A Danish Chatbot for Kaj Munk: Experiences in Developing a Chatbot for the Study of Kaj Munk’s Ordet

David Jakobsen; Simon K. Pacis; Sara F. Jalk; Peter Øhrstrøm

doi:10.47852/

Authors

David Jakobsen Department of Communication and Psychology, Aalborg University, Denmark https://orcid.org/0000-0001-9607-2508
Simon K. Pacis Department of Communication and Psychology, Aalborg University, Denmark https://orcid.org/0009-0002-1829-2622
Sara F. Jalk Department of Communication and Psychology, Aalborg University, Denmark https://orcid.org/0009-0003-7495-4835
Peter Øhrstrøm Department of Communication and Psychology, Aalborg University, Denmark https://orcid.org/0000-0001-9432-7618

DOI:

https://doi.org/10.47852/

Keywords:

Kaj Munk, LLM, LLaMA2, LLaMA3, Danish

Abstract

This study presents a pilot project developing a chatbot that is capable of conversing regarding Kaj Munk’s drama Ordet (The Word) within a Danish cultural framework. Building on the argument for a dedicated Danish language foundation model, the project explores both linguistic and cultural challenges in generating accurate and idiomatic Danish responses. Three systems based on LLaMA2 (with and without Danish fine-tuning) and LLaMA3 were compared through factual and interpretive dialogs with the play. By fine-tuning a Danish translation of the OASST2 dataset and applying retrieval-augmented generation (RAG) to the digital Kaj Munk archive, this study evaluates model performance based on correctness, fluency, and cultural adequacy. Results show significant improvements with RAG and fine-tuning yet persistent traces of English language structure and bias. Beyond linguistic accuracy, the analysis highlights the need for models trained on culturally grounded corpora that reflect Danish literary traditions shaped by authors such as Grundtvig and Munk. The findings illustrate both the potential and limitations of large language models as tools for literary interpretation and for sustaining national cultural identity in AI-mediated dialogs.

Received: 14 January 2025 | Revised: 3 September 2025 | Accepted: 22 October 2025

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

The data that support the findings of this study are openly available in Figshare at http:doi.org/10.6084/m9.figshare.30250096.

Author Contribution Statement

David Jakobsen: Conceptualization, Methodology, Validation, Formal analysis, Resources, Writing – original draft, Writing – review & editing, Supervision, Project administration, Funding acquisition. Simon K. Pacis: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Data curation, Writing – original draft. Sara F. Jalk: Validation, Formal analysis, Investigation, Data curation. Peter Øhrstrøm: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Resources, Data curation, Writing – original draft.