How to use the word frequency counter
Paste text, or press Open file (or drop a file on the box) to load a Word .docx, OpenDocument .odt, PDF or plain-text file. Files are read inside your browser — Word files are unzipped locally and PDFs read from their text layer — and never uploaded. The counter builds a ranked table of every distinct word: its rank, how many times it appears, and what percentage of all words it represents. The chart shows the top 20 at a glance; the table holds everything, sortable by count or alphabetically and searchable with the filter box. Switch to 2-word or 3-word phrases to find recurring expressions — “customer service”, “as soon as”, “climate change policy” — which single-word counts hide.
The summary tiles give context: total words, unique words, how many terms survive your filters, how many are used only once, and lexical diversity — unique words as a share of all words. A long technical manual might sit around 20–30%; a short, varied essay can exceed 60%. Very low diversity in a short text usually means repetition worth editing out.
Download a CSV — named after the file you opened — to take the full list into Excel, Google Sheets or a word-cloud generator, or Copy a tab-separated version that pastes straight into a spreadsheet. For a headline word count with readability scores, use the word counter.
What each option does
| Option | Effect |
|---|---|
| Ignore stop words | Hides very common words (the, and, of, is, it’s…) in English, French, German or Spanish. For phrases, drops any phrase that starts or ends with one, so “the river” goes but “bridge of sighs” stays. |
| Also ignore | Your own list — names, brand words, “said” — separated by commas or spaces. Phrases containing any of them are dropped too. |
| Case-sensitive | Off by default, so River and river are counted together. Turn it on to separate proper nouns from ordinary words — Bath the city from bath. |
| Include numbers | Lists figures such as 2026 or 3.5 as terms. Off by default, because numbers rarely tell you what a text is about — but they still count towards the total. |
| Keep hyphenated words together | Treats well-known as one term. Only a hyphen touching letters on both sides joins; a command-line flag such as -v after a space never glues onto the word before it. |
| Ignore accents | Merges café with cafe and naïve with naive. |
| Merge possessives | Counts river’s as river. Contractions that are stop words, such as it’s and that’s, are left alone. |
| Keep links and e-mails whole | Counts https://example.com/page or [email protected] as one term instead of breaking it into https, example and com. |
| Min. length / min. count | Hide terms shorter than a number of characters, or that appear fewer times than you set. |
Single-word percentages are shares of the total words tile — every word in the text, numbers included — and phrase percentages are shares of every phrase of that length. Both are calculated before filters, so hiding stop words or numbers never inflates the share of the words that remain.
How words and phrases are split
The counter uses your browser’s built-in Unicode word segmenter (Intl.Segmenter), the same machinery browsers use for double-click word selection. That means accented Latin, Cyrillic, Greek, Arabic and Hebrew words are split correctly, and Chinese, Japanese and Thai — written without spaces — are divided into dictionary words rather than counted as one enormous “word” per sentence. Contractions stay whole, so don’t is one term, and curly apostrophes pasted from Word are treated the same as straight ones.
Phrases are only counted within a clause: a full stop, comma, semicolon, bracket, quotation mark or line break ends the run. So in “red wine, white wine” the phrase wine white is never counted. This is what makes the phrase list useful for spotting real repeated expressions rather than accidental word pairs.
Everything happens on your device. Very long texts are processed in chunks with pauses for the browser to stay responsive, so a full novel counts in a second or two. The table shows up to 2,000 rows on screen; the CSV contains every term.
What people use word frequency for
- Editing. Find crutch words and repeated phrases — really, actually, in order to — then use the case converter or your editor’s find and replace to tidy them.
- SEO and content. Check that a page actually talks about its topic, compare your wording with a competitor’s, and spot phrases searchers use that your copy lacks.
- Research. Quick corpus checks on interview transcripts, open-ended survey answers, speeches or literature: which themes recur, and how vocabulary differs between two texts.
- Language learning. Pull the most frequent words out of a text you are about to read and learn those first — the top 100 words of most texts cover roughly half of everything written.
- Word clouds. Export the CSV and feed it into any word-cloud or data-visualisation tool that accepts word/count pairs.
If your input is a list rather than prose — one item per line — and you want to count repeated lines, the line counter reports duplicates and can remove them.
Frequently asked questions
What are stop words and should I remove them?
Stop words are the most common function words — the, a, and, of, to, is — which dominate any frequency list without saying anything about the subject. Leave the filter on to see what a text is about. Turn it off when you are studying style, writing habits or language use, where function words are exactly what you want to count.
How is the percentage calculated?
For single words, each term’s count is divided by the total words figure — every word in the text, including numbers, whether or not numbers are listed. For phrases, it is divided by the number of phrases of that length. Filters are applied afterwards, so if “river” appears 12 times in 600 words it shows 2% however many words you hide.
Can I count phrases, not just single words?
Yes. Choose 2-word or 3-word phrases at the top. Phrases never cross punctuation or line breaks, and with stop words ignored, phrases that begin or end with a stop word are dropped, which leaves meaningful expressions such as “flood defences” or “old town council” at the top of the list.
Does it work for languages other than English?
The word splitting works for virtually any language, including Chinese, Japanese and Thai, which have no spaces between words. Built-in stop-word lists cover English, French, German and Spanish; for other languages turn the filter off and add the words you want hidden to Also ignore, or raise the minimum length.
Can I analyse a Word document or PDF?
Yes. Press Open file or drag the file onto the text box. Word (.docx), OpenDocument (.odt), PDF and plain-text files are read in your browser and never uploaded. Word files give the main text and tables; PDFs give their text layer, including running headers and page numbers. The CSV is named after the file.
Will the CSV open correctly in Excel?
Yes. The file is UTF-8 with a byte-order mark so Excel shows accents and non-Latin scripts properly, uses standard quoting for terms containing commas or quotes, and prefixes any term starting with =, +, - or @ so a spreadsheet cannot mistake it for a formula. It lists every term, not just the rows shown on screen.