Ravens on the garden path
« previous post | next post »
I just ran across a particularly impressive garden path sentence in Bernd Heinrich's book RAVENS IN WINTER (p. 268); it took me several tries to get this sentence to parse grammatically:
"Even the wolverine is said to do nothing to drive ravens off that land beside it and steal its food."
(Of course parsing is no problem if the sentence is spoken. But in written form, for me at least, "and steal its food" just didn't seem to fit at first. My mis-parse was reading "off" as the head of a prepositional phrase.)
Y said,
June 11, 2023 @ 3:49 pm
On my first reading, I thought it was the land's food the ravens were stealing. I didn't understand what that meant.
A comma after "off" would fix the sentence nicely.
tim1724 said,
June 11, 2023 @ 3:54 pm
If they’d written “to drive off ravens” instead of “to drive ravens off” it would have been much more clear.
Joshua K. said,
June 11, 2023 @ 4:32 pm
It took me multiple readings of both Y's and tim1724's variations to parse them, too.
To those who remain baffled by this sentence, here's my (less elegant, but hopefully clear) variant: "It is said that when ravens land beside the wolverine and steal its food, the wolverine doesn't do anything to drive them off."
Paul Clapham said,
June 11, 2023 @ 4:36 pm
I wa parsing "land" as a noun as well. Also the wrong garden path.
J.W. Brewer said,
June 11, 2023 @ 4:42 pm
Agree with tim1724 as to the source of the problem, and as to what would have been the more natural way to express the point. Maybe for a one-word COMPLEMENT there's no meaningful difference between "drive off COMPLEMENT" and "drive COMPLEMENT off," but things are otherwise when the COMPLEMENT is a lengthy phrase and you try to break it up with the second bit of the phrasal verb stuck into the middle of it.
Robert Coren said,
June 11, 2023 @ 5:00 pm
I got this one at first reading too, although I could see how "off that land" could lead one into the weeds. But I agree with tim1724 that "drive ravens off" is bad diction, and the "off" should come before the birds.
Taylor, Philip said,
June 11, 2023 @ 5:02 pm
I have to say, the sentence was entirely transparent to me. But given Sally's observation that "parsing is no problem if the sentence is spoken", this may be because I instinctively verbalise as I read — the written word becomes a spoken word in my head, and it is the verbalisation that usually creates the mental model rather then the written text.
JPL said,
June 11, 2023 @ 6:32 pm
Sally:
Nice one!
Philip:
I think that still assumes that you have chosen the right path on the first try, since the sequence of expressions "off that land" in the serial unfolding of the written text is objectively ambiguous wrt syntactic function, but resolvable once you reach the end of the sentence.
For a fix, one could choose an unambiguously verbal (or at least non-nominal) expression, such as "alight", instead of "land". I always think it's best to make the fix as minimal as possible; here I suspect perhaps the prior text mentioned "ravens", and that determined the splitting of the verb-particle expression. I wouldn't necessarily want to always eliminate the ambiguity, as you may have noticed. Maybe I want the reader to pay attention and read the sentence carefully, all the better if the reader realizes that she has to go back several sentences later.
Bloix said,
June 11, 2023 @ 7:00 pm
IMHO, the best solution involves inserting a definite article: "to drive off the ravens that land beside it and steal its food."
JPL said,
June 11, 2023 @ 7:15 pm
PS: So how does the reader notice that the alternative interpretation (or parsing) is the wrong one, since according to the purely syntactic rules there is really nothing wrong with the sentence (either way)? How is that realization possible? (OK, what exactly is it that "doesn't make sense"?)
Sally Thomason said,
June 11, 2023 @ 7:33 pm
JPL, But there is something wrong syntactically on the garden-path reading! When I read it as the wolverine driving ravens "off that land beside it" — that is, off the land beside the wolverine — there is no referent for "its" in "its food". It can't be the ravens' food, because "ravens" is plural; it'd be O.K. syntactically if it involved the wolverine doing nothing to drive ravens off that land beside it [it = the wolverine] and steal their [= the ravens'] food, but that's not what it says. In the sentence as written, "its food" can't possibly refer to the wolverine's food, because the wolverine couldn't reasonably be supposed to want to steal its own food. (Personally, I'm impressed with the audacity of those ravens: encroaching on a carcass occupied by a huge ferocious weasel would take a lot of guts. I have never seen a wolverine in the wild, but I spend a lot of time in Montana, so I haven't given up hope.)
Garrett Wollman said,
June 11, 2023 @ 8:20 pm
I wonder if the correct parsing is more obvious to a native German speaker (like Heinrich himself — his family came to the US in 1951, when he was 10 or 11). (I've read several of Heinrich's books including his biography of his father, THE SNORING BIRD, but I don't recall stumbling over English usage.)
cameron said,
June 11, 2023 @ 9:58 pm
the German connection is interesting. the idiom is "to drive off". I could imagine that being calqued into German as aufdriven, with separable prefix. in this case the sentence is clearer in English if the "off" stays next to the "drive": to drive off ravens.
but I could see how a native German speaker would want the prefix to migrate in the direction of the end of the clause
JPL said,
June 11, 2023 @ 10:52 pm
Sally:
My "syntactically OK" alternative (garden path) interpretation admittedly requires a figurative or "poetic" context: the antecedent for "its" is "the land" ("the land's food"), and the food it grows on itself, which is meant for (in the fantasy context) non-carnivorous animals like deer or birds, not for wolverines. But given that same syntactic form, there must be other contents that would fill the category specifications without problem (e.g., "off that picnic table", where the human picnickers have temporarily abandoned it, and the wolverine leaves the stealing of the food to the ravens). I thought that the alternative path might have been compelling initially because formally (i.e., only wrt categories and including agreement and anaphora) its syntax is OK, but the clash with reality and sense comes from what is expressed.
What I was wondering is, were "ravens" introduced in the text preceding that sentence ("even the wolverines" seems to imply another, less surprising, case with ravens, with "ravens" in given information status)?
Yuval said,
June 12, 2023 @ 2:21 am
Also for the nonprojectivity files!
Sally Thomason said,
June 12, 2023 @ 9:44 am
JPL, Ah, now I see what you mean. Of course you're right. The preceding sentences are about ravens not being driven off by other predators/scavengers — coyotes, for instance. Heinrich, like (I assume) anyone who knows anything about wolverines, would have expected wolverines to be less hospitable than other carnivores in rural Maine.
Philip Anderson said,
June 12, 2023 @ 1:22 pm
@Bloix
But your solution hasn’t just inserted a definite article, it has brought the “off” forwards to after “drive”, which is what tim1724 suggested, but without the article; both sound correct to me, though I prefer it with. But “drive the ravens off that land” doesn’t change anything for me.
A simple “… drive the ravens off.” is fine, but “off” binds more tightly to the immediately-following “that land” than the earlier, but separated “drive”, with no comma or spoken pause.
Philip Anderson said,
June 12, 2023 @ 1:26 pm
My initial interpretation was that in “that land beside it”, the “it” referred to some previously-mentioned place, like a river, which the land was next too.
JPL said,
June 15, 2023 @ 6:04 pm
The comments on this one seem to have dried up, but, before it disappears "below the fold", I just want to register that this is a beautiful example, sort of a real-world, "in-the-wild" case of Chomsky's "colorless green ideas" sentence, but also involving the phenomenon of syntactic ambiguity. It allows us, I think, to reexamine from a different angle the significance of such sentences and to understand the mechanisms connecting perceivable phonological shapes and propositions, both on the level of acts of language use and on the level of the historical construction of the general systems (norms for natural language speech communities?) that make the acts possible. I'll stop so far, but I think it's a big open question. (Including questioning the idea that, "I think we are forced to conclude that grammar is autonomous and independent of meaning", and asking whether or not it is a case of hysteron proteron. It depends on what you take as the puzzle.)