Abstract
Zipf's law and issues in natural languages, programming languages and command languages related to it, are reviewed. We point out the need for a unified theory to resolve the issues. Based on Herbert Simon's creative process for scientific discovery, we evaluate four leading theories of language generation. As a result, the Simon-Yule model of text generation is selected as a promising theory. The implications of the theory related to the issues in natural languages, programming languages and command languages are discussed.

This publication has 0 references indexed in Scilit: