March 08, 2007

English words for snow


A news bulletin on NPR's Morning Edition (3/6/07) about Juneau, Alaska, discussed snow there and ended with:

Four to six inches were expected before turning to rain.

I sent this on to the Fellowship of the Predicative Adjunct as being perfectly clear, but nevertheless very troubling, and noted that it was not improved by supplying an overt antecedent snow for the missing subject in before turning to rain:

Four to six inches of snow were expected to fall before turning to rain.

You still keep reading it as saying that the inches of fallen snow would turn to rain.  So something more than simple modifier-dangling is going on here.


Let me shift to a somewhat simpler example:

(1) Four inches of snow had accumulated before turning to rain.

Awful.  We can fix things by restoring the missing subject:

(2) Four inches of snow had accumulated before (the) snow turned to rain.

Now notice that the two occurrences of snow in (2) have somewhat different senses: the first occurrence refers to snow on the ground, fallen snow, while the second refers to snow in the air, falling snow.  I'll call these the APUT sense and the QANIK sense, respectively, using the corresponding West Greenlandic Eskimo roots as labels.  (Surely after all those years of our talking about the Eskimo words for snow here on Language Log, everybody will recall the two West Greenlandic roots.)  The problem with (1) is that we're trying to use APUT snow as an antecedent for an omitted element that must be understood as referring to QANIK snow.

This problem is parallel to the problem with

(3) The chicken looked delicious in the stew, despite having been raised in inhumane conditions.

where we have an occurrence of chicken 'chicken meat' being used to supply the meaning of the missing subject in despite having been raised in inhumane conditions, which involves a kind of domestic fowl, rather than the meat of these birds.  As Jerry Sadock and I, building on observations by George Lakoff, noted three decades ago -- the paper is now available here -- both explicit anaphors and omitted elements understood anaphorically (as in the predicative adjuncts above) can be used to detect ambiguities in expressions. I conclude from the problematic (3) that there are (at least) two different, but homophonous, words chicken; one of them can't stand in for the other.  And from the problematic (1) that there are (at least) two different, but homophonous, words snow; one can't be used to supply the meaning of the other.

In other (looser) words: English, like West Greenlandic, has two basic words for snow.  It just happens that in English the words are homophones, while in West Greenlandic they're not at all similar phonologically.

A little complicating twist: using APUT snow as an antecedent for an omitted element that must be understood as referring to QANIK snow gets you in trouble, but the opposite use isn't nearly so bad:

(4)  The snow reduced visibility to only a few feet, before accumulating to almost three feet in our driveway.

What's going on here, I think, is that a reference to QANIK snow permits an inference to APUT snow, since falling snow results, very quickly, in fallen snow.  Snow in the air is just a moment away from being snow on the ground.

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at March 8, 2007 01:46 PM