Abstract
This paper introduces a new framework for processing Natural Language statements. The parallel is drawn between the Natural Language processing and the Data Mining technology of information granulation. The formalism affords consistent representation of a well-known phenomenon of 'approximate' grammatical correctness of Natural Language statements. The ontology-based information structuring is a natural complement of the syntactical information granulation. The approach is validated on some simple Natural Language statements and the directions for the future development of the system are outlined. The paper focuses on conceptual framework only and, as such, is intended to stimulate further research into the various implementation considerations that are prerequisite of large-scale applications.