Reading Japanese Without Spaces: How to Find Where a Word Ends

Japanese is written without spaces between words. The boundaries that spaces show you in English, you have to find yourself in Japanese. Japanese children can’t do this from the start either. The Ministry of Education’s commentary on the elementary school curriculum for Japanese (2017) lists, as support for pupils who find it hard to follow a text with their eyes, preparing texts “written with spaces so that the groupings and breaks between words are clear.”

If you can’t tell where a word ends, you can’t look it up. This post gives four clues for finding word boundaries, then looks at the places where those clues stop working.

Clue 1: where kanji turns into kana

The most reliable clue is a change of script. A run of kanji is usually the core of a word. The hiragana that follows is either okurigana, the kana ending of the word itself, or a particle or auxiliary.

In 先生は本を読んだ。 (“The teacher read a book.”), your eye catches the kanji 先生, 本 and 読. The は after 先生 and the を after 本 are particles.

But the switch from kanji to kana is not always a word boundary. In 読んだ, the んだ is part of the conjugation of 読む (to read) and cannot be split from 読. I think it is best to treat the kanji-to-kana switch not as a line to cut on, but as a sign that the word ends somewhere nearby. Look at the kana after the kanji one by one, and the word ends where a particle appears.

Clue 2: particles

The common particles are not many. If you know は, が, を, に, で, と, も, へ, の, から, まで and より, sentences become much easier to divide.

Most particles follow a noun. In 駅の前で友だちと会った (“I met a friend in front of the station”), the words 駅 (station), 前 (front) and 友だち (friend) come right before の, で and と.

The trouble is that the same kana also appear inside words. The は in はなし (story) and the と in ところ (place) are not particles. A well-known example is ここではきものをぬいでください. Read as ここで、はきものを, it asks you to take off your 履物 (footwear). Read as ここでは、きものを, it asks you to take off your 着物 (kimono). In text written mostly in kana, sometimes only the meaning of the sentence can decide.

Clue 3: runs of katakana

A run of katakana is usually one loanword. In コーヒー (coffee) or テレビ (TV), the place where katakana starts and stops is the word boundary.

When two loanwords sit side by side, as in ジュースサーバー (juice dispenser), the boundary is inside the katakana. Names are often divided with a middle dot, as in ジョン・スミス (John Smith).

Clue 4: punctuation and quotation marks

The full stop 。 and the comma 、 never fall inside a word. Text inside 「」 is speech or a quotation, so words break at the brackets too. It sounds obvious, but when a long sentence confuses you, look only at the part up to the first comma. It narrows what you have to think about.

Bunsetsu: how Japanese schools divide sentences

Japanese schools teach children to divide sentences into 文節 (bunsetsu), the shortest chunks that still carry meaning. The test children learn is to insert ね: wherever ね fits naturally, there is a boundary.

You can say 先生ね、本をね、読んだ, so 先生は本を読んだ has three bunsetsu: 先生は / 本を / 読んだ. A bunsetsu is usually a core word, often in kanji, followed by kana. That is the same shape as the idea from clues 1 and 2, that a word ends in the kana after the kanji.

Bunsetsu boundaries and word boundaries are not the same thing. Still, if you look words up from the start of a bunsetsu, you will usually find the word you want.

Where the clues fail

There are places where these clues help less.

Kanji compounds. In long runs of kanji like 日本語教育 (Japanese language education) or 大学病院 (university hospital), the script tells you nothing about where to cut. You divide from the front using words you know. Some compounds are listed in the dictionary as one word.

Text in mostly kana. Folk tales and children’s books use few kanji, with long stretches of hiragana. むかしむかしあるところに (“once upon a time, in a certain place”) divides into むかしむかし / ある / ところに if you know it, but clue 1 is no help. Easy text can be hard to divide.

Long conjugations. 食べさせられなかった (“was not made to eat”) is a single bunsetsu, but it stacks causative, passive, negative and past onto 食べる (to eat). You can see where the word ends and still struggle to get back to the dictionary form.

Supporting words. The いる in 読んでいる (is reading) and the こと in 行くことができる (can go) are words of their own, but their meaning is easier to take together with the word before them.

Texts in old kana spelling also write some words with different kana from today, such as いふ for いう. I wrote about that in Reading Old Kana Spelling.

Choosing what to look up

All of these clues are for deciding, before you open the dictionary, where a word begins. For learners, I think the start of a word matters more than its end. If the start is right and you select a little too much, the dictionary can usually still sort it out.

If the start is wrong, you land on a different word altogether. In ここではきものを above, starting from は gives you 履物 (footwear) and starting from き gives you 着物 (kimono). When a meaning doesn’t fit the sentence, suspect the starting point first.

Looking words up in Niho

I make Niho, a free reader for iPhone and iPad that lets you read the works of Aozora Bunko with a built-in dictionary.

In Niho, you long-press the text to select it and tap Look up. Niho tries the strings that begin at the start of your selection, shows the longest one that is in the dictionary first, and lists the shorter candidates after it. Select 先生一人 and 先生 (teacher) comes first, followed by 先. So the trick is to start your selection at the beginning of the word and let it run a little past the end. If the selection is shorter than the word, only shorter words can be found.

If you select ここではきものを starting from は, the longer word comes first, so the first answer is 履物. If that doesn’t fit the sentence, move the start of the selection one character later.

Conjugated words are traced back to their dictionary form. Look up 食べました and Niho shows 食べる, along with the steps it took to get there. A suru verb such as 勉強した goes back to the noun 勉強 (study). Conjugations in old kana spelling are not supported. For names and rare words that aren’t in the dictionary, the Jisho item in the selection menu lets you check a web dictionary.

How not to worry too much about the exact selection is in Practical Tips for Reading Japanese with Niho, and Getting Started with Niho walks through your first book.