Maverick Philosopher

The question to be answered, in a rough formulation, is this: What secures (grounds, establishes, accounts for) the peculiar unity of subpropositional constituents whereby they form a unified whole capable of having a truth-value?

At Sophist 261d ff., Plato raises the unity question and puts into the mouth of the Eleatic Stranger the suggestion that there is a fundamental difference between names and verbs, onomata and rhemata, and that "a statement never consists of names spoken in succession, nor yet of verbs apart from names." (262a) If I say, 'lion, stag, horse,' I do not express a proposition. Ditto if I say, 'runs, sleeps, walks.' But if I say, 'Socrates runs,' then I express a proposition or make a statement. So a statement is a complex in which there is a "weaving together [of] verbs with names." (262d)

The idea suggested by the Sophist passage is that if one thinks of the parts of a (declarative) sentence as all of them names, then there is no way that the parts can form a truth-valued unity. On this approach, the solution to the unity puzzle consists in realizing that some parts of speech are not names, and that they fit together with names to form unities of sense capable of being either true or false. Names name, while verbs describe; the roles are distinct but complementary so that, when a name is concatenated with a verb, the result is a propositional unity. Accordingly, those who find a problem here are simply laboring under a false assumption, namely, that all the parts of a sentence are names.

In Fregean terms, the false assumption is that all expressions are complete, when in truth some are essentially incomplete or gappy. Thus, in 'Socrates is white,' 'Socrates' is a name, and thus a complete expression, whose referent is an object (Gegenstand), whereas '___is white' is not a name but an essentially predicative expression whose referent is not an object but a concept (Begriff). On this approach, there is no need and no room for a copula. On the side of language/mind, there is a name and a predicable and they fit together without the need of a copulative tertium quid to connect them. It is as if the predicable has a gap or slot in it in which the name fits. The name completes or 'saturates' the essentially incomplete predicable expression. On the side of the world, there is an object and a concept, with the object falling under the concept, but without the need for an instantiation or exemplification relation to connect them. Since the copula 'is' on this approach is not a name, it cannot be taken to name the relation of instantiation.

This approach seems to stop Bradley's regress before it can start. If subject and predicate fit together like a plug and socket, then there is no need for a third item to connect them; and without this third item, the regress can't get started. You will recall that part of the unity problem is to secure unity without igniting a vicious infinite regress.

First Argument Against the Platonic-Fregean Option

This approach, though very attractive at first, issues in the notorious 'horse' paradox. One of Frege's fundamental ideas is that there is an unbridgeable ontological chasm between objects and concepts, names and predicables. Objects can be named, but cannot be predicated; concepts can be predicated, but cannot be named. Objects are complete or 'saturated' entities; concepts are incomplete or 'unsaturated. Thus objects and concepts complement one another, and because of this, Frege thinks that their unification so as to form a truth-valued whole is not a problem as it would be if all of the parts of a sentence were names. For if all the parts of a declarative sentence were names, both the senses and the referents of these names would be objects and a tertium quid would be needed to connect them. But if a third thing such as a relation of instantiation is brought in to connect subject-constituent and predicate-constituent, then Bradley's regress is up and running. For if I connects S and P, what connects I, S, and P? A triadic relation? And so on.

The idea, then, is that the unity problem is solved, and Bradley's regress is avoided, if the essentially predicative nature of concepts is appreciated. This implies that the referent of '___is white' is essentially gappy, and essentially unnameable. For if I were to try to name the referent of this predicable expression, using perhaps the abstract substantive 'whiteness,' or 'the concept whiteness,' the thing named would be an object, not a concept. Yet it seems clear that we do need to talk about, and quantify over, concepts. Thus there will be occasions on which we will want to say that F and G are equinumerate, or that G is included in F. So if Frege's theory implies that we cannot talk about, and make predications of, concepts, then this will amount to a serious defect in the theory.

This bring us to the 'horse' paradox: The concept horse is not a concept. For 'the concept horse,' which is in subject position functions as a name, and what names denote are objects; but no object is a concept. Hence the concept horse is not a concept. But obviously the concept horse is a concept.

This contradiction is a reason to reject the Fregean solution to the unity problem. In a nutshell, the object/concept dichotomy helps with the unity problem, but gives rise to the 'horse' paradox.

They are not the same because (1) has a property that (2) does not have, namely, the property of being truth-valued. If we assume Bivalence, then the property of being truth-valued is the same as the property of being either true, or if not true, then false. (1) is truth-valued, but none of its parts are truth-valued. Furthermore, the list of these parts, which is a mere collection of them, is not truth-valued.

The question at the level of philosophical theory about what constitutes the unity of a sentence/proposition presupposes a datum, namely, that the unity of a proposition is a peculiar unity that cannot be understood in terms of the 'unity' of a mere collection, whether this be an ordered or unordered list or an ordered or unordered set, or a mereological sum. Unfortunately, David Brightly denies what I take to be a datum, a given, a starting point for theorizing:

You haven't really given us a reason for thinking that understanding 'Socrates is white' consists in any more than having the meanings of 'Socrates', 'is', and 'white' simultaneously in mind, which is about as far as my own introspection takes me.

But surely one could have those three meanings before one's mind without having before one's mind the further thought that 'is' connects the other two. To understand 'is' by itself is just to understand the general sense of the copula. But to understand 'Socrates is white' is to understand not merely the general sense of the copulative 'is' but also the copulative 'is' as actually connecting the other two items. There is a distinction between the copulative 'is' and the copulative 'is' as actually copulating!

I am taking for granted the standard distinctions among the various senses of 'is.' There is the existential 'is,' the identitative 'is,' the copulative 'is' and perhaps the veritative 'is.' The topic at present is the copulative 'is.'

Brightly counters:

I think it would be very difficult to have the meaning of 'is' before the mind without the thought that it acts as a connective. Does 'is' have any meaning, apart from conveying tense, beyond its connecting role? In Russian, for example, in the first and second person present tense it is omitted altogether and the English pseudo-sentence 'Socrates white now' conveys the sense equally well.

It is certainly true that one cannot understand the copulative 'is' without understanding it as an actual or potential copula. But the point is that one can understand the copulative 'is' without understanding it as actually connecting two specified terms. Here is an analogy. I can easily understand that a certain cable, which I see for sale in an electronics store, is used to connect digital cameras to computers. This understanding does not require that the cable actually be in use as connecting a specified computer to a specified digital camera. And if the cable were in use connectng camera C to computer K, I could easily understand how that same cable could be used to connect some other camera to some other computer. Furthermore, supposing that every camera and every computer in the universe to be destroyed, but the cable in question to remain in existence, that cable would not cease to be the cable it is, and have the function it has, in these drastically altered circumstances.

The analogy, then, is this. The cable, hanging on a rack in an electronics store, and not connecting anything to anything, is to the copulative 'is' in general as the cable actually connecting a specified camera to a specified computer is to the copulative 'is' actually connecting a subject term and a predicate term in a sentence. Just as I can understand the function of a cable without seeing it in actual use connecting one device to another, I can understand the copulative 'is' without understanding it as connecting two specified terms.

Therefore, when I bring before my mind the mere list of a sentence's parts, I do not thereby bring before my mind the truth-valued unity of those parts. The sentence is more than the sum of its parts. It is a unity of those parts which makes of them an item that can have a truth-value.

Brightly rightly points out that the grammatical copula is dispensable. Although it is not correct English to say 'Socrates white,' one could say that and manage to convey a complete thought to one's audience. So the grammatical copula 'is' is dispensable, at least in some cases. In Turkish, both of the following are correct:

3. Deniz mavi. (The sea blue)

4. Deniz mavidir. (The sea is blue).

What examples like this show is that the grammatical copula is eliminable. (The Turkish example also shows that the grammatical copula need not be an independent or stand-alone word; it can be a suffix.) But these examples do not show that the logical copula is eliminable. Consider a variant of written English in which the grammatical copula is replaced by spatial juxtaposition in a certain order. In such a variant, there would still be a logical copula; it is just that it would be expressed, not by a written sign, but by spatial juxtaposition in a certain order.

It is essential to distinguish grammatical from logical copulae. Appreciating the distinction, we see that the dispensability of the former, assuming it is dispensable, does not show the dispensability of the latter. It is obviously the logical copula with which we are concerned in the question of the unity of the declarative sentence/proposition.

To sum up. What I called a datum above really is a datum. An atomic sentence is not a mere list of its subsentential parts. A sentence has a truth-value, and in the mouth of a speaker, says something about the world. But the individual words, whether taken singly or collectively, do not have a truth-value, nor do they say anything about the world. The 'is' in the sentence 'Socrates is white' does not occur in that sentence in the way 'is' occurs in the corresponding mere list. In the list, 'is' occurs as an isolated element, whereas in the sentence 'is' occurs both as an element and as that which actually connects subject and predicate. It occurs not as a mere potential copula, but as an actually copulating copula.

But this is all preliminary. The real fun starts when we try to work out a theory of what this peculiar unity consists in.

Much depends, of course, on just what is meant by 'proposition.' Clarity will be served if we catalog some of the main uses of the term.

1. Geachian Propositions. A Geachian proposition is

. . . a form of words in which something is propounded, put forward for consideration; it is surely clear that what is then put forward neither is ipso facto asserted nor gets altered in content by being asserted. (Logic Matters, 255)

Geach distinguishes propositions in his sense from what we may call

2. Abstract Propositions. These are the contents of tokenings of indicative sentences. Suppose I assertively utter 'Two times two is four' and Karl assertively utters 'Zwei mal zwei ist vier.' Clearly, our utterances have the same content; we express the same thought. If you deny that, then I say you deny a datum. Now one terminological issue is whether we should refer to this common content as a proposition, or as a propositional content. I prefer to say that the the tokenings of the English and the German sentence express the same proposition.

Thus I distinguish among sentence types, tokens of sentence types, and propositions. Sentence types and tokens are tied to particular languages whereas propositions are not so tied and are therefore in this sense 'extralinguistic.' The arithmetical thought expressed by Karl and me is neither English nor German, even though the sentences in which we 'couch' the thought are German and English, respectively.

3. Commenter 'Ockham' suggests that in one sense, "a 'proposition' is a sentence that (impersonally) makes an assertion."

4. 'Ockham' suggests in a second sense that a proposition is "what is asserted by a sentence."

I reject both (3) and (4), and for the same reason: Sentences don't make assertions, people makes assertions using sentences. (Analogy: Guns don't kill people, people (sometimes) kill people (sometimes) using guns.) To say that a sentence impersonally makes an assertion is to say that a sentence in the indicative mood is typically used to make an assertion; but that is not to say that the sentence makes an assertion.

Now the issue before us is whether one and the same proposition can be asserted in some contexts and left unasserted in others. If we adopt the approach of either (3) or (4), then of course there cannot be unasserted propositions. But if we adopt (1) or (2), then there can be unasserted propositions. Since I see now reason to accept (3) or (4), I agree with Geach about the Frege point.

Why is this important? One reason is that it bears on the question of existence. The issue is whether the existence of an individual belongs to it intrinsically, and thus is in some sense predicable of it, or whether existence in no way belongs intrinsically to individuals but is rather bound up with acts of judgment as Hume and Brentano maintained. At this point I refer the interested reader to the preceding post in this chain.

So on the pragmatic side we have the distinctions among speech acts, and on the semantic side, the distinctions among moods. One question that arises is whether the speech acts map neatly onto the moods. When I make an assertion, must I use an indicative sentence? Or can I make an assertion using a non-indicative sentence? And can I utter an indicative sentence and not make an assertion? Here are some theses that seem true.

T1. One can make an assertion using a non-indicative sentence.

T2. One can utter an indicative sentence and not make an assertion.

T3. One can utter an interrogative sentence and be taken by one's audience to be making an assertion.

T4. One can utter an imperative sentence and thereby make an assertion.

T5. One can utter an optative sentence and thereby ask a question.

Ad (T1). Suppose A sincerely asks, 'Does God exist?' and B replies, 'Is there an angry unicorn on the far side of the moon?' It seems that B has answered A's question, and has done so by making an assertion, an assertion more straightforwardly put by an utterance of the indicative sentence, 'God does not exist.' And yet B has made his assertion by uttering an interrogative sentence. This appears to be evidence that one can make an assertion by using a non-indicative sentence. One can make an assertion without using an assertoric sentence.

Ad (T2). Suppose Johnny is picking his nose in public, and Mommy says to Johnny, 'We don't do that.' Mommy utters an indicative sentence, and yet does not make an assertion; she issues a command.

Ad (T3). I bought a hat recently at a swap meet for $20 and I asked the lady who sold it to me, 'Do you know how much this hat would cost at a retail outfitter's?' Her reply: 'How much?' The lady took me to be making an assertion that would normally be couched in some such indicative sentence as, 'I know how much this hat would cost elsewhere.' And yet, I was asking a genuine question. (I need to write a separate post on rhetorical questions.)

Ad (T4). Davidson gives this example: 'Notice that Joan is wearing her purple hat again.' The sentence is in the imperative mood, but is used to make an assertion. An example of my own. I say something you disagree with, and you reply, 'Get out of here.' You are not commanding me to leave, but denying my assertion with a counter-assertion.

Ad (T5). Another example from Davidson: 'I'd like to know your phone number.' This grammatically optative sentence might be used to ask a question that would be put more directly as 'What is your phone number?'

This linguistic data seems to support the view that we must distinguish between speaker's meaning and sentence meaning, which is a species of linguistic meaning. Sometimes they coincide, but often they come apart, as in the above examples. What a speaker means or intends by his speech act often diverges from what sentences mean when considered in abstraction from speakers and their intentions.

Now does commenter 'Ockham' deny that we need to make this distinction?

REFERENCE: Donald Davidson, "Moods and Performances" in Inquiries into Truth and Interpretation, Oxford, 1984, pp. 109-123.

My asserting that Dan is drunk is a speech act. When I make an assertion, I do at least two things: I commit myself to the truth of what I assert, and I communicate the content of my assertion to a hearer. Let me now propose a thesis that seems true but which we might want to think about:

T1. To assert that p is to assert that p is true.

Be that as it may, suppose I assert that Dan is drunk. I do this by tokening the sentence type 'Dan is drunk.' The assertoric force of the tokened sentence type is indicated by 'is' which signals the indicative mood among other things. Now here is my question: Are indicative mood and assertoric force the same, or different? When I refer to the assertoric force of a sentence, am I referring to its indicativity, and vice versa? Or must we distinguish between indicativity and assertoric force?

Consider this exchange between speakers A and B:

A: Peter is innocent.
B: (Ironically) Yeah, right. Peter is innocent.

Although both A and B are tokening 'Peter is innocent,' only A is asserting that Peter is innocent. Now the sentence type considered by itself is in the indicative mood. So I am tempted to say that assertoric force, as a semantic component, is identical to indicativity. Both tokens have assertoric force despite the fact that only one is asserted. Now consider this example:

1. If Peter is innocent, then his conviction is unjust.

To assert a conditional is not to assert its antecedent. The antecedent, 'Peter is innocent,' is in the indicative mood. I want to say that it has assertoric force despite the fact that it is not being asserted.

What is assertoric force? I suggest that it is that property of a sentence that renders it capable of being used by an agent to say something either true or false. As far as I can see, it is the same as the property of indicativity. Sentences with assertoric force can be used without being asserted, but no sentence lacking assertoric force can be asserted. To say that a sentence is assertoric, or has assertoric force, is to say that it is of the appropriate form for the making of assertions.

Since assertoric force either is or is equivalent to the property of being either true or false, a sentence's being assertoric does not entail its being true. Nor, of course, does a sentence's being asserted by someone entail its being true. To assert a sentence is to assert it as being true, but that is not to say that an asserted sentence is true. Whatever truth is, it involves a relation to something external to sentences or propositions.

Clearly, each of these sentences can be used to express many different thoughts or propositions. If Jack says 'I love you' to Jill, the proposition expressed is different from the proposition expressed if Bill says 'I love you' to Hill. Since one and the same sentence type can be used to express different propositions, it follows that sentence types are distinct from propositions.

We must also distinguish between a sentence type and its linguistic meaning, the meaning it has in virtue of the conventions of the language to which the sentence type belongs. The four sentences displayed above have the same meaning. Since one and the same meaning is possessed by these four different sentence types, it follows that linguistic meanings are distinct from sentence types.

It follows from the two points just made that linguistic meanings are distinct from propositions. One proof of this is that one can have a complete understanding of the linguistic meaning of a sentence without knowing any proposition that the sentence has ever expressed. Suppose a Spanish speaker learning English learns that 'Mary loves Carl' means the same as 'Mary ama a Carl.' The Spanish speaker then fully understands the linguistic meaning of 'Mary loves Carl' but without needing to know any proposition, any truth or falsehood, that the English sentence has ever expressed. (See Castaneda, Thinking and Doing, p. 35) Therefore, the linguistic meaning of a declarative sentence is distinct from the proposition expressed by the sentence on some occasion of the sentence's use.

'Ockham,' blinded by his nominalist fear of reification, cannot admit this obvious distinction between linguistic meaning and proposition. This causes him to give bad arguments. For example, he writes, "In summary, the meaning of a sentence is what it says, what it says is true or false, ergo the meaning of a sentence is a 'truth bearer'." The argument is this:

1. The meaning of a sentence is what it says
2. What a sentence says is either true or false
-----
3. The meaning of a sentence is either true or false.

The argument equivocates on 'what it says.' If premise (2) is true, then what a declarative sentence says is identical to the proposition it expresses. It is important to realize that I am not assuming any particular theory of propositions. Thus I am not assuming that they are Platonic entities. I am simply insisting that we need to distinguish between the linguistic meaning of a sentence (the meaning it has in virtue of the conventions of the language to which it belongs) and the thought a sentence expresses when the sentence is uttered or otherwise tokened by a person in a definite situation. But in premise (1), the linguistic meaning of a sentence is identified with what it says. Thus 'what it says' is being used in two different ways, which fact destroys the validity of the argument.

If 'Ockham' says I am begging the question againat him, I reply that he is failing to admit an obvious distinction. The distinction is not original with me. It ought to be visible to anyone. If an a priori commitment to nominalism blinds one to so obvious a distinction, then so much the worse for an a priori commitment to nominalism.

A. Sentence-Types

Count the sentences in the following display:

The sea is blue
The sea is blue
The sea is blue.

'Three' and 'one' are both reasonable answers, but they can't both be right on pain of contradiction. To avoid contradictions we draw distinctions. In this instance we reach for the tried-and-true distinction between type and token and say that we have three numerically distinct tokens of one and the same sentence type. Now consider a different display:

The sea is blue
Deniz mavidir
Deniz mavidir
Die See ist blau
Die See ist blau
Die See ist blau.

In this display we have six sentence tokens of three sentence types belonging to English, Turkish, and German respectively. But all six tokens have the same linguistic meaning. They all mean what we English speakers mean when we think or say or write that the sea is blue.

Now my first point is that sentence types cannot be the vehicles of the truth-values. For example, the sentence-type 'The sea is blue' is neither true nor false. Which sea, and when? For a sentence-type to be true or false, all of its tokens would have to be true or false. But some tokens of 'The sea is blue' are true and some tokens are false. If Ali of Samsun looks at the Black Sea on a moonless night and utters the words 'Deniz mavidir,' he says something false. Or consider, 'He owes me ten dollars.' This is neither true nor false since it is indeterminate who 'he' and 'me' refer to, and which time the present tense of 'owes' pick out.

B. Meanings of Sentence Types

My second point is that the linguistic meanings of sentence types cannot be the vehicles of the truth-values. 'He is dead to me' and Er ist mir tot have the same linguistic meaning, but this meaning is neither true nor false for the same reason that the sentence type 'He is dead to me' is neither true nor false. No truth-value can be assigned until the reference of 'he' is fixed.

C. Sentence Tokens as Physical Phenomena

Now let's consider a token of a sentence type, a token of 'The moon is full.' Glancing up at the night sky, I say, 'The moon is full.' By uttering these words I produce a sentence token. Or I might write the words down on a piece of paper. Either way, a complex physical particular comes into existence, either a collection of sounds or a collection of marks on paper. If what we mean by a sentence token is a purely physical phenomenon, then sentence tokens cannot be true or false. For physical phenomena, considered by themselves, are devoid of meaning, amdeaning is a necessary condition of having a truth-value.

For a sentence token to have a truth-value, it cannot be a 'dead' set of marks or acoustic disturbances; it must be marks or sounds 'animated' by a mind who intends the marks or sounds to convey a meaning. And this suggests that the primary truth-bearers are items on the order of judgmental contents. But more on this later.

Let's think about the English word, not the letter, 'I.' This one-letter word is the first-person singular pronoun. It is the paradigm case of an indexical expression. The most obvious thing to say about 'I' is that its reference is context-sensitive. Tokened by BV, it refers to BV; tokened by AR it refers to AR. (An uttering is a tokening, but there are tokenings that are not utterings: there are inscribings and other encodings.) But

1. Proper names are also context-sensitive
2. Proper names are not indexicals
So
3. What makes 'I' an indexical cannot just be its context-sensitivity.

Ad (1). Since many rejoice under the name 'Phil Spector,' this expression type does not count as a genuine proper name; it is rather a use or tokening of it that does. Thus my use of 'Phil Spector' in the sentence 'Phil Spector is standing trial for murder' picks out exactly one man, the record producer famed for his 'Wall of Sound.'

Ad (2). If proper names are Millian tags whose meaning is exhausted by their reference, then a proper name is not an indexical for the reason that indexicals have a linguistic meaning, a Kaplanian character, while proper names do not. Thus 'I' has a meaning that every speaker of English understands, a linguistic meaning that can be captured in some such rule as 'I' refers to the speaker or writer. With respect to 'I' and other indexicals one can distinguish between character and content in Kaplan's jargon, but one cannot make this distinction with respect to Millian tags since their meaning is exhausted by their reference.

If, on the other hand, it is maintained that proper names express Fregean senses through which their reference is routed, then too proper names are not indexicals since there are no such senses in the case of idexicals. Indexicals are paradigms of direct reference. This is not to say that indexicals lack lingusitic meaning. Clealry, 'now' and 'here' differ in lingusitic meaning (Kaplanian character) in that 'now' picks out times whereas 'here' picks out places. The point is that indexicals refer directly.

When I say, 'I am happy,' I refer to myself but not via a special Fregean sense that my use of 'I' expresses. For if there were such a sense, what would it be? It would have to something like the property of being me, a haecceity property that I alone have and I alone could have. This property, however, appears ineffable, and suspect for this reason. Given that 'I' refers, it refers directly.

Now let's consider whether indexical expressions are token-reflexive expressions. For Hans Reichenbach, the word 'I' means the same as 'the person who utters this token'; 'now' means the same as 'the time at which this token was uttered,' etc.

In the case of 'I' it seems false that 'I' means the same as 'the person who utters this token.' Although it is true that I am the person who utters this token, if the synonymy holds, then

4. I exist

is synonymous with

5. The person who utters this token of 'I' exists.

But they cannot be synonymous since they have different truth conditions. There are possible worlds in which (4) is true, but (5) false. Surely my existence cannot depend on my uttering tokens of 'I.' (Cf. David Kaplan, Themes from Kaplan, pp. 519-520.)

To sum up. In response to Rhoda, I say first that indexicals should not be assimilated to token-reflexives, and second, that not every referring device is context-sensitive, let alone indexical.

But there is another issue that may be at the back of Rhoda's mind. It is arguable that linguistic reference is ultimately grounded in thinking reference, first-person reference. If so, it is always a subject who refers using linguistic devices; words and phrases do not refer on their own. One cannot do semantics from a third-person point of view, abstracting from the intentionality of mind. So it may be that all reference is ultimately indexed to an underlying transcendental ego.

It is plain that Columbus could not have used 'Castro's island' to refer to his landfall. The reason is not that he didn't speak English, for we could easily translate (1) into Italian and the ambiguity of (1) would remain in the translation. The reason is that Cuba had to wait until 1959 to become Castro's island whereas Columbus sailed the ocean blue in 1492. So if we take (1) to impute to Columbus a reference to Cuba under the description 'Castro's island,' then (1) is false. Nevertheless, there is a way of construing (1) so that it comes out true. One must distinguish between an external and an internal occurrence of a word or phrase and then take the occurrence of 'Castro's island' in (1) as an external occurrence.

'Castro's island' occurs externally in (1) in that it is an expression that the speaker (or writer or thinker) of (1) uses in his attribution of a belief to Columbus, but not one that Columbus could have used. To appreciate the contrast consider this example of mine:

2. JFK believed that Castro's island harbored Soviet missiles.

In (2), 'Castro's island' occurs internally in the sense that it is an expression that is used by the speaker of (2), and could also have been used by JFK, to refer to Cuba. JFK could have referred to Cuba as Castro's island whereas Columbus could not have. JFK could have had before his mind a Fregean proposition the subject constituent of which was the Fregean sense expressed by 'Castro's island.' But not so in the case of the Italian mariner.

So 'Castro's island' occurs externally in (1) but internally in (2). Now what I want to suggest is that an external occurrence of an expression in a belief context is a de re occurrence of that expression.

Since (1) is open to two readings, one true and the other false, there is pressure to disambiguate (1) by adopting a notation that prevents the ambiguity from arising. So consider

1E. Castro's island was such that Columbus believed it to be China.

What (1E) does is withdraw 'Castro's island' from the scope of a verb of propositional attitude. One virtue of (1E) is that it removes any hint that Columbus thought of his landfall as Castro's island. The speaker of (1E) refers to Cuba as Castro's island but does not impute to Columbus any reference to Cuba under this description or via the sense (Sinn) of 'Castro's island.' More simply,

1E*. Castro's island was believed by Columbus to be China.

Ignoring the past tense, we can can perhaps put it this way:

1E**. BEL (Columbus, Castro's island, being China)

where 'BEL' is a triadic belief relation, and not a dyadic one as it would be if belief were thought to connect a believer to a Fregean proposition. In this 'ontological assay' of (1E) no Fregean proposition occurs. All we have is two individuals and a property related by a three-term relation. The whole complex is a Russellian proposition.

Note that one can correctly apply Existential Generalization to (1E) and its variants. Thus one can validly infer

3. (Ex)(x was believed by Columbus to be China.)

One can also substitute coreferential terms for 'Castro's island' salva veritate. Thus from (1E*) one can infer

4. Cuba was believed by Columbus to be China.

Now if you understand the distinction between external and internal occurrences of an expression, as explained above, then it seems to me that you have all you need to understand the de re/de dicto distinction as applied to beliefs. An expression that occurs externally occurs de re, and one that occurs internally occurs de dicto.

I am not saying that some beliefs are de re and some are de dicto. I am saying that some occurrences of epxressions in belief sentences are de re occurrences and some are de dicto. I am also not saying that this is what most philosophers mean by the distinction. I am saying that this is one way of attaching sense to the distinction.

I am not committed to (1E**) as the correct ontological assay of (1E*). The notion that one can analyze belief without invoking propositions or proposition-like entities strikes me as highly dubious. But this is really a separate topic.

Now, do you see that this septad is pregnant with contradiction? By (T3), Cicero has a certain property, the property of being believed by Sam to be a philosopher. Therefore, given the truth of (T1) and (T4), Tully has that same property. But this implies the negation of (D3).

To remove the contradiction, we must reject one of the T-propositions. The D-propositions express the data of the problem. Obviously, they can't be rejected. Of course, nothing hinges on the particular example. There are countless examples of the same form. Someone could believe that 3 is one of the square roots of 9 without believing that one of the square roots of 9 is a prime number, even though 3 is a prime number.

The Fregean solution is to reject (T1). In (D1), 'Cicero' refers to its customary sense, not its customary referent, while in (D2), 'Cicero' refers to its customary referent. This implies that the antecedent of (T4) remains unsatisfied so that one cannot conclude that Tully has the property of being believed by Sam to be a philosopher.

A different solution, one proposed by Hector-Neri Castaneda, is achieved by rejecting (T2) while upholding the rest of the T-propositions. The rough idea is that 'Cicero' in all its occurrences refers to a 'thin' object, an ontological guise, a sort of ontological part of ordinary infinitely-propertied particulars. This ontological guise is not strictly identical to the ontological guise denoted by 'Tully,' but the two are "consubstantiated" in Castaneda's jargon. This consubstantiation is a type of contingent sameness. Since Cicero and Tully are not strictly identical, but merely consubstantiated, the fact that Cicero has the property of being believed by Sam to be a philosopher does not entail that Tully has this property. So the contradiction does not arise. (Cf. The Phenomeno-Logic of the I, pp. 183-186)

The issue really has nothing to do with spatially external objects. There unproblematically are such objects whatever their ultimate ontological status. Note also that ER can be true even if there are no spatial individuals. ER is simply the claim that there is a way things are independent of us: it says nothing specifically about spatial individuals.

As Searle interprets it, ER sets forth a condition on the intelligibility of discourse and thought rather than a truth condition of discourse and thought:

There are conditions on the intelligibility of discourse . . . that are not like paradigmatic cases of truth conditions. In the normal understanding of discourse we take these conditions for granted; and unless we took them for granted, we could not understand utterances the way we do . . . . (181)

Among these conditions on intelligibility is ER. It is a necessary presupposition of a large chunk of thought and discourse. What Searle is doing is giving a transcendental argument for ER. He takes it as given that a sentence like 'Mt Everest has ice and snow near the summit' is intelligible. He then inquires into what must be presupposed for it to be intelligible. For the sentence to be true, Mt. Everest must exist, and it must have ice and snow near the summit. But for the sentence to be intelligible, it is not necessary that Mt. Everest exist, or if it does exist that it have ice and snow near the summit. What is necessary is that ER be true: that there be a way things are independent of human representations. If the mountain exists, then that is (part of) the way things are, and if it does not exist, that too is (part of) the way things are. The way things are, then, is not a truth condition of any such statement as 'Mt Everest has ice and snow near the summit.' It is a condition of the intelligibility of such statements and their negations. So even if every statement asserting or implying the existence of a physical object is false, and there is no spatially external world, it is still the case that ER is true. For it is still the case that there is a way things are independent of human representations.

For Searle, then, external realism (ER) is a transcendental condition of the intelligibility of large portions of public discourse. He is aware that to have shown this is not to have show that ER is true. (194) Speaking as we do, we are committed to its being true, but that is not to say that it is true. That there is a way things are independent of human representations is presupposed by the intelligibility of much of what we think or say, but it doesn't follow that it is true.

Why not? Because its truth is conditional upon the fact that our thought and speech is intelligible. If ER is true, then it is true whether or not human representations and their intelligibility exist. But if ER is argued to transcendentally as a condition of intelligibility, then ER's truth is conditional upon the existence of human beings and their representations. So we cannot say that ER is true, but only that we must presuppose it to be true. This is not to say that without us it would be false, but what without us it would be neither true nor false.

Is Searle's position satisfactory? I'm not sure. I want to be able to say that ER is true simpliciter, or true unconditionally (i.e., not conditional upon the fact of the intelligibility of our discourse.)

But does my desire to be able to say that ER is true unconditionally make sense? Maybe not. We cannot not presuppose that there is a way things are assuming that we continue to think and talk as before. But is there a way things are? Yes, it might be said, in the only sense in which it would make sense to assert it, namely, as a presupposition of our thought and talk. That is, what we as rational beings must presuppose as being the case IS the case. The 'possibility' that it not be the case is unmeaning. No sort of wedge can be driven between the presupposing and the being. But this seems to land us in a form of transcendental idealism.

A fascinating labyrinth, this. Collateral reading: Martin Heidegger, Sein und Zeit, section 44 (c), Die Seinsart der Wahrheit und die Wahrheitsvoraussetzung.

From the same source:

Owen Gingerich, historian and astronomer emeritus at Harvard who led the committee that proposed the initial definition, called the new definition "confusing and unfortunate" and said he was "not at all pleased" with the language about clearing the neighborhood.

Gingerich also did not like the term "dwarf" planet.

"I thought that it made a curious linguistic contradiction," Gingerich said in a telephone interview from Boston (where he could not vote). "A dwarf planet is not a planet. I thought that was very awkward."

Yes indeed, in the new dispensation a dwarf planet is not a planet. But is this a "linguistic contradiction" as Gingerich says? It is if one takes 'dwarf' to be what I will call a specifying adjective, one that divides a genus into species, or a class into two disjoint subclasses. Thus if 'dwarf' is taken as dividing the class of planets into the dwarves and the non-dwarves, then 'A dwarf planet is not a planet' is a contradictory expression. Compare, 'An elm tree is not a tree,' 'A bald bachelor is not a bachelor,' 'A sharp knife is not a knife.' These are all contradictory expressions.

But there are also plenty of examples of what I will call sense-shifting adjectives. These are also known as alienans adjectives. They are so-called because they 'alienate' (literally: make other) the senses of the terms to which they are attached as modifiers. Consider the way 'decoy' when used to modify 'duck,' shifts the latter's sense. Clearly, a decoy duck is not a duck, but a piece of wood got up to look like a duck. Thus, 'A decoy duck is not a duck' is not a contradictory expression. Other examples: 'Negative growth is not growth,' 'Relative truth is not truth,' 'Artificial leather is not leather.' None of these are contradictory expressions.

So in response to Gingerich I would say that no "linguistic contradiction" hides in the bosom of 'dwarf planet.' 'Dwarf' in this usage is a sense-shifting or alienans adjective. The use of such adjectives in English is well-established and not a problem. The adequacy of the new definition of 'planet' as applying to celestial objects that clear the neighborhood around their orbits and are not satellites, however, is a question I'll leave for the astronomers.

Companion post: Nine Planets or Ten? The Empirical as Involving the Conceptual

This is a very interesting argument, and I would like 'Ockham' to tell us if it is his invention or from whom he has borrowed it. The aim is to show that there are singular concepts. I take it irreducibly singular concepts are intended, singular concepts that cannot be analyzed in terms of conjunctions of pure or qualitative properties. The concept fastest runner applies to exactly one individual at a given time in a given world, and is therefore singular, not general; but it is not irreducibly singular inasmuch as it can be analyzed in terms of multiply exemplifiable properties together with the help of some logical machinery.

It would also be nice to know whether concepts are being viewed as mind-dependent items or in some Platonic fashion. Given 'Ockham''s moniker, I'll bet on the former. If so, an objection immediately springs to mind: how can minds such as ours grasp (understand) irreducibly singular or individual concepts? How could any finite mind grasp an unanalyzable concept like Socrateity? A concept cannot be 'in' a mind without being understood by the mind it is 'in'; so an ungraspable concept would not be a concept. My singular concept, Vallicellitas if you will, is graspable by God perhaps but not by any finite mind.

Ad (1). This seems right. For example, the function, or one function, of 'Tony Blair' is to tell us which member of the Labour Party is Prime Minister.

Ad (2). This line telescopes an argument that appears to be a non sequitur. I grant that even if there is only one A, there could be another, and I grant that there can only be one N. But it does not follow that singular terms are not descriptive. If the descriptive backing is made complex enough, it seems the singularity of 'N' could be captured. Form the conjunction K of all the properties Socrates has in the actual world. This should suffice to individuate him in the actual world, i.e., to differentiate him from everything else in that world. Do the same for every possible world in which Socrates exists. Now form a monstrous disjunction D from all these Ks. This huge disjunctive property (each disjunct of which is a conjunction) is built out of descriptive elements and seems to capture Socrates' singularity.

Ad (3). This is plausible. 'Hades' and 'Cerberus' are singular terms even though there is nothing to which they refer. They are not meaningless. It is just that their meaning is not exhausted by their reference. I would conclude that they must have some descriptive backing. But this is not what 'Ockham' will conclude.

Ad (4). This follows. No problem.

Ad (5). We have no good reason to accept this since (2) has not been established.

Ad (6). Non sequitur! This is supposed to follow from (5) but does so only in the presence of some such subsidiary premise as that singular terms are significant. Otherwise, one could infer that, despite appearances, singular terms are insignificant (without meaning or sense). Imagine a Bradleyan semanticist who is out to consign singular terms to Appearance as opposed to Reality.

In sum, I think there are two main problems with the 'Ockhamite' argument. The first is that (2) has not been shown to be true. The second is that irreducibly singular concepts are creatures of darkness. I understand 'Cerberus': he is the tri-capital -- Is that a word? If it isn't, it is is now -- canine guardian of the underworld. But I cannot wrap my mind around (understand) any singular concepts if these are supposed to be irreducibly, non-qualitatively, singular. To put it brutally, there are no singular concepts to be signified.

'Santa Claus does not exist,' then, is true if and only if the history of the uses of 'Santa Claus' ends in a block, i.e., if a given use cannot be traced back in the causal chain to a real person. But surely neither declarative sentence-types nor declarative sentence-tokens can be non-derivatively either true or false. It is the proposition expressed by the tokening of a sentence-type that is either true or false. So what ought to be said is that the proposition expressed by 'Santa Claus does not exist' is true if and only if the history of the uses of 'Santa Claus' ends in a block. But then it is clear that this proposition can be true even if there is no name 'Santa Claus.' The wholly nonlinguistic fact of Santa's nonexistence cannot depend on a linguistic fact about a word.

But even if 'Santa Claus' does not end in a block, but leads back to a real person, it can still be the case that Santa Claus does not exist. How? Well, suppose that at the causal origin of the name is a real person, Saint Nicholas. Then 'Santa Claus' does not end in a block: it traces back to a real person. It is just that we use 'Santa Claus' now to refer to a legendary character (who rides a sleigh through the sky, etc.) and not to Saint Nicholas. So when I say now 'Santa Claus does not exist,' I say something true even though the history of uses of the name does not end in a block.

It could be that Saint Nick became legendary so gradually that each receiver of the name intended to use it in the way it was used by the person from whom he received it. It could be a sorites situation. With each link the intention to refer to the same thing is preserved; but at the ends of the chain the reference is different. (Cf. Leonard Linsky, Names and Descriptions, University of Chicago Press, 1977, p. 107 ff. The sorites idea is my own.)

Let's see if we can sort this out. Suppose there is a man, Ken, who knows nothing about Aquinas except that he is a prominent Catholic theologian and a saint of the Roman Catholic Church. Other facts about him are beyond Ken's ken. Ken doesn't know whether Aquinas lived after Augustine or before, the titles of the books he wrote, his principal doctrines, or anything else, except the two facts I mentioned, and perhaps what logically follows from them. Suppose Ken and I are having a brief discussion about Aquinas.

Regardless of what Geach might actually have meant in the passage quoted above, my claim is that, first, when Ken says 'Aquinas was a big name in Catholic theology,' he refers successfully to Aquinas, the same person I was talking about; and second, that Ken's use of 'Aquinas' is not routed through (mediated by) the sense of any definite description Ken can (has the ability to) supply. Does this example refute descriptivism tout court? No, but it refutes a certain kind of descriptivism. It refutes the following thesis:

T. No one ever refers successfully to an object except via a definite description (or cluster of same) that he can (has the ability to) produce on demand.

McGrew's response to this is that Ken could have supplied the definite description, 'the person other people refer to when they use the name "Aquinas"' or perhaps, 'The important Catholic thinker Bill refers to when he says "Aquinas".' If 'could have' refers to an ability of Ken's, then it is not clear that Ken has this ability. Indeed, I can simply stipulate that he doesn't have it. That being stipulated, wouldn't we agree that Ken referred successfully to Aquinas when he said, 'Aquinas was a big name in Catholic theology'? Ken referred to Aquinas even though he did not refer to Aquinas via any definite description, and also lacks the ability to concoct any such description.

We now consider Jerry, who is just like Ken in his ignorance of individuating facts about Aquinas, but is intellectually more flexible. He has the ability to concoct the definite description, 'the theologian you, Bill, are referring to now when you utter "Aquinas".' Call this description D. So if I ask Jerry who he means by his utterance of 'Aquinas,' he says, 'I mean the very person to whom you are referring when you say "Aquinas".' The question is now this: Does Jerry's employment of D specify conditions satisfaction of which would count as Jerry's successful reference to Aquinas?

It seems to me that the answer is in the negative. Jerry has no idea to whom I am referring. That is, Jerry has no idea of the properties whose joint instantiation is the person I am referring to. It follows that Jerry has no idea to whom he is referring. For Jerry is referring to the same person to whom I am referring. Since Jerry has no idea to whom he is referring, and since the descriptivist notion is that reference is to whomever or whatever satisfies the idea of the object to which one purports to refer, it follow that Jerry cannot be referring to anything.

To put it another way, Jerry cannot borrow my reference; he must enact his own. I have an idea of the properties joint instantiation of which would constitute successful reference to Aquinas. But Jerry has no idea of these properties. So Jerry cannot be said, on descriptivist ptrinciples, to be referring to anything.

I have made two points. The first is that thesis (T) above is false. The second is that on descriptivist principles there can be no borrowing of speaker's or thinker's reference: a definite description like 'the person to whom Bill is referring' cannot constitute the descriptive backing of an ordinary proper name.