The original question was simple: what if you learned language through complete situations instead of grammar tables? That question is still alive, but it’s sharpened. Most learners stall at intermediate because they know things they can’t use. The gap isn’t knowledge — it’s automaticity. Shima’s bet is that bounded repetition within meaningful contexts closes that gap, and that this works across languages and learner types.
Islands of fluency
An island is 50 to 100 sentences in a real-world context, drilled until automatic. Not a topic. Not a grammar point. A situation you can handle without thinking — ordering at a restaurant in Barcelona, navigating a train station in Tokyo, having a parent-teacher conference in Polish.
The grammar reveals itself through repetition in context. Explicit instruction is there when you want it, but you encounter the language before you’re taught rules about it. The unit of progress isn’t what you understand — it’s what you can produce without translating in your head.
What we’re learning
The approach is being tested across Spanish, French, Polish, Japanese, and Danish, with different learner profiles: heritage speakers, expats, stalled classroom learners. The interesting finding is that the island structure holds across languages, which wasn’t obvious. What started as a Japanese learning experiment turns out to be a question about how acquisition works generally.
The design itself is part of the experiment. No gamification, no streaks, no rewards. The hypothesis is that adult learners don’t need extrinsic motivation — they need the experience of fluency itself to be motivating. The whole aesthetic — warm functionalism, animations that settle and breathe, no “Great job!” energy — is testing whether you can sustain engagement through competence rather than dopamine.
Where it is now
iOS app in beta. A web-based Studio for authoring island content. Institutional and classroom models are being explored — another test of whether the approach scales beyond self-directed learners.