Meaning as truthconditions

1
Meaning as truth-conditions

• Davidson

2
Return to Frege (?)

• Frege and Russell had appealed to what
  circumstances S is true in when telling us what S
  means, but somewhat unsystematically
• Davidson invoked Tarskis work on defining true
  sentence for a formal language -gt contribution of
  structure to meaning (compositionality)

3
Frege

• Frege's views that a declarative sentence
  expresses a Thought and denotes a truth-value,
  and that the meaning of expressions is to be
  given by stating their contributions to the
  meaning of the sentences in which they occur,
• together give his approach to meaning/sense a
  central concern with the conditions under which
  sentences are true.
• Russell's analysis of definite descriptions is
  again a matter of spelling out how such
  expressions relate to the truth conditions of the
  sentences in which they occur.

4
Tarski

• Formal language (predicate calculus)
• Theory in metalanguage for L that generates all
  the T-sentences S is true in L iff S
• Used notion of satisfaction of a predicate by
  (sequence of) objects
• Reveals meaning of structures of the language L
  (see next)

5
Tarski

• Tarski showed how to define 'true sentence' for
  certain formal languages.
• Provided a theory in a metalanguage for the
  language L which would yield an infinite number
  of theorems of the form '"snow is white" is true
  in L iff snow is white' (sentences of this form
  '"p" is true in L iff p' are those that
  Tarski's Convention T requires be derivable for
  an adequate theory of truth).
• it sets out to show how the truth-conditions of
  complex sentences can be derived from their
  component parts. His fundamental notion here is
  that of the satisfaction of an n-place predicate
  by a sequence of objects.

6
Sample

• One of Tarski's examples is the idea that the
  formula x2 4 is satisfied by 2 and -2.
• This notion allows him to construct accounts of
  the way the truth-value of complex quantified
  expressions of an extensional formal language is
  derivable from the elements of the expression.
• In this way he reveals the meaning of the formal
  structures of the expressions in question.

7
Davidson

• Challenge see if we can extend Tarskis theory
  of natural language
• Needs an account of structure about same time
  as Chomskys emphasis on syntax
• the rise of Chomskyan linguisticsset out to
  provide precisely the systematic account of
  language structure that would be needed if one
  were to do for English what Tarski had done for
  the predicate calculus.
• Lycan gives some elementary ideas about syntax on
  pp. 137-9.

8
Holism

• One perhaps unexpected consequence, but one
  Davidson welcomes, is that a theory of meaning
  must be holistic we can only give a theory of
  meaning for one bit of language by giving an
  account of every bit of language.
• Interest in Davidson's programme arises from its
  commitment to extensionality.
• There seem to be sentences in natural language
  that are intensional.
• Davidson needs to construct a theory to account
  for them that does away with this intensionality.
  He has thus constructed an account of indirect
  speech
• Elsewhere he has made major contributions to the
  logical analysis of talk about actions and
  events the motivation being to fit such language
  within a system of a generally Tarskian sort.

9
Problems for Davidson from Lycan Non-declarative
sentences

• 1. Non-declarative sentences. Davidson describes
  the meaning of sentences in a language as
  resulting from the contribution of their
  components to the truth-values of the sentences.
• But a lot of perfectly meaningful sentences don't
  have truth-values. There are obvious cases
  (imperatives, questions, expressions of wishes
  optatives) and there may be some unobvious
  cases too (apparently declarative sentences that
  deserve to be interpreted as not making true or
  false claims moral or other evaluative
  statements).
• As far as the obvious exceptions are concerned, a
  common reply is to find some core that fits the
  truth-conditional story (the factor common to
  'the door is closed', 'close the door!', 'would
  that the door were closed', etc.)
• More generally hope that one can get so far with
  truth-conditions and then tell a different story
  for some special cases. 'You blew up my boat'
  gets treated by the Davidson and then we allow
  certain wh-transformations that generate
  questions why, when, etc.
• As Lycan says, you can go the other way too if
  the Davidsonian looks sufficiently attractive it
  may give you reason to deny that moral claims or
  conditionals lack truth-values.

10
Problem Testability

• 2. Testability. Often complex claims about
  logical form that Davidson produces.
• His reply is that it is difficult, but not
  impossible. We can at least see how much of the
  data is captured by the theory.
• So, for instance, Davidson's account of events
  does allow him to capture many entailments at the
  cost of quantifying over concrete events.
• He doesnt care about the ontological
  committments of the theory

11
Deictics (Indexicals)

• Deictics (Indexicals). This is another example of
  Davidson needing an elaborate account of logical
  form.
• Lycan introduces the particular topic of deixis
  by reference to the wider contrasts of semantics
  and pragmatics, distinguishing the latter into
  'semantic pragmatics' and 'pragmatic pragmatics'.
  
• the fundamental notion in any kind of pragmatics
  is the role of the context of utterance in
  contributing to the meaning of what is said.
• It is easy enough to invoke context the
  difficulty is spell out how this contribution is
  actually managed, since competent speakers don't
  have particular problems with sentences that rely
  to some extent upon their context of use.

12
Indexicals

• Lycan focuses on is deixis or the use of
  indexical or demonstrative or token-reflexive
  items.
• He raises this (p. 142) as a problem for
  Davidson. "'P' is true iff p" might be
  appropriate for some sentences, but "'I am sick
  now' is true iff I am sick now" is pretty
  hopeless.
• It makes the sentence in anyone's mouth dependent
  on my state of health. Davidson responds by
  relativizing truth to a speaker and a time "'I
  am sick now" is true as potentially spoken by p
  at t iff p is sick at t".

13
Indexicals

• Lycan's own approach is to stress the endless
  variety of factors in the context that can, on
  occasion, become relevant. Some are pretty
  obvious others more exotic.
• speaker we have seen that with 'I'.
• location 'here', 'there'.
• time the use of tenses.
• pointing gestures 'This one is more expensive
  than that'.
• assumed vantage points Lycan contrasts 'X went
  to Z' with 'X came to Z' and claims the second
  sentences can only be properly used by a speaker
  whose assumed vantage point is Z itself. There is
  also a temporal vantage point here, the time of
  the arrival of X at Z rather than the time of
  utterance.
• hemisphere it's summer in the antipodes, and we
  might talk about what is happening in Australia
  by saying 'since it's summer....'
• Lycan reports that some theorists have tried to
  capture this variety by specifying various
  parameters.
• One example uses 8 elements a possible world, a
  time, a place, a speaker, an audience, a sequence
  of indicated objects, and two other unexplained
  items.
• Lycan's point is that we cannot hope to limit the
  number of features that may become relevant.
• So he offers instead a general function (he
  writes it as Greek alpha I shall use 'D') that
  takes an expression and computes what the context
  contributes to its meaning. So if "I" always
  denotes the speaker, D("I",C) - what D assigns to
  'I' in the context C - is the speaker of C.
  D("tomorrow",C) might yield the day immediately
  following the uttering in C.

14
Indexical Complexity

• One moral is yet another recognition of the
  failure of Relationalism it fails to in the face
  of the complexities of what people can reasonably
  intend by talking about 'meaning'.
• Lycan invokes Kaplan's contrast of the
  'character' and 'content' of sentences. If Caesar
  actually said "Et tu, Brute" as Brutus stabbed
  him, we know whom the 'tu' referred to, as also
  the unexpressed reproach.
• But if we consider the myriad other uses of this
  sentence we don't usually know who was speaking,
  or to whom. So in one sense, there is a lot
  missing from our understanding but in another,
  there isn't once we know the language, we know
  all we need to know about the meaning of the
  sentence, even when it is being used simply as an
  example (and so has no referent for 'tu' and no
  real speaker).
• Character, for Kaplan, is the meaning in this
  broad sense we all have content is what you get
  when you actually plug in D and get the actual
  speaker, hearer, etc.

15
Problem Contingently co-extensive predicates

• 4. Contingently co-extensive predicates. If
  'renate' ( 'having kidneys') and 'cordate' (
  'having a heart') apply to exactly the same range
  of objects, the Davidsonian will equate them.
• Lycan replies that the theory need not contain
  nonlinguistic premises, and so will not generate
  '"renate" denotes cordates'.
• Such terms are distinguishable in certain
  constructions (modal or other opaque ones) if
  Davidson can deal with such constructions his
  theory will make the necessary distinctions. But
  that last 'if' is a big one.

16
Problem Intensionality

• 5. Intensionality. It constitutes Lycan's next
  objection in fact.
• There are a lot of linguistic contexts that do
  not seem to be truth-functional. Davidson is
  forced to offer very complex and hidden logical
  forms for such sentences sentences are
  implicitly truth-functional.

17
Problem Collapse in Grice

• Collapse into Griceanism. Strawsons critique To
  make sense of falling back on truth, one has to
  pass on to stating or asserting things, i.e.
  something to do with communication, and so the
  approach has to concede to Grice in the end.
• Lycan replies that reducing meaning to truth
  still provides a theoretical economy, even if we
  do not have anything to say further about truth.
  And there are things to be said about truth other
  than what Strawson supposes.
• So Strawson is simply assuming that Grice must be
  right somehow. At this point, Lycan suggests that
  we might take Grice as committed to
  compositionality in moving from unstructured to
  structured expressions, and so in effect
  conceding something to Davison, but providing a
  way of moving from speaker to utterance meaning
  in general.

18
Problems Extend to Possible Worlds

• Davidson can be extended using possible worlds
  but that seems to solve contingent co-extensivity
  and intensional contexts, but
• Weirdness. One might be happy using metaphorical
  talk about possible worlds to make a point. But
  this approach takes them seriously, as real
  existents in some way.
• David Lewis was notorious for advocating the
  straightforward existence of possible worlds
  they provide the best explanation of what we want
  explained, so we have to accept them, just as we
  currently accept quarks and curved space-time. To
  many of us this seems incongruous. There is
  nothing in the usage of our language that can
  offset the metaphysical baggage of possible
  worlds.

19
Problem understanding

• Understanding. Lycan reports an argument of
  Dummett and Putnam a sentence meaning is what
  one knows when one knows what a sentence means
  to know what a sentence means is just to
  understand the sentence understanding happens to
  real people and affects their behavior.
• How can knowledge of a truth-condition, a 'wide'
  property of sentences ('not in the head' as
  Putnam puts it by reference to Twin Earth
  stories), affect people?
• understanding is not merely knowledge of
  truth-conditions, so meaning is not merely
  truth-conditions.
• Lycan notes that this argument requires
  understanding to be 'narrow'. This is perhaps not
  obviously true. If it is accepted, it should make
  us question the simple 'to know what a sentence
  means is just to understand the sentence'. He
  also comments that the argument assumes wide
  concepts cannot figure in the etiology of
  behavior, but this is false.