A case of long neume pattern: Tone B verses
If neumes are so ambiguous when taken individually, what about considering larger structures involving sequences of neumes, including also the particular musical contexts in which such sequences appear? In this sense, the fundamental hypotheses underlying the SONHIS project is the presence in the Hispanic chant of mnemonic patterns encoding unambiguous melodic clichés.
In fact, pioneering works performed via visual inspection showed the existence of long overrepresented neume patterns in the responsorial Psalms verses whose standardization was identified nearly a century ago by Rojo & Prado [1] and Peter Wagner [2] and was later classified by Don Randel [3]. Such patterns constitute the so called seven verse tones (alphabetically designed from letter A to G), a nomenclature that should not be confused with the concept of tone within the oktoechos formulation for the Franco-Roman repertoire.
Therefore, our work with patterns must necessarily begin by reviewing these known standarized sequences whose study can now benefit from the capabilities of computational tools in both analytical and visualization tasks -for the later, the results of first studies on tone verses were even limited by the paper size in which the aligments of the patterns had to be displayed (Figure 1).
OCR of all 221 tone-B responsorial verses in the León Antiphoner
We will specially focus on tone B verses, not only for being by far the most common of the seven tones [5] but also because we count with a singable instance of such verse class among the chants transcribed in source Aem. 56 into the singable Aquitanian notation: the verse Dies quibus peregrinatus of the responsory Psalm Ecce ego viam whose transcription provide a fundamental cue to assist the deciphering of other tone B verses.
However, extrapolating the pitches of the Dies quibus to the many other instances of verses in tone B found in Hispanic sources is not a straightforward task since the text that is set to music is prose and the music had to be adapted to the length, accentuation and syntactic structure of each verse. As Randel states in his monumental study of these formulas:
The application of the more common melodies to their texts, especially in the León sources, follows clearly defined principles. These principles are similar to those embodied in the Gregorian responsoria prolixa in that the melodies are generally bipartite, certain formulas remaining fixed while others are adjusted to the structure of the text involved. Adjustments are made much as adjustments of the Gregorian formulas are made: by the addition of puncta and by the suppression, contraction, and division of other elements [6].
So, to study the way in which the music of the verses developes for the different textual circumtances we have to align the verses -in a similar way of what is shown in Figure 1 but now without the restrictions imposed by paper size and also assisted by alignment tools imported from the not so distant discipline of genomics. In fact, we aim to align all the tone-B verses in the León Antiphonary (source León 8) whose OCR's are listed in the index below. As a sample, see here the model of alignment we produced using (only) 33 verses encoded by the CEAP team.
Finally, let's say that, as worthy as they are, the best that these historical visual searches could accomplish was to catalogue patterns that (i) are long or very repetitive, (ii) appear on pages close together in the same book or (iii) appear in similar chants. By contrast, in order to detect patterns (especially short ones) with a medium-to-low frequency of occurrence and a high dispersion in the sources, the adoption of computer-assisted methods becomes essential.
