Martin (1992: 516-7):
Experientially mode mediates the degree to which language is part of or constitutive of what is going on. In text [4:2] for example, most of the meanings are made verbally (excepting the exophoric reference discussed above). At the dog show itself however, language would have a much smaller rôle to play in the showing area, where most of the meaning is realised through action, not words. … Showing a dog and describing how a dog is shown are in other words very different modes.
 This confuses metafunctions. The system of mode, and thus all its feature oppositions, is a resource of the textual metafunction.
 In SFL theory, mode does not "mediate". Oppositions such as constitutive vs ancillary are features of mode. Mode enables both field and tenor.
 This dimension of mode — the opposition of constitutive vs ancillary — corresponds to Hasan's (1985/9: 58) system of LANGUAGE RÔLE:
First, there is the question of the LANGUAGE RÔLE — whether it is constitutive or ancillary. These categories should not be seen as sharply distinct but rather as two end-points of a continuum.
 The exophoric references, of course, were also made verbally. Exophoric reference is a linguistic resource. In this case, the exophoric references were made to manual gestures on a table surface.
 This confuses material order phenomena: what people do, with semiotic order (meta)phenomena: what people say. This confusion is further confounded by associating the material doing with Hasan's language rôle (textual semiosis).