Word Finder

This app lets you play around with word lists in various ways:

• Random mode lets you pick words at random from a word list. It uses uniform-random sampling with replacement. This means that with a short word list, you are more likely to see repeats.

• Jabber mode (for Jabberwocky) lets you see new words spliced together from existing words: for example, like the way we combine spoon and fork to make spork. This uses n-grams with the method detailed in this write-up from 2012. My implementation then was a command-line version written in Python; the implementation here is the same algorithm, but in JavaScript, usable on desktop or mobile.

• Pattern mode is a dictionary lookup. If you type in only letters, such as example, you'll the word at the bottom if it's in the dictionary, or nothing if it isn't. You can also use a . to match a single letter, and/or * to match zero or more letters. For example, t..th will match teeth, tooth, and tenth; th*st will show you all words starting with th and ending with st.

• Anagram mode lets you see all the ways the letters of a given word can be permuted to make another existing word.

• Bee mode is a tool of last resort for the New York Times Spelling Bee (see below for philosophy).

About the word lists

• The Usual Suspects list is a hand-curated collection of words I often miss in the NYT Spelling Bee (see below).
• The Swadesh list is about two hundred core English words, originally developed for linguistic research.
• The 2K list is the General Service List published by Michael West in 1953.
• The 10K through 40K lists were found on a web search.
• The 300K list is one I’ve accumulated over the years, from multiple sources I’ve lost track of.
• The Finnegan list is taken from an online corpus of Finnegans Wake.
• There are some non-English lists I found on the web. There is a skew toward languages I have some personal familiarity with.

On-line help

Please see the hosted app, at either of the following locations, for on-line help:

• https://johnkerl.org/word-finder
• https://johnkerl.github.io/word-finder

JavaScript tooling

This app uses the Sliver JavaScript library.

This is a fun little DIY JavaScript project at the end of 2024 and the start of 2025.

Spelling-Bee Philosophy

This is a word-list app -- a tool of last resort for the NYT Spelling Bee, and other things as well. Here is my own approach:

• First: I start with the day's letters, of course.

• Second: I generally get around half the words before looking at the Grid. I consider this effectively a part of the game itself.

• Also second: The Buddy's Grid and Two-Letter List portions have all the same information as the Grid, while removing the need for scratch paper.

• Third, the Stats: these aren't hints by any means, but, they help me prioritize: if I'm missing, say, a six-letter word starting with a P, then if 80% of readers have found it, I do know the word, and I just need to search. But if only 20% of readers have found it, in my experience, that's a word I do not know (yet).

• My partner and I have developed a list of the "usual suspects" -- words that seem to appear only in the Spelling Bee, including aril, natant, tilth, and tinct. These are encoded in the Word Finder app. The usual-suspects list also includes our oh-I-cannot-believe-I-keep-forgetting that words.

• I try to complete the puzzle using these four. About half the time, I can Queen Bee on this basis: without hints.

• Fifth: the reader hints. On those days I've got a few words left, I end up taking 1, 2, 3, 5 hints; worst case, 10 or so. And looking at a hint almost always results in success.

• Sixth: even with reader hints there is occasionally a stumper. This rarely happens for me -- less than one day in ten -- in large part due to the quality of the reader hints. The fork in the road is to shrug and say I don't know what this word is, and call it a loss for the day -- or, to find a way to learn what may be a new word. Here is where word lists come into play -- and Word Finder's Bee mode is a word-list app designed for that desperate purpose.