The science behind WordLab

July 23, 2013

Word Laboratory, or WordLab for short, is a fun little application that makes up nonsense words. Interesting words, funny words, bizarre words. It’s entertaining if you enjoy words and language. It’s useful if you need to coin a new term or come up with a new product name. I simply like seeing what comes next, because it’s always a surprise.

So what makes Word Laboratory tick? What is so interesting about this simple little app? The answer: Markov Chains — we love them, and who doesn’t really?

Wait, what’s that? “What is a Markov Chain,” you ask? In the interest of keeping things — well — interesting, let’s make this short: a Markov Chain is a thingy* that can be used to model certain statistical processes in the real world by basically asking, “What are the chances that x follows y?” The collection of answers to all those “What are the chances?” questions make up a Markov Chain: the probabilities that particular chains of events will occur.

So what does this have to do with Word Laboratory, language, and made up words? Everything! If you look at the words of a language as being a series of questions like, “What are the chances that the letter e follows the letter i?” then you should start to see how this works. What is really nifty is that each language has certain statistical properties that can be analyzed and stored in a relatively small file — a “Language Word Brain” if you will. And this isn’t necessarily an exaggeration: Markov Chains can be used to model processes in the brain, and they share some qualities with those big sophisticated things they call “Neural Nets” that help computers think about things.

Word Laboratory generates made up nonsense words that pretty much sound like they could be real words because they are based on the same rules and probabilities that make up actual real words. This explains why some of the words that Word Laboratory generates are real words. And because our Language Word Brain files are small, we’ve included enough of them to generate nonsense words in English, Spanish, French, German, and Italian. And another language too.

So there you have it: Word Laboratory. It’s science.** And it’s fun.



* Not a scientific term.

** In the interest of science, Word Laboratory uses linked Markov Chains and additional (memoryless) heuristics to improve the quality of words that it generates. Fumbing*** good words.

*** Generated by Word Laboratory.

Tagged with:

Comments are closed.