends in a vowel meets another that starts with one.
  • Restore elided vowel: The opposite of elision, i.e. the full form of an elided word is restored.
  • Substitution: a word is replaced by a longer or shorter variant.
  • Deletion: a word is removed entirely.
  • Insertion: a new word is added.

Option 3: Relaxing the fit between syllables and music

As stated above, I've treated this option as a last resort, avoiding it wherever possible.

  • Melisma: A single syllable is drawn out over two separate notes or ligatures.
  • Breaking: Two syllables are sung to a single ligature.

Option 4: Changing the music

For examples of interesting musical fixes, see CSM 18, 76, 77, 79, 113, 147, 169, 181, 220, 224, 263, 293, 307, 345, 352, 367, 377 (featured above), 384 and 421.

[1]  Strictly speaking, .ood and .oron are long plicas, not ligatures, but that distinction isn't important here.