Remarks on layering 279ciples) between qualificational categories and linguistic forms, in the sense
that there is no one-to-one relation between them. Space prevents me from
elaborating this here (see for details Nuyts 2001a: 305ff.), but in strongly
generalized terms the mismatch comes in two types.
One type is involved in the phenomenon of semantic paradigms intro-
duced above: one qualificational category is expressed in a range of
different form types. There are two subtypes of this. One qualification as a
whole can be expressed in alternative forms: for example, epistemic
modality codes into adverbs, adjectives, auxiliaries, and mental state and
similar predicates.^5 Or different dimensions of one qualification can on oc-
casion be expressed in two forms, as for instance when a negative polar
epistemic evaluation is simultaneously expressed by one form denoting the
epistemic scaling and one denoting the negative polarity (cf. I doubt vs I
don’t think, or it is doubtful vs it is improbable vs it is not probable). Such
one-meaning-to-multiple-forms mappings are to a large extent due to the
multi-functionality of language use, i.e. to the interaction of the semantic
category with other functional factors pertaining to its linguistic expression
(such as, very prominently, information structure), and the different de-
mands those functional dimensions impose on structure (Nuyts 2000).
The other type of mismatch works the opposite way: one linguistic form
is used by two or more qualificational categories. This, too, comes in two
subtypes. Two qualifications may be expressed simultaneously in one
form: I think, for example, expresses epistemic likelihood and subjectivity
(and as Perkins 1983 shows, such combinations are rather common). Alter-
natively, two or more qualifications may on different occasions use one
and the same form: past tense, for instance, normally marks time, but can
sometimes express epistemic modality, viz. when it is used as a weakener
on other epistemic expressions, as in the ‘past’ modals in English (e.g. may
vs. might) or in ‘performative past’ mental state predicates (i.e. certain uses
of I thought). These kinds of mismatch are clearly not accidental or unsys-
tematic. The former subtype is no doubt due to the frequent co-occurrence
of the two qualifications in certain usage conditions (for example, subjec-
tivity and epistemic uncertainty frequently go hand in hand in spontaneous
conversation). And the latter subtype is a by-product of the principle of ex-
pressibility, viz. the long-term striving for qualificational categories to find
expression, whereby expression forms are not created out of the blue, but
are initially ‘borrowed’ from other meanings on the basis of the principle
of (most probably) metonymic transfer.^6
The mismatch between meanings and forms not only underscores the
need to separate a conceptual system of qualifications from a linguistic sys-