Vreleksá

eldin raigmore · Posted: Sat Jul 11, 2009 9:24 pm Post subject: A Resource

Where do we put a "Resources" thread?

On Conlang-L someone told me about a resource I'm having some fun with.

http://elexicon.wustl.edu/
http://elexicon.wustl.edu/WordStart.asp
http://elexicon.wustl.edu/query13/query13.asp

It's got 80000 English words and 80000 "English" non-words.
They tracked how fast people recognized whether or not it was a word, and how fast they could pronounce it, and how accurate they were in both recognition and pronunciation.
They also track some phonological and morphological facts about the word; how many phonemes, how many syllables, how many morphemes, etc.
And they track orthographical features about it, too.
They track the words' frequencies according to two different major studies.
They track something they call "bigrams". As near as I can tell, they look at each consecutive pair of letters in the word, and tell how frequent that pair of letters is; they're trying to determine whether familiarity with all the bigrams makes it easier or harder to recognize or pronounce the word.
They have things they call "phonological neighborhoods" -- other words that sound a lot like this word -- and "orthographical neighborhood" -- other words that are spelled a lot like this word. (In most cases the neighborhoods are up to 20 words big.) They tell how many "neighbors" of the word are more frequent than it is, and how many are less frequent. The idea, I think, is to figure out if a word is a lot like a lot of more familiar words, it's likely to be misread; and the existence of a lot of similar less-familiar words might also have such an effect, though a noticeably less radical effect.

-------------------------------------------

Anyway; these might give you ideas about how your own conscripts should work, at least if your conscripts are somewhat sound-based.

Also, if you just want to know the most frequent English words, you can find them here, up to the first 80,000.

Also, if you have a bunch of "non-words" -- that is, words in a conlang instead of English -- you can input a list of them and it will deduce phonological and orthographical rules based on that list, and tell you how hard or easy it predicts recognizing some other word might be.
_________________
"We're the healthiest horse in the glue factory" - Erskine Bowles, Co-Chairman of the deficit reduction commission

Hemicomputer · Joined: 04 Feb 2008 Posts: 610 Location: Calgary, Alberta

Tolkien_Freak · Posted: Sat Jul 11, 2009 10:54 pm Post subject:

I'm not quite sure I get it, but it looks good.