Syntax I

I didn’t provide much detail on the key for the practice final about the answer to the Glishen problem, and I was asked if I could elaborate. So, I will. To read this elaboration, click on the “(more…)” link below.
(more…)

By accident, I seem to have left some references to sentence (18) in the statement of Task 6 on the extra credit problem about Trees are easy to draw. Those should refer to sentence (29)—that is, the question is about how himself in (29) satisfies Principle A, and refers at one point to for John, which is also in (29).

Sorry about the typo.

So, perhaps you’re trying out the Topicalization tests, and you may well be finding that the judgments are a bit harder to make than anticipated. Let me give you a couple of notes about judgments, if this helps.

Our theory is such that at this point we only have a way to distinguish “grammatical” from “ungrammatical”, but yet our intuitions on the “goodness” of a sentence are significantly more squishy than that. One problem is that there is kind of an unavoidable tendency to give a sentence the benefit of the doubt, grammatical until proven otherwise. It’s clear what the sentence means, so it is sometimes possible to overlook that it might not be a great sentence. Still, if you can only imagine it being said by Yoda, that doesn’t really qualify it as a good English sentence.

It’s also worth being aware that wh-islands are perhaps extra-squishy, and it’s relatively easy to find examples of a wh-island violation that sound pretty good. So, for example, (1) and (2) don’t really sound too bad, although they are wh-island violations.

(1) ?Which movie did you forget whether you returned to Blockbuster?
(2) ?What did you wonder who stole? 

On the other hand, (3) sounds gut-wrenchingly horrible.

Scenario: John stole a bunch of things. Your friend remembers almost all of the details, what he stole, how he stole them, but she says that there’s one method that he used that she’s now forgotten how he used it — that is, what he stole using that method. You wish to inquire what the method was.

(3) *How did you forget what John stole?

Given that, a good strategy might be to first consider how bad the wh-question version of your topicalization test sentence is. If you find that the wh-island violation doesn’t sound too bad in a wh-question, then it’s not going to be very surprising to find that the topicalization doesn’t sound too bad either, regardless of whether topicalization is constrained by islands or not. So, you want to try topicalization out on those sentences for which wh-questions sound worst when the wh-phrase comes out of an island.

This is true with most of the islands, some are worse than others. Also, don’t forget that when you are checking complex noun phrase islands, the complex noun phrase needs to be definite. So, with either the or my or something (3-5)—indefinite DPs don’t seem to act as islands (6-7).

(4) *What did John read the book about?
(5) *What did John read my book about?
(6)  What did John read a book about?
(7)  What did John read books about?

You might also try making the sentences a bit more complicated, that can exacerbate any slight ungrammaticality. So, (8) and (9) are both adjunct island violations with basically the same structure, but (9) sounds worse. Well, somewhat, anyway.

(8) ??What did John laugh before he told Mary?
(9) ?*What did John rapidly start feeling nauseous before he told his best friend?

What the exercise is really about is coming up with sentences that would tell you whether topicalization seems to be constrained by the same islands that wh-movement is constrained by. Your final conclusion is not that important, except insofar as it follows from the judgments you report, although—as with wh-movement—there’s kind of a canonically accepted answer out in the field.

The last question on homework 11 asks you to test whether Topicalization is subject to island constraints by constructing some test sentences. Let me just say a couple of words about this to get you on the right path, if you’re not sure exactly what this question is asking for.

First, as background, we have been looking at wh-movement, which is a process by which a wh-word moves from the position where its case is checked to SpecCP. Wh-movement happens because C has a [uwh*] feature, requiring that a phrase with a [wh] feature move into SpecCP. In the process, we’ve discovered that the movement can’t escape certain constituents—the islands. So, wh-movement is impossible to a SpecCP outside when the wh-word is inside an interrogative clause (wh-island), inside a definite DP (CNP island), or inside an adverbial clause (Adjunct island).

The idea of the question on the homework is to consider a different kind of movement operation, topicalization. In a sense, it’s very similar to wh-movement—there’s a feature on C ([utop*]), requiring that a constituent with a [top] feature move into SpecCP. The question is asking you to test to see if movement to SpecCP for the purposes of topicalization is also constrained by the three islands that constrain movement of wh-words to SpecCP in wh-movement.

To do the test, you basically try putting the thing that would be topicalized inside an island, and then try to form the topicalization.

To give an example without giving away the problem, I can show you what you’d do to show that a normal declarative CP is not an island…

First, it’s not an island for wh-movement, which is demonstrated by the following question:

(1) What did John say that Mary bought?

Here, what starts off inside the embedded CP (as the object of bought) and moves up to the main clause SpecCP (and, as we’ve discussed in class, stops off in the intermediate SpecCP on the way). But the point of the question above is that it is possible to move from within a declarative complement clause to a higher SpecCP.

To show the same thing with topicalization, you’d put the element to be topicalized in an embedded declarative clause, and then topicalize it, like:

(2) Cheese, John said Mary bought.

Topicalization usually sounds a little bit better with contrast, so you’d probably want to think of a sentence like that as being followed by a sentence like But wine, John said Mary didn’t buy.

The topicalization sentence is good, from which we can conclude that topicalization (like wh-movement) can escape an embedded declarative clause. An embedded declarative clause is not an island.

For the problem itself, the idea is to come up with sentences like (2), except where the topicalized element originates in a position inside an island. To fully answer the question, it is necessary to judge whether the resulting sentence is good or not, and that task might be a bit more demanding if English isn’t your native language. You could run the sentences by native speakers (again, if you provide a contrast with a second sentence, like I exemplified above, it will sound better) if you aren’t confident, but the main thing in the problem is the construction of the test sentences, regardless of whether you feel comfortable judging their status.

I’ve posted two practice finals, which work the same way as the practice midterms did. One of the practice finals was actually last year’s final, the other is the practice final I gave out last year. The keys are also there, although of course this will be much more useful to you if you first try to work out the problems without the key, and then check to see how you did.

These are not for turning in, but they’re there for the purpose of giving you an idea of what to expect, and for inspiring questions that you might want to try to deal with next week.

I announced this in class, but for the record: if you choose to do the extra credit (Trees are easy to draw) and turn it in to me sometime prior to the final, it can count for up to two homework scores.

Here’s how that works: If you basically ace that extra credit assignment, that’s two √++es in place of your second- and third-lowest homework scores. Partial credit is possible, so even if it doesn’t raise your lowest homework scores all the way to that level, it will improve them.

But the limit on extra credit for homeworks is nevertheless two homeworks’ worth. This means that if you also turned in BUCLD talk summaries, the potential available extra credit for the new extra credit assignment is less, but then again, it’s also easier to reach √++es on both of them. If you happen to already have √++es on everything, then the extra credit won’t actually result in anything extra (but it will still have a certain character-building value, I suppose).

This is kind of complicated, sorry, but I want it to be relatively fair. So, in summary, you will have 11 homework scores total by the end of the semester. The lowest one will be forgotten. The next two lowest ones are vulnerable to being improved/replaced by turning in the extra credit assignment(s). Your highest 8 scores are whatever they are. Maybe it’s not that complicated after all.

Just in case you hadn’t noticed, the fact that she and Mary can’t be the same person in the question Which song about Mary does she like? is astounding. In case you’ve been so busy checking features and finding PRO that you haven’t had a chance to observe what an intricate set of (largely correct) predictions this model of syntax is making about our knowledge of language, it might be worth taking a moment just to marvel at it.

Ask your relatives why she can’t be Mary. They’ll be surprised that it can’t, and then if they’re so inclined to think about it, they’re not going to remember ever being taught that.

By the way, just a quick note. In the last class, we met “phases”, which have as a consequence that a wh-word that moves from a lower clause to a higher clause has to stop off partway up. You didn’t officially know this prior to homework 10, so it’s not a problem if, say, you already gave it to me and your wh-words moved in one fell swoop. Or, for that matter, if you give it to me Tuesday and they do.

I’ve finally completed the complex extra credit assignment that I was trying to have ready for Tuesday (you can download the PDF file).

It’s a relatively long excursion into what we need to suppose in order to draw a tree for the sentence Trees are easy to draw. I quite like it, actually—but it’s significantly more challenging (or at least more involved) than any homework assignment so far. It also is quite wide-ranging, making use of topics we’ve covered throughout the semester, from constituency tests up to phases. Honestly, if I’d written it over the summer, it would probably have been a take-home final.

I’m still debating how best to count this, and I’ll announce my decision about it in class on Tuesday. It will probably be worth more than just a single homework score, given how involved it is.

There is no due date on this problem, apart from the end of the semester.

I’ve gotten enough last minute questions on this one that I think I want to give you a little bit longer on this last homework. So, you can hand it in after break, although if you’re happy with it (if you understand it, you’ll probably know, there aren’t really that many “tricks” involved) and want to give it to me today, fine.

I was never planning on assigning a homework over this break anyway, so this won’t put anyone behind, really.

I was, however, planning to release an “extra credit” assignment that you could do, if you so desired. I had what I thought was a pretty nice one worked out concerning the (famous, old, and practically insoluble problem of how to analyze sentences like Trees are easy to draw), but by the time I got it most of the way done, I’d seen a flaw in the approach I was guiding you toward. So, I’ll have to post it later, once the break has begun. There won’t be a deadline for it, except just the end of the semester.