FreeTxt: Democratising Qualitative Text Analysis Through Enhanced Usability and Bilingual Capability

August 28, 2026 · 2 min read
projects

FreeTxt is an open-source toolkit for analysing and visualising qualitative free-text data in English and Welsh, built for non-expert users in the public, cultural and education sectors. This project, funded by UKRI’s Research Software Maintenance Fund, strengthens its usability, robustness and bilingual NLP capabilities.

Funder: UKRI Research Software Maintenance Fund (Round 2), via the Software Sustainability Institute
Period: 2026 – 2027
Role: Project Co-Lead (technical development and Welsh NLP evaluation), with Dawn Knight (Cardiff University)
Partners: Cadw, National Trust Wales, National Centre for Learning Welsh, WJEC, National Museums Northern Ireland, The Audience Agency, Welsh Government (Welsh Language Technologies), Welsh National Opera; advisory input from UCREL, Lancaster University
Research theme: Welsh Language Technology & Multilingual NLP


FreeTxt was originally co-designed by Cardiff University and Lancaster University, together with National Trust Wales, Amgueddfa Cymru, Cadw, WJEC and the National Centre for Learning Welsh, and released in 2024 with AHRC funding. It is now maintained solely by Cardiff University and has grown well beyond its original partners: The Audience Agency integrated it into its UK-wide Audience Answers platform in 2024, and it is used across academic, public-sector and industry settings for sentiment analysis, summarisation, and frequency and collocation analysis of survey and feedback data.

This project addresses the usability, reliability and Welsh-language gaps that now limit FreeTxt’s further growth. Planned work includes:

  • Refactoring the codebase, introducing automated testing, and improving performance for multiple simultaneous users
  • Evaluating how well English-language NLP models transfer to Welsh, and developing bespoke Welsh summarisation and sentiment analysis models where they do not
  • WCAG 2.1 accessibility improvements and login-based user workspaces for institutional deployment
  • Establishing open-source governance, contributor guidelines and a shared development roadmap with project partners

New partners joining for this phase include Welsh National Opera, National Museums Northern Ireland, and the Welsh Government’s Welsh Language Technologies department, alongside original partners Cadw, National Trust Wales, WJEC and the National Centre for Learning Welsh.


Selected Publications

Fernando Alva-Manchego
Authors
Researcher in Natural Language Processing
My research interests include text simplification, readability assessment, multilingual NLP, Welsh language technology, and NLP for education and social care.