Halliday & Matthiessen (1999: 127):
One question that arises here is the extent to which the semantics and the grammar are in phase with one another. It is conceivable that a semantic sequence of figures could be realised in the grammar by a single clause with complexes at group/word rank — i.e., the sequencing is downranked in course of realisation.
One argument in favour of exploring this possibility is that it would be possible to sort out the potential ambiguity of examples such as Henry and Anne went to the movies — grammatically, we have a group complex (Henry and Anne), but semantically, it could be either a sequence of figures (on the reading 'separately': 'Henry went to the movies and Anne went to the movies') or a sequence of elements (on the reading 'together': 'Henry and Anne went to the movies'). This was, of course, an area investigated in terms of transformational grammar with different deep structures posited for the readings 'separately' and 'together'.
It is clear that there are grammatical group complexes which cannot be expanded semantically as sequences of figures; for example, Henry and Anne chatted is not likely to be a grammatical compression of 'Henry chatted and Anne chatted'; rather, it is agnate with Henry chatted with Anne. In such cases the problem of ambiguity does not arise.