Generalizing vocable uses becomes a process of attenuating its effectiveness.
Attenuation, as a process, includes more than the extension of certain relations, it also includes changes in the type of relations. The notion of attenuation is not the familiar notion of temporal increase in the extension of a relation. From a model-theoretic point of view, we think of an n-adic relation as a set of n-tuples of objects, and if we choose, we come to think of a relation as the union of a set of sub-relations temporally ordered by inclusion, say, and so to speak of the extension of the relation varying over time. But from the same model-theoretic standpoint, attenuation represents more than just a change in extension. In the first place, it includes changes in the semantic types of objects, and therefore changes in the semantic types of relations. Ultimately attenuation includes the introduction of uses of the vocable which, though n-adic, are not relational (that is, joining nominals) but rather connectival (that is, joining main clauses). The relatively primitive but of
No one BUT Harry spoke
eventually gives rise to the but of
Harry spoke, BUT Hilda did not.
Thus the notion of attenuation requires resources for its
explication beyond those required even for temporal variability in the
extension of a relation. Extensionally speaking, an n-ary relation is
just a set of ordered n-tuples. To say therefore that the extension
of n-ary relation R varies over time is either to utter an oxymoron or
to speak elliptically. Conversationally R acquires members, but
within the standard semantic representation, that conversational idea
is accommodated by speaking of the n+1-ary relation 
in which the
n+1th position is a temporal index. What attenuation and its eventual
outcomes require is more radical than this, since attenuated R begins
to acquire n-tuples of vocables of different semantic types. What was
relational vocabulary may acquire pre-connectival uses; that is, it
may find n-tuples of vocables in its extension that are the semantic
representatives of noun phrases and noun clauses rather than of nouns
(terms).
At this point, it would be semantically, even conversationally, more convenient to admit defeat and introduce systematically variant notation.