Here, the approach of Halliday and Hasan’s (1976) taxonomy is adopted; that is, a cohesive tie is where one linguistic element presupposes the other for its interpretation, and cohesion is provided by the resolution, or decoding of this presupposition. The surface components that contribute to cohesion may be language specific (Blum-Kulka, 1986); in English, cohesion is classified into five types: reference, substitution, ellipsis, conjunction and lexical cohesion. These relations contribute to the total unity of the text, which is thus characterized by the number and type of ties, and the density and types of chains these ties form in longer texts. What is measured, though, is not strictly speaking cohesion itself, but the nature of the referential cohesive forces that may contribute to text quality (e.g., to coherence or accessibility), assuming that the ties are used appropriately.