182 J. Lachlan Mackenzie
ing a model of clause production. He nevertheless refers to it as a grammar,
not least because the processes described culminate in a modified,
slimmed-down version of FG, including a rather orthodox expression rule
component. He has thus clarified the stages that take us from the first
communicative intention through to the representation of the predication.
Bakker and Siewierska (this volume) have now provided an existence
proof of the possibility of regarding the expression rules, too, as following
production. They assume a fully specified underlying representation of the
clause which is static and declarative in style. But the expression rules they
propose can and should be seen as dynamic, working top-down, depth-first,
and - crucially, as in actual language production, left-to-right.
The trend towards a re-interpretation of (at least part of) FG as a pro-
duction model is reflected, too, in Hengeveld’s (this volume) call for a
conversion of FG from a bottom-up grammar taking the predicate frame as
its starting point and adding layers in an upward fashion to a top-down
grammar, “starting with the communicative intention and ending with the
articulation of the linguistic expression”. This call cannot be divorced from
the increasing interest in reconciling Functional Grammar and discourse
analysis. This urge to include discourse considerations in the overall model
arises from a desire to achieve ‘pragmatic adequacy’ rather than ‘psycho-
logical adequacy’. Nevertheless, we are aware that psychologists, too, have
been moving away from the examination of individual utterances towards
considering ‘arenas of language use’, as Clark (1992) has it, and the entire
range of activities that occur in those arenas. There may be less distinction
between ‘psychological’ and ‘pragmatic’ adequacy than in the early days
of FG.
The relationship between discourse and grammar can, I believe, be seen
in three different ways, all of which are being tried out in current work.
Firstly, we may seek to model discourse and grammar in separate com-
partments of our theory; we will then try to link them through an interface
(this is essentially the position taken in work by Kroon 1997 and by Vet
1998). Secondly, we may examine the proposition that discourse is struc-
tured in analogy to clausal grammar – this is how I read Chapter 18 of Dik
(1997) and, in different ways, the earlier and recent work of Hengeveld
(1997, this volume) and Moutaouakil (this volume). Thirdly, and this is the
position that seems most consistent with a production-mimicking approach,
we can see discourse production as a dynamic process occurring in real
time and the expression of the clause as a similarly real-time process.
Clark (1996: 285) has pointed out that “Utterances are often viewed as
the prerogative of speakers – products that speakers formulate and produce