204 Peter Harder
focus of interest of the Danish Functional Grammar group under the
heading ‘iconicity and structure’).
Let us look at the analysis of the ambiguous English sentence (3) above.
In terms of linguistically represented content, there is presumably no ambi-
guity in English: under both readings, it is a declarative clause with two
argument positions, the latter being filled by an embedded clause. In the
FDG analysis there is no level at which this identity of coded content is ap-
parent, which is why (cf. above) I would like to extend the representational
level to accommodate the linguistically represented content of I am afraid
in its interactive function.
If we do this, the interesting facts about the ambiguity can be brought
out by interfacing the two different interpersonal readings with the same
linguistically represented content. It is generally assumed (cf. Hannay and
Mackenzie 1996: 114) that the main information is standardly put in main
clauses, yielding an unmarked ‘iconic’ matching relation between the in-
terpersonal level and the level of linguistic representation. If we see the
information in the subclause as a subact in its own right (it may be some-
what implausible, but not, I think, out of the question; I propose it for
purposes of illustration rather than as a hypothesis), this yields the ‘inter-
face’ analysis of the main features of the ‘emotion report’ reading shown in
Figure 2 below.
In English, there is no linguistic trigger of the discrepant second read-
ing; it is simply a fact of usage. In Spanish, where the same structural
discrepancy is found, the indicative está provides a coding explanation, be-
cause it overrules the motivation for a subjunctive that emerges from the
matrix verb, declaring everything with its scope to be ‘real’ – which is dis-
crepant with the standard influence of a mental matrix verb.