This article first specifies a definition of “text” that fits its present purpose; then it attempts to contribute, in the light of the demands involved in its treatment and the difficulties involved in learning it, to an explanation of its complexity, particularly its double construction, which is both linear and reticulated. In order to do that, the study will recall the conditions of textual progression along the linear axis, where the reader establishes semantic, syntactic, thematic, logical and argumentative connections between words and propositions that follow each other; and the conditions of its organic composition, which allows the reader to incorporate the words, phrases and paragraphs so constituted into a global structure (sequences, parts of the text, its general structure) as a function of textual models he or she may have experienced. But the paper insists upon the fact that, in order for these two types of organisation – linear sequence and hierarchical inclusion – to constitute a text, it is necessary for them to be inscribed in an enunciative context and a communicative project that give the text its origin, its finality, and its function. In view of the development of the use of New Information and Communication Technologies, it examines to what extent these new supports, in terms of their nature as well as their modes of functionality, require or lead toward new linguistic and cognitive strategies for treatment of text, especially in learners, children and/or non-native speakers, and how these learners can perceive and construct the local and global coherence (semantic, logical, argumentative…) of a document presented in a hypertextual form.