Agreement is pervasive in some languages, and absent in others. Chinese has no agreement whatsoever, English has next to none. The most common type of agree-ment is that of verbs with their subjects. Some languages even have the verb agree with the direct object (Hungarian, Mordvin (a language spoken in Russia, related to Hungarian), Potawatomi (an American Indian language)). Other languages have the verb agree in addition with the indirect object (Georgian). Agreement is typically in person and number, but often also in gender. Above we have seen that definiteness can also come into the picture. Adjectives sometimes agree with the nouns they modify (Latin, German, Finnish), sometimes not (Hungarian). There is no general pattern here. This is one of the things that one has to accept as it is.
Even though there is a way to account for the syntax of questions in terms context free rules, by far the most efficient analysis is in terms of transformations. Typically, a transformation is the movement of a constituent to some other place in the tree. This lecture explores how this works and some of the conditions under which this happens.
Movement
We have learned that in English the transitive verb requires its direct object im-mediately to its right. This rule has a number of exceptions. The first sentence below, (235), displays a phenomenon known as topicalisation, the second, (236), is a simple question using a question word.
Air pilots Harry admires.
(235)
Which country have you visited?
(236)
We could of course give up the idea that the direct object is to the right of the verb, but the facts are quite complex. For example, no matter what kind of constituent the question word replaces (subject, object, indirect object and so on), it is at the first place even if it is not the subject.
Alice has visited Madrid in spring to learn Spanish.
(237)
What has Alice visited in spring to learn Spanish?
(238)
Who has visited Madrid in spring to learn Spanish?
(239)
When has Alice visited Madrid to learn Spanish?
(240)
Why has Alice visited Madrid in spring?
(241)
We see that the sentences involving question words differ from (237) in that the question word is in first place and the verb in second place. There is a way to arrive at a question in the following way. First, insert the question word where it ought to belong according to our previous rules. Next, take it out and put it in first position. Now move the auxiliary (/has/) into second place. We can represent this
as follows, marking removed elements in red, and newly arrived ones in blue:
Alice has visited what in spring to learn Spanish?
(242)
What Alice has visited what in spring to learn Spanish?
(243)
What has Alice has visited what to learn Spanish?
(244)
A more standard notation is this:
Alice has visited what in spring to learn Spanish?
(245)
What Alice has visited in spring to learn Spanish?
(246)
What has Alice visited to learn Spanish?
(247)
(The underscore just helps you to see where the word came from. It is typically neither visible nor audible.) The first notation is more explicit in showing you which element came from where (assuming they are all different). However, nei-ther notation reveals the order in which the movements have applied. It turns out, though, that this is irrelevant anyhow. (The tree structures do not reveal much about the derivation of a sentence either, and that is mostly considered irrelevant detail anyway.)
The good side about this proposal is that it is actually simple. Very little needs to be done to save the original approach. However, now we have a two-stage approach to syntactic structure: first we generate representations with a context free grammar and then we mix them up using certain rules that move constituents.
But in principle this is what we have done with phonological representations, too.
First we have generated deep representations and then we have changed them according to certain rules. Thus, we say that the context free grammar generates deep syntactic representations, but that the rules just considered operate on them to give a final output, the surface syntactic representation. The rules are also referred to as (syntactic) transformations.
Wh-Movement
Let us investigate the properties of the so-called Wh-Movement. This is the trans-formation which is responsible to put the question word in front of the sentence.
Question words are also referred to as wh-words, since they all start with /wh/
(/who/, /what/, /where/, /why/, etc.). At first blush one would think that syntac-tic transformations operate on strings; but this is not so. Suppose the original
sentence was not (237) but
Alice has visited which famous city in Mexico to wait (248)
for her visa?
Then the output we expect on this account is (249). But it is ungrammatical.
Instead, only (250) is grammatical.
∗Which has Alice visited famous city in Mexico to wait (249)
for her visa?
Which famous city in Mexico has Alice visited to wait (250)
for her visa?
It is the entire DP that contains the question word that goes along with it. There is no way to define that on the basis of the string, instead it is defined on the basis of the tree. To see how, let us note that the sentence (237) has the following structure.
(Some brackets have been omitted to enhance legibility.)
Alice [has [[visited [which famous city in Mexico][to (251)
wait for her visa]]]]?
Now, /which famous city in Mexico/ is a constituent (it passes for example the tests À – Ã). Moreover, it is the object of the verb /visited/. The word /which/ is a determiner, and the smallest phrase that contains it is the one that has to move.
Wh-Movement I.
Only phrases can be moved by Wh-Movement. What moves is the least phrase containing a given wh-word. It moves to the beginning of a clause (= CP).
This specification is imprecise at various points. First, what happens if there are several wh-words? In English only one of them moves and the others stay in place;
the choice of the one to move is a bit delicate, so we shall not deal with that ques-tion here. In other languages (Rumanian, Bulgarian, Hungarian are examples) all of them move. Second, what happens if the wh-word finds itself inside a sentence that is inside another sentence? Let us take a look.
(252) Mary thinks you ought to see what city?(deep structure)
Here the wh-phrase moves to end of the higher sentence (and notice that some-thing strange happens to the verb too):
(253) What city does Mary think you ought to see?
However, some verbs dislike being passed over. In that case the wh-phrase ducks under; it goes to the left end of the lower sentence.
∗What city does Mary wonder you have seen?
(254)
Mary wonders what city you have seen.
(255)
So, let us add another qualification.
Wh-Movement II
The wh-phrase moves to the beginning of the leftmost phrase possi-ble.
We shall see further below that this is not a good way of putting things, since it refers to linear order and not to hierarchical structure.