You are on page 1of 11

3rd Lecture: Noun Phrases I

LT 3209 Syntax
Prof. Matthias GERNER
Room B 7622
mgerner@cityu.edu.hk
http://www4.lt.cityu.edu.hk/~mgerner/
Q-8

Instructor:
Office:
e-mail:
Web:
Assignment Locker:
Outline of session:
1 N-bar categories

1.1 Intermediate Constituents (IC)


1.2 Empirical evidence for N-bar
General Structure of Noun Phrases
2.1 Complements and Adjuncts
2.2 Optional Constituents

1 N-bar categories
In the first lecture we have argued that there are two types of categories: word-level categories and
phrase-level categories:
Word-level categories: N = Noun; V = Verb; A = Adjective; P = Preposition; ADV = Adverb; M =
Modal; D = Determiner etc.
Phrase-level categories: NP = Noun Phrase; VP = Verb Phrase; AP = Adjective Phrase; PP =
Prepositional Phrase; ADVP = Adverbial Phrase etc.
1.1 Intermediate Constituents (IC)
In this lecture we postulate a wide-ranging internal structure for all Noun Phrases and argue that in
addition to word-level and phrase-level categories, we should acknowledge a third category layer in
between word-level and the phrase-level. We will now argue in favor of the existence of constituents
that are larger than words but smaller than phrases.
Consider the following example that has become famous in the Generative Literature in the last 60
years.
(1) The king of England
Lets analyze its structure. First of all, there is no doubt that it is a noun phrase, as it can take the
genitive inflection which is a feature of noun phrases (not of bare nouns).

(2) The king of Englands crown


Concerning the direct constituents of the noun phrase [the king of England], it seems obvious that [of
England] is a PP constituent. There are two or three tests that might reveal this status. For example, it
is possible to coordinate [of England] with another similar PP.
(3) The king [PP of England] and [PP of Wales]
Furthermore, it can also function as the shared constituent in a Shared Coordination:
(4) He is the king and she is the queen [PP of England].
These tests give credit to the claim that [of England] is a PP constituent of the overall Noun Phrase.
But is it an immediate constituent of the noun phrase? Well, not exactly. Linguists have debated this
question abundantly and proposed the following working hypothesis:
NP
D

the

PP

king

P
of

N
England

The difficulty is to name appropriately the node for [king of England] termed by ?. An obvious
suggestion is to say that [king of England] is just another Noun Phrase with the following tree
structure.
NP
D
the

NP
king of England

But the replacement test shows that it is not a noun phrase at an equal footing with [the king of
England]:
(5) The king of England
*king of England

opened the parliament.


2

(6) They crowned

the king of England

yesterday.

*king of England
Furthermore, if we were to allow [king of England] as Noun Phrase and also [the king of England] as
Noun Phrase, it would imply that noun phrases can be derived from other noun phrases by adjoining
a determiner. We would thus have to admit a generative deduction rule like:
(7) NP D NP
But such a rule is recursive (in that the symbol NP occurs on both sides of the arrow). Applying this
rule a number of times we will receive multiple determiners, such as a the that king of England.
The structure obtained is represented by the following tree diagram:
NP
D

NP
D

NP
D

NP

The problem with these structures is that they do not represent any spoken English. It is
ungrammatical in English to compound more than one determiner. Examples include:
(8) (a) *the the king of England
(b) *a the king of England
(b) *the this king of England
(c) *our your king of England
(d) *an our king of England
(e) *that their king of England
True, have said some linguists, but remarked that the ungrammaticality of several sentences above is
semantic in nature. For example the concepts of a and the are contrary. Should we prevent
syntactic rules like (7) from generating sentences such as (8) (a-e), since the role of syntactic rules is
to generate syntactically well-formed structures? The question we should then ask is whether there is
any double-determiner NP that is well-formed. Or, are all double-determiner combinations ill-formed?
Well, there is no double-determiner construction in English that is well-formed. There is, however,
one subgroup of Determiner+Possessive sequences that are ill-formed, but that can be interpreted
semantically.
(9) (a) *a my book
(b) a book of mine
3

(10) (a) *this your car


(b) this car of yours
(11) (a) *some your friends
(b) some of your friends
The fact that these ill-formed (a) versions can be semantically interpreted shows that the
ungrammaticality of double-determiner constructions is not semantic in nature but rather syntactic
although there are also instances of semantic incompatibilities (e.g. a <> the or indefiniteness <>
definiteness).
The consequence of this discussion is that double-determiner constructions are syntactically
ill-formed and that determiners do not directly modify noun phrases but something smaller than noun
phrases, though larger than single nouns.
In the 1950s, when Structural Linguistics, the predecessor of Generative Grammar, was in full swing,
a linguist called Harris invented a number system that labels the different constituents of the noun
phrase. Adopting his system we might argue that the noun king is a constituent of type N0, that the
constituent labeled by ? = [kind of England] is a single phrasal expansion of king and hence of type
N1. The full noun phrase [the king of England] is a double phrasal expansion of the head noun king
and thus of type N2.
(12)
N2
N1

D
the

N0
kind

PP
of England

In this notation N2 corresponds to NP in the old system, N1 does not have a counterpart, whereas N0
corresponds to N. In addition to this numerical superscript notation, there are two other fully
equivalent writing notations due to two other linguists, i.e. Chomsky (1970) and Jackendoff (1977).
These notations are notational variants, different ways of saying the same thing.
NUMBER NOTATION

BAR NOTATION

PRIME NOTATION

(Harris 1951)
N0 (N-Zero)
N1 (N-One)

(Chomsky 1970)
N
N (N-Bar)

(Jackendoff 1977)
N
N' (N-Prime)

N2 (N-Two)

N (N-Double-Bar)

N'' (N-Double-Prime)

Given that these three notational systems are entirely equivalent, it is not surprising to find that they
4

are used interchangeably. For example Jackendoffs book uses the bar notation in the title (it is called
the X-bar syntax), but uses the prime notation throughout the rest of the book.
For typographic reasons, we will not use the bar notation, but the prime notation. The latter is easier
to represent in Word, though we will use N-bar, when we speak about given phenomena. This is
current practice in Generative Grammar. All three notations coexist without contradiction. We will
thus say that [king] is an N, that [king of England] is an N-bar and that [the king of England] is an
N-double-bar. At the same time we will represent tree-diagrams with primes:
(13)
N''
D
the

N'
N
kind

PP
of England

1.2 Empirical evidence for N-bar


Until now we have only seen a justification for the fact that [king of England] is not a full noun
phrase. Is there now real empirical evidence for the existence of N-bar, for the existence of an
independent category [king of England]? Or is it a mere theoretical construct of linguists?
Yes there is evidence. We may simply apply the usual syntactical tests and will see that sequences
like [king of England] can be submitted to syntactical tests. From these tests we will have to
conclude that [king of England] has the status of constituent, since only independent constituents can
undergo such tests.
First, the sequence [king of England] can undergo the test of Ordinary Coordination with another
similar sequence.
(14) Who would have dared defy the [king of England] and [ruler of the Empire]?
In this construction N-bar is coordinated not N-double-bar. Ordinary coordination of N-double-bar
would lead to expressions such as [the king of England] and [the ruler of the empire]. Moreover,
N-bar sequence [king of England] can function as the shared constituent in cases of Shared
Coordination. Consider for example
(15) He was the last and (some people say) the best [king of England].
5

Additional support comes from the Proform test. In Proform tests, only unitary constituents, i.e.
constituents that reveal an internal unity, not just any fragment, can be replaced by a proform. In the
following three sentences, only the N-bar constituent [king of England] can be replaced by a proform,
but not the N-constituent [king] and not the N-double-bar constituent [the king of England].
(16) (a) *The [king] of England defeated the one of Spain
(b) The present [king of England] is more popular than the last one
(c) *Respecting [the king of England] is better than rebelling against one.
We can say that one is an N-bar proform, because it can exactly stand for N-bar constituents and not
for mere N-constituents or N-double-bar constituents.
So, both the Coordination tests and especially the Proform test provide strong empirical support for
the N-bar analysis. The phrasal constituent [king of England] is therefore not only a theoretical
construct of linguists, but a real phrasal constituent that is recognized by the language through
empirical data. We shall henceforth assume that there is indeed an intermediate constituent called
N-bar that is larger than N but smaller than NP. There are thus three types of nominal constituents: N,
N' and N''.
2

General Structure of Noun Phrases

2.1 Complements and Adjuncts


The aim we are pursuing in this lecture is to feature the general structure of Noun Phrases, abstract
enough to represent the internal composition of every noun phrase. Through what we have seen so far,
we are tempted to pose the general structure of noun phrases as follows:
(17)
N''
Determiner

N'
N

Postnominal PP

We might say that Postnominal PPs expand N-constituents into N-bar constituents and that
Determiners expand N-bar constituents into N-double-bar constituents. It is rather straightforward to
admit that we have a general higher-level structure of the following type:
(18)
N''
Determiner

N'
6

But what is the lower-level internal constituent structure of N'? Is it true that all postnominal PP
and indeed all postnominal phrases of any kind have essentially the same constituent structure?
(19)
N'
N

Postnominal Phrase

Well, not all postnominal phrases have the same constituent structure and status. In traditional
grammar, one distinguishes between Complements and Adjuncts. In spite of our general skepticism
about everything traditional, we have to admit that intuition coming from traditional grammar is
well-founded sometimes.
The following two phrases constitute examples for Complements and Adjuncts:
(20) (a) a student [of Linguistics] (= Complement)
(b) a student [with long hair] (= Adjunct)
What is the point of distinguishing between Complements and Adjuncts? The postnominal phrase [of
Linguistics] tells us something about the activities of the head noun. The head noun student is
derived from the verb study. The constituent [of Linguistics] can be transformed into a Complement
of the verb study:
(21) (a) He is [a student of Linguistics].
(b) He is [studying Linguistics].
In a similar way, the constituent [of England] is a Complement of [king]. Taking [king] as a semantic
equivalent of [ruler] we may paraphrase:
(22) (a) He is [the king of England].
(b) He is [ruling (over) England].
This type of paraphrase is unavailable for postnominal phrases like [with long hair], as we can see in
the following example:
(23) (a) He is [a student with long hair].
(b) ?He is [studying long hair].
Although (23b) is grammatical, it is not a paraphrase for (23a). The constituent [with long hair] does
not tell us anything about what the student is studying; it merely serves to give us additional
7

information about the student. The kind of postnominal phrase found in (20b) or in (23a) is called
Adjunct.
The obvious question to ask now is what are the structural properties of the Complement-Adjunct
distinction and how do Complements and Adjuncts differ from the other class of nominal modifiers
we are already familiar with, namely Determiners? The main claim we can make about their
differences is that
(24) (a) Complements expand N into N-bar
(b) Adjuncts expand N-bar into N-bar
(c) Determiners expand N-bar into N-double-bar
In an alternative wording, we may represent visually the general structure of the Noun Phrase as
follows:
(25)
N''
Determiner

N'
N'

Adjunct

Complement

We can see from (25) that Determiners are sisters of N-bar and daughters of N-double-bar. Adjuncts
may function both as sisters and daughters of N-bar. This means that Adjuncts resemble
Complements in that both are daughters of N-bar; but Adjuncts differ from Complements in that
Adjuncts are sisters of N-bar, whereas Complements are sisters of N.
In a similar way, Adjuncts resemble Determiners in that both are sisters of N-bar, whereas they differ
from Determiners in that Determiners alone are daughters of N-double-bar, while Adjuncts bear a
daughter relationship only to N-bar. Another wording of the same situation is through respective
Phrase Structure Rules that generate Determiners, Adjuncts and Complements.
(26) (a) N'' D N' [called the Determiner Rule]
(b) N' N' PP [called the Adjunct Rule]
(c) N' N PP [called the Complement Rule]
When you have a look at the tree diagram above and at the Phrase Structure Rules, you will notice
that this structure makes a claim about the relative order of Complements and Adjuncts in a phrase.
Complements will always stand closer to the head noun than adjuncts, especially in constructions in
8

which both collocate. Consider for example:


(27) (a) the student [of Linguistics] [with long hair]
(b) *the student [with long hair] [of Linguistics]
If we were to present the structure of (27b) we could draw the tree diagram. The problem with this
tree is that its branches cross. In the previous session, we have encountered the NO CROSSING
BRANCHES CONSTRAINT, which disallowed branches in a tree diagram to cross. This constraint
accounts for the ungrammaticality of noun phrases such as (27b) and for the relative order of
Complements and Adjuncts.
(28) *the student [with long hair] [of Linguistics]
N''
D

N'

the

N'
N

PP (Adjunct)
with long hair

PP (Complement)

student

of Linguistics

It follows that Complements must occur closer to their head nouns than Adjuncts. Let us now
generate the phrase-marker (tree diagram) of (27a) by strictly applying the Phrase Structure Rules:
(26) (a) N'' D N' [Determiner Rule] {1st Step}
(b) N' N' PP [Adjunct Rule] {2nd Step}
(c) N' N PP [Complement Rule] {3rd Step}
(29) the student [of Linguistics] [with long hair]
N''
D

N'

the
N
student

N'

PP (Adjunct)

PP (Complement)

with long hair

of Linguistics
9

2.2 Optional Constituents


Following me up to this point you might have got the impression that every noun phrase is composed
of a Determiner, a Complement and an Adjunct. However, any of these constituents may be optional.
What I mean with optional is that there might be noun phrases without Determiners, without
Complements or without Adjuncts. I do not mean with optional that every Noun Phrase can be used
with a Determiner as well as without Determiner, or can be used with a Complement as well as
without Complement, or with an Adjunct as well as without Adjunct. The conditions that determine
whether a Determiner can or cannot be omitted from a Noun Phrase will not bother us in this course.
The questions that should interest us during this course are
-

What is the structure of Noun Phrases that lack Determiners?,


What is the structure of Noun Phrases that lack Complements?,
What is the structure of Noun Phrases that lack Adjuncts?.

Concerning Noun Phrases that lack Determiners, let us consider the following phrase:
(30) Students of Linguistics with long hair
First question we should ask: what is the categorical status of this sequence? Is it an N-double-bar
constituent, an N-bar constituent or something else? Well, it is a full Noun Phrase (a N-double-bar
constituent). We can prove this with the usual syntactical tests, for example with the
Sentence-Fragment test.
(31) Speaker A: What kind of students do you hate teaching?
Speaker B: [Students of Linguistics with long hair]
The Proform test delivers a similar result:
(32) [Students of Linguistics with long hair] sometimes think the world owes them a living.
In fact [Students of Linguistics with long hair] is a full Noun Phrase. Now, how can we account for
the fact that [Students of Linguistics with long hair] lacks determiners?
Simply by replacing our set of rules by another rule. We may replace the rule N'' D N' by the two
rules:
(33) (a) N'' D N'
(b) N'' N'
The adopted writing fashion now is to collapse both rules into the following rule:
10

(34) N'' (D) N'


The sequence [Students of Linguistics with long hair] would thus correspond to the following tree
diagram.
(35) [Students of Linguistics with long hair]
N''
N'

N
students

N'

PP (Adjunct)

PP (Complement)

with long hair

of Linguistics

It is not just Determiners which are optional constituents of Noun Phrases. Complements and
Adjuncts may be optional as well as shown in the following examples:
(36) (a) a student [of Linguistics] [with long hair] (with Complement and Adjunct)
(b) a student [of Linguistics] (with Complement, without Adjunct)
(c) a student [with long hair] (without Complement, with Adjunct)
(d) a student (without Complement and Adjunct)
How can we generate such reduced Noun Phrases? We may amend the Phrase Structure Rules.
(37) (a) N' N' PP [new Adjunct rule which is the same as the old one]
(b) N' N' [new Adjunct rule: new add]
(c) N' N PP [new Complement rule which is the same as the old one]
(d) N' N [new Complement: new add]
We may collapse these rules into the following rules
(38) (a) N'' (D) N' [called the Optional Determiner Rule]
(b) N' N' (PP) [called the Optional Adjunct Rule]
(c) N' N (PP) [called the Optional Complement Rule]

11

You might also like