tlhIngan-Hol Archive: Fri Aug 05 01:23:04 1994
[Date Prev][Date Next][Thread Prev][Thread Next]
Re: Scrabble letter frequencies
- From: Mark Reed <[email protected]>
- Subject: Re: Scrabble letter frequencies
- Date: Fri, 5 Aug 1994 13:15:07 -0400 (EDT)
- In-Reply-To: <[email protected]> from "David E G Sturm" at Aug 5, 94 02:22:09 am
The list I generated did include the affixes, but only once each.
Perhaps a compromise between the two methods would generate the best results?
Instead of counting characters in a text or the lexicon, take several
texts and generate a "word" list from them, where "word" includes affixes
as well as the base word. This list would include each such word from
the texts only once. Then you count the characters in this word
list.
I think this is what you want for Scrabble. It takes into account the fact
that the suffixes may appear in conjunction with several different words, while
avoiding the problem whereby frequently-used words would artificially bump
the frequency counts of their constituent characters. (The fact that
the word 'e' shows up an awful lot doesn't make ['] and [e] any more valuable;
you can still only make 'e' out of them. But the fact that the *suffix* -'e'
shows up a lot does make them more valuable, because you can add -'e' to
a large number of words to make new words.)
(Side note: Would sticking -'e' on the end of a noun be a legal play? I
suppose it's not really any worse than sticking -s on an English word which
forms its plural that way, which is legal . . .)
-Mark