DSM Workshops at U Fribourg

A two-day workshop taught by Stephanie Evert for the doctoral programme Multilingualism: Acquisition, Education, and Society at Université de Fribourg on 1–2 October 2026.

The first workshop gives a theoretical and practical introduction to the foundations of distributional semantic models (DSMs), including extensive hands-on experience with the wordspace package for R. Participants will explore how collocational profiles and distributional vector representations are related, and what they can tell us about language and word meaning. They will have the opportunity to discuss potential applications of distributional semantics in their own research.

All slides, example scripts, and exercises can be downloaded from the Course materials page. See Software & data for installation instructions and some larger example data sets.
Materials are currently being revised and will be available for download by the start of the workshop.

• 9:00–10:30 Introduction: from collocations to DSMs
  • slides: Part 1 (slides, handout ) – code: hands_on_part1.R
• 10:50–12:15 Creating a DSM: taxonomy of parameters
  • slides: Part 2 (slides, handout) – code: hands_on_part2.R
• 14:00–15:30 DSM exploration & discussion
  • exercise: hands_on_part2.R (explore DSMs and their parameters) – group discussion: are DSMs useful for your research?
• 15:50–17:00 Hands-on: build and use your own DSM
  • slides: Part 2 (building a DSM) – code & exercises: hands_on_part2_input_formats.R

• 9:00–10:30 Evaluation of DSMs: tasks, data sets, results
  • slides: Part 3 (slides, handout) – code: hands_on_part3.R
• 10:50–12:15 Hands-on: DSM evaluation
  • hands-on exercise: hands_on_part3.R – extended exercise: hands_on_part3_extended.R (for those with patience and fast CPUs)
• 14:00–15:30 DSMs in linguistics: meaning by collocation vs. distributional similarity
  • slides: Part 4 + Part 5 (mathematical insights) – code: – group discussion
• 15:50–17:20 Discussion & outlook
  • spillover material – group discussion – outlook on embeddings – code: (Schütze 1998)