recover our innocence, and each generation born and raised with this augmented
language inherits our original metaphysical sin. So endeth the lesson.
Now for the maxims. Language did not evolve so that we could do metaphysics.^61
Communicating efficiently about the wonders of our internal worlds and the dangers
of our shared common world are of much more practical interest to us than having a
language that corresponds with the ultimate structure of reality, and so it is unsur-
prising that meanings for our expressions—even those cherished by ontologists—
were determined at least as much by the former as by the latter. There are meanings
available that serve these practical purposes and enable us to truthfully communicate
about dream characters,fictional entities, holes and shadows, and so on. If there are
purely linguistic considerations that mandate not taking sentences like“there are
holes”or“there are numbers”at face value, so be it—but I doubt that there are. Do
not let concerns of ontology determine your theory of the semantic values of natural
language expressions. If there is widespread apparent acceptance of various“prob-
lematical ontological commitments,”seek not paraphrases of them but rather accept
them as they are unless other linguistic data decisively tell against the appearances.
Rather,fight the ontologicalfight on the ontological playingfield: I do not doubt that
there are holes—there are—but I also do not doubt that there are notreallyholes.
Similarly, do not doubt whether there are numbers, properties, or propositions.
There are prime numbers between 2 and 9; so there are prime numbers. Wolves and
whales have many anatomical properties in common; so there are properties. Ross
believed what Elizabeth said; there is something that can be both believed and said,
i.e., a proposition. In each case, the premise is true and the inference valid. Believe the
conclusion. Then turn to the pressing question of what manners of being these
entities enjoy.^62
It might even turn out that a quantifier that includes numbers, properties, pro-
positions, as well as concrete entities, is more fundamental than the one employed by
our Edenic ancestors. From their perspective, then, therereallyare things that there
are not—a possibility discussed in section 5.5. Ontological discovery and semantic
change of ontological vocabulary can go hand in hand. If an extended meaning
relative to the meaning for the quantifier we have actually selected is more natural,
and sufficiently many members of our linguistic community discover this fact even if
inchoately, we should expect a meaning shift.
(Note that though there is a formal possibility of it being true that there are really
things for which there are not, I think it is unlikely that this possibility obtains.
Suppose for the sake of argument that there are two meanings for an existential
quantifier, the one that we have actually selected and the extended one that by
assumption is more natural. Suppose that the range of our actual quantifier is
exhausted by entities of kinds K1–K10. Unless there are widespread, frequent, sincere
(^61) As van Inwagen (1990: 130) points out, that’s not what natural languages are for.
(^62) Compare with Schaffer (2009).