On Kripke and Statements
On Kripke and Statements
On Kripke and Statements
S aul A. Kripke is one of the most creative and influential philosophers in the
twentieth century. It is not an exaggeration to say that he helped change the
face of analytic philosophy in the last half of the century. His contributions to logic
and philosophy are broad, deep, and long lasting. Since it is not feasible in an article
like this to mention and discuss all of his contributions, I shall not attempt to do
so.1 Instead I will focus on what seems to be a problem for Kripke’s position with
respect to certain necessary a posteriori truths and true negative existentials. I shall
tentatively suggest that within Kripke’s work a solution to the problem in ques-
tion can be found provided one is willing to distinguish statements from proposi-
tions. First, the problem.
Kripke’s claim that Hesperus is Phosphorus is both necessarily true and known
only a posteriori has produced a wide range of responses. I offered the following
argument against Kripke’s position that there are necessary a posteriori truths of
this sort (Fitch 1976):2
1. For an overview of Kripke’s contributions, see my forthcoming book Saul Kripke soon to
be published jointly by Acumen Publishing Limited (London) and Princeton University Press
(Princeton, NJ).
2. G. W. Fitch, “Are There Necessary A Posteriori Truths?” Philosophical Studies 30 (1976):
243–247. This is actually a slightly revised version of the argument. The revision consists in a mod-
ification of premise (3). In the original argument premise (3) was a substitution principle that gov-
erned the use of rigid designators. The principle is overly broad for what it was designed to do,
namely establish premise (3). Also I shall ignore in this section of the paper delicate issues con-
cerning the possibility of Venus’s failing to exist.
295
296 G. W. Fitch
(I) (1) The terms “Hesperus” and “Phosphorus” are rigid designators.
(2) The objects of knowledge are propositions.
(3) If a and b are co-extensive rigid names then Èa = b˘ expresses the
same proposition as does Èa = a˘.
(4) We know a priori that Hesperus is Hesperus.
Hence,
(5) We know a priori that Hesperus is Phosphorus.
Kripke’s argument for the falsity of (5),5 according to Soames, only shows that (4b)
is not known a priori and not that (4a) is not a priori. Soames says:
3. Scott Soames, Beyond Rigidity: The Unfinished Semantic Agenda of Naming and Neces-
sity (Oxford: Oxford University Press, 2002).
4. Nathan Salmon, Frege’s Puzzle (Cambridge, Mass.: MIT Press, 1986).
5. Saul Kripke, Naming and Necessity (Cambridge, Mass.: Harvard University Press, 1980).
Here is the argument that Soames is referring to: “The evidence I have before I know that Hes-
perus is Phosphorus is that I see a certain star or a certain heavenly body in the evening and call
it ‘Hesperus,’ and in the morning and call it ‘Phosphorus.’ I know these things. There certainly is
a possible world in which a man should have seen a certain star at a certain position in the evening
and called it ‘Hesperus’ and a certain star in the morning and called it ‘Phosphorus’; and should
have concluded—should have found out by empirical investigation—that he names two different
stars, or two different heavenly bodies. . . . So in that sense we can say that it might have turned
out either way. Not that it might have turned out either way as to Hesperus’s being Phosphorus.
Though for all we knew in advance, Hesperus wasn’t Phosphorus, that couldn’t have turned out
any other way, in a sense. But being in a situation where we have exactly the same evidence, qual-
itatively speaking, it could have turned out that Hesperus was not Phosphorus; that is, in a coun-
terfactual world in which ‘Hesperus’ and ‘Phosphorus’ were not used in the way that we used
them, as names of this planet, but as names of some other objects, one could have had qualita-
tively identical evidence and concluded that ‘Hesperus’ and ‘Phosphorus’ named different objects.
. . . So two things are true: first, that we do not know a priori that Hesperus is Phosphorus, and
are in no position to find out the answer except empirically, Second, this is so because we could
have evidence qualitatively indistinguishable from the evidence we have and determine the ref-
erence of the names by the positions of the two planets in the sky, without the planets being the
same” (pp. 103–104).
On Kripke and Statements 297
imagined world do not know the proposition they use the sentence Hespe-
rus = Phosphorus to express, for the simple reason that the proposition they
use the sentence to express is false in their world. But this does not show
that the different proposition we use the sentence to express isn’t known by
us; nor does it show that it isn’t known by us independent of empirical inves-
tigation. (Soames 2002, p. 8)
It is clear that Soames is assuming that the objects of knowledge are propositions.
Moreover, Soames and I both hold that the propositions semantically expressed
by the sentences “Hesperus is Hesperus” and “Hesperus is Phosphorus” are one
and the same. Additionally, Soames holds that the semantic content of a proper
name is just its referent and thus Soames agrees with (early) Russell that Hespe-
rus with all its clouds is a constituent of the proposition expressed by (4a). So the
problem for Kripke’s position seems relatively clear: How can Kripke distinguish
the claim that Hesperus is Phosphorus from the claim that Hesperus is Hesperus?
In a recent paper, Nathan Salmon argues that Kripke’s account of true singular
negative existentials is flawed.6 One example that Salmon focuses on is the nega-
tive existential (his numbering):
. . . the account fails to solve the problem. The “that” clauses “that Holmes1
plays the violin” and “that Holmes1 exists” are no less problematic than
“Holmes1” itself. Kripke concedes, in effect, that if a is a thoroughly nonre-
ferring name, then propositional terms like Èthe proposition that a is bald˘
are also thoroughly nonreferring. The account thus analyzes a negative exis-
tential by means of another negative existential, generating an infinite
regress with the same problem arising at each stage. (Salmon 1998, p. 297)9
So the problem for Kripke’s views is that if there is no proposition that Holmes
plays the violin or that Holmes exists (on the relevant use of ‘Holmes’), then the
claim that there is no true proposition that Holmes exists, as well as the claim that
8. For a more detailed account of Kripke’s view, see Salmon’s paper (1998).
9. Salmon’s objection is similar to David Braun’s objection to what he calls the Metapropo-
sitional View, which is in essence the view that Salmon attributes to Kripke (“Empty Names,”
Noûs 27 (1993): 449–469). Braun also objects, in this paper, to what he calls the No Proposition
View, which is a version of the view I suggested for empty names in “Non Denoting,” Philosoph-
ical Perspectives 7 (1993): 461–486.
On Kripke and Statements 299
according to the fiction Holmes plays the violin, cannot be true. They both contain
empty expressions: the clause that Holmes exists and the clause that Holmes plays
the violin. Yet this was supposed to be the way we understood the true reading of
(0) as well as the true reading of the claim that Holmes plays the violin.
The connection between the objections to Kripke’s account of true negative exis-
tentials and his claim that Hesperus is Phosphorus is a necessary a posteriori truth
is that both are based on the relationship between the claimed truths and the rel-
evant propositions (or lack thereof). If there is no such proposition expressed by
the sentence “Sherlock Holmes exists”, then there can be no true claim that there
is no true proposition that Sherlock Holmes exists. If the proposition expressed
by “Hesperus is Hesperus” is the same proposition as expressed by “Hesperus is
Phosphorus”, then, given that the proposition is knowable a priori, it is not the
case that one can know that Hesperus is Phosphorus only a posteriori. I shall focus
for the moment on necessary a posteriori truths, returning to true negative exis-
tentials shortly.
A careful reading of Naming and Necessity shows that in those lectures
Kripke only argues that there are statements that are necessary and knowable a
posteriori, not propositions. When Kripke’s lectures were published in book form
some ten years after they were given, Kripke was aware of the problem his account
of necessary a posteriori truths appeared to have and he commented on this
problem in the preface to the book version:
Clearly, Kripke is expressly refraining from adopting any view concerning propo-
sitions, although he does think that there is an important difference between the
English sentences in question. Soames is, of course, well aware of these comments
by Kripke, but he sees no way that Kripke can avoid commitment to propositions
given his claims that certain statements are knowable only a posteriori or his fre-
quent references to necessary truths. Soames says:
It is certainly true that faced with the considerations mentioned by Soames, Kripke
needs to say something about the nature of these truths that are necessary and
known only a posteriori. It is also true, I think, that when Kripke presented the
lectures (and even ten years later when the preface was written), he was unclear
as to the exact nature of these truths. Whether or not these truths are propositions
in the sense that Soames, Salmon, myself, and others accept was something that
Kripke was uncertain about.
Soames sees only two possibilities for these truths—sentences and proposi-
tions. Sentences don’t work since it is contingent that the sentence “Hesperus is
Phosphorus” expresses a truth as Kripke himself took pains to point out. As
Kripke says, “one could have had qualitatively identical evidence and concluded
that ‘Hesperus’ and ‘Phosphorus’ named different objects” (Kripke 1980, p. 104).
But propositions don’t work either for the reasons given in section 1.
One might question premise (4). Everyone seems to have assumed that
premise (4) is true. I say that the premise is “unexceptional” and Soames says that
the proposition that a = a is “surely knowable a priori.” But just how sure are we
that the proposition that Hesperus is Hesperus is knowable a priori? Most of those
who accept Kripke’s arguments against descriptivism hold that the proposition
that Hesperus is Hesperus is a Russellian proposition or what David Kaplan called
a “singular proposition”. Singular propositions have individuals as constituents
and hence the proposition that Hesperus is Hesperus contains Venus with all its
clouds as a constituent. How could anyone have knowledge of such a proposition
a priori? Knowledge of such a proposition would seem to require knowledge of
the existence of a contingently existing particular, namely Venus.
It is important to see here that this problem about singular propositions is
not the problem that Kripke’s distinction between “weak” and “strong” necessity
was designed to solve. One might reject the claim that the proposition that Hes-
perus is Hesperus is necessary on the grounds that its truth depends on the exis-
tence of Hesperus. However, as Kripke and others have noted, we can avoid this
problem simply by assuming the proposition in question is the conditional that if
Hesperus exists, then Hesperus is Hesperus. Such a proposition is true even when
considered under counterfactual circumstances where Hesperus fails to exist. Still,
this proposition is also a singular proposition involving Venus as a constituent. One
cannot have a priori knowledge of the existence of a particular whose existence is
only contingent.
But do we need to have a priori knowledge of the existence of Venus in order
to have a priori knowledge of the proposition that Hesperus is Hesperus? There
is no simple answer to this question even assuming that the proposition in ques-
On Kripke and Statements 301
tion is a singular proposition.10 What does seem clear is that what can be known
a priori is that the sentence “If Hesperus exists then Hesperus is Hesperus”
expresses a true proposition in our language. One wonders if the sentence/propo-
sition confusion that Soames points out with respect to (4a) and (4b) is not present
in the uncritical acceptance of (4). That is, one wonders if there is not an implicit
inference being used to justify (4) in the same way that occurs in the argument for
the falsity of (5), according to Soames. In both cases one argues from the fact that
it is known a priori/a posteriori that a certain sentence expresses a truth in our lan-
guage to the claim that the proposition so expressed is known a priori/a posteri-
ori. This is not an acceptable inference, however.11 Yet, it does seem obvious that
we know that Hesperus is Hesperus, if Hesperus exists. Indeed, I would suggest
that to the extent that it is obvious that we know a priori that Hesperus is Hes-
perus it is equally obvious that we know only a posteriori that Hesperus is Phos-
phorus. One can bite the bullet by rejecting one of the “obvious” claims, as Soames
does. Or maybe there is something to the idea that is only hinted at by Kripke that
we may be able to distinguish statements from both sentences and propositions
and thereby find some middle ground. First, we need to clarify our notion of a
sentence.
A sentence, as many philosophers use the phrase, is an uninterpreted string
of symbols. This accounts for the way Soames presents (4b). He remarks that a
certain sentence expresses a truth in our language. So the very same sentence might
express a falsehood in another language and he takes Kripke’s argument5 to be
an argument to that effect. There is a counterfactual situation where the expres-
sions “Hesperus” and “Phosphorus” are introduced that is qualitatively identical
to the actual situation, but the names are introduced to name distinct objects. In
such a situation the sentence “Hesperus is Phosphorus” would not only fail to
express a truth, but also it would not be a sentence of our idiolect either (because
one of the names fails to be a name of Venus). Whether or not it is a sentence of
English in the described counterfactual circumstances depends on the nature of a
natural language.
On one view of a natural language, English might have contained a term
“Phosphorus” for the planet Mars without containing a term “Phosphorus” for the
planet Venus. This view treats languages as analogous to persons and other objects
commonly thought to have both essential and accidental features. So, one could
hold that it is an accidental feature of English that “Phosphorus” designates Venus
and thus the English sentence “Hesperus is Phosphorus” might have expressed a
different proposition from the one it actually expresses as illustrated by Kripke’s
example.
10. Here is an argument that you cannot know a priori that x is x: If a is a completely empty
name (i.e., does not designate anything), could one know the proposition that if a exists then a =
a? Would there be any such proposition? It seems reasonable to hold that one needs to know that
there is an a in order to know any proposition that purports to be about a and of course such
knowledge cannot be had a priori if a is a contingently existing object.
11. This is a point emphasized by Keith Donnellan in his paper “The Contingent A Priori
and Rigid Designators” in P. French et al., eds., Contemporary Perspectives in the Philosophy of
Language (Minneapolis: University of Minnesota Press, 1979).
302 G. W. Fitch
(3*) If a and b are co-extensive rigid names then the proposition that the
statement that a = b is true is the same proposition as the proposition
that the statement a = a is true.
But it is not obvious that (3*) is true. On the conception of statements being con-
sidered the statement that Hesperus is Hesperus is not the statement that Hespe-
rus is Phosphorus, and hence the proposition that the statement that Hesperus is
Hesperus is true is not the proposition that the statement that Hesperus is Phos-
phorus is true.
But is it the case that the proposition that the statement that Hesperus is
Phosphorus is true is necessarily true? Could not the statement have been false?
Of course, the sentence “Hesperus is Phosphorus” might have been false, had it
expressed a proposition different from the one it did express. It might have
expressed the proposition that Venus is Mars and hence expressed a false propo-
sition. But the statement that Hesperus is Phosphorus cannot be used to assert the
proposition that Venus is Mars. The names “Hesperus” and “Phosphorus” are fixed
in our use to designate Venus.12 In order for the statement that Hesperus is Phos-
phorus to be false, it must be used in a given context to assert a false proposition.
But the only proposition that can be semantically asserted with its use is the propo-
sition that Venus is Venus. There is considerably more that needs to be said about
the nature of statements if the defense of Kripke’s position is to have any hope of
succeeding. Before we become involved in the details, however, it is helpful to see
how the introduction of statements can avoid the objections to Kripke’s position
with respect to true negative existentials.
Recall that the problem for Kripke’s view with respect to (0) (Sherlock
Holmes does not exist) is that (0) does not express a proposition and hence under-
standing (0) as
is unhelpful because (0*) itself does not express a proposition (and hence cannot
be true). Thus, there is no true reading of (0) along the lines suggested by Kripke.
This objection is based on the view that the clause “that Holmes exists” in (0*)
purports to designate (express) a Russellian proposition. (0*), however, can be
read in a different way. We can understand (0*) as asserting that no true proposi-
tion can be asserted by using the statement that Holmes exists. The distinctions
made above with respect to the uses of the name “Sherlock Holmes” need to
honored here as well. There are, if Kripke, etc. are correct, many interpretations
of the sentence (0) even within our conventional language use. So there is no single
statement expressed by (0*) in English but rather a number of different statements
including there is no true proposition asserted by the statement that fictional
12. There are some complications here. The uninterpreted “Hesperus” may very well have
many designations even in our language (as does the term “John Smith”). I believe this is a result
of ambiguity (as opposed to indexicality), but I realize that not everyone holds this view. I shall
simply ignore this complication and assume that the terms in question have single designations
in my language.
304 G. W. Fitch
13. David Kaplan, “Demonstratives,” in J. Almog, J. Perry, and H. Wettstein, eds., Themes
from Kaplan (Oxford: Oxford University Press, 1989), pp. 481–563.
306 G. W. Fitch
One might be tempted to suppose that there is some more general thing that
the sentence “says” in every context—namely, the proposition expressed by
the speaker is F (or some such thing). But this will not do. Our notion of
“what a sentence says” is tied to what speakers who assertively utter the sen-
tence say. (Soames 2002, p. 104)
While I am not as convinced as Soames that the notion of “what a sentence says”
is as precise as he seems to think,14 I agree with him that in uttering the sentence
“I am a philosopher” one does not thereby assert the proposition that the speaker
or agent of the context is a philosopher. If what a sentence says is what one asserts
using that sentence (with a given interpretation in a given context), then what a
sentences says is not the statement expressed by that sentence. But statements are
not intended to replace propositions as what we assert when we assertively utter
a sentence. Statements are intended to fill a “meaning” gap between sentences
viewed as complex syntactical items and what we assert when we utter such items.
There is a meaning that remains constant across contexts when English speakers
(speaking in English) utter “I am a philosopher” that is not captured by the propo-
sition asserted by the speakers in the various contexts.
Even if one agrees that certain sentences have a kind of meaning that is not
captured by the propositions asserted using those sentences, questions remain
about the semantic significance of this kind of meaning and whether sentences
containing proper names have a linguistic meaning distinct from the content of a
name. Because a full discussion of the role of the linguistic meaning in semantic
theory will take us far afield, I shall leave that issue for another time and focus
instead on the question of whether proper names have a linguistic meaning dis-
tinct from their referent (if there is one).
Unlike indexicals, it is often thought that the only kind of meaning that a
proper name has is the semantic content of the name and that, for Millians, is
simply the referent of the name. This is a cornerstone of the objections offered by
Soames and Salmon to Kripke’s position. So, even if one can distinguish the state-
ments that [pointing to an object in the sky] is Hesperus and Hesperus is Hesperus
on the grounds that they differ in linguistic meaning (character), a similar claim
cannot be made with respect to the statements Hesperus is Hesperus and Hespe-
rus is Phosphorus. The latter two statements have the same linguistic meaning
(character) and hence statements conceived of as the meanings of sentences
cannot solve our problems. On Kaplan’s formal definitions of character and
content, presented in his paper “Demonstratives,” the two sentences (“Hesperus
is Hesperus” and “Hesperus is Phosphorus”) do have the same character and the
14. See David Lewis’s discussion of what a sentence says in his paper “Index, Context and
Content” in Papers in Philosophical Logic (Cambridge: Cambridge University Press, 1998),
pp. 21–44.
On Kripke and Statements 307
same content. In more recent work, however, Kaplan has come to doubt that they
even share a content let alone that they have the same linguistic meaning.15
Because I shall claim that names do have linguistic meaning, I want to separate
linguistic meaning from Kaplan’s formal notion of character (while retaining, I
hope, the intuitive idea behind Kaplan’s notion of character).
If the linguistic meaning of an indexical like “I” is related to the rule asso-
ciated with its use, then perhaps the linguistic meaning of a name is related to the
rule associated with its use. The meaning of an indexical is given by the rule that
determines the referent of an indexical in a given use. So the natural place to look
for the meaning of a name is at the rule that determines the referent of the name
in a given use.16 If Kripke and others are correct, the referent of a given use of
a name is determined via a chain of reference-dependent uses that leads to a
baptismal ceremony at which the name is introduced. It is a chain of reference-
dependent uses because a given use of a name is based on one’s learning the name
and when one learns a name one intends to use it to refer to what the person
from whom one learned the name intended to refer. All who use an established
name assume that at some time and in some way there was a baptismal ceremony
where the name being used was introduced even if the details of that ceremony
are not (usually) known. This suggests that the meaning of a given name is given
by the rule that fixes the referent of a given use of the name via the chain of ref-
erence-dependent uses to the name’s introduction or the baptismal ceremony.
Because the rule that fixes the reference of uses of distinct names leads to distinct
introductions or baptisms, part of the meaning of a name is fixed by the details of
its introduction even though one need not know the details in order to be a com-
petent user of the name. This is similar to the fact that one need not be an expert
on trees to use the terms “beech” and “elm” nor does one need to know the details
of the rule governing the use of “you” in order to be competent user of that expres-
sion. One can be competent in using expressions without knowing in detail the
rules that give the linguistic meaning of those expressions. One need only have a
general sense of the rules in the sense that one can distinguish appropriate from
inappropriate use.
Kripke himself suggests that the sentences “Hesperus is Hesperus” and
“Hesperus is Phosphorus” differ in meaning in some way since he holds them not
to be interchangeable. In fact, he suggests that the “mode of fixing the reference”
15. In his paper “Words” (Proceedings of the Aristotelian Society [Supp.] 64:93–117) Kaplan
says the following in a footnote (# 6, p. 95):
I have come to think that two sentences whose syntax—perhaps I should say, whose
logical syntax—differs as much as “a = a” differs from “a = b” should never be regarded as
having the same semantic value (expressing the same proposition), regardless of the seman-
tic values of the individual lexical items “a” and “b.”
In this quote Kaplan uses the phrase “semantic value” to mean semantic content in the sense used
in the text. What I am suggesting is that such sentences do differ in semantic value (linguistic
meaning), even though they are used to assert the same proposition (content).
16. The names I am discussing are all simple proper names. The situation is somewhat more
complex if we include descriptional names such as “the Willis Avenue Bridge”. I ignore this com-
plexity in what follows.
308 G. W. Fitch
accounts for the difference in our “epistemic attitude toward the sentences
expressed.” I suggest that the rule that fixes the reference of a given use of a name
can provide us with the linguistic meaning of the name and the rule involves the
baptismal ceremony. Hence names like “Hesperus” and “Phosphorus” do differ in
linguistic meaning as a result of distinct introductions. The same can be said of
“Sherlock Holmes” and other fictional names. They also have linguistic meaning
and that meaning is determined by how they are introduced. In the case of fic-
tional names there is no referent, but there remains the baptismal ceremony. Sen-
tences containing the nonreferring use of these names still have a meaning and
hence they can express statements even though they cannot be used to assert
propositions. So on one reading of Kripke’s claim that there is no true proposition
that Sherlock Holmes exists, Kripke is asserting a proposition about a statement.
What he is asserting is that the statement that Sherlock Holmes exists cannot be
used to assert a true proposition.
If, as I have suggested, proper names do differ in linguistic meaning, then the
objections that Salmon, Soames, and others have raised against Kripke’s examples
of necessary a posteriori truths and his account of true negative existentials can
be avoided. They can be avoided but not without some cost. The cost is that what
seemed to be a sentence about an object turns out to be a sentence about a state-
ment. In the case of negative existentials, the price does not seem (to me) too high
to pay. If we want to truly say that Sherlock Holmes does not exist, there can be
no object that our statement is about. That is part of the point of making the state-
ment in the first place.
The interpretation of the sentence “Hesperus is Hesperus is an a priori truth”
might seem to some a bit more expensive. The view I suggest holds that this sen-
tence be understood to express the proposition that the statement that Hesperus
is Hesperus always expresses a true proposition (in its contexts of use) is know-
able a priori. Thus, the proposition that is claimed to be known a priori is one
about a statement about Venus and not simply the proposition about Venus itself.
Yet even here one should be uneasy with the alternative view that a Russellian
singular proposition involving a contingent object is something that is knowable
a priori. Such knowledge appears to involve a priori knowledge of the existence
of a contingent object. How could that be? So even in these cases I do not find
the cost too high.17
17. I would like to thank Ted Guleserian, Tom Blackson, Peter French, and Stew Cohen for
helpful discussion and comments on an earlier draft of this paper.