Counting Words in Japanese and Chinese: Why Characters Rule

No spaces between words

English uses spaces to separate words, and a word count can simply count the groups. Japanese and Chinese are written without spaces between words, so a sentence looks like one unbroken line of characters. A counter that splits on spaces would report a whole paragraph as one word.

Reading such a text, a person recognises the word boundaries from experience. A computer needs rules or a dictionary to do the same, and the result is never as certain as in English.

Why characters are the common measure

Because of this, texts in Japanese and Chinese are normally measured in characters. A school essay may be set at 800 characters, a publisher quotes length in characters, and translation is often priced per source character. The number of characters is exact and the same in every tool.

Japanese counts need a choice as well. They use three scripts: kanji, hiragana and katakana. Writers and counters count each symbol as one character, whatever its script, and punctuation marks such as 。and 、 count as well.

How a browser finds words

Modern browsers include an interface called Intl.Segmenter, which follows the Unicode rules for text segmentation. For Japanese and Chinese it uses a dictionary-like approach to find likely word boundaries. This counter uses it to report a word count for such text.

The Unicode standard itself says that languages without spaces need more than the default rules and that Chinese and Japanese are even more complex than Thai or Khmer. The result is a good estimate, not a definition, and another tool can return a slightly different number.

What the word count means

A word count for Japanese is useful for comparing drafts or for getting a feel for the density of a text. It is not suitable for a strict requirement unless the recipient uses the same method. If you have to meet a limit, ask whether it is in characters.

The same applies to the most-used words list. In Japanese, short grammatical particles such as の and は appear very often and can fill the top of a frequency list, which is why a ranking of such text is mainly informative for longer words.

Reading and speaking time

Reading speed in Japanese is usually quoted in characters per minute rather than words. A figure of around 400 to 600 characters per minute is common for silent reading, and this counter uses 500. For speaking it uses 300 characters per minute, which agrees with a standard speed for reading aloud given in a Japanese educational report.

These are estimates, and real speeds depend on the text, because kanji-dense writing is read more slowly than a text with much hiragana.

Character limits on platforms

Platform rules treat these languages differently. X counts every Japanese, Chinese and Korean character as 2, so a post can hold at most 140 of them. Google Ads counts double-width characters as 2 in headlines and descriptions. A plain counter reports each character as 1, so the platform may count double.

When you write for such a platform, multiply the plain count by 2 to estimate the weighted figure and check the final text in the composer.

Half-width and full-width characters

Japanese text can contain full-width letters and digits, which are wider, next to half-width ones. They look different but each is one character in a plain count. Some systems, including some forms and ad platforms, count a full-width character as two and a half-width one as one.

If a limit seems to be reached too early, check whether full-width characters are used. Converting digits and Latin letters to half-width form is a common fix.

An example

Take the Japanese sentence 今日は天気がいいので、公園を散歩します。 It has 20 characters including the comma and the full stop. A space-based counter would report it as one word. The segmentation in a browser typically splits it into about eleven word-like units, but another method might split it differently.

Both statements are true at once: the character count is exact, and the word count is a reasonable estimate. That is why the character count is the safer figure for a requirement.

Mixed-language text

Many Japanese texts contain Latin words, numbers and symbols, such as brand names or units. A counter treats these in the normal way: a group of letters or digits between spaces is one word. The surrounding Japanese is then segmented by the browser.

This means the word total of a mixed text combines an exact count for the Latin part with an estimate for the Japanese part. For reading time, the counter uses words for the Latin part and characters for the Japanese part and adds them.

A practical routine

• Use the character count as the main measure.
• Use the word count as an estimate only.
• Check platform rules for weighted characters.
• Keep a margin when the text mixes full-width and half-width characters.
• Confirm the final length in the target system.

The counter shows both the characters and the estimated words for Japanese and Chinese text, so you can use whichever measure the task requires.