Halliday & Matthiessen (1999: 242-3):
If we now relate [transcategorisations] to the types of element, we find that in some instances the semantic nature of the transcategorisation is clear. For example,
flake - thing: 'turn into flakes' - process;shake – process: shaky 'tending to shake' – quality,shaker 'that which shakes (= vessel in which dice is shaken)' - thing;awake - quality, awaken 'cause to become awake' - process;analyse - process, analyst 'one who analyses' - thing.
We can gloss these in everyday terms, without recourse to technicality. In other instances, however, the nature of the change is less clear. What for example would be the semantic interpretation of shakiness, awakening, analysis, development? Here we find ourselves using precisely the terms of our own metalanguage in the definition: 'quality of being shaky', 'process of being awake, or causing to become awake', 'process of analysing, developing'.When this happens, it is a signal that a phenomenon of this other kind — quality, or process — is being treated as if it was a thing. The grammar has constructed an imaginary or fictitious object, called shakiness, by transcategorising the quality shaky; similarly by transcategorising the process develop it has created a pseudo-thing called development.What is the status of such fictitious objects or pseudo-things? Unlike the other elements, which lose their original status in being transcategorised (for example, shaker is no longer a process, even though it derives from shake), these elements do not: shakiness is still a quality, development is still a process — only they have been construed into things. They are thus a fusion, or 'junction', of two semantic elemental categories: shakiness is a 'quality thing', development is a 'process thing'. All such junctional elements involve grammatical metaphor.