Remarks on layering 291follows. Performative expressions linguistically render the conceptual out-
put of operations of the central control unit directly. But descriptive,
commitment-less expressions do not result from an operation of the control
system, but render qualifications which have been represented and stored
as part of the speaker’s knowledge of the world.^21 This does not mean,
however, that the distinction can simply be grasped in terms of an opposi-
tion between an operation and a stored representation. For, obviously, an
operation over an SoA still results in a value being assigned to the SoA;
and, as was implied in Section 5.2, this value must even be placed in long-
term storage. What really differentiates qualifications with speaker com-
mitment from those without, then, is this: the former, after having been
executed, remain ‘validated’, i.e. they remain coded within the layered sys-
tem, and this is probably the case as long as the speaker is not forced to
reassess them. This does not apply to the latter. Descriptive qualifications
involving the speaker’s own former views originally involved validated
qualifications in the layered system, which have become invalidated by in-
formation acquired later. But for some reason they have not simply been
erased from memory, but have been stored as knowledge about one’s for-
mer ideas. Descriptive qualifications involving another speaker’s views, on
the other hand, are acquired purely through perception, either because one
has been informed about them, or because one has inferred them from
other observations. So they immediately enter the conceptual system as
knowledge about facts in the world, i.e. as a representation of an SoA.
When a speaker brings up a performative qualification in communica-
tion, however, the ‘operational’ status of the qualification, probably
unavoidably, becomes acute again. Let us take the example of epistemic
modality again. A speaker will probably only bring up his/her degree of
epistemic commitment to an SoA in a discourse if the status of the SoA in
this regard is not obvious, to him/herself, or because the hearer turns out to
hold a different view, or because there is otherwise new information rele-
vant for the speaker’s view. Thus, one may reasonably assume that an SoA
in long-term storage in conceptualization is normally certain knowledge.
That is, the epistemic slot in the layered system over it is set for certainty.
Consequently, its epistemic status is not an issue and is not brought up, at
least not by the speaker, when talking about the SoA. When knowledge is
epistemically less than certain to the speaker, however, it is inherently un-
stable, and (s)he will keep trying to resolve the uncertainty. This means
that (s)he keeps processing information potentially relevant to the SoA,
thereby epistemically reassessing it. (One may actually wonder whether
uncertain information is normally stored for more than a short to interme-