How to Count Words and Characters (and Stay Under the Limit)

Counting words sounds trivial until one program says 18 and another says 15 for the same paragraph. This guide covers what counts as a word and a character, the usual limits (posts, SMS, search titles), how to estimate reading and speaking time, and how to trim a text that runs over.

Updated:

What counts as a word

For a counter, a word is a run of letters or digits separated from the next by spaces or punctuation. The doubtful cases are what make two programs disagree: hyphenated words, numbers with a decimal point or comma, times of day, email addresses and links.

The ABC SyG Word counter uses your browser's word-segmentation rules. In our tests, "well-known" counts as two words, "3.50" as one (the whole number), "9:30" as two (9 and 30), and English contractions such as "don't" as one. Email addresses and links can be split into more than one word. Word processors usually count anything without a space as a single word, so a hyphenated word counts as one there.

If you have to meet an exact minimum or maximum (a school paper, an article), check how the program used by the person reviewing it counts, and leave a margin of 3 or 4 percent for differences in criteria.

Characters with spaces, without spaces, and line breaks

Characters "with spaces" count everything: letters, digits, punctuation, spaces and line breaks. Characters "without spaces" leave out spaces, tabs and line breaks. Which one you need depends on who sets the limit: social networks and SMS count spaces; some contests and translation jobs are measured without them.

Two details catch people out. First, an accidental double space counts as two characters. Second, a line break counts as one character when counting with spaces. And an emoji counts as one visible character in the tool, although many platforms count it as more, as shown below.

The sample paragraph, counted the way the tool counts it
TextWordsWith spacesNo spaces
I tried the new café downtown: excellent! It costs $3.50 and opens at 9:30. Don't miss it.189074

A worked example, counted by hand

Take the text from the table above. For the words, read and number them: I (1), tried (2), the (3), new (4), café (5), downtown (6), excellent (7), It (8), costs (9), 3.50 (10), and (11), opens (12), at (13), 9 (14), 30 (15), Don't (16), miss (17), it (18). That gives 18 because the colon splits "9:30" into two; if you counted it as a single word the total would be 17. The dollar sign is not a word.

For characters without spaces, add up each chunk between spaces: I (1), tried (5), the (3), new (3), café (4), "downtown:" (9), "excellent!" (10), It (2), costs (5), "$3.50" (5), and (3), opens (5), at (2), "9:30." (5), "Don't" (5), miss (4), "it." (3). The total is 74. With 17 chunks there are 16 spaces between them: 74 + 16 = 90 characters with spaces. Paste the text into the Word counter and you will see exactly those numbers under Words, Characters and Characters (no spaces).

Common limits (approximate)

Limits change over time and every platform has special rules, so treat these as a reference and check the current one before you publish.

WhereApproximate limitWhat to keep in mind
Post on X (Twitter)280 charactersLinks count at a fixed length (about 23 characters) and some emojis or scripts count as more than one
SMS160 characters (GSM alphabet) or 70 (with special characters or emoji)If it doesn't fit, it is split into shorter parts (roughly 153 and 67 characters per part)
Instagram caption2,200 charactersOnly the first few lines show before "more"; put the important part first
Page title for search enginesAround 60 charactersGoogle measures pixels, not characters: a "W" takes more room than an "i"
Meta descriptionAround 155 charactersAlso measured in pixels, and Google may rewrite it

About SMS: the GSM alphabet covers basic Latin letters and digits but not everything. Many accented letters, non-Latin scripts and emoji are outside it. A single such character makes the whole message switch to the wider encoding, and the limit drops from 160 to 70. That is why a short text with one emoji can go out as two messages.

Reading time and speaking time

Adults reading silently average roughly 200 to 250 words per minute, and speaking aloud with pauses and emphasis runs about 130 to 160. These are averages: technical writing reads slower and a children's story faster.

The Word counter shows Reading time (calculated at about 220 words per minute) and Speaking time (about 130). For 1,000 words, reading comes out at 4 min 33 s (1,000 ÷ 220 = 4.55 minutes) and speaking at 7 min 42 s (1,000 ÷ 130 = 7.69 minutes). If you must present in 5 minutes, your script should not pass about 650 words (5 × 130), before counting pauses, slides or questions.

If all you want is that estimate, the Reading Time Calculator shows both times with the same method.

How to trim a text that runs over

  1. Paste the text into the tool's box and note how many characters or words you need to remove.
  2. Start with what nobody will miss: double spaces, repeated greetings and clarifications, filler adverbs such as "really" or "basically".
  3. Look at Most repeated words: the list appears when a word shows up at least twice and ignores very short and common filler words. If a term appears ten times, some sentences can probably be merged.
  4. Swap long constructions for short ones ("in spite of the fact that" becomes "although", "in order to" becomes "to").
  5. Watch the number drop as you delete. When you reach the limit, leave a character or two of margin in case the platform counts differently.

For a post, for example, count with the tool and then paste into the platform to confirm: each network has its own rules for links, emojis and mentions.

Counting the words in a PDF

If the text lives inside a PDF, copying it page by page is slow and usually brings odd line breaks. The PDF Word Counter reads the text of each page and gives you words, characters with and without spaces, sentences, paragraphs, reading time and words per page.

There are two practical uses. You can type a range in Pages to count (for example, 2-10) to measure only the body of a paper, without the cover or the bibliography. And you can compare the total with an assignment limit. Keep three limits in mind: it only counts real text, so a scanned PDF (images of pages) gives zero and needs text recognition (OCR) first; headers, footers and page numbers count as words; and columns, tables or words hyphenated at the end of a line can cause a small difference from Word.

Privacy and limits of the tool

The count happens in your browser: what you type or paste is not uploaded or stored. That helps when the text is a private draft, a letter or a work note.

Limits to keep in mind: in languages written without spaces, such as Japanese, the count depends on the browser's segmentation and may differ from other programs; reading time is an estimate based on an average pace; and each platform counts in its own way, so the tool's number is a reference, not a guarantee.

Frequently asked questions

Does a hyphenated word count as one word or two?

It depends on the program. The ABC SyG Word counter follows your browser's rules and usually counts "well-known" as two words; word processors count it as one. If you have a strict limit, use the same program the reviewer will use.

Do spaces count as characters?

Yes, in the "with spaces" count. The tool also shows Characters (no spaces) for when the limit is measured that way. Line breaks count as one character in the count with spaces.

How many characters fit in an SMS?

About 160 if it only uses the GSM alphabet, and 70 if it includes special characters, many accented letters, or emoji. If it goes over, it is split into shorter parts, which many plans bill separately.

How long does it take to read 1,000 words?

Roughly 4 to 5 minutes silently (the tool calculates 4 min 33 s at 220 words per minute) and about 7 to 8 minutes aloud (7 min 42 s at 130). It is an average, so your own pace may differ.

How many words does my PDF have if it's a scan?

The tool will report that there is no text to count, because a scan is an image of the page. You first need to run the PDF through text recognition (OCR) and then count.