Towards a speaker model of Functional
Grammar
Dik Bakker and Anna Siewierska
- Introduction
According to the standard model of an FG grammar as presented in Dik
(1997: 58ff.) a linguistic utterance is constructed in three stages.^1 First of
all the relevant lexical elements and terms are selected from the Fund, if
necessary after predicate and/or term formation (Stage I). Then from these
basic elements, an underlying representation is constructed in an inside-out
fashion, starting with the layer of the nuclear predicate and ending with the
pragmatic (clausal) layer. In the process, the corresponding operators and
semantic, syntactic and pragmatic functions are assigned to each layer
(Stage II). And finally, once this pragmatico-semantic construct has been
completed, this is converted into a prephonetic string by the expression
rules (Stage III).
The Functional Discourse Grammar model (FDG) as presented in
Hengeveld (this volume) proposes an alternative for the organization of
Stages I and II above, and indeed an extension of the latter. An important
distinction between FDG and the standard model is the separation of
pragmatic and semantic information into two separate levels. The prag-
matic top layer of the underlying representation in the standard model has
been separated from the now purely semantic Representational Level (RL)
and added as the inner layer to an organizationally higher and purely prag-
matic Interpersonal Level (IL). In addition, the interpersonal level has two
higher, discourse-related layers, viz. the Act and the Move. The informa-
tion contained in the IL and RL is constructed in a top-down, outside-in
fashion rather than the inside-out direction (upward layering) followed
throughout in the standard model. The result is that pragmatics goes before