In various research fields of the humanities and the social sciences, cooccurrence relationships between two or more linguistic items have been used as a measure of the interrelatedness (distance, intimacy, etc.) of the linguistic items, and many measures, indexes, coefficients of cooccurrence relationship have been proposed and applied to natural language data. In the past researches, however, methodological examination of the proposed measure of cooccurrence has been neglected. For example, such aspects of the proposed measure as their applicability to different data and the limitation of their applicability have been given little attention. In order to create a measure of coocurrence between two lingustic items which is universally applicable, the present paper first discusses various methodological aspects of possible easures of cooccurrence, and then proposes a new measure of cooccurrence of two liguistic items, called coefficient of cooccurrence (C).