LING 101: Syntax

This page was prepared by William Idsardi as a study resource for his LING 101 course at UDel. You may find it useful when preparing for exams. Note that the material covered by these notes may not exactly match the topics covered in LING 101 when taught by other instructors. To return to the "Study notes" page use the navigation bar at the left.

How do we COMBINE WORDS to make SENTENCES? Syntax uses trees (just as in morphology) but the trees are built on WORDS instead of morphemes.
Words are the fundamental units of sentences.
The laws of combination for words are the syntactic rules.

Morphology Review

Recall that words have internal structure (they are comprised of morphemes):

            Noun
            /   \
   Adjective    -ity
  /      \
un-    Adjective
         /    \
      Verb    -able
        |
     believe
Meaning: ity( un ( able (believe) ) )

         "The property of not being able to be believed"

Recall also that each affix is added according to a morphological rule.

Remember that the rules are specification of part of a tree:

            Noun
            /   \
   Adjective    -ity
Meaning: "The property of being Adjective"

Sentence Structure

Sentences are broken down into sub-parts (PHRASES) in the same way.
We know that there is structure in sentences separate from the meaning of the sentence because of the difference between "well formed nonsense" (1) and "total gibberish" (2):
1. Colorless green ideas sleep furiously.
2. Green sleep furiously ideas colorless.
Neither (1) nor (2) have any discernable meaning.
But (1) "sounds better" than (2)
We could perform more sophisticated psychological tests to investigate this, for example how well can people remember (1) versus (2).
We explain the difference between (1) and (2) by showing that (1) obeys the rules of English sentence structure (syntax) whereas (2) does not.

Structural Ambiguity

Recall that "un-tie-able" is an ambiguous word, with two meanings that correspond to different internal word structures:

    Adjective
     /     \
   Verb    -able
   /   \
un-    Verb
        |
       tie
Meaning: able( un (tie) ) = "can be untied"

    Adjective
   /        \
un-         Adjective
            /       \
         Verb       -able
           |
          tie
Meaning: un( able (tie) ) = "can't be tied"

The same kind of structural ambiguity can be found in sentences as well.

Consider the two possible trees for the phrase "synthetic buffalo hides"

           o
       /       \
    o            o
    |        /      \
synthetic buffalo hides
Meaning: "buffalo hides that are synthetic"
         "synthetic" has SCOPE over "buffalo hides"

              o
          /       \
         o          o
     /      \       |
synthetic buffalo hides
Meaning: "hides from synthetic buffalo"
         "synthetic" has SCOPE only over "buffalo"

Combining the same words in different ways yields different meanings.
So far syntactic structural ambiguity can explain a few jokes. But many ordinary sentences can be structurally ambiguous.
Consider: "Big cats and dogs must be on a leash."

There are two meanings for the phrase "big cats and dogs":

          o
    /     |  \
   o      o   o
  / \     |   |
big cats and dogs
Meaning: "dogs and big cats"
         "big" has scope only over "cats"

     o
  /      \
 o        o
 |    /   |   \
big cats and dogs
Meaning: "big cats and big dogs"
         "big" has scope over "cats and dogs"

The important question in interpreting the phrase is what does "big" have scope over.
Same principles apply for sentences such as:
"The boy saw the man with the telescope."

The question is: What is the scope of "with the telescope"? Does it modify only "the man" or does it modify "saw the man"?

        o
    /        \
   |          o
   |      /      \
   |     |          o
   |     |    /   /     \
   |     |   |   |       o
   |     |   |   |    /      \
   o     |   |   |   |        o
  / \    |   |   |   |     /    \
the boy saw the man with the telescope
Meaning: The boy saw the man.  The man had a telescope.

        o
    /        \
   |           o
   |      /    |       \
   |     |     |          o
   |     |     |      /      \
   o     |     o     |        o
  / \    |   /   \   |     /    \
the boy saw the man with the telescope
Meaning: Using the telescope, the boy saw the man.

Phrasal Categories

So far we have just been drawing diagrams, without naming the phrases.
Now we would like to add descriptive labels (names) to the phrases.
Words have category labels (Noun, Verb, Adjective, Adverb, Article, Preposition,...).
Phrases are constructed out of a "head" plus other material into:
- Noun Phrase (NP)
- Verb Phrase (VP)
- Prepositional Phrase (PP)
- Sentence (S)
- ...

So a tree diagram will include the phrasal labels:

        S
    /      \
   |         VP
   |      /      \
   |     |         NP
   |     |    /   /  \
   |     |   |   |    PP
   |     |   |   |   /  \
   NP    |   |   |   |   NP
  / \    |   |   |   |  /  \
Art  N   V  Art  N   P Art  N
 |   |   |   |   |   |  |   |
the boy saw the cat on the mat
Meaning: The boy saw the man.  The man had a telescope.

The SUBJECT is the NP directly under S (in this sentence, "the boy").
The OBJECT is the NP directly under VP (in this sentence, "the cat").

Recursion

Sentences can contain sentences:

      S
 /         \
|           VP
|    /             \
|   |               S
|   |       /     /    \
|   |      |     |     VP
|   |      |     |   /   \
NP  |      |     |  |    NP
|   |      |     |  |   /  \
N   V    Comp    N  V  Art  N
|   |      |     |  |  |    |
I believe that John is a student

Recursion can go on indefinitely:

This is the cat
    that ate the rat
        that ate the cheese
            that was sold by the man
                that lived in the city
                    that was on the river
                        ...

Because syntax is recursive, there is NO longest sentence.
And the number of sentences in any language is INFINITE.
Thus syntax uses a small (finite) number of rules to generate an infinite variety of sentences.

In addition to sentences inside sentences we can also find other instances of recursion: VP inside VP, PP inside PP, NP inside NP:

       NP
  /   /  \
 |   |    PP
 |   |   /  \
 |   |   |   NP
 |   |   |  /  \
Art  N   P Art  N
 |   |   |  |   |
the cat on the mat

As with recursion involving S's we can put NP's inside NP's indefinitely too:

The cat
    on the mat
        near the door
            beside the window
                ...

Constituency Tests

How can we tell what the structure of a sentence is?
How do we know where the phrases are?
How can we prove it?
How can we justify it?
The answer is to use CONSTITUENCY TESTS (constituent = phrase).
There are five basic tests for phrases:
1. Substitution by anaphors (pronouns)
2. Substitution by question words
3. Ability to stand alone
4. Co-ordination
5. Movement

Substitution by anaphors

Recall the structually ambiguous sentence, "The boy saw the man with the telescope."

(1)     S
    /        \
   |          VP
   |      /      \
   |     |         NP
   |     |    /   /     \
   |     |   |   |       PP
   |     |   |   |    /      \
   NP    |   |   |   |        NP
  / \    |   |   |   |     /    \
Art  N   V  Art  N   P   Art     N
 |   |   |   |   |   |    |      |
the boy saw the man with the telescope
Meaning: The boy saw the man.  The man had a telescope.

(2)     S
    /        \
   |           VP
   |      /    |       \
   |     |     |         PP
   |     |     |      /      \
   NP    |     NP    |        NP
  / \    |   /   \   |     /    \
Art  N   V  Art  N   P   Art     N
 |   |   |   |   |   |    |      |
the boy saw the man with the telescope
Meaning: Using the telescope, the boy saw the man.

We can substitute pronouns for the noun phrases in these sentences.
For example we can substitute "he" for "the boy" to make the sentence
He saw the man with the telescope.
The ability to substitute an anaphor (pronoun) for a string of words is one justification for that string being a constituent.
Let's try substituting for "the man" using "him":
The boy saw him with the telescope.
Is the new sentence still ambiguous?
NO!
Why not?
Answer: "the man" is a constituent all by itself only in tree (2). In tree (1) "the man" is an NP only with the extra phrase "with a telescope". Therefore, we predict that substituting "him" for "the man" is only possible with tree (2). Therefore, the meaning of the new sentence should only be the meaning associated with sentence (2), that is:
Using the telescope, the boy saw him.
This is exactly correct.
Notice that this means that substitution with anaphors is also an effective way to diagnose some cases of structural ambiguity.
We can also find anaphors for VP, "do":
A: The boy saw the man with the telescope.
B: I did too.
C: He did?
D: He did so!
We can also find anaphors for some PP's, for example "there":
The cat sat on the mat.
The cat sat there.
However, not all phrases have anaphor substitutes in all languages.

Making Questions

This test is very similar to the anaphor test. However, question words also usually move to the front of the sentence.
The boy saw the man.
Who saw the man? (substituting "who" for "the boy")
Who did the boy see? (substituting "who" for "the man")
The cat sat on the mat.
Where did the cat sit? (substituting "where" for "on the mat")
What did the cat sit on? (substituting "what" for "the mat")

Ability to Stand Alone

Frequently in conversation it is not necessary to speak in entire sentences. Sometimes a piece of a sentences is enough, such as in answer to a question. We find that people do speak in phrases, so this is another test for phrases.
A: Where did you eat lunch?
B: In the park. (= I ate lunch in the park.)

Here is the structure for the sentence:

      S
 /        \
|          VP
|   /    /     \
|  |    |       PP
|  |    |     /   \
NP |    NP   |    NP
|  |    |    |   / \
N  V    N    P Art  N
|  |    |    |  |   |
I ate lunch in the park.

Notice that the strings of words that we have identified as phrases, "the park", "in the park" are valid answers to the question.
But strings of words which are not phrases, "lunch in", "in the", "lunch in the park" are NOT valid answers to the question.
For every question that you can form using the question test there should be an answer that uses the substituted phrase alone:
The cat sat on the mat.
Question: Where did the cat sit? (substituting "where" for "on the mat")
Answer: On the mat.
Question: What sat on the mat? (substituting "what" for "the cat")
Answer: The cat.

Co-ordination

Another test to see if a string of words is a constituent is to conjoin a parallel string of words with a conjunction (such as "and").
The cat sat on the mat.
The cat AND THE DOG sat on the mat. (testing "the cat")
The cat sat on the mat AND IN THE BOX. (testing "on the mat")
In some cases you may have to alter the sentence SLIGHTLY:
The cat is eating.
The cat AND THE DOG are eating.

Movement

We will see later how some syntax rules can move phrases around.
For the time being we will use a template for movement tests:
It was X that (S - X)
where X is the test phrase and (S - X) is the rest of the sentence
The cat sat on the mat.
It was THE CAT that SAT ON THE MAT.
It was THE MAT that THE CAT SAT ON.
It was ON THE MAT that THE CAT SAT.
In testing VP, there is a slight change, using the anaphora "do":
It was SIT ON THE MAT that THE CAT did.
Notice that non-phrase strings are no good:
* It was THE CAT SAT that ON THE MAT.
It is important to notice that (S - X) is NOT necessarirly a phrase.

Using Constituency Tests

Consider the following sentences:
1. My uncle went out the door.
2. My uncle threw out the door.
Does "out the door" function the same in both sentences?
How do you know?
Answer: The meanings are different. In (1) "out the door" represents a location, whereas in (2) "the door" is being "thrown out".
We can prove the difference by testing whether "out the door" is a constituent in each sentence.
Anaphor test:
My uncle went THERE. OK
My uncle threw THERE. NO
Question test:
WHERE did my uncle go? OK
WHERE did my uncle throw? NO
Stand alone test: (as answers to the above questions)
Where did my uncle go? Out the door. OK
What did my uncle throw out? Out the door. NO
Co-ordination test:
My uncle went out the door AND INTO THE YARD. OK
My uncle threw out the door AND INTO THE YARD. NO
Movement test:
IT WAS out the door THAT my uncle went. OK
IT WAS out the door THAT my uncle threw. NO
So, "out the door" passes the tests in "My uncle WENT out the door." but not in "My uncle THREW out the door."
Notice that the two sentences do not have totally different structures. In both cases "the door" is a noun phrase. And both sentences have verb phrases, "went out the door" and "threw out the door".
It is "out" that is behaving differently in the sentences.
There is one more phenomena (called "Particle Movement") that distinguishes the two sentences:
My uncle threw the door OUT.
*My uncle went the door OUT.

The structures for the two sentences are:

         S
       /   \
      /      VP
     /      /   \
    /      |      PP
    |      |     /  \
   NP      |    /    NP
  / \      |   |    / \
Pro  N     V   P  Art  N
 |   |     |   |   |   |
My uncle went out the door

         S
       /   \
      /      VP
     /      /  \    \
   NP      |    |     NP
  / \      |    |     / \
Pro  N     V   Part  Art  N
 |   |     |    |     |   |
My uncle threw out   the door

Phrase Structure Rules

Recall that in morphology we had trees for individual words--

   Noun
   /  \
Verb   |
 |     |
bake  -er
"someone who bakes"

And we generalized word structure patterns into word structure rules--
```
   Noun
   /  \
Verb  -er
```

We are going to do exactly the same generalization for syntax, finding RULES for the various pieces of the tree--

        S   S is immediately above NP and VP
       /  \      
      /    VP  VP is immediately above V and PP
     /    / \    
    /    |   PP   PP is immediately above P and NP
   /     |  /  \    
  NP     |  |   NP   NP is immediately above Art and N
 /  \    |  |  /  \
Art  N   V  P Art  N
 |   |   |  |  |   |
The cat is on the mat

  NP 
 /  \    is a valid piece of a tree in English
Art  N

The immediate constituents allowed by a language are stated in PHRASE STRUCTURE RULES, such as:
```
  NP       PP       S
 /  \     / \     /  \
Art  N   P  NP   NP  VP
```
There is another notation for phrase structure rules, also called REWRITE RULES.
- NP ==> Art N
- PP ==> P NP
- S ==> NP VP

Some Phrase Structure Rules for English

S ==> NP VP
S ==> NP Aux VP
where Aux is an auxilliary verb, like "be", "have", "do"
We can combine theses two rules into a single rule with a choice (the Aux is optional) as:
S ==> NP (Aux) VP
NP ==> (Art) (Adj)* N (PP) (S)
Nouns can be modified by
- Articles ("that man")
- Adjectives ("the big round red ball")
  the asterisk (*) indicates that there can be more than one
- Prepositional Phrases ("the pillow on the couch")
- Sentences ("the man that I knew")
- Only the Noun is required in an NP, except is the Noun is singular, then there must also be an Article--
  Tigers eat.
  The tiger eats.
  *Tiger eats.
The book distinguishes between S and S'. We will ignore that distinction.
S ==> (Comp) NP (Aux) VP
PP ==> P NP
PP ==> P PP
The order of constituents is a CHOICE languages make. In English PP's are in the order P NP, Japanese uses the reverse order, NP P.
VP ==> V (NP) (PP) (S)
Verbs can have different kinds of complements:
- NP -- The bat hit THE BALL.
- PP -- The cat sat ON THE MAT.
- S -- I know THAT JOHN IS TALL. ("that" is a COMPlementizer)
- NP PP -- John put THE VASE ON THE TABLE.
- NP S -- John convinced MARY THAT HE WILL WIN.
However, not all verbs can be used with all expansions the the VP rule. Some verbs require certain complements (similar to the rule that singular Nouns require an Article).
John put the book on the table.
*John put the book.
*John put on the table.
*John put.
When we learn the verb "put" we must also learn what complements it requires and allows.
In this case put requires an object NP and a goal PP, [ __ NP PP ].
This information is stored with the verb in the mental dictionary (lexicon).
It is called the SUBCATEGORIZATION FRAME.
Verbs (a CATEGORY of words) contain cases like "put" which belong to a SUBCATEGORY, and require NP PP in their VP.
Verbs also impose semantic requirements on their complements and subjects.
The tiger ate the apple.
%The rock ate my lunch
%The cat ate sincerity
(% signifies semantic ungrammaticality)
The verb "eat" requires an animate subject and a concrete object.
Other uses of "eat" are metaphoric, e.g. "The dryer ate my sock."
Notice that some verbs can appear without any complements (VP ==> V), e.g. "Mary cried."
And some verbs require that no object be present:
*Mary slept the bed.

Transformational Rules

Once we have built a basic tree, we then might want to change it, for example to turn it into a question.
1. John is going to school.
2. Is John going to school?
What happened between (1) and (2)? "Is" moved to the front.
How did we make the yes/no question?
What change did we make?
A Hypothesis: Move the first Aux to the front of the sentence.
This works for this case, but we need to TEST the hypothesis.
In order to test the idea "first" we will need sentences with at least two Auxilliary verbs.
Let's try this one: "The boy who WAS sleeping WAS dreaming."
Let's move the first WAS to the front:
*Was the boy who __ sleeping was dreaming?
No good, what about the other:
Was the boy who was sleeping __ dreaming?
That's the right question.
So it is NOT the first Aux that moves.
Another Hypothesis: Move the last Aux to the front of the sentence.
Works for these two cases. TEST it again, with this sentence:
"The boy WAS dreaming about a bear that WAS dancing."
Move last Aux: *Was the boy was dreaming about a bear that __ dancing?
Move first Aux: Was the boy __ dreaming about a bear that was dancing?
Therefore it's not the LAST Aux either.
So what is it?
Answer -- Move the HIGHEST Aux in the tree to the front of the sentence.

That is, move the Aux that is right under the top S in the tree.

            S
         /     \      \
     NP          \      \
   / |   \         \      \
  /  |        S      \      \
 |   |      / | \      \      \
 |   |   NP   |   VP     \     VP
 |   |   |    |    |      |     |
Art  N   N  Aux    V     Aux    V
 |   |   |   |     |      |     |
The boy who was sleeping was dreaming.
<=========================|

         S
      /  |     \
    /    |          VP
   |     |        /    \
   |     |       /        PP
   |     |      |       /    \
   |     |      |      |       NP
   |     |      |      |    /  /  \
   |     |      |      |   |   |     S
   |     |      |      |   |   |   /  \   \
  NP     |      |      |   |   |   |    |    VP
 /  \    |      |      |   |   |   |    |     |
Art  N  Aux     V      P  Art  N  Comp Aux    V
 |   |   |      |      |   |   |   |    |     |
The boy was dreaming about a bear that was dancing.
<========|

So it is NOT a property that you can hear (first, last) that determines which Aux to move.
Rather, it is a property of the sentence structure tree (HIGHEST).
But you can't hear the tree.
Other tranformations (in English) in the book (Study them!)--
- Wh-movement ("What is John eating __ ?")
- Passive ("John was seen by Mary.")
- Subject-Verb agreement

Word Order

Recall that languages can choose the order of the constituents in a phrase structure rule.
English: PP ==> P NP
Japanese: PP ==> NP P
Given the two basic rules for English:
1. S ==> NP VP (introducing subject NP)
2. VP ==> V NP (introducing object NP)
we can say that the overall word-order in a simple sentence is Subject-Verb-Object or SVO.
There are two choices for each rule:
1. Sentence
  1. S ==> NP VP
  2. S ==> VP NP
2. Verb Phrase
  1. VP ==> V NP
  2. VP ==> NP V
This gives four basic orders:
1. SVO: English
2. VOS: Huave
3. SOV: Turkish
4. OVS: Barasano
Unfortunately, other orders such as VSO (Irish) and OSV (Apurina) also exist.
To get VSO and OSV the verb must MOVE. We will not consider that rule here.

Author: William James Idsardi Email: <idsardi@udel.edu>
Last Updated: Sun, Jan 21, 1996