Meaning, Truth-Conditions and Defeasibility

[J L Speranza]

In this post I shall address some of the topics raised by R M Petty in
his interesting communication. I'm still elaborating on the issues as the
thread develops, thus my remarks should in no way be seen as conclusive.
Also, if my overall tone is rather critical, that may well be due to my
current scepticism as to how constructive one can be in this field. I hope
my post may clarify my position regarding some terminological or
definitional issues.
In 'Re: Truth & Other Conditions' (#5019), R M Petty writes: "Both [J L
Speranza] and M J Murphy seem to be using the term 'defeasible' [as in
'defeasibility-condition' JLS] in a way that is not familiar to me". Also M
J Murphy in 'Dummett & Various Conditions' (#5018) writes: "I am assuming
... 'warranted assertibility' [conditions] and 'defeasibility-conditions"
are the same thing. ... J L Speranza sounds like he sees a distinction". In
his reply to M J Murphy (#5019), R M Petty characterises a
'defeasibility-condition' as different from both a truth-condition and a
warranted-assertability condition, adding that it may well still be
"different than what J L Speranza has in mind".
Well, I should say his characterisation is pretty accurate as per what
I did have in mind. His point, contra M J Murphy, is to suggest that such a
defeasibility condition would NOT be a one under which a truth-condition
would be subsumed: "...When one has WARRANT TO ASSERT ['p']
one has reason to believe that ['p'] is TRUE. The DEFEASIBILITY-CONDITIONS
would be those ... that would make ['p'] FALSE. One could have warrant to
assert ['the cat is on the mat'] if one knows that the cat in question has a
habit of sleeping on the mat after having just eaten, and one knows that the
cat has just eaten. The GENERAL DEFEASIBILITY-CONDITION ... is the cat NOT
actually being on the mat; a SPECIFIC DEFEASIBILITY-CONDITION may be that
the cat was distracted by some shiny object".
Excursus. While believing R M Petty's explication covers my view pretty
accurately, admittedly I was thinking originally of 'defeasility' as it is
used as a technicism (e.g by S C Levinson's Pragmatics) to label a feature
of the so-called nonmonotonic inference pattern. Thus, in '(All) birds fly;
the penguin is a bird; Therefore, the penguin flies', the 'therefore' is
formalised by a twisted arrow, saving it from being a 'fallacy'. The root of
nonmonotonicity is of course in the premise, '(All) birds fly', a defeasible
statement if there ever was one which gets defeated every time the objection
is raised, 'Hey, the penguin is a bird and it don't (sic) fly'. Monotonicity
thus relates to 'ceteris paribus' statements: 'it's all things being equal
that (all) birds fly'. Defenders of nonmonotonic usually reasoning bring
psycholinguistic evidence that noting the stereotype of 'bird' is a flying
creature, and that it's universal quanitification that is processed (however
qualified as ceteris paribus), rather than a corresponding particular one
('Some/most birds fly').
Now, how can one state defeasibility-conditions in a full way so as to
provide the 'meaning' of a given utterance 'p'? To start with, it may be
worth noting that the Anti-Realist, pointing as he is to a limitation of
truth-conditional semantics, is certainly diverging from a conservative
approach such as H P Grice's (in
Studies in The Way of Words). For Gricean pragmatics, the central meaning
('dictive component') of 'p' is
shaped by 'p''s truth-condition. Anything beyond such condition may still be
accounted for part of the utterer's meaning (broadly, a 'conversational
implicature'). Now, this diverges from saying such 'implicature' will
include specific warranted-assertability or defeasibility conditions.
Rather, it will take the shape of that aspect of meaning that is derived
from the truth-condition and which has to be inferred assuming a general
presupposition of rationality. Thus, by uttering 'the cat is on the mat', an
utterer may conversationally implicate some figurative extension to that
utterance (e.g. that some nasty woman is being punished by her pimp - as my
Longman Dictionary allows), it would require a different scenario that some
specific defeasibility condition is being implicated.
R M Petty distinguishes between of 'general' vs 'specific'
defeasibility-conditions. P M Petty describes one such general
defeeasibility condition for 'the cat is on the mat' as being 'the cat is
not on the mat'. Granted, he uses the 'ontological mode' of a state of
affairs of the cat not being on the mat. The 'defeasibility-conditional
semantics' for 'p' would thus list as Tarski theorem making up the meaning
of 'p' that not-p. Perhaps bringing in specific defeasibility conditions is
more promising. R Petty lists as such one, 'The cat has been distracted by a
shiny object'. I guess the proponent of a defeasibility-conditional
semantics would be however happy to specifiy the Tarski theorem as one that
does not mention any specific condition, i.e. as 'p' means "ceteris paribus"
p.
With regards to Schiffer's treatment of Dummett, R M Petty writes: "My
intuition is that [Schiffer] has not quite got Dummett right, although ...
this is really nothing more than a guess". Re: Schiffer's
counterexample to Dummett, (1) Thales did not begin to walk until he was 13
months old, he writes:
"My intuition is that [the 'Thales' sentence] is an EFFECTIVELY UNDECIDABLE
statement"
Maybe. Schiffer's point is that it is also MEANINGFUL, and so that the
onus probandi is on Anti-Realist Semantics as how to explain such semantic
dimension.
R M Petty writes: "[1] is EFFECTIVELY UNDECIDABLE for the reasons
Schiffer gives ... There is no clear method of decidablity. As such one
ought to speak of [warranted] ASSERTABILITY-conditions rather than
truth-conditions. It strikes me that Schiffer's description DOES MEET with
the conditions of warranted assertion: ... (1) is grammatically well-formed,
speaks of a figure that is part of our cultural history, and
supplies information about the biological development of said figure. I am
sure this is not an exhaustive list of assertability-conditions even if it
is an accurate one".
Myself, I would take the exegesis referred to by R M Petty (that (1)
speaks of "a figure that is part of our
cultural history and supplies information about the biological development
of said figure") to be about the 'meaning' of (1), rather than its
warranted-assertability conditions. 'Assertability-conditions' (i.e. the
utterer finding himself in a position that warrants the assertion of (1))
would rather include to be the utterer's belief that 1. There is a written
work from the time of Thales, and 2., that that written work says among
other things, as translated into English, that Thales did not begin to walk
until he was 13 months old'. Now, this can always be refuted by some
SPECIFIC DEFEASIBILITY-CONDITION that shows that the written work (referred
to in the warranted-assertability condition) is proved to be a fake (Perhaps
similarly for 'the cat is on the mat').
R M Petty writes: "What makes me think this is a misunderstanding ... of
Dummett [by Schiffer] is that Schiffer writes as if VERIFICATION is
interchangeable with TRUTH. ... He commonly pairs [verification] with
falsification ... I think he has misunderstood the point of saying that (1)
is EFFECTIVELY UNDECIDABLE".
It seems to me that Schiffer is NOT concerned with (1)'s truth-condtion,
only its direct-verification conditions. I take that (1)'s truth conditions,
as per Schiffer and Davidson would merely mention the
paratactical Tarskian theorem ('Thales did not begin to walk until he was 13
years old' is true iff Thales did not begin to walk until he was 13 years
old), or else some specification of the right-hand side of such
biconditional (eg. 'the philosopher called Thales did not begin to
articulate his feet to move his body about and without the aid of his hands
until at least the 390th day after he was born).
R M Petty writes: "It strikes me that the point of classifying some
statements as EFFECTIVELY UNDECIDABLE is to separate those statements for
which we cannot TELL if they are true, false, or have some third value." He
quotes from Dummett's 'Truth' "to the effect that, for those statements that
are EFFECTIVELY UNDECIDABLE, or for those statements that are properly
subject to 3 truth-values, [...] it may be better if we just assume they are
FALSE":
"In most philosophical discussions of truth and falsity, what we really
have in mind is the distinction between a designated and an UNDESIGNATED
VALUE, and hence choosing the names 'truth' and 'falsity' for particular
designated and undesignated values RESPECTIVELY will only OBSCURE THE ISSUE"
(PL p63, TOE p14app) (Emphasis mine. JLS).
P M Petty writes: "... I admit that I am not entirely comfortable with my
understanding of this passage ... but I read the class of DESIGNATED
TRUTH-VALUES as interchangeable with the class of effectively decidable
statements and UNDESIGNATED as EFFECTIVELY UNDECIDABLE statements"
Allow me to disagree. Note that Dummett is using 'respectively'.
'Undesignated' seems to mean there just 'false', i.e. the class of
utterances one can DECIDE to be not 'true'. I'd take the
'designated/undesignated' terminology as a logical technicism, brought e.g.
to mark eg a formal vs informal interpretation of a
calculus. Thus, one may have ''p and q' is true iff both conjuncts are true'
(or ''p & q' is 1 iff both p and q are 1'), with 'false' (or 'O') as
UNDESIGNATED. Alternatively, in ''p and q' is false iff at least one conjunct
is false' (or ''p and q' is 0 iff at least one conjunct is 0') it is rather
'true' (and '1') which are undesignated.
P M Petty adds: "It may be better to assume EFFECTIVELY UNDECIDABLE
statements are FALSE in order not to OBSCURE THE ISSUE [as per Dummett's
wording], 'the issue' being whether or not we have a method of decidability".
Well, wouldn't THAT obscure the issue? :). I would think that if some
utterance is held to be FALSE, then it is DECIDADBLE, in fact DECIDED that it
is NOT TRUE. The problem (or Dummett's Dilemma with the Effectively
Undecidable as I may call it) is that, by definition, there is no way (or
algorithm) of decididing if a given 'p' is true or false (eg Schiffer's
Thales sentence). The paradox being that such utterances still 'mean'.
R M Petty: "How Schiffer's terminology may be leading to a
misunderstanding is that his use of 'verification' may have roughly the same
meaning as 'method of decidability', provided the former is meant to convey
that we have the effective ability to verify that ['p'] is either true or
false. ... To verify ['p'] would, then, be to determine whether or not [it]
is true or false. Unfortunately, Schiffer couples
'verify' with 'falsify' as if to verify something is to find it to be true.
It may be that Schiffer's terminology is an example of the exact type of
OBSCURANTISM that Dummett refers to in the passage above"
Allow me to disagree again. I don't quite think that in equating
'verify' as 'finding
something to be true' Schiffer is being obscure. But then I don't think
Dummett is in any case having such case of 'obscurantism' in mind. Dummett's
reference to 'obscuring the issue' seems to be related to the
designated/undesignated distinction, i.e. to the idea that replacing talk of
'truth' by talk of 'designated value' (where this can be 'true', '1', or
what not) would be (in an essay entitled 'Truth') a case of 'obscurus per
obscurius'.
R M Petty: "[This] terminology may allow Schiffer to slide from the
issue of whether or not ['p'] is CAPABLE of being found to be true or false
to an attempt to find ['p'] true or false, as if there were no distinction.
This is the value, and point, of Dummett's distinction and one I hope I have
made it clearly".
I agree there IS a distinction between a factual statement ('p' is
verified') and a modal one ('p cannot be verified), and Dummett has indeed
brought to focus the class of effectively undecidable statements. Schiffer's
contribution is merely to point out that those statements still seem to have
a 'meaning'... Note that in Remnants of Meaning, Schiffer is examining what
constructive proposal may be behind anti-realism, namely, what a
defeasibility-, direct-verification- , or
warranted-assertability-conditional semantics may have to offer.
R M Petty writes: "I should add that, [for] ... effectively decidable
statements, [warranted-assertability] conditions and truth-conditions may
very well be interchangeable, or at least very similar".
Are they? How similar (in meaning) are: 'the felis domesticus is
situated upon the rag' and 'the
cat has just eaten and we know where he habitually goes after that', which I
take (respectively) as something like a truth- and a warrant-assertability-
condition for an utterance which that does NOT look terribly as an
effectively undecidable utterance?
R M Petty writes: "[For] EFFECTIVELY UNDECIDABLE STATEMENTS,
assertablity-conditions and truth-conditions are definitely NOT
interchangeable".
I second that. I would add that, as the strong point of anti-realism, a
way out of Dummett's dilemma with the Effectively Undecidable would be to
regard that the very talk of a truth-condition for such statements would be
pretty much NONSENSE. I agree in this with B M Majors in that Dummett seems
to be following a double strategy (the two horns of his dilemma?). One thing
is to demistyfing the notion of a truth-condition (suggesting that it may be
understood in ways other than the realist one), another thing to eliminating
it (at least for the problematic case of the effectively undecidable).
With apologies for the very long post. I realise I could edit and
re-edit my Spanglish, but I rather send it before too long... :)

With all best wishes,
J L Speranza
Buenos Aires, Argentina.

=============================================================================
| DISCUSS | Analytic Philosophy | <http://www.shore.net/~vanegas/analytic/> |
|---------------------------------------------------------------------------|
| For information about your subscription to Analytic, send mail to |
| <analytic...@shore.net> or point your WWW browser to the URL above. |
|---------------------------------------------------------------------------|
| Analytic's Second Annual Essay Contest is now open. Essays will be |
| received until June 15, 1999. The winning author will receive $250. |
| See <http://www.shore.net/~vanegas/analytic/prize.html> for details. |
-----------------------------------------------------------------------------