When Donald Davidson died aged 86 in 2003, he had been a towering figure
in philosophy for more than forty years. Ever since the 1960s he has shaped
philosophical discourse by publishing a number of hugely influential essays,
covering topics in semantic theory, philosophy of language, mind, and action,
as well as epistemology and ethics. These essays are now collected in five
volumes, of which two were published after Davidson’s death under the guidance
of his wife, Marcia Cavell. At least the two collections published in the
eighties, Essays on Actions and Events
(1980) and Inquiries into Truth and
Interpretation (1984), can safely be counted amongst the most important
philosophy books of all time.

Because Davidson has made all his important contributions to
philosophical research in the form of essays, it may be surprising to learn
that Davidson completed a book before his death. Truth and Predication, as it is called, comes with sleeve notes
advertising it as the distinguished author’s "first proper book".
There is no need at this point to discuss what constitutes a ‘proper’ book, but
it is worth having a look at the genesis and content of Truth and Predication. In 1989, Davidson gave a series of three
lectures on truth at Columbia University. These three lectures were published
in 1990 as an issue of The Journal of
Philosophy under the title "The Structure and Content of Truth".
This text constitutes the first half of the new book. The second half of Truth and Predication is based on a
series of four lectures delivered at several institutions in 2001 and 2002.
These four chapters on the problem of predication have previously been
unpublished. The book had been completed and accepted by Harvard University
Press when Davidson died. Final revisions that Davidson intended to make in
response to comments from James Higginbotham, Tyler Burge and Charles Parsons
were partly incorporated into the text, partly added in bracketed footnotes.

Because the first part of the book has been previously published and
widely discussed, it is the second part on the so-called ‘problem of
predication’, surely, that ought to attract our attention. Because this second
part of the book does not draw on the material on truth in the first part — or
only in the sense that it makes substantial use of the term ‘truth’ — it can
easily be read and discussed on its own. In what follows, I will restrict my
attention to this second part of Truth
and Predication, that is, to predication.

There is a number of different ways in which chapter four, "The
Problem of Predication", articulates the problem Davidson takes himself
and the philosophical tradition to be dealing with. Essentially, it is the
question how subjects and predicates combine to form sentences that have
various familiar characteristics and uses. An adequate account of predication
must, according to Davidson’s formulations, explain "the semantic role of
the copula", the "nature of predication" (p. 76), the
"unity of the proposition", "the nature of judgment",
"how names or other singular terms [are] related to predicates" or
"how a substance is related to its attributes" (all p. 77). Because
the question underlying these formulations was introduced to Western philosophy
by Plato (p. 83), Davidson devotes the first half of chapter four to showing
how the problem arose in the context of Plato’s theory of forms, and then turns
to Aristotle.

The historic treatment of the problem continues in chapter five, where
Davidson considers the accounts of Russell, Strawson, and Sellars. Russell’s
and Strawson’s views (at least one of Russell’s views, that is), together with
Plato’s and Aristotle’s, are rejected for the same reason: By assuming that the
semantic roles of both names and predicates consist in referring to some
entity, they all share an assumption that Davidson sees as leading into a
vicious regress. According to this family of views, the semantic difference
between names and predicates is based on the difference between the kind of
entity the respective terms refer to, e.g., particulars in the case of names,
properties or universals in the case of predicates. But however these two
entities are characterized, it remains to be explained how they are related,
and at this point Davidson diagnoses a regress. Wilfrid Sellars fares slightly
better in that he is credited with realizing that it is generally a bad idea to
conceive of predicates as designating anything at all. However, Sellars’ view,
too, is rejected.

The solution to the problem, which is prepared and presented in chapters
six and seven, comes in two parts, of which the first is owed to Frege, the
second to Tarski. Frege’s contribution is the thought that predicates can be
thought of as ‘incomplete’ or ‘unsaturated’ functional expressions, mapping
objects (the referents of the proper names that ‘complete’ or ‘saturate’ the
predicates) onto truth values. Davidson acknowledges this assimilation of predicates
to functional expressions as an important step forward, but insists that Frege
himself did not solve the problem of predication. This negative assessment is
based on two readings of Frege (both by Michael Dummett), which are both
rejected. Call them ‘hard’ and ‘soft’. The hard Fregean view that predicates
are actually functions, mapping objects (the referents of proper names) onto
objects (truth values), is deemed unattractive for familiar reasons (e.g.,
assimilation of sentences to proper names). The soft Fregean account, according
to which predicates are functional in character, but not really functions
(because their values are not objects) is judged explanatorily unsatisfactory
(pp. 136-140). I will get back to this assessment later.

Tarski’s contribution to the solution to the problem of predication is
sketched on just the final five pages of the book. It is based on Davidson’s
trademark idea that Tarski’s truth definitions for formal languages can be used
to explain the meanings of subsentential expressions in natural languages.
Accordingly, the semantic role of predicates consists in being completed or
satisfied by certain entities in certain orders. For some predicate to be
satisfied is for the sentence of which it is part to be true. The ordered
couple <Tom, Jerry>, for example, satisfies the predicates x is a cat and x is chasing y if and only if the sentences ‘Tom is a cat’ and ‘Tom
is chasing Jerry’ are true, while the ordered couples <Jerry, Tom>, or
perhaps <Jerry, King Kong> do not satisfy these predicates.

This is, of course, just the kind of explanatory use of Tarski’s truth
definitions that Davidson has always advocated. The solution to the problem of
predication, Davidson contends, has been worked out "unrecognized as such,
almost seventy years ago" (p. 77). Presumably because the solution has
been around for so long, and because Davidson has been pointing at it ever
since he published "Truth and Meaning" in 1967, the new book offers
only the briefest of sketches of how Tarski’s truth definitions can be used to
explain predication. The meagre presentation of the solution, combined with the
effort previously spent on historically tracing the problem and the failed
attempts to solve it, comes across as a rhetorical anticlimax. However, the form
here perhaps fits the message: This is it, and yes, sorry, it’s all there is to
say.

But slim solutions, too, ought to be carefully motivated and explained.
Enjoyable as it is to read how Davidson deals with the problem of predication,
he should have said more about how the alleged solution is meant to work, and
why it is acceptable as a solution. It would have been particularly helpful to
see Tarski’s solution as related to the soft Fregean view, which seems to share
important features of the ‘solution’ while itself being diagnosed
unsatisfactory. When discussing the soft Fregean view that predicates are like
functions in that they map entities onto truth values, Davidson questions
"whether predication has now been explained" (p. 139), because this
proposal amounts to just the claim that "the predicate does just what we
know predicates do". However, at the same time he is confident that it is
explanatorily sufficient to say that predicates are ‘what is true’ of entities
in the context of a Tarski-style truth definition. More details should have
been given on the difference between Tarski and a soft reading of Frege.

Once it is conceded that the sections on Plato, Aristotle, Russell,
Strawson, Sellars and Frege are too sketchy to be of much use to the researcher,
all the reader walks away with is a handful of criteria of adequacy for alleged
solutions to the problem and Davidson’s assertion that Tarski’s truth
definitions, properly adapted, provide the solution. The solution is not spelt
out, and the criteria of adequacy are not argued for. Moreover, there is a
tendency to tacitly promote truth from an obvious candidate for a criterion of
adequacy for possible solutions to the only candidate to play a basic role in
such solutions. This would amount to a promotion indeed, which can be
challenged and should be justified. There is, of course, an important
difference between semantic projects that accept as a criterion of adequacy
that they have to explain what it is for a proposition to be true, and theories
that treat some notion of truth as explanatorily basic. Davidson often glosses
over this difference, for example by saying that one thing we can learn from
the "history of failures" is "how central the concept of truth
is to any solution" (p. 155).

Davidson ends his introduction to this book by musing whether readers
will be happy with his own, i.e. Tarski’s, solution to the problem of
predication: "Not surprisingly, I think that the best approach is some
version of Tarski’s truth definitions, though for several reasons this is not a
conclusion with which everyone will be happy, and probably no one will be
completely content with it." (p. 6) This worry may or may not apply,
depending on the reader’s views in philosophy of language. What can be said for
sure, however, is that no one will be completely content with Davidson’s presentation of what may or may not be
the solution, for all we can say after having read the book, to the problem of
predication.

© 2005 Dominique Kuenzle

Dominique Kuenzle is a PhD
candidate at the University of Sheffield.  He works on pragmatist accounts
of conceptual content and is interested in rational, discursive and epistemic normativity,
its ‘continental’ critics and rationalist defenses.

Categories: Philosophical