How words are counted, and why word counters disagree

Every word counter has to decide what a word is. Here are the hard cases, how common tools handle them, and the exact rules our counter follows.

Paste the same paragraph into three word counters and you can get three totals. None of them is broken. Each one has made a different decision about what a word is, and most never tell you which.

For a 2,000-word essay the gap is usually a handful of words. It starts to matter when a limit is strict, when a text is full of numbers or links, or when it mixes English with Chinese or Japanese.

The basic idea

Almost every counter starts the same way. It splits the text into pieces and counts the pieces that look like words. The disagreements come from two questions:

What separates words? Spaces, obviously. But what about hyphens, dashes, slashes, and periods?
What counts as a word? Letters, yes. But what about numbers, symbols, and emoji?

Microsoft Word, Google Docs, and most web counters answer these differently. The cases below are where you will see it.

The tricky cases

Hyphenated words

“Well-known”, “mother-in-law”, and “e-mail” each read as one word. Most counters agree, because a hyphen inside a word is not a space. A few simple counters split on every non-letter and get two or three.

Contractions and possessives

“Don’t”, “it’s“, and “Maria’s” are one word each. A counter that splits on apostrophes would count “don” and “t” separately. Good counters keep the apostrophe when it sits between letters.

Numbers

Is “2026” a word? Word processors say yes, and so do we. A year, a page number, or a price is part of the sentence and takes up space on the page.

Decimals and thousands separators are where tools split. “3.5” is one number, but a counter that breaks on periods sees two. “1,000” can fall apart the same way at the comma.

URLs and email addresses

A link such as example.com/guides/word-count has no spaces, so a word processor that splits only on spaces counts it as one word. A counter that splits on punctuation may count four or five. Email addresses behave the same way. If your text is full of links, expect the widest disagreement here.

Dashes

An em dash with spaces around it (“the plan — if it works — ships Friday”) never causes trouble, because the spaces do the splitting. An em dash without spaces (“the plan—if it works—ships Friday”) is the problem. A counter that splits only on spaces sees “plan—if” as one word. A counter that treats the dash as punctuation sees two.

Slashes

“And/or” and “either/or” are the classic case. Split on spaces only and you get one word. Treat the slash as a separator and you get two.

Chinese, Japanese, and Korean text

Chinese and Japanese are written without spaces between words. A counter that splits on spaces would count a whole Chinese sentence as one “word”. Microsoft Word avoids this: its total counts each Chinese or Japanese character as one word (its statistics list “Asian characters” separately and add them to the word total). That is not a linguistic word count, since many Chinese words are two characters long, but it is consistent and widely used.

Korean usually puts spaces between words, so the issue shows up most with Chinese and Japanese.

Emoji

An emoji carries meaning, but nobody writing to a word limit expects 🎉 to use part of it. Some counters count emoji as words; many ignore them.

What Op-Ed Insight counts

Our word counter and the editor use one rule everywhere:

A word is a run of letters or digits. Apostrophes and hyphens inside the run are kept, so “don’t” and “well-known” are one word each.
Numbers stay whole. “3.5”, “1,000” and “10:30” are one word each.
Links, email addresses and initialisms are one word. A web address starting with “http” or “www”, an email address, and “U.S.” or “e.g.” each count once. A bare “example.com/guides” without either prefix is split at its punctuation.
Each CJK character counts as one word, which matches how Microsoft Word’s total behaves.
Emoji are not words. They still count as characters in the character counter.
Everything else separates words. Spaces, dashes, slashes, and other punctuation all end a word.

That last rule means “plan—if” is two words and “and/or” is two words. We think that matches what a reader sees. You may disagree, which is fine, as long as you know the rule.

Here is how a few tricky inputs come out:

Input	Our count	Why
don’t	1	Apostrophe inside a word is kept
well-known	1	Hyphen inside a word is kept
e-mail	1	Hyphen inside a word is kept
3.5	1	Decimal number stays whole
1,000	1	Thousands separator stays inside the number
and/or	2	Slash separates words
plan—if	2	Em dash separates words
plan — if	2	Spaces separate words
東京	2	Each CJK character is one word
Great job 🎉	2	Emoji are not words

When the count really matters

If someone gives you a strict limit, such as a scholarship essay, a grant form, or a journal abstract, the count that matters is theirs. A few habits help:

Find out which tool they use. Many application portals count in their own text box. Paste your draft there before the deadline, not after.
Leave a margin. If the limit is 500 words, aim for 490. A few words of difference between tools will not sink you.
Watch the risky inputs. Unspaced dashes, slashes, URLs, and numbers with separators are where counts drift. If you are right at the limit, rewrite one of them rather than argue about it.

Tip: If you are a few words over, cut a filler phrase before you cut an idea. “In order to” becomes “to”, and “at this point in time” becomes “now”. Our wordy-phrases list has many more.

Word counts are a rough measure of length, not a measure of quality. Use them to stay inside a limit, then spend your attention on whether each sentence earns its place.