FG and the dynamics of discourse 213information, is fed by the interpersonal and expression levels and simultane-
ously feeds the representational level in order to enable later reference to
earlier acts and expressions.
This discourse- and cognitively oriented expansion of FG (see also
Harder, Connolly, Verstraete, this volume) parallels the trend of other
functionalist approaches, notably Systemic Functional Grammar (SFG) and
West Coast Functionalism (WCF; e.g. Cognitive Grammar, Construction
Grammar, Emergent Grammar, Conversational Analysis, etc.),^2 as well as
the more formally-oriented proposals of e.g. Lexical Functional Grammar
(LFG), Generalized Phrase Structure Grammar (GPSG) or Head-Driven
Phrase Structure Grammar (HDPSG). Some common ground thus seems to
be emerging, seeking speaker-oriented models that treat the speech produc-
tion process as running from intention to articulation (Levelt 1989),
thereby assigning discourse and cognition corresponding levels of repre-
sentation. In these emergent approaches linguistic expressions (and by
implication universals) seem to arise out of the interplay of language-
specific properties (e.g. lexico-grammatical semantics, constructional
meaning) and factors that are not specifically linguistic, such as constraints
on attention and memory, the dynamics of discourse, co(n)textual charac-
teristics, channel restrictions etc., as processed by the human cognitive
system, despite the limitation that at present there is no full-blown answer
as to how this system works or what it looks like.
Possibly to some extent sprouting from this limitation, there are aspects
of FDG – as Hengeveld concedes – that deserve further elaboration in or-
der to become a fully-fledged framework. Indeed, in line with Bakker and
Siewierska (this volume), one misses a more detailed account of the role of
the lexicon in the model, of the processes whereby semantic representa-
tions come into existence, and of the reasons why left-to-right expansions
are only assumed for the interpersonal level when we have every reason to
suspect that these restrictions equally affect the entire production process.
Furthermore, FDG claims that in the coding of his/her communicative in-
tention, the speaker relies on entities of the third, second, first or zero order
(i.e. propositional contents (p), states of affairs (e), entities or individuals
(x), and properties (f)), but at the same time the possibility is admitted that
all entity types may also be expressed in a non-hierarchical way (i.e.
through lexical items). One could find in this last procedure an argument to
support the view that, instead of a threefold hierarchical ordering of IL, RL
and EL, only the latter is required (see Nuyts and Fortescue, this volume,
for concomitant suggestions to minimalize the model), with the lexicon and
the grammar forming a continuous spectrum of symbolic assemblies or