Halliday & Matthiessen (1999: 207-8):
The Attribute is not a prototypical participant We have already noted that as it stands it cannot function interpersonally as Subject. On the other hand, it can easily be instated as a participant by adding the noun or the noun substitute one(s) to the nominal group which realises it: this is a dry plate, this is a dry one. The fact that the thing can be instated as the Head of a nominal group serving as Attribute illustrates the point already made: the quality does not construe a separate class of thing, it presumes this class from the environment. Thus this is heavy means that it is heavy relatively to whatever class of thing it has been assigned to; compare the truck was very heavy / a very heavy one (i.e. "heavy for a truck') with the chair was very heavy / a very heavy one (i.e. 'heavy for a chair'). Hence it is not possible to re-instate the thing where the Attribute occurs as resultative in a process of doing: we do not say I've wiped it a dry plate or I've wiped the plate a dry one.