Using the IPA transcriptions

The phonetic transcriptions of the Cantigas use the International Phonetic Alphabet, and are provided on this website in conjunction with the main Pronunciation guide, which you should read through if you have not yet done so.

The following features of the IPA transcriptions are worth noting:

  • The position of the stress accent is not indicated. However, in the main texts the stress is marked with the acute or circumflex accent in all cases where it is not predictable from the basic rules given in the Pronunciation guide, e.g. ángeos [ˈanʤe.os] with initial stress, and porê [poˈɾeŋ] with final stress. The performer therefore has all the necessary information already, although for completeness I do intend to mark stress in the phonetic transcriptions too, in the near future.
  • Unlike the optional bullets in the main text which indicate all syllable boundaries, the IPA syllable separator (the period .) is used only between consecutive vowels in hiatus, e.g. cre•en•ça [kre.enʦa].
  • Spaces are inserted only when word boundaries and syllable boundaries coincide, e.g. co•m' an•ge•os can•ta•da [komaŋʤe.os kantada].
  • In a diphthong, the semivowel or glide (if any) is marked with the inverted breve below, e.g. viu [βiu̯], mia [mi̯a]. Where metrical diphthongs result from synalepha or synaeresis, both vowels may be strong, e.g. on•de a [ondea] (no period between [e] and [a]).
  • Where synalepha and synaeresis compress two identical vowels, the result is marked as a long vowel, e.g. tii•a [tiːɲa]. In spoken recital or rhythmically free singing a slight extension of the vowel would be appropriate.
  • Apart from the previous point, I don't explicitly mark vowel length on single syllables anywhere else, since any variation that might have occurred in the natural speech of the time was certainly not phonemic, and made no difference at all to the understanding of a word.
  • Although vowels would naturally be pronounced with some nasal quality before a nasal consonant ([m], [n] or [ŋ]), vowel nasalization in medieval Galician-Portuguese is no more phonemic than vowel length—it is a secondary effect that's completely predictable from the consonant. Nasality is therefore not explicitly marked on vowels in my transcriptions. This is the case even for vowels that are written with the tilde (ã, ĩ, õ and ũ) in the main text, which are better analysed as biphonemic—vowel + velar nasal consonant—and are therefore transcribed as [aŋ][eŋ], [iŋ], [oŋ] and [uŋ]. The [ŋ] may then nasalize the vowel as a secondary effect (giving [ãŋ], [ẽŋ] etc.), but this is no different from what happens in a word like dan, transcribed [daŋ] with predictable nasalization to [dãŋ].
  • Syllable-final but not word-final nasal consonants m and n are shown assimilated to a following labial or dental consonant: they appear as [m] before [b], [m] or [p]; and as [n] before [d], [ʣ], [t] or [ʦ]. Elsewhere, including at the end of a word, they appear as the velar nasal [ŋ]. For example, en•men•tar [emmentaɾ], a•pou•sen•tar [apou̯zentaɾ], men•gua•da [meŋgu̯ada], on•rra•da [oŋrrada], vir•ge [βiɾʤeŋ]
  • The alveolar trill [r] is always doubled to [rr] (e.g. te•rra [tɛrra], rei•no [rrei̯no]) in order to distinguish it more clearly from the flapped or tapped [ɾ], even though a single [r] is technically more correct. Don't overdo it on the length, though.
  • I'm using the older ligatures [ʧ], [ʦ], [ʤ] and [ʣ] to represent affricates, even though these are no longer official IPA. This is simply for practical reasons: the recommended replacements, where the stop and fricative components are written separately but with an overtie, don't work reliably in all browsers.
  • Melismata are expanded in the transcriptions. This is not really desirable, but it has turned out easier to keep them than to remove them, and for the moment it's not really worth the extra programming effort for so few examples (see the full list).

Finally, a plea: As with the syllable bullets, try to use the IPA transcriptions only as a reference when studying new cantigas, nothing more. If possible, don't rehearse by singing directly from the transcriptions, and never perform directly from them. At the point where you have an audience, you should always be thinking about the meaning of the words, not the sound, and the normal spelling is by far the best medium for this. There's also a risk that over-reliance on IPA can lead to confusion in the performer's mind between what's a phonetic transcription and what's normal spelling, with the result that a word like guisa, transcribed [giza], ends up being said as [ʤiʣa] due to double application of the pronunciation rules.

[1]  In fact, even in the word-final case, -m or -n can be assimilated to a labial or dental consonant at the beginning of the next word. I don't mark this, however, because it's unnecessary for good pronunciation, and because I think that transcribing the same word differently depending on what follows is probably counter-productive in a performer's edition: it will just cause unnecessary confusion.