Word Counter
Character & Reading Time

How Words and Characters Are Counted

Counting text sounds trivial until you compare two tools and get different answers. The disagreements come from a handful of edge cases that every counter has to make a decision about.

A word here is a run of characters separated by whitespace. That means a hyphenated compound such as "state-of-the-art" counts as one word, a number like "1,250" counts as one word, and an em dash surrounded by spaces is treated as a separate token by some tools and skipped by others. Contractions are always one word. URLs and email addresses count as one word each, which can make a technical document's count look lower than it reads.

Characters are counted two ways: with spaces and without. Most publishing and social platforms count with spaces, while many academic and translation contexts quote characters without spaces. The difference is roughly a sixth of the total for ordinary English prose, which matters when a limit is tight.

Sentences are detected by terminal punctuation, so abbreviations like "Dr." or "e.g." can inflate the count slightly, and paragraphs are separated by blank lines. These are approximations by nature; no counter can perfectly parse natural language, and small variances between tools are normal rather than a sign that one is broken.

Platform Limits Worth Knowing

Two of these deserve emphasis. Search engine title and description limits are measured in pixels rather than characters, so a line of capitals will truncate earlier than the same character count in lowercase; treat the numbers above as guides and check the rendered result. And a single emoji in an SMS silently switches the whole message to a character set that more than halves the per-segment allowance, which is how a "short" message ends up billed as three.

Reading Time and Speaking Time

Reading time estimates assume a silent reading speed for adults of roughly 200 to 250 words per minute for general prose. Dense technical or academic material runs slower, often 100 to 150, because comprehension rather than decoding is the bottleneck. Reading on a phone tends to be slower than on paper, and scanning is much faster than reading, which is why published "5 min read" labels are a rough courtesy rather than a measurement.

Speaking time is a different and more reliable number, because the physical act of articulation sets a floor. Conversational speech runs about 130 to 150 words per minute. Presentation delivery, where you are pausing deliberately for emphasis and letting points land, is closer to 100 to 130. Audiobook narration is typically 150 to 160. If you are writing a three-minute conference talk, aim for around 400 words rather than 500, because nerves make people speed up and then over-run when they correct.

Academic and Professional Word Counts

Where a word count is enforced, the definition of which words count is set by the institution and varies considerably. Commonly excluded are the title page, abstract, reference list, bibliography and appendices. Commonly included, and often forgotten, are footnotes, in-text citations, table captions and figure legends. Tables themselves are treated inconsistently: some departments count every word in a table, others exclude tables entirely.

Most institutions allow a tolerance of around ten per cent, and many apply a penalty beyond it. Because the rules differ so much, the only reliable approach is to check the specific handbook rather than assume, and to run your count with the same inclusions your marker will use. A counter can tell you how many words are in the text you paste; it cannot know which parts your department excludes.

Keyword Density in Context

Keyword density is simply how often a term appears as a percentage of total words. It was once treated as a target to optimise, and advice circulated about hitting specific percentages. That approach is long obsolete. Modern search systems evaluate meaning and topical coverage rather than repetition counts, and text engineered to hit a density figure reads badly to humans while offering no ranking benefit.

The density readout is still worth glancing at, but for the opposite purpose: as a diagnostic for accidental over-repetition. If one phrase is appearing at three or four per cent, the writing has probably become repetitive and would read better with variation. Used that way it is an editing signal, not a target.

Where This Counter Is Approximate

Whitespace-based word counting does not work for languages written without spaces between words, including Chinese, Japanese and Thai. For those, character counts are the meaningful measure and the word figure should be ignored. Counts here may also differ slightly from Microsoft Word, which applies its own rules to hyphenated terms, numbers and text inside certain fields. Neither is wrong; they are different conventions. If a submission is judged on Word's count, verify the final figure in Word.

Your Text Is Never Uploaded

Drafts are confidential by default. Unpublished articles, client proposals, dissertations under embargo, legal correspondence and internal announcements all routinely get pasted into online counters, and most of them transmit that text to a server to process it. This one does not. Counting happens character by character in your browser as you type, using JavaScript that loaded with the page. Nothing is uploaded, nothing is stored between visits, nothing is logged, and no part of your text is attached to an analytics event. Disconnect from the internet and keep typing: the counter carries on working, which is the simplest proof there is.

Frequently Asked Questions