An Active-Passive Puzzle

UPDATE (7/31):  The following entry is deeply confused.  But I will leave it up for the sake of the  commenters, David Gordon and AJ,  who refuted it.  In my defense I will say something Roderick Chisholm once said about himself in a similar connection, namely, that I wrote something clear enough to be mistaken.

..............

The following two sentences are in the active and passive voices, respectively:

1. Tom said that someone was in the vicinity.

2. Someone was said by Tom to be in the vicinity.

Both sentences 'say the same thing,' i.e., express the same proposition, the same thought, the same Fregean Gedanke.  Aren't active-to-passive and passive-to-active transformations in general truth- and sense-preserving?  But the two sentences have different entailments. 

(2), which is de re, entails that someone was in the vicinity.  (1), which is de dicto, does not entail that someone was in the vicinity.  But if the two sentences have different entailments, then they cannot express one and the same proposition.

The puzzle expressed as an aporetic triad:

A. (1) and (2) express the same proposition.

B. (2) entails a proposition -- Someone was in the vicinity -- that is not entailed by (1).

C. If p, q are the same proposition, then for any proposition x, p entails x iff q entails x.

The limbs of the triad are individually plausible but collectively inconsistent.

How do we solve, or perhaps dissolve, this puzzle?

Related articles

Knowledge and Belief: An Aporetic Triad

Metaphysical Grounding and the Euthyphro Dilemma

Are Propositions Counterexamples to Brentano's Thesis?

A Being-Knowledge Antilogism

What Problem Does Literary Fiction Pose?

Modal Logic

Comments

[Ed]

The same proposition cannot have different entailments. As I pointed out in a comment to your earlier post, the difference, if there is any, has to be caused by a different reading or sense.

Suppose that p both entails and doesn't entail q. If it doesn't entail q, then it is possible that p is true and q false. But if it does entail q, this is impossible. Thus something that is possible is also impossible.

[BV]

Right, so which of (A)-(C) do you reject?

[Eric]

I wonder if we might reject (A) by assuming that propositions are structured entities, and that the structure of any particular proposition is a consequence of the structure of the sentence that expresses it.

So, if we say that voice is a structural property of sentences, then it's plausible to infer that 1. and 2. will possibly express different propositions. And if we suppose that the de dicto/de re distinction (here) applies to propositions, then the fact that the propositions expressed by 1. and 2. are possessed of distinct properties supports the claim that (A) is false.

But now we need to explain the fact that (1) and (2) seem to express the same proposition (i.e. say the same thing -- in other words, we need an error theory to explain why (A) seems to be true). We might explain this by appealing to the distinction between what sentences are commonly taken to express in natural language, and what they express in virtue of their logical form, which is revealed on analysis. So, while the sentences

(3) Elizabeth II was born in London

and

(4) The queen of England (in 2014) was born in London

seem to say the same thing when considered at the level of natural language, they express different propositions on e.g. a Russellian analysis (if we suppose 'Elizabeth II' is a 'logically proper name', and not a disguised description, and 'the queen of England' is a definite description, not a title), in which case (3) is object dependent, while (4) is object independent. Similarly, (1) and (2) seem to 'say the same thing' at the level of natural language, but express differently structured, and hence distinct, propositions.

(That's my tentative suggestion -- I'm a neophyte when it comes to philosophical logic, so I hope it isn't complete nonsense!)

[Eric]

Oops, that should be:

[While (3) and (4)] seem to say the same thing when considered at the level of natural language, they express different propositions on e.g. a Russellian analysis (if we suppose 'Elizabeth II' is a 'logically proper name', and not a disguised description, and 'the queen of England' is a definite description, not a title), in which case [EDIT:the proposition expressed by] (3) is object dependent, while [EDIT: the proposition expressed by] (4) is object independent.

(I appreciate the fact that this is one place where (what is often described as) pedantry is a virtue!)

[David Gordon]

I wonder whether one might reject (A) in this way:

(2) is not the correct passive voice equivalent of (1), nor is (1) the correct active voice equivalent of (2)

The passive voice equivalent of (1) is: That someone was in the vicinity was said by Tom.

The active voice equivalent of (2) is: Tom said of someone that he was in the vicinity.

[BV]

Hi David,

I think you're right! My lust to find insolubilia may have led me astray.

[BV]

Eric,

Thanks for your comment. But I think David Gordon got to the heart of the matter.

[AJ]

I guess I'm an idiot, but I simply don't see how #1 & #2 as formed have different entailments. Can you make this more clear? I've asked an intelligent, philosophically inclined fellow, and neither does he. Very bothersome. What gives? What are we missing?

[BV]

AJ,

If Tom knew that someone was in the vicinity, then someone was in the vicinity. But if Tom said that someone was in the vicinity, then it doesn't follow that someone was in the vicinity.

Ah, now I get it. (2) does not entail that someone was in the vicinity either.

Now I must confess publically that I made two bad mistakes in this post, the one exposed by David Gordon above, and the one that I think you are alluding to. Mea culpa, mea maxima culpa.

My excuse? I wrote the post very quickly.

But in my defense I will say this: My post was clear enough to be refutable!

[london ed]

We are no way out of the woods yet. (1) and the natural reading of (2) are obviously not equivalent (as I was the first to point out, in fact, due to the different entailments). But we still have the problem of active-passive equivalent when proper names are the accusative of the active verb. Consider

(1) Bilbo, who[m] Tolkien depicts as being in his old age …

(2) Bilbo, who is depicted by Tolkien as being in his old age …

In the first example, the relative pronoun 'who' is the object of the verb 'depicts'. This is different from the example above, where a 'that' clause is the object of 'says'. It follows that when we convert active to passive, that 'who' is the subject. (If it were in Latin, the first 'who' would be the accusative 'quem', the second would be the nominative 'qui'). So there is no David Gordon solution available here, i.e. we can't make a that-clause the subject.

But the corresponding sentence we would be talking about in Tolkien is "Bilbo is in his old age", i.e. Tolkien does not write "Bilbo is depicted by me as being in his old age". Van Inwagen uses this type of example to prove that 'Bilbo' in (2) cannot have the same function as it does in LOTR. Whereas in LOTR it either refers to a hobbit or to nothing (to nothing, according to Van Inwagen), in (2) it must refer to an abstract object. This idea underlies Van Inwagen's entire theory, and he returns to it constantly.

However he is wrong in this case. Consider a continuation of (2):

(3) Bilbo, who is depicted by Tolkien as being in his old age, is a kindly old hobbit

The subject of the copula 'is' is Bilbo. But he is also, via the relative pronoun, the subject of 'is depicted'. This is impossible, on Van Inwagen's theory. The subject of 'is depicted' must be an abstract object. The subject of 'is a kindly old hobbit' is either something else (a Meinongian whatsit) or nothing at all.

*I changed 'who' to 'whom' to satisfy pedants.