I talked with a few people today about what exactly I was after in part 4, so let me say a couple of things about that here (I know it’s kind of late relative to when the homework is due, but I’ll try to post comments like this as soon as someone asks about them).
First: don’t forget about the previous post I made about this: There is a typo in the instructions when I talk about what meanings you should get in Parts 1-3. Parts 1 and 3 have the same meaning, Part 2 has a different meaning.
Second, none of these constituency tests are going to fail outright. You’ll be creating 6 sentences, 2 each for parts 1-3, and they should all be grammatical sentences. But there is a nuance when it comes to what your test sentences can mean, which is what is explored in part 4.
So, here’s the idea about part 4:
The sentence in the homework has a structural ambiguity that arises from the fact that the phrase from Belgium could be attached in a couple of different places. For simplicity, let me forget about the chocolates and concentrate on the box: I got a box from Belgium. It’s possible for from Belgium to modify box, so then what we’re talking about is a Belgian box (a box from Belgium). It’s also possible for from Belgium to specify something about the getting—I got, from Belgium, a box.
These two meanings correspond to two different syntactic structures. In the first, the PP from Belgium is lower in the tree, and in the second, it is higher. But the way you pronounce both of these trees is the same, so we feel the sentence to be ambiguous—the words in that order could correspond to either tree, so to either meaning.
Part of what is meant by having a different structure is that different things will be constituents. Something that is a constituent in one tree might not be a constituent in the other tree. And the constituency tests are a way to find out what a constituent is. So, when we are testing a box from Belgium to see if it is a constituent, we are implicitly testing to see if it is a constituent in both trees. If the test passes for the first meaning (Belgian box), then that means that a box from Belgium is a constituent in the first tree. If the test passes for the second meaning (received from Belgium), then a box from Belgium is a constituent in the second tree. But if it only passes one, then it is probably a constituent only in one of the two trees. So, if the test sentence (like It’s a box from Belgium that I got) only corresponds to one of the meanings (Belgian box), then we preliminarily take that to mean that a box from Belgium is a constituent only in the “Belgian box” tree, but is not a constituent in the “received from Belgium” tree.
So, that’s what we’re doing when we are considering the meanings that the test sentences can have—we’re trying to determine which tree the string of words we’re testing is a constituent in.
Now, one last thing: from Belgium is a constituent, in both trees. But yet when we test it, it seems that the test sentences only have one meaning (the “received from Belgium” meaning). That’s what the little note I added in on Part 4 relates to—there is an additional condition on sentences that says, essentially, you can’t topicalize or cleft something that is inside an NP. For example, if you have an NP like the book about squirrels that John bought in a sentence like I read the book about squirrels that John bought, you can’t form a cleft using squirrels (*It’s squirrels that I read the book about that John bought or form a topicalization using squirrels (*Squirrels, I read the book about that John bought), even though we know that squirrels is a constituent (since it’s just one word). The reason you can’t use either test is that squirrels is inside the noun phrase, and topicalization and clefting simply doesn’t work for constituents inside a noun phrase.
If you think this through just a little bit further, I expect that you’ll see why testing whether from Belgium is a constituent wound up unambiguously having the “received from Belgium” meaning, even though from Belgium actually is a constituent in the structures for both the “received from Belgium” and the “Belgian box” meanings.
Anyway, I hope that helps make what’s going on here seem a bit less of a mystery. I still want you to try to write out a short explanation of why the test sentences wind up not being ambiguous, but what I’ve explained here should take you most of the way through that part.