ESTONIAN ACADEMY
PUBLISHERS
eesti teaduste
akadeemia kirjastus
PUBLISHED
SINCE 1997
 
Acta cover
Acta Historica Tallinnensia
ISSN 1736-7476 (Electronic)
ISSN 1406-2925 (Print)

Research article
The Historian and Modern Statistical Methods: Applications to Thirteenth-century Livonian Chronicles; pp. 173–204
PDF | https://doi.org/10.3176/hist.2026.2.02

Author
Nicolas Bourgeois
Abstract

The article examines the application of statistical and machine-learning-based tools to two thirteenth-century Livonian chronicles: the Chronicle of Henry of Livonia and the Livonian Rhymed Chronicle. It highlights both the potential and the limitations of these methods for the analysis of medieval texts. The article discusses key challenges, including sparse and multilingual corpora, difficulties of evaluation, and methodological complexity. The article argues that the principal value of increasingly complex computational methods lies less in autonomous historical interpretation than in their integration into transparent, verifiable, and collaboratively curated scholarly infrastructures. It therefore proposes principles for designing digital tools and metadata that are accessible, reliable, and open to continuous scholarly verification.

References

1. M. Piotrowski. Natural Language Processing for Historical Texts. (Synthesis Lectures on Human Language Technologies, 5). Springer, Cham, 2012.
https://doi.org/10.1007/978-3-031-02146-6.

2. N. Bourgeois, M. Cottrell, S. Lamassé, M. Olteanu. Search for Meaning Through the Study of Co-occurrences in Texts. – Advances in Computational Intelligence. IWANN 2015. Lecture Notes in Computer Science, 9095. Ed. by I. Rojas, G. Joya, A. Catala. Springer, Cham, 2015, 578–591.
https://doi.org/10.1007/978-3-319-19222-2_48.

3. N. Bourgeois, A. Pellet, M. Puren. Using Topic Generation Model to Explore the French Parliamentary Debates During the Early Third Republic (1881–1899). – Digital Humanities in the Nordic and Baltic Countries Publications, 2022, 4, 2, 35–51.
https://doi.org/10.5617/dhnbpub.11259.

4. S. Hochreiter, J. Schmidhuber. Long Short-term Memory. – Neural Computation, 1997, 9, 8, 1735–1780.
https://doi.org/10.1162/neco.1997.9.8.1735;
A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, I. Polosukhin. Attention is All You Need. – Proceedings of the 31st International Conference on Neural Information Processing Systems, 2017, 6000–6010.
https://dl.acm.org/doi/10.5555/3295222.3295349;
J. Devlin, M.-W. Chang, K. Lee, K. Toutanova. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. – Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Vol. 1 (Long and Short Papers). Association for Computational Linguistics, Minneapolis, 2019, 4171–4186.
https://doi.org/10.18653/v1/N19-1423.

5. A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, G. Krueger, I. Sutskever. Learning Transferable Visual Models from Natural Language Supervision. – Proceedings of the 38th International Conference on Machine Learning. Vol. 139 of Proceedings of Machine Learning Research, 2021, 8748–8763.
https://doi.org/10.48550/arXiv.2103.00020.

6. Liv-, est- und kurländisches Urkundenbuch nebst Regesten. Erster Band 1093–1300. Hrsg. von F. G. von Bunge. Kluge und Ströhm, Reval, 1853.

7. See Crusading and Chronicle Writing on the Medieval Baltic Frontier. A Companion to the Chronicle of Henry of Livonia. Ed. by M. Tamm, L. Kaljundi, C. S. Jensen. Ashgate, Farnham, 2011.

8. See A. V. Murray. The Structure, Genre and Intended Audience of the Livonian Rhymed Chronicle. – Crusade and Conversion on the Baltic Frontier, 1150–1500. Ed. by A. V. Murray. Ashgate, Aldershot, 2001, 235–251; A. V. Murray. Formulaic Language in the Livonian Rhymed Chronicle: Set Phrases and Discourse Markers in Middle High German History Writing. – Amsterdamer Beiträge zur älteren Germanistik, 2019, 79, 86–105; A. V. Murray. The Livonian Rhymed Chronicle as a Transitional Text: Formulaic Language in Middle High German Verse History. – Rilce: Revista de Filología Hispánica, 2020, 36, 4, 1324–1343.
https://doi.org/10.15581/008.36.4.1324-43.

9. Heinrichs Livländische Chronik = Heinrici chronicon Livoniae. Bearb. von L. Arbusow, A. Bauer. Hahnsche Buchhandlung, Hannover, 1955.

10. Livländische Reimchronik, mit Anmerkungen, Namenverzeichnis und Glossar. Hrsg. von L. Meyer. F. Schöningh, Paderborn, 1876.

11. W. Jezierski. Politics of Emotions and Empathy Walls in Thirteenth Century Livonia. – Making Livonia: Actors and Networks in the Medieval and Early Modern Baltic Sea Region. Ed. by A. Mänd, M. Tamm. Routledge, London, New York, 2020, 113–142. See also W. Jezierski. Emotional Bonding and Trust during Sieges, Twelfth–Thirteenth Centuries. – Risk, Emotions, and Hospitality in the Christianization of the Baltic Rim, 1000–1300. Brepols, Turnhout, 2022, 175–205, esp. 192–203.

12. This and all subsequent translations from Henry’s chronicle are taken from The Chronicle of Henry of Livonia. Trans. by J. A. Brundage. The University of Wisconsin Press, Madison, 1961.

13. The scikit-learn documentation provides a useful overview of these various algorithms:
https://scikit-learn.org/stable/modules/clustering.html (accessed 5/10/2026).

14. D. Blei, A. Ng, M. Jordan. Latent Dirichlet Allocation. – Journal of Machine Learning Research, 2003, 3, 993–1022.
https://www.jmlr.org/papers/volume3/blei03a/blei03a.pdf.

15. This should come as no surprise to historians, who are accustomed to developing concepts that make sense for studying a given period, even though the people living at the time would certainly never have used them themselves – feudalism, for instance.

16. Whereas words can be directly embedded in a latent space, modern language models generally operate on tokens rather than words. Tokenisation differs fundamentally from lemmatisation: whereas lemmatisation reduces each word to a common lexical form while leaving the syntactic structure of the sentence unchanged, tokenisation represents a sentence as a sequence of tokens. A token is an abstract unit that typically corresponds to part of a word, although it may sometimes correspond to an entire word or, in some cases, to several words. Crucially, the division of a word into one, two, or even several tokens is not guided by grammatical or linguistic considerations, but is determined primarily by the statistical distribution of linguistic units in the corpus on which the model is trained.

17. L. van der Maaten, G. Hinton. Visualizing Data Using t-SNE. – Journal of Machine Learning Research, 2008, 9, 2579–2605.
https://www.jmlr.org/papers/volume9/vandermaaten08a/vandermaaten08a.pdf;
T. Mikolov, K. Chen, G. S. Corrado, J. Dean. Efficient Estimation of Word Representations in Vector Space. – International Conference on Learning Representations, 2013.
https://doi.org/10.48550/arXiv.1301.3781.

18. The Chronicle of Henry of Livonia. Trans. by J. A. Brundage; The Livonian Rhymed Chronicle. Trans. by J. C. Smith, W. L. Urban. Indiana University, Bloomington, 1977.

19. More complex architectures can blur this distinction, particularly when language models are used to evaluate the outputs of other language models. Such approaches may raise difficult questions even for machine-learning specialists, and most historians are unlikely to need to venture into this particularly treacherous territory.

20. Liv-, est- und kurländische Urkundenregesten bis zum Jahre 1300. Hrsg. von F. G. v. Bunge, mit Ergänzungen von L. Arbusow jun., neu herausgegeben von F. Benninghoven, Geleitwort von P. Johansen. Hamburg, 1959.

21. A. Karjus. Machine-assisted quantitizing designs: augmenting humanities and social sciences with artificial intelligence. – Humanities and Social Sciences Communications, 2025, 12, 277.
https://doi.org/10.1057/s41599-025-04503-w.

Back to Issue

Choose new content alerts