How Word Counters Count Words, and Why Results Differ

The basic rule

Most word counters follow one simple rule: a word is a group of characters separated from the next group by white space. Spaces, tabs and line breaks all count as separators. Count the groups and you have the word count.

That rule is the one used by most word processors, and it is why the sentence “The quick brown fox jumps” has five words. It works well for English and other languages that put spaces between words, and it needs no dictionary.

The method is also why different tools usually agree to within a handful of words on an ordinary text. The disagreements, covered below, come from special cases rather than from the basic rule.

Hyphens and apostrophes

The rule leaves a few cases open, and the usual answer is to follow the spaces. “Well-known” has no space inside, so it is one word. “It’s” and “don’t” are one word each. “Mother-in-law” is one word, and so is a long compound such as “state-of-the-art”.

A dash with spaces around it, such as the one in this sentence — is a different matter. A counter that only splits on spaces may count the dash as a word of its own, while one that requires letters or digits in a word does not. This counter counts only groups that contain at least one letter or digit, so a lone dash or a row of symbols is not a word.

A slash is treated the same way. “and/or” has no spaces, so it is one word. If you write “and / or” with spaces, the slash on its own is not a word because it contains no letter or digit, and the count is two.

Numbers, dates and symbols

Numbers count as words in most tools. “In 2026 we sold 1,500 units” has six words. A date written as 20.10.2026 is one word, because it has no space, and a price such as $9.99 is one word too.

Symbols on their own are a gray area. An ampersand between two words, a bullet at the start of a list item or an emoji standing alone may or may not be counted, depending on the tool. A counter that requires a letter or digit leaves them out, which usually matches how a person would count.

Why two tools differ

When the same text gives 1,203 words in one tool and 1,198 in another, the cause is nearly always one of the points above. One tool counts a lone dash and another does not. One splits “well-known” into two words and another keeps it as one. One counts numbers and another skips them. A small difference of this kind is normal and does not mean that one tool is broken.

If a number matters, such as a hard limit for an assignment, check which tool the recipient uses. When in doubt, aim for a margin of a few per cent below the limit.

For a rule such as “between 500 and 600 words”, a margin on both sides is the safest way to stay inside the range whichever tool is used to check it.

Languages without spaces

Japanese and Chinese do not put spaces between words, so splitting on white space would count a whole sentence as one word. Counting such text needs a method that finds word boundaries. Modern browsers include one: the Intl.Segmenter interface follows the Unicode text-segmentation rules and splits text into words for many languages.

The Unicode standard itself notes that languages without spaces need more than the default rules and that Chinese and Japanese are especially complex. The result is an estimate. A segmentation based on a dictionary may group characters differently from a human editor, which is why counts for these languages can vary a little between tools. Many publishers in these languages set limits in characters for exactly that reason.

Headings, lists and tables

A word counter works on plain text, so structure disappears and only the words remain. Headings, list items and table cells are counted as words like any other text, as long as they contain letters or digits. Page numbers, headers and footers in a document are not part of the pasted text, but footnotes can be, depending on how the text was copied.

If an assignment excludes the title, the bibliography or the footnotes, remove them before counting, or count them separately and subtract. A word processor’s own status bar usually counts the whole document by default.

A practical tip

Paste the final version of your text into the counter, read the word total, and compare it with the requirement. Check the characters, sentences and paragraphs at the same time to spot a text that is too dense or too fragmented.

Keep the counting rule in mind when you edit. Joining two words with a hyphen lowers the count by one, and splitting a compound raises it. Those small changes are fine for style, but they should never be used to meet a limit that a reader will check by hand.