Martin (1992: 518):
It is important not to confuse the semiotic space under construction here with either of two independent dimensions; the interpersonal distinction between proposals and propositions, and the experiential distinction between activities and things. Pursuing the MOOD opposition first, monitoring texts for example can be 'imperative', telling someone what to do (e.g. ærobics class), or indicative, telling someone (e.g. a small child) what is going on (or is about to).
 To be clear, "the interpersonal distinction between proposals and propositions" is one of speech function (semantics), not mood (lexicogrammar). The difference between the systems is stratal.
 A text that tells someone what to do is not a monitoring text. Synonyms of 'monitor' include:
observe, watch, keep an eye on, keep track of, track, keep under observation, keep watch on, keep under surveillance, surveil, check, keep a check on, scan, examine, study, record, note, oversee, supervise, superintend
 In an ærobics class, the language that 'tells people what to do' is instructing, not monitoring. Further, the rôle of language (Hasan 1985/9) in such a situation type is ancillary, not monitoring. (Martin distinguishes monitoring from ancillary.)
 Telling someone what is about to happen is predicting, not monitoring.
The fundamental confusion here is that the discussion is presented as theorising mode, the systems of the textual metafunction at the level of context. Here, instead, the discussion is concerned with the interpersonal dimension of linguistic content — semantics (speech function) and lexicogrammar (mood) — of registers. The confusion is thus simultaneously along three theoretical dimensions: stratification, metafunction and instantiation.