FG from its inception 51PR5, i.e. the problem of discourse, is specifically addressed by FG 4.
Hengeveld (this volume) seeks to deal with discourse phenomena by in-
cluding discourse-sized units in the interpersonal (and possibly a higher
rhetorical) layer and by feeding linguistic information into the modular
‘communicative context’. That is, he attempts to combine the upward-
layering approach with the modular approach (cf. Mackenzie 2000).
The most significant improvement in FG 4 is the notational severing of
the model from FG’s sentence-centric shackles. I say ‘notational’ since it is
ironic that in FG 0 , Dik (1968: 199) advocates “no restriction to sentence as
highest unit” and in FG 1 he writes (1978a: 15) that “FG is meant to cover
any type of linguistic expression, to the extent that the internal structure of
that expression is governed by grammatical rules. It is thus not restricted to
the internal structure of sentences ... ”. FG 4 merely fulfils what Dik always
envisioned for FG. The motivation for this change, however, arose through
FG’s encounter with linguistic phenomena, not with a desire to align the
notation with the vision. Further studies need to be done to see how benefi-
cial the model is for the description of discourse.
Our next concern is PR6, or the problem of psychological adequacy.
Hengeveld suggests that FG 4 has “a higher degree of psychological ade-
quacy” (this volume: 2) than FG 3. This is justified on the grounds that it
starts from the interpersonal layer and works down towards the expression,
while FG 3 moves from the lexicon to predicates through to full expres-
sions.
It would seem that when Hengeveld talks of psychological adequacy he
is actually addressing the problem of the underlying representations: how
are we to interpret the model, what is it a model of? But this poses a prob-
lem: FG 4 is strictly a model of grammar, verified and refined by linguistic
data, and applied to linguistic data; it proves its worth by how good a job it
does of explaining grammatical phenomena. It is not a model of speech
production.
Thus the model itself is agnostic with respect to psychological proc-
esses. Linguists may attach symbolic meanings to the model that are
analogous to cognitive-linguistic phenomena but they are just that, sym-
bolic meanings. Therefore I am not convinced that FG 4 (or FG0-3) can be
psychologically adequate in any meaningful sense. This may trouble some
(and relieve others), so I will present a brief sketch of three ways of under-
standing the relationship between psychology and FG 4.
Firstly, we would do well to study Dik’s original use of psychological
adequacy in FG 1. Using LIPOC as an example (Dik 1978a: 192–212), we
see that LIPOC derives from linguistic data as a general word-ordering