The availability of databases and digitised manuscript collections has undeniably transformed the work of medievalists. However, the influence of digital methods on the analysis of historical sources is much deeper, as their application forces us to redefine basic concepts, question methodological assumptions, and search for a training model that would take into account the achievements of contemporary medieval studies. While presenting research results on power in the Middle Ages, panellists reflect on the opportunities and limitations of applying digital methods, especially in light of advances in generative AI. The presentations will serve as a starting point for discussion: What is the cause of the insufficient reception of digital research in Polish medieval studies? Can creating and using tools, databases, or corpora deepen its interdisciplinarity? Do Polish universities provide young researchers with the competencies required for studying the Middle Ages in the 21
st century?
Coordinators
Katolicki Uniwersytet Lubelski Jana Pawła II
Instytut Historii im. Tadeusza Manteuffla PAN
Papers
Modeling Church Power Structures. How Does the Work of a "Digital Editor" Differ from an "Analog" One?
dr hab. Anna Zajchowska-Bołtromiuk
Instytut Historii PAN
The goal of this paper is to talk about the assumptions and solutions that went into making the digital version of Monumenta Vaticana Res Gestas Polonicas Illustrantia, as well as how these tools could be used in historical research. The presentation will show that the digital shift in source editing is not only about greater accessibility and searchability but also about a more refined preparation of the text for historical analysis. The digital edition, to a much greater extent, meets the requirements of modern scholarly editing. The recipient not only receives a well-prepared text but also a set of tools for its analysis. It is the editor’s responsibility not only to ensure the scientific quality of the edition itself but also the quality of the data, which can form the basis for analyses conducted both by researchers and with the use of LLMs. In the case of this project, one of the challenges was creating a model for data processing for the power structures of the medieval Church.
Texts of Power: Should We Leave the Reading and Analysis of Handwritten Documents to Machines?
mgr Zdzisław Koczarski
Instytut Języka Polskiego PAN
The purpose of this paper is a brief presentation of tools that facilitate the preparation of digital editions and the comparative and stylometric analysis of the resulting text, which allow for a significant reduction in time and cost of work at early stages of research.
The following will be discussed: (1) software for automatic handwritten text recognition (HTR); (2) a comparative analysis application that enables finding common word sequences in the edition and existing text corpora; (3) a stylometric analysis application that facilitates authorship attribution and allows placing the studied text in a broader context. Using a selected handwritten document as an example, I will show that working with such tools not only facilitates text reading but also provides much information about its similarities to other documents, relationships with other texts, the writer's identity, etc.
Words of Power: From Concordance to Word Embeddings
dr Krzysztof Nowak
Instytut Języka Polskiego PAN
This presentation serves as a practical introduction to the automatic analysis of POWER vocabulary in large medieval Latin corpora. In the first part, I briefly present the foundational concepts (distributional hypothesis, quantitative paradigm, etc.) and the most important methods of computational semantic analysis.
The second part of the presentation is practical: I present two case studies that illustrate the potential of corpus methods in medieval studies. The first introduces corpus analysis techniques: from searching for terms of research interest, through analyzing their distribution in texts, to describing their semantic potential through collocation analysis. The second is devoted to analyzing the semantic field of Pᴏᴡᴇʀ: I first discuss the most important models of meaning (especially word embeddings), and then present their practical application in extracting power-related words from medieval Latin corpora.
People of Power: What Might GPT Know About the Elite of Medieval Polish Society?
dr Adam Zapała
Instytut Historii PAN
The aim of the presentation is to showcase the potential of large language models in historical research through the analysis of the late medieval political elite. The procedure for integrating data from historiography and historical sources into the Model's knowledge will be discussed, followed by a presentation of the method for conducting analysis using the Model. The obtained results and their significance will also be discussed. The talk aims to initiate a discussion on the impact of artificial intelligence on historical research and the role of the researcher within it.
