Guides 06

How word counts work in different languages

English and Korean are counted by spaces; Japanese and Chinese have no spaces. Why word counts differ from tool to tool.

Who decides where a word ends?

In English word processors, the word count is generally the number of chunks separated by spaces. ‘It's a well-known fact.’ has 4 space-separated chunks, so it is 4 words. The hyphenated ‘well-known’ counts as one. This method is simple and predictable, which is why it is widely used for English assignment and paper length requirements.

Korean: eojeol units

Korean also uses spaces, so counting the same way gives the number of eojeol (a word together with its attached particles and endings). ‘나는 학교에 갑니다’ (I go to school) is 3 eojeol. Linguistically it can be broken down further into morphemes, like ‘나/는/학교/에/가/ㅂ니다’, but everyday word counting treats each eojeol as one word. This tool counts by eojeol too.

As a result, a Korean text tends to have fewer words than an English text with the same content, because particles and endings attach to the preceding word to form a single eojeol. This is also why the length of Korean writing is usually given in characters rather than words.

Japanese and Chinese: no spaces

Japanese and Chinese are written without spaces within a sentence. Counting by spaces would turn an entire sentence into 1 word. So this tool splits chunks containing Chinese characters and kana into words using the internationalization segmenter built into the browser (Intl.Segmenter). This segmenter estimates word boundaries from a dictionary, as in ‘東京/に/行き/ます’ (going to Tokyo).

However, what counts as one word differs between dictionaries and segmenters. The same sentence can give different word counts in different tools, so the length of Japanese and Chinese text is usually set in characters. Japanese manuscript paper and Chinese manuscript length requirements are also based on characters.

Numbers and symbols

This tool counts a chunk as a word only if it contains at least one letter or digit. ‘2026년’ (the year 2026) is 1 word, while a standalone ‘—’ or ‘·’ is not a word. For word frequency, punctuation at the start and end is stripped and the word is lowercased before tallying, so ‘Text,’ and ‘text’ are counted as the same word.

Why word counts differ between tools

If a submission has a word-count limit, the surest approach is to do a final check with the program the recipient specifies (for example, a particular word processor).

Last updated: 2026-09-23

Open the character counter →