Who makes it
SveLingo is built and maintained by Wenlin Fan, a software engineer based in Sweden who started the project to keep the words from his own Swedish classes. The code and the dictionary pipeline live on GitHub. There is no company behind the site, no advertising and no investors; it is free because it started as the tool its maker wanted while learning Swedish.
How the dictionary is built
Every entry starts from the open Swedish lexical resources SALDO and SALDOM, which give the headwords, their senses and the inflection paradigms. From those paradigms we generate the complete form table for each word — hund, hunden, hundar, hundarna — and attach the grammar rule that produces each form, so a word page can link straight to the chapter of the grammar book that explains it.
Example sentences come from a corpus of about 800,000 Swedish sentences aligned to senses. The phrases shown under a headword are kept only when that phrase actually occurs in the corpus; anything a language model suggested but the corpus could not confirm is left out. Pronunciation transcriptions come from the Braxen and NST pronunciation lexicons, and the audio is generated with a Swedish neural voice — clearly useful for hearing a word, not a substitute for a native speaker. CEFR levels are estimated by a model calibrated against the Kelly list and reviewed word by word for the core wordbooks.
The wordbooks — Core A1 to C1 and SFI B, C and D Essentials — are curated by hand from frequency data and course syllabi. Word pages that do not yet meet the quality bar (a sense with an example, audio, a form table) are kept out of search engines on purpose.
How the review works
Saved words are scheduled with an SM-2 based spaced-repetition scheduler: the first review the next day, the second six days later, then the previous gap multiplied by an ease factor between 1.3 and 2.5 that moves with how confidently you answered. The formula is documented in how SveLingo works, and you can try it there.
What we do with your data
The dictionary needs no account. An account stores your saved words, wordbooks and review history so they can sync between the web and the iPhone app. Analytics are opt-in on the app and described in the privacy policy. Questions and corrections are welcome through the contact page; entries are corrected by hand, and corrections roll into the next dictionary build listed on the updates page.