Quick summary:

It may seem like switching the verbs in a sentence like "Stanley seems to be alone" to "Stanley wants to be alone" shouldn't cause too many changes in how the sentence is structured syntactically. However, since "seem" and "want" belong to different verb classes, swapping causes a lot about the sentence to change! "Seem" is a raising verb: it can take Stanley from the lower clause, and move it up to the front of the sentence. "Stanley" is more associated with "alone" in terms of its theta role in the sentence than "seem"; in fact, "seem" doesn't have a role to hand out to its subject, as can be seen from the availability of the meaningless "it" as a subject, instead: "It seems Stanley is alone." The "it" doesn't do anything beyond filling in the subject position; there is no "it" doing any seeming.

On the other hand, in "Stanley wants to be alone", Stanley is the one doing the wanting... but he's also the one being alone. And that poses a problem: how can Stanley have two roles at once? That's not allowed under theta theory - every noun can only do one thing. So rather than violate that, we propose that "want" is a control verb. Unlike with "seem", in a sentence like ours here, "Stanley" never started out in the lower clause at all. Instead, Stanley started off in the main, upper clause, and controls a silent pronoun, which we'll call PRO, in the subject position of the lower clause: "Stanley wants PRO to be alone." But it's not just that this lets us keep the one noun / one theta role idea! There's also other evidence for the existence of PRO, like the differing behaviours of seem vs. want with regard to idioms, or phrases that have a special meaning unrelated to the meaning of the parts that make them up. "Seem" as an upper clause verb doesn't disrupt idiomatic meanings, but "want" does, because sentences with "want" have a silent sentence in the middle of the idiom chunk. Try forming sentence with "The apple (seemed / wanted) to fall far from the tree." Only one of them has the idiomatic meaning still.

So even if on the surface, the two sentences look similar, underneath, there are some big differences. It just goes to show, it's worth looking under the syntactic hood!


Extra Materials:

In discussing the effects that our word choice has on the syntax of our sentences, we made the distinction between subject raising and subject control. The first process involves the subject of a main clause starting off somewhere lower in the tree — perhaps as the subject in some other clause altogether — and then making its way up the tree. The second process looks similar on the surface, but instead involves the subject of a main clause starting off high up in the tree and staying there, while kind of ‘remotely controlling’ the meaning of some lower, unpronounced pronoun whose presence is indirectly detectable.

    (1) Raising (movement with a trace): [IP Stanleyi [VP is likely [CP ti to play the game]]]

    (2) Control (with a silent pronoun): [IP Stanleyj [VP is eager [CP PROj to play the game]]]

As the first partially bracketed diagram shows, Stanley starts off as the agent of the verb “play” in the lower CP, and then moves up to the main IP to become the subject of “likely”, due to the requirement that all English clauses have subjects. In the second case, Stanley is the agent — or more accurately, the experiencer — of being eager in the main IP, while his silent accomplice PRO does the job of being the agent of “play” in the lower CP.

But you might ask whether other kinds of raising or control exist, and you’d be right to! Both object raising and object control can happen, too. Let's start with object control! This involves the object of a verb controlling the meaning of the silent pronoun PRO in some lower part of the tree. For this to happen, we need to look at ditransitive verbs — verbs with two objects, where one does the controlling and the other contains the ‘controllee'. Take the verb “convince”.

    (3) The Narrator convinced [NP Stanley] [CP that all the other employees were gone]

Here, it’s clear that the verb really needs two objects: someone to be convinced of something, and the thing that someone actually comes to believe. But, we can switch that second clausal object for a clause without any obvious subject and arrive at the following:

    (4) The Narrator convinced [NP Stanleyk] [CP PROk to avoid the stairs]

The overall structure has remained largely intact, but now the subject of the embedded clause has been replaced with a silent PRO, whose referent is Stanley.

We know that this can't be an instance of raising, because it fails our idiom test. Taking a look at the sentence in (5), it seems as if the intended meaning, the one where somebody's past misdeeds have come back to haunt them, is unavailable. It just sounds like the Narrator's feeling a bit lonely and wants some company -- even if he has to settle for birds.

    (5) The Narrator convinced the chickens to come home to roost.

Also, Stanley and the following clause can’t all be part of the same embedded structure, with Stanley as the subject. It’s clear from constituency tests that “Stanley to avoid the stairs” just doesn’t form a single contiguous phrase. For instance, it can’t be moved to the front in a cleft sentence, which is why (6) sounds just horrible.

    (6) *It was Stanley to avoid the stairs that the Narrator convinced.

So, “convince” is one of a set (along with “ask” and “tell” and “persuade”) of ditransitive object control verbs.

Of course, object raising exists, too! It just usually goes by the name passivization. That’s right! Verbs in the passive voice essentially behave like raising verbs. That’s because when a verb takes on that little passivizing suffix (often -en or -ed), as in “was written” or “was destroyed”, its agent kind of gets “fired” from the job. That is, the normally obligatory doer of the action now becomes entirely optional.

    (7) “The Narrator wrote the story” “The story was written (by the narrator)”

But passive sentences still have subjects, because no matter whether it's active or passive, English still isn’t a null-subject language. So, the object of the verb can get ‘stolen’ from down below, in much the same way that the subject of a raising verb gets pilfered.

Passive sentences even have the option of using an expletive pronoun — “it” or “there” — in place of the recently demoted agent noun phrase (though you might notice that the object ends up preceding the verb anyway, which suggests that verb phrases are more complex than we thought). Have a look at the expletive passive sentence in (8):

    (8) “The player took many paths” “There were many paths taken

And we know that movement must be taking place, because passive sentences pass our idiom test. The sentence in (9) retains the intended meaning of some secret being divulged.

    (9) “Someone let the cat out of the bag” “The cat was let out of the bag”

Thus, by way of subject raising, passivization, and the various flavours of control, we find that language ends up exploiting all the possible combinations of verb behaviour! What’ll it think of next?



So how about it? What do you all think? Let us know below, and we’ll be happy to talk with you about raising and control of whatever kind you like. There’s a lot of interesting stuff to say, and we want to hear what interests you!

blog comments powered by Disqus

Previous Topic:                                                                                                                                                                                                            Next Topic:

Future Tense                                                                                                                                                                                                           Coming soon!