NEUMES

NEumed Unicode™ Manuscript Encoding Standard



Preliminary Taxonomy of Neume Forms

Louis W. G. Barton and Ruth Ripley

University of Oxford

Revision: 1 October 2001




Taxonomic Principles


  1. Abstract neumes, not glyphs

    A key strategy of the NEUMES project is that transcriptions should enocde abstract neumes rather than glyphs. A glyph is a graphical rendering; it is closely related to the notion of a font for display or printing. An abstract neume is a classificational concept that results from our speculative reconstruction of the 'individuation scheme' that might have been in a scribe's minds when translating between aural image and written symbol.

  2. Two layers of digital encoding

    1. The low-level data representation, should capture the 'intensional' forms of neumes, viz., the abstractions of neume entities that we speculate scribes used to "individuate" aural phenomena into gestural entities. These neume forms we shall call abstract neumes. A test of membership in the set of abstract neumes shall be whether an item of transcription information is strictly needed for music-theoretic analysis. Abstract neumes shall be encoded in the 16-bit Unicode™ Standard, specifically in its Private Use Area, which was reserved by the Unicode™ Consortium for purposes such as this. The principal benefits of this strategy are (a) disambiguation, and (b) processing efficiency during 'random access' of data streams. We anticipate that an analysis program will typically need to do 'complex traversal' of data, whereas a visual redering program can process a data stream in strictly sequential order.

    2. The high-level 'wrapper' encapsules the abstract neumes for transmission and data storage, and shall include all transcription information that is needed to suppport all the anticipated uses of the data, but shall exclude transcription information that is represented in the abstract neumes. For file-format compatibility with a broad variety of other types of data, the encapsulation layer shall be coded in XML (Extensible Markup Language), our particular scheme for doing so being called NeumeXML. XML uses tags (similar to HTML tags), which (worst case) must be read sequentially from the beginning of the file to disambiguate them. This is not a problem for visual rendering, because visual documents are typically coded so that they can be layed out from top to bottom and left-to-right. Processing efficiency is not a stated goal of XML. XML tags are typically ASCII characters (essentially, those found on a typewriter). Between the tags we can interpose neumed Unicode characters.

  3. Abstruse data representation


  4. The low-level data representation is not intended to be 'human readable', in the sense that it shall contain many non-ASCII characters. This 'human non-readability' is entirely normal for Unicode data. Unicode (which has complete and unambiguous character codes to cover every written language on Earth) has tens of thousands of code points (i.e., distinct and unambiguous codes), whereas ASCII has only 256 code points. For visual rendering, it is the responsibility of the applications program (such as a Web browser or word processor) to render Unicode characters visually as glyphs.

  5. 'Lossless' representation vs. ease-of-use


  6. In Phase One of the NEUMES project, we expect to develop data-entry and data-manipulation software mainly for the purpose of testing the data representation. These programs will rely on the user's having enough musicological sophistication to correctly use all the neume abstractions in the data representation. Eventually (likely in Phase Two of the project), we expect to develop data-entry software suited to less-skilled users, who would likely be able only to do a visual transcription without understanding the proper grouping of neumes. An 'assist' program will translate the user's input to the full complexity of the data representation by means of programmed rules.

  7. What constitutes a 'neume'?

  8. An important topic for us to resolve is the difference between a neume and a neumatic sign. As I understand it, Cardine holds that a neume is a melodic 'gesture' that may be written as several neumatic signs in series. For example, is the following Aquitanian notation a single neume or three neumes?



  9. Compound neumes

  10. Some compound neume glyphs may have a one-to-one relationship with abstract neume forms. Other compound glyphs may be rendered by the display or printing program such that a sequence of two or more abstract neumes comprises a single glyph. Taxonomic distinction between compound neumes in the data representation, and compound neumes rendered by the applications program, should be made on the basis of the 'intensional' form, viz., whether the scribe likely conceived of the compound neume as an entity of cognitive individuation. In principle, compound neumes could be coded in the data representation using (what the Unicode calls) 'combining characters'; an example of 'combining characters' in ordinary text is diacriticals.



Various Questions


Q: Intervals for compound neumes: maximum interval is a 6th?

Q: In cases where the rendering of a compound neume depends on context, can all contextual information be derived from rules, or is it necessary to encode such contextual information in (second-order) NeumeXML?

Taxonomy of Neume Forms across Neume Families

Below is listed the set of neume forms I came up with. This list must be reconciled with the Oxford Chant Group table (bottom of this page). Next to some of these forms are questions in brackets [ ] that Tom and/or I had.

A. Number of forms, name of form, and comments:

4 virga and substitute-style virga [Problem: do we need a substitute style?]: both in plain and liquescent forms

4 punctum and substitute-style punctum [Problem: do we need a substitute style?]: both in plain and liquescent forms

50 pes (or podatus) at intervals of 2, 3, 4, 5, 6: ending can be plain, with apostropha, with oriscus, with apostropha and oriscus, or with oriscus and apostropha: each plain or liquescent

50 clivis (or flexa): same variants as for pes

50 torculus (or pes flexus): same variants as for pes [Problem: mixed intervals]

50 porrectus (or flexa resupina): same variants as for pes [Problem: mixed intervals]

36 scandicus at intervals of 2, 3, 4, 5 for the first part: ending can be plain, with apostropha, with oriscus, with apostropha and oriscus, or with oriscus and apostropha: each plain or liquescent

6 climacus (virga followed by one to three puncta "normally," can be composed if more): plain forms (3) and liquescent forms (3)

2 tripunctum: plain and liquescent forms [Check fact: same as trigon?]


B. Auxiliary signs:

1 custos

C. Forms that could be composed using 'combining characters':

? quilisma (ends something else; first note is a special sign followed by a note higher; quilisma is always attached to something else).
? salicus [Problem (ask other musicologists): can be composed? intervals? liquescent?]
? pressus [Problem: needed but must discuss later: composed? intervals? liquescent?]
? tractulus (like pes at interval of 2nd up, or unison, or liquescent) [can be composed? intervals? liquescent?]

Forms in blue are those that need to be reconciled with the Oxford taxonomy.

bi-virga
bi-punctum
torculus resupinus flexus
porrectus flexus
scandicus flexus
climacus resupinus
porrectus subpunctis
pes subpunctis
pes praepunctis
virga prae-bipunctis
torculus praepunctis
porrectus praepunctis
apostropha
oriscus
clivis (or strophica) (can be composed) [intervals?] [liquescent?]
torculus (or pes flexus strophicus) (can be composed) [intervals?] [liquescent?]
dis- or tristropha praepunctis (can be composed) [intervals?] [liquescent?]
pes stratus (can be composed) [intervals?] [liquescent?]



Oxford Chant Group

Data Representation and Taxonomy


The table, below, shows the neume taxonomy currently in use by the Oxford Chant Group (OCG), as provided by Dr. Ruth Ripley. This taxonomy was designed especially for a database of unheighted neume transcriptions.

Two salient principles of the OCG data representation scheme are that:
  1. it is text-based, viz., a second-order data representation (see Barton, "Computer Encoding of Neumed Manuscripts");


  2. it is strongly tied to a set of glyph images (see definition of "glyph," above).
An OCG transcription file consists of many lines of ASCII text delimited by the newline character. A single neume descriptor consists of a 'group' of (one or more) lines in series. These groups are arrayed in the file in the 'natural' order of reading from the source document, as follows.
§ Composition of a 'group'

In the following discussion, I have marked with an asterisk (*) the items that appear in the table, below, which you can use for reference.

A 'group' consists of a primary line, followed by (zero or more) 'detail' lines. The primary line has the following scheme: Following the primary line come the 'detail' lines. One 'detail' line has the following scheme: The last line of a 'group' contains the 'relative position string' for the notes in this neume. *


§ Explanation of certain (second-order) codes

'Note' length options: 'Note' position options:
§ Default values

In the OCG data-entry program, 'default' values are entered automatically if one does not specify a value. The note defaults depend on the type of neume and (sometimes) on the previous neume. Defaults currently are as follows.
§ Taxonomy (OCG)

Neume name Number of notes Relative positions string Detail number Detail text Detail action

apostropha 1 ?
climacus 3 ? - -  1 c1 make first note short
2 t1 make first note long
3 ep1 make first note long
4 l3 make third note long
5 l23 make second and third notes long
6 big leave notes unchanged (simply records the shape of the neume)
climacus resupinus 4 ? - - + 1 c1 make first note short
2 t1 make first note long
3 ep1 make first note long
4 l3 make third note long
5 l23 make second and third notes long
6 big leave notes unchanged
(simply records the shape of the neume)
clivis 2 ? - 1 t+ep make both notes long
2 ep2v make second note long
3 ep2h make second note long
clivis + pressus minor 4 ? - ? - 1 t+ep make first two notes long
2 ep2v make second note long
3 ep2h make second note long
gravis 1 -
oriscus 1 ?
pes 2 ? + 1 t2 make second note long
2 ep2 make second note long
3 diag make both notes long
pes combipunctis 6 ? + + + - - 
pes quassus 2 ? +
pes stratus 3 ? + = 
pes subbipunctis 4 ? + - -  1 c12 make first two notes short
2 t2 make second note long
3 t12 make first two notes long
4 l12 make first two notes long
5 l2 make second note long
6 diag make first two notes long
porrectus 3 ?  - +  1 ep3 make third note long
porrectus flexus 4 ?  - + -  1 ep34 make third and fourth notes long
2 l34 make third and fourth notes long
3 clong make all four notes short
porrectus subbipunctis 5 ? - + - - 
pressus maior 3 ? = -  1 c2 make second note short
2 t2 make second note long
pressus minor 2 ? - 1 joined no change to notes
punctum 1 ?
quilisma 1 +
salicus 3 ? + + 
scandicus 3 ? + +  1 epx3 make all three notes long
2 l12ep3 make all three notes long
scandicus flexus 4 ? + + - 1 c34 make third and fourth notes short
2 t34 make third and fourth notes long
3 ep34 make third and fourth notes long
4 ept34 make third and fourth notes long
5 ep4v make fourth note long
6 ep4h make fourth note long
scandicus flexus resupinus 5 ? + + - + 1 ep5 make fifth note long
torculus 3 ? + - 1 3down no change
2 ep23noc1 make first note short and second and third long
3 c1ep23 make first note short and second and third notes long
4 t23noc1 make first note short and second and third notes long
5 c1t23 make first note short and second and third notes long
6 c1t23ep23 make first note short and second and third notes long
7 t23ep23noc1 make first note short and second and third note long
8 l3 make third note long
9 ep3v make third note long
10 ep3h make third note long
11 stretched make first note short and second and third notes long
torculus resupinus 4 ? + - + 1 t234noc1 make first note short and second, third and fourth notes long
2 ep234noc1 make first note short and second, third and fourth notes long
3 l234noc1 make first note short and second, third and fourth notes long
4 ep4 make fourth note long
torculus resupinus flexus 5 ? +  - +  -
tractulus 1 ?
trigon 3 ? + -
virga 1 ?
virga cumbipunctis 5 ? + + - - 
virga praebipunctis subtripunctis 6 ? + + - - - 
virga strata 2 ? = 1 c2 make second note short
2 t2 make second note long
other 1 ?