Conditionals as Illocutionary Operators Ariel Cohen arikc bgu ac

Conditionals as Illocutionary Operators Ariel Cohen arikc@bgu. ac. il Ben-Gurion University Israel

Outline l l l 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

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

Exclusive conditionals l Suppose: l l l A fire breaks out in Ted’s house at 6 am. 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.

Vagueness? Note that the presence of an exclusive interpretation is not dependent on the predicate being vague: (2) If there is anybody Hilda kissed, it’s Marvin. = Hilda could only have kissed Marvin (if she kissed anyone at all). l

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. l The pronoun may be the result of the exclusive reading, not the cause. l

Theories of conditionals l l 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.

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

Are conditionals objective or subjective? l l l Gibbard (1981): if individuals have access to different evidence, then one may justifiably utter A>B, while the other may utter A> B Lewis (1976) and Bennett (2003): A>B is not merely about the speaker’s belief, but about the world. So, conditionals are relativized to individuals and evidence, yet not subjective…

The point-of-view problem From whose point of view is a conditional evaluated?

Illocutionary theories l l 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.

Conditional assertion? l l Conditional assertion (Edgington 1995; De. Rose and Grandy 1999): a speaker who utters A>B 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.

Putting forward? l l Von Fintel (2003) considers (and abandons): a speaker who utters A>B 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.

Embedding l l 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)

1. Conditionals are hard to embed Disjunctions (Edgington 1995) (4) #If A then B, or if C then D l Antecedents of other conditionals (Gibbard 1981): (5) #If Kripke was there if Strawson was there, then Anscombe was there l

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. l

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 (De. Rose and Grandy 1999) l

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)

Illocutionary theories and dialogue l l Both illocutionary theories: if one utters A>B, 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).

A cloud of propositions? Von Fintel and Gillies (2010) reject illocutionary theories: l Conditionals are relativized to individuals l A>x. B says that all A-worlds compatible with X’s evidence are B-worlds l Exclusive conditionals are again not accounted for.

A cloud of propositions? A speaker who utters A>x. B is putting into play a “cloud” of conditional propositions: (i) A>speaker. B (ii) A>hearer. B (ii) A>group. B l The speaker asserts the conditional on the basis of the (standard) truth of (i). l The hearer reacts as if (ii) was asserted l

Clouds of propositions and dialogue l l l 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?

Conditionals as assertions l l l My Proposal: Use standard assertion instead of weak speech acts or clouds of asserted propositions. The speaker who utters A>B actually asserts B.

The commitment problem l l l However: according to standard theories of assertion, if one asserts B, one is committed to the truth of B. But one who utters A>B is not committed to the truth of B. How can one assert B without being committed to the truth of B?

Assertion and commitment l l l 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.

Degrees of belief l l Belief is not an all or nothing: 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.

Assertion l l 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.

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)) l

A solution to the commitment problem l l l When one utters A > 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.

A solution to the commitment problem l l l 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 speaker’s confidence that B is true.

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

Conditionals as illocutionary operators l l The conditional does not contribute anything to the content of the proposition It expresses the strength with which it is asserted. An utterance of A>B is: A(B, P(B) P(A)).

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)

Conditionals in dialogue l l Sometimes the hearer’s decision is not made at the time of utterance, and is deferred to some future time. This is often the case with conditionals

Conditionals in dialogue (14) If Carolyn wins the lottery, she will be happy. l A(happy(c), P(happy(c))≥P(win(c))) l Suppose P(win(c)) is very small, so before the lottery (14) tells us very little—no update. l But if Carolyn eventually wins, P(win(c))=1. l Then P(happy(c))=1, and the hearer may update the common ground with happy(c).

A solution to the point-of-view problem l l l 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?

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

A solution to the point-of-view problem l l l 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

A solution to the point-of-view problem l l l The sincerity condition of a conditional A>B 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.

A solution to the point-of-view problem l l 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.

Exclusive conditionals: the solution (1) If there’s anything I hate, it’s waking up early. (2) If there is anybody Hilda kissed, it’s Marvin. l Claim: In all exclusive conditionals, the consequent B entails the antecedent A. l Thus, necessarily, P(A)≥P(B). l But, like with all conditionals, P(B)≥P(A). l It follows that P(A)=P(B).

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

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

Conclusion l l l Solution to the commitment problem: one can assert B without being committed to it, if one’s 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.

References Bennett, J. 2003. A Philosophical Guide to Conditionals. Oxford University Press. De. Rose, 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 (DGf. S), 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.

References Kratzer, A. (1986). `Conditionals', CLS-22. Krifka, M. 2001. `Quantifying into question acts. ' Natural Language Semantics 9: 1, pp. 1– 40 Lewis, D. : 1976. “Probabilities of Conditionals and Conditional Probabilities. ” Philosophical Review, 85: 297– 315. Searle, J. R. 1965. `What is a speech act? ’ In M. Black, editor. Philosophy in America, 221– 39. 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).