Reading Time: Different Languages Are Counted Differently
Most reading-time tools count words by splitting on spaces. Japanese and Chinese do not put spaces between words, so that method scores an entire paragraph as one word. Korean does use spaces, but the unit it produces is not the unit English word-per-minute figures were measured in.
The "5 min read" label on an article, the estimated length of a talk, the runtime of a piece of narration — all come from the same calculation. Measure how much text there is and divide by how fast it is read.
It sounds simple, and it goes wrong at the first step: deciding what counts as one unit. Splitting on spaces works for English, and applying that same rule to other languages produces a number that means nothing.
Languages without spaces
Japanese and Chinese do not separate words with spaces. Splitting on whitespace therefore counts a whole paragraph as a single word — an 89-character Japanese passage comes out as about a third of a second's reading.
Korean does use spaces, but the unit between them is the eojeol, a word plus its attached particles. A single Korean chunk can correspond to several English words, so applying an English words-per-minute figure makes Korean text look shorter than it is.
This tool therefore counts CJK characters separately — Hangul syllables, Han characters and kana — charges them at a rate suited to reading by character, and adds the result to the time for any Latin-script words present. A document mixing Korean and English is timed correctly for both halves rather than by whichever script the counter happened to see.
Counting sentences has the same trap
Treating a full stop as the only sentence ending means a Japanese text registers as one sentence no matter how long it runs, because Japanese ends sentences with the ideographic full stop 。 rather than a period.
This tool recognises the ideographic full stop and the fullwidth question and exclamation marks as well. Korean mostly uses the ordinary period so it was less affected, but the difference is large for Japanese material.
The speed settings
Reading speed varies more between people than between scripts. These are starting points, not promises.
| Setting | Latin script | CJK script | Suits |
|---|---|---|---|
| Slow | 150 words/min | ~375 chars/min | Unfamiliar subjects, study material |
| Average | 200 words/min | ~500 chars/min | General prose |
| Fast | 300 words/min | ~750 chars/min | Familiar subjects, skimming |
| Speaking | 130 words/min | ~325 chars/min | Talks and narration |
Speaking time is the more useful number
Reading time is broadly indicative. It varies enormously between people, and the same person skims one article and studies the next. The label at the top of a blog post is a sense of scale rather than a prediction.
Speaking time is far more practical. The pace of reading a script aloud varies much less between people, and a talk slot or a video length is a real constraint you have to hit. 130 words a minute is a clear, deliberate delivery — most people speed up under pressure.
If you are writing for a three-minute video, aim for a calculated speaking time slightly under three minutes. Real recordings add breaths and pauses that the arithmetic does not.
Good uses for this
- •Checking whether a talk fits the slot you have been given.
- •Sizing a narration script to a target video length.
- •Producing the reading-time label for an article.
- •Checking essay or report length — both with and without spaces are shown.
- •Comparing the length of a text before and after translation.
Extended FAQ
My Korean text shows zero words.
That is expected. Hangul is counted by character and reported under the CJK figure; the word count covers only Latin-script text, so pure Korean shows none. The reading time is calculated correctly from the character count.
Why does this disagree with other sites?
Because most of them split on whitespace and nothing else. For Korean or Japanese text they will report a far shorter time than the real one. For English the difference is rounding.
Where does the characters-per-minute figure come from?
It is set against the commonly quoted figure of roughly 500 characters a minute for adults reading Korean or Japanese prose. It is an approximation, and the spread between individual readers is wider than the difference between scripts.
Are emoji and symbols counted?
They are included in the character total but not counted as words or CJK characters, and they do not contribute to the reading time.
Is my text stored?
No. Everything is calculated in your browser and nothing is transmitted.