LahTex: Slovene Easy-to-Read Corpus for Text Simplification Using Artificial Intelligence

LahTex

Project Code: V6-2523

The project addresses the issue of limited access to information among groups with lower reading literacy, such as persons with special needs, immigrants, and the elderly. Easy-to-read texts, incorporating linguistic adaptations that facilitate comprehension, are essential for the more active inclusion of persons with special needs in society and serve as a means of supporting the integration of immigrants. The key challenges in the field of Easy Language in Slovene primarily arise from the lack of adequate infrastructure and resources. At present, no specialised Easy Language corpus exists in Slovene, which significantly hampers the potential for systematic research and the development of effective tools and methodologies for text simplification. An additional challenge lies in the heterogeneity of the target groups — their needs, abilities, and expectations vary considerably, necessitating different levels and types of linguistic adaptation. Simultaneously, the potential of artificial intelligence as a means of fostering inclusive communication remains largely underexploited in the Slovene context. A major limiting factor is the absence of appropriate training data required for the development of high-quality models for simplified text generation. The primary objective of the project is therefore the creation of the first easy-to-read corpus in Slovene, which will enable empirical research into easy-to-read Slovene and support the development of artificial intelligence tools for text simplification. The corpus will comprise authentic easy-to-read texts, machine-translated texts, and artificially generated easy-to-read texts. Drawing on this dataset, a text simplification system will be developed based on the Slovene large language model GaMS, tailored to different target groups through various easy-to-read subcorpora. By developing resources and tools in the domain of easy-to-read Slovene, the project contributes to digital inclusion and the promotion of human rights by facilitating the creation of accessible content for vulnerable populations.

Project objectives

• Define a conceptual framework of easy-to-read and plain language with respect to different target groups of users.
• Establish the first specialized corpus of easy-to-read Slovene.
• Use artificial intelligence tools to generate simplified versions of texts in Slovene.
• Develop a set of core vocabulary and a basic grammar of easy-to-read Slovene.
• Disseminate research findings to the professional and wider public.

The project’s objectives are directly aligned with several strategic documents of the European Union, which emphasize the importance of linguistic accessibility, digital inclusion, and human rights. The need for accessibility of information is already mentioned in the European Charter for Regional or Minority Languages from 1992, which promotes the protection of linguistic diversity and access to information in a comprehensible form, including intra-language versions such as easy-to-read. In the Strategy for the Rights of Persons with Disabilities 2021–2030, the European Commission explicitly highlights the need for accessible communication for all citizens, the development of easy-to-understand content in public and private services, and the promotion of digital accessibility, also with the help of artificial intelligence tools. The Digital Education Action Plan 2021–2027 further supports innovation in digital learning and accessibility, focusing on innovative learning tools, among which are also artificial intelligence tools for linguistic simplification.

Researchers:

• Institute of Contemporary History:
  • Vojko Gorjanc
  • Matej Klemen
  • Filip Dobranić
• Faculty of Computer and Information Science:
  • Marko Robnik Šikonja
  • Aleš Žagar
  • Tinca Lukan
• Faculty of Arts, University of Ljubljana:
  • Špela Arhar Holdt
  • Iztok Kosem
  • Simon Krek

Documents:

• Data Management Plan (slo)
• Contract for text contributors (slo)
• Communication Strategy (slo)

Results:

• Stanje virov in korpusov lahkega branja v Evropi – November 2025
• Konceptualni okvir lahkega branja

Presentations and events:

• Jezikanje: Novi korpus lahkega branja v slovenščini, 25. 05. 2026
• Srečanje projekta LahTex, 9. 6. 2026