June 24, 2007

Nearly and almost

[This is a guest posting by Jerry Sadock, following up on Mark Liberman's two postings about "nearly no".]

These two items are almost synonymous, but not necessarily nearly synonymous. They may well be truth conditionally equivalent, "almost P" and "nearly P" both entailing "not P" (Atlas, Horn, but pace Sadock) and "close to P". Thus "Eric knows almost 500 languages" and "Eric knows nearly 500 languages" would both be false either if Eric knows 501 languages or if Eric knows no more than 400. But there is some kind of difference between them as revealed by the differential goodness of Almost no one was there and Nearly no one was there.

The difference would then have to be a difference in nuance, or connotation, or (more technically) conventional implicature. That difference, it seems to me, has to do with expectations: Nearly n connotes that n exceeds (hence is better than) what was expected or hoped for, while almost n does not conventionally connote any particular desire, hope or expectation, but easily supports a conversational implicature to the same effect as the conventional implicature associated with nearly. So, for example, if I say I have nearly $10 in my wallet I suggest that that's a lot and since it isn't, the sentence is strange in most contexts. Compare Molly has nearly $10 in her piggy bank, a lot for a three year old, perhaps. But if I say I have almost $10 in my wallet, while not a lot for a man of my means, it would be a fine thing to say if we each wanted a tall iced latte and you asked me how much I had on me. As another example, almost tolerable is much more natural than nearly tolerable since the latter suggests that being merely tolerable is better than we could have hoped for. In the right context, however, it's OK. I could describe the noonday temperature in Tucson in July as "nearly tolerable" if it was 101º since I expected it to be absolutely intolerable, 113º say, and something approaching tolerable is better than that.

If this is so there is a reason for the distinction in out-of-context acceptability between examples like Almost no one was there and Nearly no one was there. The first can be a hedged estimate, pure and simple, and hence it's OK. The second suggests that zero attendance exceeds our expectations, which in most contexts is odd. What did we expect, that a negative number of people would be there? But cobble up a context in which zero does exceed expectations, and it becomes OK. This would ordinarily involve scale reversal, because negative quantities exist only in the realm of mathematics. Let's say I've organized a boycott on Humvees. I hope and expect that my campaign will be successful and only a small number will be sold. My expectations would then be exceeded if none are sold and then Nearly no Humvees were sold last month seems fine in that context.

Some of the examples of nearly no on the web clearly show such reverse expectation or hope, like this one:

Nearly none of the known near-earth objects have any chance of hitting the earth, and there are just two or three such objects that researchers can't dismiss yet, said Marsden. (link)

One would hope that one of these babies won't slam into our planet and send us the way of the dinosaurs. But there are a lot of these puppies out there, so the best we could hope for is that only a few stand a chance of closing our chapter for good. None would be a big relief -- the most we hoped for, so to speak. The last clause in the quotation, there are just two or three such objects that researchers can't dismiss yet, is revealing. It shows that there's an active effort to get the number down, hence the scale reversal.

Posted by Arnold Zwicky at June 24, 2007 02:02 PM