Radical: Davidson
Radical: Davidson
Radical: Davidson
by Donald DAVIDSON
Kurt utters the words "Es regnet" and under the right conditions we
know that he has said that it is raining. Having identified his utterance
as intentional and linguistic, we are able to go on to interpret his
wonds: we can say what his words, on that occasion, meant. W e could
we know that would enable us to do this? How could we come to
know it? The first of these questions is not the same as the question
what we do know that enables us to interpret the words of others.
For there may easily be something we could know and don't, know-
ledge of which would suffice for interpretation, while on the other
hand it is not altogether obvious that there is anything we actually
know which plays an essential role in interpretation. The second ques-
tion, how we could come to have knowledge that would serve to yield
interpretations, does not, of course, concern the actual history of lan-
guage acquisition. It is thus a doubly hypothetical question: given a
theory that would make interpretation possible, what evidence plau-
sibly available to a potential interpreter would support the theory to
a reasonable degree? In what follows I shall try to sharpen these
questions and suggest answers.
The problem of interpretation is domestic as well as foreign: it
surfaces for speakers of the same language in the form of the question,
how can it be determined that the language is 'the same? Speakers of
the same language can go on the assumption that for them the same
expressions are to be interpreted in the same way, but this does not
indicate what justifies the assumption. All understanding of the
speech of another involves radical interpretation. But it will help
keep assumptions from going unnoticed to focus on cases where inter-
pretation is most clearly called for: interpretation in one idiom of
talk in another.
314 Donald Davidsoa
for the simplest sentences (which they clearly did not), they did not
touch the problem of extending the method to sentences of greater
complexity and abstractness. Theories of another kind start by trying
to connect words rather than sentences with non-linguistic facts. This
is promising because words are finite in number while sentences are
not, and yet each sentence is no more than a concatenation of words:
this offers the chance of a theory that interprets each of an infinity of
sentences using only finite resources. But such theories fail to reach
the evidence, for it seems clear that the semantic features of words
cannot be explained directly on the basis of non-linguistic phenomena.
The reason is simple. The phenomena to which we must turn are the
extra-linguistic interests and activities that language serves, and these
are served by words only in so far as the words are incorporated in
(or on occasion happen to be) sentences. But then there is no chance
of giving a foundational account of words before giving one of sen-
tences.
For quite different reasons, radical interpretation cannot hope to
take as evidence for the meaning of a (sentence an account of the
complex and delicately discriminated intentions with which the
sentence is typically uttered. It is not easy to see how such an approach
can deal with the structural, recursive feature of language that is
essential to explaining how new #sentences can be understood. But
the central difficulty is that we cannot hope to attach a sense to the
attribution of finely discriminated intentions independently of inter-
preting speech. The reason is not that we cannot ask necessary ques-
tions, but that interpreting an agent’s intentions, his beliefs and his
words are parts of a single project, no part of which can be assumed
to be complete before the rest is. If this is right, we cannot make the
full panoply of intentions and beliefs the evidential base for a theory
of radical interpretation.
We are now in a position to say something more about what would
serve to make interpretation possible. The interpreter must be able
to understand any of the infinity of sentences the speaker might utter.
If were are to state explicitly what the interpreter might know that
would enable him to do this, we must put it in finite form. * If this
requirement is to be met, any hope of a universal method of inter-
pretation must be abandoned. The most that can be expected is to
explain how an interpreter could interpret the utterances of speakers
of a single language (or a finite number of languages): it makes no
316 Donald Davidson
We should, I think, consider (E) as evidence that (T) is true. Since (T)
is a universally quantified conditional, the first step would be to gather
more evidence to support the claim that:
(GE) (x)(t)(if x belong8 to the German speech community then
“Es regnet” at t if and only if it is raining near x at t ) )
(x holds true
testable only at the sentential level. The more subtle gain is that very
thin evidence in support of each of a potential infinity of points can
yield rich results, even with respect to the points. By knowing only the
conditions under which speakers hold sentences true, we can come out,
given a satisfactory theory, with an interpretation of each sentence.
It remains to make good on this last claim. The theory itself at beet
gives truth conditions. What we need to show is that if such a theory
satisfies the constraints we have specified, it may be used to yield inter-
preta t ions.
than before about how to interpret if all we knew was that a certain
sequence of sentences was lthe proof, from some true theory, of a
particular T-sentence.
A final suggesltim along these linas would be to say that we can
interpret ,a particular sentence provided we know a correct theory of
truth that deals with the language of that sentence. For then we know
not only the T-sentence for the sentence to be interpreted, but we also
know 'the T-sentences for all other sentences; and of course, all the
proofs. Then we would see the place of the sentence in the language
as a whole, we would know the role of each significant part of the
sentence, and we would know a great deal about (the logical connections
between this sentence and others.
The suggestion fails. For how can it help in interpreting a single
sentence to know the (truth conditions of others? Of course, if we learn
that a speaker also holds other sentences to be true or false, that may
be a help. Indeed, enough more such information, and interpretation
certainly will be possible. But enough more such information, and the
theory isn't needed, for all that went into the theory was information
about sentences held true under vanious circumstances. The point of
the theory is to digest this fund of evidence and to deliver it in a form
useful for the interpretation of isolated utterances. W e must conclude,
I think, that relativizing a T-sentence to a proof or theory is no help:
if the theory does what it is designed to do, T-sentences taken alone
must provide all we need for interpretation.
If we knew that a T-sentence satisfied Tarski's Convention T , we
would know that it was true, and we could use it to interpret a sentence
because we would know that the right branch 'of the biconditional trans-
lated the sentence to be interpreted. Our present trouble springs from
the fact that in radical interpretation we cannot assume that a T-
sentence satisfies the translation criterion. What we have been over-
looking, however, is that we have supplied an alternative criterion:
this criterion is that the totality of T-sentences should (in the seme
described above) optimally fit evidence about sentences held true by
native speakers. The present idea (is that what Tarski assumed outright
for each T-sentence can be indirectly elicited by a holistic constraint.
If that constraint is adequate, each T-sentence will in fact yield inter-
pretations.
A T-sentence of an empirical theory a f truth can be used to inter-
pret a sentence, then, provided we also know that the T-sentence i5
Radical Interpretation 827
entailed by some true theory that meets the formal and empirical,
criteria. It is not necessary to know what the theory is in a particular
case, only that it is such a theory. For if the constraints are adequate,
the range of acceptable theories will be such that any of them yields
some correct interpretation for each potential utterance. To see how
it might work, accept for a moment the absurd hypothesis that the
constraints narrow down the possible theories to one, and this one
implies the T-sentence (T) discussed previously. Then we are justified
in using this T-sentence to interpret Kurt's utterance of "Es regnet"
as his saying that it is raining. It is not likely, given the flexible nature
of the constraints. that all acceptable theories will be identical. When
all the evidence is in, there will remain, as Quine has emphasized, the
trade-offs between the beliefs we attribute to a speaker and the inter-
pretations we give his words. But the resulting indeterminacy cannot
be so great but that any theory that passes the tests will serve to yield
interpretations.
FOOTNOTES
1. Here and throughout this paper my debt to the work of W.V.O. Quine will
be obvious. The term "radical interpretation" is meant to suggest a strong
kinship with Quine's "radical translation" (Word and Object, Cambridge, Mass.
1960). Kinship is not identity, however, and "interpretation" in place of " trans-
lation" marks one of the differences: a greater emphasis on the explicitly
semantical.
2. At one time I was convinced that unless such a finitely characterized theory
could be provided for a language, the language couldy not be learned by a
creature with finite powers. (See Donald Davidson, Theories of Meaning
and Learnable Languages", in Proceedings of the 1964 International Congress
for Logic, Methodology and Philosophy of Science, Amsterdam 1966, pp. 383-394.)
This still seems to me likely to be right, but Georg Kreisel has made me realize
that the point is not obvious.
3. The idea of a translation manual with appropriate empirical constraints as a
device for studying problems in the philosophy of language is, of course, Quine's.
This idea inspired much of my thinking on the present subject, and my proposal
is in important respects very close to Quine's. Since Quine may not have
intended to answer the questions I have set, the claim that the method of
translation is not adequate as a solution to the problem of radical interpretation
may not be a criticism of any doctrine of Quine's.
4. Alfred Tarski, "The Concept of Truth in Formalized Languages", in Logic,
Semantics, Metamuthematics, Oxford 1956.
5. For a discussion of how a theory of truth can handle demonstratives, and how
Convention T must be modified, see Scott Weinstein, "Truth and Demonstra-
tives", Nods (forthcoming, 1974).
328 Donald Davidson
6. See John Wallace, "On the Frame of Reference", Synthese, Vol. 22 (1970).
pp. 61-94.
7. Tyler Burge, "Reference and Proper Names", Journal of Philosophy, Vol. 70
(1973), pp. 425-439.
8. Gilbert Harman, "Moral Relativism Defended", forthcoming.
9. John Wallace, "Positive, Comparative, Superlative", Journal of Philosophy,
Vol. 69 (1972), pp. 773-782.
10. Donald Davidson, "On Saying That", Synthese, vol. 19 (1968), pp. 130-146.
11. Donald Davidson, "Causal Relations", Journal of Philosophy, vol. 64 (1967),
pp. 691-703.
12. Donald Davidson, "Quotation", unpublished. Forthcoming as a chapter of The
Structure of 'Truth, Oxford.
13. Michael Dummett, Frege, London 1973.
14. Readers who appreciate the extent to which this account parallels Quine's account
of radical translation in Chapter 2 of Word and Object will also notice the
differences: the semantic constraint in my method forces quantificational struc-
ture on the language to be interpreted, which probably does not leave room for
indeterminacy of logical form; the notion of stimulus meaning plays no role in
my method, but its place is taken by reference to the objective features of the
world which alter in conjunction with changes in attitude towards the truth of
sentences; the principle of charity, which Quine emphasizes only in connection
with the identification of the (pure) sentential connectives, I apply across the
board.
15. This idea, and others rejected here, will be found in various articles of mine:
see "Truth and Meaning", Synthese, vol. 17 (1967), pp. 304-323, "Semantics for
Natural Languages", in Linguaggi nella Societd e nella 7ecnica, Milan 1970,
pp. 177-188, and "True to the Facts", Journal of Philosophy, vol. 66 (1969),
pp. 748-764.
Donald Davidson
The Rockefeller University
and Princeton University
New York 10021