Compteur de Mots

The counter defines a word as any unbroken run of non-whitespace characters separated by one or more whitespace characters — spaces, tabs, or newlines. Under this rule, a hyphenated term like 'state-of-the-art' counts as one word because no whitespace separates its parts. An em dash surrounded by spaces, as in 'he left — and never returned', acts as a separator, so 'left' and 'and' are counted separately. Apostrophes inside contractions like 'don't' and 'they're' do not split the word. Possessives like 'company's' count as one word. A standalone number like 42 or 3.14 counts as one word. A currency symbol attached to a number, like $99, also counts as one word. This behavior matches the counting logic used by Microsoft Word, Google Docs, and most editorial style guides for basic word counts. Where it diverges is in edge cases. Some academic submission systems count hyphenated compounds as two words. Some journalism tools count numbers differently. If your target platform uses a stricter definition, paste a known paragraph of exactly one hundred words from that platform and compare both counts. Use the difference to calibrate. Practical tip: for academic word limits, subtract five percent from the tool count as a conservative buffer, since grading software sometimes counts slightly higher on edge-case punctuation than browser-based counters do.

The default reading time calculation uses 238 words per minute, which is the median silent reading speed found across multiple peer-reviewed studies of adult readers working through modern expository prose. That number is a center-of-distribution figure, not an upper bound. It produces a realistic estimate for a general blog audience. However, reading speed varies significantly by content type and reader background. Dense technical documentation, academic papers, legal contracts, and medical literature typically slow readers to 100 to 150 words per minute because each sentence carries a higher cognitive load. Readers for whom the language is not their first language and children in grades four through eight also read at 120 to 180 words per minute on average. The tool shows three figures to give you a useful range: a fast-reader estimate at 300 words per minute, a median estimate at 238 words per minute, and a slow-reader estimate at 150 words per minute. For voice-over and narration scripts, the relevant figure is 130 to 150 words per minute, which is standard broadcast narration pace. For podcast ad reads, 160 words per minute is common. For audiobooks, the range is 150 to 180 words per minute depending on genre. Choose the figure that best matches your delivery channel. Practical tip: for a conference talk, divide your word count by 130 to get a reliable speaking time in minutes, since people speak more slowly than they read silently.

Yes. The counter has been tested against inputs up to one million characters, which covers roughly 180,000 English words at average word length. That is longer than any standard novel — Tolstoy's War and Peace runs about 580,000 words in translation, which would require multiple pastes, but any single chapter or section will process without issue. All counting logic runs as synchronous JavaScript in the browser's main thread. When you paste a very large block of text, you may notice the page freeze for a fraction of a second while the browser parses the input and hands it to the counter. On a modern desktop browser, inputs up to 100,000 words typically parse in under 200 milliseconds. On older mobile hardware, a 50,000-word paste may take up to one second. The numbers return accurately as soon as the parsing completes. The tab does not crash, and no content is lost. If you need to count several long sections in sequence, use the Clear button between each paste rather than selecting all and replacing. Replacing while old text is still in the area can occasionally produce a momentary character-count discrepancy during the transition. Practical tip: for very large manuscripts — 200 pages or more — paste chapter by chapter and note each word count. This also gives you a per-chapter breakdown, which is useful for balancing a book's pacing before submission to an editor or agent.

No part of your text is ever uploaded, transmitted, or stored. The counter runs entirely in your browser using client-side JavaScript. Every keystroke, every paste, and every character change is processed locally inside the current tab. There is no WebSocket connection that streams text to a backend. There is no debounced POST request that sends batches of content to an analytics server. There is no autosave function that uploads a copy to cloud storage. You can verify this directly by opening your browser's Developer Tools, switching to the Network tab, and typing into the input field. The only network activity you will observe is the initial page load — HTML, CSS, and JavaScript assets. No subsequent requests carry your text. This isolation matters considerably for certain categories of content. Drafts under a publishing embargo cannot legally be shared with third-party services before the release date. Unpublished academic work may violate journal submission policies if transmitted through outside tools. Legal briefs, employment contracts, patient notes, and internal financial documents are often governed by confidentiality agreements that prohibit upload to external platforms. Software source code under a closed-source license has similar restrictions. For all of these, a locally executed counter is the only compliant choice. Practical tip: if you work with regulated documents, confirm your organization's policy on browser-based tools and confirm that no browser extensions you have installed intercept clipboard data before it reaches the input field.