Martin (1992: 98-9):
The basic discourse option organising Table 3.2 has to do with phoricity (from endophoric, exophoric, homophoric, anaphoric, cataphoric). The nominal groups in [3:1] have been organised semantically into phoric and non-phoric classes according to whether their grammar signals the identity of the participant they realise as recoverable or not.
Table 3.2. Coding recoverability in [3:1]
IDENTITY NOT RECOVERABLE IDENTITY RECOVERABLE indefinite article: pronoun: I. I, I, I, I, I, I, me, me, my mum,him, it ['the tiger'] a Hippopotamus…………………... he, he, him a gorilla…………………………… I(t) a baby gorilla a watch demonstrative: this man definite article: the zoo, the zoothe tiger proper name: Mum
 Martin here neglects to acknowledge that the term 'phoricity' comes from his major source, Du Bois (1980: 226), who coined it as a cover term for the types of reference in Cohesion In English (Halliday & Hasan 1976). Its use by Martin here is merely a rebranding of Halliday's 'reference' — the Greek-derived morpheme 'phor-' and the the Latin-derived morpheme '-fer-' are cognates.
 As previously demonstrated, in SFL terms, Martin's focus on "definiteness" confuses nominal group deixis (non-specific vs specific) with cohesive reference (identifiability). As a result, he mistakes entire nominal groups for reference items, as shown in Table 3.2, attributing phoricity to nominal groups instead of reference items, and leading to the absurdity of interpreting Mum as a reference item. The analysis of reference, below, from Halliday (1994: 317), makes a clear distinction between reference items and the nominal groups in which they figure:
 The intrusion of the word 'semantically' is an unwarranted and misleading fudge in Martin's argument for a discourse semantic model of reference. Table 3.2 is a classification of grammatical units ('nominal groups') according to grammatical criteria ('whether their grammar signals the identity of the participant they realise as recoverable or not').