A distinction must be made between the philosophical theory of conceptual analysis and the historical
philosophical movement of Conceptual Analysis.
The theory of conceptual analysis holds that concepts - general meanings of linguistic predicates - are the
fundamental objects of philosophical inquiry, and that insights into conceptual contents are expressed in
necessary 'conceptual truths' (analytic propositions). There are two methods for obtaining these truths:
(1) direct a priori definition of concepts;
(2) indirect 'transcendental' argumentation.
The movement of Conceptual Analysis arose at Cambridge during the first half of the twentieth century, and
flourished at Oxford and many American departments of philosophy in the 1950s and early 1960s. In the USA its
doctrines came under heavy criticism, and its proponents were not able to respond effectively; by the end of the
1970s the movement was widely regarded as defunct. This reversal of fortunes can be traced primarily to the
conjunction of several powerful objections: the attack on intensions and on the analytic/synthetic distinction; the
paradox of analysis; the ‘scientific essentialist' theory of propositions; and the critique of transcendental
arguments. Nevertheless a closer examination indicates that each of these objections presupposes a covert appeal
to concepts and conceptual truths. In the light of this dissonance between the conventional wisdom of the critics on
the one hand, and the implicit commitments of their arguments on the other, there is a manifest need for a careful
re-examination of conceptual analysis.
1 Origins and career of Conceptual Analysis
Many of the elements of Conceptual Analysis are present already in John Locke's Essay Concerning Human
Understanding (1689) - in his doctrines of general ideas and definitions (decompositions of complex general ideas
into sets of simple ideas); in his distinction between 'trifling' and 'instructive' universally certain propositions; and
in his closely related distinction between 'intuitive' and 'demonstrative' knowledge ( Locke, J. §§3-4). An even
more important source of influence is Immanuel Kant's Critique of Pure Reason (1781/1787). There Kant makes
three crucial sets of distinctions. The first is between 'analytic' and 'synthetic' propositions, that is, between
propositions true by virtue of conceptual content alone, and propositions true by virtue of conceptual content
together with a non-conceptual semantic element ('intuition'). The second is between a priori (necessary,
experience-independent) and a posteriori (contingent, experience-dependent) truths. The third is a threefold
division between proofs by empirical methods, 'constructive' proofs in mathematics, and 'transcendental' proofs.
Transcendental proofs establish the truth of non-mathematical synthetic a priori propositions by showing how the
natural sciences - and human experience itself - presuppose a set of primitive pure concepts or 'categories' (
Kant, I. §4).
Kant's important idea that conceptual truths can be either analytic a priori or synthetic a priori is effectively erased
by Gottlob Frege in his Foundations of Arithmetic (1884). Frege's overriding philosophical aim is to put
mathematical proof on a firm footing by reducing the truths of arithmetic to analytic truths of logic. In view of this,
the proper goal of an analysis is the production of non-circular, explanatory, yet meaning-preserving general
definitions of fundamental concepts - as exemplified in Frege's famous definition of a number as a class of
equinumerous classes ( Logicism; Frege, G. §§7-8). Analytic a priori truths for Frege are propositions that
follow deductively either from the self-evident, unprovable laws of pure logic alone, or else from the laws of logic
together with logical definitions.
Frege's method was enthusiastically developed and subtly transformed by G.E. Moore in what may be regarded as
the first phase of Conceptual Analysis. Moore supplemented Frege's austere logicism with a Kant-inspired
attentiveness to the multiplicity of different sorts of concepts, propositions and logico-semantic relations, and with
a predilection for arguments resting on appeals to common sense ( Moore, G.E. §§3-4). Moorean analysis then
travelled from Cambridge to Oxford, where J.L. Austin and Gilbert Ryle added to it a special focus on the uses and
abuses of ordinary language. This led directly to the vigorous growth in the 1950s of the second phase of
Conceptual Analysis, sometimes also called 'Oxford Philosophy'. Conceptual analysis was exported to the USA in
the 1950s and early 1960s, primarily through the writings of H.P. Grice and P.F. Strawson. While it found a niche
for a time in many American philosophy departments, it did not ultimately survive. It was attacked on several
fronts by leading American philosophers (most damagingly, perhaps, by W.V. Quine - see §3 of this entry for
details), and by the end of the 1970s had largely succumbed.
2 The theory and methods of conceptual analysis
The career of Conceptual Analysis was rather brief and embattled, but its underlying philosophical theory,
conceptual analysis, should be analysed and judged on its own merits. For simplicity's sake, we can think of
conceptual analysis as defined by the conjunction of the following five theses:
(1) The content thesis. A concept is a general content possessing intrinsic, individuating structures and relations
(an intension), and having a corresponding application either to sets of actual or possible objects (an extension), or
to other concepts.
(2) The linguistic thesis. A concept is the meaning of a predicate-expression; and all such words have meanings
only in the context of whole sentences used (first and foremost) in making statements in ordinary discourse.
(3) The modal thesis. Every true proposition expressing conceptual interconnections is necessary and analytic.
(4) The knowledge thesis. Purely conceptual inquiry produces important a priori knowledge. This knowledge is
expressed in analytic propositions known to be true either by (a) direct definitional analysis of conceptual contents,
or by (b) indirect ‘transcendental' arguments.
(5) The metaphilosophical thesis. All fundamental philosophical errors arise from misunderstandings of concepts,
and can be corrected only by proper conceptual analyses.
The first two theses convey a theory of concepts. Being general, concepts play the role traditionally assigned to
universals ( Universals). Yet because they are ontologically dependent upon ordinary language, concepts are
not otherworldly, Platonic entities. And because concept-possession depends upon linguistic use and mastery,
concepts are immediately accessible to all competent speakers.
Concepts bear necessary relations to one another and also have fixed internal structures; these relations and
structures are open to the process of analysis; and a capacity for analytical insight is guaranteed by linguistic
mastery. Concepts, however, are of two quite different sorts: 'categorematic' and 'syncategorematic'.
Categorematic concepts (for example, 'bachelor' or 'being taller than') are 'material' intensional contents that
uniquely and independently determine concept-extensions. Syncategorematic concepts, by contrast, are 'formal'
intensions that apply in a rule-like way to other concepts or conceptual complexes. These in turn are of two sorts:
(1) ‘logical concepts' (such as 'conjunction') expressing logical operations; and
(2) 'categorical concepts' (such as 'objecthood') expressing higher-order conditions of the applicability of
lower-order concepts.
Analytical insight into categorematic and syncategorematic concepts permits the capture of both non-logical and
logical truths (for example, 'bachelors are unmarried males' and ' » (P & » P)') within the general class of
conceptual truths.
The modal thesis tells us that all conceptual truths are analytic and necessary; such truths reflect conceptual
contents alone and bear no connections to the disposition of things in the actual world or any possible world. They
are therefore 'topic-neutral'. This makes is relatively easy to see why, as the knowledge thesis asserts, the
cognition of analytic propositions is a priori: the insight into conceptual content requires no appeal to empirical
facts or individuals. And certainly in the case of such simple definitional propositions as 'bachelors are unmarried
males', it appears to be the case that a direct awareness of conceptual identity - guaranteed by linguistic
competence and the grasp of word-synonymy - requires no appeal to experience in order to be known. But the very
idea of a conceptual identity is not so simple as one might think; nor does insight into conceptual truth always
result from definitional inquiries alone. This is manifest in three ways.
In the first place, definitional truths do express conceptual identities or synonymies of words, but the criterion of
identity cannot be merely that concepts are identical, or words synonymous, when they share the same extensions
necessarily. The concepts ‘creature with a heart' and 'creature with a kidney', for example, share actual
extensions, but are clearly not identical. And as C.I. Lewis first pointed out in ‘Modes of Meaning' (1943-4) there
are also concepts - such as ‘equilateral triangle' and ‘equiangular triangle' - that are necessarily co-extensional, but
not precisely identical. Hence a stricter criterion of conceptual identity, involving an isomorphism of the
concepts' internal structures, must be invoked.
In the second place, there are analytic propositions expressing conceptual relations that reflect only partial
identities of concepts, for example: (A) 'bachelors are males'. And most logical truths appear not to reflect either
complete or partial conceptual identities. Here, however, it is possible to appeal to a criterion of analyticity used
by Kant, namely that the denial of an analytic proposition leads to a contradiction. This is closely connected with
the idea that when terms in partially definitional propositions are replaced by their full definitions, or perfect
synonyms, logical truths will result. Thus substituting 'unmarried males' into (A) for 'bachelors' produces the
logical truth, (A*) 'unmarried males are males'. The denial of (A*) is obviously logically contradictory. So an a
priori grasp of conceptual identities and logical concepts appears to be sufficient for knowledge of definitional
propositions and logical truths alike.
Third, however, there are conceptually true propositions, such as (B) 'nothing can be simultaneously coloured in
two different ways all over' and (C) 'the world as we experience it contains reidentifiable objective particulars in a
single spatiotemporal scheme', that do not seem to reflect logical truths, or even complete or partial identities of
concepts, but rather conceptual connections of a somewhat different sort. Here we are strongly reminded of Kant's
view that some conceptual truths are not analytic, but instead synthetic. And indeed, although conceptual analysts
generally eschew the existence of the synthetic a priori, this is precisely where the appeal to transcendental
arguments comes in. A transcendental argument aims to show that a proposition P (say, (B) or (C)) is conceptually
true because it is presupposed by another proposition Q (say, 'a is red; so it is not green' or 'a is not being
perceived by me now; but it is still over there just the same'), which is taken by hypothesis to be perfectly
acceptable and a ‘paradigm case' of some class of statements. Not only is P a necessary condition of the truth of Q,
but more profoundly P is a necessary condition of the real possibility or meaningfulness of Q - of its being true or
false in the first place. This is because the concepts expressed in P are categorial concepts having a 'conceptual
priority' over the concepts expressed in Q, which is to say that the concepts in P have a central place in the overall
'conceptual scheme' by which language-using human beings organize their common sense experience in the ways
exemplified by Q. Thus the conceptual truth (B) expresses an insight about the very nature of human experience of
colour; and the conceptual truth (C) expresses an insight about the very nature of human perception of objects.
Transcendental arguments thus extend the scope of conceptual analysis from the mere definitional or logical
exploration of conceptual contents (sometimes also called 'philosophical grammar'), towards insights into first
principles expressing the ‘conceptual geography' of the common sense world ( Transcendental arguments).
The metaphilosophical thesis follows directly from the other four. Concepts govern the ways we think about all
things and other concepts; thus not only all philosophical truths, but also all philosophical errors, are conceptual.
The two methods of conceptual analysis - definitional and transcendental - must be employed not merely as means
of philosophical insight but also for the unmasking and diagnosis of conceptual confusions.
3 Five fundamental objections
The many different lines and styles of criticism directed against conceptual analysis tend to converge on five basic
objections:
(1) The flight from intensions. If concepts are linguistic intensions, then obviously any sceptical argument showing
that intensions do not exist will undermine the linguistic thesis. Just such an argument has been influentially
promoted by Quine, in two parts. First, intensions are said to be either ontologically 'mysterious' or purely
psychological entities that intervene between language (or linguistic behaviour) and reference, and should be ruled
out of any properly logical and scientific approach to semantic issues. Second, all the explanatory roles
traditionally played by intensions - as what words signify, as truth-vehicles, as grounds of synonymy, as grounds
of modality, as objects of the propositional attitudes, and as objects of philosophical analysis - can be functionally
mimicked by logical or linguistic devices that make no appeals whatsoever to intensional entities (Intensional
entities).
(2) The death of analyticity. Perhaps even more famous than Quine's attack on intensions is his attack, in ‘Two
Dogmas of Empiricism' (1951), on the very idea of analyticity ( Quine, W.V. §8). Setting aside logical truth,
Quine argues that non-logical analyticity is based on the concept of synonymy. But every plausible attempt to give
Routledge Encyclopedia of Philosophy, Version 1.0, London and New York: Routledge (1998)
Conceptual analysis
an explanation of synonymy (by appeal to the notions of definition, linguistic interchangeability, or semantical
rules) ends either in circularity or vacuity. In the absence of a clear account of synonymy, no clear boundary
between analytic and synthetic (factual, contingent) propositions can be established. If sound, this argument forces
the rejection of the modal thesis ( Analyticity).
(3) The paradox of analysis. In ‘Moore's Notion of Analysis' (1942), C.H. Langford points up a deep difficulty in
the conception of a definitional analysis. In order for a proposition expressing the results of such an analysis to be
correct or true, it must establish a complete or partial identity between concepts. But if an identity is so
established, then the very same concept, wholly or in part, redundantly shows up twice in the same conceptual
truth, as expressed by two different words or phrases. Thus every correct definitional analysis of a concept is
non-informative and trivial; and the very project of definitionally analysing a concept is epistemically pointless. If
true, it follows directly from the paradox that the first part of the Knowledge Thesis, which states that all
conceptual truths express important a priori knowledge, is false.
(4) Scientific essentialism and the contingency of conceptual truths. Enshrined in the linguistic thesis, the modal
thesis and the knowledge thesis, are claims to the effect that the meanings of words are conceptual intensions, and
that conceptual truths are analytic, a priori and necessary. But it has been influentially argued by Hilary Putnam
(1975) that the extensions of some general words - 'natural-kind' terms such as 'water' or 'cats' - are not in fact
determined by their corresponding concepts. The extension of a natural-kind word, says Putnam, is instead
determined by a strict relation of identity between the natural kind and the microphysical stuff that locally
predominates in the samples used by scientists in their empirical investigations ( Reference §3). The stuff's
physical microstructure - say, water's being H2O - is its scientific essence; and propositions expressing this
essence - say, 'Water is H2O' - are necessary and a posteriori. But this immediately implies that the conceptual
propositions expressed by the use of sentences including natural-kind terms will not be necessary truths. For
example, consider the apparently necessary (because analytic by partial definition) proposition 'Water is a liquid'.
The natural-kind word 'water' will pick out only whatever stuff in a given world has the microstructure H2O. But
it is possible that on a different world, under different physical conditions, the stuff that is H2O and a liquid here
on Earth might look very different and have very different surface properties: it might be solid, for example. So the
conceptual proposition 'Water is a liquid' is false in that possible world; and thus it is only contingently true in the
actual world, even if grasped a priori.
(5) Transcendental arguments presuppose verificationism. Even supposing that definitional conceptual truths are
empty tautologies, and not always necessary, still conceptual truths gained by transcendental arguments would
remain cognitively significant and modally secure. But as Barry Stroud (1968) has pointed out, the theory of
transcendental arguments assumes a strongly verificationistic theory of meaningfulness for concepts and
propositions. Verificationism, however, is afflicted with insurmountable problems. So transcendental arguments
are semantically suspect, and the second part of the Knowledge Thesis would thereby seem to be undermined too.
4 The inescapability of conceptual analysis
On the assumption that these criticisms are sound, things look very bleak for conceptual analysis. And it is true
that the movement of Conceptual Analysis did eventually break up under the weight of the criticisms just
described. But a closer inspection reveals a striking feature of the philosophical dialectic: In order to gain the
acceptance of any argument aimed against conceptual analysis, it appears that the critic must finally appeal to the
truth of some premises that implicitly invoke concepts and conceptual truths.
To take only one central example. Quine's famous arguments against intensions and analyticity all assume the
notion of a logical truth. According to Quine in 'Truth by Convention' (1936), a logical truth is a sentence that
contains certain words (logical constants) 'essentially': these words preserve their interpretations across every
uniform assignment of values to the non-logical constants in the sentence, ensuring that it 'comes out true' no
matter what. Now logical constants, with their 'essential occurrence', are semantically equivalent to the conceptual
analyst's logical concepts; and in this way Quinean logical truths are (covertly) conceptual truths. Moreover,
although Quine suggests in ‘Two Dogmas of Empiricism' (1951) that even logical truths are revisable, he states in
his later Philosophy of Logic that ‘every logical truth is obvious, actually or potentially' ([1970] 1986: 82), and
that the very attempt to deny a logical truth would involve a change of meaning of the logical constants. In other
words, logical constants and logical truths are ineliminable parts of any rational conceptual scheme recognizable as
our own. This recognition is epistemically equivalent to what conceptual analysts mean by the a priori grasp of a
conceptual truth; the only difference is that whereas most analysts hold that logical truths are known and justified
by direct conceptual insight, Quine persuasively appeals instead to a transcendental proof.
If sound, this argument smoothly generalizes. No philosopher can do without logic; and if logic is itself necessarily
such as to contain concepts and conceptual truths that are grasped a priori, and whose existence and validity can be
established only via transcendental argument, then no philosopher can ultimately avoid the analysis of concepts.
Supposing that conceptual analysis is - even in this minimalistic way - philosophically inescapable, the demand for
a re-examination and re-working of its basic theses seems self-evident.