Detrended fluctuation analysis in natural languages using non-corpus parametrization

The existence of long-range correlation in English and Korean had been reported by Montemurro et al. and Jaemi Bhan et al. This work extends this line of research to other languages (mainly Turkish), compares the differences between a meaningful text and a randomly created text by means of detrended fluctuation analysis and diffusion entropy analysis using a newly proposed parametrization which does not depend on corpus. The results imply a unique long-range correlation for each language analyzed, although similarities exist for languages of the same family.