Martin (1992: 112):
So, within undirected reference there is a contrast between presuming the identity of the participant being realised and optionally presuming the identity of group [sic] of participants to which it belongs. This opposition is illustrated in the difference between [3:32] and [3:33]:
[3:32] The boy took the frog home.[3:33] The boy took the smallest frog home.
In [3:32] the frog signals that the identity of the frog is recoverable; but in [3:33] the smallest frog signals not only that the identity of the frog the boy took home is recoverable (reminding phoricity), but that the identity of the group of frogs from which it was selected is presumed as well (relevance phoricity).
To be clear, these clauses are Martin's constructions only. The actual text (p147) is:
[3:88] xxvii and they saw his pet frog and another frog and lots of baby frogs.
xxviii Tommy took one of the baby frogs home with him.
 This misunderstands demonstrative reference and rebrands the misunderstanding as reminding phoricity (discourse semantics). It is the non-specific demonstrative reference item the that presents the identity as recoverable. The (implied) repetition of frog is lexically cohesive.
 This misunderstands comparative reference (grammar) and rebrands the misunderstanding as relevance phoricity (discourse semantics). Comparative reference, whose items function in the domain of adverbial groups as well as nominal groups, is not realised through superlatives. Again, the cohesion here is lexical, through the (implied) repetition of frog.
It will be seen later that Martin's misunderstanding of grammatical reference leads to his confusing it with lexical cohesion, a textual lexicogrammatical system which he rebrands as IDEATION, an experiential discourse semantic system.