Zipf’s law—another view

In many scientific disciplines phenomena are observed which are usually described using one of the various versions of the so-called Zipf’s Law. As no single formula could be found so far which would be able to sufficiently fit to all data more and more variants (modifications, ad-hoc formulas, derivations etc.) are created, each of which agrees quite well with some given data sets but fail with others. The present paper proposes a new approach to the problem, based on the assumption that every data set which displays a Zipf-like structure is composed of several system components. A corresponding model is presented and tested on data from 100 texts from 20 languages.