Research data
The research datasets planned for Sanamahti, the principles of openness, how to cite and the attribution of the Kotus word list. Nothing is published yet.
No data from Sanamahti has been published yet. Download links will appear on this page when the first dataset is published. Until then, the page says what is planned and on what terms.
Status on 22 October 2026. The first edition of the living lexicon is in use in the game, but the measurement system is still at the design stage, and the licence of the open data has not been chosen yet.
Planned datasets
In the design, the living lexicon gives rise to four kinds of data:
- Word familiarity. For each word, the number of “know” and “don’t know” responses, the number of pseudo-word responses and the guess-corrected share.
- Surface forms. Rejected swipes, classified as an inflected form, a derivative, a compound or unknown, with the number of swipes and an estimate of the number of different players.
- Editions. The numbered weekly editions and their changes: words added, demoted, modified and rejected.
- Decision log. Cases, decisions, rationales, rule references and evidence with sources and dates.
None of these datasets exists yet. It has not been decided which of them will be published for download, or in what format. How the data comes about, and its limitations, are described on the method page.
Openness and licence
The aim is for the method, the editions and the decisions with their rationales to be public, and for the aggregate datasets to be downloadable and citable. The licence has not been chosen yet. The licence and the content of the first release will be decided before publication and recorded on this page.
The data produced by players is intended to be published as aggregates. They will contain no rows for individual players and no player identifiers. The name of the player who first found a word is shown only with that player’s consent.
If your research needs the data in a particular format or with a particular scope, say so now through the contact form. The publication format has not been locked.
How to cite
Citation instructions will be published with the first dataset. The editions are numbered, so a citation can point to a specific edition.
Until then, the project can be cited by the address of this page and the date of access. The author is Alex Liao.
Source data and the Kotus word list
The base of the living lexicon is Nykysuomen sanalista 2024, the word list of contemporary Finnish published by the Institute for the Languages of Finland (Kotus). Kotus publishes the list under the Creative Commons Attribution 4.0 International licence (CC BY 4.0). Four fields of the list are used: headword, homonym number, word class and inflection type.
Source: Nykysuomen sanalista 2024. Kotimaisten kielten keskus (Institute for the Languages of Finland). Licence CC BY 4.0. Retrieved on 13 October 2026 from https://kaino.kotus.fi/lataa/nykysuomensanalista2024.txt (SHA-256 bfa3d6a630fdff100d7765ee057f55f9b5b0c25045dbdfa809d3e4d1b4473cb3, 104,743 rows). The base was taken into use in the game in edition 1 on 22 October 2026.
Sanamahti modifies the material: a versioned layer of Sanamahti’s own changes is added on top of the Kotus list. For every word it is recorded whether it comes from the Kotus list, from the game’s earlier word list or from a decision made by Sanamahti. The Kotus list is the base and counts as strong evidence, but Sanamahti’s lexicon can differ from it.
Sanamahti is not a Kotus project, and Kotus has not taken part in it. Definitions from Kielitoimiston sanakirja, the Kotus dictionary of contemporary Finnish, are not copied or paraphrased. Word meanings are written by AI and labelled as AI-written.
The plan is also to use Yle’s news archive in the Language Bank of Finland and Wikimedia material as evidence sources, if their terms of use allow it. The attributions will be recorded here when their use begins.