On a global level, an attempt will be made to relate relatively macro-level intuitions about properties of texts to more micro-level notions and to empirically testable lexicogrammatical properties. The strategy will be to (a) partly reduce an intuitive notion of ‘information distribution’ in texts and sentences to more technical and better understood notions of ‘information structure’, ‘informational density’ and ‘grammatical metaphoricity’, and (b) operationalize these latter notions in such a way as to make them empirically testable on electronic corpora, using the ‘shallow’ concepts of ‘explicitness, density, and directness’ as properties of semantics-to-grammar mapping in sentences. It will furthermore be suggested that aspects of intuitive qualities of texts, such as ‘content orientation’ vs. ‘interactant orientation’ and others can be partly modelled in terms of (a) and (b). The argumentation in this paper thus proceeds from intuitive text-level notions to more technical and clause-based concepts, and from these concepts to operationalizations. It then moves ‘upwards’ again to text-level properties, exploring the relationships between the two levels. Finally, an outline is attempted of some implications for studies of language contact and language change.