Tuesday, October 05, 2010

If you aren't dead, this is a relevance conditional

There's an lovely relevance (or biscuit) conditional in the latest edition of BBC Radio's More or Less, a programme about statistics. It's in the programme broadcast on 1st October 2010, currently available here, and eventually to be archived here.

The sentence uttered:
If you are still alive, that was Matt Parker, the stand-up mathematician.

The context:
Introducing a guest item about the risks of death associated with different activities, the host announces that “there is a 0.00003 % chance that you will die while listening to Matt's essay.” (This is at about 20:40 minutes into the programme.) At the end of the essay (around 24:00), the host picks up again with the sentence quoted above.

What does it mean to say that this is a relevance conditional? Well, the way that the sentence is used here, it is clear that the main clause of the conditional (‘that was Matt Parker, the stand-up mathematician’) is true regardless of whether the proposition given by the if-clause ‘you are still alive’ is true or false. So it is more like the second group of examples below than the first, or the third, and like the other relevance conditionals it sounds very odd indeed if you put then in:

?? If you are still alive, then that was Matt Parker, the stand-up mathematician.

Normal (hypothetical) conditionals
If you heat water, (then) it boils.
If John has restocked it, (then) there’s beer in the fridge

Relevance conditionals
If I may be honest, (??then) you are not looking good.
If you are thirsty, (??then) there’s beer in the fridge

Factual conditionals
If it is stupid (then) you shouldn’t bother with it.
If he’s so smart (then) why isn’t he rich?

These examples and the labels for the different uses of conditionals (they are not really different types*) are from Bhatt, R. & Pancheva, R. (2006). Conditionals. In M. Everaert & H. C. van Riemsdijk (Eds.), The Blackwell Companion to Syntax. (pp. 639–687). Oxford: Blackwell.

Why the alternate name, biscuit conditionals? That is what philosophers usually call them, following J.L. Austin's example:

There are biscuits on the sideboard if you want some.

(from p 212 of Austin, J.L. (1970) Ifs and cans. In Philosophical Papers, 2nd ed. London: Oxford University Press: 205–232.)

There is a post on relevance conditionals at Language Log, if you want to know more, with excellent references to the recent scholarly literature.

*Since (e.g.) there might be circumstances in which your being thirsty causes** there to be beer in the fridge (or vice versa***) and in those cases an utterance of “If you are thirsty, (then) there’s beer in the fridge,” will express a factual normal, i.e. hypothetical conditional. See p 406 of DeRose, K. & Grandy, R. E. (1999). Conditional assertions and ‘biscuit’ conditionals. Noûs, 33(3), 405-420.

**Other relations than causality are possible, but that is a topic for another post...

***For example, we know that Smith puts salt in your drinking water sometimes and this makes you thirsty (but you never get thirsty otherwise). We also know that he only does this if he knows that there is beer in the fridge. Then we have good grounds to think that it is true to say “If you are thirsty, (then) there’s beer in the fridge.”


Andreas said...

What about conditionals like these:

(1) If Chopin's love life got worse as he got older, his music got better.

(2) If a Rolls Royce is not exactly a family car, it certainly is fast.

(3) If dogs are a poor substitute for friends, they're cheaper than kids.

Are these biscuit conditionals?

nick said...

I think that these are a fourth* type (or rather, use) of conditionals.

I take it that the hallmarks of this use are that what is in the if-clause is (pragmatically) presupposed – and generally has to be accommodated – whereas what is in the main clause is (typically) asserted.

(Or in relevance-theoretic terms, such a use makes it manifest that the speaker is committed to what is expressed by the if-clause, and makes manifest what is in the main clause).

We're therefore on the line of the truth-table for material implication where p is true, q is true and (therefore) p->q is true.

Here are a couple of attested examples:

If there are tensions within the coalition, and more pointedly within the Liberal Democrat party itself, an analysis in the Liberal magazine is likely to pour fuel on the fire.
(This could have been meant as a standard, hypothetical conditional, but wasn't.)

If Jean Nouvel's polemical Quai Branly ethnological museum in Paris was a viciously brilliant exercise in atmospheric complexity designed to stick two fingers up at post-colonial orientalists, the Doha museum has steered an almost perfectly balanced course between academic and popular viewing conditions.

* Since Bhatt and Pancheva also list factual conditionals (following Iatridou, also noting that Haegeman calls them premise conditionals).

The example they give is:

If Fred is (indeed) so smart, why didn’t he get the job?

They say: "Factual conditionals are somewhat harder to distinguish from hypothetical conditionals. According to Iatridou (1991), the conditional clause in a factual conditional is presupposed to be true. Haegeman (2003) notes that while in a hypothetical conditional the antecedent clause is integrated into the speech act of the matrix clause, the antecedent in a factual conditional has an independent illocutionary force."
And later:
"Factual conditionals, on the other hand, carry the presupposition that someone (other than the speaker) believes the proposition expressed by the if-clause to be true"

Factual conditionals are OK with 'then', like ordinary hypothetical conditionals. Andreas' type resist 'then' (it tends to turn them into weird-sounding hypotheticals), as biscuit/relevance conditionals do.

On the other hand, Andreas' type seem to pattern with the factuals in presupposing the antecedent in one way or another, or having two separate illocutionary forces, while hypotheticals and relevance conditionals do not, if I am not mistaken.


Actually there's at least one more type, the kind that are better rephrased with 'even if' or 'still' or both:

(Even) If the US halts the bombing, then North Vietnam will (still) not agree to negotiate. (Stalnaker)

nick said...
This comment has been removed by the author.