A New Architecture for Functional Grammar (Functional Grammar Series)

FG from its inception 47

illustrating the working of the expression level leaves open the possibility
of alternative interpretations of the model.
Let us now turn to PR2: the problem of the underlying representations.
It is difficult to evaluate FG 4 ’s treatment of this problem since Hengeveld
does not discuss it. He only writes that “... the construction of linguistic
expressions can be interpreted as a decision-making process on the part of
the speaker” (this volume: 10).
Nevertheless, one could argue that Hengeveld addresses this problem
indirectly by introducing a third interpretation of underlying structures. Let
me explain. The interpersonal level clearly represents the intention of the
speaker, indicating discourse moves, discourse acts, ascriptive acts, and
referential acts. The variables for these do not have ‘meaning’ per se, as
they are outside C, “the information communicated in a discourse act”
(Hengeveld this volume: 8). Neither could they be triggers for the expres-
sion rules to generate the correct linguistic expression. In a word, the
interpersonal layer appears to be representing the pragmatics of communi-
cation, but pragmatics in a particular way. On the one hand, the
interpersonal layer represents pragmatics in a very broad sense, including
the illocutionary and intentional aspects of language use. On the other, it
represents pragmatics in a rather narrow sense, as Hengeveld restricts the
interpersonal layer to representing only grammatically coded reflections of
communication. So although a language may be used for dozens if not
hundreds of different types of discourse acts, illocutionary acts, intentions,
and so forth, it is only those with a unique grammatical expression that will
be represented.
It should be clear from my historical survey that the solution to PR2 is
not simply to clarify the status of URs, but in a principled manner to iden-
tify and untangle the components of URs and thus work out what belongs
where. And although Hengeveld does not address PR2 explicitly, his clear
separation of what can be understood as pragmatic, semantic, and syntactic
layers paves the way, in my opinion, for a resolution of this problem. In
other words, Hengeveld’s three layers correspond to the three possible in-
terpretations of the single-layered UR of FG1-3. To illustrate, let us return to
our three example sentences illustrating definiteness, this time with a pos-
sible FG 4 -like annotation:

(23) a. Mr Smith went home grumpy.
interpersonal: Decl (R: ‘Mr Smith’: ‘grumpy’)^39
representational: (dsx: Mr Smith: grumpy)
expression: (NP: /mstə smθ/ ... AdjP /grmpi /)