Word Counter

Paste your text, press Count, and the word total is highlighted at the top of the result card. Below it sit eleven more rows. Each one answers a slightly different question about the same text, so it helps to know exactly what each one counts.

• Characters is every character in the box, spaces, tabs and line breaks included. This is the figure most character limits on forms and social platforms refer to.
• Characters without spaces removes all whitespace (spaces, tabs, line breaks) and counts what is left. Some publishers and translation agencies quote prices or limits on this figure.
• Unique words counts each distinct word once, ignoring capital letters. "The", "the" and "THE" are one unique word.
• Sentences counts the stretches of text that end in sentence punctuation or a line break.
• Paragraphs are blocks of text separated by at least one blank line. A single line break without an empty line between does not start a new paragraph.
• Lines is the number of line breaks plus one, so a text with no line breaks is one line.
• Average word length is the mean number of characters per word, and average sentence length is the mean number of words per sentence. Long sentences are the first thing to look at when a text feels hard to read.
• Longest word shows the single longest word in the text, which is handy for spotting a pasted URL fragment or a typo that glued two words together.
• Reading time and speaking time turn the word count into minutes and seconds.

A second table lists the most frequent words. Both tables can be copied, saved as CSV, or downloaded as an image. If you edit the text or change the language after counting, the results clear and you press Count again, so what you see always matches what is in the box.

There is no single official definition of a word, which is why two counters often give slightly different totals for the same essay. This one uses a simple rule you can predict: a word is a run of letters or digits. Once a word has started, apostrophes and hyphens inside it keep it whole.

That rule leads to these results:

Punctuation on its own never counts, so a dash between spaces, a bullet or an ellipsis adds nothing. Only the plain hyphen key glues words together; an en dash or em dash typed without spaces (word—word) splits into two words. If a pasted draft is full of long dashes you would rather turn into commas or plain hyphens, an em dash remover can clean them before you count.

Sentences end at ., !, ? or …, at the Chinese and Japanese full stops 。！？, at the Hindi and Bengali marks । and ॥, or at a line break. A closing quote or bracket right after the punctuation stays with its sentence. This also means a heading or list item on its own line counts as a sentence, and an abbreviation like "e.g." ends one early.

Word processors make different choices. Some treat "3.5" as one word, some count a lone dash, some split "well-known". None of these choices is wrong, they are just different. When a teacher or editor sets a limit, a gap of a few words between tools is normal and rarely matters. If it does, check with the tool the limit will be measured in.

Chinese, Japanese and Thai are written without spaces between words, so a counter cannot find word boundaries by looking for gaps. Proper segmentation needs a dictionary of the language, and this tool does not use one. Instead it estimates the count from the number of characters.

• Chinese (simplified and traditional) and Japanese: every 1.6 Han, hiragana or katakana characters count as one word.
• Thai: every 4 Thai characters count as one word.
• Latin letters and digits mixed into the text, such as brand names, model numbers or English terms, are counted as normal words and added on top.

So a Japanese paragraph of 160 kana and kanji plus the word "iPhone" comes out at about 101 words. Treat these figures as estimates. Real word lengths vary with the topic and writing style, and a technical text with many long compounds will be overcounted compared with a dictionary-based tool.

For strict limits in these languages, the Characters row is the number to trust. Most Chinese and Japanese assignments, publishers and platforms set limits in characters anyway, and the character count here is exact.

The most-frequent-words table is less useful for these scripts: without segmentation, each unbroken run of characters between punctuation is treated as one item, so you will mostly see whole phrases rather than single words.

Korean does use spaces, so it is counted exactly. Each space-separated unit counts as one word, and particles attached to a noun do not add extra words. Vietnamese is also counted by spaces, but because each syllable is written separately, a two-syllable word like "Việt Nam" counts as two.

The language picker controls this. Detect automatically looks at the script first and then votes on common words, falling back to the page language when the result is unclear. If you pick Other, no estimate is applied, and each run of Chinese, Japanese or Thai characters counts as a single word.

Reading time is the word count divided by 200 words per minute. Speaking time divides the same count by 130 words per minute. Both are shown in minutes and seconds, rounded to the nearest second.

The 200 figure is deliberately a little conservative. A large review and meta-analysis of reading rate by Marc Brysbaert put typical silent reading of English non-fiction at roughly 238 words per minute for adults, with fiction a bit faster. Rounding down leaves room for readers who skim less, read on a phone, or are reading in a second language. Speaking rate is slower because every word has to be pronounced and listeners need pauses; 130 words per minute is a comfortable pace for a presentation or a voice-over, not a fast newsreader.

Use these as planning numbers, not promises. Dense technical writing, unfamiliar vocabulary and long sentences slow readers down, while light blog posts go faster. When speaking, people who pause for slides, emphasis or audience reactions often finish later than the estimate, so if you have a strict ten-minute slot, aim for a script that comes out around eight and a half minutes here.

The same speeds are used for every language. Words in different languages carry different amounts of meaning, so a minute of German and a minute of Spanish are not strictly comparable, but the tool does not try to adjust for that. For Chinese, Japanese and Thai the times are based on the estimated word count, so they inherit its uncertainty.

The Most frequent words table lists the ten words that appear most often, with how many times each was used and its share of all the words in the text. It is the quickest way to see which words you lean on without noticing.

To keep the list useful, a few things are left out:

• Common function words for the selected language, such as "the", "and", "of" or "with" in English. These top every list and say nothing about your writing.
• Words shorter than three characters.
• Pure numbers, so a list of years or prices does not crowd out real words.

Words are compared without regard to capitals, so "Market" and "market" are counted together. The share is calculated against the full word count, function words included, so a word used 12 times in a 600-word essay shows 2%. Ties are listed alphabetically.

Two common uses:

1. Editing an essay or article. If "important", "really" or the same technical term shows up near the top with a share well above the rest, you have found the words to vary. Repetition is often invisible when you reread your own draft, and very obvious in a table.
2. Checking keyword focus. For a web page or product description, the table shows whether the term you want to be found for actually leads, or whether a side topic took over.

Picking the right language matters here, because it decides which stopword list is applied. With Other, no function words are removed, only short words and numbers. For Turkish, case folding follows Turkish rules, so "I" pairs with "ı" and "İ" with "i" instead of being merged wrongly.

Heavy repetition of certain stock phrases is also one of the signals readers associate with machine-written text. If you want to look at that side of a draft, an AI writing pattern detector that uses the same word counting rules can point out transitions, hedges and filler phrases sentence by sentence.

Everything happens in your browser. When you press Count, the text is processed by code already loaded on the page; it is not sent to a server, logged or stored. That matters when you are checking an unpublished manuscript, a thesis chapter, a client's copy under NDA, or a cover letter you would rather keep to yourself. Once the page has loaded, counting does not depend on a connection at all.

The tool is free, with no account, no sign up and no daily limit on how many texts you count. There is no character cap set by the tool either; a long document is limited only by what your browser comfortably handles in a text box, which in practice covers full essays, articles and book chapters.

When you need to keep a record, both result tables have export options. CSV gives you a file you can open in a spreadsheet, useful for tracking the length of several drafts or comparing chapters side by side. Image saves the table as a picture you can drop into a report or send to a teacher or editor as proof of length. You can also copy any single value from the counts table.

Reset clears the text and the results in one step, which is the quickest way to start on the next document.