FG from its inception 43(18) (ςi: [ ω 3 [ ω 2 [ ω 1 predN (xi) θ 1 ] θ 2 ] θ 3 ])
Hengeveld (1992) is an important article. Firstly, he incorporates Dik’s
(1989: 50) zero-level ‘f’ variable into the representation for all predicates
and an illocutionary ‘F’ variable. Thus the layered structure, with operators
added, is:
(19) E 1 : [π 4 F 1 : ILL (F 1 ) (S) (A) (π 3 X 1 : [predication] (X 1 ))] (E 1 ))
π 2 e 1 : [(π 1 f 1 : predβ (f 1 )) (ω x 1 : (f 2 : predN (f 2 )) (x 1 ))n] (e 1 )Hengeveld justifies the ‘f’ and ‘F’ variables using his characteristic ar-
gument: because a predicate and illocution may serve as an antecedent for
anaphoric reference, a variable for the antecedent needs to be included in
the UR. Secondly, Hengeveld clarifies the place of adverbs in Functional
Grammar and provides functional definitions for parts of speech. Finally,
we see the first hint that the representational level is not always necessary
when Hengeveld briefly mentions that interjections such as Ouch! may
have a structure as follows:
(20) E 1 : (fi: ouchInt (fi)) (EI)
Hengeveld (1997) introduces a rhetorical layer above the interpersonal
layer, which consists of a discourse (D), a discourse type (T), and a move
(M). Hengeveld changes (S) and (A) in the illocutionary frame to (P 1 )S and
(P 2 )A and allows recursion at all levels to indicate the hierarchical structure
of discourses. This approach to discourse became known as the ‘upward
layering approach’. The layered structure is accordingly:
(21) D 1 : [(T 1 : TYP (T 1 )) (M 1 : [speech act] (M 1 ))n] (D 1 )
(E 1 : [π 4 F 1 : ILL (F 1 ) (P 1 )S (P 2 )A (π 3 X 1 : [predication] (X 1 ))n] (E 1 ))n(π 2 e 1 : [(π 1 f 1 : predβ (f 1 )) (ω x 1 : (f 2 : predN (f 2 )) (x 1 ))n] (e 1 ))nIn FG 4 the Move ‘M’ has moved down into the interpersonal level and
the rhetorical level is missing. However, it would seem that Hengeveld still
believes in such a level since he writes, “there are higher levels of dis-
course organization which are not captured here” (this volume: 5).