tap to hear

학습 끝 · The learn list is empty

These words are only on this device.

나의 성장 · Your growth

Every word you have proven on a test at least once — a record of how far you have come.

학습자 · LEARNER

북알라바마 한국학교 · 한국어 실험실

—

학습 기록부 · Record of study

단어 · Words: 0
최고 훈장 · Highest: —
상태 · Standing: —
확인 · Checks: 0
학습 일수 · Days: 0
자신감 · Mean conf.: —

훈장 · Commendations

다시 볼 열 단어

needs work

Practice only — nothing here is recorded. The drip meets these words cold later, and that is the measurement we trust.

가장 잘 아는 열 단어

strongest

Your surest words, by confidence. These have earned their place in the long-term lane and come back rarely.

Your words, day by day

단어 기록 · Word history

지금 일상 대화 속 내용어의 0.0%를 알고 있어요
You currently know 0.0% of the content words in everyday spoken Korean.
이 숫자에 대해 더 알아보기 · Learn more about this number

명예의 전당 · The Hall of Names

Everyone who has earned a medal, forever. Nobody is ranked — being here is the achievement, and it never comes off.

불러오는 중 · loading…

열 개를 안다고 하셨어요.

Voice

이 숫자는 어디서 왔나 · Where this number comes from

Short version: it is arithmetic on 5.18 million words of Korean dialogue, and we will show you the whole calculation.

What it counts

It is the share of content words — the words that carry the meaning: nouns, verbs, adjectives. Not the share of sentences you would understand. Knowing every brick is not the same as reading the blueprint.

So "30% of the content words" means that in a typical spoken sentence, roughly three out of every ten meaning-carrying words are ones you already own. The grammar holding them together is a separate job.

Where the raw counts come from

We started with a public frequency list built from Korean subtitles — 5.18 million running words, 50,000 distinct forms. Korean conjugates heavily, so every form was reduced to its dictionary word first: 먹어요, 먹었다 and 먹는 all count as one word, 먹다. Then each word gets its share of all the content words in that corpus.

Why some very common words are missing

The 250 most frequent words in Korean are grammatical glue — 하다, 있다, 것, 되다. They are everywhere and they are nearly impossible to draw. We set them aside on purpose and say so, rather than quietly counting them to inflate the number. They come back later, as phrases, where they actually make sense.

The arithmetic

Every word carries its own percentage. 나 by itself is 3.44% of all content words spoken. Add them up in order and the curve looks like this:

first 50 words	22.0%
first 100 words	30.0%
first 300 words — this deck	36.9%
first 500 words	40.3%
all 1,500 picturable words	47.0%

Notice how steep the start is: the first fifty words do more work than the last thousand. That is the entire reason this deck is ordered the way it is, rather than alphabetically or by textbook chapter.

That 47% at the bottom is the ceiling — the most any picture-based deck can reach, because the rest of spoken Korean is grammar and glue that no illustration can teach. The second number on the progress page is simply your position against that ceiling: how far you are through the part pictures can teach.

What we are not claiming

This is our arithmetic, not a published statistic. There is no official token-coverage curve for spoken Korean. Any figure you find online claiming one is unsourced — including, we suspect, some that look authoritative.

Subtitles are translated speech, not overheard conversation. They lean toward drama vocabulary and they over-use pronouns, because English needs pronouns and Korean drops them. To correct for this the ranking is blended with the Korean government's official learner list (국립국어원, 5,965 words); 1,296 of our 1,500 appear on it. That fixes most of the skew. Not all of it.

Coverage is not comprehension. Owning 47% of content words will not let you follow 47% of a conversation. It means the words you meet most often are the ones you already know — which is what makes the next sentence learnable instead of overwhelming.

Check us

Every word in the deck carries its own percentage and its rank, and they sum to the numbers above. If a number here ever looks wrong to you, we would rather hear it than have you quietly assume we rounded in our own favour.