claims. Hence, we take for granted that what look like statements about God do have a
truth value and go on from there to raise questions about the predicates involved.
The first question is this: Why is there a problem? Predicates applied to God—“makes,”
“knows,” “loves,” “forgives,” “speaks”—are all very familiar. Why should there be a
problem about our understanding of them?
To see why there is a problem here we need to realize that the above terms are typical of
those applied to God in that they are borrowed from elsewhere. We learn what it is to
make or know something, to love or forgive someone, to speak to someone from our
experience of and interaction with other people. We then understand God's making,
knowing, or forgiving, if we do and to the extent we do, by some sort of extension of our
understanding of these terms in their human application to their use in application to God.
And so the basic problem is: What kind of extension?
Is it necessary that we borrow terms learned in another sphere of discourse for talk of
God, or could it be otherwise? Could we establish theological predicates from scratch on
their home ground, just as we do with terms for speaking of human beings? No, the
existing order is our only alternative, and for the following reason. We have the kind of
cognitive access to human beings that undergirds a common vocabulary for speaking of
each other, but we lack that support for speaking of God. A parent can tell by observation
when the child is perceiving another person talking or making something, and this makes
it possible to introduce the child to the established meanings of “speak” and “make” in
their human application. But we can't do anything analogous vis-à-vis God. Even if the
child can be aware of God's speaking to her or forgiving her or comforting her, the parent
can't tell when the child is aware of this unless the child informs the parent of it. And that
presupposes that the child has already learned how to apply these
end p.232
terms to God. Thus, there is no possibility of building up a theological vocabulary from
scratch. To be sure, once we have a stock of divine predicates that have been derived
from their human originals, special theological terms, like “grace” or “omnipotence” or
“indwelling” can be introduced on their basis, perhaps with the help of the learner's
participation in religious practice. But there is no possibility of cutting loose completely
from the human prototypes and doing the whole thing on its own. So we are stuck with
the problem of how we can derive terms suitable for theological use from terms originally
applied to human beings.
The simplest way is to make no change at all. Apply the terms to God in just the same
sense as that in which we apply them to human beings. When terms are used in the same
sense in two or more applications, we speak of applying them univocally. Note that this
option does not require us to make the absurd assumption that God is just like a human
being in all respects. Why shouldn't we use “know” or “power” or “good” with exactly
the same meaning in human and divine applications, while fully recognizing that God has
infinitely more knowledge and power and goodness than any human being? But there are
strong reasons for denying complete univocity across human and divine discourse, given
plausible ways of assigning meanings to the relevant terms in their human applications.
Let me make explicit some constraints that govern this discussion. First, the senses of
terms applied to God must be construed in such a way that it is at least possible that they