[hist-analytic] Grice and Carnap on Analysis

Roger Bishop Jones rbj at rbjones.com
Tue Feb 9 07:09:44 EST 2010


We need some focus JL!
Analysis.
I've attempted a fairly general response on this message, but I will have to 
be more draconian on the rest if I'm to get any progress in the rest of my 
life.

On Sunday 07 Feb 2010 00:51, Jlsperanza at aol.com wrote:
> Re: "reductive analysis" as used by Grice WoW:RE versus 'reductionist
> analysis':
> 
> "I don't object to it. As I understand it reductive analysis is a way of
> analysing aspects of languages, and this is not something I do
>  systematically (I  might be obliged to do it ad hoc in the course of a
>  conversation, but natural  languages are never the subject matter of my
>  philosophy, so I don't do this kind  of thing
> systematically)."
> 
> I'm going to use "NL" for "natural language". It's not a construction I
> use, but see Jones does. I'll oppose it to FL, formal language, which I
>  don't use it either but he does.
> 
> I think that the _linguistic_ version of 'reductive analysis' is just
> _one_: surely one that appealed to Grice. But in his day, they were never
>  sure what 'linguistic' meant. Recall that I think Bergmann had coined
>  'linguistic revolution', or 'linguistic turn'. EVERYTHING was linguistic.
>  To use your charmer: it was a linguistic TOE. But they used 'linguistic'
>  so freely that they  sometimes engaged in what irritated Mundle (The
>  critique of
> LINGUISTIC  philosophy, Oxford, 1973) -- a charming book by an opponent
>  that I rather read  any day before some exercises in silly linguistic
>  analysis. This man has charm):  He analyses abuses of
> 
>    'the grammar of ...'
>    'the logical syntax of ...'
> 
> While Grice and Strawson would be seen as merely analysing usages, etc.
> they were, or thought they were into concepts. Kemmerling found out this
>  the hard way. He wrote his PhD in German on 'meaning'. Surely he had to
>  use 'meinen'  (German for meaning). He found that none of the Gricean
>  clauses applied to  'meinen'. He is still my favourite Continental German,
>  if you can believe  that).
> 
> Jones:
> 
> "I am like Carnap in that my deliberations about language are either  about
> language in general (e.g. in discussing concepts like analyticity, though
> this only applies to descriptive language), or else they are about formal
> languages. In the latter case I tend to prescribe a semantics, and though
> mathematical logicians might define a logical system and then do something
> like  analysis to determine what are its models., I don't think they call
>  this analysis, and I'm confident they wouldn't call it reductive
>  analysis."
> 
> Good. I think ONE way to approach this is via ch. ii of my PhD thesis. Just
>  joking. There, I analyse Grice's claim to analyse 'and' versus the
> logician's  AND. The implicature "and then" as in
> 
>    "Oh, Mary -- she is very fine. She got married and had a  child."
> 
> It may well turned out that the proceeding was: she did have a child, a
> single married, suffered quite a bit for it, and eventually married the
>  vicar. But now she is, oh, so very fine."
> 
> For Grice, 'and' implicates 'and then'. So, in NL. In FL it never does.

I think the strongest you should say is never has, but you have to be close to 
ominiscient to do that there are so many FL's about.
There isn't a any reason in principle why a formal language might not have an 
"and then" connective, though it would have to be a temporal logic.

> Even in NL it never does. Why, well because in his system Q -- he only
>  cared to  compile such a FL when tributing Quine, in Davidson/Hintikka,
>  Festchrift for  Quine -- 'p & q' is defined truth-functionally. There are
>  two ways of doing  this: via Grentzen-type inference rules, and via
>  truth-tables. Both yield the  same, syntax-based or semantic-based
>  analysis of "&". Think of PM Principia  Mathematica -- and their heris as
>  Grice calls them. Think of the definition of  the iota operator, as in
> 
>    ix.Kx & -Bx
> 
> the king of France is not bald. Wrong logical form: correct logical  form:
> 
>    -(ix.Kx & Bx)
> 
> -- by using the SECOND logical form, you avoid the silly implicature: the
> king of France is NOT bald: there's none!

This looks like a muddle to me JL.

Iota is usually a "VBTO", a variable binding term operator, and would be used 
to make a definite description (ix. Kx) from a predicate such as
"King of France"
The whole would then be rendered:

	~B(ix. Kx)

If you want to demonstrate Russell's theory of descriptions by such a notation 
then its tricky, because the bound variable in the description has to be used 
outside (the description) for the assertion, so there isn't any easy way to 
abbreviate the full statement, which would have been something like:

	(there exists a unique x such that Kx), and ~B x

Problems with the scope of x there, so you spell it out :

	exists x s.t. Kx and (forall y Ky => y = x) and ~Bx


I don't think there is any strictly correct transcription into a FL of 
something which in natural languages yields sentences of doubtful status.

I don't think that it is true to say that:

	The King of France is not bald

means the same as:

	Its not true that the King of France is bald

The latter is less contentious than the former, to which I would not assent.
Neither Russell's theory of descriptions not the use of a definite description 
operator gets this distinction, for those of us who don't like to affirm the 
falsity of the  "the king of france is bald".

> ----
> 
> 
> Jones:
> 
> "Well I do want to know why Grice objects to reductionist analysis. I  can
> see in specific cases that it doesn't work. I have never been in the least
> tempted to suppose that material objects can be analysed into phenomena.
> But on the other hand, fundamental physics seems to be seeking a "TOE" and
> there is no reason in principle why such a TOE might not involve just one
> kind of individual substance. The claim that a theory is a TOE takes a bit 
>  of understanding, and whatever  the relationship between the TOE and the 
>  rest of science there will be a temptation to call it some kind of
>  reduction, which by Grice would then count
> as reductionist? From my scant knowledge of  Grice I would be surprised if
> he came out against the possibility of a single  substance TOE (if that
>  were thought to be scientifically tenable). What do you  think?"
> 
> I THINK one problem may be Hitler.
> 
> Suppose the TOE explains all that Hitler did.

I don't think that a physicist with a TOE need be committed to that claim.

> But we may still find  that
> this is insufficient. We want to analyse history in terms of the intentions
> of the historical agents. I don't think proponents of TOE are going to go
> over  the details. So they may analyse Hitler as a general case of
> pharmaceutical even  phenomena. In broader terms, what I think Grice did
>  object -- especially in the  closing section of The conception of value --
>  the devil of scientism in his  Method: from the banal to the bizarre -- is
>  that an explanation in terms of the  most fundamental physical theory --
>  call it TOE -- may leave us cold when we are  dealing at ANOTHER level of
>  theory.
> It seems he would say that the choice of 'theoretical concepts' is just
> that. The TOE works with some THEORETICAL (hence TOE) objects. But folksy
> wisdom  has its own theoretical objects. The theory of the folk (and Grice
>  could get  repetitive here -- in later years he became more and more a
>  defender of the  common woman) will work for the folk in a way which the
>  TOE may not?
> 
> The schematics of how a stratum of theory -- which he calls C -- may be
> insufficient for another stratum C' is out there in the paper. I have
> discussed  it elsewhere. If I find the relevant quote I will provide. Since
>  he is not, as  he usually is not, being dogmatic.

This seems to me the be an argument against the possibility of a TOE to which 
all else is linguistically reducible.
I accept this, but I think the idea of a TOE which does not offer that is 
coherent and sufficient to understand the physicists concept of TOE.
The is some more subtle form of "reduction" involved here (if we want to call 
it a reduction).
Carnap did embrace something like this in his latter attitude towards the 
relationship beween physicalistic and phenomenalistic language.
Personally I would not call that relationship a reduction at all.
But the idea of a TOE is interesting because it does exemplify something that 
I am inclined to call a reduction, or an analysis (this is really what  or 
similar to Russell's analyses of mind and matter are)


> Jones:
> 
> 
> 
> "Yes, I completely misconstrued this last time, thinking it was  offered as
> a general pattern for the analysis of propositions "p", but it is  rather a
> specific analysis of the meaning of "means that",
> At least I hope  that the intended content of that sentence is that "means
> that" means  "intends that", because if it is construed as telling us
> something
> about the  meaning of p (which is how I at first took it) then I would have
> to
> raise  objection."
> 
> Right. The idea that 'p' works as a dummy here Grice conceived when
> pressurised by Richards (in Repy to Richards) to reply to the challenge of
> vascous circle. Sic. Quinion has just distributed a malaprop on vascuous
>  circle and cannot resist.
> 
> --- The problem with Tarski also appealed to Grice since you mention:
> 
> "But Tarski does not in fact offer this as a viable method of defining  the
> semantics of a language, it serves in his paper in a description of
> certain difficulties which arise in defining the semantics of natural
>  languages which Tarksi considers so serious that he completely abandons
>  any such enterprise."
> 
> Right. A pity, though.

Tarksi's reasons for dumping natural languages here are global (the liar 
paradox) rather than local the semantics of particular details.
I think that one can hope to cast light on the meaning of little fragments of 
the language even though there is this big problem making the whole fit 
together.
Also I don't think the liar is as big a problem as Tarski took it to be.
The real problem with natural languages from the point of view of semantics is 
that they are chameleons, totally adaptable to their context (especially the 
speaker and hearers), so one can doubt that there is any definite language 
there at all.

> One thing Grice found failed with Tarski is things
> like
> 
>    "What the policeman said is true".
> 
> I have seen policemen, and I think this works. They usually say the  truth.
> I think it's part of their training or something. They could actually be
> discharged if they don't ("Do you know the way to the bank?" -- "Five
>  blocks to  the right, two to the left" -- myself, I usually lie rather
>  than confess I don't  know. Just joking). But Grice considers, words:
> 
>    Tarski is unable to symbolise
>        what the policeman said is true.
> 
> --- "Suppose what he said is "Monkeys can talk". He goes on in WoW:iii. The
>  idea is that a disquotational theory of truth alla Tarski, is like
> Strawson's in  terms of the illocution of 'ditto-ing', unable to cope with
> 'embedded contexts'.  Etc.
> 
> On the other hand, our formal devices (conjunction, disjunction,
> conditional, etc) ARE truth-functional, so we know about 'true' all we need
>  to  know about it, and no need for a Polish logician, respected as he was,
>  to go the whole hog and bring Aristotle to justify his Polish
>  considerations alla Lukasiewicz.
> 
> Jones:
> 
> Re: the compositionality of
> 
>     'the cat is on the mat'
> 
> or
> 
>    'monkeys can talk'
> 
> Jones:
> 
> "I'm afraid I don't get this. I feel that I am some way off  understanding
> this bit of Grice."
> 
> I was merely dropping the compositionality thing, since S. R. Bayne had
> mentioned it, and you were asking for a clarification. But I suppose that
> expecting clarification of Grice deciphered by yours truly is a bit too
>  much. I was trying to say that in predicate-calculus, it's usually
> 
>    Fa.
> 
> an individual with feature F. This means that "p" is decomposed into "F"
> and "a". And there is a way in which questions of 'meaning' apply to "F"
>  and "a"  -- what is the meaning of "F"? What is the meaning of "a"?
>  Usually, the latter  is deemed nonsense. Proper names have no
>  meaning/sense, only reference. And Fs  are given the sense of their
>  extension. But in any case, it seems proper to be  able to apply 'means'
>  to sub-sentential, sub-propositional parts. The  compositionalists, as I
>  understand them, are wanting to say that the meaning of  'p' is a
>  composite of the meaning of 'F', 'a', and the syncategoremata involved. 
>  Not for Grice.

Yes, but the question is, why not?
What cases does he have in mind as violating compositionality.

The notion of compositionality you describe does not seem materially different 
from the one which I find in semantics in computer science.
But for that kind of semantics it is easy to prove that any semantics can be 
rendered compositionally.
That's if you get free choice of what the "meanings" are.
You just have to make sure that the meanings have enough information in them 
that you don't every have to look back at the original syntax to discover the 
meaning of the whole of which they are a part.
If you can't achieve that technical effect more elegantly (and normally it 
would not be problematic) then you can contrive it by including a copy of the 
syntax in the denotation.

So technically, if you allow yourself enough freedom about the meaning of 
"meaning" (i.e. the semantic domains in your denotational semantics), then you 
can always make a semantics compositional.
However. if you wanted the things denoted to be "meanings" as these are 
understood in ordinary language, then that could be tricky.
> 
> Re: 'cause' as efficient cause in much of the positivistic thought:
> 
> "This aspect of logical positivism is new to me. I don't think I have read
> any of their writings on causation.
> One expects positivists to be rather  instrumental, so what you say is
> surprising to me."
> 
> Good point. I would have to read more about them, then. That was the story
> I was taught! Recall philosophers, especially when cross-examining, need a
> good  story even if false. When I had to pass my course of Metaphysics,
>  with a flying  A+, I must say (the official teacher was ill for a whole
>  year, so I was  cross-examined by Mario Presas, Osvaldo Guariglia and
>  Ezequiel de Olaso) it was  all about the refutation of positivism by, of
>  all people, Romano Harre.
> 
> And Madden!

I'm glad no-one every asked _me_ about them!

> I had to read that bore of a book. They say that Hume (even if that's were
> your heart is) is wrong: 'cause' is not as Hume thought it wasn't.

So he was right?
(all he said after all, was that causal connections are not necessary, and by 
that he clearly(!) meant, logically necessary)

> Grice was a metaphysician. So here it is where we may reconcile Carnap  and
> Grice. Grice thought that there was a thing called theoria-theoria or prote
>  philosophia, first philosophy alla Aristotle. Thus, he would have objected
> to  your earlier, colloquial idea that the TOE is into the ultimate
> 'substantia'.

I don't understand the objection.

> Grice would have been particular as to whether we _do_ mean
> 'substance',  hypokheimenon of Aristotle. What if it's a mere wavicle?
> 
> Recall that for Eddington -- and Grice discusses this in "Eddington's
> Tables" in his Actions and Events -- it's not really 'substantial' table
>  that is the true table, but a table made up of wavicles. The idea of the
>  ultimate item  of matter fascinates physicists.
> 
> Indeed, when I hear physicists speak -- as I often do, on Discovery Channel
>  -- talk about the big bang, and the multiverses where laws other than
> Einstein's  or Newton's hold -- I call them philosophers. I call them
>  Thales, if pressed.

Well, I agree, there is so much metaphysics in there and so little 
appreciation of the "fact".

> Etc.
> 
> So, yes. I would think that
> 
> 1. We don't need to retreat to a pragmatist escapade. Most of our notions,
> e.g. analyticity, Grice claims at the end of his day, are "pragmatist" in
> nature: to say that they are pragmatist is no excuse to deny them a
> truth-value,  since a truth-value is after all a type of value. Grice was
>  enamoured with this  constructive idea of 'value' as pervading it all
>  (especially since he saw  metaphysicians have to be REMINDED that value
>  exists, he said).

Carnap has a good story on how pragmatic decisions can be incorporated into 
language, and then yield analytic truths.

> 2. We perhaps need, as philosophers, to, however, keep some respect for
> 'folk wisdom'. Grice was conservative here, and he would say that, whatever
> TOE  claims, the ideas of the woman of the street should be given proper
> consideration, if only, as it were, _in_ the street.

I don't think there is any more of an issue here, with a TOE, than with the 
rest of science.
People sometimes are just wrong, and science might correct them.
But there is a danger that a scientist might misconstrue ordinary language as 
saying something which he knows to be false, and then attempt to correct him 
where is is not in error.

RBJ



More information about the hist-analytic mailing list