Halliday & Matthiessen (2014: 645-6):
In fact we might have (disregarding the scansion, of course) any of the following sequences:
four-&-twenty blackbirds ... the blackbirds began to sing
" the birds began to sing
" the creatures began to sing
" they began to sing
the reference item they being simply the most general of all. Compare ankylosaur ... creature in the following description of a dinosaur:
As an added means of self-defence the ankylosaur had a club on its tail. The creature may have been able to swing the club with great force and aim a savage blow at an enemy.
Such instances are typically accompanied by the reference item the. This interaction between lexical cohesion and reference is the principal means for tracking a participant through the discourse. Instead of the creature we might simply have had the personal reference it; and earlier in the same text, we find such examples:
Ankylosaurus – the “fused lizard” – was the largest of the ankylosaurs, but in spite of its size and frightening appearance it fed only on plants.
Blogger Comments:
To be clear, the source of the notion of 'tracking a participant through the discourse' — which Matthiessen again fails to acknowledge — is not Halliday, but Martin (1992), further elaborated in Martin & Rose (2007). This misunderstanding derives from Martin's confusion of textual reference with ideational denotation (and deixis), and consequently, his confusion of reference with lexical cohesion. Evidence here (Martin 1992) and here (Martin & Rose 2007). In Martin's model, speakers are said to track themselves through a text, and participants are said to be tracked by being omitted.