Founder
White
Founder of Rhythm Word
About
The gap is not what you know. It is what you can use.
Rhythm Word started with a quiet frustration. The same word would come back in an Anki deck for the tenth time, the same dry definition, the same example sentence, and it still would not arrive when it was needed. The deck had trained recognition and nothing else. That is the real gap in vocabulary study: passive vocabulary is what you recognize when you read, active vocabulary is what comes out when you speak. Almost every study method only grows the first one.
So we re-built the vocabulary app from a different premise. You name a goal, an exam or an interview or the English you actually need, and the word list and every sentence are built for it. Every review opens with a sentence that has never been seen before, calibrated to your current level, so you read meaning instead of recognizing a card. Then the session ends in production: speaking, writing, and scenario dialogue ask you to use the word yourself. Recognition is where most apps stop. Here it is the halfway point.
What we believe
- Recognition is not knowing. A word you can pick out of four options is not a word you own. Productive knowledge has to be practised under production conditions, which means saying it and writing it, not seeing it again.
- Comprehensible input wins. Stephen Krashen's i+1 hypothesis is forty years old and still right: input one step beyond your level produces acquisition.
- Context, not isolation. Words live in collocations and scenarios, not on flashcards. You learn "take the train" faster than you learn "train (n.)".
- Spacing is non-negotiable. Without retrieval near the point of forgetting, vocabulary decays. FSRS-4.5 is the best public scheduler we have right now.
- Above B2, vocabulary is the bottleneck. Coxhead, Nation, Laufer all agree. Grammar is mostly fixed by intermediate level; what separates a Band 6 from a Band 8 is words you know.
- Learning should be private and yours. Your progress data is encrypted, never sold, never used to train anything that is not your own scheduler.
Who builds it
The product, briefly
- iOS, Android, and the web, plus a Chrome extension.
- A word plan built around the goal you name, not a generic frequency list.
- Production checks on every session: speaking, writing, and scenario dialogue.
- Six target languages: English, Spanish, French, German, Japanese, Korean.
- Eleven UI languages.
- Twelve preset learning goals plus custom goals.
- Six review engines covering swipe, spelling, pronunciation, fill-in-the-blank, matching, rapid fire.
- Article generation, offline mode, home-screen widget.
- Free to learn with no word cap and no expiry. A quarterly or annual subscription unlocks the advanced features; local-currency pricing varies by region.
Where to read more
For AI agents
Full machine-readable documentation: /llms-full.txt. Short summary: /llms.txt.