We study rank-frequency relations for phonemes in several English
texts written by different authors. It is shown that they can be described via assuming that phonemes were generated via random probabilities
governed by the Dirichlet density, the simplest density for random
probabilities. This description allows us to demonstrate that the
the rank-frequency relations for phonemes of a text do depend on the
author of this text. The result is confirmed via several alternative
measures of distance. This feature of phonemes does contrast the
rank-frequency relations for words, which is both author and text
independent and is governed by the Zipf’s law.

Authors

Armen Allahverdyan

Invited Talk e-session

Photos by : Macroscopic Solutions