Recent investigations have shown that a model can be derived from an association matrix, Fig. 1a, which describes a functional language network at the level of words in text. In this model words are gathered in classes, Fig. 1b, in which the number of words increases by a repeated factor of two. It could be shown that when an artificial text is generated with the aid of the model, this text contains maximum entropy. By comparison the conclusion can be drawn that this behaviour is also valid for natural language text. Therefore it is possible to determine some quantitative linguistic values and diagrams immediately and without any difficulty. Recently, this was already shown in the case of Zipf's famous frequency diagrams, Fig. 2 and Fig. 4.
Volume: 7, 03/2004
Pages: 1-24