Comment clauses and FDG 207ence in structure, the English mechanisms share with the Spanish mood
distinction the element of undermining the power of the matrix clause over
the subclause – something which is important in understanding the rela-
tionship between grammar and discourse categories in such cases (cf.
Bolinger 1968).
The absence of that is also significant: that can be understood as an op-
erator which (like ‘declarative’) takes scope over the entire proposition (cf.
Harder 1995), reifying it into an entity (while ‘declarative’ uses it to con-
vey a piece of information). Therefore the absence of that when the matrix
clause is in the comment clause position is different from cases where that
is optionally absent. By being preposed without the complementizer, the
subclause is effectively promoted into a full declarative clause.
The realignment of matrix and subclause does not bring about a full
parallelism between discourse status and grammatical status. Distributional
tests would still class it with matrix clauses. This is not so obvious with I
am afraid, however, so for ease of illustration let us use the formal version
I fear instead. In sentence (6):
(6) John is ill, I fear.
the verb fear is prima facie without a second argument, and the only
way of bringing about a canonical dependency situation is by having a
grammatical interpretation under which John is ill remains embedded in
relation to the verb fear, while at the same time I fear is treated analogi-
cally with subclauses in terms of discourse status.
An investigation of this field, I suspect, will lead to the need for many
subdivisions of the representational layer of organization with many differ-
ent links to both the expression and the interpersonal levels. The work of
pursuing this angle will also require a reconsideration of the status of the
sentence from a discourse perspective. Hansen (2001: 128) takes up this
issue, quoting an example from German from Franck (1985: 234):
(7) Das war / also im Jahre 1907 / bin ich geboren.
that was / then in:the year 1907 / was I born
‘approx. So that was in 1907 I was born.’
Here we see a constituent with what Franck called ‘double-bind’ status,
serving both in a sentence begun in the previous instalment and a sentence
completed in the following instalment. Hansen’s conclusion is that the sen-
tence should perhaps be replaced as the relevant unit of analysis in favour