Martin (1992: 97-8):
The Tagalog rôle marking system interacts with participant identification in that the Theme is almost always "definite" — and so translated into English as the.
… and it turns out that Tagalog does not have a [definite/indefinite] article system as part of its grammatical resources for tracking participants. Like English it does have demonstratives, pronouns and proper names … . And unlike English, the Theme in Tagalog is almost always definite.
The [this] means that the work of identifying participants in English is done mainly through the nominal group, whereas in Tagalog the process is undertaken through co-operating clause and nominal group systems … .
In the remainder of this chapter English resources for participant identification will be outlined. To begin the discourse semantics of 'definiteness' will be presented, followed by a review of the implications of stratifying discourse semantics and nominal group systems.
 Martin switches between 'identification' and 'tracking' as if they are equivalent, without providing any supporting argument. This might be seen as a variant of the logical fallacy known as proof by assertion. See the previous post for the distinction between the two.
 In SFL theory, the 'definite/indefinite' contrast is theorised as the contrast between specific and non-specific DEIXIS in the system of DETERMINATION (Halliday & Matthiessen 2014: 366). Its inclusion here demonstrates that Martin has confused the (structural) nominal group system with the (non-structural) system of cohesive reference. It will be seen that this misunderstanding pervades this chapter and further undermines the theoretical validity of his system of IDENTIFICATION.
 In SFL theory, there is no "article system", because 'article' is form (a word class), not function.
 Given that Tagalog has specific Deictics and pronouns, it is likely that it is these that serve the function of demonstrative and personal reference.
 If it is true that 'the Theme in Tagalog is almost always definite', then the unlikely claim is that the nominal groups realising Theme almost never have non-specific Deictics; that is, the counterparts of English each, every, both, all, neither, no, either, some, any, etc.