Syntactic Theory

Second Edition

Geoffrey Poole
 © Geoffrey Poole 2002, 2011

*
All rights reserved. No reproduction, copy or transmission of this
publication may be made without written permission.
No portion of this publication may be reproduced, copied or transmitted
save with written permission or in accordance with the provisions of the
Copyright, Designs and Patents Act 1988, or under the terms of any licence
permitting limited copying issued by the Copyright Licensing Agency,
Saffron House, 6-10 Kirby Street, London EC1 N 8TS.
Any person who does any unauthorized act in relation to this publication
may be liable to criminal prosecution and civil claims for damages.
The author has asserted his right to be identified as the author of this
work in accordance with the Copyright, Designs and Patents Act 1988.
First edition 2002
Second edition 2011
Published by
PALGRAVE MACMILLAN
Palgrave Macmillan in the UK is an imprint of Macmillan Publishers Limited,
registered in England, company number 785998, of Houndmills, Basingstoke,
Hampshire RG21 6XS.
Palgrave Macmillan in the US is a division of St Martin's Press LLC,
175 Fifth Avenue, New York, NY 10010.
Palgrave Macmillan is the global academic imprint of the above companies
and has companies and representatives throughout the world.
Palgrave® and Macmillan® are registered trademarks in the United States,
the United Kingdom, Europe and other countries.
ISBN: 978-0-230-24393-4 hardback
ISBN: 978-0-230-24394-1 paperback
This book is printed on paper suitable for recycling and made from fully
managed and sustained forest sources. Logging, pulping and manufacturing
processes are expected to conform to the environmental regulations of the
country of origin.
A catalogue record for this book is available from the British Library.
A catalog record for this book is available from the Library of Congress.
10 9 8 7 6 5 4 3 2 1
20 19 18 17 16 15 14 13 12 11
Printed and bound in Great Britain by
MPG Group, Bodmin and Kings Lynn
Phrase
Structure and
Constituency

Now that you have some idea what our general approach to 'language' is going
to be and seen some reasons why studying language and languages in the man-
ner of the natural sciences might be an interesting way to proceed, it's now
time to get on with the theory-building part of things. We'll start by looking at
some very simple sentences and exploring what you already know about them
in virtue of being a native speaker of a language, English in this case.
One of the things that you know about the sentences of your language is
that they're made up of words and that certain of those words go together to
create larger units. Since you don't always know consciously what you know
about your language, we'll look at various traditional tests which can tease out
what you know about the structure of sentences.
We'll then introduce phrase-structure rules and phrase-structure trees. Phrase-
structure trees are used to illustrate graphically the structure of a given sen-
tence. We'll then refine our phrase-structure rules and trees in order to better
account for what you know about the different ways words are organized in the
structure of sentences.

2.1 A simple starting point

Let's start by taking a look at an ordinary English sentence like (1), with a view
towards exploring some of the things that you know about it:

(1) The student will meet her friend at the station.

The first observation to make with respect to (1) will strike you as the most
blindingly obvious thing I could possibly say. I claim that part of what you
know when you know your language is that a sentence like (1) is made up of
words.
I know what you're thinking: 'Of course (1) is made up of words! How could
it possibly be otherwise?' However, if we were to look at a sound spectrograph
of (1), a visual representation of the sound energy created when (1) is produced,

21
22 PHRASE STRUCTURE AND CONSTITUENCY

you would see that there are no breaks in the sound energy corresponding to
where the word breaks are. It's all one continuous flow. Therefore, even some-
thing as simple as the fact that you perceive (1) as having words is due to the
cognitive system that you have for processing language. You might be inclined
to think that you know about word breaks just because the breaks are there in
the writing. However, this is just an accident of English. Plenty of other writ-
ing systems (like Japanese katakana) put breaks after every syllable rather than
every word. Word breaks are also easily perceived by speakers who are illiter-
ate, nor do you have any problem 'hearing' word breaks in sentences with new
words or words that you've never seen written.
However, you know more than just that the sentence in (1) is made up of
words. You also know intuitively that certain words in (1) belong, in traditional
terminology, to the same part of speech. That is, student, friend and station are
all nouns, at is a preposition, meet is a verb, and so on. You might again think
that this is something that you've been taught explicitly at school, but in fact all
you're doing at school is being given labels for things you already know about.
We'll take those intuitions over fairly directly, although we'll refer not to 'parts of
speech', but rather to a word's syntactic category. Here are some examples of elements
from various syntactic categories. As you can see, the correspondence between syn-
tactic category and traditional 'part of speech' is pretty straight forward:

Table 2.1 Syntactic categories

Category Example

Noun table, sincerity, station

Preposition at, on, in, by, over
Adjective green, intelligent, generous
Verb give, put, run, sleep, disturb
Auxiliary can, should, must, will
Determiner the, a, this, those, several, my

The only one that might be slightly unfamiliar is determiner. This is a sort of
catch-all category which includes what are traditionally called articles (the, a),
demonstrative adjectives (this, that, many, several), and the like. However, don't
worry now about memorizing the syntactic category of every word. It'll become
second nature in no time.
Now, not only do you know that each of the words in (1) belongs to a
certain syntactic category. You also know that certain strings of words in
(1) go together to form larger units and that other strings of words don't.
For example, the two words at the beginning of the sentence the student
go together to form a noun phrase (NP). On the other hand, the string of
words student will meet does not form any kind of unit. Her friend and the
station are also examples of noun phrases in (1). The nouns student, friend
and station are the heads of these three noun phrases respectively. They are
the most important elements - the pivots around which the entire phrase
turns. Phrases can also be parts of larger phrases. For example, the noun
 A SIMPLE STARTING POINT 23

phrase the station combines with the preposition at to form the prepositional
phrase (PP) at the station. In this case, the preposition at is the head of the
prepositional phrase.

Give the syntactic category of each word in the following phrases. What is the
syntactic category of the entire phrase (e.g. noun phrase, adjective phrase,
etc.)? Which word is the head ofthe phrase?
(a) the book about linguistics
(b) at the end of the platform
(c) see the answer
(d) aware of the problem

Phrase-structure trees and phrase-structure rules

So it seems as though part of what you know when you know English is that
there are two 'levels' of organization. There is the word or 'head' level and the
phrase level. Each phrase has a head, although not every word in the sentence
will be the head of a phrase.
In order to express your knowledge about the structure of phrases, we could
simply list the information:

(2) at the station (PP)

a. at: preposition (P) (head of phrase)
b. the station: NP
i. the: determiner (Det)
ii. station: noun (N) (head of phrase)

However, a more compact way of illustrating this organization is to use a

phrase-structure tree:

(3)

Det N

I I
at the station
24 PHRASE STRUCTURE AND CONSTITUENCY

(3) indicates precisely the same information as (2). The entire phrase at the
station is a PP, which is itself made up of a preposition combining with an
NP. This particular NP consists of a determiner and a noun. Therefore, the
phrase-structure tree makes a specific claim about what you know about your
language. To draw the tree in (3) for the phrase at the station is to claim that
your cognitive system assigns that structure to it.
However, you know more about your language than this. It's not just that
this particular PP is made up of a P and an NP (and perhaps there are oth-
ers which aren't). Rather, you know that all PPs are made up of a P and an
NP. What your cognitive system seems to possess is the following phrase-
structure rule:
(4) PP ~ P NP

(4) doesn't just make a claim about a particular PP, it says that all well-formed
PPs are composed of a preposition followed by an NP. If we claim that this
rule forms part of what you know when you know your language, then we
can account for not just the PP in (3), but an infinite number of other PPs as
well:

(5) to the store, at many banks, from a friend, towards the future, etc.

Similarly, the tree in (3) claims that you know something about the particular
noun phrase the station. It's made up of a determiner plus a noun. However, we
can account for more by assuming that you know the following phrase-structure
rule:

(6) NP~DetN

This rule accounts for your knowledge of the station as well as an infinite
number of other NPs:

(7) the book, that aeroplane, a review, several decisions, etc.

A well-formed NP can also have an optional preceding adjective phrase (AP),

as well as an optional following PP:

(8) The (green) books (about magic) are helpful on occasion.

Therefore, the rule in (6) should really be at least (9):

(9) NP ~ Det (AP) N (PP)

Having looked a little bit at the individual phrases from the sentence
in (1), let's now see if we can draw the phrase-structure tree for the entire
sentence. My personal feeling is that it's easier to work from the bottom up
when first learning to draw phrase-structure trees, so let's start with the
words and their syntactic categories, including now verbs (V) and auxiliaries
(Aux):
 A SIMPLE STARTING POINT 25

(10) Det N Aux v Det N p Det N

I I I I I I I I I
The student will meet her friend at the station

As we mentioned above, the student, her friend and the station are all noun
phrases, so let's put those in:

(11) NP NP NP
~ ~ ~
Det N Aux v Det N p Det N
I I I I I I I I I
The student will meet her friend at the station

We also know that Ps go together with NPs to form PPs, so let's hook at and
the station together:

(12) pp

~
Det
NP

N Aux v
~
Det
NP

N p
fA
Det N
I I I I I I I I I
The student will meet her friend at the station

Now meet is a verb, and is therefore the head of a verb phrase (VP). Her friend
is the direct object of the verb, and so therefore must be in the verb phase.
The PP at the station gives the location where the meeting action took place,
and so therefore is also in the verb phrase. You might think the auxiliary
should also go in the verb phrase, but it doesn't, for reasons we'll discuss in
Chapter 3.

(13) NP Aux VP

A
Det N V
~
NP PP

Det
AA N P NP

A Det N

I I
The student will meet her friend at the station
26 PHRASE STRUCTURE AND CONSTITUENCY

We're now just left with the subject NP, the auxiliary and the VP. These things
go together to form the sentence as a whole:

(14)

NP Aux VP

Det
A N
~
v
NP pp

~ ~
Det N P NP
~
Det N
I I
The student will meet her friend at the station

So now we have a phrase-structure tree associated with the sentence in (1).

However, as before, we want to make claims about not just the sentence in (1)
but an infinite number of other sentences. What we therefore want to do is
take what we've done so far and generalize it. If we do that, we have the phrase-
structure rules in (15):
(15) a. S ~ NP Aux VP
b. NP ~ (Det) (AP) N (PP)
c. VP~ VNPPP
d. PP~ PNP
Just as with our NP rule, though, we're going to need more rules for well-formed
VPs. In (15), we have only one rule for VP expansion, VP ~ V NP PP. However,
there are many more possible VPs than just this one. Consider the VPs in (16):
(16) a. John slept (VP ~ V)
b. Mary gave Bill the book (VP ~ V NP NP)
c. Ivan went to the museum (VP ~ V PP)
If we were going to try to account for a substantial fragment of English using
phrase-structure rules, we'd need a lot more rules of the sort in (16). However,
we won't be pursuing this at any length here because we're going to see in
Chapter 3 that the phrase-structure rules (in the form that we've seen so far) need
to be substantially modified anyway. However, this chapter is just about getting
used to the idea that words are organized into larger groups and how phrase-
structure rules work in general, so we'll keep using them for the moment.
 CONSTITUENCY 27

2.2 Constituency

So far, we've been leaving things on a somewhat intuitive level when it comes
to the details of sentence structure. We all share the feeling that, for example,
at the station forms some kind of a unit, a PP in fact, whereas other strings of
words don't form a unit. Let's call these units constituents. Student will meet, for
example, doesn't feel like a unit of any kind.
Feelings and intuitions are all well and good, and it's true that they play
an important starting role in just about any scientific project. However, it's
also true that intuitions can be wrong. For example, I have the intuition that
every morning the sun rises in the sky while the earth remains perfectly still.
On some level, I know that that intuition is wrong, but that doesn't alter
the fact that the intuition is very strong indeed. Therefore, it would be nice
if we could devise some tests to back up our intuitions as to which groups
of words do or do not form constituents. To that end, we'll take a look at
some traditional tests which are used to indicate constituency: (1) the sub-
stitution test, (2) the coordination test and (3) the movement test. Some
are a little bit more reliable than others for reasons which we'll see in later
chapters.

The substitution test

The first test that we'll look at is called the substitution test. This constituency
test relies on the fact that groups of words which function as constituents in the
sentence can generally be replaced in a sentence by a single word. This confirms
that the group of words forms a constituent. Depending on exactly the single
substitution word you choose, you can even identify the syntactic category of
the constituent you're dealing with.
Pronouns, for example, can be used to tell whether a given string of
words constitutes an NP constituent. If a given group of words can be
replaced by a pronoun, then that group of words is not only a constituent,
but is specifically an NP. Consider the contrasts between the sentences in
(17):

(17) a. The Belgian scientist who won the Nobel Prize at the age of 22
retired today.
b. She retired today.
c. *The Belgian scientist who she today.

The pronoun she is able to replace the entire string of words the Belgian sci-
entist who won the Nobel Prize at the age of22. On the other hand, as (17c)
indicates, she cannot replace the string of words won the Nobel Prize at the
age of22 retired. This suggests that the string of words does not form an NP
constituent.
The Belgian scientist who won the Nobel Prize at the age of 22 is not
the only string of words in (17a) which can be replaced by a pronoun. For
28 PHRASE STRUCTURE AND CONSTITUENCY

example, we could take the Nobel Prize and replace it by the pronoun it,
creating (18):

(18) The Belgian scientist who won it at the age of22 retired today.

You'd have to have been talking about the Nobel Prize during a previous part
of the conversation in order for (18) to be an appropriate thing to say, of course.
Compare (18) with (19) for example:

(19) *The she won the Nobel Prize at the age of22 retired today.

Unlike (18), (19) is just an unintelligible jumble. This is because she has
replaced Belgian scientist who in (17a), and these words do not form an NP
constituent.

Using pronouns such as he/she, it, them, etc. identify all of the NPs in the fol-
lowing sentences:
(a) The new gorilla's attendant put a tasty banana on the purple table.
(b) The person responsible for security left the visiting dignitaries in a minus-
cule antechamber.

Just to have a general term to refer to these single words that can substitute
for others, let's refer to them as 'proforms'. Pronouns substitute for NPs, but
there are other proforms which can substitute for other syntactic categories.
For example, words such as here, there or then can substitute for certain
kinds of PPs. Consider the contrasts in (20):

(20) a. Bill's book about economics in the eighteenth century was found
in the dining room.
b. Bill's book about economics then was found in the dining room.
c. Bill's book about economics in the eighteenth century was found
there.
d. *Bill's book about there century was found in the dining room.

Again, we see that the proforms then and there can only substitute for a
constituent which is a PP, like in the eighteenth century or in the dining room. A
group of words which is not a constituent, like economics in the eighteenth cannot
be substituted for. Nor can there or then substitute for a constituent which is not
a PP, like Bill's book about economics:

(21) *Then was found in the dining room

 CONSTITUENCY 29

Use PP preforms to identify the PPs in the following sentences:

(a) Mary put the book in Sue's jacket on Thursday.
(b) The agent who left the computer on the table at lunchtime was dismissed
on the very day he was hired.

The coordination test

Conjunctions, elements like and, or and but, can serve as a very useful aid
in determining the constituent structure of a sentence, since only constitu-
ents can be conjoined. Furthermore, both elements to be conjoined must
be of the same syntactic category (i.e. two Ns, two APs, etc.). Sequences of
words which are not constituents cannot be conjoined. Consider the fol-
lowing sentence:

(22) John's expose in The Guardian gave new hope to the government's
opponents.

We could use the substitution test from the previous section to identify several
strings of words as constituents. For example, fohn's expose in The Guardian is
a constituent, as it can be replaced by the proform it.

(23) It gave new hope to the government's opponents.

We could, however, also use the coordination test to confirm this result by
attempting to coordinate fohn's expose in The Guardian with another NP. If
the result is grammatical, than this would confirm that fohn's expose in The
Guardian is indeed a constituent:

(24) [John's expose in The Guardian and the subsequent revelations] gave
new hope to the government's opponents.

In (24), fohn's expose in The Guardian has been coordinated with another NP
the subsequent revelations, indicating that it is indeed a constituent.
We get the same results if we try to coordinate other constituents. For
example, we can also have coordinations like (25a) (VP-coordination) and
(25b) (PP-coordination):

(25) a. John's expose in The Guardian [gave new hope to the government's
opponents and put fear into ministers' hearts].
b. John's expose in The Guardian gave new hope [to the government's
opponents and to seekers of justice throughout the world].
30 PHRASE STRUCTURE AND CONSTITUENCY

On the other hand, if we were to take a string of words which does not form
a constituent, the resulting coordination would be ungrammatical:

(26) *John's expose in The Guardian [gave new and renewed old] hope to
the government's opponents.

The movement test

In addition to substitution and conjunction, it's also possible to use move-
ment as a test for constituency. We'll talk about movement in greater detail
in Chapter 6. For the moment, all that's really important to notice is that cer-
tain groups of words seem able to be moved around as a unit in the sentence,
whereas others cannot. For example, consider (27):

(27) John will leave a book on the table.

It is possible to take various constituents in (27) and move them to the front
of the sentence, giving them an emphatic or contrastive interpretation. This is
called topicalization:

(28) a. On the table John will leave a book (but on the chair he will leave
some paper).
b. The book John will leave on the table (but the letter he will put in
the bin).

On the other hand, if we try to take a string of words which does not form a
constituent and topicalize it, the result is ungrammatical:

(29) a. *Book on, John will leave the table.

b. *On the, John will leave the book table.

Unfortunately, the movement test is the least reliable of the three tests that
we've mentioned because movement in general, and topicalization in particu-
lar, are subject to a number of constraints which have nothing to do with con-
stituent structure. For example, topicalization of a VP is rather odd in most
dialects of English:

(30) Leave the book on the table, John will.

 PHRASE-STRUCTURE TREE 31

Probably the most well known 'speaker' of a dialect in which this is possible is
the character Yoda from the Star Wars films. However, there is a serious point
we can make. In writing dialogue for that character, the writers never topical-
ize non-constituents, so even if Yoda's topicalization is a little freer than most
dialects of English, he would never produce sentences like those in (29).
Because of these 'outside' influences on movement, the movement test can
only be used in a 'positive' way. That is to say, if something passes the move-
ment test, then it definitely is a constituent. If it doesn't pass the movement test,
then you can't conclude anything. It might be a constituent, or it might not. As
an example of this, notice that the table in (27) is clearly a constituent (being an
NP). It passes the substitution and coordination tests with flying colours (31):

(31) a. John left a book on it.

b. John left a book on the table and the chair.

However, if we use the movement test, (32) is rather odd:

(32) *The table, John will leave a book on, but the box, John will put a
record in.

By contrast, the phrase left a fails the substitution and the coordination tests,
and is definitely not a constituent. It also fails the movement test:

(33) a. *John it/there book on the table.

b. *John left a and threw the book on the table.
c. *Left a, John book on the table (but put a, John record in the box).

So the movement test should be used with a little caution. Only constituents
pass the movement test, but both constituents and non-constituents can fail
it. A quick confirmation with the other tests is often a good idea if there is any
doubt.

2.3 The elements of a phrase-structure tree

and their relations

Having seen a few elementary phrase-structure rules and phrase-structure trees

in the previous section, we need to introduce some terms for talking about
phrase-structure trees and the relations between the elements in them, which
32 PHRASE STRUCTURE AND CONSTITUENCY

are called 'nodes'. We'll be using these terms so frequently throughout the book
that they'll become second nature in no time.
We'll start off in a seemingly aggressive vein with the term dominate.
One node dominates another when you can trace a path from the first node
to the second one moving only downward through the tree. Take a look at
the sentence in (34) and the phrase-structure tree representation for it in
(35):

(34) The agent will go to Madrid.

(35) s

NP Aux VP
~ v
~pp
Det N
~
P NP

I I
The agent will go to N
I
Madrid

Notice that the S node dominates every other node in the tree. Starting from
S, you can get to any of the other nodes in the tree by moving only downward.
On the other hand, the PP node dominates only the P node and NP node
associated with Madrid. The PP node does not, for example, dominate the NP
node associated with the agent. You can't get from the PP node to the NP node
associated with the agent by only going down. You need to go up through VP
and S before going down.
A more specific type of domination is immediate domination. A node imme-
diately dominates every node which you can get to by moving down exactly
one level. So while S dominates every other node in the tree, it immediately
dominates only the NP the agent, the Aux will and the VP go to Madrid.
Similarly, the VP node dominates five nodes (V, PP, P, NP and N), but it only
immediately dominates V and PP.
As a point of cultural interest, nodes in a phrase-structure tree are always
female rather than male. It's not uncommon to see the nodes which some-
thing immediately dominates referred to as its daughter nodes. Less often,
you'll see a node which immediately dominates something referred to as its
mother node. One place, however, where the matriarchal term is used very
frequently is for sister nodes, that is, daughters of the same mother, as in (36)
or (37):
 PHRASE-STRUCTURE TREE 33

(36) pp

P
~ ~ NP
~
Det N
I I
in the city

(37) VP

V
~
~ NP
I
N
I
destroy Madrid

Notice that these relational terms can give us a way of defining a constituent.
Any group of heads which are exhaustively dominated by a given node (i.e. there
is a node which dominates every one of those heads and no others) is a constitu-
ent. For example, we know that the group of words to Madrid is a constituent,
and that's represented in the tree in (35). The PP node dominates the P to and
the N Madrid, and no other heads.
The last relation between nodes that we'll look at in this section will play
a major role in the chapters to come: c-command. C-command was first for-
mulated by a linguist named Tanya Reinhart (1976), and has since come to
play a fundamental role in syntactic theory. The notion of c-command seems
relevant to almost every phenomenon. (We'll see most of the common ones at
some point in this book.)
C-command is usually defined in the following way:

(38) A node u c-commands another node f3 if and only if the first branch-
ing node which dominates u dominates [3.

In the tree in (35), for example, the subject NP node (dominating the agent)
c-commands every other node in the sentence apart from S itself. The first
branching node which dominates the subject NP node is the S node, and that
S node dominates every other node in the sentence. However, the NP Madrid
c-commands only the preposition to. The first branching node dominating
Madrid is the PP node, and that dominates only the preposition. (Just as
a side note, even though it technically satisfies the definition, nodes don't
c-command themselves or any nodes that they themselves dominate.) Using
the relational terms of this section, we could put (38) another way by saying
34 PHRASE STRUCTURE AND CONSTITUENCY

that a node c-commands (I) its sisters and (2) every node that its sisters
dominate.

Draw the tree for 'The man will see the book'. Indicate three pairs of nodes
which are sisters. What nodes does Aux c-command?

Constituents, phrase-structure trees and

the 'triangle' notation
Having discussed various ways of referring to the elements of phrase-struc-
ture trees, I want to end this section by briefly introducing a notational device
that we'll be starting to use in drawing phrase-structure trees and which is
something we'll use a lot during the rest of the book. As we've been discuss-
ing, phrase-structure trees encode a lot of information. But often, when we're
explaining a particular point or showing how a particular analysis works, not
all of that information is equally important. We might want to focus on certain
aspects of the structure of the sentence, but not really be interested in other
details. When we don't want to worry about the details of a particular constitu-
ent because they aren't relevant for our discussion, we'll draw a triangle under
the node in the phrase-structure tree.
So, let's say that we were talking about the sentence in (34) again. The com-
plete tree is in (35), but if we were just making the point that the subject NP
and the VP c-commanded each other, we might draw an abbreviated phrase-
structure tree like (39):

(39) s

NP Aux VP

L I ~
the agent will go to Madrid

The tree in (39) indicates that the agent is an NP and go to Madrid is a VP,
but doesn't indicate the internal structure of these constituents, because
that's not what we're focusing on. As we move on to more complex syn-
tactic analyses and arguments, it'll be handy to be able to take some of the
structure and 'put it off to the side' in this way. However, when drawing
trees for exercise purposes, for example, it wouldn't be appropriate to use
triangles, since the point of the exercise is to show exactly what the com-
plete structure is.
X' -Theory and
Functional
Categories

In the previous chapter, we talked about the fact that, as a native speaker of a
language, you know that sentences are not merely 'beads on a string' - with
one word coming right after the other with no larger organization beyond that.
Instead, what you know is that the elements of a sentence are organized hier-
archically into larger groups, which we called 'constituents'. In this chapter,
we'll talk about a specific way of expressing this knowledge, called X'- Theory,
and look at some of its implications. We'll also introduce some new syntactic
categories, called functional categories (as opposed to lexical categories like N
and V) and examine the role that they play in expressing your knowledge about
your language.

3.1 Introduction to X'-Theory

So far we've done pretty well getting our phrase-structure rules in order, but
you may have noticed a problem creeping into things, particularly when we
introduced the X' level. The problem is redundancy. The changes we made to
the NP phrase-structure rules have a lot in common with the modifications
that we later made to the VP phrase-structure rules.

(1) a. NP -----* (Det) N'

b. N' -----* (AP) N' (PP)
c. N' -----* (AP) N (PP)

(2) a. VP-----* V' (XP)

b. V' -----* V' (XP)
c. V' -----* V (XP)

This is not just a problem with VP and NP. It's a general problem with phrase-
structure rules of the sort that we've been looking at so far. For example, every
NP that we've seen so far has a head N. Every AP has had an A head. Every PP
has had a P head, and so on. We've never seen cases where a VP has an A head,

61
62 X'-THEORY AND FUNCTIONAL CATEGORIES

for example, nor do there seem to be any cases like that in languages of the
world. This might seem like a completely trivial point. I can hear you saying,
'Well, it's obvious. You couldn't have a verb phrase if it didn't have a verb in it
somewhere!' However, it's important to be surprised by 'obvious' things and to
distinguish what's obvious from what the theory tells you should be possible.
If the cognitive system that you possess as a native speaker of a language
can be correctly described as having rules of the form V' ---+ V (XP), why aren't
there rules of the form V' ---+A (XP)? Or perhaps NP---+ P' (XP)? From a for-
malism point of view, there's nothing to distinguish rules like (2) and (1) from
countless other possible rules which we never find. And our theory at this point
doesn't explain why.
It also looks as though our phrase-structure rules are missing a generaliza-
tion. All NPs immediately dominate an N'. All VPs immediately dominate a
V'. All PPs immediately dominate a P'. There's a pattern here.
As first suggested by Chomsky (1970) (a paper termed 'monumentally import-
ant' by Webelhuth (1995b: 18)), we might be able to eliminate the redundancy
and increase our explanatory power by generalizing the phrase-structure rules.
Put another way, Chomsky proposed that what you have as part of your cogni-
tive system is a general format for phrase-structure rules which are independent
of any syntactic category (N, A, V, etc.). So, instead of having rules like (I) and
(2), we have the general schema in (3). This approach is known as X'-Theory
(where X, Y and Z are just variables standing for any syntactic category):

(3) a. XP ---+ (Spec) X'

b. X' ---+ X' (YP)
c. X' ---+ X (ZP)

Although the X can stand for N, V, A or P, you must fill in the X's on both sides
of the arrow with the same choice in any given instance of (3a), (3b) or (3c), so
VP must dominate a V', an A' can only immediately dominate another A' or
an A, and so on. In (3a), 'Spec' is short for 'specifier', a term which we'll define
properly in the next section.
If (3) is part of the cognitive system which enables you to speak and under-
stand your language, then we can explain why many potential rules (like
VP ---+A' PP) aren't found. With the schema in (3), they can't be created. Every
phrase, whether it be NP, VP, AP or PP, must by the rules in (3) ultimately end
up with anN, V, A or P head.
Also, using the general format in (3) eliminates the redundancy among cat-
egories. The reason why all VPs immediately dominate a V' is because they're
following the general schema in (3).

3.2 Inside the XP

Having looked at some general terms which are used to describe relations
between nodes in a tree in the previous chapter, let's take a look at some config-
urational terms relating to the structure of phrases now that we've introduced
 INSIDE THE XP 63

the X' -system. Again, these terms will become second nature in no time, so
don't worry if they seem a little complicated at first.

(4) NP

Det N'

N'

from London
N adjunct

I
student of Greek
head complement

Let's start at the bottom. The most important part of the phrase is the head,
which we've mentioned before in discussing the two-level system. Within the
X'-system, it's common to see the head X of an XP written as 'X"', which is read
as 'X-zero'. (The idea is that the single-bar level is level 1, so one level below
that is level zero.)
Another very common configurational term when discussing XPs is the
complement. Looking at rule (3c) of the general X' schema, the complement is
the optional sister ZP to the head X". A complement is always defined relative
to a head. So in the NP that we drew the tree for in (4), the complement is the
PP of Greek. That PP is the sister to the head N" of the NP. The complement
often corresponds to what would traditionally be called the 'object' of the head,
particularly with verbs and prepositions.
Not all heads have complements, which is why the complement is optional
in the X'-schema rule in (3c). However, when they do, it's often because they
are obligatory for that head. Consider the two different VPs in (5):

(5) a. John slept

b. John destroyed the town

The verb slept in (5a) does not have a complement, but the verb destroyed does
in (5b). The complement is its direct object, and it is obligatory.

(6) *John destroyed

64 X'-THEORY AND FUNCTIONAL CATEGORIES

Identify the head and the complement of the head within the following XPs:
(a) jealous of Federico's achievements
(b) into the fray
(c) buy the record

Working our way upward in the diagram in (4), we distinguish comple-

ments from adjuncts. Whereas complements, when they appear, are usually
obligatory, adjuncts are optional. Adjuncts are immediately dominated by X'
and a sister to X', and so are introduced by rule (3b) above. Being immediately
dominated by X' and a sister to X' they can be easily identified using the sub-
stitution tests discussed in Chapter 2. Consider the VP in (7):

(7) destroyed a town on Friday

Using the do so test, we see that destroyed a town is a V' and that destroyed a
town on Friday is a V'.

(8) John destroyed a town on Friday and ...

Godzilla did so on Saturday.
Godzilla did so as well/so did Godzilla.

Therefore the tree for the VP in (7) is (9).

(9) VP

I
V'

V' pp

v NP
6
on Friday

I
destroyed
D
a town

Of the two NPs, then, the town is a complement, as it is the sister to the head V
of the VP, and the PP on Friday is an adjunct. And they would seem to accord
with the characterizations given above. The complement of destroy is obliga-
tory. It's part of the essential meaning of destroy. (We'll talk about this more in
the next chapter.) On Friday, by contrast, is optional. It simply gives additional
 INSIDE THE XP 65

information about when the destroying the town action took place. In the tree
for the NP in (4), the PP from London is an adjunct.
The fourth and final important configurational term within the XP is the
specifier. Semantically, it often gives more information about the head or makes it
more specific, but for our purposes it is simply the element which is immediately
dominated by XP and is a sister to X' which we saw in rule (3a) of the X' schema.
So in the tree for the NP in (4), the determiner the is the specifier of the NP.
It's important that you remember the 'immediately dominated by XP' part
of the definition though, because that's what distinguishes specifiers from
adjuncts. Both adjuncts and specifiers are sisters to an X', but only the specifier
is immediately dominated by XP. Consider (10), with the tree in (11):

(10) the book about linguistics on the table

(11) NP

Det N'

N' pp

~pp
N'

the
I
No

I
book
D about
linguistics
on the table

Notice that both the determiner the and the PP on the table are sisters to anN'.
However, only one of them is the sister to an N' which is immediately domi-
nated by NP. That's the determiner. Therefore, the is the specifier of the NP in
(10). Both the PPs about linguistics and on the table are adjuncts.

Identify the specifiers in the following XPs. Which of the various PPs are
adjuncts?
(a) several recordings of whales on cassette
(b) three miles down the road
(c) rather appalled by the situation in New Jersey
66 X'-THEORY AND FUNCTIONAL CATEGORIES

3.3 Functional categories

In the previous sections we gave arguments for adopting the X'-system in (12)
as a description of what you know about your language:

(12) a. XP ---+ (Spec) X'

b. X' ---+ X' (YP)
c. X' ---+ X (ZP)

However, in doing so, we focused on phrases like NPs and VPs. The majority
of the categories that we've seen so far, specifically N, V, A and P, are usually
referred to as lexical categories. These contrast with the so-called Junctional
categories Inflection (I), Complementizer (C) and Deteterminer (D). In this
section, we're going to argue that functional categories, just like their lexical
counterparts, also have structures fully consistent with the X' schema just out-
lined. We'll also argue that X' -Theory is just as valid for seemingly more exotic
languages like Turkish and Japanese.

From Aux and S to IP

When we were discussing X' -Theory earlier on in the chapter, you may have
been wondering what happened to the following phrase-structure rule:

(13) S ---+ NP Aux VP

This phrase-structure rule really looks like a problem. If we're right to say that
(3) is the general format that all phrase-structure rules have to follow, then
we'll have to change (13). It doesn't fit the X' format at all. Within the X'
system, there's always a one-to-one relationship between heads and phrases.
Every NP has a head No in it and every head No is the head of some NP and
so on. However, in (13), we've got an S on the left side, but it doesn't appear to
dominate anything of category S. On the right side, all we've got are NP, Aux
and VP. Going the other way, the Aux category would seem to be a head also,
but there's no Aux' or AuxP anywhere to be found.
We can kill two birds with one stone if we say that what's really going on
is that S is essentially an 'Aux-phrase' which has Aux as its head. Also, just
to make a terminological change to bring us into line with the way linguists
talk, we'll also need to change the name of the Aux node to 'Inflection', which
is usually abbreviated to 'I' or 'Infl'. With the addition of an intermediate I'
category, we can replace (13) with something much neater and consistent with
X'-Theory:

(14) a. IP---+ NP I'

b. I' ---+ I" VP

The subject of the sentence is now the specifier of IP (Inflection Phrase) and
the VP is the complement ofi 0 •
 FUNCTIONAL CATEGORIES 67

Although there is no element which can act as a proform for 1', we can use
the coordination test to show that IP needs three levels of structure.

(15) John will read the book and may agree with the conclusions.

What we have in (15) is a conjunction that includes the auxiliary and the VP,
but not the subject. That suggests that these elements form a constituent to the
exclusion of the subject. However, the phrase-structure rule in (13) claims that
these elements do not form a constituent. For that reason, (13) seems to be an
incorrect characterization of what you know when you know English. On the
other hand, the X' -consistent phrase-structure rules in (14) do predict that the
auxiliary (inn and the VP do form a constituent, namely I'.

Using IP, draw the tree for the following sentence:

Mary will read several newspapers.

In those instances where there is no auxiliary, we'll assume that the I" head of
IP contains two pieces of information: the tense of the sentence (e.g. past or
present, since future tense is indicated by the auxiliary will) and the subject-
verb agreement information, as in (16). It is because tense and subject-verb
agreement are marked by inflectional endings on the verb that the whole
sentence is called 'IP'.

(16) John went to the store.

IP

NP I'

6John I VP

I I
V'
[ 3rd sg]
~pp
past

V'

v
I

I
went
6
to the store
68 X'-THEORY AND FUNCTIONAL CATEGORIES

To see concretely why Io needs to have tense and agreement information, let's
examine a construction called 'VP-ellipsis'. Under certain circumstances, when
two sentences are conjoined, the second VP can be deleted:

(17) Mary can't go to the store but John can.

In (17), the VP in the second conjunct is identical to the VP in the first con-
junct (go to the store), but it remains unpronounced, leaving behind the auxil-
iary can. But consider what happens when there is no overt auxiliary:

(18) I buy books every day but John doesn't.

Because there is no auxiliary, the dummy verb do must be inserted in the sec-
ond conjunct. However, notice that the dummy verb shows up in specifically
the present tense (doesn't vs didn't) and third person singular form (doesn't vs
don't). Therefore, it must be the case that the tense and subject-verb agreement
information are (inflectionally) present in Io even when there is no auxiliary to
'spell them out'.
In order to remember this, when drawing a phrase-structure tree with no
auxiliary, be sure to indicate the tense and agreement information in 1°. Thus
the sentence in (19) has the tree in (20):

(19) You buy books on Friday.

(20) IP

NP I'

~
You 10 VP

I I
r2nd sg] V'

~pp
[present

V'

~A
V NP on Friday
I
buy
6
books

In addition to the cases where there is no auxiliary, we also need to say some-
thing about cases where we have more than one auxiliary, as in (21):

(21) You will be buying books on Friday.

 FUNCTIONAL CATEGORIES 69

We'll analyse these cases using what are called 'VP-shells'. We'll see another
use for VP-shells in Chapter 9, but for the moment we'll keep the focus on
how they provide a structure for sentences with multiple auxiliaries. The first
auxiliary is in I", but any other auxiliaries that you might have are within the
VP, as a series ofV's, each of which takes a VP as its complement, until you get
to the main verb:

(22) IP

NP I'

D
You I VP

I I
will V'
~
V'
~VP
V

I
be

~NP
I
V'
Lon Friday

V
16
buying books

The first auxiliary will is in e, but the second auxiliary be is a verb, which takes
a VP headed by buying as its complement. If we had had another auxiliary, say
will have been buying, then will would be in I 0 , with have as a verb taking a VP
headed by been as its complement. Been would then in turn have taken a VP
headed by buying as its complement.
VP-ellipsis again suggests that this analysis is on the right track.
Consider (23):

(23) You have been working hard, but John hasn't.

This suggests that be working hard is a VP, since it can be elided under
VP-ellipsis. Working hard is also a VP, as (24) is also possible:

(24) You have been working hard, but John hasn't been.
70 X'-THEORY AND FUNCTIONAL CATEGORIES

Complementizers and CP
Consider the following sentences in light of the discussion in the previous
section:

(25) Mary will meet her friend at the station.

(26) I was wondering whether Mary will meet her friend at the station.

Clearly (25) is a sentence, so it must be an IP. It seems equally clear that the
italicized string in (26) is also an IP. The question is what to do about whether.
Clearly, we can't just say that the V' wonder takes an IP complement, as in (27):

(27) V'

V' IP

I
wonder

That would predict that (28) should be grammatical, but it isn't:

(28) *I wonder Mary will meet her friend at the station.

In order to account for (26), while not allowing (28), what we need to say is
that the IP Mary will meet her friend at the station combines with some other
category, whose head is whether.
Elements like whether or that (as in I know that Bill is here) are traditionally
referred to as complementizers. They get that name because they are added to
the beginning of an IP in order to help the IP fit in as the complement of a
verb. If the complementizer C (whether in this case) is the head of CP, then
X' -Theory tells us that it must have a specifier and a complement in a three-
level structure:

(29) CP

specifier C'

co complement

I
whether
 FUNCTIONAL CATEGORIES 71

In this case, the IP Mary will meet her friend at the station is the complement of
C. It has no specifier, although we'll come to some examples later which do.
Putting all of this together, we would draw the tree for (26) as (30):

(30) CP
I
C'

C
~IP
~I'
NP

I~P
I
V'

~CP
yo
I
C'

C
~IP
I~
I was wondering whether Mary will meet her
friend at the station

You may be wondering why we've started off with a CP in (30). It's certainly
true that sentences don't start with complementizers:

(31) *That I was wondering whether Mary will meet her friend at the
station.

However, this seems to be an idiosyncratic feature of English. There are many

other languages in which main clauses can start with complementizers. In
Spanish, just to pick one example, main clauses can begin with a complementizer
when the sentence expresses a wish (32), or when something is being repeated
or emphasized (33).

(32) Que se calle Juan

that self silence-3sg Juan
'Let Juan keep silent.'
(Demonte and Fernandez-Soriano 2009: (4c))
72 X'-THEORY AND FUNCTIONAL CATEGORIES

(33) Que no quiero verla

that not want-lsg to see her/it
'I don't want to see her/it.'
(Butt and Benjamin 2000: 448)

We haven't really encountered any non-English data before, so a quick word is

in order regarding how to read the examples. The first line is the sentence in
the original language, converted to a Roman alphabetic script if necessary (as
with, for example, Japanese or Russian). The second line is a morpheme-by-
morpheme 'translation'. For example in (32), the Spanish verb calle contains the
root 'to silence' and a third person singular, present tense marker. The third line
is a translation of the original example sentence into grammatical English.
Although we're focusing on English (since this book is aimed at speakers of
English), we're assuming that the basic structure of the sentence is provided by
UG and so therefore should be the same in all languages. So we don't want to
be misled by idiosyncracies of English. (This is going to come up again later
on in this chapter, and in an even more important way in the next chapter,
when we talk about the concept of 'Case'.) We'll see in later chapters anyway
that there are some English sentences which do start with CP, so, really, the
idiosyncratic nature of English here is in fact only apparent. From here on, we'll
be assuming that all English sentence trees start with CP.

Getting into Det: The DP hypothesis

In the previous sections we've been trying to tighten up our phrase-structure
assumptions a bit to bring them into line with the X' system, eliminating heads
like Aux that had no associated phrases and eliminating 'phrases' like S which
didn't seem to immediately dominate an element of the same category.
Continuing this exercise, let's examine the element Det, which is still out-
side the X' system. Determiners are clearly heads, not phrases, but we've just
been sticking them in the specifier of NP position without saying anything
more about them. In this section, we'll examine some arguments that Dets, in
fact, have associated D's and determiner phrases (DPs). However, it will turn
out that it's not as simple as sticking a DP in the specifier position, as in (34):

(34) NP

DP N'

I
No

There have been various arguments given in the literature that what we've been
calling noun phrases are actually determiner phrases. That is, the structure of
the book actually has a determiner head which takes an NP as its complement:
 FUNCTIONAL CATEGORIES 73

(35) DP

I
D'

NP

Abney (1987) pulls together some observations regarding gerundive 'Pass-

ing' constructions in English to argue for the existence of determiner phrases.
An example of the gerundive 'Poss-ing' construction in English is (36):

(36) [John's building a spaceship] caused a huge commotion.

There's a simple question about phrases like fohn's building a spaceship which
turns out not to have an obvious answer: what syntactic category is it? Is it an
NP? Or is it something sentential, like an IP or a CP?
On the one hand, (36) as a whole looks like it must be an NP as it can turn
up in NP positions. In (37), we see that a 'Poss-ing' phrase can be a subject, and
in (38) the object of a preposition, which is impossible for a CP:

(37) a. I wondered if [John's building a spaceship] had upset you.

b. I wondered if [NP the book ] had upset you.

(38) a. I told you about [John's building a spaceship].

b. I told you about [NP the book].
c. *I told you about fcp that John built a spaceship].

However, it also seems clear that building a spaceship represents a VP. In particu-
lar, the '-ing' element of the gerundive 'Poss-ing' is not a noun which has been
derived from a verb, like destruction or referral (from destroy and refer respect-
ively). Instead it has all the characteristics of a verb. It takes direct objects, for
example, without requiring any kind of preposition, such as of

(39) a. John destroyed the spaceship.

b. John's destroying the spaceship (caused a panic).
c. *John's destruction the spaceship (caused a panic).
d. John's destruction of the spaceship (caused a panic).

The '-ing' element also takes so-called 'verbal particles', which can shift position
just as they would with a 'real' verb:

(40) a. John explained away the problem/John explained the problem away.
b. John's explaining away the problem/John's explaining the problem
away (didn't satisfy his critics).
c. *John's explanation away of the problem/*John's explanation of
the problem away ...
74 X'-THEORY AND FUNCTIONAL CATEGORIES

What we seem to be stuck with are two pieces of a puzzle that don't fit together.
If we just bolt together what we've got so far in a Frankenstein's monster-like
fashion, it's this:

(41)

NP 2 VP

D
John's
~
building a spaceship

Unfortunately, this is not a structure which is allowed by X'-Theory.

Explain why (41) is not allowed by X'-Theory.

The problem is that NP 1 is a noun phrase, but it fails to immediately dominate

anN'. NP 1 doesn't seem to have an No head either.
However, perhaps the sentence-like nature of'Poss-ing' gerundives can offer
a clue as to what to do. Consider the tree for (39a):

(42) IP

NP I'

Io VP

I I
[ 3rd sg]
V'

~NP
past

v
I ~
John destroyed the spaceship

Here, we have a subject john and a VP destroyed the spaceship. A mediating role
is played by the functional category I. It 'hooks up' the VP with its subject. It
 FUNCTIONAL CATEGORIES 75

allows the two to get together. If we assume that there's some mediating cat-
egory in the gerundive 'Poss-ing' construction, that would provide us a way out
of the X' -Theoretic problem that we had in (41):

(43) XP

NP X'

VP

~
John's building a spaceship

The only question now is 'What is "X", our noun phrase correlate of I?'. Well,
there aren't really too many options. It's at least possible that X is some category
which never has any overt words, but that would be an unpleasant thing to have
to say. Even the other functional categories like I and C have at least some actual
words which belong to them (modals like can and will for I and complementizers
like that, whether and ijfor C), and it's not completely clear how a child learning
English would even figure out that this category existed if there were no overt
words of that category. However, it really looks like everything falls together
nicely if we assume that the mystery category is Det, the determiner, and what
we've previously thought of as NPs are actually DPs.

(44)

D'I

VP

, ~
John s building a spaceship

More on DPs: possessives, proper names and pronouns

In this section, I want to tie up a few potentially loose ends concerning this
re-analysis ofNPs as DPs. We'll start with a slightly counter-intuitive analysis of
possessive DPs like our book. We'll then talk about what the implications of the
DP hypothesis are for the analysis of bare plurals, proper names and pronouns.
76 X'-THEORY AND FUNCTIONAL CATEGORIES

Possessive DPs
Given the discussion in the previous section concerning the 'Poss-ing' construc-
tion, you can probably guess what our analysis of possessive DPs like John's book
is going to be. We want to emphasize the parallelism between John's book and
John's building a spaceship, so the structure for the two of them will be essen-
tially identical. In fohn's building a spaceship (44),John is the specifier of the DP,
the head D of the DP contains the possessive marker 's, and the complement
of D is the VP building a spaceship. The analysis of fohn's book is identical, only
with NP, instead ofVP, as the complement of the determiner 's.

(45)

D'I
7i
John Do
II NP
6book
's

This leads, though, to an interesting, and maybe slightly counter-intuitive,

analysis ofDPs in which the possessor is not itself a DP, like john or the student,
but instead is a pronoun, as in (46):

(46) his book

The only real difference between fohn's book and his book seems to be that the
former has a more specific third person singular masculine entity as its posses-
sor (John vs he). Yet, at least on the surface, there's a possessive 'sin the former
but not the latter.
What we'll suggest is that fohn's book and his book are actually more parallel
than the English morphology might suggest, and that the tree for (46) is actu-
ally (47), fully parallel to the tree in (45) above:

(47)

D'I
~
he Do
II NP

's
6book
 FUNCTIONAL CATEGORIES 77

Just as in (45), the DP his book has a specifier. But rather than John it is the
third person singular pronoun he. The head D of the DP contains the posses-
sive marker s, and it takes the NP book as its complement. In other words, as
far as the syntax is concerned, his book is 'really' hes book. Hes book is realized
as his book because of a low-level morpho-phonological rule of English which
says that hes is 'spelled out' as his. In a similar way, our book is 'really' wes book,
their book is 'really' theys book, etc.
For the moment, we'll have to leave only the conceptual argument based
on simplicity, parallelism and theory-neatness considerations as a reason for
assuming (47). To be fair, that's not nothing. The alternative is one in which
johns book and his book have rather different structures. If, for example, we
were to make his a determiner, then one DP would have a specifier and the
other wouldn't. If, on the other hand, his is a specifier, it's not a DP, since
his doesn't occur in places where other DPs do (like subject, object of a pre-
position, etc.).
Furthermore, the 'low-level morpho-phonological rule of English' that we
have to assume which changes hes into his doesn't seem too unreasonable either.
The possessive forms his, our and their, like many other pronoun forms, are
clearly the reflection of an earlier stage of English, and are something speakers
of Modern English simply have to memorize. We will, however, come back to
the question in Chapter 5 and see that Binding Theory provides an empirical
argument for the analysis that we've proposed. So if you're unhappy about (47),
hang on and you might be a little happier later.

Proper names and bare plurals

A sort of theme is beginning to emerge about ways in which what we see on

the surface can be slightly misleading when it comes to the analysis of cer-
tain phenomena. In this section, that theme will continue when we turn to
proper names and bare plurals. The question is what to do with elements like
Alice or tigers. They clearly have to be DPs because they appear in typical DP
positions:

(48) a. I like Alice/tigers.

b. Bill left all his money to Alice/tigers.

Nonetheless, they do not have any overt determiner, and, in the case of proper
names in English, cannot have an overt determiner:

(49) *I like the Alice.

However, the interpretation of bare plurals does seem to include something

determiner-like. (50a), for example, is paraphrasable as (SOb):

(50) a. Cows eat grass.

b. All cows eat grass.

Neither is the paraphrase limited to all: other contexts appear to allow other
paraphrases with other determiners, like most or some ...
78 X'-THEORY AND FUNCTIONAL CATEGORIES

(51) Cats are smaller than dogs/Most cats are smaller than most dogs.

(52) People approached the UFO cautiously/Some people approached the

UFO cautiously(, but a few stayed back).

Furthermore, in the case of proper names, other languages do allow them

to be used with (definite) determiners. Southern varieties of German, Ancient
and Modern Greek, and certain non-prestige dialects of Castilian Spanish are
all examples. We're assuming that the determiner system, and functional cat-
egories in general, form part ofUG. So, bearing in mind the discussion about
description and explanation back in Chapter 1, we want to check a wide range
of data to make sure that we're not being misled. Therefore, we will assume
that bare plurals and proper names are DPs which contain a null determiner,
as illustrated in (53).

(53) DP

I
D'

D NP

I I
<I> N'
I
N

I
Alice I tigers

Pronouns
Moving on from proper names, you might assume that we would analyse pro-
nouns in the same way. (We haven't been explicit about their structure so far,
indicating them merely as a DP with a triangle.) However, it looks as though
pronouns have some important differences. First, pronouns can occur together
with Ns in a way that proper names can't:

(54) We linguists have got to stick together.

These cannot themselves occur with a determiner, so can't, for example, be

analysed as compound nouns:

(55) a. the rose garden

b. *the we linguists

Also, pronouns can occur together with adjectives and nouns:

 COORDINATION AND X'-THEORY 79

(56) You blue aliens need to fill in these forms.

This suggests that pronouns are actually the reverse of proper names. Rather
than being OPs which contain a null determiner, the pronouns themselves are
Os (which take a null NP sister when they appear alone).

(57) OP

I
0'

0 NP

I I
we N'
I
N

I
<P

3.4 Coordination and X'-Theory

The final functional category which we'll consider in this chapter is conjunc-
tions, as illustrated by (58):

(58) John and Bill went to the store.

In Chapter 2, when we were illustrating the coordination test, we weren't

explicit about what phrase-structure trees we were assuming for coordinated
structures. Our hypothesis is that X' -Theory is provided by UG and creates
a general framework which all phrase structure conforms to. If that's correct,
then we would hope to find an X' -consistent way of incorporating sentences
like (58).
Various researchers have suggested (59) as a way of meeting our requirements:

(59) ConjP

Conj'

Conj