284 Jan Nuyts
semantic division in (4) – one not present in the FG system, however – lies
in between time and deontic modality. Deontic and epistemic modality and
evidentiality (and emotional attitude, if it is included in the system) all in-
volve (different) kinds of speaker attitudes, i.e. explicit statements of
degrees to which the speaker is committed (in different ways) to the SoA.
Qualifications from time downward cannot be called ‘attitudinal’, however
(cf. Nuyts 2001a: 344ff.).
These stacks do not bring any new ‘logic’ into the system: as will ap-
pear from their definition above, they fit perfectly into the general rationale
provided in Section 4.1. This also means that the borders between them do
not constitute radical breaks (for the rationale implies graduality). Consider
the middle stack. Quantificational aspect does not change the properties of
the SoA as such, but nevertheless it still draws fairly closely on those inter-
nal properties, since it requires a comparative check whether there is
identity between instances of a potentially recurrent SoA. Temporal and
spatial qualification, however, involve a specification of or situation in the
external dimensions of time and space of an SoA, for which purpose the
internal structure of the SoA matters much less. If we jump to the upper
stack, deontic modality brings in the issue of speaker commitment, which
naturally arises due to the increased role of interpretation that emerges
from the clash between the SoA and other, external knowledge. But in a
way, deontic modality also still situates the SoA, viz. in the social world
and its moral values, without putting its reality status as such at issue.^12
Epistemic modality, however, does not situate the SoA any more, but
rather broaches the question whether it needs to be situated at all. Clearly,
then, the breaks in the system are no more than somewhat more drastic
qualitative jumps culminating from an accumulating number of smaller
jumps underneath them, which all fit into the same general logic.
4.3. Explaining scope extensions in terms of the rationale behind the
system
The rationale behind the layered system discussed in Section 4.1 also natu-
rally explains the differences in the scope of qualifications, not only
relative to each other, but also over domains of the qualified SoA. That is,
in the present account scope is a natural by-product of the logic inherent in
the system. FG deals with the scope of qualifications over domains of the
SoA by relating the qualifications to different hierarchical levels in clause
structure (the predicate, the predication, etc.). This is an essential dimen-
sion of the FG account which I have neglected so far. However, this kind