I am getting read to do the Sex & the City analysis. In the "Linked" book I'm listening to, they mention the idea of a "semantic web" i.e., the relationship between words in a sentence, which words connect to which others. I'm basically going to make the same kind of map (I think) for a couple of scripts, so it's not so revolutionary as far as that goes. However, what I hope will emerge is a kind of memetic clustering above the semantic web. Anyway, I'm going to start with joint shared sentence membership and see what happens.
I was thinking about Japanese and Chinese this morning -- neither language uses spaces -- and was thinking that maybe a semantic/memetic analysis of those two might simply be done at the character level. The word-character boundary is blurred, and in English we sort of take the "word level" as established, otherwise we could take a kind of "character proximity" graph and start one level lower, which would in principle give us something akin to the "word web" of English deriving from nothing more than the character relationships, so the "words" themselves would be sort of hidden in the web of characters, which is what you should see in Chinese and Japanese, without knowing a priori that such-and-so character next to another character actually represents a word, or that such kana character is actually a declension of the Chinese character next to it, or that a given character in Chinese has only grammatical marking potential, e.g., 嗎 ("ma") or 的 ("de") and such.
Anyway, hopefully, or theoretically, all of that stuff would come out in the wash. I may do a character-level analysis of English and Japanese and see what happens, at some point. Maybe on a shared, known text such as the Bible or Alcoholics Anonymous or something.