POLISH JOURNAL OF PHILOSOPHY
Vol. V, No. 1 (Spring 2011), 39-57.
P. F. Strawson on Predication
Danny Frederick
Independent researcher
14 Willow Tree Drive, Seaview, Isle of Wight, PO34 5JG, UK
Abstract. Strawson offers three accounts of singular predication: a grammatical, a
category and a mediating account. I argue that the grammatical and mediating accounts
are refuted by a host of counter-examples and that the latter is worse than empty. In
later works Strawson defends only the category account. This account entails that
singular terms cannot be predicates; it excludes non-denoting singular terms from being
logical subjects, except by means of an ad hoc analogy; it depends upon a notion of
identification that is too vague; and it is unnecessarily complicated, relying on analogies
where a more uniform explanation should be possible. But I show how the account can
be corrected to avoid all these difficulties and to provide an accurate account of singular
predication.
1. Introduction
In this paper when I speak of a singular predication, I will be referring to a
kind of sentence, rather than to the proposition expressed by the sentence.
Similarly, I will always use ‘logical subject,’ ‘subject,’ ‘logical predicate,’
‘predicate’ and ‘singular term’ to refer to linguistic expressions. A singular
predication is a sentence in which an individual is characterised in some
way, or in which several individuals are specified to be in some relation to
each other. For example, ‘Aquinas thinks’ is a singular predication in which
Aquinas is characterised as thinking. ‘Aquinas’ is a singular term and is the
logical subject of the sentence, and ‘thinks’ is the monadic predicate. ‘Plato
is older than Aristotle’ is a singular predication in which ‘Plato’ and
‘Aristotle’ are singular terms and logical subjects, and the expression ‘older
than’ is the relational predicate (in this case, a two-place predicate, unlike,
for example, ‘between,’ which is a three-place predicate). In contrast, a
general predication has an expression for generality in place of the logical
subject, as in ‘Everyone thinks’ or ‘Someone is older than Aristotle.’ But
this explanation of singular predication is vague. For example, we might
say that in ‘Aquinas thinks’ the individual attribute, thinking, is
characterised as belonging to Aquinas, in which case ‘thinks’ would turn
out to be the singular term and logical subject and ‘Aquinas’ the predicate.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
40
P. F. Strawson on Predication
Providing a clear and accurate account of singular predication has proven
difficult.
Strawson initially offered three related accounts, namely, grammatical,
category and mediating, but in later work he focused on just the category
account. I will argue that all three accounts fail, but that the category
account, despite its formidable difficulties, can be rectified. In his various
presentations of this account over many years Strawson’s terminology
varies. What he at some times calls ‘properties,’ he at other times calls
‘universals,’ ‘kinds,’ ‘general characters’ or ‘general concepts.’ Whereas in
some places he says that expressions ‘identify’ entities, in other places he
says they ‘introduce,’ ‘signify’ or ‘specify’ entities. I think it is plain that
such variations are intended to be merely verbal rather than substantial, so I
will use ‘property’ and ‘identify’ consistently throughout.
In section 2, I will expound and criticise the grammatical account. In
section 3, I will explain the category account. In section 4, I will set out and
criticise the mediating account. In section 5, I will criticise the category
account before showing how it can be amended quite simply to provide an
adequate account of singular predication. I conclude the discussion in
section 6.
2. The Grammatical Account
On the grammatical account (Strawson, 1959, pp. 146-158), a singular
predication is a sentence in which two expressions, each identifying an
entity, are propositionally combined, that is, combined in such a way as to
express something which is capable of being true or false. The subject is a
noun or noun-phrase: it identifies an entity in the substantive style. The
predicate is a verb or verb-phrase, identifying an entity in the verbal style.
Since the role of the verb in natural languages is to signify combination into
a proposition, we may further say that what distinguishes subject from
predicate is that the latter, but not the former, identifies an entity in the
propositional style. That is, the predicate, in addition to identifying an
entity, also carries the propositional symbolism. This, says Strawson, would
give us a sense in which the predicate is incomplete vis-à-vis the subject.
For while each expression is incomplete in that it falls short of being a
sentence, the predicate, by carrying the propositional symbolism, cries out
for completion. Thus, in each of the following sentences,
(1) Socrates thinks,
(2) Socrates is wise,
(3) Socrates is a philosopher,
the subject is ‘Socrates,’ which is a noun which identifies Socrates. The
predicates are the verb or verb-phrases, ‘thinks,’ ‘is wise’ and ‘is a
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
Danny Frederick
41
philosopher,’ which identify, respectively, thinking, wisdom (or being wise)
and philosophership (or being a philosopher).
However, this account fails for the following reasons. First, we may
split (3) into ‘Socrates is’ and ‘a philosopher,’ in which case the former
expression identifies Socrates in the verbal style while the latter is a nounphrase which identifies philosophership in the substantive style. Yet we do
not want to say that ‘Socrates is,’ as it occurs in (3), is a predicate. Strawson
(1959, pp. 158-159) says that the grammatical account does not entail this
unwanted result, because ‘a philosopher,’ unlike ‘the philosopher who
taught Plato,’ does not identify a specific entity. However, he is inconsistent
here. For if, as he says, ‘is a philosopher’ as it occurs in (3) identifies
philosophership in the verbal or propositional style, then ‘a philosopher’ as
it occurs in the same sentence should also be held to identify
philosophership, but in the substantive style. The fact that ‘a philosopher’
does not identify a specific concrete entity, such as Socrates, is irrelevant.
Second, in
(4) Socrates is the philosopher who taught Plato,
(5) Socrates is Socrates,
‘Socrates is’ attaches to a noun or noun-phrase. But unlike the noun-phrase
‘a philosopher’ in (3), each noun-phrase or noun following ‘is’ in (4) and
(5) identifies a specific concrete entity, in fact, the same concrete entity as
is identified by the grammatical subject, ‘Socrates.’ So ‘Socrates is,’ in (4)
and (5), qualifies as a predicate on the grammatical account. Strawson is
prepared to accept this result because, he says, in such sentences, ‘Socrates
is’ has the force of ‘Socrates is identical with,’ so what it identifies is not
Socrates, but being identical to Socrates. But this is another arbitrary shift
between concrete and abstract entities. Strawson tries to evade the first
objection by averting his eyes from the abstract entity identified by ‘a
philosopher’ and pointing out that it identifies no concrete entity; and he
tries to evade the second objection by averting his eyes from the concrete
entity identified by ‘Socrates’ and pointing out that it identifies an abstract
entity (being Scorates, or being identical to Socrates). His responses to
these two objections are entirely ad hoc.
Third, consider the following sentence:
(6) The rose is the colour of the book.
Here Strawson cannot say that the ‘is’ is the ‘is’ of identity, on pain of
absurdity. Yet if we split the sentence into ‘The rose is’ and ‘the colour of
the book,’ then the latter is the subject on the grammatical account and the
former is the predicate. This gives the result that in (6) we are predicating
the rose of the colour of the book, which is definitely the wrong way
around.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
42
P. F. Strawson on Predication
Fourth, we can happily split sentences like (1) into substantive subject
and verbal predicate. But traditional logicians (for example, Arnauld, 1662,
pp. 103-108) would split sentences like (2) and (3) into three parts: subject
(‘Socrates’), predicate (‘wise,’ ‘a philosopher’) and the sign of
propositional combination (the copula, ‘is’). They would do something
similar for sentences (4), (5) and (6). Here, the verbal or propositional
symbolism is separate from both subject and predicate. So it could be
objected that the grammatical account forces us to lump together separate
elements (predicate and copula) that natural language has been kind enough
to separate.
Although Strawson mistakenly says that the grammatical account works
for natural languages, he regards it as ultimately unsatisfactory (Strawson,
1959, pp. 162-166). For, he says, we could have spoken a language in
which there were no nouns or verbs, but in which entities were identified by
means of neutral expressions. Let us pretend that ‘Raleigh’ and ‘smoking’
are such neutral expressions. We can imagine a language in which
concatenation of two such expressions results in a list, whereas
propositional combination is signified by a bracket. But in addition to the
convention ‘Raleigh (smoking),’ in which propositional indication goes
along with the predicate, there are other possible conventions. In ‘(Raleigh)
smoking’ the propositional symbolism is connected with the subject. In
‘(Raleigh smoking)’ propositional indication is carried by neither subject
nor predicate. In consequence, the grammatical account does not explain
the logical distinction of subject and predicate in a singular predication.
3. The Category Account
Strawson mentions a traditional doctrine to the effect that particulars can
appear in discourse as subjects only, never as predicates; whereas properties
can appear either as subjects or as predicates (Strawson, 1959, p. 137). But
he suggests that, whereas this doctrine appears to base the particularproperty distinction on the subject-predicate distinction, perhaps the order
of explanation ought to run in the opposite direction (Strawson, 1959, p.
161). This leads him to the category account, which explains singular
predication in terms of the types of the entities identified by subject and
predicate. On this view, the basic case of a singular predication is a
sentence, such as ‘Socrates is wise,’ which consists of an expression
identifying a spatio-temporal particular which is propositionally combined
with an expression which identifies a property of such particulars. The
sentence is true if and only if the particular exemplifies the property. The
subject of the sentence is the expression identifying the particular; the
predicate is the expression identifying the property.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
Danny Frederick
43
Just as particulars exemplify properties, so properties exemplify other
properties, as in ‘Wisdom is a virtue’; and these latter properties exemplify
further properties in their turn, and so on. In short, we have the idea of a
hierarchical classification of entities into types, with any entity being of
lower type than any entity which it exemplifies. Spatio-temporal particulars
are thus entities of the lowest type (1971a, 68-69). Accordingly, any
sentence in which two properties are identified, and which is true if one of
the properties exemplifies the other, is also, by analogy with the basic case,
to be called a ‘singular predication.’ The expression identifying the property
of lower type is the subject and the expression identifying the property of
higher type is the predicate. Further, since abstract objects, like numbers,
also exemplify properties, this permits a further analogical extension of the
terminology of ‘subject’ and ‘predicate’ in application to singular numerical
statements, such as ‘Six is even;’ and so on (Strawson, 1959, pp. 171-172;
1971a, p. 69; 1974, pp. 35-37).
Strawson says that on this account the truth of the traditional doctrine
stated at the start of this section is assured. For spatio-temporal particulars
can exemplify properties, but they are not themselves the sorts of things
that can be exemplified. But particulars may be part of what is predicated,
because of the analogy between, for example, ‘is married,’ which identifies
a property, and ‘is married to Socrates,’ which identifies a property-cumparticular (Strawson, 1959, pp. 171-172).
Strawson also says that this account explains a number of asymmetries
between subjects and predicates that have been noted by logicians
(Strawson, 1971b, pp. 96-98, 101-114; 1974, pp. 5-9). The first
asymmetries concern negation:
(n1) by negating a predicate we get a new predicate, but by negating a
subject we do not get a subject;
(n2) by negating a predicate in a singular predication we obtain the
negation of the sentence, but it is not the case that by negating the
subject we obtain the negation of the sentence.
The other asymmetries concern composition. Consider the following four
sentences.
(c1)
(c2)
(d1)
(d2)
Tom is both tall and bald.
Both Tom and William are tall.
Tom is either tall or bald.
Either Tom or William is tall.
The sentences (c1) and (d1) are singular predications with compound
predicates. The sentence (c1) is true if and only if the conjunction ‘Tom is
tall and Tom is bald’ is true. Similarly, (c2) has the same truth conditions as
the conjunction ‘Tom is tall and William is tall.’ The sentence (d1) is
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
44
P. F. Strawson on Predication
equivalent to the disjunction ‘Tom is tall or Tom is bald.’ Similarly, (d2) is
true if and only if the disjunction ‘Tom is tall or William is tall’ is true. So
far, the second sentence of each pair is parallel to the first, so it seems as if
we may regard these second sentences as singular predications with
compound subjects. But differences emerge once we consider negation.
Strawson says that if we apply (n2) to these sentences, while in (c1) and (d1)
we obtain the negation of the sentence if we negate the compound
predicate, this is not so for (c2) and (d2). In these latter cases, we have to
negate the compound subject to obtain the negation of the sentence. This
disqualifies (c2) and (d2) from being singular predications and so it shows
an asymmetry of subjects and predicates with regard to composition.
The explanation of the asymmetries regarding negation and composition
is quite straightforward on the category account. For, with respect to
singular predications of the basic class, the subject identifies a spatiotemporal particular, whereas the predicate identifies a property which such
particulars may exemplify. And there are appropriate asymmetries between
those classes of entity. For every such property, P, and every particular, p,
there is another such property, P*, such that, necessarily, p exemplifies P if
and only if it does not exemplify P*. But it is not the case that any
particular is such that, for every property which it may exemplify, there is
another particular such that, necessarily, the first particular exemplifies the
property if and only if the second particular does not. Similarly, with
respect to composition, the properties that particulars may exemplify stand
in implication relations to other such properties vis-à-vis particulars, but it
is not the case that particulars stand in such implication relations to other
particulars vis-à-vis the properties which they may exemplify. For example,
it is necessary that there is a property (being tall and bald) that is possessed
by all particulars which possess both tallness and baldness. But it is not
necessary that there is a particular that has all the properties that both Tom
and William have. And it is necessary that there are disjunctive properties
(for example, being tall or bald) which are possessed by all particulars
which are either tall or bald, but it is not necessary (or even possible) that
there is a particular that has every property that either Tom has or William
has. Analogous explanations referring to similar facts about entities of
different types will explain the asymmetries of subjects and predicates for
singular predications which are not of the basic class (Strawson, 1971b, pp.
101-114; 1974, pp. 23-29, 35-37).
Strawson (1974, pp. 30-31) offers an explanation based on the category
account for the appeal of the grammatical account. What he feels needs to
be explained is that, in singular predications in natural languages, the
predicate sometimes, though the subject never, absorbs the symbolism of
propositional combination or, to put it another way, takes the form of a
verb. His explanation runs as follows. If we negate the predicate of a
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
Danny Frederick
45
singular predication, we obtain the negation of the sentence; so negation
and the predicate have an affinity for each other. It would be natural to
express the negation of a sentence by modifying the symbolism of
propositional combination; so negation and the form of the verb have an
affinity for each other. As a consequence, there is a mediated affinity
between the predicate and the function of indicating propositional
combination; and this makes it natural, though by no means compulsory,
that the predicate should take the form of the verb.
Finally, Strawson notes some tensions between the category account and
the grammatical account. The first concerns cases where proper names
appear in predicative positions. Thus a gesture might be described as
Napoleonic or a person might be described as a Hitler. But these are clearly
cases where a particular gives its name to a (possibly complex) property
borne saliently by that particular. Another type of case is where something
is described as being American, for example, because it has the property of
having some close connection with America (Strawson, 1959, pp. 173-174).
A more difficult case is represented by a sentence like
(7) Wisdom Socratises,
if used as an alternative to ‘Socrates is wise.’ For the category account
might lead us to say that, in (7), ‘Wisdom’ is the predicate and ‘Socratises’
the subject, for it is the latter that identifies the entity of lower type. But the
grammatical account declares the verbal ‘Socratises’ to be the predicate and
the substantive ‘Wisdom’ to be the subject. Strawson suggests that the
conflict could be resolved by construing ‘Socratises’ as short for ‘is a
characteristic of Socrates.’
I will criticise and correct the category account in section 5.
4. The Mediating Account
Strawson notes the following contrast (1959, pp. 180-186). Each expression
that identifies a spatio-temporal particular presupposes an empirical
proposition. But it is not the case that each expression identifying a
property presupposes an empirical proposition. The intended contrast seems
to be something like this. The expression ‘wise’ identifies a property merely
in virtue of its meaning. But ‘the philosopher who drank hemlock’
identifies a particular only on condition that a certain empirical fact obtains,
namely, that just one philosopher drank hemlock. Given the Description
Theory of Names, to a version of which Strawson subscribed (1959, p.
181), what holds of definite descriptions that identify particulars holds also
of names for particulars. But the point can be made independently of
adherence to that disputable theory. For it is not sufficient for a name to
identify a particular that the name should have a meaning, that there should
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
46
P. F. Strawson on Predication
be something that counts as understanding the name. There must also be a
particular in an appropriate contextual relation to the use of the name; and
whether or not this is so depends upon the empirical facts. For example,
Leverrier thought that an unseen planet was responsible for the
irregularities in the orbit of Mercury. He introduced the name ‘Vulcan’ to
refer to the unseen planet. But the empirical proposition presupposed by the
use of the name turned out to be false: Vulcan does not exist, so ‘Vulcan’
identified nothing. (It should be noted that I am using ‘Vulcan’ here as
Leverrier and his contemporaries used it. Recently, ‘Vulcan’ has been given
as a name to a planet outside of our solar system.) Something similar can be
said for demonstrative expressions identifying particulars. For example,
someone suffering from a delusion might use the demonstrative expression
‘that man’ when there is no man there.
Some philosophers have argued that a genuine name, or a genuine
singular term, cannot have meaning or be understood unless there is an
object which it identifies. On this view, if it is true that someone knows the
meaning of such an expression, then it follows that the expression identifies
a particular. But these philosophers have to admit that when there is no
particular, there can still be an illusion of understanding, an illusion which,
from the point of view of the victim, is indistinguishable from real
understanding, as in the cases of Leverrier and the deluded person just
described. So, on the view of these philosophers, when someone
subjectively appears to understand a singular term purporting to identify a
particular, there is still an outstanding empirical question of whether this
apparent understanding of the meaning of the expression is real
understanding (that is, whether the expression actually identifies anything).
The same sort of empirical question does not always arise in the case of
expressions identifying a property. So Strawson’s point would still hold
even from the point of view of these philosophers; indeed, their view has
similarities to Strawson’s own later point of view (1974, p. 58).
It should be noted that Strawson is denying that expressions identifying
properties always presuppose an empirical proposition. He is not affirming
that they never do. Strawson gives examples of such expressions which do
presuppose an empirical proposition. ‘The quality most frequently
attributed to Socrates in philosophical examples’ happens, as a matter of
fact, to identify wisdom. ‘The disease which kept John from work last
week’ may identify influenza (Strawson, 1959, p. 186).
Thus Strawson is led to his mediating account: a subject of a singular
predication is an expression (short of being a sentence) which, in
identifying an entity, presupposes an empirical proposition, while a
predicate is an expression (also short of being a sentence) which identifies
an entity without in this way presupposing an empirical proposition.
Strawson says that this account harmonises with the category account
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
Danny Frederick
47
because it has the consequence that expressions identifying spatio-temporal
particulars can never be predicates. It also implies that some expressions
identifying properties will be predicates and some will be subjects, which is
also required by the category account (Strawson, 1959, pp. 187-188).
However, despite Strawson’s claims, the mediating account is actually
inconsistent with the category account. For on the latter account, an
expression identifying a property may, in principle, appear either as subject
or as predicate, depending on the type (or level) of entity identified by the
other expression with which it is combined to form a sentence. That is to
say, on the category account, subject and predicate are determined
relatively to each other. But on the mediating account, one can determine
whether an expression is a subject or a predicate in isolation, independently,
on the basis of whether it presupposes an empirical proposition. In fact it
can easily be seen that the mediating account leads to many classifications
which are in conflict with the category account and which are also
mistaken. For example, consider the following sentences, one of which we
have considered already:
(6) The rose is the colour of the book,
(8) Courage is a virtue.
Intuitively, and by the category account, both of these are singular
predications, with subjects, ‘The rose’ and ‘Courage,’ and predicates, ‘the
colour of the book’ and ‘a virtue.’ Yet on the mediating account, (8)
consists of two predicates and (6) of two subjects, since ‘Courage’ does not
presuppose an empirical proposition while ‘the colour of the book’
presupposes that the book exists and has just one colour.
Strawson considers an example like (8). His response (Strawson, 1959,
p. 189) is to say that ‘courage,’ as it appears in (8) should be classified as a
subject because (8) is true if and only if the entity identified by ‘courage’
exemplifies the property identified by ‘a virtue.’ He would presumably
make a similar response to (6). But this response undermines the mediating
account. For we are here compelled to say that an expression which is a
predicate on that account, must yet be classified as a subject because it is
classified as a subject by the category account. Similarly, an expression
(‘the colour of the book’) which is a subject on the mediating account must
yet be classified as a predicate because it is classified as a predicate by the
category account. In short, sometimes the mediating account gives the same
result as the category account and sometimes it does not. When it does not
give the same result, it is wrong and should be ignored. When it does give
the same result, it is not needed. The mediating account appears to be worse
than empty.
Strawson sees the contrast between expressions which presuppose an
empirical proposition, and those which do not, as exhibiting a sense in
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
48
P. F. Strawson on Predication
which the former are ‘complete’ and the latter ‘incomplete.’ For both types
of expression are incomplete in that neither is a sentence, so neither
explicitly expresses a proposition. But the former do presuppose, and thus
present, a proposition while the latter do not. Strawson thinks it would be
fitting if expressions which are ‘incomplete’ in the sense just noted were
also ‘incomplete’ in the sense of being bearers of the propositional
symbolism. So he offers his mediating account as an explanation for the
appeal of the grammatical account.
However, this supposed explanatory value of the mediating account
appears to be illusory. We have seen that the grammatical account fails. For
example, where a predicate is an adjective (like ‘wise’) or an indefinite
noun-phrase (like ‘a philosopher’), it does not have a verbal form and so
does not qualify as a predicate on the grammatical account. If the mediating
account were to explain the allure of the grammatical account, one might
expect it to fail in the same cases in which the grammatical account fails.
But adjectives and indefinite nouns may identify their properties without
presupposing an empirical fact, and thus qualify as predicates on the
mediating account. Similarly, if we split
(3) Socrates is a philosopher
into ‘Socrates is’ and ‘a philosopher,’ the former expression carries the
propositional symbolism and so counts as a predicate on the grammatical
account; but as it presupposes an empirical proposition (‘Socrates exists’) it
counts as a subject on the mediating account. The two accounts classify the
second part of the sentence, ‘a philosopher’ in opposite ways, too (despite
Strawson’s shifty denial noted in section 2).
However, even in the cases in which the two accounts lead to the same
classification, the supposed explanation appears to be a sham. For it
depends on the application of the same word, ‘incomplete,’ to the features
by which each account identifies the predicate. But these are different
features for each account. For the grammatical account, ‘incomplete’ means
carrying the propositional symbolism. For the mediating account,
‘incomplete’ means not presupposing an empirical proposition. But despite
applying the same label, Strawson gives no account of any connection
between the two kinds of incompleteness. Indeed, it may be thought that an
expression which presupposes a proposition, and so which presupposes a
propositional tie, has a greater affinity for the propositional symbolism than
one that does not. But this would give the opposite conclusion to the one
Strawson wants. So Strawson’s claim of explanation is spurious.
Therefore, the mediating account appears to be mistaken; and
Strawson’s contention that the category account is ‘explained and
underpinned’ by it (Strawson, 1959, p. 245) seems preposterous. It seems
likely that Strawson came to see this himself since, in his later works, he
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
Danny Frederick
49
continues to defend the category account but not the mediating account (see
Strawson, 1971a; 1971b, p. 97; 1974, p. 9).
5. Critique and Correction of the Category Account
On the category account, the basic case of a singular predication is a
sentence in which two expressions are propositionally combined, one
identifying a spatio-temporal particular and one identifying a property of
such things, the former expression being the subject, the latter being the
predicate. The sentence is true if and only if the entity identified by the
subject exemplifies the entity identified by the predicate. In other cases of
singular predication the subject is an expression identifying a property or
other abstract entity, while the predicate is an expression identifying a
property of such entities.
We saw in section 2 a type of sentence that is problematic for
Strawson’s grammatical account, namely:
(5) Socrates is Socrates.
This sentence also appears to be problematic for Strawson’s category
account because it contains an expression, ‘Socrates,’ which identifies a
spatio-temporal particular, and which appears in the sentence first as subject
and then as predicate. Both subject and predicate identify a spatio-temporal
particular.
Strawson’s response would be the same as his response to (5) when
considered as a counter-example to the grammatical account: he would
maintain that, in (5), the ‘is’ is the ‘is’ of identity, so that (5) is elliptical for
‘Socrates is identical to Socrates.’ In this expanded version of (5), we have
a subject identifying a particular, Socrates, propositionally combined with a
predicate identifying a property, being identical to Socrates. Alternatively,
we have two subjects, each identifying Socrates, propositionally combined
with a two-place relational predicate identifying the property of identity.
However, this response depends upon the contention that ‘is’ is
ambiguous, as between predication and identity. Strawson offers no
argument for this contention. Lockwood (1975) has argued that the
contention goes back to Frege (1892, pp. 43-44), who needed it to save his
theory from refutation by the existence of sentences such as (5), but that
Frege’s arguments for the contention are invalid. It may also be argued that,
if an expression is ambiguous, then one would expect to have to translate it
into other languages, now one way, now another (Kripke, 1979, p. 19), but
the putative ambiguity in ‘is’ is preserved as we pass from Frege’s German
to English and to other languages. There is not space here to discuss this
matter in any detail. But we can acknowledge that, since (5) appears to have
a natural construal as a predication with a singular term as predicate, it is
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
50
P. F. Strawson on Predication
potentially a counter-example to the category account, and that an account
that dispenses with the supposed ambiguity in ‘is’ would be preferable.
(Rejection of the claim that ‘is’ sometimes means the same as ‘is identical
to’ does not entail that the latter expression has no use or that it does not
express the relation of identity.)
Searle (1969, pp. 115-118) says that if we accept Strawson’s talk of
predicates identifying properties, then we must also accept that subjects
which identify particulars also identify properties. For example, in the
singular predication,
(9) The philosopher who drank hemlock was wise,
the subject identifies the particular, Socrates, while the predicate identifies
the property, wisdom (or being wise). But in the same sense in which the
predicate identifies a property (via its meaning), the subject identifies a
number of properties, namely, philosophership, drinking and hemlock.
Strawson explicitly accepts this point:
the concept green can equally be said to be specified [i.e., identified] in each of the
three following English sentences: ‘The door is green,’ ‘The green door is locked,’
‘Green is a soothing colour’… (Strawson, 1974, p. 21)
but he fails to see that it is a threat to the category account. Since he
originally (1959, p. 181) adhered to a version of the Description Theory of
Names, by Searle’s argument Strawson is committed to saying that not only
definite descriptions, but also names and, in fact, all expressions identifying
spatio-temporal particulars, also identify properties.
We can consider a reply that Strawson might have made to Searle’s
objection. Thus, it could be said that, whereas the definite description in (9)
has semantic parts which identify properties, the expression as a whole
identifies a particular; and it is the whole expression that appears in (9) as
logical subject. But this reply fails. For the definite description taken as a
whole also identifies a property, namely, being uniquely a philosopher who
drank hemlock.
It may be objected that being uniquely a philosopher who drank hemlock
is a strange kind of property in that it can be exemplified by at most one
thing. But we must admit such properties, since they result from the fact,
upon which Strawson draws for his explanation of the asymmetries between
subjects and predicates, that properties have logical relations to each other.
Further, Strawson is happy to admit being identical to Socrates as a
property, as we have already seen, and that property, too, can be
exemplified by at most one thing. Indeed, the category account would
naturally lead us to say that being uniquely a philosopher who drank
hemlock is a property, because it is identified by the predicate in the
singular predication,
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
Danny Frederick
51
(10) Socrates is unique in being a philosopher who drank hemlock.
Another response that Strawson might have made to Searle’s point would
be to stipulate that, in a singular predication, an expression is a subject just
so long as it identifies a particular, no matter what else it identifies. This
would make the definite description in (9) a subject, not a predicate.
However, it would also make the predicate of (10) into a subject. For the
predicate in (10) does not only identify a property, it also identifies
Socrates, just as the subject of (9) does. A slightly different example would
be ‘John is taller than everyone else,’ the predicate of which identifies the
tallest person.
So the category account seems to be refuted. Singular predications of the
basic class were to involve subjects which identified spatio-temporal
particulars and predicates which identified properties of such particulars.
But we have seen that the predicates sometimes identify particulars
(whether or not we accept the supposed ‘is’ of identity), and the subjects (at
least on Strawson’s view) always identify properties.
A further problem for the category account is raised by non-denoting
singular terms. For ‘Vulcan is a planet’ and ‘The author of Principia
Mathematica is English’ are singular predications. But Principia
Mathematica had two authors, and it turned out that Vulcan did not exist.
So, these sentences would not qualify as singular predications on the
category account, for while the predicates identify properties that may be
exemplified by particulars, the subjects do not identify particulars. This
also, incidentally, bars the expressions from being subjects on Strawson’s
other two accounts (each of which require of a subject that it identify an
appropriate entity). Strawson (1959, p. 228) insists, however, that such
sentences should be classified as singular predications, apparently because
the singular terms presuppose a proposition. This presumably supplies one
of the analogies on which Strawson’s account is wont to rely.
Cases like this should make us uneasy. Strawson offers an account of
singular predication which fails to capture some singular predications. But
instead of regarding these sentences as counter-examples to his account, he
looks for and finds some analogy on the basis of which to grant them the
status of singular predication in some extended sense. The manoeuvre
seems ad hoc and to that extent it robs Strawson’s account of explanatory
value.
I think that all these problems can be resolved by abandoning
Strawson’s notion of identification and replacing it with Mill’s distinction
between connotation and denotation (Mill, 1843, book 1, chap. 2, section
5), though without accepting Mill’s denial of connotation to names. The
logical content of an expression is that aspect of its meaning that is relevant
to the inferences in which it may figure. A term is an expression the logical
content of which is a property. The term connotes that property. A one-
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
52
P. F. Strawson on Predication
place term denotes each thing that exemplifies the property it connotes; a
two-place term denotes each ordered pair of things that exemplify the
relational property it connotes; and so on. A singular term connotes a
property that can be exemplified by at most one thing, so it has a unique
denotation if it has any denotation at all.
Strawson’s notion of identification is too loose because it may be taken
to pick out either the connotation of a term or, in the case of a singular term,
its unique denotation. Yet an attempt to restrict the notion of identification
either exclusively to connotation, or exclusively to unique denotation
(where there is such an entity), would ensure that we sometimes get the
wrong result. For, in the case of a singular term, we sometimes want it to be
the connotation that is identified while in other cases we want it to be the
denotation. Strawson’s category account only seems to work, when it does,
because Strawson shifts silently between connotation and denotation in
whatever way is required to give the right result. But once we have Mill’s
distinction, we can specify when it is a singular term’s connotation, and
when it is its unique denotation, that is relevant by means of the
specification of the truth-condition for a singular predication. In the
simplest case, where the singular predication is monadic, we have the
following:
Two terms, T1 and T2, are combined into a singular predication =df T1 is
singular; and the combination is true if and only if the thing denoted by T1
exemplifies the property connoted by T2. T1 is the subject-term; T2 is the
predicate-term. The property connoted by the predicate-term is predicated
of the thing denoted by the subject-term.
This account naturally accommodates the counter-examples to
Strawson’s category account. First, there is no embarrassment caused by
the fact that singular terms may identify both properties and particulars;
indeed, it is central to the account that a singular term may have both
connotation and unique denotation. Second, the fact that a singular term
might not denote is similarly part of the account; for what makes a term
singular is not that it actually denotes something but that its connotation is
such that it can have at most one instance. Third, the account permits
singular terms to appear as predicates in singular predications. Thus,
(5) Socrates is Socrates,
is a singular predication with ‘Socrates’ as subject-term in its first
occurrence and as predicate-term in its second occurrence, in which being
Socrates, or Socrateity, is predicated of Socrates. (If someone insists that, in
(5), ‘is’ is not simply the copula but means ‘is identical to,’ my account will
still classify (5) as a singular predication, but as a dyadic, instead of a
monadic, one.)
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
Danny Frederick
53
The connotation of a denoting proper name is therefore tantamount to a
haecceity, though different connotations may be tantamount to the same
haecceity, as in the case, for example, of the connotations of ‘Hesperus’ and
‘Phosphorus.’ In contrast, the connotation of a non-denoting name, such as
‘Vulcan,’ is not tantamount to a haecceity, because we can grasp a
connotation that is tantamount to a haecceity only if we have an appropriate
contextual connection to the thing which instantiates it.
The distinction of subject-term and predicate-term is a distinction of
logical role, unlike the distinctions between terms, such as singular term
and non-singular term, which are distinctions of logical category. One
category of terms, namely, singular terms, can play either logical role. One
and the same singular term, with one and the same connotation (and thus
the same denotation), makes differing contributions to the truth conditions
of a singular predication depending on whether it occurs as subject-term
(focusing on its denotation) or as predicate-term (focusing on its
connotation). These different contributions are specified by the semantical
explanation of combination into a singular predication: the sentence is true
if and only if the denotation of the subject exemplifies the connotation of
the predicate. So, while the connotation of a term is that aspect of its
meaning that is relevant to inferences in which sentences containing it may
occur, the term’s logical role in those sentences is also relevant. This is true
not only of concrete singular terms such as ‘Socrates,’ but also of abstract
singular terms, such as ‘wisdom.’ Thus, the sentence ‘Wisdom is wisdom’
is true if and only if the denotation of the subject-term, wisdom, exemplifies
the connotation of the predicate-term, namely, being wisdom.
It might be objected that some expressions, such as ‘unique in being a
philosopher who drank hemlock,’ qualify as singular terms because they
denote at most one thing, but cannot appear as subjects of singular
predications because they are of the wrong grammatical form. But this is a
‘parochial’ objection. Such terms can be logical subjects, but the
requirements of our grammar mean we have to give them the form of a
noun before they can appear as grammatical subjects. With regard to the
term in question, we could just prefix the four words ‘the person who is.’
Thus we will recognise the same logical term in different grammatical
guises. This is somewhat similar to the way in which we count the logical
predicate as the same in the sentences, ‘Socrates is a human’ and ‘Socrates
is human’ (we ignore the indefinite article), or in the sentences ‘Peter lives’
and ‘Peter is living’ (ignoring the verbal differences).
Our account may seem to challenge the traditional doctrine that
Strawson expresses as: ‘particulars can appear in discourse as subjects only,
never as predicates’ (Strawson, 1959, p. 137). The source of the doctrine
appears to be Aristotle (2007): ‘that which is individual and has the
character of a unit is never predicable of a subject’ (section 2); ‘primary
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
54
P. F. Strawson on Predication
substance is not predicable of anything, it can never form the predicate of
any proposition’ (section 5). However, it is a consequence of our account
that spatio-temporal particulars cannot be predicated, since while they may
be the denotation of a predicate, it is the connotation of the predicate that is
predicated, and the connotation of a predicate is always a property. And
when Aristotle says that a ‘primary substance’ cannot be a predicate he
cannot mean simply that a primary substance cannot be a bit of language:
he must mean that a primary substance cannot be a property of other things,
which is to say, again, that it cannot be predicated. So there is nothing in
our account of predication to impugn the hoary metaphysical wisdom of
Aristotle. Further, our account plainly preserves Strawson’s explanation of
the asymmetries of subjects and predicates with respect to composition and
negation by reference to the differences between properties and their
instances.
For simplicity I have been speaking so far only of monadic predications.
The more general explanation is as follows.
A singular predication =df a combination of an n-place term and an
ordered set of n occurrences of m singular terms (n ≥ m) which is true if and
only if the things denoted by the n occurrences of the singular terms, taken
in order, exemplify the property connoted by the n-place term. The m
singular terms, in each of the n occurrences, are the subject-terms; the nplace term is the predicate-term.
This account gives a uniform explanation of singular predication, subject
and predicate, rather than the patchwork of basic case and multiple
analogies that Strawson’s account offers. Although we will have to qualify
this in a moment, this is surely a further advantage.
Strawson is aware of this type of objection to the complexities of his
view and he makes a reply to it:
the appearance of any type of item other than a particular in the role of subject is
dependent upon, or presupposes, its capacity to appear in another propositional role.
We could not predicate anything of happiness unless we could predicate happiness
of people. We could not predicate anything of a number unless we could predicate
having a certain number of a set. We could not predicate anything of propositions
unless we framed propositions in which nothing was predicated of propositions. The
propositional role of the particular, on the other hand, is…always that of subject.
(Strawson, 1971b, p. 114)
Even prescinding from the dubiety of the last statement, one surely wants to
protest that such considerations as these are not logical but epistemological.
There is no incompatibility in endorsing a simple and uniform account of
the logical distinction between subject and predicate, on the one hand, while
on the other hand urging the epistemological primacy, amongst singular
predications, of those in which the subject identifies a spatio-temporal
particular. One could even quite consistently go on to maintain the
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
Danny Frederick
55
metaphysical primacy of spatio-temporal particulars over other kinds of
thing. So Strawson’s stated reasons for preferring the convoluted category
account to a more elegant variant do not seem to be cogent.
There is, however, a logical reason for introducing convolutions into an
account of singular predication based on properties and exemplification.
For, if we admit the property of being a property that does not exemplify
itself, we will also admit a paradox; namely, that if this property
exemplifies itself, then it does not, and if it does not, then it does. Russell,
in order to avoid such paradoxes, proposed a rigid hierarchy of types such
that a property could be exemplified only by an entity of lower type and
could only exemplify a property of higher type. If we accept such a
hierarchy, then it seems we also have to accept a hierarchy of types for the
property of exemplification. For the property of exemplification that holds
between particulars and the properties they exemplify would be of a
different (lower) type to the property of exemplification that holds between
these properties and the properties that they exemplify. As a consequence,
there could be no more than an analogy between singular predications
whose subjects identify entities belonging to different types. For whatever
reason, Strawson’s category account reflects such a rigid type-hierarchy.
However, various ways of avoiding paradox without enforcing such a rigid
hierarchy have been proposed, and others are being investigated. These
permit properties to be exemplified by entities of different types, which
may enable us to say, for instance, that if John is dyslexic and dyslexia is
interesting, then John is interesting too, and in the same sense. As a
consequence, while some complications will doubtless need to be added to
our proposed simple and uniform account of singular predication to secure
it from paradox, it seems that these should be far less cumbersome than the
profusion of analogies endemic to Strawson’s account.
6. Conclusion
Strawson initially offers three accounts of singular predication:
grammatical, category and mediating. In later works he defends only the
category account. I have argued that all of Strawson’s accounts fail.
Strawson admits that the grammatical account has counter-examples and
that it is not satisfactory from a logical point of view. But he thinks it is
relatively successful as applied to natural languages. I have argued that
there are more counter-examples than Strawson notices and that the
account’s supposed success depends on artificially lumping together the
predicate with the propositional symbolism even though natural language
has separated the two.
I have shown that Strawson’s mediating account is worse than vacuous,
often giving a wrong classification of subjects and predicates. It also fails to
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
56
P. F. Strawson on Predication
provide any explanation for the appeal of the grammatical account, despite
Strawson’s claim to the contrary. For one thing, it leads to different results
to those provided by the grammatical account; and for another, the
supposed explanation depends on application of the same label to quite
different features of expressions.
Strawson’s category account entails that singular terms cannot be
predicates, which appears to be false. It excludes non-denoting singular
terms from being logical subjects, except by means of an ad hoc analogy. It
involves a notion of identification that hovers vaguely between connotation
and unique denotation. Its reliance on basic case and analogy seems
unnecessarily complicated. The account’s vagueness and its liberal
invocation of analogies make it ad hoc and seriously undermine its
explanatory value.
But it seems easy to correct the category account by invoking Mill’s
distinction between connotation and denotation and explaining a (monadic)
singular predication, simply, as a combination of a singular term and
another term into a sentence which is true if and only if the denotation of
the former exemplifies the connotation of the latter.
References
Aristotle (2007). Categories. (E. M. Edghill, Trans.). Adelaide:
ebooks@Adelaide. Downloaded on 27 April 2010 from
http://ebooks.adelaide.edu.au/a/aristotle/categories/
Arnauld, A. (1662). The Art of Thinking. (J. Dickoff & P. James, Trans.).
Indianapolis: Bobbs-Merrill (1964).
Frege, G. (1892). On Concept and Object. In P. Geach & M. Black.(Eds.
and Trans.), Translations from the Philosophical Writings of Gottlob
Frege (pp. 42-55). Oxford: Blackwell (1980).
Kripke, S. (1979). Speaker’s Reference and Semantic Reference. In P.
French, T. Uehling, & H. Wettstein (Eds.), Contemporary Perspectives
in the Philosophy of Language (pp. 6-27). Minneapolis: University of
Minnesota Press.
Lockwood, M. (1975). On Predicating Proper Names. Philosophical
Review, 84, 471-498.
Mill, J.S. (1843). A System of Logic J. M. Robson (Ed.). Toronto:
University of Toronto Press (1974).
Searle, J. (1969). Speech Acts. Cambridge: Cambridge University Press.
Strawson, P. F. (1959). Individuals. London: Methuen.
Strawson, P. F. (1971a). Singular Terms and Predication. In his 1971c, pp.
53-74.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.
Danny Frederick
57
Strawson, P. F. (1971b). The Asymmetry of Subjects and Predicates. In his
1971c, pp. 96-115.
Strawson, P. F. (1971c). Logico-Linguistic Papers. London: Methuen.
Strawson, P. F. (1974). Subject and Predicate in Logic and Grammar.
London: Methuen.
Please purchase PDF Split-Merge on www.verypdf.com to remove this watermark.