By: Claire Lavarreda, DITI Research Fellow & PhD Candidate in History
I. Overview
The Dragon Prayer Book Project is an ongoing endeavor to study, transcribe, translate, and encode a medieval manuscript at Northeastern University. Led by Dr. Erika Boeckeler, the Dragon Prayer Book Project has three main fields of research related to the book, including transcription, conservation, and microbiology. Written and illustrated by Dominican nuns at the Convent of Saint Catherine in Nuremberg Germany, the date of creation is sometime after 1461.1 The book contains over six hundred pages, with numerous musical annotations and vivid embellished letters.
Over the years, students have been working diligently to transcribe and translate the Dragon Prayer Book according to TEI (Text Encoding Initiative) guidelines. The current Dragon Prayer Book XML schema was created by Dr. Julia Flanders. It was customized from the general TEI guidelines and intended for simple transcription.2 At its core, TEI (along with other markup languages) is intended to organize and represent texts. It is mostly linear and hierarchical in nature, making it a clear choice for encoding content like paragraphs, but slightly more complex for marginalia, given that marginalia has many non-linear features – such as diverse locations and meanings.3 While markup languages and maginalia are not inherently misaligned, projects like the Dragon Prayer Book must utilize TEI in creative and interpretative ways to accurately represent their manuscripts.
In Fall of 2024, I joined the Dragon Prayer Book Project in a support role as part of my research assistantship with the Digital Integration Teaching Initiative. After speaking with Professor Boeckeler and establishing the current needs of the DPBP, it was determined that a three-fold marginalia project would be ideal, with deliverables to include:
(1). An inventory of all the marginalia in the Dragon Prayer Book, containing text translations, identification of the language, scribe, and location, and links to translation sources.
(2). A written document detailing categories and methods for encoding the marginalia.
(3). The encoded marginalia as encoded in the Oxygen XML editor & published on GitHub.
In order to gain familiarity with the project, I attended the weekly DPB meetings and practiced translating several pages from Latin to English in Oxygen.4 By Spring 2025, I moved fully into the marginalia aspect of the project, completing the inventory before creating the guidelines and encoding the marginalia.
II. Establishing A Theory of Marginalia for the Dragon Prayer Book Project
In order to develop a method for encoding marginalia, it is necessary to have a theory of marginalia first.5 Essentially, what does marginalia mean to the Dragon Prayer Book? Is it considered a general note, a correction, or a direct part of the text? Is it important to preserve the physical order of the marginalia, and if so, what assumptions does the project make regarding reading order? Complicated questions arose during my research process, necessitating the creation of a basic framework before doing any actual encoding. The following principles established a working theory for my project:
(1). The Dragon Prayer Book Project prioritizes content over form: I determined that the DPBP is more concerned with the manuscript’s content than the physical representation. This decision was based on the pre-existing encoding and transcription practices for the main text. For example, <choice> and <reg> are used heavily throughout the encoded pages, and vivid embellished letters are not treated as visually distinct despite being large and colored. The content – specifically, the readability of the content – is a priority for the project.
(2). The encoding process for marginalia must be flexible in order to integrate into the existing encoded pages: As I worked through the documents, it became clear that some previous students had their own systems for encoding marginalia. There were several instances of <choice> and <seg>, and in the interest of preserving their work, I needed to create an encoding process that didn’t disrupt what had already been done. Furthermore, the encoded marginalia needed to be “editable” for future students or project decisions.
(3). The marginalia should be classified beyond “note,” but does not need to be complicated (no need for elements like <locus> or <decoNote> at the moment): This principle was determined in conversation with Dr. Erika Boeckeler during one of our weekly meetings. After reviewing some potential elements, Dr. Boeckeler provided helpful feedback on what was actually needed.
With the above guidelines established, I was able to proceed with the rest of my project.
III. Process: Categories, Elements, & Encoding
The first step in the encoding process was to create an inventory of all the marginalia in the DPB. In addition to images of the marginalia and their corresponding page numbers, I added potential definitions, translations, languages, and scribes. The next step was to categorize the nature of the marginalia using general TEI Manuscript Description elements and attributes. <Note> is the grounding element, with further specification made through <corr>, <add>, <del>, <figure>, and <lang>.6 Attributes used to differentiate scribal notes and the original author’s notes are represented with @type= “authorial” and @type= “editorial”.7
With all marginalia found and categorized, I then chose a “system” for encoding that would prioritize the content of the notes. I decided to model the encoding after the notes system used in the Women Writers Project, which places all notes within a <hyperDiv> at the beginning of the manuscript. These notes are assigned a unique xml:id, which correspond to the individual notes (which have the matching xml:id) throughout the text.8 This ensured that the main text would be easily “readable” (not bogged down with in-line notes), while preserving the important content. I then assigned the marginalia an @xml:id, a @type, and an @xml:lang, with the final product visible below.
IV. Results
V. Future Considerations & Final Thoughts
The Dragon Prayer Book Project has involved many participants, and will continue to do so. With this in mind, there are several things future students will need to address, including (1). Research on different scribes, (2). Establishment of a personography, which would then entail pointing to scribal personography entries using @hand, (3). Musical annotation, (4). The use of <hyperDiv> versus <standOff>, and (5). Assigning target attributes with their full pointers. These considerations are also the product of encoding non-linear and manuscript features in TEI, requiring specification for different types of handwriting, script, and authors. Further, students will need to examine the previous encoders’ uses of <choice> and <seg> for marginalia that are scattered throughout the text.
Given the expansive and ever-changing nature of the project, it is likely that the current marginalia encoding system will need to be adapted and specialized. However, the Dragon Prayer Book Project now has a theory of marginalia and a basic system to build upon, providing a foundation for future annotation work. Links to the final research project follow: the inventory, these guidelines, and the edits made in Oxygen, which are visible on GitHub.
Endnotes:
1. Ian Thomsen, “Deciphering the Medieval Secrets of the Dragon Prayer Book,” Northeastern Global News, Oct. 3, 2019, https://news.northeastern.edu/2019/10/03/exploring-dragon-prayer-books-medieval-prayers-and-chants-recited-by-german-nuns/
2. Julia Flanders, “Dragon Prayer Book Schema,” Github, 2022, dragon-prayer-book/schema/dragon_schema.xml at main · bloucha/dragon-prayer-book · GitHub
3. These include varied note locations (on the top, bottom, or side of a page; over text, etc) and different meanings/intentions (as a correction, addition, unrelated thoughts). I want to thank the brilliant and accomplished Dr. Avery Blankenship for her wonderful explanation of marginalia and encoding. Her insights, resources, crash-course, and advice were critical to this project. For more on Avery, check out her website: https://www.averyblankenship.com/.
4. The translation process is relatively straight-forward and rarely relies on online translators, like Google or DeepL. A student first selects an available page (already transcribed in Latin), opens it in Oxygen, and then searches for the Latin prayer/hymn in the Office Book for Dominican Sisters, the Dominican Missal and the Douay-Rheims Bible, in addition to second-tier sources. The student then enters the English translation in between the <ab type=“translation” xml:id=“p0566_translation” corresp=“p0566_translation“></ab> tags, noting the source used.
5. TEI Consortium, eds. “Manuscript Description,” Guidelines for Electronic Text Encoding and Interchange, 1.7.0. January 24, 2025, http://www.tei-c.org/Vault/P5/1.7.0/doc/tei-p5-doc/en/html/DS.html#DSFLT
6. “Note” is a general umbrella for all notes and marginalia, which are further defined by <corr> (correction), <add> (addition), <del> (deletion or omission), <figure> (indicating a drawing, table, or some other non-text symbol), and <lang> (referring to the language of the text).
7. While I was creating the inventory, I noted the presence of more than one scribe. However, I am not an expert in medieval German manuscripts, and I felt that encoding the scribal differences was beyond the scope of this project. “Type” was sufficient enough to distinguish between the author’s edits and the general scribal edits. Further research is needed to determine if there are, in fact, several annotators, at which point the DPB Project could consider assigning them unique identifiers and developing a personography.8. Women Writers Project, “Internal Documentation: <note>,” WWP, accessed April 4, 2025.
