Fwd: Explanation for Zipf's law..

["Subbarao Kambhampati" <rao@asu.edu>]

I think I stated the explanation for zipf's law wrong in the class. 

Specifically, Zipf's own explanation for the observed zipfian distribution of word frequency in a language is 
that it is a natural tradeoff between the speaker's interest in using a *SMALL* vocabulary and the listener's interest
in hearing a *LARGE* vocabulary (so different shades of meaning can be easily disambiguated). [I think I inverted it 
in the class..]

There have been alternate explanations of course. People even showed that even if we just randomly type letters with spaces,
the "emergent" language will have a zipfian distribution!

I added a slide on this to the lecture notes.

rao