Updated 2 November 2026

Data sources

Wordloader translates nothing on its own. Everything you see on a word card is taken from the dictionaries of Slovenian research institutes, and every word shows where its data came from. If a word is in none of the dictionaries, the app says so.

Sentences for the case exercises

SUK 1.1 corpus, a manually annotated training corpus of Slovene. Jožef Stefan Institute and University of Ljubljana. Licence CC BY-SA 4.0, source: hdl.handle.net/11356/1959.

This gives the Slovene sentences and the case and number in them. Our selection is shared under the same licence. Russian translations, situation descriptions and explanations were picked by AI and carry a dot.

Stress, grammar, forms, frequency

Sloleks 3.1, the morphological lexicon of Slovene. Centre for Language Resources and Technologies, University of Ljubljana (CJVT UL). Licence CC BY-SA 4.0, source: hdl.handle.net/11356/2080.

This gives the dictionary form, the stress, the part of speech, gender and aspect, every case and verb form used by the case drill, and how common the word is in the Gigafida corpus. The learning order follows that frequency: common words first, rare ones later.

Senses and definitions

SSKJ², the Dictionary of Standard Slovene, second edition. Fran Ramovš Institute of the Slovenian Language, ZRC SAZU, published at fran.si. The dictionary is not published under an open licence.

eSSKJ, the growing dictionary of contemporary Slovene, from the same institute and the same portal. Licence CC BY-NC-ND 4.0: non-commercial use only, no derivatives. It supplies words that SSKJ² does not have yet.

Russian translations

The Russian-Slovene dictionary by Janko Pretnar. ZRC SAZU, published at franja.si. Licence CC BY-SA 4.0. Every Russian equivalent on a card is verified by a reverse lookup in this dictionary.

How we treat the dictionaries

An entry is requested one at a time, and only when somebody actually needs the word; it is then stored on our side so that other people’s servers are not asked twice for the same word. We do not download whole dictionaries. We have written to the Institute of the Slovenian Language about the terms for a paid version and are waiting for their answer.

What the language model does

Sorting the Russian equivalents into the senses of a Slovene entry is done by the Claude model from Anthropic. It does not invent translations: it is given the dictionary entries and its job is to distribute the existing equivalents across the senses. Anything a dictionary does not confirm is marked on the card with a dot, so you can tell a dictionary entry from a suggestion.

Thank you

Apps like this one exist only because Slovenian institutes open up their data. Thank you to CJVT at the University of Ljubljana and to the Fran Ramovš Institute of the Slovenian Language at ZRC SAZU for that work. If you hold rights to this material and see something wrong in the app, write to alex@alexvolkov.com and we will put it right.

Back to the home page