Halliday & Matthiessen (1999: 267-8):
It is clear that, among the various elements involved, the relators are the most unstable in terms of their susceptibility to metaphoric transformation. They are, as it were, the first to leave; and they travel farther than the rest.We can perhaps link this property of relators to their status in the overall ideation base. Relators construe the highly generalised logico-semantic relations of expansion that join figures into sequences: elaborating, extending, enhancing. We have remarked already on the fact that these relationships of expansion pervade very many regions of the semantic system: they are manifested in the organisation of figures of being, in the types of circumstantial element that occur within a figure, in the taxonomy of 'things', and elsewhere, as well as of course in their 'home' region of the construal of sequences, as links between one figure and another. This led us to characterise the categories of expansion as "transphenomenal" and as "fractal":
- transphenomenal in the sense that they re-appear across the spectrum of different types of phenomena construed by the ideational system; and
- fractal in the sense that they serve as general principles of the construal of experience, generating identical patterns of organisation of variable magnitude and in variable semantic environments.
It is these characteristics of relators that make them particularly liable to migrate: to be displaced metaphorically from their congruent status (as paratactic and hypotactic conjunctions) and to appear in other guises in other locations — as minor processes (in circumstantial elements), as processes, as qualities and as things. Thanks to this metaphoric instability, relators are able to play a central part in the re-construal of experience that is a feature of the discourse of the sciences — that makes these discourses possible, in fact, and hence provides the semiotic foundation for the construction of scientific knowledge.