How Cotobara Lessons Work and How to Set Your Pace

In Cotobara, "lessons" and "reviews" are two different things. A review is a word you already learned coming back on schedule. A lesson is meeting a word for the first time. This post is about lessons: what happens inside one, and how many you should be doing.

Which words you get, and in what order

You don't pick your words. Cotobara picks them, and the ordering is the whole point.

Candidates are ranked by a combination of JLPT level and real-world frequency. N5 words outrank N4 words, and common words outrank rare ones, so what you get next is roughly "the easiest useful word you don't know yet."

Importantly, that ordering is a priority, not a wall. A very common N4 word can arrive before a genuinely obscure N5 word, because in practice you will hear the common one first. You're not locked out of a level until you've cleared every last word beneath it, which is how learners end up grinding through vocabulary they will never actually meet.

Two settings shape the pool:

• Starting level: set it to N3 and Cotobara stops offering the easier stuff. Half-levels like N3.5 skip the most common half of the level below.
• Rare-word filter: raise it and vocabulary below your frequency threshold is left out entirely.

Both live in Settings, and neither disturbs words you've already started.

A lesson is five words, and every card asks you something

Lessons come in batches of five. That's small enough to hold in your head and finish in a few minutes, and large enough that the words get spaced apart inside the lesson itself.

A lesson is not "read five cards, then take a quiz." It's a run of steps, and none of them are passive.

Meeting a word starts with a guess. You see the Japanese with furigana and a pronunciation button, and four possible meanings. You pick one before you're told anything. This is a pretest, not a test: a wrong guess helps as much as a right one, nothing is scored, and the app says so on the card. Attempting an answer before seeing it is one of the most reliably useful things you can do to a fact you're about to learn.

The moment you guess, the card opens as feedback:

• The English meaning beside the word, plus the alternate meanings the grader will accept later.
• The word breakdown: kanji components, how the parts combine, a mental model, usage notes (more on those here), sitting under the meaning it explains, not hidden behind a tap.
• An example sentence with natural-sounding audio, where you can hover most words for a reading and meaning (coverage isn't complete and it's occasionally wrong).
• A row of chips with the frequency badge and a bookmark toggle, and an own-wording field if you want to add your own accepted answer up front.

Checks, interleaved

Then the lesson comes back for it. Each word is checked one word later: word one is checked right after word two is taught, and so on, with the last word's check at the end. So every word is retrieved twice inside a single lesson, with a gap in between rather than seconds apart. It's the same principle the review schedule runs on, compressed into five minutes.

The checks are multiple choice, but not the multiple choice you're expecting. Wherever the word appears verbatim in its own example sentence, Cotobara masks it (＿＿＿) and asks you to fill the blank. You're choosing a word in context, not matching a Japanese string to an English one. Where a cloze question isn't possible, it falls back to a meaning question. The wrong options are picked to be close in length to the right one, so nothing is given away by shape.

Two rules:

• Miss a check and the word's learning notes open right there, at the moment of failure, and the word comes back later in the lesson.
• A word you miss twice isn't asked a third time. It's learned along with the rest and comes back through your review queue tomorrow, which is where a word you don't know belongs. A lesson that grows every time you get something wrong is a lesson with no end you can see.

A journey rail across the top shows the whole run: a tile per word, a dot per check, and a dashed node where a miss has been recycled. You can always see how much is left.

Checks deliberately do not autoplay audio before you answer; a question that pronounces its own answer isn't a test. Once you've answered, autoplay follows your setting like everywhere else.

When you finish, the five words enter your review schedule, and the schedule takes over. If you have allowance left, "Keep learning" starts the next batch.

Your very first lesson ends in a practice review

Lessons ask you to tap; reviews ask you to type. On day one, nothing used to bridge those, so a beginner's first typed card arrived hours later, alone, with their schedule riding on it.

Now the first lesson you ever finish sends you back to the dashboard with your reviews card lit up and holding those same words. Press it and you go through the real review cards, same questions graded as you answer, except that nothing is written back. No schedule moves, nothing promotes or demotes. It exists so the review screen is somewhere you've already been by the time it counts.

Choosing your daily limit

This is the most consequential number in the app, and most people set it too high.

Every new word you learn today isn't a one-time cost: it's roughly a dozen reviews spread across the next year. Ten new words a day is a comfortable-feeling lesson session and, a few weeks later, a review queue that greets you with hundreds of cards every morning. That queue is where people quit.

Some rough guidance:

• 3–5 a day: the sustainable default. About 1,000–1,800 words a year, which is a genuine JLPT level or more.
• 10 a day: a good pace if you're studying seriously and have 20–30 minutes daily, every day, including bad days.
• 20+ a day: only worth it if you're pushing toward a fixed exam date and you have honestly checked that you can clear the reviews it generates.

New words are a Premium feature: a paid plan goes up to 100 a day and lifts the review cap entirely. The free plan keeps the beginner course and reviews for 50 words a day, so a deck you have already built stays yours to study.

Let the app tell you when you're wrong about your pace

You don't have to guess. Cotobara's dashboard tracks your rolling seven-day accuracy and nudges you when your numbers say your pace is off:

• Consistently above 90%? You have headroom. It'll suggest raising your daily limit.
• Below 80%? You're outrunning your own retention, and it'll suggest lowering it, or drilling your recent mistakes first.
• Between 80 and 90% it says nothing, on purpose. That band is the sweet spot, and an app that nags you inside it is just noise.

The nudges only appear once you have enough reviews in the window to mean something. Accuracy is the honest signal here: not how many words you started, but how many you actually kept.

Start for free and do your first lesson now. Or see everything else Cotobara does.