## Sunday, May 11, 2014

You can follow this conversation by subscribing to the comment feed for this post.

Thanks for these comments Bill. I think you have your finger on the main challenge (and hence the main interest) of the theory.

>>Consider the the first-person singular pronoun, 'I.'

Can we set these aside for now? Indexicals are easily dealt with. (And yes, I should have specified this, my mistake).

>>What you are saying is that 'Frodo,' though empty, has a meaning, but this meaning is wholly reducible to the purely syntactical role it plays in the above argument.

Yes, but the idea of a 'syntactical role' is complex. Consider:

(A) There is a character in LOTR called 'Frodo'. Frodo is a hobbit.
(B) I have a pet called 'Frodo'. Frodo is a dog.

Clearly we can't, and we don't, infer 'some hobbit is a dog'. This is because the first Frodo 'refers back' to 'character in LOTR', whereas the second refers back to 'a pet'. What is this back reference? It can't be to an object existing in reality, because the first example is about something which doesn't exist in reality. Ergo the back reference must reduce to syntax. There must be purely syntactical rules that allow us to generate the inference 'some character in LOTR is a hobbit' from the two sentences in (A), and the inference 'some pet of mine is a dog' from the two sentences in (B).

'Syntax' is slippery of course. My understanding is as follows. If there is a set of rules or an algorithm that would allow a computer, armed with an understanding of English, to crack through the text of LOTR and sort out all the referring expressions into sets corresponding to the different characters, then the semantics of referring expressions reduces to 'syntax'.

Note this is not 'eliminativist' wrt semantics. Clearly names and referring expressions have a meaning. But we can reduce their meaning to syntax, i.e. linguistic rules, perhaps very complex. Demonstratives and other indexicals are an apparent exception which I would like to set to one side for now.

>>But then what distinguishes the meaning of 'Frodo' from that of 'Gandalf'?

Consider

(*) There is a wizard called 'Gandalf', and there is a hobbit called 'Frodo'. Frodo lives in the Shire.

From this we can infer that some hobbit lives in the Shire. If we changed the second 'Frodo' to 'Gandalf', we could not make this inference, although we could make the inference that some wizard lives in the Shire. Changing the name changes the meaning of the whole, ergo the names 'Frodo' and 'Gandalf', in that context, have different meanings. Or rather, their meaning is completely reducible to the inference they generate.

Although there is a prima facie distinction between reduction and elimination, you may find it difficult to keep your reduction from collapsing into an elimination. (I have written many posts on this problem.) You don't want to be eliminativist about meaning, but you may end up being eliminativist willy-nilly, i.e., nolens volens.

Further, you may not even be able to get as far as a reduction. If you show that for every mental state there is corresponding brain state, that doesn't show that mental states reduce to brain states. Correlation is not identity. If for every student there is a chair, it doesn't follow that students reduce to chairs. In general, a function that maps each x onto a unique y does not reduce the xs to the ys.

Suppose you show that every name plays a unique syntactic role, that would not suffice to show that each name just is its syntactic role.

And if you did show this, then, by the first point, you would have eliminated meaning.

Meaning is irreducible because meaning comes from mind and mind is irreducible.

Bill,

I assume that in the phrase "...that would not suffice to show that each name just is its syntactic role." you meant to say "...that would not suffice to show that [the meaning of] each name just is its syntactic role."

>>Meaning is irreducible because meaning comes from mind and mind is irreducible.

That is a labyrinth I would like to avoid. Whether or not meaning is mind-dependent, the question is whether some meanings are 'object dependent'. Some have argued either that the object is trapped in the meaning (as Kaplan jokingly suggests), or that there is some kind of irreducible relation between the meaning and the object, such that, with the object destroyed, the meaning is destroyed too.

Bill, do you agree (1) that in this mini-story

(*) There is a wizard called 'Gandalf', and there is a hobbit called 'Frodo'. Frodo lives in the Shire.

the meaning of the story is changed if we change the name 'Frodo' in the second sentence to 'Gandalf'? And do you agree (2) that, if so, this must be because the names 'Gandalf', and 'Frodo', as used in the story, have different meanings? And (3) do you agree that at least part of the change in meaning is explained by difference inference we make from the story? E.g. with the name 'Frodo', we infer that some hobbit lives in the Shire. With the name 'Gandalf', we infer that some wizard lives in the Shire.

The question is also whether there is any more to the meanings of these empty names than the inferences they generate. Do we also need 'objects' to explain the meaning of 'Frodo' and 'Gandalf' as those names are used in the story above?

Peter,
You are exactly right. That is what I meant to say.

>>Suppose you show that every name plays a unique syntactic role, that would not suffice to show that [the meaning of] each name just is its syntactic role.

Now I understand. Let's take that as a topic for another post.

Ed,

You think Kaplan was joking? I don't think so. Don't be misled by the word 'trapped.'

Yes I agree with your three points. But you are making a far more radical claim. You are suggesting that there is nothing more to the meaning of 'Frodo' than the inferential role that that name plays in the context of TLR. No one is going to deny that there is some connection between the meaning of this name and the inferential role it plays. The latter is or is close to a Moorean fact. You, however, are making a wild theoretical claim. It is arguably as wild or bold as what Meinongians claim. They say that some things have no being at all! You say that there are meaningful non-logical expressions whose meanings reduce to syntactic roles!

I should like to avoid both of these wild theories.

Fearing the Jungle, you embrace the Razor.

Note that you cannot move directly from the denial of objects to inferential semantics. Why can't a name have sense without reference?

I can understand an inferential semantics of logical connectives/constants such as the ampersand and the wedge. But names are non-logical expressions. What you have said so far gives me no reason to think that the meaning of a name can be reduced to its inferential role.

At the end of the day you may be an eliminativist about meaning, nolens volens.

>>You say that there are meaningful non-logical expressions whose meanings reduce to syntactic roles!

Not quite, but depends on what we mean by 'syntactic'. We agreed, did we not, that to resolve the problem of ambiguity we had to include meaning within the ambit of logical form. We distinguish 'Frodo' (as in LOTR) from 'Frodo' (guy in Ohio).

>>Note that you cannot move directly from the denial of objects to inferential semantics. Why can't a name have sense without reference?

Well surely it can: an object-independent sense. However, I deny that there is any more to that sense than is needed to generate that inference.

Let me work on another post to discuss by email, if that's OK with you.

PS we can’t approach this discussion without at some point looking at Searle’s Chinese Room argument. Searle argues that we cannot reduce semantics to syntax, IMO without being entirely precise about what ‘syntax’ or ‘semantics’ really means. I do a lot of translation work and very often have to refer to a dictionary. So I follow the rule: look in the Latin dictionary for the word I want to translate, and find the English word that corresponds to it. Then I have translated from English to Latin. Few translators are fluent enough to do without a dictionary, yet can’t we say we understand Latin?

I think I'll do a post on the Chinese Room. But for now:

It is hard to see what your translation adventures have to do with the Chinese room argument. For one thing, Searle stuck in the room knows no Chinese at all, indeed he cannot even identify the characters as Chinese rather than Japanese, but you know Latin fairly well. You know Latin and you know English, hence your translating, even when you must have recourse to a dictionary, is not a mechanical manipulation of symbols that are (to you) meaningless. Let us say you come across the Latin *cicuta.* From the context you suspect it refers to a poison or something harmful to humans, but you don't know that it means hemlock. So you consult the dictionary. But you did know that the Latin word is feminine, a noun, a Latin word, a word for something harmful, the rough pronunciation of the word , etc. Searle in his room knows none of these sorts of things. It is not essential to his thought experiment that he know even that the symbols he manipulates are characters of Chinese, or even of a natural language as opposed to say Esperanto.

Ed,

First, you compare refuted scientific theories to philosophical theories. In particular, you compare direct-reference semantics to eyebeam-theory of perception as well as a few other refuted scientific theories. I think this comparison is not apt. Unlike refuted scientific theories which are generally "dead for good", philosophical theories seem to exhibit the exceptional aptitude of *resurrection*: they are down (for a period), but never out. Look at the history of philosophy.

Second, I do not see how you can avoid across the board a semantics that is object-oriented. At some point or other language must link to reality. O/w I do not see how meaning can emerge. Consider the problem of empty-names or the subclass of fictional-names (your primary concern). What is the problem with them? Well, the problem is that unlike names that have a referent in a language as well in most uses, empty/fictional names lack a referent. So the very problem of empty/fictional names arises by comparison to non-empty names which do have a referent. So even in stating the very problem of fictional-names you cannot avoid but comparing them to names that do refer. Hence, word-object relations are unavoidable.

Third, Bill has been harping on the idea that you cannot tease out meaning from syntax or so called "inferential-semantics" beyond what can be done regarding the meanings of logical constants. Consider this problem. Imagine a narrative with only two syntactic expressions: 'Fa'; 'Ga'. You may go on and derive as many inferences from these two expressions until you are blue in the face (there are of course infinitely many, strictly speaking) and you won't know what 'F' and 'G' mean; whether they have the same meaning or different meanings; whether one entails the other; and so on. (This is just the Chinese Room argument all over again [I think]). So this is Bill's challenge to you: How can you get any meaning of the non-logical terms from "inferential semantics" (forget the word 'syntax' in this context) provided inferential-semantics means the traditional inferences sanctioned in Prop. and First-Order Quantification calculi? And you cannot assume here the values true/false as given to you in advance, for these depend upon meaning of the parts of the bearers of true/false.

The comments to this entry are closed.

## Other Maverick Philosopher Sites

Member since 10/2008

## April 2018

Sun Mon Tue Wed Thu Fri Sat
1 2 3 4 5 6 7
8 9 10 11 12 13 14
15 16 17 18 19 20 21
22 23 24 25 26 27 28
29 30