Introduction

Background and intention

This paper attempts to establish a new approach with graph networks, which we refer to as the cluster grammar, as a computational linguistic grammar. The main content of this paper is the construction of the theory and the verification of its potential for application to computer science.
We begin by introducing the graph networks. The graphs are considered as the essential elements for language abilities. The innate abilities which we introduce in order to elucidate the acquisition of the language ability, generation and processing of sentences are given as the graph theoretical algorithms. The theory includes a sort of verification problem as well as the Universal Grammar introduced by Chomsky, however the theory is intended to be applied to information technology where the given algorithms and methods can be developed further to have computers process languages with more developed data structure.
This sort of graph networks has a similarity in the semantic network introduced by Richard Richens in 1956. The cluster grammar not only gives such a structure on morphemes, as well as a different structure on signifiés.