glossjarAbout glossjar
glossjar teaches the most common Lithuanian words, a few every day. Rate each word and a spaced-repetition scheduler (FSRS) brings it back just before you would forget it.
How it works
Every day you get 10 new words (change it in Settings) from the 5000+ most frequent ones, plus the words that are due for review. The FSRS algorithm schedules them. Your rating decides when a word comes back:
- Again
- You forgot it. You see it again in about a minute, and it counts as a lapse.
- Hard
- You remembered it, but with effort. The gap until the next review is shorter.
- Good
- You remembered it. The gap is normal and grows every time you get it right.
- Easy
- You knew it instantly. The gap is long, weeks.
Why it works
Memory fades along a forgetting curve: fast at first, then slower. Without review, most of a new word is gone within days.
Reviewing a word just before you would forget it strengthens the memory the most, and every review pushes the next one further away. That is why the scheduler shows each word at the right moment, not all the time.
A few minutes a day beats cramming: the words stay for good instead of vanishing after the test.
Where the words come from
The word list is the most frequent Lithuanian words, counted and lemmatised by us from the frequency lists of the Corpus of Contemporary Lithuanian (CCLL) and the OpenSubtitles 2018 word list. Genders, plurals, forms and stress come from English Wiktionary and hunspell_morphology_lt. A native speaker checked every word, its meanings and its pronunciation.
CCLL lemma frequency lists: V. Petkevičius, Vytautas Magnus University, CLARIN-LT 2025, CC BY 4.0. FrequencyWords (OpenSubtitles 2018): Hermit Dave, CC BY-SA 4.0. English Wiktionary via kaikki.org: CC BY-SA 4.0. hunspell_morphology_lt: Virginijus Dadurkevičius, MPL 1.1 / GPLv3+ / LGPLv3+. We changed and combined them, and share the resulting word list (words, ranks, genders, plurals and forms) under CC BY-SA 4.0.
CCLL · FrequencyWords · kaikki.org · hunspell_morphology_lt · CC BY 4.0 · CC BY-SA 4.0
Every pronunciation is a synthetic, machine-generated voice, not a human recording, checked by a native speaker. The voices are Google Cloud Text-to-Speech and two Piper voices: Ingutė (lt_LT-ingute-medium), Robertas & Claude, grown from Reginutė, under the NewGenLTU Open RAIL-D License 1.0 (Lithuanian State Digital Solutions Agency); and Reginutė (lt_LT-reginute1-medium), which is trained on the LIEPA corpus (Vilnius University, CC BY 4.0). Their stress dictionary is built from LIEPA and g2p-lt-lexicon (Arunas Smaliukas, CC BY 4.0).