Language Log

A post post post toast post toast post toast post toast post

In his original toast post Eric observed that the slippery piece of toast that always hits the floor jelly side down is odd, or at least implies multiple misfortunes for each piece of toast. In my post toast post toast post I noted that the problem is that always is stuck deep in a relative clause, and cannot manage to get hold of the piece of toast at the head of the clause. It would need to do that in order to get wide scope, so as to say for each slippery piece of toast that it hit the floor jelly side down.

But Mark was not satisfied. In his post post toast post toast post toast post he pointed out that sometimes always does seem to do the job that I apparently said was impossible.

Here are two of Mark's examples:


Is there some difference between the slippery piece of toast that and a light and crispy toast that which allows always to get wide scope in the second case, but not in the first? The answer is no. But there is a crucial difference between the examples which gives a similar effect. The slippery piece of toast has a limitation that a light and crispy toast lacks.

Don't get me wrong: I like pieces of things: that piece of paper, a piece of the Berlin wall, and of course, the slippery piece of toast. Nothing against them, no siree. They come in all sorts of shapes and sizes, and, they are almost invariably just delightfully tangible. Oooh, you know what? I'm holding a piece of something in my hand right now and crinkling it.... yum. Even the ones that aren't tangible are eminently givable. I would love to give Mark a piece of my mind. And some of them are so terribly useful. Don't know what we'd do without them. So I hate to criticize the little lovelies. But no more beating about the bush, I'll just come right out with it. Pieces of things don't make good natural kinds. (Ouch: it was no fun saying that, but sometimes as a professional linguist you've just got to ignore your own personal feelings and come right out with the facts.)

Here are some potential kind denoting expressions: dinosaurs, tall buildings, freak thunderstorms, Danish crispbread, Mark Liberman's blog postings, the long vacation, toast. I say "potential" cos while you can use these expressions to refer to a kind, you can also use them to refer to sets of things, or, in the case of crispbread, a mass of the stuff. What's the difference between a kind denotation and a set or a mass? The ultimate source on this question is Greg Carlson's 1977 UMass PhD dissertation Reference to Kinds in English. There are ways to tell whether you've got a kind NP on your hands, since some predicates like kinds, but not other things. The classic one is is/are extinct, but I prefer is/are more common than ever, and has three subtypes, which are more general in their applicability.


What sorts of NPs can't refer to kinds? Well, here are some: Mark, my cat, a certain friend of mine. You can tell they are not good kinds, because they don't fit in at all well with my two favorite kind predicates:


So now we come to all those pieces of things. These, unfortunately for the original slippery toast example, do not make good kinds:


Here is a slight complication: Elster, the author of the original slippery toast sentence, really was trying to conjure up the notion of a kind, i.e. an awkward kind of food. But I believe he was trying to conjure up that kind by referring to an imagined prototypical example. If it refered to anything, I guess the slippery piece of toast would refer to this prototypical exemplar of the kind, not to the kind itself. This is an awkward idea to get straight, and given the stark results of the above kind-denoting predicate diagnostics, I'm just going to gloss over it. Grant me that the slippery piece of toast cannot refer directly to a kind. So what? We'll have to get a little technical now. Consider an example Mark found:


Here the indefinite a fun and fruity wine can be paraphrased as a fun and fruity kind of wine. But always does not take wide scope over this indefinite, which would allow it to quantify over kinds of wine. That is, the meaning of the relative clause in this example does not look like the following gloss, even after pragmatics has filled in the fact that we are talking about events in which someone drinks or tastes the beverage in question:


No,  what always quantifies over is events involving that particular kind of wine. The meaning of the whole NP a fun and fruity wine that always pleases us can be given as:


The meaning of the relative clause, RC, should then be:


The crucial point here is that you don't drink or get pleased by kinds as such. Rather, you drink or get pleased by exemplars of the kind. That's what it means to drink or get pleased by a kind. So because of the way we interpret predicates applied to kinds, the meaning of the relative clause, RC, is equivalent to:


Let's check one of the above toast examples from Mark, a light and crispy toast that always hits the floor marmalade side up. This must mean:


Here RC is:


Hey presto - a light and crispy piece of toast gets to do something a slippery piece of toast couldn't! I realize that I haven't been able to explain this puzzle without introducing some technicalities, and I've left many further complications out of the discussion. None the less, I hope the outlines of the explanation are clear. Here is the whole story in five hard and fast bullets best taken straight to the head:

• The slippery piece of toast is not kind denoting;
• A light and crispy toast is kind denoting;
• It isn't kinds that hit the floor but exemplars of the kinds;
• Always cannot quantify over the kind introducing expression, but since events involving the kind actually involve exemplars of the kind, always manages to quantify over those exemplars;
• That's a neat trick. Try to remember it next time you are trapped on a scope island with no toast.

Posted by David Beaver at July 11, 2004 04:51 AM