Digital resources in the Social Sciences and Humanities OpenEdition Our platforms OpenEdition Books OpenEdition Journals Hypotheses Calenda Libraries OpenEdition Freemium Follow us

Distant Diplomatics in Catalonia, Andorra and France

The RomTur-.project collects historical patrimony in the border region of Catalonia, Andorra, and France. Charters are part of this, and the project team is creating a databse for this purpose. They invited international colleagues for a conference on“La documentation médiévale (IX – XII siècles) : création, bases de données et gestion numérique” in Barcelona 2-4 décembre 2025.

romtur-Logo


This included by presentations by me (“Charters as a data lake? Current state of distant diplomatics”), and by our advisory board members Antonella Ambrosio (“Tracce notarili tra testo, scrittura e spazio: fonti del mezzogiorno medievale (secoli X-XII) in prospettiva digitale”) and Sébastien Barret (“Les bases de données françaises d’actes pour le Xe-XIIe siècles, entre fonds d’archives anciens et collections numériques”). There were several other highly interesting contributions on automatic transcription, online charter collections, or large scale databases by Marlène Helias-Baron, Olivier Canteaut, Vera Isabell Schwarz-Ricci, Coral Cuadrada, Alicia Fornés, Miriam Palomba, Joana Maria Pujadas-Mora, Miguel Calleja, Cristina Jular, Michelle Bubenicek, Xavier Costa, Maria Soler, Maria de la Pau Serrat, Esther Travé Allepuz, Mariona Coll, Nilo Pedrazzin, Eric Mermet, Davide Gherdevich, Sébastien Poublanc, Ramon Alberch (International Council of Archives), Antoni Albacete, Alan Capellades, Rafel Ginebra, Joan Maria Quijada, and, of course, the core team of the project: Ignasi Baiges, Daniel Piñol, Flocel Sabaté, Romain Saguer, Jaume Marcé, Clara Renedo. They covered many aspects of the problem: hard core diplomatics, computer science, archival practice, and cultural history.

In my contribution, I described, how we are working on an environment that supports distant diplomatics with methods that could be conceptualised with the “data lake” term. The business informatics use of the term is certainly beyond our scope. However, we face the same problems in how preprocess the highly variable data that downstream tasks don’t get messy.

Putting samples of distant diplomatics work, for instance by Miguel Calleja, who gave his own presentation at the conference, Nicolas Perreaux’ CEMA corpus, and Michael Gervers work with the DEEDS corpus, Georg gave some – rather arbitrary – examples of possible research question we could try to answer with distant diplomatics methods:

  • Scribes of papal charters correlate with recipients?
  • Who and when the „das ist unser ernstlich meynung“ clause is used?
  • Show me all notarial signs used by those active in Cologne in the first half of the 15th century!
  • Has a charter text been drafted by the chancery (individually or re-using a formulary) or by the petitioner/receipient? („Empfängerausfertigung“)
  • Is there a correlation between issuers and notaries?
  • Is there a correlation a between issuer and clauses?
  • What are the shared patterns of charter production in a specific region?
  • What we want to know …
  • Where does the language of papal charters impress local charter practice earliest / most?
  • By which legal practice (notaries, common law, roman law, …) is charter language most influenced?
  • Is charter layout a local, personal, regional, or supra-regional development? Where and when does it start?

Monasterium.net and other large scale charter corpora can help answering this type of questions, if they get in the right scale by aggregating data from different sources and processing them on at least some basic levels – and this is precisely the process, business informatics call “data lake”. He explained how we in the DiDip project contribute to “data distillage” aiming at reducing the heterogenity of the collected data. For instance by automatic annotation of visual areas on the images of charters : the charter itself, the text on the charter, seals, decorations. Or by identifying text lines, transcribing them automatically, cleaning automatically transcribed texts, geo-referencing places of issuing etc. All of this needs sophisticated methods to exclude garbage: places in Southern America are highly improbable for issuing a 14th century charter preserved in Austria.

DiDip is tackling this for instance at the image level, where we introduced into the automatic detection of areas on the image a “ignore” class for pins holding the parchment during digitisation, for instance, and a class for the calibration card that supports measuring the real size of the document even with an automatic approximation.

Tamás suggests enhancing the quality of HTR through XLM-Roberta and BART-Base based postprocessing, finding error candidates by comparing to linguistic probability and suggesting better text. Similar to the first step, other methods can support the cleaning by identification of the rubbish: a simple language detection method can already tell if a passage is close enough to readable text and by this assigned to a language with high probability. Distance to the preserving archive can suggest outliers in geo-referencing of places of issuing.

These methods give already first insights into the data: the distance between place of issuing and location of the preserving archive is also a marker of geographic outreach of the issuing party. In the monasterium.net corpus obviously Rome and Avignon play a significant role. Vienna is important by the Habsburg rulers? The read dots in Northwest of Spain reflect history of archival concentration (in the National Archives of Madrid) or place of unexpected importance?

There are more of these kind of domain agnostic methods: a generic LLM request on summarizing dirty HTR can be not only incomplete but misleading, but it can also return more detailed information from a early-modern German charter than the archivist had time to record. Daniel Luger and Florian Atzenhofer-Baumgartner were able to identify a reference to Durantis Speculum iudicale, when they shuffled texts around that were suspect to be related to arbitration by sharing multiple word sequences with known arbiratration related documents (Luger & Atzenhofer 2024).

What is to learn? The data lakes of huge charter corpora can be cleaned up and studied with computational methods – if we know what to ask.

You can find his presentation here.


OpenEdition suggests that you cite this post as follows:
Georg Vogeler (February 2, 2026). Distant Diplomatics in Catalonia, Andorra and France. DiDip. Retrieved March 7, 2026 from https://didip.hypotheses.org/4599


You may also like...