Reconciling corpora from enthusiast and scholarly communities: a sustainable approach to augmenting the Electronic Corpus of Lute Music

Authors: Crawford, Tim / Lewis, David

Date: Friday, 8 September 2023, 9:15am to 10:45am

Location: Main Campus, L 1.202 <campus:note>


In this paper we report on the augmentation of the Electronic Corpus of Lute Music (ECOLM) – an online collection of encoded music – with material originating from different communities with somewhat different priorities. These differences are reflected in the selection of music, the style of the encodings themselves, the level of annotation of the music, and the amount and quality of supporting metadata. Since the new material contains metadata and transcriptions for tens of thousands of items, an important consideration is sustainable and scalable approaches to the integration, improvement and management of the resulting dataset.

Music for the western lute was written in various forms of tablature notation. The repertory (c.1500-c.1800) is very large, perhaps up to 100,000 items surviving in both printed and manuscript sources. Although many are duplicated, these are very rarely identical copies. A large number are lute arrangements of well-known vocal music, often with highly virtuosic elaborations added to the sequences of notes in the original voice-parts.

Variation between versions, and the relationship between arrangements and their models, can provide musicologists with evidence for studies of style, influence and performance practice. To use the methods of digital humanities we need large datasets of encoded lute music, ideally in parallel with contemporary vocal and keyboard music. Digital editions can also allow the music to be presented in a variety of alternative forms from the same encoding: perhaps in a different style of tablature, as a staff-notation transcription intelligible to non-players, or as audio playback.

The ECOLM project began in 1999 to work on ways of handling lute music in machine-readable formats taking advantage of current web technology. A major step forward since those early days has been the establishment of an MEI module for encoding tablature. Using Verovio in a web browser allows lute tablature to be displayed in its Italian and French versions directly from MEI files, and there exists a software converter between most encoding formats used for lute tablature.

ECOLM currently contains around 2,000 academically-curated items, a small subset of the lute repertory. We are augmenting this corpus greatly by incorporating resources intended for a different community – that of lute-playing enthusiasts, whose priorities in certain respects differ from those of academics.

Enthusiast lutenists generally want to play from ‘performing’ editions of lute works; these may correct errors, regularise barring and repeats, reconstruct missing passages, and even combine sections from different sources. Often these interventions are made without comment. A significant number of players will also happily download and play from facsimiles of the original sources as PDFs, accepting or ignoring the often frequent errors or idiosyncracies of those sources. Some, notably professional performers preparing for concerts or recordings, may be prepared to go to the trouble of comparing a performing edition note-for-note with a facsimile, but many will not.

Some 8,000 pieces forming the contents of various published editions have also been donated to us; these comprise the encodings used to generate printed tablature, plus accompanying material (unstructured metadata) in Word or PDF files. The majority were originally published by the UK Lute Society in newsletters or tablature editions distributed to members over several decades (the machine-readable encodings have often been made or revised by the editor, John Robinson, more recently). The supplied metadata is generally quite complete, although over time (e.g. as manuscripts in private hands have changed owner or found their way into public collections) some inevitable discrepancies have crept in.

Meanwhile, we are also adding a larger collection of about 20,000 pieces from an online web-resource, curated and largely encoded by Sarge Gerbode but contributed to by a number of other enthusiasts, including many encodings of entire sources. An accompanying spreadsheet contains basic metadata for much of the collection. While the spreadsheet provides useful information about the sources used by the encoders, it has little scholarly pretension and has not been automatically updated as the resource grows.

Finally, we hope to incorporate a large metadata repository for lute sources which includes full inventories of around 900 sources, plus encoded tablature incipits for over 60,000 pieces. It covers the historical repertory nearly completely, but is of variable accuracy and consistency. On the other hand, it is the only online resource of its kind, offering something like a parallel for lute tablatures to RISM’s indexes of polyphonic vocal music sources from the Renaissance.

Contribution Type