[hist-analytic] L-truth, A-truth: Carnap and Grice

Roger Bishop Jones rbj at rbjones.com
Wed Mar 3 10:53:38 EST 2010

On Tuesday 02 Mar 2010 22:31, Jlsperanza at aol.com wrote:
> In a message dated 3/2/2010, rbj at rbjones.com writes:
> "at the last in the Schilpp volume he concedes defeat to
>  the  Quine-Tarski conspiracy on use of the term "Logical
>  Truth" and starts using  A-true instead of L-true. In
>  "Meaning Postulates" he seems to use the "L-"  concepts
>  throughout."
> Interesting. So, in the new rewrite the "Meaning
>  Postulates" would then become 'analytic'.

Meaning postulates would always have been analytic, as (well  
logical) truths for Carnap.

What's happening here is not that the meaning of either of 
these terms is changing as far as Carnap is concerned but 
just that he is coming to terms with the fact that he is the 
odd man out on what "Logical Truth" should mean and deciding 
to stop using the term.  Quine and Tarski want it to be a 
narrower term than analyticity (a difficult position for Quine 
to coherently adopt) but can't actually make their minds up 
what it should mean.  They want logical truths to be true 
taking account only of the meanings of the "logical" 
constructs but not of "non-logical" constructs, but they can 
offer no definite account of which constructs are and are not 

It might be worth mentioning that this switch from L to A on 
Carnap's part is a small part of a rather larger 
reformulation of his semantics in the Schilpp volume, the 
primary effect of which is to disarm Quine's objection in 
"Two Dogmas" to the definition of analyticity for specific 
languages rather than in general terms.  It is not plausible 
that Quine could not see that Carnap's semantics could 
easily be reorganised to meet his objection, but he made it 
anyway as if it were substantive, and it is something which 
Carnap should have done all along.
I have only just carefully read through that short section 
(just a couple of pages 900-901), and it is not above 
criticism, but the points I could raise against it are also 
straightforward to correct.

>  Personally, I
>  don't think it was a bad move at all: from "L"  to "A".
>  It does make you wonder about meaning, though. Suppose I
>  say I use 'snow' to mean snow CONTINGENTLY, i.e. not
>  really 'analytically'.

I don't myself think this a proper use of the terms 
contingently and analytically.  You are using them as 
qualifiers for actions or events, but they are properties of 
propositions or sentences in context.
Also you are mixing your qualifiers here, unless you are 
conceding the coincidence of analyticity and necessity
(which I would encourage!).

Maybe that's not what you were doing.
You want it to be contingently true that snow means what it 
does, and your wish is granted, it is!
But it doesn't make any difference to the status of truths in 
the "object language", it only makes a difference to meta-
theoretic claims.

Thus, "snow is white" will be analytic, but that would 
normally be contingent.  It depends how you say it, how you 
identify the language.  If you say:

	"In the English language 'snow is white' is analytic"

Then to discover whether this is analytic or synthetic you 
have to dig into the meaning of "English language", but I 
think it is likely to be synthetic.  (assuming "English 
language" means something like "the language predominantly 
spoke in England".

>  I do it because my parents taught
>  me to. I could have used "Arthur" (Harrison, Intro to
>  the Philosophy of Language -- Macmillan, ""Arthur" we
>  could use to refer to snow -- the idea that a natural
>  kind should  not be thus named is a convention we should
>  sometimes _flout" (or words -- what a  genius of insight
>  Harrison is -- born Sussex, this his main work, a
>  treasure in  my Swimming-Pool Library). And suppose I
>  use 'white' CONTINGENTLY too.  So, 'all  snow is white'
>  is true iff all snow is white. "All snow is white" as
>  ANALYTIC  meaning postulate?

If it is a "meaning postulate" (a term I deprecate, since to 
call it a postulate suggests something more speculative than 
one expects in a prescription of meaning, and to use 
"meaning postulate" as the name of a definition invites 
improper definition and encourages this wayward idea that 
analyticity is an attribute which we can arbitrarily assign 
to whatever propositions we would like to be necessary) then 
It will be analytic and necessary.  It makes no definition 
how we know that it is a "meaning postulate" or how it came 
to be one, these are meta-theoretic, and it is the truth and 
status of the sentences/propositions of the object language 
which concerns us. 

>  Well, yes, but relying on a
>  few contingent things, like my  choice of labels... etc.
>  So, I would think that L-truth vs. A-truth, may be read 
>  as a move towards a _stronger_ position?

No, its a purely verbal adjustment.
The issues you are raising are not real problems in Carnap's 
position they just need explaining.  Which possibly I might 
have done?

I liked the tweedledum stuff.


More information about the hist-analytic mailing list