Conditionals as Illocutionary Operators

1
Conditionals as Illocutionary Operators

• Ariel Cohen
• arikc_at_bgu.ac.il
• Ben-Gurion University
• Israel

2
Outline

• Exclusive conditionals
• Existing theories fail to account for them
• Two general problems with conditionals
• 1. The point of view problem
• 2. The commitment problem
• Conditionals in dialogue
• The solution conditionals modify sincerity
  conditions

3
Exclusive conditionals

• (1) If theres anything I hate, its waking up
  early.
• Does not mean merely that I hate waking up early.
• Exclusive meaning Waking up early is the only
  (contextually relevant) thing I could possibly
  hate.
• Maybe I hate waking up early, maybe I hate
  nothing, but there are no more options.

4
Exclusive conditionals

• Suppose
• A fire breaks out in Teds house at 6am.
• The fire destroys the house and all his property,
  giving him severe burns.
• It also causes him to wake up.
• Ted cannot utter (1), except sarcastically.
• The reason is that Ted hates getting his property
  destroyed and being severely burned more than
  waking up early.

5
Vagueness?

• Note that the presence of an exclusive
  interpretation is not dependent on the predicate
  being vague
• (2) If there is anybody Hilda kissed, its
  Marvin.
• Hilda could only have kissed Marvin (if she
  kissed anyone at all).

6
The pronoun?

• The exclusive reading of the conditional is not
  dependent on the existence of a pronoun
• (3) If there is anything I hate, I hate waking up
  early.
• The pronoun may be the result of the exclusive
  reading, not the cause.

7
Theories of conditionals

• No classical approach to conditionals can account
  for exclusive conditionals.
• For simplicity, assume the speaker hates
  something, so the antecedent of (1) is true in
  the actual world.

8
Theories of conditionals

• Material implication since A is true, the
  conditional AgtB reduces to B
• Lewis/Stalnaker since A is true in the actual
  world, which is the closest world to itself, AgtB
  again reduces to B.
• Kratzer (1986) since the speaker knows what she
  hates, A is true in all epistemically accessible
  worlds, and AgtB reduces to B.
• Must rethink the meaning of conditionals

9
Are conditionals objective or subjective?

• Gibbard (1981) if individuals have access to
  different evidence, then one may justifiably
  utter AgtB, while the other may utter Agt?B
• Lewis (1976) and Bennett (2003) AgtB is not
  merely about the speakers belief, but about the
  world.
• So, conditionals are relativized to individuals
  and evidence, yet not subjective

10
The point-of-view problem

• From whose point of view is a conditional
  evaluated?

11
Illocutionary theories

• Partly because of the point-of-view problem, it
  has been proposed that conditionals do not have
  truth values.
• Their effects are, instead, at the level of the
  speech act.

12
Conditional assertion?

• Conditional assertion (Edgington 1995 DeRose and
  Grandy 1999) a speaker who utters AgtB asserts B
  if A is true, but asserts nothing at all if A is
  false.
• But then (1) is again reduced to the assertion
  that the speaker hates waking up early.

13
Putting forward?

• Von Fintel (2003) considers (and abandons) a
  speaker who utters AgtB puts forward B and is
  committed to B following from the evidence taken
  together with A.
• But then the speaker of (1) is again simply
  committed to the proposition that she hates
  waking up early.

14
Embedding

• The main argument against illocutionary theories
  conditionals can be embedded.
• But
• 1. Conditionals are not easy to embed, and can
  only do so in restricted environments
• 2. Speech acts actually can be embedded in
  certain environments (Krifka 2001)

15
1. Conditionals are hard to embed

• Disjunctions (Edgington 1995)
• (4) If A then B, or if C then D
• Antecedents of other conditionals (Gibbard 1981)
• (5) If Kripke was there if Strawson was there,
  then Anscombe was there

16
2. Speech acts can sometimes be embedded

• Conjunctions (Edgington 1995, Krifka 2001)
• (6) If A then B and if C then D
• (7) a. Which dish did Al make? And which dish did
  Bill make?
• b. Eat the soup! And drink the tea!
• c. How beautiful this is! And how peaceful!
• d. I hereby baptize you John. And I hereby
  baptize yóu Mary.

17
2. Speech acts can sometimes be embedded

• Consequents of conditionals
• (8) If A then if B then C
• (9) a. If she calls, let me know!
• b. If she calls, what shall I say?
• c. If you help me, I promise I will repay you.
• d. If you go to New York, watch out for the
  taxi drivers (DeRose and Grandy 1999)

18
Embedding is no argument

• Conditionals do not go into truth-functional
  contexts, or into each other, easily, then The
  facts square at least as well with the hypothesis
  that conditionals do not have truth values as
  with the hypothesis that they do. (Edgington,
  1995, p.284)

19
Illocutionary theories and dialogue

• Both illocutionary theories if one utters AgtB,
  one does not make an assertion of B, but
  something weaker (conditional assertion or
  putting forward).
• But how can such weak speech acts have a role in
  a dialogue?
• How can they affect the common ground?
• Illocutionary theories never really explain how
  dialogue works (von Fintel).

20
A cloud of propositions?

• Von Fintel and Gillies (2010) reject
  illocutionary theories
• Conditionals are relativized to individuals
• AgtxB says that all A-worlds compatible with Xs
  evidence are B-worlds
• Exclusive conditionals are again not accounted
  for.

21
A cloud of propositions?

• A speaker who utters AgtxB is putting into play a
  cloud of conditional propositions
• AgtspeakerB
• AgthearerB
• AgtgroupB
• The speaker asserts the conditional on the basis
  of the (standard) truth of (i).
• The hearer reacts as if (ii) was asserted

22
Clouds of propositions and dialogue

• The speaker utters the cloud based on her belief
  in one proposition, but the hearer responds to
  another proposition from the cloud.
• Does not sit well with standard views of
  assertion (Stalnaker 1978)
• Why would the hearer react to something that is
  clearly not put forward to update the common
  ground?

23
Conditionals as assertions

• My Proposal
• Use standard assertion instead of weak speech
  acts or clouds of asserted propositions.
• The speaker who utters AgtB actually asserts B.

24
The commitment problem

• However according to standard theories of
  assertion, if one asserts B, one is committed to
  the truth of B.
• But one who utters AgtB is not committed to the
  truth of B.
• How can one assert B without being committed to
  the truth of B?

25
Assertion and commitment

• To answer both the point-of-view and the
  commitment problems, we need to take a closer
  look at assertion.
• Why does it seem so compelling that when one
  asserts B, one is committed to B?
• Searle (1965) all speech acts have sincerity
  conditions, and the sincerity condition of an
  assertion is that the speaker believe the
  asserted proposition.

26
Degrees of belief

• Belief is not an all or nothing thing one can
  believe a proposition to a varying degree.
• One may be certain that B, think that B is
  likely, entertain the possibility that B
• Vanderveken (1990-1) sincerity conditions,
  including belief, have degrees of strength.
• The strength of belief in an asserted proposition
  can be modified by various illocutionary
  operators, e.g., adverbs like presumably or
  certainly.

27
Assertion

• An assertion operator A(C,S)
• C the content of the asserted proposition
• S its degree of strength in terms of (epistemic)
  probability.
• Unmodified assertion has a default (high) value
  illocutionary operators can modify that.

28
Assertion

• For example
• (10) a. William is lying.
• b. A(lie(w), P(lie(w)) high))
• (11) a. William is certainly lying.
• b. A(lie(w), P(lie(w)) very-high))
• (12) a. William is presumably lying.
• b. A(lie(w), P(lie(w)) low))

29
A solution to the commitment problem

• When one utters A gt B, one does assert B.
• But the belief of the speaker in the truth of B
  can be at a varying degree of strength, from
  certainty to skepticism.
• Hence, it does not follow that the speaker is
  committed to B.

30
A solution to the commitment problem

• The speaker is only committed to B if A is true.
• The speaker is committed to B to the extent that
  she is committed to A
• the more likely A is, the higher the speakers
  confidence that B is true.

31
A solution to the commitment problem

• (13) If it rains, the track will be muddy.
• The speaker asserts that the track will be muddy
• But she is not committed to this proposition.
• Her confidence that the track will be muddy is
  greater than or equal to her confidence that it
  will rain.

32
Conditionals as illocutionary operators

• The conditional does not contribute anything to
  the content of the proposition
• It expresses the strength with which it is
  asserted.
• An utterance of AgtB is
• A(B, P(B)?P(A)).

33
Conditionals in dialogue

• Why would the hearer care about the strength of a
  proposition?
• The hearer has to decide whether to update the
  common ground with the asserted proposition.
• This decision is based on various sources of
  evidence.
• One of these sources is the strength of belief in
  the proposition (Wolf and Cohen 2009)

34
Conditionals in dialogue

• Sometimes the hearers decision is not made at
  the time of utterance, and is deferred to some
  future time.
• This is often the case with conditionals

35
Conditionals in dialogue

• (14) If Carolyn wins the lottery, she will be
  happy.
• A(happy(c), P(happy(c))P(win(c)))
• Suppose P(win(c)) is very small, so before the
  lottery (14) tells us very littleno update.
• But if Carolyn eventually wins, P(win(c))1.
• Then P(happy(c))1, and the hearer may update the
  common ground with happy(c).

36
A solution to the point-of-view problem

• Degrees of belief, by their very nature, are
  subjective.
• But the point-of-view problem indicates that
  conditionals are not really subjective.
• How can we objectivize belief?

37
A solution to the point-of-view problem

• Following Wolf and Cohen (to appear)
• The degree of belief need not be that of the
  speaker.
• The sincerity condition may require that people
  with sound judgment would come to believe, on the
  basis of the available evidence, the asserted
  proposition.

38
A solution to the point-of-view problem

• Each individual reasoner 1?j?m is assigned a
  weight, vj, indicating how good a reasoner he or
  she is.
• Pmixture is a mixture model of these reasoners
• the weighted sum of these individual
  probabilities

39
A solution to the point-of-view problem

• The sincerity condition of a conditional AgtB is
  P(B)?P(A).
• But this probability does not have to be
  subjective, but rather a mixture model of good
  reasoners.
• Different reasoners may have different opinions,
  and, in particular, be exposed to different
  bodies of evidence, hence their judgments may be
  different.

40
A solution to the point-of-view problem

• Different speakers may refer to different
  reasoners, or assign different weights to the
  same reasoners.
• This accounts for the relativization of
  conditionals to speakers and evidence, without
  making them entirely subjective.

41
Exclusive conditionals the solution

• (1) If theres anything I hate, its waking up
  early.
• (2) If there is anybody Hilda kissed, its
  Marvin.
• Claim In all exclusive conditionals, the
  consequent B entails the antecedent A.
• Thus, necessarily, P(A)P(B).
• But, like with all conditionals, P(B)P(A).
• It follows that P(A)P(B).

42
Exclusive conditionals the solution

• P(?x hate(I,x))P(hate(I,early))
• So hating something is the same as hating to wake
  up early.
• I hate nothing (in context) besides waking up
  early.
• This is compatible with hating nothing then both
  probabilities are zero.

43
Conclusion

• Conditionals are modifiers of the sincerity
  conditions of speech acts
• A speaker who utters AgtB asserts B with a
  strength of belief at least as high as the
  strength of belief in A.

44
Conclusion

• Solution to the commitment problem one can
  assert B without being committed to it, if ones
  strength of belief is not high.
• Solution to the point-of-view problem the belief
  is objectivized as a mixture of the beliefs of
  competent reasoners.
• Exclusive conditionals the strength of belief in
  the consequent is equal to that of the
  antecedent, resulting in a zero probability for
  all alternatives.

45
References

• Bennett, J. 2003. A Philosophical Guide to
  Conditionals. Oxford University Press.
• DeRose, K. and Grandy, R. 1999. Conditional
  assertions and biscuit conditionals. Noûs 33
  405-20.
• Edgington, D. 1995. On conditionals. Mind, 104,
  235-329.
• von Fintel, K. 2003. Epistemic modals and
  conditionals revisited'. A talk given at the
  UMass Linguistics Colloquium.
• von Fintel, K. and A. Gillies (2010) The
  subjectivity of conditionals in a new light.
  Presented at the Workshop on Subjective Meaning
  Alternatives to Relativism, at the Annual
  Conference of the German Society for Linguistics
  (DGfS), Humboldt-University, Berlin.
• Gibbard, A. 1981. Two Recent Theories of
  Conditionals', in W.L. Harper, R. Stalnaker, and
  G. Pearce (eds.), Ifs, D. Reidel, Dordrecht, pp.
  211-247.

46
References

• Kratzer, A. (1986). Conditionals', CLS-22.
• Krifka, M. 2001. Quantifying into question
  acts.' Natural Language Semantics 91, pp. 140
• Lewis, D. 1976. Probabilities of Conditionals
  and Conditional Probabilities. Philosophical
  Review, 85 297315.
• Searle, J. R. 1965. What is a speech act? In M.
  Black, editor. Philosophy in America, 22139.
  Allen and Unwin, New York.
• Vanderveken, D. 1990-1991. Principles of language
  use, Cambridge University Press
• Wolf, L., and A. Cohen, to appear. Clarity as
  objectivized belief. In Vagueness and Language
  Use, ed. by P. Egré and N. Klinedinst (Palgrave
  Studies in Pragmatics, Language and Cognition)
  Palgrave Macmillan, Basingstoke.
• Wolf, L. and A. Cohen (2009) Modal adverbs as
  negotiation chips. Sprache und Datenverarbeitung
  33(1-2).