Halliday & Matthiessen (1999: 201-2):
It will be clear that both these strategies, the expansion of word to group and the accumulation of contrasting words, are strategies for assigning qualities to things: we can gloss a slipper, for example, as a soft supporting garment worn by either sex at the lower extremities of the body. In general the qualities involved will come from the same semantic domains in either case; and there is often fluctuation between the two strategies within the language — we sometimes talk of slippers and sometimes of soft shoes. …
And although we have presented the two strategies as discrete, there are of course intermediate modes of construing that form a continuum between the two: noun-compounding (more syntagmatic) and morphological derivation (more paradigmatic). So we find compounds such as pushbike, motorbike; and derivational series like cycle : bicycle, monocycle, tricycle. It is not difficult to invent new categories if we need them such as bikelet or megabike. Note also the morphological strategy for deriving casual terms from formal ones: bicycle > bike (cf. omnibus > bus).
The sort of strict taxonomy that is typically associated with related series of nominal groups is often a feature of special registers of the language. The limiting cases of such taxonomies are those found in the specialised technical registers of science and technology; these include some which are partially or even wholly designed in a conscious exploitation of the grammatical resources involved. The "things" that are construed in this way include the more abstracts concepts of a scientific theory, the virtual objects that are postulated to explain the more arcane phenomena that impinge on human experience.