142 Francis Cornish
With this subsuming, controlling relationship between the higher-level
Interpersonal Level and the lower, intermediate-level Representational one,
which forms a message on the basis of the pragmatic structuring specified
at the Interpersonal Level, Focus assignment or the lack of such assignment
may be shown to have an effect on the semantic nature of the predicator
selected at the lower Representational level, something which was impos-
sible or very difficult to show under the standard model of FG (Dik 1997a:
Ch. 13).^24 Bolkestein (1998: 204–206) likewise gives Latin examples
where the assignment of Focus function to given constituents in a clause
may block Subject or Object assignment to certain other constituents. Thus,
in her words (p. 205), “the location of main Focus must be part of the input
for the rule that takes care of syntactic function assignment”. All these phe-
nomena show clearly that pragmatic function assignment must take place
before, and not after, the stages at which semantic and syntactic properties
are established within the clause. In any case, as the various tests, such as
the Question-Answer test for Topichood and potential Focus domains,
clearly indicate, information from the wider discourse context in which a
given clause is set is crucial for the assigning of pragmatic functions to
relevant constituents within it.
The structure of the discourse corresponding to a particular Move, then,
is established at the Interpersonal Level in the new model, in terms of a co-
herence relation with respect to the last such Move, and to the wider
discourse and thematic structure as specified by the Communicative com-
ponent. The mode of message management adopted at this level will in part
determine at the lower Representational Level a particular underlying
clause structure, with Topic and Focus assignments marked (as appropri-
ate) as a function of the message structure indicated at the Interpersonal
Level. Finally, the Expression rule component will convert these specifica-
tions into an actual object-language expression (but now, there is no
requirement of a one-to-one relation between a type of systematic coding
in the language in question and a given Topic or Focus function: Topic ref-
erents, in particular, are established as such independently of a given
coding type, and indeed are already marked as such at a more language-
exclusive level of representation – the Interpersonal Level). So topics (ref-
erents) may be such without there needing to be dedicated language-
specific coding devices for expressing them, the Interpersonal and Repre-
sentation Levels now being distinct and containing non-isomorphic types
of units.