Halliday & Matthiessen (1999: 270-1):
We can then summarise our interpretation of grammatical metaphor as follows.(1) There is an increase in textual meaning, since participants have the most clearly defined status as information: in particular, they can be construed (by the thematic and information systems) into a 'backgrounded + foregrounded' pattern which maximises the information potential of the figure.(2) There is a loss of experiential meaning, since the configurational relations are inexplicit and so are many of the semantic features of the elements (e.g. engine failure : an engine / engines // the engine / the engines; failed / fail / will fail &c.)(3) There is a further loss of experiential meaning, since the categories of experience become blurred (failure is not most obviously felt as a 'thing', otherwise it would have been construed as such in the first place); the construction of reality becomes a construction of unreality, detached from ordinary experience and hence inaccessible and remote.(4) There is however a gain in the potential for experiential information, because the participant, more than any other element, can be expanded in respect of a wide range of semantic features; this enables anything construed as a thing to become part of an experiential taxonomy which embodies far greater generalisation about the overall nature of experience.