Conference Agenda

Undoubtedly, MEI is one of the most, if not the most versatile data model for encoding music notation, covering an increasing number of notational systems for various epochs and regions, and different scholarly interests and research questions on such music. In chapter 1.3.5 of the MEI Guidelines, it is recommended that “in production, it is best to use a customized version of MEI, restricted to the very needs of a project.” However, there is no documentation so far on how to properly enrich the MEI model in those cases where the current standard does not adequately cover a given research interest.

Obviously, the first step would be to reach out to the MEI Community and to ensure that there are no conceptually adequate, but insufficiently apparent solutions available. Such cases are not unlikely if a solution would depend on features of MEI that have not been widely explored. However, it is possible and legitimate for a research project to have specific needs that are beyond the scope of the current MEI framework. A reason for that might be the exploration of a repertoire or feature currently unsupported by MEI. In that case, it is probably best to continue the discussion with the MEI Community and jointly seek to find a model that neatly fits into the existing MEI framework. Such an enhancement may require changes to current MEI, so this isn’t necessarily an addition only, and may result in breaking changes to other parts of the MEI framework. Eventually, project requirements may contradict existing design choices – they are intentionally not covered, or addressed in ways which do not cover the needs of the example at hand. Reasons for that might be complications in other parts of MEI caused by the implementation of this model, or potential inconsistencies between these parts. In such cases, there is a need to find a data model which extends beyond MEI’s current scope without interfering with the standard MEI model.

Our presentation seeks to shed light on this critical gap in the documentation of MEI. We will discuss best practices for individual extensions to the standard, aiming to emphasize the inherent challenges faced by researchers and projects encountering modelling challenges that are not currently addressed by the MEI framework. Addressing this documentation gap is crucial for fostering innovation and accommodating diverse scholarly interests within the music encoding community.

Efforts to encode music have long recognized the differences between how note information can be conceived of logically, how the notes are to be performed, and how they are to be conveyed in standard music notation. Rich encoding schemes such as MEI allow description in each of these domains—the logical, gestural, and visual—although not always elegantly in the same file. Contrapuntal keyboard music is a challenging genre for multi-domain encoding, particularly with regard to voice. There is not always consistent mapping from staff and stem direction to contrapuntal voice nor do staves necessarily indicate which hand is playing, especially if there is hand-crossing. Furthermore, occasional chordal textures can defy the assignment of individual notes to a particular voice.

Bach’s Goldberg Variations provide a rich inventory of examples of all these challenges. This paper will illustrate those challenges and potential solutions in the context of an early-stage project to build an interactive viewer / player for the Goldberg Variations that allows exploration of the work through different domain lenses. The specific contrapuntal structures of these variations allows for a detailed exploration of multi-domain encoding. We abstract away from the conventions of music notation to support a more detailed exploration of the analytical domain which can be rendered in a variety of different ways. As well as the encoding and web application itself, the paper will also discuss Python code used to convert and map between the different domains of encoding.

The power of working across the logical, gestural, visual, and analytical domains stretches what we are able to do but also brings up fascinating issues for how to encode music in a machine-readable and human-readable way to maximize how much information can be presented and analyzed.

MEI’s development to date has understandably focused on the digitization of visually-based musical documents. It thus has provided a parallel to historical musicology’s long-standing (since the 19th century) conception of musical scores as musical works. The past quarter-century, in particular, has seen an expansion of musicological research to include performance studies based on the use of audio-visual recordings as primary source documents. Not least in the context of current considerations to expand the coding options of the format to include audiovisual elements, the limits of this current structuring become apparent. The question therefore arises whether the existing MEI structures are sufficient to respond to this performance oriented turn, or if they should be questioned and adapted as necessary in order to better align the data format with current developments and considerations in musicological research and neighbouring disciplines.