Halliday & Matthiessen (1999: 265):
Participants are realised by nominal groups, which allow more or less indefinite expansion (through the univariate structure of modification). This expansion is the grammar's way of constructing taxonomies of things: grouping them into classes, assigning properties to them, quantifying them and then uniquely identifying any individual thing, or any number, set or class of things, in relation to the 'here-&-now' of the speech event.
The expansion involves open sets of things and qualities, realised by lexical items; but it can also capture a circumstance, realised by a prepositional phrase, or an entire figure, realised by a clause, and put it to use as a quality in describing or identifying such a thing or set of things, e.g. this unique 20-piece handprinted china dinner service with optional accessories never before offered for sale at such a bargain price.Qualities are attached to things, and so contribute to this overall expansion. They also have possibilities of expansion of their own, by submodification (at least for intensity, but sometimes along other lines as well: very long, longest; dark blue, red hot). …This grammatical potential for taxonomising is complemented by the lexical potential of the nominal group for construing feature networks. So by construing any phenomenon of experience as a thing, we give it the maximum potential for semantic elaboration.