January 06, 2007

Conlang sizes

While looking for the lexicon size of Lojban I found Category:Conlangs by Lexicon Size - Langmaker and decided to plot lexicon size of a number of constructed languages versus their rank (i.e. by size order).

conlangs.png

Surprise! A power law seems to be involved, at least for languages bigger than 100 words. I get an exponent of a=-1.2420, which in turn implies that the frequency of conlangs of size S scales as S-1.8052 (see here for explanation). Maybe I should rush off and send a paper to Nature? :-)

Addendum: Arnt Richard Johansen asked whether there was a correlation between conlang popularity and vocabulary size. Plotting the popularity ranking versus the size we get a nice cloud of uncorrelated points:

popvoc.png

(same thing when using the direct popularity scores) Looking at the popular languages suggests that many of them are popular for being referred to in films, novels or games rather than actual use.

On the other hand, popularity score vs rank looks like a power-law with exponent is -0.4105.

rankscorepopular.png

I haven't found any good (i.e. big) dataset on number of speakers, but given that few conlangs have huge communities I would expect a lot of statistical noise due to the many one-person languages.

Posted by Anders3 at January 6, 2007 02:40 AM
Comments