Jason (jcreed) wrote,
Jason
jcreed

If you happen to be at CMU or another institution that subscribes to Nature, you can (and I recommend you do) read the full version of "Quantifying the evolutionary dynamics of language", a story that made its way around the blogoworld a little while ago. It's a nice little statistical result, corroborating long-held evolutionary suspicions that irregular verbs probably regularize more slowly the more commonly they are used. Their result is: in fact the regularization rate (in English, sampled at about 800 AD, 1200 AD, and 2000 AD) is very close to the inverse square-root of the verb's frequency. I'd be really interested in seeing a communicative model that yielded this rate, and somehow wouldn't be too surprised if someone could cook up a nice one — the square root has somehow the right sound to it, given the Θ(n2)-ishness of complete graphs and network effects and stuff like that.
Tags: language
Subscribe
  • Post a new comment

    Error

    Anonymous comments are disabled in this journal

    default userpic

    Your reply will be screened

    Your IP address will be recorded 

  • 0 comments