Constituency
and Constituency Tests
Md. Nazrul Islam
ID:
1142007
M. A in ELT under MAPW, Department of English
Jahangirnagar University, Savar, Dhaka
I am grateful to my honorable course instructor Associate Professor and Chair Dr. Sanyat Sattar,
Jahangirnagar University, Savar for his incomparable and invaluable guidelines for
making this short thesis.
Abstract
This article discusses structures of syntax.
Syntax is one of the basic levels of linguistics. Constituency is the most
import part of syntactic structures. Basically syntactic structures depend on
this constituency. Without the basic knowledge of constituency test, no one can
acquire fundamental skills about syntax. So I have tried to clear the
constituency test although I have some bindings to gain appropriate knowledge.
Here I would like to discuss a little about constituency other than my target
article. Firstly, single words are constituents, but frequently we see some
exceptions like certain contractions, certain possessives. For example, You,
come here! Secondly, complete sentences are constituents. For instance, Come
here! Thirdly, any sequence of words which can be functionally replaced by
a single word must be a constituent. For example, The woman with red hair
won first prize. She won it. Fourthly, any sequence of words
which can stand alone in answer to a question. For instance, Who stole the
money? The man. Moreover, any sequence of words conjoined to an
independently identifiable constituent is a constituent. i.e. Mary helped the
man in the green hat and Michael. Finally, any sequence
of words which can be moved as a unit and still produce a grammatical sentence
with the same meaning is a constituent. Such as, I worked on the ancient text with
great care. With great care I worked on the ancient text.
CONSTITUENCY TESTS
a) The
students loved their syntax assignments.
We can notice
that on a purely intuitive level there is some notion that certain words are
more closely related to one another. For example, the word the seems to
be tied more to the meaning of students than it is to loved or syntax.
A related intuition can be seen by looking at the sentences below. b) The students loved their phonology
readings.
c) The students
hated their morphology professor.
Compare these
sentences to the ones in (b) and (c). We will see right away that the relationship
between the students and their syntax assignments and the
students in the first sentence and their phonology readings in (a)
is the same. Similarly, the relation between the students and their
morphology professor in (c), while of a different kind (hating instead of
loving), is of a similar type: There is one group (the students) who are
either hating or loving another entity (their syntax assignments or their
morphology professor). In order to capture these intuitions (the intuition
that certain words are more closely connected than others, and the intuitions
about relationships between words in the sentence), we need a more complex
notion. The notions we use to capture these intuitions are constituency and
hierarchical structure. The notion that the and students
are closely related to one another is captured by the fact that we treat
them as part of a bigger unit that contains them, but not other words. We have
two different ways to represent this bigger unit. One of them is to put underlines
around units:
d) the
students
The other is to
represent the units with a group of lines called a tree structure:
 |
e) the students
These bigger
units are called constituents. A definition for a constituent is given here:
A string of
words that functions together as a unit is called constituent. In other words, a group of expressions
within a larger phrase that form a unit is known as constituent. For
example,
f) The fluffy cat was sleeping on the desk.
Constituency is
the most important and basic notion in syntactic theory. Constituents capture
the intuitions mentioned above. The “relatedness” is captured by membership in
a constituent. As we will see it also allows us to capture the relationships
between constituents alluded to in (c) and (d).
Constituents
don’t float out in space. Instead they are imbedded one inside another to form
larger and larger constituents. This is hierarchical structure. Foreshadowing
the discussion below a bit, here is the structure we’ll develop for a sentence
like this:
S
NP VP
D
N V PP
The man
eats
P
NP
at Adj N
fancy
restaurants
When
we studied before ‘Syntax as a science’, we learnt that linguistics in general
(and syntax specifically) up to the criterion of the scientific method. That
is, if we make a hypothesis about something, we must be able to test that
hypothesis. Now, we have proposed the hypothesis that sentences are composed of
higher level groupings called constituents. Constituents are represented in
tree structures and are generated by rules. If the hypothesis of constituency
is correct, we should be able to test it in general (as well as test the
specific instances of the rules).
In
order to figure out what kinds of tests we need, it is helpful to reconsider
the specifics of the hypothesis. The definition of constituents states that
they are groups of words which function as a unit. If this is the case, then we
should find instances where groups of words behave as single units. These
instances can serve as tests for the hypothesis. In other words, they are tests
for constituency. There are a lot of constituency tests listed in the
syntactic literature. We are going to look at only five constituency tests
here: answering to question, clefting, coordination, substitution, and
deletion.
Firstly, we can identify constituency by answering to question. We may ask as
many questions as possible related to the target sentence. The answers will
help to find the constituency. For example,
g) The man stole the money.
Who stole the money? The man.
What did the man do? Stole the money.
What did he steal? The money.
Movement
is our second test of constituency. If
you can move a group of words around in the sentence, then they are a
constituent because you can move them as a unit. Some typical examples are
shown in (h), (i) and (j). Clefting (h) involves putting a string
of words between It was (or It is) and a that at the
beginning of the sentence. Preposing (i) (also called pseudoclefting)
involves putting the string of words before a is/are what or is/are
who at the front of the sentence. Passive forms are shown in (j).
Briefly, it involves putting the object in the subject position, the subject in
a “by phrase” (after the word by) and changing the verb form (for
example from kiss to was kissed).
h)
Clefting: The cat was sleeping on the desk.
It was on the desk that the cat was sleeping.
It was the cat that was sleeping on the desk.
*It was on the that the cat was
sleeping.
i) Preposing: Big bowls of beans are what I like.
(from
I like big bowls of beans.)
j) Passive: The big boy was kissed by [the slobbering dog].
(from
The slobbering dog kissed the big boy.)
Again,
the movement test is only reliable when you keep the meaning roughly the same
as the original sentence.
Coordination is our third constituency test. This test is when the
words strings are joined by a coordinating conjunction “and” and they act like
a single unit. This unit also works as an argument.
k) John and the man went
to the store.
l) *John and very blue went to the
store.
If
you can co-ordinate a group of words with a similar group of words, then they
form a constituent.
Moreover,
the smallest constituent is a single word, so it follows that if you can substitute
a group of words with a single word then we know that group forms a
constituent. Consider the italicized NP in (m) and (n), it can be replaced with
a single word (in this case a pronoun). This is the substitution test.
For example,
m)
The man from NY flew only ultra-light planes.
n)
He flew only ultra-light planes.
There
is one important caveat to the test of substitution: There are many cases in
our rules of optional items (those things marked in parentheses like the AP in
NP → (D) (AP) NP). When we replace a string of words with a single word, how do
we know that we aren’t just leaving off the optional items? To avoid this
problem, we have to keep the meaning as closely related to the original as
possible. This requires some judgment on our part. None of these tests are
absolutes.
The
final test we will use is the deletion. By this test some parts of the
sentences can be deleted. The remaining part is grammatical. For example,
o) We went there to buy snacks. We went
there.
* We went buy snacks
We looked at
the idea that sentences are hierarchically organized into constituent
structures. We represented these constituent structures in trees and underlined
diagrams. We also developed a set of rules to generate those structures, and
finally we looked at constituency tests that can be sed to test the structures.
Parts of speech are the labeling system for constituent structure.
References
Andrew Carnie,
Syntax, The University of Arizona (online edition)
Sportiche,
Dominique (1988) A Theory of Floating Quantifiers and its Corollaries for
Constituent
Structure. Linguistic Inquiry 19.3: 425-449.
Uriagereka,
Juan (1998) Rhyme and Reason: An Introduction to Minimalist Syntax.
Cambridge Mass: MIT press.
Williams, Edwin
(1994) Thematic Structure in Syntax. Cambridge: MIT Press.
