Syntax I

I’ve gotten a few questions about the question on the practice midterm in which you are to indicate which sentence corresponds to the structure given. The three sentences are I tripped on Commonwealth Ave., I sprinted to class, and Water spilled on me.

So as not to give it away, I’ll hide this discussion “below the fold”. To read more (after you’ve done what you’re going to do with that practice midterm), click on the “(more…)” link below.
(more…)

A couple of minor typos were called to my attention on the example derivation of Pat must not have been sending flowers to Chris that I posted a while back. Neither should have caused too much confusion, but for the record: the [uN*] feature of little v was not crossed off in the tree when it was checked, and in fact survived un-crossed-off all the way until the end. It was actually checked when the NP Pat was Merged with v′ (as noted in the commentary text), I just forgot to strike it out in the tree, and then copied and pasted after that.

Also, there were two step 5s for some reason.

I’ve revised it and will post the revised version, but there’s really no need to print it out again for these changes if you already have a copy.

I’ve talked to a couple of people who have not been completely clear about when to apply Principle B and Principle C of the Binding Theory. It’s straightforward once you see it, I think, but it always causes confusion (so it’s ok if you’ve been confused—but I hope that now you won’t be).

Binding theory is about binding, first and foremost. And binding is defined as follows. X binds Y if and only if X c-commands Y and X and Y are coindexed.

Let me repeat that first part. If and only if X c-commands Y.

Now, what you are doing when you apply Binding Theory is this: You are looking at a particular NP in a sentence. If it is an anaphor, you apply Principle A. If it is a pronoun, you apply Principle B. If it is a name, a definite NP, or pretty much anything else except a pronoun or anaphor, you apply Principle C.

To apply Binding Theory, you look at that NP. Then you look above it in the tree, that is, you look at what c-commands that NP. If there is something above it in the tree (something that c-commands your NP) that is coindexed, then you have found something that binds your NP. Then, you check the appropriate condition. If your NP is an anaphor, that something you found has to be nearby, within the same clause. If your NP is a pronoun, that something has to be far away, not within the same clause. If your NP is a name, that something will result in a violation of Binding Theory (Principle C) regardless of where it is.

Finding where Principle A is relevant is easy enough. You see -self, you think “Principle A.” But the configurations where Principle B and Principle C are relevant very often both contain both a pronoun and a name.

(1) She₁ saw Mary₁ in the photo.
(2) Mary₁ saw her₁ in the photo.

Sentence (1) is a violation of Principle C. Sentence (2) is a violation of Principle B.

Sentence (1) is not a violation of Principle B. There is a pronoun, and it is coindexed with the NP Mary, but it doesn’t matter. Mary does not c-command the pronoun in (1), so Mary doesn’t bind the pronoun. Hence, the pronoun in (1) is properly free in its binding domain. Do not confuse coindexation with binding: Binding requires c-command. Once again: X binds Y if and only if X c-commands Y and X and Y are coindexed.

Just a note, the key for homework 5 has now been posted, on the syllabus page.

If you print your handouts, homeworks, keys, and whatnot in the computer lab in the basement at 111 Cummington, be aware that sometimes the fonts don’t come out quite as they should. For example, I have seen a midterm key that was printed there, and all of the “bar” symbols had been replaced by “thetas” (so, it looked like there was a T, projecting to a Tθ, projecting to a TP—where, of course, that middle one should have been T′). Keep your eyes out for weird-looking stuff if you are printing in that lab, and know that it might be there. Ask me if you’re not sure about something you see on a printout from there.

There were a couple of errors on the Fall 2004 midterm key, and a new version is now posted (if you haven’t downloaded it yet). The errors were in 6d (V would initially also have a [uN*] feature), and 9b (the sentence should have started with It had been… rather than It has been..).

I’ve had a couple of questions about how to attach adjectives, and it’s probably worth saying something about that.

Adjectives are modifiers, like adverbs are. So, the way they attach is with Adjoin. The idea is that you have an NP, like students, and you want to modify the NP with an adjective, like happy. The result should still be an NP, though, because you can put happy students in all the same places in a sentence that you can put students. The result of adjoining happy to students would look like this:

      NP
     /  \
  AdjP   NP
 happy   students

I wrote “P” after these because they are both maximal projections—neither has any strong features left to check, so neither will project any further. You can’t in general adjoin like this until you’ve checked all of the strong features, meaning that this type of adjunction (for adverbs and adjectives) will always occur at the XP level. This would sometimes be called “XP-adjunction.”

What might happen next, after you’ve adjoined happy to students is that the NP happy students might be Merged into an argument position (to get a θ-role), e.g., as sister of V (in a sentence like I saw happy students) or as the specifier of v (in a sentence like Happy students passed the midterm). That is, once you’ve adjoined happy to students, the NP happy students functions as an inseparable unit, an NP that can get a θ-role, or move to the specifier of TP. (For whatever reason, this might be different with all and both, which do seem possible to leave behind, as in students will all come to class).

I did get an interesting question about the relationship between this and head-adjunction, of the kind that you get when you move V to v or an auxiliary to T. Head-movement (movement of a head up to a higher head) also winds up creating an adjunction structure that looks very much like the structure of happy students above. But there are a couple of differences.

For historical reasons, it is always assumed that XPs adjoin to XPs and heads adjoin to heads—you don’t have an XP adjoining to a head, or a head adjoining to an XP. This distinction made a little bit more sense back in the days of X-bar theory, where anything you take out of the lexicon projected a structure that included a head, an intermediate projection, and a maximal projection. In our theory, however, the distinction between “head” and “maximal projection” is often quite blurry, since the very same node in the tree can be both a head and a maximal projection.

When you draw a head-adjunction, of the kind that arises when V moves to v, you draw it like this:

    v
   / \
  V   v

Because v is a head here (unambiguously so, it projects to a v-bar, and its sister is a VP), when you adjoin V to it, you are adjoining the head V to the head v. This is why no P was written by V here. In a sense, this is partly because the head V already projects its features to a VP, lower in the structure. When you move the head away, it doesn’t get to project again, and since it doesn’t even have the possibility of projecting when adjoined to another head, we don’t label it as VP in the structure above. But, mainly, it’s just because the only thing you adjoin to a head is another head.

Another point about head-movement (adjunction of a lower head to a higher head) is that it differs from XP-adjunction in that head-movement does occur in order to check a feature. XP-adjunction doesn’t.

I sometimes draw head-movement as happening from V to v in the step before the step in which the Agent is Merged. Merging the Agent checks a (strong) selectional feature of v, and by the reasoning I’ve presented in class about this, it should be that adjunction should wait until there are no more strong features to check on the head to which you are adjoining. Since head-movement does check a feature (such as the [uV*] feature of v), the best we can do is make that the last strong feature checked (it can’t wait until there are no more strong features to check, because it has to check one of them itself). That is, perhaps the Agent should really be Merged first, before V moves to v. It doesn’t really matter either way, but for consistency perhaps I’ll start doing that.

Probably the best way to distinguish between head-adjunction and XP-adjunction would be in terms of the features that get checked, then. Head-adjunction (which will arise from moving a lower head to a higher head) happens only when the higher head has a strong feature that needs to be checked. XP-adjunction happens only when there are no features left that need checking. Practically speaking, XP-adjunction will be cases of adjunction of a modifier, and head-adjunction will be cases of adjunction of a moved head.

The term “complex head” really only refers to the latter case, the one where a head adjoins to another head due to movement. The structure for happy students given above is not considered to be a complex head because, although it does happen to be made of heads, the NP students actually serves as a maximal projection. It’s not going to project any further, it doesn’t have any strong uninterpretable features to check. In that respect it’s different from v.

It’s possible that all of this has not really made the situation any clearer, but I’m hoping that, if you were wondering about this, you have at least a little bit more to go on.

When I’m writing up the keys and homework, I generally use Arboreal, which is a (not free, but pretty cheap) font. With it, I can go pretty quickly and do what I need to do.

There are a few tree-drawing tools on the web too that you might look at. I’ve really only glanced at these. They may be more trouble than they’re worth, but if you’re adventurous, you can try them out. Note, though, that sometimes the assumptions they make about the syntax aren’t exactly the same as the assumptions we’re making (e.g., they might provide an “IP” where we would use a “TP”, that sort of thing).

The Syntax Student’s Companion is a Java applet that lets you draw trees. I’m not sure how to save them, but perhaps it’s possible.

The phpSyntaxTree program will convert labeled brackets into trees for you.

The TreeForm project is another tree-drawer, of the same basic type as the Syntax Student’s Companion, runs on Mac and Windows. But I can’t seem to get it to do anything, perhaps the “alpha” designation is serious.

There are others out there, if you find one that seems useful, let me know…

There is (understandably, I’d say) a little bit of confusion with respect to the representation of tense on modals, let me say a little bit about that:

Modals seem to come in pairs that are related in a way that looks a lot like the way present and past tense verbs are related. May/might, shall/should, will/would, can/could, … Plus, there is also the “Sequence of Tense” argument that Adger gives on pp. 159-160. It seems that might behaves like a past tense verb, and may does not.

When it comes to the question of the meaning, things are potentially a bit dicier. Certainly, I could leave tomorrow doesn’t seem like it is the past, isn’t that what I could have left this afternoon would mean?

I think this is a tricky thing to get a handle on, but there might be some contribution of “pastness” to could, something lacking from can. Of course, whether I can leave tomorrow or I could leave tomorrow, the fact of the matter is that the leaving is tomorrow, not now or in the past. But the tense is on the modal, not on leave. We might think of this as a difference in when the possibility was. I could leave tomorrow so long as sometime before now, it was possible that tomorrow I leave. An hour ago, for example, it might already have become clear that among the ways the world could turn out, at least one of them is one in which tomorrow I leave. Or, try, this:

Three hours ago, I bought and scratched a lottery ticket, won a brazillion dollars, sold it to buy a plane ticket to São Paulo, packed my bags, cabbed over to Logan, stood in front of the jetway door impatiently. Twenty minutes ago, somebody bumped into me and, it appears, has stolen all of my cards, money, and ID. And my plane ticket. It’s 7:40. The plane leaves in twenty minutes.

So, I could go to São Paulo at 8:00 (20 minutes from now). But I can’t now (go to São Paulo at 8:00, 20 minutes from now).

You see? I can’t express that as I can go to São Paulo because that’s just false under this scenario. It’s important for can that the possibility hold now. For could on the other hand, the possibility needs to have held in the past, even if it doesn’t hold now. Present, past; can, could.

Of course, that is perhaps a bit oversimplified. It may be easier to tell this kind of story for can/could than for other modals. There are also complex interactions, it appears, with the modal base (so if the modal has an epistemic reading, a deontic reading, etc.). But perhaps this can serve as sufficient justification for now.

That is, might, should, could, would are all past tense, they all have a [uInfl:] feature that gets valued by [tense:past] on T.

I’ve talked to a couple of people who didn’t quite feel comfortable with the idea that inside one clause there might be another. But, nevertheless, inside one clause, there might be another.

We have a Hierarchy of Projections (HoP) that says:

T > (Neg) > (M) > (Perf) > (Prog) > v > V

This shouldn’t be taken to imply that there can only be one T per sentence, however. It’s easy to see that you can have one sentence inside another: Take any sentence (for example, Shyanne is engaged to Steve) and you can put in inside another one with said (for example, Natalie said (that) Shyanne is engaged to Steve) and yet another one with heard (for example, Ripley heard (that) Natalie said (that) Shyanne is engaged to Steve), and so forth.

We’ll get to how to incorporate the word that into the structure. For the moment, we’re just dealing with nonfinite clauses in the homework.

In the main clause, there is just one T. But it is possible for a verb (say, for example, want) to have a TP complement, a whole clause. What Carlisle wants Ripley to confess means is that Carlisle wants something to be the case, and what it is this: Ripley confesses. That is, the Proposition expressed by the embedded TP is just one of the arguments of the verb want, and has pretty much the same status as the Theme cheese does in Wallace wants cheese.

So, if you have a TP inside a sentence, in argument position (where it is getting a Proposition θ-role, for example), the Hierarchy of Projections basically starts over for that TP. Inside that TP, the order of projections is T > (Neg) > (M) > (Perf) > (Prog) > v > V, but it’s independent from the sequence in the main clause.