Class 2

I. INTRODUCTION TO “INTRODUCTION TO SYNTAX”

SYNTAX: the discipline of Linguistics which is concerned with how sentences are put together.

Consider the following utterance by a three-year-old child acquiring Turkish:

1. “Gitmek istiyorum.”
go.inf want.pres.prog.1.sg.
‘I want to go.’

An unremarkable sentence, indeed. But the knowledge that the child must have in order to
produce this sentence is not unremarkable at all.

He/she minimally has to know the following:

i. The words (their sounds and meanings),
ii. The fact that the verb gitmek ‘go’ precedes the verb istemek ‘want’ (the child does not
say *“Istiyorum gitmek”),
iii. The fact that the first verb must appear in the infinitive (non-agreeing) form (the child
does not say: *“Gitmem istiyorum” or *“Gidiyorum istiyorum”),
iv. The fact that the second verb must appear with an appropriate agreement suffix, -um
(the child does not say *“Gitmek istemek”, *“Gitmem istemek” or such),
v. The fact that –um is an appropriate agreeing suffix for the first person singular (the
child does not say *“Gitmek istiyorsun”).

Given that the child produces a correct sentence of Turkish, we must assume that the child
knows all this, i.e. that he/she possesses rather complex knowledge of how Turkish works.

The child is only three years old: he/she has not had any classes on the structure of Turkish nor
has anybody been teaching the child the rules of the language.

If we asked the child to tell us what rules he/she used to put together the sentence above, we
wouldn’t, I’m afraid, get very far. He/she would not be able to explain how he/she assembled the
sentence. OK, well, but the child is only three years old – he/she can hardly explain anything.

However, imagine a person who is 30, has never gone to school but speaks Turkish as his/her
mother tongue, is also (not surprisingly) capable of producing the same sentence (and many
more!).

Thus, we must assume that the thirty-year-old also must possess the same complicated
knowledge. And yet, if we asked this person to explain to us how he/she assembled the sentence,
we again, I’m afraid, would not get very far!

1
 Class 2

Given this situation, we are presented with at least two puzzles:

1. What form does the knowledge required to master the syntax of one’s language take?
2. How come that a three-year-old comes to possess this knowledge so early?

Let us first ponder the first question.

We will start by assigning a particular role to syntactic knowledge. Let’s go back to the sentence
in (1)/(2).

2. “Gitmek istiyorum.”
go.inf want.pres.prog.1.sg.
‘I want to go.’

What we know about this sentence (on the face of the matter) is that it has:
• a particular sound [gitmekistijorum],
• a particular meaning: “I want to go.”

We will start by hypothesizing that what connects the sound and the meaning is syntax (this is
not obvious, but we will take it as a sound hypothesis). Sentences are put together by a set of
procedures, which are part of our minds.

Syntax: A level of linguistic representation that mediates between sound and meaning, where
words are organized into phrases and sentences.
Carnie (2002), pg. 20

Our next hypothesis will be that these procedures are governed by a set of rules, which form a
grammar. Broadly speaking, these rules apply to a set of lexical items and assemble them into
phrases. They also apply to a set of phrases and assemble them into sentences. Thus, the
syntactic rules generate sentences. This is why this approach to language is called
GENERATIVE GRAMMAR.

As we saw previously, our knowledge of the procedures/rules that generate sentences, although
complex and intricate, is entirely subconscious or tacit.

Our task, as syntacticians, is to figure out what it is that we subconsciously know about the
syntax of (our) language.

Let us now return to the second question: How can a child acquire the knowledge of syntax of
his/her language so early?

Noam Chomsky (MIT), who is the father of generative grammar, made a (still very
controversial!) claim that this is so because language is pre-wired into human brain, in the same
way in which walking is, for example.

2
 Class 2

Human babies are born with their minds already prepared for the acquisition and usage of
language. This initial state of human mind, the capacity it has for language, is called
UNIVERSAL GRAMMAR.

Thus, language seems to be innate.

II. PRINCIPLES AND PARAMETERS IN A NUTSHELL

If language is innate, then how come languages are so different from one another?

Principles and Parameters approach (roughly):

UG consists of a set of PRINCIPLES (universal properties of language) some of which may be
parametrized (PARAMETERS: options that every language learner sets for him/herself based on
the properties of the input/linguistic data that surrounds him).

For example:

Principle: Every sentence has a subject.

can
Parameter: Subject be null (unpronounced).1
can’t

3. English: *“ Am going.”
4. Turkish: “ Gidiyorum.”

The hope is that with enough Parameters, the number of possible combinations covers the
diversity of human languages.

III. SYNTACTIC STRUCTURE

We said that syntax is what mediates between sound and meaning, syntactic rules create or
generate sentences.

So, when you utter a sentence, you assemble the words and the phrases together, and then
phonology and semantics “look at” the output of syntactic procedures and interpret this output.
• The result of the phonological interpretation is the stream of sound.
• The result of the semantic interpretation is the meaning.

In this course, we are going examine the syntactic procedures involved in the production of
sentences, and look at what the output of this computation is.

1
This particular parameter is called the Pro-drop parameter.

3
 Class 2

When we utter a sentence, we intuitively group some words in it together, to the exclusion of
others. For example:

3. The man who lives next door works for the government.

4. Words intuitively grouped together (the “surprised” reaction to this sentence) might
involve some of the following phrases:
▪ Next door
▪ The government
▪ Who lives next door
▪ The man who lives next door
▪ Works for the government
▪ For the government, …

5. However, we do not intuitively group together the following words (they would not be
uttered by someone who is surprised to hear this sentence):
▪ *Door works
▪ *For the
▪ *Man who
▪ *lives next

Why do some combinations (those in 4) sound good, and some combinations (those in 5) sound
bad?

There are several possibilities here:

i. It’s all random.
ii. There is an algorithm that helps us decide which words to group together, and this
algorithm is based on whether a certain word can be grouped together with the word on
its left or the word on its right (for example, the word next can only be grouped with the
word on its right, and not on its left, etc.)
iii. There is an algorithm, and it is based on which position in the linear order of words a
particular word occupies (for example, the one-but-last word can be grouped with the last
word, etc.)
iv. There is an algorithm, and it is based on something that is not immediately visible when
we hear the sentence, but has to be discovered: structure.

If possibility (i) is correct, then we can all go home, because there is no way for us to ever
scientifically study language.

4
 Class 2

Possibility (ii) is testable. We can also refute it. In 3 (= 6), the word next is grouped with the
word on its right:

6. The man who lives [next door] works for the government.

However, in 7 below, the word next cannot be grouped with the word on its right.

7. What I did *[next is] slap him on the face.

Possibility (iii) is also testable, and refutable. In 3 (= 8), the one-but-last word word is grouped
together with the last word:

8. The man who lives next door works for [the government].

But in 9 below, the last word cannot be grouped together with the one-but-last.

9. I will call *[you tomorrow].

Let us now look at the possibility (iv), namely, that syntax in fact creates hierarchical structures
that are, of course, not visible, but are nevertheless very real.

Let’s test this possibility with a simpler sentence:

10. Mary loves Bob.

A very simple structure that we can imagine is the following:

11.

Mary

loves Bob

What are the words that can be grouped together here (be uttered in surprise at this sentence)?

12. Mary? ✓
Bob? ✓
Loves? ✓
Loves Bob? ✓
Mary loves Bob? ✓

13. *Mary loves? 

5
 Class 2

Let’s look again at the structure in 11.

14. 
 
Mary
 
loves Bob

What we have here is a syntactic tree.

Like any tree, it has branches, in our case four of them.

At each end of a branch, there is a node. There are five nodes total in this tree.

One node is such that it has branches below it, but no branches above it. Which one is it?
This node is called the root node.

Two nodes are such that they have branches below it. Which ones are those? and
These nodes are called non-terminal nodes. Thus, the Root node is also a non-terminal node.

Finally, there are three nodes such that they have branches above them, but not below them.
Which ones are these? , , and . These nodes are called terminal nodes.

IV. RELATIONS THAT HOLD IN A SYNTACTIC STRUCTURE

We will say that nodes 2 and 3 have the same mother node (1). Similarly, nodes 4 and 5 have the
same mother node (3).

We will call nodes 2 and 3 sister nodes. Similarly, nodes 4 and 5 are also sisters.

Nodes 2 and 3 are going to be called daughter nodes of node 1. Similarly, nodes 4 and 5 are
daughters of node 3.

15. IMMEDIATE DOMINANCE:

A node X immediately dominates a node Y iff X is a mother of Y.

16. DOMINANCE:
A node X dominates a node Y iff X is a mother of Y or X dominates a mother of Y.

6
 Class 2

Given this definition of dominance, what are the nodes that are dominated by node 1 in our tree?

17.

And what are the nodes that are dominated by node 3 in our tree?

18.

Nodes 4 and 5 do not dominate any other nodes. However, they dominate some words: node 4
dominates the word loves, and node 5 dominates the word Bob.

V. CONSTITUENTS AND TESTS FOR CONSTITUENCY

Ok, now what words does node 1 dominate? It dominates everything that is dominated by the
nodes that node 1 dominates.

1 dominates 2,  it dominates the word Mary.

1 dominates 3, not obvious what that means yet, but let’s note that dominance is transitive.
1 dominates 4,  it dominates the word loves.
1 dominates 5,  it dominates the word Bob.

So, our node 1 dominates the words: Mary, loves, and Bob.

What words do the nodes 2, 3, 4, and 5 dominate?

19. Node 2:

20. Node 3:

21. Node 4:

22. Node 5:

Let us now go back to 12 and 13. These are the lists of possible and impossible intuitive
groupings of words in our sentence Mary loves Bob.

7
 Class 2

We will note that for every possible grouping of words, there is a node in the structure that
dominates all of the words grouped together and nothing else:

23. Mary? ✓ = node 2

Bob? ✓ = node 5
Loves? ✓ = node 4
Loves Bob? ✓ = node 3
Mary loves Bob? ✓ = node 1

24. *Mary loves?  No node in the tree dominates the words Mary and
loves without dominating something else as well,
namely Bob.

We will say that the words which are grouped together form constituents.

25. A CONSTITUENT is a chunk of a syntactic tree that includes a node and all that it dominates.

So, it seems that words that can be grouped together form constituents.

Let us now go back to our sentence in 3 (= 26). The syntactic tree that we might draw for this
sentence is given in 27.

26. The man who lives next door works for the government.

27. 

 

the  works 

man  for 

who  the government

lives 

next door

Can you figure out why the following words cannot be grouped together?
→ *Door works → *Man who
→ *For the → *lives next

8
 Class 2

Note that the structure that we posited for the sentence in (26) is hierarchical, i.e. it contains
various levels, so some words are ‘higher’ than others. For example, the word man is higher than
the word next.

If we had posited a flat structure – a structure in which all the elements are daughters to the same
node, as in (28), we would have no explanation why, for example, the words door and works
cannot be grouped into a constituent.

28. 

the man who lives next *[door works] for the government

Speakers somehow know what words can be grouped together.

Given that words that can be grouped together are constituents, then speakers must have some
kind of knowledge of what constituents there are in a particular sentence.

Constituents crucially make reference to structure. Therefore, if speakers know about

constituents, then they must know about the structure. This means that the structure must be real.

There are tests that can confirm whether certain strings of words are constituents or not:
• Substitution or replacement test
• Stand alone test
• Movement test

SUBSTITUTION TEST: only constituents may be replaced by a single word.

29. a. The man who lives next door works for the government. (node in 27)
b. He works for the government.

30. a. The man who lives next door works for the government. (node in 27)
b. The man who does works for the government.

31. a. The man who lives next door works for the government. (node in 27)
b. The man who lives next door works for it.

9
 Class 2

STAND ALONE TEST: if the words can stand alone as a response to a question, then they probably
form a constituent.

32. a. Who works for the government?

b. The man who lives next door. (node in 27)

33. a. Where does the man who works for the government live?
b. Next door. (node in 27)

MOVEMENT TEST: only constituents may be moved. When a constituent is moved, it appears in a
place in the sentence which is different from what its usual, or canonical position is.

CLEFTING:
34. a. The man who lives next door works for the government.
b. It is the government that the man who lives next door works for.

35. a. Mary likes the boy with dark hair.

b. It is the boy with dark hair that Mary likes.

PREPOSING OR PSEUDO-CLEFTING:
36. a. Mary likes the boy with dark hair.
b. The boy with dark hair is who Mary likes.

PASSIVE:
37. a. Mary likes the boy with dark hair.
b. The boy with dark hair is liked by Mary.

38. a. A big black dog bit the boy.

b. The boy was bitten by a big black dog.

None of these tests is perfect. Sometimes two tests give you different results. This is because
more is going on in syntactic structures than meets the eye. When figuring out whether a string
of words forms a constituent or not, ALWAYS APPLY MORE THAN ONE TEST.

10