Long-Range Statistical Correlations in Human Language: A Case Study in Persian Language

Complex structure of human language enables us to exchange very complicated information. This communication system obeys some common nonlinear statistical regularities. We investigate four important statistical features of Persian language. We perform our calculations for adopted works of six famous Persian litterateurs. Zipf’s law and Heaps’ law, which imply well-known power-law behaviors, are established in this language, showing a qualitative inverse relation with each other. Furthermore, the informational content associated with the words ordering, is measured by using an entropic metric. This metric can be applied in words relevancy ranking process. We also calculate fractal dimension of words in the text by using box counting method. The fractal dimension of each word, that is a positive value less than or equal to one, exhibits its spatial distribution in the text. Generally, we can claim that the Persian language follows the mentioned statistical laws, like the other languages studied in previous works.