Towards a speaker model of FG 327pression rules. We will take as our point of departure the proposals relating
to the expression component made in Bakker (2001) rather than the stan-
dard expression rules as presented in Dik (1997). The major aspects of this
dynamic model of expression will be presented in Section 2. Section 3 will
briefly discuss linguistic modelling in general. Furthermore, we give a
critical discussion of some aspects of the FDG model in this light. Sugges-
tions are given to clarify some of our points. In Section 4 we will look at
the implications of FDG for our revised expression rules. We will integrate
the two models, and simulate the production of a short stretch of spoken
discourse on the basis of this integral dynamic model. Finally, in Section 5
we draw some conclusions for the overall organization of an F(D)G gram-
mar and its further development.
- Overview of the dynamic model of expression
In their standard form, the FG expression rules come in three phases. First,
starting out from a fully specified underlying representation, which con-
tains the predicates and the relevant functions and operators, the
grammatical elements are generated. This gives us both the independent
grammatical morphemes such as determiners, auxiliaries and prepositions,
and the respective kinds of bound morphemes, such as tense markers and
agreement suffixes. In the second phase, all lexical and morphological ma-
terial is ordered into a linear string. In the third phase this string is given its
final phonological shape.
In Bakker (1994; 1999) it is shown that there are two major shortcom-
ings to the way the expression rules are traditionally organized. Firstly the
expression component as it stands undergenerates, i.e. it cannot produce
certain forms that actually occur in languages. This endangers the descrip-
tive adequacy of an FG grammar in a rather obvious sense. A clear, and
ubiquitous example of undergeneration is provided by constructions in
which form and order interact. Examples (1) from Breton (Borsley and
Stephens 1989) and (2) from Dutch are a case in point.
(1) a. Ar vugale ne lenn-ont ket levrioù
the children PCL read-3PL not books
b. Ne lenn ket ar vugale levrioù
PCL read not the children books
‘The children do not read books.’