44 Matthew P. Anstey
6.2.2. Some additional problems
Hengeveld’s 1997 model is still quite different from FG 4. We need to re-
view briefly the factors that motivated the other changes in the FG 4 model.
Bolkestein (1985a) is probably the first author to articulate clearly the
problem of discourse (PR7) in FG. She notes how discourse-level consid-
erations affect the code of language. Many other studies considered
discourse within the FG framework, but it was probably Kroon (1995) who
established this problem in the centre of contemporary FG debate. She pro-
vided the important definitions of ‘discourse move’ and ‘discourse act’
adopted in FG 4.
Harder (1989) is to my knowledge the first to point out that the ‘E’ vari-
able cannot be interpreted as referring to a speech act in the way that the
other variables refer to their respective referents, because an utterance is a
speech act. Bolkestein (1992) also remarks that ‘E’ refers ambiguously
both to a clause, which is a product of a speech act, and to a speech act it-
self. Vet (1998) demonstrates that the use of ‘E’ in direct speech implies
that the speaker ‘says’ a speech act. Hengeveld (this volume) therefore
makes two important changes: he changes ‘E’ to ‘A’ (a discourse act) and
he introduces the expression layer to allow the UR to refer to the language
code, as he explains in his Section 6.2.
Van der Auwera (1992: 336) argues that Hengeveld’s use of variables is
at times “an ontological mix-up” between reference and denotation. This
criticism motivates Hengeveld’s (this volume) introduction of referential
and ascriptive acts in the interpersonal level, indicated by ‘R’ and ‘T’ re-
spectively.
Moutaouakil (1996) and Mackenzie (1998) extend Hengeveld’s (1992)
treatment of sentence fragments that omit certain layers in the layered
structure altogether but still remain discourse acts. This explains why in
FG 4 the presence of a particular layer does not entail that all lower layers
within its scope have to be present.
Vet (1998) and Van den Berg (1998) suggest that discourse phenomena
are better handled by a modular approach. Vet’s motivation for the modu-
lar approach is that the phenomena of verbal interaction “obey different
rules” from those of the linguistic expression code. Vet also points out that
such a module is necessary regardless of discourse phenomena, to account
for pragmatic illocutionary conversion for instance. The introduction of the
communicative context in FG 4 is Hengeveld’s modular proposal.