Forgetting the words you looked up
Most words have to be met many times in reading before they stay. Each time you look up a word you have saved, Fusen shows how many times you have saved it, with the sentences you saved it from.
Counted from the Aozora Bunko text of Kokoro, about 160,000 characters, November 2026. Updated .
Why looked-up words don't stay
You look a word up, understand the sentence, and a few pages later you meet the word again and can't remember what it means. In the first months of reading Japanese books, this is the usual experience.
Waring and Takaki measured it in 2003. Fifteen Japanese learners of English, at intermediate level or above, read the graded reader A Little Princess. Beforehand, the researchers had replaced 25 common words in it (house, face, money and others) with made-up spellings, so that no reader could know them. Each occurred between once and 18 times.
Asked straight after reading to write down the meanings, the learners managed 4.6 of the 25 on average (about 18%); three months later, 0.9. For the words met 15–18 times, the share fell from 42% to 6%. Words met fewer than eight times were almost never given a meaning, whether straight after reading, a week later or three months later. The authors conclude that one book leaves very few new words behind, and that vocabulary grows only from a great deal of reading (the paper).
The learners were reading English, but the lesson carries over to Japanese: forgetting a word met once or twice is normal, and a word takes many meetings in reading to stay.
Look it up again when it comes back
Tap + beside a definition, and Fusen saves the word with the sentence you met it in. The next time you look it up:
The popup marks the word Saved 2×: you have saved it twice before.
Tap the mark to see the sentence from each time you saved it, and the book each one came from.
Tap one to read the whole passage again.
The popup is shown in the screenshots in Learn Japanese by reading it. Looking a word up again doesn't take you off the page, and the popup lists the sentences you met it in before; reading them again makes the meaning easier to recall.
Common words come back within a book
The counts below are from Natsume Sōseki's Kokoro (『こころ』), the full Aozora Bunko text: about 160,000 characters, or roughly 273 pages at 600 characters a page. A "new word" here is a noun, verb, adjective or adverb outside the 5,000 most common words, leaving out names and numbers: a word that a reader who knows those 5,000 would still have to look up.
The first dozen or so pages are the hardest, with over twenty words you haven't seen before on each page; after that the number falls by about half. Further on, most of the book's new words have already appeared, and fewer and fewer appear for the first time.
In the first 27 pages (the first tenth of the book), 494 new words appear for the first time. 65% of them appear again later, and 142 (29%) appear five times or more in the book. 打ち明ける (to confide, to confess), for example, appears 32 times, from the first page to the last:
They take a long time to reappear, though. Of the words first met in those 27 pages that appear again, half take more than 45 pages to do so. By then you have probably forgotten what you looked up, and it is time to look it up again.
A book's words and a frequency list
Vocabulary lists are usually ordered by how often each word occurs in a large body of text. The counts above use wordfreq's Japanese frequencies, drawn from Wikipedia, film and TV subtitles, web pages, Twitter and Reddit. Words that are common there are not necessarily common in novels:
| Word | Rank in the general list | In Kokoro |
|---|---|---|
| 黙る (to fall silent) | 2,226 | 40 times |
| 手紙 (letter) | 2,660 | 74 times |
| 答える (to answer) | 3,235 | 94 times |
| 住民 (residents) | 2,010 | 0 |
| 分野 (field, area) | 2,049 | 0 |
| 獲得 (acquisition) | 2,091 | 0 |
Working through this list as far as word 2,100 would teach you 住民, 分野 and 獲得, but not yet 黙る, 手紙 or 答える. The bottom three never appear in Kokoro; the top three appear 208 times between them.
A longer list doesn't solve this. Kokoro has 318 new words that appear five times or more, and 186 of them (58%) rank outside the list's top 10,000: someone who had learnt 10,000 words from it would still not know more than half of the new words this book keeps using.
Genre changes the words too. Of the 50 most frequent new words in Miyazawa Kenji's Night on the Galactic Railroad (『銀河鉄道の夜』), only four are among Kokoro's 50 (到頭、じっと、下りる、気の毒). Miyazawa's are words of travel and the night sky, such as 汽車, 天の川 and 野原; Sōseki's include 叔父, 書物 and 打ち明ける. Miyazawa's vocabulary is unusual in its own right, so part of the difference is the author's.
The more you read, the closer the words you meet come to the frequencies of the kind of writing you read: read modern novels, and the words that keep coming back are those of modern novels. The key words of popular titles are listed in What to read, worth a glance before you start one.
Rare words and rare meanings can wait
59% of Kokoro's new words appear only once in the book. Look them up, understand the sentence and read on; there is no need to stop and learn them.
Some of them are common elsewhere: of those 1,650 words, 359 rank within the general list's top 10,000. A word that is common enough will turn up again in the next book.
The same goes for meanings. A dictionary often lists several senses for one word: JMdict gives 考える seven. A book usually uses one or two of them, and understanding the one used in the sentence in front of you is enough.
A definition is not the word
A definition approximates a word in another language. Languages divide meaning differently, so several Japanese words often map onto the same English word.
思う and 考える are both glossed "to think" in JMdict. 考える is thinking something through, over a span of time; 思う is closer to a judgement, a feeling or a passing reaction. In these two sentences from Akutagawa Ryūnosuke, swapping one for the other would change what is meant:
- 治修はしばらく黙ったなり、何か考えているらしかった。Akutagawa, 『三右衛門の罪』: Harunaga sat silent for a while, apparently thinking something over.
- 彼は勿論「おや」と思った。Akutagawa, 『お時儀』: "Oh?" he thought, of course.
嬉しい and 楽しい are both "happy" in JMdict. 嬉しい is gladness at something that happens, such as finally meeting someone or hearing good news; 楽しい is a time or an activity being enjoyable, such as a meal together or a trip.
- やっと金さんにも遇える時が来たのだ、嬉しい。Akutagawa, 『奇怪な再会』: At last the time has come to see Kin-san again. I'm so glad.
- 三人は楽しく昼飯の卓についた。Arishima Takeo, 『或る女』: The three sat down cheerfully to lunch.
Differences like these are hard to learn from definitions. Meeting a word in many sentences, and seeing what it goes with and where it is used, brings your sense of it closer to a Japanese reader's. When a use puzzles you, tap Explain with AI: Fusen AI explains what the word means in that sentence and how it feels.
Loanwords are a trap of their own for English speakers: they look like English but often mean something else. マンション is a block of flats, not a mansion, and スマート often means slim. Met in context, the English reading stops making sense.
Words to learn on purpose: Anki
Some words rarely appear in books, but you want to keep them: terms from a field you care about, or phrases you like. Fusen can send a collection's words to AnkiDroid, with the sentence, reading and definition on each card, and Anki schedules the reviews. See Review in Anki.
Questions
Does Fusen have a review feature?
Not spaced review. Fusen tells you when you look up a word you have saved, and shows the sentences you saved it with. For scheduled review, send the words to Anki.
Should I still learn from a word list?
When you start reading, knowing a few hundred to two thousand of the most common words makes the first book manageable. After that, the words your books keep using fit what you read better than a list does; carrying on through a list spends a lot of time on words that rarely appear in novels.
How do I go over the words saved from one book?
Open Words in the notebook and filter by that book's collection to see every word saved from it; tap a word to see the sentence you saved it with.
How were these numbers counted?
Sudachi splits the text, and words are counted in their dictionary form; common-word ranks come from wordfreq's Japanese list. The method is the same as for the title pages: see What to read.
Kokoro and Night on the Galactic Railroad are both about a hundred years old and use spellings rarely seen in modern books, such as 今迄 and 到頭, which pushes up the share of words outside the general list's top 10,000. Counted the same way, Dazai Osamu's later No Longer Human (『人間失格』, 1948) has 54% of its new words met five times or more outside the top 10,000, against 58% for Kokoro.
For reading every day, see Keep going.