Random word generator
An English word from a list, drawn when you tap.
Recent words
Words from this visit show up here.
Each tap is a new pseudorandom draw via crypto.getRandomValues. Not atmospheric noise. How the draw works
Put this tool on your own page
Paste this where you want the tool to appear. It runs the same draw in your visitor’s browser, with no account and nothing sent to a server.
About this page
Tap Generate for an English word, or for several at once. The random word generator draws from a list bundled with this page, rather than from a server, and it does not glue random letters into something that only looks like a word. You can ask for one word or up to twenty. You can keep every part of speech, or limit the draw to nouns, verbs, adjectives, or adverbs. Animals, foods, and colors are smaller lists for when the room wants a category instead of a grammar label. Leave the length boxes blank for any length, or set a shortest and a longest size, both counted in letters. Every word that survives those filters is equally likely. A common word is not favored over a less common one that made the same list. Words in one tap are drawn without replacement, so a count of five is five different words. The page does not add a definition, a plural, or a sentence. It shows the lemma that was drawn, in lowercase, because that is how the list stores it. Nothing is chosen while the page is loading. There is no account. The list is a subset of WordNet 3.0 from Princeton University. The methodology page quotes that database's license and the rule that keeps the draw uniform.
Related
Questions
Is every matching word equally likely?
Yes. Once you pick a list and a length, the generator takes one fair index into the words that remain. The draw uses crypto.getRandomValues in your browser, then rejection sampling so the leftover remainder of a 32-bit value cannot favor the low end. ChanceKit does not claim atmospheric noise, hardware entropy you can audit from this page, or a casino-style provably fair setup. Read the methodology before you use a result for anything that matters. The list itself was narrowed before you arrived: a lemma had to be a single alphabetic word that a tagged corpus counted at least twice, and an editorial exclusion list drops ethnic slurs and strong profanity, plus lemmas whose qualifying senses are sexual assault or explicit sex. That list is not a complete filter. Words about conflict, religion, and the body can still appear, and so can clinical words such as sex and nude. Inside the words you can still draw, frequency in print does not change the odds. "Person" is not due more often than a rarer noun of the same length.
Can I choose a part of speech or a category?
The List menu is the filter. Any word uses the combined lists. Nouns, verbs, adjectives, and adverbs are the parts of speech WordNet records for those lemmas. Animals, foods, and colors are smaller editorial lists for a game or a classroom round: dog is an animal, apple is a food, and red is a color. A word can sit in more than one list. Chicken is both an animal and a food. Choosing Animals does not also apply the noun filter. The category is the whole pool. If you wanted only five-letter foods, set the list to Foods and set both length boxes to 5.
How does the word length filter work?
Length is the number of letters in the lemma, not syllables and not the length of a definition. Blank boxes mean any length. A minimum of 4 and a maximum of 6 keeps words of 4, 5, or 6 letters. Both ends count. If you ask for a length the list does not contain, the page says so and does not invent a word to fill the gap. Twenty words in one tap are refused when fewer than twenty words match, because this page draws without replacement and will not repeat a word to hit the count.
Where does the word list come from?
WordNet 3.0, a lexical database from Princeton University, copyright 2006. ChanceKit ships a subset, not the full database: phrases, punctuation, and words the corpus barely tagged are left out. The animal, food, and color lists are a further edit so the category matches what a person expects. Egg is listed under foods even though its highest-tagged sense is biological. The methodology page states the license, including the as-is disclaimer, and links the Princeton license text. You can read that page before you rely on a word for anything you would have to explain.
Will the same word come up twice?
Not inside one tap. Five words are five different lemmas. The next tap starts over and may repeat a word from the history strip. The strip is a record of this visit, and it does not remove those words from the pool. If you are dealing a unique set across several taps, cross off the ones you have already used. Refreshing the page clears the strip. ChanceKit does not keep an account of the words you generated.
What should this random word generator not be used for?
Do not use it as a password, a seed, or a secret. The output is a dictionary word shown on a screen, which is a weak secret even though the index was drawn with Web Crypto. Do not use it to pick lottery numbers or bets. A class can use it with no signup, on a projector, as long as the teacher is willing to see an ordinary English word that might be uncommon or blunt. The exclusion list is not a children's dictionary and it does not define the word it shows. For a name, use the name picker. For a single letter, use the letter generator. For an integer, use the number page.
Last updated