I spent yesterday at the MIT/UMass/UConn/Brown Semantics Workshop. It was lots of fun, even if it has a slightly unwieldy name. In the future the conference will be the Southern New England Workshop in Semantics (SNEWS), and we’ll try to present long drawly papers at the end to live down to the name. Next year the conference will be at Brown for the first time, and I encourage everyone in the area who’s interested in semantics to attend.

It was lots of fun seeing all these semantics papers. Despite all the presentations being by graduate students (at least I think they were all grad students, apologies if I’m misclassifying anyone) the quality was as high as I’m used to from philosophy conferences. It was also really nice to see people presenting theories with testable consequences, and then watching them get tested during discussion. Whatever their other virtues, it’s rare to see that at metaphysics or epistemology presentations.

The best presentation was by Pranav Anand of work he’s doing with Andrew Nevins. They make a convincing case that Zazaki, a language spoken in Turkish Kurdistan, contains monsters. (An early version of this paper, attributed just to Nevins, is online here.)

In Zazaki the translation of (1) can mean either (2) or (3).

(1) Melanie said that I’m an idiot.
(2) Melanie said that Brian is an idiot.
(3) Melanie said that Melanie is an idiot.

There were two surprising qualifications of this. Apparently the monstrous readings are not available for all propositional attitude verbs. So (4) must mean (5).

(4) Melanie believes that I’m an idiot.
(5) Melanie believes that Brian is an idiot.

And there is a coordination constraint on the readings of the indexicals. They must either all behave monstrously, or all behave ‘normally’. So if I’m reporting to Ted an utterance of Melanie’s to Charlotte, then (6) can mean (7) or (8), but not (9) or (10).

(6) Melanie said that I owe you money.
(7) Melanie said that Brian owes Ted money.
(8) Melanie said that Melanie owes Charlotte money.
(9) Melanie said that Brian owes Charlotte money.
(10) Melanie said that Melanie owes Ted money.

Given those results you might think that we have a confusion of direct with indirect speech reports here, but they test for that hypothesis and make some convincing arguments that no such confusion is happening. They then have some very complicated examples involving embedded reports to test various hypotheses about how to understand these monsters.

There’s been a few other papers arguing for the existence of monsters in various languages, but these results seem much more convincing than any others previously presented. The paper is just an excellent combination of theoretical and applied work, and hopefully it will be in print somewhere prominent soon.

I also learned about one other surprising result that I’d never heard before. Present a five year old with the following scenario: there are four horses, three of them are ridden by boys and the fourth is not, and there is nothing else in the picture. Then ask the child whether the following sentence is true or false: Every boy is riding a horse. A large percentage of the children will say this is not true. There’s various hypotheses about what is driving this, and a few results about what modifications of the experimental setting will improve children’s results.

The writers in this field commonly distinguish between child-like and adult-like answers to the questions, but it seemed to me that’s being too generous to adults. After all, adults make a very similar mistake in the Wason Selection Task. It would be interesting to know whether the same things that improve performance in the Wason Selection Task improve performance among children.