a naturalistic Trinity. He even includes the unity of the three persons in one God under
the guise of the essential unity of all these aspects of nature.
The second group is extremely varied. The early twentieth-century Spanish-American
philosopher George Santayana took religious doctrines as primarily symbolic of value
commitments and attitudes. In Reason in Religion (1905) he distinguished two
components of a religious doctrine, or “myth,” as he preferred to say. There is (1) an
evaluation of some sort, which is (2) expressed in the form of a picture or story. For
example, the Christian “myth” of God's incarnation in Jesus Christ and his sacrificial and
unmerited death on the cross to atone for the sins of men can be regarded as a symbol of
the moral value of self-sacrifice. That moral conviction can be expressed much more
forcefully and effectively by that story than by just saying “Self-sacrifice is a noble
thing.” Santayana also considers religious myths to have the function of guiding our lives
in certain directions. This directive function is emphasized in Braithwaite (1955). He
takes religious statements “as being primarily declarations of adherence to a policy of
action, declarations of commitment to a way of life” (80). We also find such an approach
in the American theologian Gordon Kaufman. He says that the question of the existence
of God is a question of the viability and appropriateness of an orientation, a true or valid
understanding of human existence (1993, 35–46).
It is clear that much speech about God does have these expressive and directive
functions, and if we have discarded the truth claims that are ordinarily taken to undergird
those functions, the latter will be all that is left. But we will be forced into these
reconstruals by the verifiability criterion only if more traditionally con
end p.224
strued statements about God are not empirically confirmable, and only if empirical
confirmability is a necessary condition of factual meaningfulness. Because I have
presented reason for rejecting the latter, the argument from verificationism against the
possibility of factual truth claims about God can be ignored, and we can proceed to
consider problems that arise with respect to such truth claims.
3. Autonomy of Religious Language?
The next problem on the agenda is whether, as suggested by Wittgenstein and others,
religious “language” is so completely distinct from other uses of language as to constitute
a separate “language game,” with its own battery of concepts, criteria of intelligibility,
criteria of truth, and so on. The most powerful of the current voices that sound this note is
D. Z. Phillips (1970, 1976). In a long series of books he repeatedly insists that religious
beliefs are held subject only to criteria that are internal to religious discourse. He takes
this to imply not only that the traditional arguments for the existence of God have no
bearing on the acceptability of religious beliefs, but that with respect to religions like
Judaism, Christianity, and Islam that rest on claims about historical events, ordinary
historical research has no bearing on their acceptability. This seems strongly
counterintuitive. How could reasons for and against the existence of God be irrelevant to