I did something like that for multiword anagrams: scoring them by their probability in a word-bigram model of English. It's expensive because I had it pick the

I did something like that for multiword anagrams: scoring them by their probability in a word-bigram model of English. It's expensive because I had it pick the best permutation of the words, by brute force. It's far from a measure that could really judge "this is a great anagram", but there are so astronomically many boring ones to filter out that it's still a big help.