Quine: Two dogmas of empiricism

1
Quine Two dogmas of empiricism

• Nyelv és elme, 2007 márc. 21.

2
1. Intro analyticity

• Meaning is something over and above extension.
• Essentialism If there are essences, then it is
  entities in the world that have them. Essential
  properties, when assigned a semantic duty, can
  become meanings. Human has an extension, but
  this extension is determined, delineated by the
  essence of being human (e.g., being intelligent,
  according to Aristotle).
• Quine meanings should not be regarded as
  objects. They should not be construed analogously
  with reference.
• Instead, the business of a theory of mening is to
  provide criteria for the synonymy of linguistic
  forms.

3

• Analytic means true by meaning. However, this
  is not exactly the same as logical truth.
• (1) Unmarried men are unmarried men.
  (uncontroversial a logical truth.)
• (2) Bachelors are unmarried men.
• Now, (2) can be turned into a logical truth (1)
  if we suppose that bachelor and unmarried man
  are synonymous.
• ? If we wish to clarify analyticity, we confront
  the problem of synonymy.

4
Carnaps proposal

• Take a set of atomic statements of a language.
  (Say, these are simple sentences with the logical
  form of applying predicates to arguments.) So we
  have a finite set of sentences S1,SN.
• Assign a truth value to each such sentence (? a
  state description). This can be done in
  different ways.
• The TV of complex sentences is derived from the
  atomic ones by laws of logic.
• An analytic statement is then one which comes out
  true under every state description. E.g., (S1 ?
  S2)or(S1 S2)

5
Problem

• No two members of the atomic set can contain
  synonyms.
• For else SiJohn is married and Sj John is a
  bachelor can both be assigned TRUE in a state
  description. But then, the derived complex
  sentence No bachelors are married is not true
  in every state description, therefore it is
  synthetic.
• The proposal is good for logical truth, but not
  for the second form of analyticity which
  presupposes synonymy.

6
2. Definitions

• Why cant we define bachelor by unmarried
  man, thereby securing synonymy?
• Three varieties definitions
• V1) Lexicography. We write down what is used in
  the language as synonymous. (So this presupposes
  synonymy.)
• BTW, are bachelor and unmarried man
  synonymous? Any counterexamples that come to
  mind?
• V2) Explication. The goal is to paraphrase, and
  sharpen the definition, to supplement or refine
  the meaning of the definiendum especially in
  some unusual contexts. Example How should we use
  bachelor in the contexts that serve as
  counterexample above? OR spelling out
  psychological concepts like thinking, memory,
  etc.

7

• Quine says, explication also presupposes
  synonymy. For the goal is to preserve the range
  of linguistic contexts in which the definiendum
  and the definiens can replace one another, and
  extend the replaceability by refining the
  definiens.
• More examples of explication
• (3) Amnesia is a loss of memory or learning
  ability.
• Well, refine major types of amnesia constitute
  losses only of certain kinds of memories and
  learning ability.
• Or consciousness is divided by some into access
  consciousness and phenomenal consciousness, both
  being described in detail.
• Quine What we improve upon here is the
  antecedently existing synonymy. We wish to extend
  synonymy by explication. So we presuppose it.

8

• V3) Abbreviation. This is really transparent, and
  does not rest on synonymy.
• Thus, either definitions capture only logical
  truths, or they presuppose the notion of
  synonymy.
• As an example of the former, think of programming
  languages the assembler code of a processor is a
  very simple language, and the definitions of the
  even simplest operations of the computer in it
  are quite complicated. High-level programming
  languages contain elementary operations - like
  fprintf in C that are themselves constituted
  by a complex assembler code.
• Note, however, that mathematical definitions
  like the definition of a convergent series can
  still be illuminating, or explicative.

9
3. Interchangeability

• Starting idea If two words A and B are synonyms,
  then they can replace each other in every
  linguistic context except direct quotation, salva
  veritate.
• A refinement we are interested in cognitive
  synonymy. Idiosyncratic associations, poetic
  quality do not count from the point of view of
  (cognitive) synonymy.
• If All and only bachelors are unmarried men is
  analytic, then bachelor and unmarried man are
  cognitive synonyms.
• But now we want an account of synonymy which does
  not presuppose analyticity.

10
Is interchangeability sufficient for cognitive
synonymy?

• (4) ??x(Bx ? Bx) a logical truth
• (5) ??x(Bx ? UMx)
• This would be the criterion of cognitive
  synonymy.
• (Plus, in this case, the following sentence
• (6) ?x(Bx ? UMx)
• is analytic.)
• BUT now the problem is that synonymy is defined
  as substitutivity in linguistic contexts that
  contain necessary. But what is this operator?
• Prefixing a sentence S with necessarily yields
  a truth only if S is, well, analytic. What was
  thrown out of the door comes back through the
  window.

11
Drop necessarily

• Now the language becomes extensional, and
  substitutivity salva veritate only assumes
  extansional agreement, for example, the truth of
  (6) above
• (6) ?x(Bx ? UMx)
• But this wont do. Extensional agreement does not
  rest on meaning, only on matters of fact that
  could be otherwise. (E.g., dog and Johns
  favourite animal may be coextensive, but surely
  not identical in meaning!)
• Thus if we drop necessarily from our language,
  then substitutivity is insufficient for synonymy.
• In sum, cognitive synonymy presupposes
  substitutivity PLUS intensional operators (like
  necessarily), which in turn lead us back to
  analyticity.

12
Pause for a moment

• Note the contrast between Kripke and Quine
• Kripke embraces possible worlds, essentialism,
  and de re necessity.
• Quine rejects all of these notions (for Quine,
  necessity is exhausted by derivability in logic
  and the theory of numbers). This results in a
  rather austere ontology, and a skeptical view of
  semantics.
• For Kripke, intensions, propositions, and
  necessity can all be accounted for in terms of
  possible worlds.

13
Give it another try

• If analytic is secured non-circularly, then
  necessity can be spelled out in terms of
  analyticity (see above), and synonymy can be
  spelled out as substitutivity in a language which
  contains the intensional notion necessarily.
• According to this line, singular terms A and
  B are synonyms if A is B is analytic.
• Statements p and q are synonymous if p?q is
  analytic.

14
4. Semantical rules and analyticity

• Idea Perhaps the analytic-synthetic distinction
  becomes clear in a precisely formulated
  artificial language, plus certain semantical
  rules.
• Quines reply here the problem is kicked
  upstairs, instead of being solved. semantical
  rule creates just as big a problem as
  analytic.
• Truth is a matter of both language and fact. So,
  perhaps analytic statements are those which are
  somehow devoid of the factual content? This leads
  us to the second dogma.

15
5. The second dogma the verification theory of
meaning

• The theory the meaning of a statement is the
  method of empirically confirming or refuting it.
• An analytic statement is one which is confirmed
  regardless of the output of any method of
  confirmation.
• Two statements are synonymous iff their method of
  confirmation is the same.
• We can perhaps bring this down to the level of
  words two words are synonymous iff, whenever one
  replaces the other in a sentence, we get a
  synonymous sentence.
• If this works, then, via synonymy, we save
  analyticity as well.

16
What are the methods of confirmation?

• Radical reductionism (Locke, Hume) Every
  meaningful sentence is translatable into a
  statement about immediate experiance. To each
  synthetic statement there corresponds a unique
  range of sensory events that adds to the
  likelihood of its truth.
• A term is meaningful only if it names a sense
  datum or a compound of sense data.
• Even when thoroughly formalized (Carnap), this
  method did not even get close to defining, in
  terms of sense data, simple statements such as
  There is a table in front of me.

17
Connection between the two dogmas

• Analytic statements are those which stand
  confirmed in any circumstances of sensory events.
• Quines critique there is such a thing as
  empirical confirmation, but its units are not
  statements (let alone words), but rather, whole
  theories.
• Our knowledge of the world is thought of as a
  network of statements (sentences). The periphery
  is constituted by sentences which are more
  susceptible to sensory input. The center is less
  so. However, sometimes even the most central
  elements of the network may be affected (quantum
  mechanics and the excluded middle).
• Analyticity becomes a matter of degree.

18
Verificationism

• Of the above, Quine accepts the idea that meaning
  is determined by methods of confirmation.
  However, he thinks that confirmation is holistic
  sentences are not supported in isolation.

19
II. Quine on language and linguistic knowledge

• Supposing that there are rules that guide
  language users, the linguist should be able to
  discover them (and rule out the alternatives).
• Quine This cannot be done, and the suggested
  conclusion is that there is no unique set of
  rules governing language use. Let alone knowledge
  of such rules.
• Quines argument radical translation is
  impossible, and that prevents us from attributing
  knowledge of linguistic rules.

20
Radical translation

• Imagine you come across a new tribe in the
  Brazilian jungle, and the task is to interpret
  their language to find out word-meanings and
  rules of grammar.
• This situation is entirely different from
  second-language learning. There are no
  familiar-sounding words, no teacher who speaks
  both languages, etc.

21

• Walking along one day on the newly-discovered
  coast of Australia, Captain Cook saw an
  extraordinary animal leaping through the bush.
     "What's that?" he asked one of the aborigines
  accompanying him.   "Uh - gangurru." he replied
  - or something like that. Captain Cook duly noted
  down the name of the peculiar beast as
  'Kangaroo'.Some time later, Cook had the
  opportunity to compare notes with Captain King,
  and mentioned the kangaroo.   "No, no, Cook",
  said King, "the word for that animal is 'meenuah'
  - I've checked it carefully.   "So what does
  'kangaroo' mean?""Well, I think," said King "it
  probably means something like 'I don't know'...

Source http//www.consciousentities.com/gavagai.h
tm
I dont know
22
Gavagai

• The anthropologist sees a rabbit hopping out of
  the grass, and hears the informant saying
  Gavagai.
• This could mean a lot of things, not just
  Rabbit, or Look, a rabbit. For instance
• (i) Be quiet (do not frighten it off)
• (ii) That animal is dangerous (the informant saw
  rabid rabbits earlier)
• (iii) Look, a furry animal
• (iv) What animal is that?
• (v) That is not a rabbit (error)
• Additional evidence can rule out some of these
  hypotheses if the informant speaks loudly that
  makes (i) unlikely if he approaches the animal,
  that rules out (ii). Etc.

23

• Further progress Pick up a rabbit, point at it,
  and ask the informant Gavagai?
• The informant says a short word in response
  does it mean yes or no?
• Perhaps Gavagai means not a rabbit, and the
  informants response means no, or Gavagai means
  rabbit, and the reponse means yes (or Gavagai
  means rabbit, and the response means no, because
  the informant is in error).
• The interpretation of one utterance depends on
  that of the other or that of many others. The
  first step toward holism.

24

• Each entry in the translation manual depends on
  the whole system including many other beliefs of
  the informant. Any experience can be made
  consistent with a given entry by making
  adjustments elsewhere in the system.
• If, on seeing an elephant approaching the
  informant says Gavagai excitedly, he may still
  mean rabbit, assuming that he believes that
  approaching elephants make rabbits hiding in the
  grass hop out and escape.
• Or what if the informant tries to mislead the
  anthropologist?
• Verificationism and sentence meaning Quine
  argues thus
• (i) Evidence for the truth of a sentence is
  identical with the meaning of the sentence
• (ii) Due to holism, sentences in separation have
  no confirmation conditions
• (iii) Sentences do not have meaning.

25
Quines conclusion

• The linguist attempting the radical translation
  could in principle construct various incompatible
  systems which make equally good sense of the
  speakers.
• Manuals for translating one language into
  another can be set up in divergent ways all
  compatible with the totality of speech
  dispositions, yet incompatible with one another.
• Two possible interpretations
• (1) There is a fact of the matter regarding the
  rules that govern the speakers utterances, but
  there is no way to find it out (underdetermination
  ). Empirical data in general underdetermine our
  theories.
• (2) There is no fact of the matter about the
  rules. There is nothing there to be found out
  (indeterminacy)
• But what reason can there be to advance from (1)
  to (2)?

26
Eliminativism about meaning

• There is no fact about rules governing language
  this, for Quine, includes facts about meaning.
  The argument, as we saw, starts from radical
  translation.
• The idea of indeterminacy is closely linked to
  Quines conviction that notions like innate
  linguistic knowledge, I-language, and the like,
  are untenable myths.
• Language, for Quine, is social behaviour, and for
  reasons stated above, verbal behaviour cannot
  determine a unique theory of meaning. (There will
  always be different and mutually incompatible
  systems of interpretation.)
• I hope everyone sees how the Chomskian assumption
  of I-language and innate knowledge affects this
  view.

27
Stimulus meaning

• Sentences, although devoid of real meaning
  according to Quine, still have stimulus meaning.
• The affirmative stimulus meaning of a sentence S
  is the class of stimulus patterns that would
  prompt assent to S. (This class includes real
  rabbits, rabbit representations, and other things
  which indicate rabbits.)
• Negative stimulus meaning is the class of stimuli
  that prompt dissent.

28
Stimulus meaning and real meaning

• Alas, says Quine, stimulus meaning does not
  capture real meaning, that is, the situations
  in which the expression in question is correctly
  used. There are four possibilities

Expression not used correctly
Expression correctly used
?1 The guide feels the smell of bears in the
forest, and hears bearish noises. He warns his
cohort Look out, there is a bear here. But
in fact he is mistaken, there is no bear
nearby. ?2 Cases in which we do not recognize a
familiar object due to unfamiliar circumstances.
(Gaurisanker and Everest)
Affirmative stimulus meaning present
?1
Affirmative stimulus meaning absent
?2
29
Another argument to the same effect

• Whether words have stimulus meaning at all is
  questionable. (Unlike sentences which at least
  have stimulus meaning.)
• Can words prompt assent/dissent? Well, at least
  in the context of pointing and naming they might.
  In such a case one could argue that it is the
  utterance of the word embedded in the context of
  pointing which has the logical form of a simple
  sentence. (E.g., point at an object and say
  bycicle, and thereby mean that is a bycicle.)
• Apply this to radical translation You point at a
  rabbit and utter Gavagai, the informant shows
  obvious signs of assent. OK, word meaning
  deciphered?
• Quine What if the informant means by Gavagai
  temporal rabbit slice, or undetached rabbit part?

30
No stimulus meaning? no meaning

• For Quine, language is a social art that is,
  exclusively behaviour.
• Utterances mean by their use being conditioned to
  sensory stimuli. There are no facts
• - about language beyond behaviour, and
• - about meaning beyond sensory stimulation.
• Which means, if real meaning is supposed to be
  something over and above stimulus meaning, then
  it is a fiction.

31
What is Quines positive view?

• Synonymy is not a notion which we can make sense
  of in this framework.
• When we paraphrase a sentence to resolve its
  ambiguity, what we obtain is not a synonymous
  sentence, but rather one that is less ambiguous
  (and therefore more informative).
• Canonical notation in logic (regimentation)
• Notation systems (the formalizing of natural
  language) has to be a partial notation for
  disccourse on all subjects.
• Canonical re-writing has to be shallow one need
  not introduce more logical structure than what
  seems useful. Embedded in the notation logically
  simple components may correspond to terms with
  highly complex content.
• ?x(human(x) ? mortal(x)) we may not be
  interested in what it is to be mortal, for
  instance. In which case, we make no effort to
  spell out formally what is predicated by this
  word. We do so only if there is a point.

32
Regimentation and science

• The simplification and clarification in logical
  theory is an algorithmic plus conceptual task.
• Logical reduction aims at simplifications of our
  scientific conceptual schemes it is a quest of
  ultimate categories or the most general traits of
  reality.
• In this, logic is similar to physics both are
  about the world, and at the same time both
  logical and physical theories contain elements of
  convention (and conventional elements in both
  theories are controlled by reality in some way).

33
Modality

• According to Quine, the concept of necessity
  comes from that of analiticity.
• Quine suggests that no sharp distinction between
  analytic (logical) truths and synthetic ones
  (reports of the world) can be maintained. Such a
  distinction is not compatible with his view of
  how language works.
• The associative network of sentences (ANS)
  picture
• - The objects of propositional attitudes are
  sentences (and not propositions).
• - The sentences that we can entertain constitute
  the nodes of an associative network. The
  sentences are associated on the basis of either
  causal or logical connections. The connec-tions
  are associations, and as such they are
  conditioned, but how they become conditioned
  depends on the logical and causal relations of
  the sentences.

34

• Thus, there is no sharp analytic-synthetic
  distinction, but there are sentences that are
  less easy to dispense with.
• Dropping a logical law together with its
  connections would severely distort the whole
  system that is, sentences about laws of logic
  are quite central in the ANS.
• However, notice that it makes little sense to
  declare both 224 and Bachelors are unmarried
  men analytic, therefore necessary. What
  immediately comes to mind is that the latter
  sentence has counterexamples, it is somehow less
  firmly secured than the former.
• No kidding, 224 would be very hard to reject,
  it would likely turn our system upside down.
• Still, quantum mechanical research raised the
  idea of revising the principle of the excluded
  middle another, apparently rock-solid principle
  of logic and math.

35

• It is logically necessary that 224 has a
  considerable intuitive force, but It is
  logically necessary that a cyclist has two legs
  sounds awkward nothing of what we are inclined
  to call logical necessity is involved here.
• The closer we approach the periphery of the ANS
  the more obviously what we have thought to be
  logical necessity turns out to be mere rules of
  use.

36
Intentionality

• When we turn from stimulus synonymous occasion
  sentences to the construing of the meanings of
  terms, we face indeterminacy there will be many
  distinct hypotheses which are underdetermined by
  verbal dispositions.
• Traditionally, the notion of intentional content
  was supposed to solve the indeterminacy
• ?rabbit-elicited mental states, perceptions and
  sentences were supposed to be about rabbits - to
  pick out rabbits uniquely and unambiguously.
  (This would be an essentialist or universalist
  picture.)

37

• But the whole story about the indeterminacy of
  translation shows that there seems to be no
  straightforward way in which such generalized
  properties as rabbitness and the like could be
  derived.
• No cross-cultural invariance in stimulation-term
  transitions can be expected instead, there are
  many different possible transitions (determined
  to a great extent by the culture- or language
  specific structure of the ANS), all with
  different outcomes.
• To summarize, the Quinian view offers partial,
  more context-dependent treatments to the problem
  of indeterminacy, and not a gratuitous or at
  least suspicious overall notion (like
  intentionality).

38
Where to escape if we dont like this story?

• If we argue successfully that language is more
  than mere behaviour, then that substantially
  weakens Quines position about language.
• 1. Convincing evidence about creative language
  use in children, and poverty-of-the-stimulus type
  of arguments would tend to refute Quines thesis
  in Word and Object
• In acquiring it i.e., language we have to
  depend entirely on intersubjectively available
  cues as to what to say and when.
• 2. The acquisition of word meanings (Gleitman,
  Landau, etc.) there are governing implicit
  assumptions, like a new name is that for a so far
  unknown object whole object is chosen,
  identification is based promarily on shape, etc.)
• Note that these studies were actually inspired by
  Quines work.
• 3. Any convincing argument/evidence for universal
  grammar.

39
(No Transcript)
40
(No Transcript)
41
(No Transcript)