Halliday (1985) emphasizes the non-structural resources of cohesion in the discourse that may involve elements of any extent, and that may hold across gaps of any extent, both within the clause and beyond it, without regard to the nature of whatever intervenes.

Following the guidelines for using Halliday and Hasan’s (1976) taxonomy, a cohesive tie is established where one element of a tie presupposes the other, and cohesion is provided by the resolution of this presupposition, that is, the element is effectively decoded by referring to another. The definitions of cohesion and reference in many cases overlap. Halliday and Hasan (1976) claim that a distinguishing feature of reference as compared to other forms of cohesion is that it is “overwhelmingly nominal in character” (43), except in the case of some demonstrative and comparative adverbs. Cohesion is generally believed to lie in the continuity of reference to something else in the discourse, where reference involves items that indicate that there is some information to be retrieved from some other part of the text. It is not seen as important to know the part of speech used to encode the relationship; its characteristics can be captured in that it has the semantic property of definiteness, or specificity. Reference can be regarded as phoric in the sense that there is an item that refers and there is an antecedent (anaphoric) and/or a postcedent (cataphoric) (Halliday & Hasan, 1976; Laczkó, 2002; Tolcsvai, 2001). However, Tolcsvai’s (2001) approach differs in that items taking part in phoric referential relationships may be various in their explicitness, they can be full NPs, pronouns, morphemes, as well as zero. Zero reference is covered in Halliday and Hasan’s (1976) taxonomy slightly differently under ellipsis, full NPs are only considered as pointing to other text segments for the other types of cohesion (especially lexical cohesion), while morphemes are disregarded under the category of reference.