Encoding the Oldest Western Music
Problems of Encoding
This project describes a system for digitally encoding neumes and corresponding text in parallel aligned documents in order to create a digital, diplomatic edition of chant texts with neumes. Neumes are graphic marks denoting relative changes in pitch; they predate staff notation and are written above text. Each neume can be marked with performance variations, called episema or liquescence. They also include musical directions abbreviated in Latin, with 15 significative letters such as ‘t’ for ‘tenere’ to indicate holding a note longer. There are at least a dozen styles of neumes, each of which has its own set of graphical symbols, like different fonts, to represent the same neumes.
A diplomatic edition of a neumed chant text must record the neumes as characters, not as absolute pitches. It also must align neumes with text, as they are visually aligned by syllable in chant manuscripts.
The ‘Virgapes’ system is based on a four-part encoding scheme for neumes that is flexible, extensible, and universal.
We have also developed a parallel document structure to align separate documents of text and neumes.
The ‘Virgapes’ System
In the Virgapes encoding, each neume is represented with a four-part code point. Each part is an integer standing for an aspect of the neume. The first integer denotes the number of pitches in the neume. The second integer is an arbitrarily assigned identifier within that group.
The system is flexible; it can expand to accommodate new or lesser known neumes. The third integer indicates the presence of episema, 1 for presence, 0 for absence. Likewise, the fourth notes liquescence in the same binary pattern.
For example: virga is a one-pitch neume, encoded as 220.127.116.11 in absence of episema or liquescence; pes is a neume of two ascending pitches, encoded (if liquescent) as 18.104.22.168.
The inclusion of episema and liquesence allows editors to note graphic marks indicating performance changes without imposing meaning. Our system also allows for specified searching: for all instances of virga or only instances of virga with episema, depending on the needs of analysis.
Parallel Aligned Documents
In addition to encoding neumes, we align transcriptions of neumes with transcriptions of texts. In a manuscript, this is done graphically: the neumes appear above the text.
In our digital editions, we create two parallel documents aligned by canonical citation using a Canonical Text Services (CTS) URN system to uniquely identify each passage.
With this, the two documents share a work hierarchy and a passage hierarchy. Consider the URN: ‘urn:cts:chant:antiphonary.einsiedeln121.text:11.introit’. The CTS namespace is ‘chant’ for the domain of chant texts. The group is ‘antiphonary’ for the type of chant book. The specific work is Einsiedeln 121. The last section notes the version, text or neume.
The second portion of the URN system is a passage hierarchy, which subdivides the work hierarchy. A parallel would be the act, scene, and line in plays. It first identifies the feast day using numbers delineated in the
Antiphonale Missarum Sextuplex.
Then, the subsection: introit, verse, etc, with further identifying numbers for graphically separated passages.
The URN system provides a citation scheme to align the texts; within the documents they are aligned by syllables, as each syllable must have at least one neume. This also provides a check for our encoding—there must be equal syllables in the text and neume document.
Digital encoding of neumes allows for advanced searching and analysis. With our two-part encoding solution, it is possible to search for repeated musical sequences, to determine if Zipf’s law applies to neumes, or to analyze musical texture based on the neume:text ratio.
Among chant scholars, the most important digital resources are the manuscript databases of the Cantus Index network. (See http://cantus.uwaterloo.ca/) These datasets include information about the manuscripts themselves in addition to the encoding of text and music. Unlike the Cantus system, however, we encode staffless neumes without imposing interpreted equivalences to later musical notation on staffs.
Of the XML systems, the most significant is the Music Encoding Initiative (MEI) (http://music-encoding.org/documentation/3.0.0/neumes). It is largely inspired by work at Tübingen. Our system also allows encoding of basic neumes with extended properties (liquescence, episema) in a specified syntax, enabling us to take account or ignore these properties in computational manipulation. Neither the current XML schemes nor the Cantus Index allow these properties to be optional.
Called ‘Virgapes’ for the first one and two-note neumes,
These are available from our Github repository: https://github.com/HCMID/chant.
This system was developed as part of the CITE Architecture for the Homer Multitext Project 2010-18, and applied to this project:
A standard chant reference work compiled by Dom Hesbert in the early 20
th century. It contains transcriptions of six important sources of Gregorian chant.