BRUCE L.

LILES
AN
INTRODUCTORY
TRA NSFORMA TIONA L
GRAMMAR
AN
INTRODUCTORY
TRANSFORMATIONAL
GRAMMAR
Prentice-Hall International, Inc., London
Prentice-Hall of Australia, Pty. Ltd., Sydney
Prentice-Hall of Canada, Ltd., Toronto
Prentice-Hall of India Private Limited, New Delhi
Prentice-Hall of Japan, Inc., Tokyo
Bruce L. Liles
University of Missouri, St. Louis

AN

INTRODUCTORY
TRANSFORMATIONAL
GRAMMAR
Prentice-Hall, Inc., Englewood Cliffs, N.J.
For Robert W. Ackerman

© 1971 by Prentice-Hall, Inc. Englewood Cliffs, New

Jersey. All rights reserved. No part of this book may
be reproduced in any form or by any means without
permission in writing from the publisher.

13-502294-0 (C)
13-502286-X (P)

Library of Congress Catalog Card Number: 70-122388

Printed in the United States of America.

Current printing (last number):

10 987654 3 2 1
 PREFACE

Ten years ago a person could read most of the important works in
linguistics in a few months, and the scholar could afford to buy most of them
for his library. Since that time research in all phases of linguistics has
progressed so rapidly that scholars who have been keeping up with develop¬
ments in their discipline are finding it difficult to read all the relevant works
published each year. The person wanting to begin a study of linguistics is
overwhelmed by the amount of material available, much of it far too
technical for him to understand. The present book is directed toward this
person, and it is limited in scope to English transformational grammar.
The student of transformational grammar needs both theory and
applications. This book attempts to fuse the two. Also, it shows the reader
that there are still many aspects of the English language that are poorly
understood and that are being investigated. The exercises at the end of the
chapters allow the reader to work with the material that has been discussed
and to use the techniques he has learned.
I would like to express my gratitude to the people who helped me in
writing this book: to David H. Grady, who suggested that I write it and who
worked with me in the initial stages of production; to Frances Cassel and
James Foster, who read the manuscript and offered useful suggestions; to
Wanda Giles, who proofread the final copy. It would be impossible to list
all the books and articles in journals that have affected my thinking; their
influence is found on almost every page of this book.

v
 CONTENTS

PART ONE PHRASE STRUCTURE I

chapter I grammars of english 3

Traditional Grammar 3
Structural Grammar 5
Transformational Grammar 6

chapter 2 the structure of the sentence 10

chapter 3 the auxiliary 19
chapter 4 lexical features 27
Divisions of a Transformational Grammar 27
Transitive and Intransitive Restrictions 30
Subject-Verb Restrictions 33
Determiner and Noun Restrictions 36
The Lexicon 39

PART TWO TRANSFORMATIONS I 41

chapter 5 the negative transformation 43
chapter 6 question transformations 51
chapter 7 transformational processes 59
Adverbial Movement 60
Indirect Objects 62
Imperatives 64
chapter 8 the passive transformation 67
VII
viii contents

PART THREE TRANSFORMATIONS II 74

chapter 9 principles of transformational grammar 75
chapter 10 compounding, deletion, and pro forms 83
chapter I I relative constructions 91
chapter 12 sentences as noun phrases < 102
%

PART FOUR PHONOLOGY 115

chapter 13 transcription 117
chapter 14 phonological features I 127
chapter 15 phonological features II 133
chapter 16 syntax and stress 141
chapter 17 stress and pitch 150
chapter 18 phonological rules 157
Bibliography 162
Index 165
part one

PHRASE STRUCTURE
 chapter one
GRAMMARS OF ENGLISH

TRADITIONAL GRAMMAR

Until the sixteenth century, Latin was the primary language of scholarship
in England and the rest of Western Europe. The only grammars studied in
English schools, therefore, were Latin grammars, which were designed to
give Englishmen the skills needed to read, write, and sometimes converse
in this lingua franca of Western Europe. During the sixteenth and seven¬
teenth centuries, grammars of English began to appear. Since these early
grammars were studied exclusively as an aid to learning Latin, they
rigorously followed Latin models, although the structure of English is quite
different from that of Latin.
For an example of differences between these languages, we can look at
how they indicate such relationships as performer and receiver of action.
In Latin these relationships are expressed primarily by the forms of the
words. In the sentence Puer virum videt, “The boy sees the man,” we
know that puer, “boy,” is the one performing the act and virum, “man,” is
the one receiving it because there is no ending on puer, but - urn is added to
vir to give virum. The sentence would mean the same thing if the word
order were reversed: Virum puer videt. To alter the meaning, one must
change the forms of the words, not their order; hence, Vir puerum videt
means “The man sees the boy,” since vir has no ending and puerum ends
in - um. In contrast, by the sixteenth century English was relying exclusively
on word order to indicate these relationships. The sentence The boy sees
the man shows that the boy is the performer because the word boy precedes
the verb; the man is the receiver, since man follows the verb. A change in
word order produces a change in meaning (The man sees the boy) or a
meaningless sentence (Sees the man the boy). Logically, a grammar of
Latin should discuss the forms of words, whereas one of English should give
more emphasis to the order in which they are arranged. The purpose of the

3
4 chapter one

early grammars of English, however, was not to provide an accurate de¬

scription of the language, but rather to serve as a basis for the study of Latin
grammar. English word order was largely ignored, and the meaning of such
relationships as actor and receiver was emphasized as a preface to Latin
forms.
During the Renaissance, Latin was replaced as the language of
scholarship by English and the other Western European languages. Because
of this new role for English, some Englishmen by the late seventeenth and
early eighteenth centuries were greatly concerned with refining their
language. They felt that English had somehow become “corrupt” and that
it was in need of purification. Through the use of logic they hoped to return
English to an “undefiled” state. With unlimited zeal and imagination but
limited knowledge about language change and the history of English, they
added the s we now have in island and the b of do u bt and debt*)*; they made
other changes based on equally misunderstood etymologies. They reasoned
that since a subjective complement means the same thing as the noun it re¬
names, people should say It is I rather than It is me, that one should use the
nominative he in Bill wanted to be he but the objective him in Bill wanted it to
be him. Unfortunately, they were silent about whether to use the nominative
or the objective in We discussed its being -. By logic this should be
We discussed its being my, the possessive my agreeing with its. They
reasoned out rules for shall and will, should and would. They legislated
against ending sentences with prepositions and against splitting infinitives.
These rules were based entirely on logic; no attention was given to what
educated people were actually saying. For their classification of words and
sentences, they followed the patterns set by grammars of the preceding two
centuries.
From this tradition developed the English grammars used in schools
during the nineteenth and twentieth centuries. This traditional grammar
is best known to many people in the United States from high-school text¬
books, college handbooks, Plain English Handbook by Walsh and Walsh, and
Descriptive English Grammar by House and Harmon. It followed Latin
grammar in concentrating on parts of speech that are subcategorized
according to case, person, number, gender, mood, tense, etc. These concepts
are informative in a study of Latin, but many of the categories are hard to
justify for English. Word order was usually ignored. Sentences were
classified as simple, compound, complex, or compound-complex; clauses
were classified as independent, noun, adjective, or adverb; phrases were

| English iland was respelled island by analogy with French isle, although the two words
are not related etymologically. The b in doubt and debt was added by analogy with Latin
dubitum and debitum, although Old French had lost the b in these words before English
borrowed them.
 grammars of english 5

prepositional, participial, gerund, or infinitive. Many school grammars

used the study of English grammar as nothing more than a background for a
study of punctuation, subject-verb agreement, pronoun case forms, and other
matters of usage.

STRUCTURAL GRAMMAR

During the nineteenth century, as scholars began studying and

comparing large numbers of languages, many of them radically different in
structure from Latin, they saw that traditional grammar was inadequate.
Some missionaries and other people describing exotic languages assiduously
forced them into the framework of Latin grammar, but others realized the
shortcomings of the tradition. This dissatisfaction with traditional grammar
continued into the twentieth century, when Jespersen and Poutsma found
it necessary to make significant departures from traditional grammar in
their monumental works on English, f Many teachers became disillusioned
with the grammar they were teaching when they discovered that it would not
account for many ordinary sentences that are encountered in modern
newspapers and magazines; for exercises they were limited to carefully
selected sentences in their texts. They often found that in their teaching
they were relying more upon observations they had made about the language
than upon explanations in the texts.
Discontent with traditional grammar grew to such proportions that
during the second quarter of the twentieth century a new approach to the
study of language evolved: structural linguistics. Followers of this approach
felt that it was necessary to study the structure of a language as objectively
as possible without reference to any other language, and they felt that
meaning was a poor guide to the analysis of structure. Instead of talking
about what a noun means (“the name of a person, place, or thing”), for
example, they began looking for other devices to identify nouns. In a
sentence such as The arguments became heated, they said that the word
arguments can be recognized as a noun because it has a plural ending,
because it has the suffix -ment, because it follows the determiner the, and
because it precedes the verb became. They attempted to analyze other
grammatical elements in terms of structure rather than meaning.

f Otto Jespersen, A Modern English Grammar on Historical Principles, 7 vols. (Copenhagen:

Ejnar Munksgaard, 1909-1949). H. A.Poutsma, A Grammar of Late Modern English (Groningen:
P. Noorhoff, 1914-1926). Jespersen and Poutsma belong neither with the traditional
grammarians discussed in the preceding section nor with the structuralists. Gleason in his
Linguistics and English Grammar (New York, 1965) calls them “scholarly traditional”
grammarians.
6 chapter one

The structuralists based their conclusions upon analyses of sentences

that they had collected from native speakers of English. One of the most
remarkable of the structuralists was Charles C. Fries, who obtained access
to letters written to a government agency as a corpus for his analysis presented
in American English Grammar (New York: Appleton-Century-Crofts, 1940).
Later he obtained permission to record telephone conversations, unknown
to the people talking. The results of this study''were published in The
Structure of English (New York: Harcourt, Brace and World, 1952). Because
of Professor Fries’ integrity and efficiency, access to the materials was
carefully controlled. The illustrative material in the published results of
the studies had all names replaced by dashes, and sentences that could
identify the speaker because of their content were carefully excluded.
Professor Fries’ work was widely read and emulated by later structuralists
such as Francis, Hill, and Stageberg.f
The work of the structuralists brought attention to the word linguist,
a term that had previously been used to designate someone who studied
languages, particularly someone who spoke several different languages.
But after the publication of Leonard Bloomfield’s Language in 1933, the term
linguist became specialized to mean only the person concerned with the new
scientific study of language. The word linguistics became popular as the
name of his discipline. Until the late 1950’s, the terms structuralist and
linguist were practically synonymous in America. The structuralists made
many praiseworthy contributions to the study of language. They challenged
current attitudes and began the scholarly study of language that has rapidly
increased during the last two decades.

TRANSFORMATIONAL GRAMMAR

Starting formally in 1957 with the publication of Noam Chomsky’s

Syntactic Structures, a new approach to the study of language was inaugurated.
This newer grammar has gone under various names: generative, transformational,
generative-transformational, and transformational-generative. This is the kind of
grammar presented in this book, and the term transformational is used for
consistency and brevity. Scholars have been so impressed by the importance
and potential of transformational grammar that since 1957 the majority of
published studies of English syntax have used this approach. During the

| Archibald A. Hill,Introduction to Linguistic Structures (New York: Harcourt, Brace and World,
1958). W. Nelson Francis, The Structure of American English (New York: The Ronald Press
Company, 1954). Norman C. Stageberg, An Introductory English Grammar (New York: Holt,
Rinehart and Winston, Inc., 1965).
 grammars of english 7

past ten years there have been many developments in transformational

grammar, and there will certainly be more in the future. Almost every
aspect of language is still being examined. There is so much controversy
among linguists that many people use the plural in speaking of transfor¬
mational grammars:; there is no single transformational grammar which is
accepted by all scholars in the field. Nevertheless, there are many ideas
which most transformationalists do accept.
The transformational grammarian is not content with describing what
he finds in a corpus of sentences collected from native speakers. He feels
that his grammar should enable one to produce all the sentences of a language,
and he is as interested in possible sentences as he is in the ones actually
recorded. Since the number of possible sentences in English or any other
language is infinite, no one could have heard all of them. Yet native
speakers of English understand new sentences such as I spilled milk in the
bathtub and He left his shoe polish in the refrigerator. Every day the
native speaker hears, reads, and creates new sentences, sentences which
seem so ordinary that he is not aware that they have never been used before.
An adequate grammar of English should enable a person to produce not
just those sentences that have been said in the past, but all the sentences
that a native speaker is capable of creating or understanding. In addition,
the grammar should not generate sentences that a native speaker would
reject, such as *The man horrified the door or *Boy on the roof is.
Notice the asterisk in the last two sentences. This mark is used to
indicate that a sentence or a part of a sentence is ungrammatical; that is, no
native speaker of the language would intentionally use it.f By grammatical
we are not referring to standard and nonstandard usage. He ain’t going is
grammatical for some people (there are native speakers who say it), but it is
nonstandard (educated speakers do not normally use it). *He not is going
is not grammatical, since no native speakers use this construction. Calling
sentences like He ain’t going and Irregardless of what you think, he taken it
grammatical does not mean that the linguist is encouraging anyone to use
them when applying for a job. He is merely restricting the meaning of the
term grammatical to structures that are normally used by native speakers of
English. He refers to acceptability among various social groups or in
various contexts as usage.
The transformationalist is more concerned with the system that
underlies the language than he is with the actual speech of an individual
at any given time. All speakers occasionally stammer, make false starts, use
wrong words, get words out of order, and change constructions in midcourse.

f The word intentionally is important. Native speakers do make ungrammatical sentences,

especially when they are excited or tired. These are normally looked upon as unintentional
slips.
8 chapter one

In addition, speech may be affected by physical surroundings, emotions,

memory limitations, distractions, or other features such as chewing gum in
the mouth of the speaker. It is language (the underlying system), not actual
speech output, that is of primary interest to the transformationalist. Another
way of stating this is to say that he is interested in the speaker’s competence,
or knowledge of the language, rather than in his performance, or actual use
of it.
In some respects transformational grammar is similar to traditional
grammar. Transformational grammar assigns each sentence an underlying
structure that is called a deep structure. Some traditional grammars used a
similar concept in speaking of “understood” elements. For example, they
said that Tom is taller than I has the underlying form Tom is taller than I
am tall and that imperative sentences such as Come here have an under¬
stood subject you. Transformational grammarians agree, but apply this
idea of underlying structure to every sentence and express it in a more
abstract form than traditional grammarians did.
As transformationalists began studying deep structures, they noticed
that languages which are quite different on the surface often show many
similar features in their deep structures. Some linguists are now investigating
the possibility that there is a universal deep structure underlying all lan¬
guages. Sentences having this universal deep structure are converted into
the sentences of particular languages by a process known as transformation.j*
Although this line of investigation is new and offers almost limitless pos¬
sibilities for further research, enough discussions of it have been either
published or presented as papers at professional meetings to make the idea
of a universal deep structure seem plausible. It will probably be several
years before we understand universal deep structure well enough to incor¬
porate information about it into an introductory text. A slightly earlier
concept has, therefore, been adopted for this text. The term deep structure
will be used to refer to an English deep structure, not a universal one.

exercises

A. Make up five sentences that you think you have never heard, read, or said before.
B. Select any book at random and make a list of the sentences you find repeated in it.
Set yourself a reasonable time limit or number of pages. What does your list
suggest about the number of sentences possible in English ?

f The concept of a universal grammar was discussed during the seventeenth and eighteenth
centuries and was then ignored for two centuries. Many linguists now regard this early idea
as essentially correct. For a discussion of the theories of the seventeenth and eighteenth
centuries, see Noam Chomsky, Cartesian Linguistics (New York: Harper and Row, 1966).
 grammars of english 9

C. Reread this chapter and mark the sentences you think you have read or heard
before. Why were you able to understand the sentences that were totally new
to you?

D. Listen to a conversation that you are not actively participating in and find
examples of sentences that do not accurately reflect the speaker’s competence.
Are false starts, incomplete structures, and other mistakes very common? Why
do they normally go unnoticed ?

E. We could make an analogy ^to linguistic competence and performance with a

student’s performance on a test. Show how his performance may not accurately
reflect his competence, or knowledge of the subject matter. How many of these
interfering factors also prevent a person’s performance from accurately reflecting
his competence in language?
 chapter two
THE STRUCTURE
OF THE SENTENCE

If you examine the sentence Yes, my neighbor has seen the dog, you notice
that the words are arranged in a definite order. Any other arrangement is
ungrammatical:

1. *My neighbor has seen yes the dog.

2. *Yes, neighbor my has seen the dog.
3. *Yes, my neighbor the dog has seen.
4. *Yes, my neighbor seen has the dog.

Because of the obvious importance of word order in English, we will be

calling attention to this process throughout the book.
Not only do the words have a prescribed order, but they also cluster
together in groups within the sentence. My neighbor forms a group, as does
the dog; on the other hand, neighbor has and seen the do not form complete
groups. If we break the sentence into natural groups, the first break will be
between yes and my neighbor has seen the dog. The second group can be
divided into my neighbor and has seen the dog. My neighbor can be
divided into my and nei ghbor. Sentences in English, then, are formed not
only of words in a specific order, but also of words arranged in hierarchical
groupings, in which words combine to form groups which are in turn parts
of larger groups.
Among these groups my neighbor and the dog seem to be the same
kind of structure since they can substitute for each other:

1. Yes, my neighbor has seen the dog.

2. Yes, the dog has seen my nei ghbor.

10
 the structure of the sentence II

We will use names to label structures so that we can show which ones are
similar. My neighbor and the dog are called noun phrases since a noun is the
chief word in each.f
Almost all fields of study have abbreviations for terms that are often
repeated; this grammar is no exception. Instead of writing sentence many
times in our rules, we use the abbreviation S; similarly, sentence modifier is
abbreviated SM, noun phrase is NP, and verb phrase is VP. Rules in a trans¬
formational grammar are expressed in the following way:

S -> (SM) Nuc

Nuc -> NP + VP

The arrow means ‘‘consists of” or “is to be rewritten as.” These rules say
that in English a sentence consists of a sentence modifier and a nucleus; a
nucleus (Nuc) consists of a noun phrase and a verb phrase. The parentheses
around SM mean that this element is optional; i.e., the sentence may or may
not contain it. Since Nuc is not in parentheses, every sentence must include
a nucleus. Notice that there is no plus mark between parenthesized elements
and adjoining elements. In the second rule the fact that NP and VP are not
in parentheses indicates that both are necessary for every sentence in
English. The rules also indicate the order in which elements must be
selected: the sentence modifier must come first, then the nucleus; in the
nucleus the noun phrase must come first, then the verb phrase.
A sentence modifier is a word or group of words like yes, no, certainly,
naturally, maybe, perhaps, possibly, in fact, to be sure, or obviously. In
Yes, that woman drinks coffee, the SM is yes, and the Nuc is that woman
drinks coffee; in the Nuc, the NP is that woman, and the VP is drinks coffee.
As you may have noticed, noun phrase is the name of the structure that func¬
tions as the complete subject of a sentence; the verb phrase is the structure that
functions as the complete predicate. This distinction between structure and
function is the same one traditional grammarians were making when they
called a word a noun or pronoun by structure but subject or direct object by
function. In the sentence Of course John snores, the SM is of course and
the Nuc is John snores; the NP is John, and the VP is snores. In the
sentence John snores, the optional SM has not been selected. Notice that a
noun phrase may be a single word (John ran) or a group of words (The
little boy ran).
Our use of the word rule is different from that used in traditional
grammar. A rule for a transformationalist is not an explanation of how to

f We are not using the word phrase in the same sense that traditional grammarians used
it.
12 chapter two

punctuate a sentence or how to avoid errors. Rather, it is a direction for

forming a sentence or part of a sentence. The rules in a transformational
grammar will specify which combinations of words are grammatical
sentences.
In addition to rules that generate the sentences of English, we also
have a means of representing the exact choices that are made in the deri¬
vation of specific sentences. This is known as a tree. The sentence Yes,
that man drinks milk is represented as follows:

NP VP

yes that man drinks milk

The sentence That man drinks milk is represented without the optional
element SM:

S
I
Nuc

NP VP

that man drinks milk

Sentences in English are not composed of mere sequences of words; rather,

as we noticed at the beginning of this chapter, they are composed of words
that cluster together, often in complex hierarchies. In the above sentence,
that man drinks milk is one cluster, which in turn is composed of two
subordinate clusters: that man and drinks milk. Notice that the tree shows
this arrangement.
Before proceeding to the noun phrase, you should practice drawing a
few trees:

1. Certainly, I know the answer.

2. He has gone.
3. The fireman fought the fire.
4. Apparently Tom is sick.
5. Unfortunately the fish died.
 the structure of the sentence 13

All rules in a transformational grammar are numbered: PI, P2, P3, etc.
P stands for phrase structure, the name for this section of the grammar. There
will be some skips in the numbers given in the rules at this time, since some
of them will not be needed until later in the chapter. Also, some rules will
be expanded in later chapters of this book. So far we have two rules:

PI: S —> (SM) Nuc

P2: Nuc NP + VP

A noun phrase always contains a nominal (N), which may be a

pronoun (he, she, we, I, they, you, etc.), a name (John, Mrs. Smith, New
York, The Grapes of Wrath, etc.), or a common noun (book, bottle, egg,
table, girl, etc.). In Chapter Four of this book we will discuss ways of
distinguishing these three kinds of nominals. For now they are all classified
as Ns. Some nominals are preceded by determiners (Det), such as the,
a, that, this, these, those, etc.; some nominals may be in the plural (PI).

P7: NP —► (Det) N (PI)

The parentheses around Det and PI mean that these elements are optional.
A later section of the grammar will tell us to read book + PI as books,
egg — PI as eggs, dish T PI as dishes, deer -f PI as deer, child + PI as
children, etc.
The tree contains all of the information from the rewrite rules for
each sentence generated. Here is the representation for Those apples look
green:

The triangle under VP tells us that no further analysis is being given for
look green.
14 chapter two

Before proceeding further, draw trees of the following:

1. An apple lay on the ground.

2. Those pearls look genuine.
3. Surely you can go with us.
4. The ducks are noisy.
5. Certainly those sheep ran fast.

Many of the terms used in transformational grammar are similar in

meaning to those used in traditional grammar: subject, article, demonstrative,
sentence, etc. Some of them, such as determiner and noun phrase, are new. In
addition, certain terms from traditional grammar are used by transfor¬
mationalists but with different meanings. We have already encountered
one of these: verb phrase. To the traditionalist the verb phrase in the sentence
The man must have found the note is must have found; to the transfor¬
mationalist it is must have found the note. This does not reflect any
disagreement between the two schools of thought; they are merely using the
same term to refer to different ideas. There will be several other terms that
are used with meanings different from those found in traditional grammar.
Let us now examine the components of the verb phrase:

P3 : VP —> Aux -f- MV (manner) (place) (time) (reason)

This means that a verb phrase consists of (or “is rewritten as”) an auxiliary
(Aux), a main verb (MV), and optional adverbials of manner (rapidly,
with ease), place (there, at home), time (then, at noon), or reason (because
of the noise). In the sentence The man will drive carefully in town today
because of the ice, the Aux is will, the MV is drive, manner is carefully,
place is in town, time is today, and reason is because of the ice.
Later we will have a great deal more to say about the English auxiliary,
but for the present only one element will be introduced:

P4: Aux —> tense

P5: tense —(Presentj

(past j

These two rules say that every auxiliary contains tense, and that tense is
either present or past. In the sentence The woman sang to me, the tense is
past; in The woman sings to me it is present. In They are at home, the
tense is present; in They were at home it is past. You will notice that tense
is not a separate word, as are the other elements we have been considering.
We list gave as past -f- give (i.e., “the past tense of give”), gives as
 the structure of the sentence 15

present + give, give (as in We give parties often) as present + give.

Every verb will be preceded by a symbol for tense.
Later when we are discussing transformations, we will need to dis¬
tinguish between be and other verbs. A few examples will show the reason
for this. He was my friend has a related negative form He was not my
friend, in which the word not is found after was, a past form of be. The
related negative of He saw my friend does not follow this pattern. Instead
of *He saw not my friend, which would be parallel to He was not my
friend, in Modern English we say He didn’t see my friend, with the addition
of do. Similarly, in questions we find a difference between be and other
verbs. He was my friend has the related form Was he my friend? He saw
my friend does not have as its related form *Saw he my friend? but rather
Did he see my friend?

' be (place
fNP
P6: MV
|ap
v (NP)

This rule says that an MV is to be rewritten as any one of the following

structures:

1. be + NP This book is a text.

2. be + Place Betty was in the car.
3. be -j- AP She was very rude.
4. V John ran.
5. V + NP Bill sold the tickets.

Whenever we speak of verbs, we are excluding forms of be (am, is, are, was,
were, be, been, being). You can probably think of other structures following
verbs, such as Susan seemed happy and We considered Susan lucky. These
will be treated in a later chapter.
AP is an abbreviation for adjective phrase, which consists of an optional
intensifier (Intens) such as very, extremely, rather, etc. and an ad¬
jective (Adj) such as old, happy, green, etc. Here is the rule:

P8: AP —>• (Intens) Adj

If we were writing a complete grammar of English in which we were

trying to describe all details, we would include many more rules. For
16 chapter two

example, we could distinguish among the following kinds of adverbials:

1. The adverb (rapidly, cheerfully, lengthwise).

2. The prepositional phrase (in the yard, at school, over the fence).

3. The uninflected word (here, there, today).

* *
\

4. The noun phrase (as in We walked a mile).

Since this book is intended only as an introduction to transformational

grammar rather than an exhaustive grammar of English, we will use the
terms manner, place, time, and reason for any of these structures. They function
alike in the rules that follow.
In case you are having trouble classifying the words following verbs
as noun phrases, adverbs, etc., a few suggestions may be helpful. Only a
noun phrase may contain a determiner or be in the plural. In the sentence
John turned the wheel, we know that the wheel is a noun phrase because
of the determiner the; also, we would not be altering the structure if we
changed wheel to wheels. Some nouns do not have plural forms: She drank
the milk. There is no such form as *miiks with this meaning: *She drank
the milks. As you will see in Chapter Four, many abstract nouns and
proper nouns do not take determiners or plural endings: John admires
courage. You will learn to recognize these nouns with practice. Except for
a few obvious proper nouns and pronouns, all the noun phrases in the
exercises following this chapter are readily recognizable by the presence (or
possibility) of determiners and plurals.
An adjective phrase may contain an intensifier: They were rather
friendly; They are stubborn. In the first sentence, rather friendly is
recognizable as an adjective phrase because of the intensifier rather. In
the second sentence, stubborn is an adjective phrase because it could be
preceded by an intensifier such as very. Also, friendly and stubborn may
be compared: friendlier and more stubborn.
Some adverbials may also be accompanied by intensihers or be
compared, but not adverbials of place: They are here, but not *They are
very here or *They are more here. Adverbials of place may be replaced
by here or there: They were in the kitchen or They were there. Most
adverbials of place are prepositional phrases, and prepositional phrases
contain noun phrases, such as the kitchen in the above sentence. In classifying
structures, do not be concerned with elements contained within the one you
are examining. To the bank in We went to the bank is a prepositional
phrase used as an adverbial of place; we can substitute there for it. It is
irrelevant that it contains a noun phrase. Another way of stating this is to
 the structure of the sentence 17

say that the bank is a noun phrase, but that to the bank is a prepositional
phrase used as an adverbial of place. |
We can incorporate all of the information contained in our rewrite
rules in trees:

Nuc

NP VP

she

Notice that the tree shows the structure of the sentence. We call each of the
places at which a symbol is written a node and say that a node dominates
anything connected by lines below it. From the tree we can see that John
is an N, since the N node dominates it. Similarly, an is a Det, be an artist
is an MV, present be an artist is a VP, etc. On the other hand, John
present be is not dominated by any single node and is, therefore, not a
structural unit.

f Some grammars would list to the bank as an adverbial of direction. In this text we will
not be distinguishing between place and direction, since they are both replaced in questions
by where.
18 chapter two

exercises

A. Draw trees of the following sentences:

1. The boy ate a hamburger greedily.
2. Bobby is quite intelligent.
3. Of course, the car is in the garage.
4. Naturally those ducks sat there for an hour.
5. No, those women planted the garden yesterday.
B. Without thinking of any particular sentence, begin with rule PI and work
through the rewrite rules until you can no longer rewrite anything. When you
are allowed a choice, select any of the alternatives at random. Then select words
that will fit the structure you have produced.

G. Some traditional grammars point out that many adverbs end in -!y and answer
the question how. To which class of adverbs do these criteria apply? According
to this information, what part of speech should sickly be in She was sickly most
of the time ? If-ly is added to a noun, such as love, friend, or heaven, is the result
an adverb? What part of speech normally results from -ly added to an adjective,
such as sudden, careful, or peaceful? Are adverbs the only words that answer
how?
 chapter three
THE AUXILIARY

The only element in the auxiliary that we have seen so far is tense. We now
need to expand our rewrite rule so that it will include such auxiliaries as
those in the sentences We had been eating and They must have been
looking at us. The sentences in the left column below contain nothing but
tense in the auxiliary; those in the right column have had something added
to them. Analyze these expanded sentences to see what has been added:

1. The bird sings. The bird is singing.

2. The bird sang. The bird was singing.
3. I eat sandwiches. I am eating sandwiches.
4. I ate sandwiches. I was eating sandwiches.

In the sentences on the left, tense is attached to the main verb: si ngs (present)
and sang (past), eat (present) and ate (past). In the sentences on the right,
however, the main verb does not change with variations in tense; it remains
singing or eating, regardless of which tense is selected. Variation in tense
is shown, rather, in the auxiliary, which is a form of be: is singing (present)
and was singing (past), am eating (present) and was eating (past). Along
with this form of be that has been added to the auxiliary there is another
morpheme (a unit that cannot be broken into smaller grammatical units):
the present participial ing. This morpheme is attached to the word that
immediately follows the auxiliary be: is singing, was eating. We can now
make the first of several expansions of the auxiliary:

P4: Aux —► tense (be + ing)

This means that every auxiliary contains tense. The elements be and ing
are optional, but if they are chosen, both must be selected and they must
come in this order, following tense. Ing is attached to the word that follows

19
20 chapter three

it. By ing we are not indicating the pronunciation of this morpheme; we

are simply using it as a symbol for present participle. The tree- for She is
singing looks like this:

Now examine the following pairs of sentences. Again, those on the

left have just tense in the auxiliary, but those on the right have been
expanded:

1. We take medicine. We have taken medicine.

2. We took medicine. We had taken medicine.
3. Ann drinks milk. Ann has drunk milk.
4. Ann drank milk. Ann had drunk milk.
5. I am here. I have been here.
6. I was here. I had been here.
7. He has the answer. He has had the answer.
8. He had the answer. He had had the answer.

Again, you will notice that with the addition of some element in the
auxiliary besides tense, the tense morpheme is no longer attached to the
main verb, but rather to the other auxiliary: have or has (present) versus
had (past). For the time being forget about time in relation to tense; we shall
have more to say about that later. Since the only change in form that can
be made in have taken or has taken is to had taken, we say that have taken
and has taken contain the present-tense morpheme, had taken the past-
tense morpheme. Instead of be + ing, this time we have added a form of
have in the auxiliary, and with it we have added en (the past-participial
morpheme) to the following word (taken, drunk, been, had in the sentences
above). We represent this morpheme with the symbol en regardless of the
actual form of the past participle. En -f- eat is eaten, en -}- hear is heard,
en -f- drink is drunk, en T hit is hit, etc. By en we are not indicating the
pronunciation of any part of the past participial form of any verb; it is
 the auxiliary 21

merely a convenient, short symbol so that we do not have to write past par¬
ticiple of each time we use it. We could now rewrite rule P4 as follows:

Aux —► tense (have -f en)

The only reason we do not is that we have already seen that be + ing may
also be part of the auxiliary.
We know that either be T ing or have -|- en may exist between tense
and the verb. We now need to^see whether they are mutually exclusive or
whether both may be selected for the same auxiliary. Examine the following
sentences:

1. They had been singing songs.

2. They have been singing songs.
3. She has been smiling at me.
4. She had been smiling at me.
5. *She was having smiled at me.
6. *They are having sung songs.

It is obvious that both be -f- ing and have -j- en may exist in the same
auxiliary. When both are present, have + en comes first. Tense is attached
to have. We can expand rule P4 like this:

Aux —> tense (have + en) (be + ing)

This means that every auxiliary must contain tense. An auxiliary may
contain both the other pairs of elements, one of them, or neither. If have is
selected, then the next word must contain the morpheme en; if be is selected,
the next word must contain ing. Whichever elements are selected, they must
come in the order in which they are listed in the rewrite rule. Tense is
attached to the element that immediately follows it. Here is a tree for a
structure in which both have and be have been selected as auxiliaries:

he past have en be ing smile

22 chapter three

The past form of have is had, the en form of be is been, and the ing form of
smile is smiling; the sentence is He had been smiling.
We need to make one final addition to the auxiliary. Examine the
following sentences:

1. I can give the answer now.

2. I could give the answer yesterday.
3. They will stop soon.
4. They would stop soon.
5. She may be joking.
6. She might be joking.
7. We shall be leaving.
8. We should have been leaving.
9. You must be going.

These sentences have their auxiliaries expanded by the addition of the

following words: can, could, will, would, may, might, shall, should, must.
You should recognize the other elements of the auxiliary: have + en and
be 4~ ing. We call these new words modals and use the abbreviation M
for them. Notice that modals precede other auxiliaries and that the word
following a modal is in the uninflected (or infinitive) form.
In Modern English, tense with modals presents a problem. In earlier
English, can, will, may, and shall were present-tense forms and could,
would, might, and should (notice the d or t) past-tense forms. In a few
structures, such as I can do it now and I could do it yesterday, the tense
distinction is still evident, although the idea of time is more firmly expressed
by the adverbials now and yesterday than by the verb. But such obvious
distinctions are rare today. Most sentences with modals, in fact, seemingly
do not contain tense in referring to time, and we could probably justify a
rule saying that an auxiliary contains tense or a modal, giving our rewrite
rule for the auxiliary as:

AuX ^ { tense } (h°''e + en) (be + lng)

But such a rule ignores the tense relationship that does exist between such
pairs of sentences as I can do it now and I could do it yesterday. We will,
therefore, continue to use a rule that has become accepted:

P4: Aux —► tense (M) (have -f en) (be -f- ing)

Regardless of meaning, classify the following modals as present: may, can,

will, shall, must; these as past: might, could, would, should.
 the auxiliary 23

For most verbs all combinations of the auxiliary are possible. There
are some nonaction verbs, however, that do not exist with be -f- ing: seem,
hear, know, hate, understand, etc. (*She was seeming frightened). The
copula be as in He was a nuisance does not take the auxiliary be -f- ing,
although there is another verb be (He was being a nuisance) which expresses
action. This feature will be treated as a peculiarity of certain verbs and will
be included in their dictionary entries. We do not need to change rule P4
for these exceptions.
Here is a tree for the •'sentence He
could have been running. It contains the
maximum expansion of the auxiliary.
You have probably wondered about
our treatment of tense in some sentences.
We have said that He has seen us con¬
tains present tense, but He had seen us
contains past tense. It is obvious that both
sentences describe action that occurred
during past time. Also we have called the
tense in He will go present and He would go past, although they both clearly
indicate future time. In fact, we have not even mentioned a future tense.
Traditional grammar lists three basic tenses. It defines present tense
as an expression of action occurring at the present time, past tense as action
occurring during the past, and future tense as action occurring during
future time. These tenses are represented by the following forms of the
verb: I see (present), I saw (past), I shall see (future). In addition, there are
three perfect tenses, indicating completed action and formed by the addition
of have en to the auxiliary. Many traditional grammars stop with these
six tenses; but others add three progressive tenses, which indicate continuing
action and are formed by the addition of be Y mg to the auxiliary. Tradi¬
tional grammars generally assume that there is a strict correspondence
between time and the form of the verb in English.
Such a correspondence does not exist in actual usage. Present time
may be represented by the present-tense form of the verb as in Now I
understand or Here comes Sally, but it is usually represented by the present
progressive form: He is sleeping; They are leaving now; I am eating. For
politeness we sometimes use “future tense” for a present action, as a check¬
out clerk may say, “That will be $6.50, please.” The normal meaning of
the present-tense form is not present time, but rather action that is habitual
or universally true (i.e., past, present, and future time): Tim is a policeman;
Water freezes at thirty-two degrees; I eat lunch at one o’clock every day.
For many verbs it is impossible to state an action occurring specifically in
the present time with the simple present tense: *1 eat the cake now (cf. I
eat cake often and I am eating the cake now).
24 chapter three

Past time is normally expressed by past tense: John ran to town; He

had already finished when I came in; In those days I ate a big dinner. But
sometimes present tense is used to express past action, as I see that he has
been defeated, or When Benjy hears the golfer cry, “Here, caddie,” he
begins crying, or Europe enters a renaissance during the twelfth century
(the historical present). The present perfect tense reflects past time in spite
of its name: He has gone; We have finished.
Future time is as often expressed by other means as it is by will and
shall ; in fact, it may be expressed by any modal or by any tense except past,
'with adverbials such as tomorrow indicating the time.

1. We leave for New York tomorrow. (Present)

2. If she stays, we must leave. (Present)
3. Ask her if she will stay. (Future)
4. Ask her if she is going to stay. (Present progressive)
5. He has his chance tonight. (Present)
6. He is leaving soon. (Present progressive)
7. When she comes, we’ll leave. (Present; Future)
8. He is about to go. (Present)
9. She is to accompany him. (Present)
10. What would you do if you had a flat tonight? (Past)

No one denies the existence of past, present, and future time. We do

say that the forms of verbs and auxiliaries in English are not so consistently
related to these three times as are the forms of the Latin verb. Time in
English is often expressed by other means than the tense of the verb. By
form there are only two tenses in English: present and past. Remember that
by tense we mean the form of the first auxiliary that follows the symbol
tense. If there is no such auxiliary, then tense is shown by the form of the
main verb. Tense is a concept of form, not meaning.
Elaborate rules for the use of shall, will, should, and would have been
devised by earlier textbook writers. The rules first appeared in the seven¬
teenth century and have been perpetuated with embellishments to the present
day. Many writers of high-school English texts and of college handbooks
are notorious for merely copying rules that have become traditional and
never questioning their validity. Most of these texts say that shall and should
are normally used for first person, will and would for second and third
persons. To show determination, one is to reverse this order, using will and
would in first person, shall and should in second and third. Then there are
many other rules that vary from text to text, such as the one requiring shall
in the question Shall you be here tomorrow? since the anticipated answer
is I shall.
 the auxiliary 25

These rules offer a good example of the utter futility of trying to

legislate English usage. In spite of assiduous drill by schoolmarms for two
centuries, the rules have never been followed by many educated people.
Most of the better books on usage today have adopted the more sensible
approach of describing what educated people actually say, rather than
trying to prescribe rules that seem “logical” or “in good taste” to someone
while he is thinking about language in his study. It is true that in spoken
British English shall and shan't do occur often with first person pronouns.
In spoken American English, however, shan't is almost never heard, and
shall appears formal, emphatic, or even affected.
No doubt the main reason for the infrequent use of shall in spoken
American English is that this modal did not become widely used in London
English until after most of the English speaking settlers of what is now the
United States had emigrated. The later rise in popularity of shall in England
did not spread to this country. Also, in American English both shall and
will in speech are contracted in sentences that are not given particular
emphasis. I'll go is the normal spoken form, with I will go or I shall go
showing either formality or emphasis. In fact, shall because of its infrequent
occurrence stands out in spoken American English and gives emphasis or
determination for all persons, as McArthur’s “I shall return” clearly
indicates. The only unemphatic use of shall in spoken American English is
in questions like Where shall we sit? and Shall we dance? But even these
are now being replaced by such forms as Where would you like to sit? and
Would you like to dance? Except for the emphatic form, shall is rarely
found in spoken American English.
In formal American English, usage is divided among educated speakers
between shall and will for first person. Although I shall and we shall occur
frequently in formal written English, it is not at all unusual to find I will and
we will in articles in professional journals and in formal correspondence from
English professors in leading American universities and from editors of
major publishing firms, particularly when rapport with the reader is
especially desirable.
Should, like shall, is used differently in British and American English.
In unpretentious American speech, should always means “ought to.” To
most Americans, I should be glad to have you visit me either sounds
affected or seems to indicate reluctance. In formal writing, the same
practice is observed. Would with first person as in I would be glad to hear
from you is the usual form.

exercises

A. Draw trees of the following sentences:

1. Mabel might be upstairs.
2. Evidently he had read the report reluctantly.
26 chapter three

3. That man must have lacked courage.

4. The members would have chosen the leader then.
5. Harold might have been sitting in his room.
6. He will be a minister next year.
7. Yes, Murphy has been very quiet.
8. The firemen will be having trouble soon.
B. After each of the following formulas there are three sentences that may or may
not have the structure described by the formula. ‘Jf a sentence follows the
notation given, write correct beside it. If it does not follow this notation, write
the formula that it does follow.
1. NP + present + M + V + NP
a. The student should use the library.
b. The car will hit the child.
c. He can sing the song.
2. NP + past + M T have + en + be + place
a. The cow has been in the field.
b. They should have been eating.
c. People could have been running in the rain.
3. NP + present + have + en + V + NP
a. We will have caught him.
b. They have chosen John.
c. We are electing him.
4. NP + past + M + be + NP
a. He is the owner.
b. Everybody has gone home.
c. They would have been conservatives.
5. NP + present + M + have + en T be + ing + V + place
a. John is in the barn.
b. Jack will have been running in the house.
c. She must have been ill.

C. Write the formula for each of the sentences below, make the addition Or deletion
that is called for, and write the resulting sentence. Example:
Sally was playing. NP T past + be + ing + play
Add have + en: NP + past + have + en -f be + ing + play
Sally had been playing.
1. I ate then. (Add be + ing)
2. They drank the wine. (Add will)
3. We had gone to the window. (Delete have + en)
4. You could have listened to me. (Add be + ing)
5. They might be disappointed. (Delete may)
6. Sam was telling a lie. (Add have + en)
7. You could have been repairing the clock. (Delete be 4- ing)
8. I was going. (Add shall)
9. I could see him well. (Delete can)
10. The waitress must be laughing at us. (Add have + en)
 chapter four
LEXICAL FEATURES

DIVISIONS OF A
TRANSFORMATIONAL GRAMMAR

In the last two chapters we developed the phrase-structure rules for English.
These rules begin with a single symbol S and progress through ordered
expansions until no further expansions are possible. The rules are repeated
below:

PI: S -> (SM) Nuc

P2: Nuc —► NP + VP

P3: VP —> Aux -f MV (manner) (place) (time) (reason)

P4: Aux —>■ tense (M) (have + en) (be + ing)

P5: Tense —*■ ( Present J

[ past J

\
NP
be AP
P6: MV
Place
1,V (NP) /

P7: NP -> (Det) N (PI)

P8: AP —> (Intens) Adj

28 chapter four

By starting with rule PI and progressing in order through the rules,

selecting or rejecting optional elements, we may expand S in the following
way:
S

PI Nuc

P2 NP + VP * %
V

P3 NP + Aux + MV

P4 NP + tense -j- be -j- ing + MV

P5 NP + present + be + ing -f- MV

P6 NP -j- present + be + ing + V + NP

P7 Det -J- N -j- present -j- be -)- ing -j- V + NP

P7 Det -fi N + present + be + ing -f- V + Det -f- N

P8 Does not apply

There are no longer any symbols that can be rewritten since nothing in the
final line occurs to the left in any of our rewrite rules. Symbols such as
Det, N, or present that cannot be rewritten by the phrase-structure rules
are called P-terminal symbols. A sequence, or string, of these symbols is called
a P-terminal string. A P-terminal string such as

Det -f- N + present + be + ing + V + Det + N

describes the structure that underlies a grammatical sentence of English,

but it is not itself a sentence. It is merely a string of elements, some of which
may be out of order and none of which are in their phonetic shape. The
phrase structure also tells how the elements in the P-terminal string relate
to each other; in other words, it specifies a structure, as exemplified by this
tree:
S

Nuc

NP VP

Det N Aux MV

/ \
Det N present be ing V Det N
 lexical features 29

Included in the phrase-structure section of the grammar is a lexicon

or dictionary, from which we replace such elements as N and V with words:

Det — N + present + be + ing -j- V + Det + N

The + boy + present — be + ing + read + a -j- book

These, of course, are not the only words we could have chosen for this
structure; the following choices are among the many others that could have
been made:
✓

1. A secretary is mailing the letter.

2. The painter is painting the house.

We next proceed to the semantic section of the grammar to provide a

meaning for the structure. The lexicon will have given meanings for the
individual words. The semantic section will tell us how these meanings are
combined to provide a meaning for the sentence as a whole. Since very
little is known about how the semantic component works, it will not be
discussed further in this book.
The phrase-structure rules do not generate finished sentences of
English. In fact, they provide only a small number of elementary structures
that underlie the sentences of English. To rearrange, delete, or add
structures, we need transformational rules. For example, we can provide a
rule for the structure:

the boy present be ing read a book

that rearranges it as follows:

present be the boy ing read a book

This is the order for the question Is the boy reading a book? The rearranged
structure is illustrated by this tree
S Such transformational rules will be dis¬
cussed in detail in the next two parts of
Nuc
this book.
Even with the application of
transformational rules, we still do not
have a grammatical sentence of English.
Another set of rules is needed:
phonological rules that tell us how to
pronounce the string that we have
generated. These rules will be discussed
in Part Four, and they will tell us how
combinations of words are pronounced;
for example, present -f- eat is eats,
30 chapter four

past -f- drop is dropped, etc. (Notice that inflectional morphemes, such
as past, en, and ing, remain before the verbs since they are merely symbols,
not phonetic representations.) After we apply these rules and those for the
addition of stress and pitch, we have a sentence: Is the boy reading a book?
We are describing rules that will enable us to produce the sentences of
English. We call this collection of rules a grammar. The grammar is organized
into three sections or components:
V

1. The syntactic component contains the phrase-structure and trans¬

formational rules and provides the structure of the sentence.
2. The semantic component operates on the P-terminal string after entries
from the lexicon have been added and gives the sentence its meaning.
3. The phonological component operates on the sentence after all trans¬
formations have been applied and gives the sentence its final form.

If we have enough rules in each of these divisions, we can generate all

of the sentences of English, including all potential sentences that have never
been spoken before. But the ideal grammar not only produces all of the
grammatical sentences of a language; it produces no ungrammatical ones.
The grammar as we have stated it so far would produce such unacceptable
structures as la, 2a, 3a, and 4a.

la. *The smoke vanished the table,

b. The smoke vanished.
2a. *An event occurred the man.
b. An event occurred.
3a. *The man brought,
b. The man brought a book.
4a. *He proved,
b. He proved his point.

Rule P6 as we have stated it rewrites MV as be plus one of several structures

or as V plus an optional NP. Sentences la and 2a are ungrammatical
because this optional NP has been selected; 3a and 4a are ungrammatical
because it has not been selected.

TRANSITIVE AND
INTRANSITIVE RESTRICTIONS

The lexical entry for each verb will have to state whether it permits a
following noun phrase or not. Vanish and occur will be listed in the lexicon
with the feature [ —_NP ], and bring and prove will be listed with
[ -j-_NP]. The underline indicates the position filled by the verb,
 lexical features 31

before an NP. The plus indicates that the verb in question can fill the
position; the minus indicates that it cannot. Some other verbs with the
feature [ T_NP ] are shoot, wax, and tear, as illustrated in the following
sentences:

1. He shot a rabbit.
2. I waxed the car.
3. Someone tore the page. „

Some verbs with the feature [ -f-_NP ] at times have indefinite

noun phrases deleted:

1. Larry ate (something).

2. The students wrote (something).
3. He drove (something).

Other verbs like these are watch, steal, and read. Whether these verbs have
a noun phrase following them or not, there is no alteration in meaning.
These verbs still have the feature [ -f-_NP ] even when the noun phrase
is not present. The feature merely states that a noun phrase is permitted
after the verb with the meaning given.
One word of caution is in order. The classification of a verb depends
upon its meaning in the sentence under consideration. There is a difference
in meaning in He ran fast and He ran the store. We set up run in our
lexicon as two separate entries, or as two separate words. The entry with
the meaning as in He ran the store has the feature [ +_NP ]; the entry
with the meaning He ran fast does not have this feature, but rather
( —_NP]. This distinction may become complicated for verbs whose
meanings are metaphorically extended. Notice fly in the following sen¬
tences :

1. The bird flew out of the room.

2. The angry woman flew out of the room in a rage.
3. Last week I flew in an airplane.
4. Last week I flew an airplane.
5. The dust flew everywhere.
6. He flew off the handle.

We will list fly in our lexicon six different times, one for each meaning.
Similar metaphoric extensions can be seen in rise (The sun rose; The boy
rose; Prices rose) and die (The man died; The car died).
32 chapter four

Verbs with the feature [ —_NP ] may be followed by nothing or

by optional adverbials, as in the following sentences:

1. The nurse was sleeping. The nurse was sleeping soundly.

2. The boy fell. The boy fell to the ground.
3. She hurried. She hurried because of the rain.
* %

Of course, as we said in the preceding paragraph, different lexical entries

may have the same form. Hence, hurry in She hurried is not the same word
as hurry in She hurried us off to work. Some other examples of verbs that
do not permit following noun phrases are vanish, sweat, pause, rise, and
occur. These verbs are traditionally called intransitive verbs, as opposed to
transitive verbs, which do take following noun phrases. Many intransitive
verbs can on occasion be followed by a noun that is similar in form to the
verb: He slept a deep sleep; He dreamed a dream (i.e., nouns that tradi¬
tional grammar called cognate accusatives). Since these are apparently the
only nouns that may follow these verbs, they do not affect the classification.
A complete grammar would probably introduce them by transformational
rules.
Some intransitive verbs must be followed by adverbials if the sentence
is to be grammatical:

1. We lay there. *We lay.

2. He lurked outside. *He lurked.

Unlike the simple intransitive verbs, these must have an adverbial of place
following them. Other examples are lie (recline), step, and glance. These
verbs have the feature [ -f_place]. The adverbial is not optional.
Sometimes to avoid repetition we delete elements that are otherwise essential.
*We sneaked, for example, is ungrammatical by itself. It may occur,
however, in the context How did you get past the guard? We sneaked.
Underlying We sneaked is the complete sentence We sneaked past the
guard. Past the guard has been deleted to avoid repetition. This kind of
deletion does not affect our classification.
There are also transitive verbs that must have adverbials of place
following the noun phrase:

1. He handed the paper to me. *He handed the paper.

2. She set the book there. *She set the book.
3 The doctor laid it there. *The doctor laid it.
 lexical features 33

These verbs have the feature [ +_: NP place ], meaning that an adverbial
of place must follow the NP.
With these features—[ -f-_ NP ], [ —_ NP ], [ +_ place ],
and [ +_NP place ]—we can account for the native speaker’s intuition
that sentences such as *We persuaded and *Someone shivered the man
are ungrammatical.

SUBJECT-VERB RESTRICTIONS

Earlier we said that the phrase-structure rules provide P-terminal

symbols such as N and V and that we select items from the lexicon that are
designated by these symbols. If we select just any N or V, we could produce
the following sentences:

1. The man drove a car.

2. An eraser fell to the floor.
3. *A door sneezed.
4. *He surprised my curiosity.

As the last two examples show, we may not select just any noun or verb, but
we must impose certain restrictions.
Notice the following sentences:

1. *The coffee prayed.

2. *Those ants talked to me.
3. *The umbrella coughed.

These are all unacceptable because the subject noun phrases are nonhuman
and the verbs require human subjects like the following:

1. The woman prayed.

2. Those children talked to me.
3. My uncle coughed.

It is, therefore, relevant to our grammar whether a noun names a human

or not. In the lexicon to indicate that a noun represents a human, we will
34 chapter four

assign the feature [ + human ]; all other nouns will be [ — human ]. Thus,
woman, bo/, and aunt are [ + human], and cat, ivy, and door are
[ — human]. Verbs that do not permit nonhuman subjects have the
feature [ — [ — human ]_]. The underline indicates the position of the
verb, after a noun with the feature [ — human ]. Verbs such as pray, talk,
faint, worship, complain, praise, acknowledge, and thank will be listed
with this feature to limit them to [ + human ] subjects. Other verbs will
not be marked for this feature.
We can thus explain why *The door fainted and *The dog praised me
are ungrammatical. In children’s stories or other imaginative literature, of
course, door and dog may become human. Some people will argue that
all of the verbs in the last paragraph may have animate nonhuman subjects;
normally they do not make this claim for all animals, but only for pets or
certain other animals of which they are particularly fond. Rather than
extend the features of the verb for these people, it seems better to say that
the distinctions human and nonhuman have been erased in certain cases
or perhaps that dog, horse, etc., sometimes have the feature [ + human ]
(when they are the subjects of verbs such as worship and acknowledge).
The features we have given will account for the grammaticalness of
sentences la, 2a, and 3a and for the ungrammaticalness of lb, 2b, and 3b:

la. The man bit me.

b. *The street bit me.
2a. A child ran rapidly into the street.
b. *A rock ran rapidly into the street.
3a. The cook tasted the soup,
b. *The chair tasted the soup.

However, if we restricted such verbs as bite, run, and taste to human

subjects, we would not be able to produce the following sentences:

lc. The horse bit me.

2c. A dog ran rapidly into the street.
3c. The cat tasted the soup.

Since these are grammatical sentences, limiting these verbs to human subjects
would be too strong a restriction. Animals as well as humans can bite, run, or
taste something. We need, therefore, another classification of nouns that
includes humans and animals. Dog, horse, girl, man have the feature
[ + animate ]; rose, rug, and rock have [ — animate ]. Verbs that do not
permit inanimate subjects have the feature [ — [ — animate ]_].
 lexical features 35

The need for a final restriction between subjects and verbs can be seen
in the following sentences:

la. His coat lay on the bed.

b. *His curiosity lay on the bed.
2a. The cake tasted good,
b. *Pettiness tasted good.
3a. An accident occurred. „
b. *A snake occurred.
4a. Six hours elapsed,
b. *Six dogs elapsed.

Sentences lb and 2b have abstract subjects with verbs that require concrete
subjects; 3b and 4b have concrete subjects with verbs that require abstract
subjects. Generally speaking, concrete nouns name objects that can be
touched or seen (cat, apple, woman, etc.). Abstract nouns do not (hap¬
piness, honesty, courage, etc.). Abstract nouns are not physically located in
space. Concrete nouns have the feature [ + concrete ]; abstract nouns have
[ — concrete ]. A verb that does not permit abstract subjects has the feature
— [ — concrete]_] (sit, rise, lie, etc.); one that does not permit
concrete subjects has the feature [ — [ + concrete]_] (occur, befall,
elapse, etc.). A verb such as amaze, which may occur with either concrete
or abstract subjects, will not be marked for this feature.
The four features [ — [ — human ]_], [ — [ — animate ] _],
[ — [ — concrete ] _], and [ — [ + concrete ]_] will be given in
the lexical entries of only those verbs which are restricted as to the kinds of
subjects they may follow. Other verbs, such as seem and appear, will be
listed without any of these restrictions since they apparently occur freely
with concrete or abstract, animate or inanimate, human or nonhuman
subjects.
The verb features we have given are illustrative rather than exhaustive.
For example, we could have a classification for verbs that require animate
objects (surprise, astonish, terrify, frighten, etc.). Or we might group
those verbs together that require living subjects: humans, animals, and
plants (grow, die, thrive, etc.). Some verbs take only abstract objects
(guess, pretend, announce, etc.). A complete classification of this nature
would go far beyond the scope of this book.
Some verbs are even more specific: evaporate must have a subject
that is a liquid; read and write must have objects pertaining to writing.
Other verbs have equally severe limitations on their subjects (disperse,
dissolve, corrode, etc.) or their objects (sing, whistle, say, etc.). A complete
lexicon would describe these idiosyncracies.
36 chapter four

Here are some verbs with their syntactic features as they would be
given in a lexicon:

admire eat
[ + NP] [+ NP]
[-[-- human ] ] — [ — animate ] ]
<

seem occur
[ - NP] [--NP]
[ — [ + concrete ] ]

DETERMINER AND NOUN RESTRICTIONS

We have now classified nouns as human or nonhuman, as animate or

inanimate, and as concrete or abstract. These features are needed in
accounting for restrictions that hold between nouns and verbs. There are
also restrictions on which determiners a noun may take:

la. *1 saw bug on floor,

b. I saw bugs on the floor.
2a. *The honesty is an admirable trait.
b. Honesty is an admirable trait.
c. *Honesty is admirable trait.
3a. *The William entered the room,
b. William entered the room.

To account for the differences between the grammatical and the ungram¬
matical sentences above, we must discuss additional noun features.
Nouns are either common [ -j- common ] or proper [ — common ].
The traditional definition of a proper noun as the name of a particular
person, place, or thing and of a common noun as any one of a class works in
many cases: Joe vs. boy, Omaha vs. city, etc. On the other hand, the names
of seasons of the year, which are traditionally considered common nouns
and are not capitalized, are as specific as names of the months, which are
proper nouns and are capitalized. Also, the devil seems to be as specific as
Satan. Normally proper nouns do not take plurals or articles, the word the
in such names as The Hague and the Amazon being taken as part of the
name and not a determiner. But even here there are problems. No one
would question calling Mr. Smith and Mrs. Smith proper nouns, but when
we speak of them together as the Smiths, we have added both an article and
a plural morpheme. The following sentences also present problems: The
 lexical features 37

Americans landed three Sundays ago; A Canadian has two Chevrolets; There
are two Jims in my room; He is not the John Smith who lives next door to
me. That there is some real system underlying the distinction common and
proper can be seen in a sentence such as Blamp opened the door. The
native speaker recognizes Blamp as the name of a person (or pet), although
it is a new word to him. Our grammar should be able to draw similar
conclusions. No doubt much of our confusion is caused by a too close
association of the noun designation proper with an initial capital letter.
The differences between common and proper nouns is a subject that needs
further research.
Some nouns name objects that can be counted, whereas others do not.
The sentence I read a book can be altered without changing the structure to
I read two books or three books or four books. Book has both a singular
and a plural form, and various numbers can be placed in front of it. We
call book a count noun and say that it has the feature [ -f- count ]. In the
sentence I drank a glass of water, the word water cannot be made plural,
even if we alter the number of other nouns and verbs in the sentence, nor
can we add numbers directly in front of the word. Unlike glass in this
sentence, water is a noncount noun and has the feature [ — count]. The
following sentences provide further examples of this feature:

1. The horse bit me. [ + count]

2. Those doves flew away. + count]
3. The horse ate some oats. — count]
4. The ground needs moisture. — count]

It is possible for a noun to be count in one sentence but noncount in

another. In the sentence The ground needs water, water exhibits the
feature [ — count]. But in He has seen the waters of the world, it
exhibits [ + count ]. The word water does not mean exactly the same thing
in the two sentences. Notice how the following sentences vary this feature:

1. He ate two eggs. [ + count ]

2. He got egg on his shirt. [ — count ]
3. He threw me a stone. [ T count ]
4. The house was made of stone. [ — count ]
5. I spilled a Coke. [ + count]
6. I spilled a glass of Coke. [ — count ]

Noncount nouns, such as those in the even-numbered sentences above,

38 chapter four

often seem to mean substance or material (they are often called mass nouns),
whereas the count nouns mean the individual items. A dictionary usually
lists these meanings separately. We will treat the count noun stone in
sentence three as a different lexical entry from the noncount stone in
sentence four.
Sometimes a noun is noncount although we could count the items it
names. Furniture, for example, in He owns much furniture, is noncount,
since we cannot say one furniture, two furnitures, three furnitures, etc.
Yet we can count the pieces of furniture. Similarly, sand in He got sand in
his shoe is a noncount noun, although we can count grains of sand.
Some abstract nouns are noncount and do not permit determiners or
plural morphemes: *The honesty is good; *They have courages. Other
abstract nouns, such as idea and trait, are count nouns and take determiners
and plural morphemes freely.
The features [ + count ] and [ — count ] determine which determiners
may accompany a noun. If the noun is [ — count], a may not be used:
*A happiness was felt; *A furniture was in the room. If the [ — count ]
noun is [ — concrete], normally the, this, and that may not be used:
* I felt the sadness; * He was full of that enthusiasm. Nouns that are [ — count]
and [ + concrete] take the freely: The oil is leaking on the furniture.
Count singular nouns may take a or the; they may not exist without any
determiner: A man sat in the room; *Man sat in room. Count plural nouns
may be preceded by the or by no determiner; they may not be preceded by
a: The students came in cars; *A students came in a cars.
Features of a noun can be represented in a matrix, in which we have
rows of positively or negatively specified features. Below are the features
for the boldfaced nouns in the following sentences: Mary loves truth. The
woman owns a cat.

woman Mary cat truth

N + + + +
common + —
+ +
count + —
+ —

concrete + + + —

animate 1
“T + + —

human ‘ + + — —

A matrix such as this contains much redundant inforamtion. When we

classify woman as count, concrete, animate, and human, we are giving
much obvious information, since all human nouns are necessarily count,
concrete, and animate as well. By omitting all redundant information, we
 lexical features 39

can turn our matrices into more useful forms:

woman Mary cat truth

N + + + +
common + — + +
count
concrete —
animate +
human + . +

A complete grammar would contain rules such as one that automatically

applies the features [ + animate ] and [ -f- concrete ] on all nouns with the
feature [ + human]. All features that are predictable by rules are left
unspecified in the matrix.

THE LEXICON

The lexical features that we have presented in this chapter will be

included in the lexicon, or dictionary. Although there are still many
unanswered questions regarding the form of an ideal lexicon, the kind of
information it should contain is clear. First, there will be a citation form of
the word, expressed in phonetic notation. Second, the idiosyncratic proper¬
ties of the word will be given; these will include the features we have presented
and all irregularities (the plural of foot is feet, the past of sing is sang, etc.).
Third, the meanings or definition of the word will be given. The syntactic
component of our grammar will make use of the features of the word, and
the phonological component will make use of the phonetic citation form
and of inflectional irregularities. The semantic component will use the
definitions and the features.
Lexical entries will look something like the following:

1. /maus/ ' + N a small rodent

[ — rule M5 ] T common
/maus/ -j- pi —>- /mais/ -f- animate
— human

2. /maus/ ~ + N a timid woman

[ — rule M5 ] T common
/maus / -f- pi —► I mais/ + human

3. /maus/ " + N a black eye

[ — rule M5 ] + common
/maus/ + pi —> /mais/ + count
+ concrete
— animate
40 chapter four

The direction [ — rule M5 ] tells us not to apply the rule for regular noun
plurals, which would yield * mouses. Current dictionaries seem to indicate
the plurals for entries two and three as we have given them. This infor¬
mation may be inaccurate; for these uses there may be no plural forms, or
the plural may be mouses.

exercises

A. Use features to explain why the following sentences are ungrammatical:

1. *The perseverance is a virtue.

2. *He has read book.
3. *A birds flew into the room.
4. *Despair dropped to the floor.

5. *The eagle prayed for an hour.

6. *They handed the book.
7. *We vanished the spot.
8. *The tree coughed loudly.
9. *The bread dripped.
10. *My boss elapsed.

B. Give the features of the boldfaced words:

1. The monkey chewed the food slowly.
2. A student coughed loudly.
3. He glanced at the water.
4. Bob handed a ruler to me.
5. The accident occurred yesterday.

C. Examine the following sentences and decide why some of them are ungrammat¬
ical. What generalizations can you make about much, many, fewer, and less?
1. Much energy was spent on this project.

2. Many apples were in the basket.

3. *Much children were in the room.
4. *Many dandruff was in his hair.
5. *He ate less apples than I did.
6. He has fewer friends than I have.
7. She has less confidence than Jane has.
8. *He has fewer poise than I.
part two

TRANSFORMATIONS I
 chapter five
THE NEGATIVE TRANSFORMATION
✓

The phrase-structure rules can produce the structures underlying such

sentences as Those boys might have been swimming in the lake and The
manager wrote a letter. They cannot produce such structures as the follow¬
ing:

1. The manager didn’t write a letter.

2. Did the manager write a letter?
3. Who wrote a letter?
4. What did the manager write ?
5. A letter was written by the manager.
6. Because the manager wrote a letter . . .
7. The letter that was written by the manager . . .
8. The letter written by the manager . . .
9. For the manager to write a letter . . .
10. The manager’s having written a letter . . .

All of these structures seem to be related in some way to The manager

wrote a letter. The same relationships are found in all of them: the manager
is the one who performed the act of writing, and a letter is the result of this
action. In spite of dfferences in form, there is a similarity in meaning in all
the structures. Transformational rules are used to produce these changes in
form.
Earlier we listed several sentence modifiers: yes, no, etc. To these we
add not, which distinguishes a sentence such as John could sing well from
the negative sentence John could not sing well. By selecting the SM not,
we can derive a structure as shown on page 44. This gives not John
past can sing well, which is not grammatical. It would be grammatical if
we changed the word order to John past can not sing well (John could not

43
44 chapter five

sing well). We now need to introduce

two new terms: deep structure and surface
structure. A structure generated only by
phrase-structure and lexical rules, such
as not John past can sing well, is a deep
structure. A deep structure that has
been transformed into a grammatical
English sentence, such as John could not
sing well, is called a surface structure.
All grammatical English sentences are
surface structures; underlying each one
is a deep structure. In later chapters
we will modify this statement somewhat. Both deep and surface structures
are more abstract than we are presenting them here, since phonological
rules have not yet been applied. However, at this stage it is easier to follow
the transformational rules if we use a form such as not John could sing well
than the abstraction that is the actual deep structure.
All native speakers of English are capable of producing an infinite
number of surface structures, or grammatical English sentences. Our
purpose is not to teach you how to make negative sentences; any normal
five-year old can do that perfectly. Our main purpose in studying these
structures is to learn to make accurate observations about how English
operates. By negative we are referring to sentence negation, not word
negation. That is not probable has the SM not. That is improbable does
not have this SM and is, therefore, not part of this immediate study, although
the processes we are employing in this chapter could easily be extended to
account for such negative forms as improbable.
By selecting the optional SM not, we can generate a number of deep
structures like those on the left below:

1. not Jerry could hear me Jerry could not hear me.

2. not Bill has received it Bill has not received it.
3. not they are going with us They are not going with us.
4. not they have been doing it They have not been doing it.

We need to formulate a rule to transform the deep structures on the left to

the surface structures on the right. At first glance you might say something
like “Move not to the position after the auxiliary.” In the first sentence,
not Jerry could hear me, this would work, but the third sentence would
become *They are ing not go with us (remember that the ing which
accompanies the auxiliary be is part of the auxiliary). We must refine our
observations. In the surface structure, the negative particle not follows part
 the negative transformation 45

of the auxiliary, but not all of it. In fact, it follows only the first auxiliary
after tense. In not Jerry could hear me, could is a case of past + can. Can
is the first occurring auxiliary; therefore, not follows it in the surface
structure. We use the abbreviation Aux1 for the first auxiliary that comes
after tense. In the following sentences Aux1 has been marked:

1. they past can be ing go

2. we present have en eat

3. those dress PI present must have en be red

So long as the auxiliary contains something besides tense (a modal, have, or

be), the first element following tense is Aux1. Our transformational rule
for the correct placement of not should read something like this: “Move
not to the position following the first auxiliary after tense.” Since we will
have a number of transformational rules, it will be advantageous to write
them in a conventional abbreviated form. The rule can be stated as follows:

not + X + tense + Aux1 + Y => X + tense -j- Aux1 + not + Y

The information on the left of the arrow describes the structure to which the
rule is applicable: one with the SM not and an Aux1. If either of these
conditions is not met, the rule does not apply. The information on the right
of the arrow describes the structure after the change has been made. The
double arrow means that this is a transformational rule rather than a
phrase-structure rule. Whereas phrase-structure rules merely expand
elements, such as Nuc into NP and VP, transformational rules rearrange,
delete, add, or substitute elements, thereby altering the underlying structure
of the sentence. The symbol X stands for anything coming between not and
tense, such as another sentence modifier or a noun phrase. Since the rule
operates the same way regardless of what follows not, we can simplify our
rule by using the symbol Xfor any structure coming between not and tense.
Similarly, Y stands for anything following Aux1. This may be other aux¬
iliaries, a verb, and anything that follows a verb. Since the same process
applies regardless of what follows Aux1, we can improve the rule by using
the symbol Y for this.
For the deep structure not they present can hear you, we can illustrate
the rule in the following way:

not X tense Aux1 Y > X tense Aux1 not Y

not they present can hear you they present can not hear you

This gives They can not hear you, after the phonological rules have been
46 chapter five

applied. This process can be illustrated with trees. Here is the deep
structure:

N Aux MV

not they present can hear you

The negative transformation rearranges the tree to provide the following

surface structure:

Nuc

they present can not hear you

After the application of phonological rules, we have the sentence They can
not hear you.
Before going further, you should practice with the following structures.
Remember that you are trying to understand the process and the mechanics
of the rule.

1. not those apples were smelling rotten

2. not Estelle would have done that
3. not you are reading fast enough
4. not Lucille will have finished by then
5. not we had heard the news
 the negative transformation 47

Some sentences have only tense in the auxiliary and, therefore, no

Aux1. For these sentences the rule we have given will not apply. We need
to write new rules for these sentences; then we will see how we can modify
the rule that we have already formulated. Examine the following deep
structures on the left and their corresponding surface structures on the right:

1. not they are our friends They are not our friends.
2. not Jane was friendly Jane was not friendly.
3. not the bird was there " The bird was not there.

Like the sentences with Aux1, these demand a rearrangement of the structure.
This time not is placed after be instead of after an auxiliary. Notice that in
these sentences be is not an auxiliary, since there is no verb following it and
since there is no ing on the next word. The be in these sentences is part of
the MV. We write this rule as follows:

not + X + tense -|- be -j- Y => X + tense -f- be J- not + Y

This rule operates on the following deep structure:

Adj

not the answer past be true

The rule transforms the deep structure into the following surface structure:

Adj

the answer past be not true

The answer was not true.
48 chapter five

We have now covered those sentences with some element in the

auxiliary in addition to tense; of the sentences with no such auxiliary, we
have covered those that contain be as part of the MV. That leaves only those
sentences with only tense in the auxiliary and with verbs other than be in
the MV. The following structures illustrate the transformation involving
these verbs:

1. not we play often We do not play often.

2. not they taste the salt They do not taste the salt.
3. not Terry eats early Terry does not eat early.
4. not the janitor did it The janitor did not do it.
5. not the man sees me The man does not see me.

In the surface structure not comes before the verb and after tense, which is
attached to do. If we omit do from the surface structure, we obtain the
following:

1. We present not play often.

2. They present not taste the salt.
3. Terry present not eat early.
4. The janitor past not do it.
5. The man present not see me.

These are not grammatical sentences, since not cannot be altered to show a
contrast between past and present. To provide a grammatical sentence, we
add the word do. Although this word has no lexical meaning, it can carry
the tense morpheme: do and does in contrast to did. These sentences
require two rules:

not + X + tense + V + Y => X + tense + not + V + Y

X T tense + A -f- Y => X + tense + do + A + Y

The negative rule moves not between tense and the verb. In the second
rule, A stands for any morpheme other than an auxiliary or a verb. Any
time tense appears before any such morpheme, we add the word do to carry
the tense. To convert the deep structure not we present jump here into a
surface structure, we apply the negative and do transformations as shown in
the following trees.
 the negative transformation 49

SM Nuc

N VP.

N Aux' MV place

tense V

not we present jump here

The negative transformation applies to this deep structure to produce the

following intermediate structure:

Nuc

Now the do transformation applies to produce a surface structure:

After the application of phonological rules this becomes We do not jump

here.
50 chapter five

We have now given three rules for the negative transformation in

English, as follows:

not + X + tense -f- Aux1 -f Y => X + tense -f Aux1 -f- not Jr Y

not -f I-f tense + be -f tense -f be -f- not -f- Y
not + X + tense + V + Y => X -f- tense + not + V -f T

There is a great deal of repetition among these rules. We have a means of

combining the three:
Aux1 Aux1 -f- not
not -j- X -f tense be Y -> X + tense be T not
V not + V

The square brackets indicate that corresponding items must be selected on

both sides of the arrow. If Aux1 is selected on the left, then Aux1 -f not must
be selected on the right; if be is selected on the left, then be -f- not must be
selected on the right; if V is selected on the left, then not -f- V must be
selected on the right.

exercises

A. Transform the following deep structures into surface structures by applying the
negative transformation and, where applicable, the do transformation:
1. not John present be in the room
2. not of course the children past can go with us
3. not those chairs present need paint now
4. not she present look tired
5. not they present be the leaders
6. not the boys present will have arrived by then
7. not no that cat present resemble my sister’s cat
8. not they present have seen me here
9. not Patsy past be friendly
10. not his foot past become infected

B. Study the following sentences and determine what governs the use of some and
any:
1. I read some new magazines. *1 read any new magazines.
2. Some people were at the door. *Any people were at the door.
3. It isn’t in any of the drawers. *It isn’t in some of the drawers.
4. He didn’t have any money. *He didn’t have some money.
5. Some dogs were in the yard. *Any dogs were in the yard.
C. Write five original deep structures that contain the SM not. Be sure that you
can account for all of them with the phrase-structure rules, and use variation
in the selection of auxiliary elements. Then transform them into surface structures.
 chapter six
QUESTION TRANSFORMATIONS

Transformation is the process that converts deep structures into surface

structures. The negative transformation involves a rearrangement of
structure, as when we move not to the position after the first occurring
auxiliary or after be. In the case of sentences containing only tense in the
auxiliary and a verb other than be, the word not is shifted to the position
after tense, and then another transformation adds do. This second trans¬
formation involves addition, rather than rearrangement of structure. In
addition to these two processes, a transformation may delete material from
the deep structure:

He can’t sing well, but I can sing well.

He can’t sing well, but I can. (sing well is deleted)

Or it may substitute something for a morpheme in the deep structure:

When Jane saw me, Jane screamed.

When Jane saw me, she screamed.

Some transformations involve a combination of these four processes:

rearrangement, substitution, deletion, and addition. Phrase-structure rules
do not perform any of these processes; rather, they expand elements (NP,
Aux, etc.) into their constituents.
English has two main kinds of questions: those that are answered yes
or no (Are you ready? Did he leave? Should I stop?) and those that are
answered by other words (Where are you going? Whose book are you
reading? What is her name?). It will become obvious as we go along that
these two kinds of questions are formed differently. We call the first type
yes/no questions, the second type WH questions, since many of them begin with
words with the first letters wh. In addition, there are questions such as Sue
is going, isn’t she? and Sue isn’t going, is she? Then there is the echo

51
52 chapter six

question: You heard him come in? We will not be treating these kinds of
questions, although they can easily be handled by the same processes we
use for yes/no and WH questions.
It would be possible to derive yes/no questions from related declaratives
such as the following:

1. Tom is sick. Is Tom sick? ' %

V

2. They have already left. Have they already left?

3. He heard us. Did he hear us?

With this approach we would have the same morphemes in both structures
(except for do in Did he hear us?), but the transformation would change
the meaning of the sentence. Is Tom sick? does not mean the same thing as
Tom is sick. A principle of our grammar is that transformations affect the
form of a structure but not the meaning. By means of the negative trans¬
formation we derived He will not go from not he will go, both of which
have the same meaning. He will not go is not derived from He will go, but
is merely similar to it.
Likewise, Tom is sick cannot be the deep structure for Is Tom sick?
although the two are similar. Just as the idea of negation must be present in
the deep structure of a negative sentence, so the idea of interrogation must be
present in the deep structure of a question. This idea is expressed by the
SM Q, which indicates that the structure is a question. More specifically,
it may be interpreted as meaning “I request that you answer yes or no to the
question. . . .”
The sentences on the left below are deep structures that have the
sentence modifier Qy those on the right are surface structures:

1. Q she could sing well Could she sing well ?

2. Q^the book has become wet Has the book become wet?
3. Q_the bell is ringing now Is the bell ringing now?

The process of forming questions, like that of making negatives, is not new
to you. When you were very young, you incorporated into the grammar
that you were learning certain rules for forming questions and negatives.
These rules are still part of your grammar, but you are probably not
conscious of the intricacies of this grammar, and your observations about
it may be inaccurate. You might say that you are moving the verb in front
of the subject noun phrase to form a question. This process, of course, would
give *Could sing she well? for the first question above, and you would
begin refining your observation, as you did in describing the negative
transformation. Tense and the first auxiliary (tense -f- Aux1) have been
 question transformations 53

placed in front of the noun phrase in the surface structure; Q has been
deleted. This transformation, like the negative, involves a rearrangement of
elements. This part of the yes/no rule can be written as follows:

Q, + NP -f tense + Aux1 2 3 + X => tense + Aux1 -j- NP -f- X

This rule will apply to a deep structure like this:

Q the bell present be ing ring now

It produces the following surface structure:

Is the bell ringing now?

Now we will examine sentences with no Aux1:

1. Q, the men are lucky Are the men lucky ?

2. Q, he was our supervisor Was he our supervisor ?

3. Q Betty is at home Is Betty at home ?

When tense is the only element of the auxiliary and the main verb is be, the
subject noun phrase changes position with tense and be, and Qis deleted:

Q, + NP + tense + be + X => tense + be + NP + X

54 chapter six

So far we have seen a parallel between this transformation and the

negative. Sentences with an auxiliary other than tense behave similarly to
sentences with no such auxiliary but with be as the main verb. Sentences
with only tense as the auxiliary and with verbs other than be behave
differently under the negative transformation. Let us see whether this
parallel is extended to the yes/no transformation:

1. Q John read my letter Did John read my letter?

2. Qthe teachers eat here Do the teachers eat here ?
3. Q, she knows my name Does she know my name ?

If we omit the meaningless word do from the surface structures, we are left
with structures such as past John read my letter. When there is no
auxiliary other than tense and the main verb is not be, tense and the noun
phrase change places, so that Q she present know my name becomes
present she know my name. Since present cannot be attached to she, we
apply the do insertion rule: present do she know my name. These steps
can be shown as follows:

Q she present know my name

By the yes/no transformation this becomes:

S
 question transformations 55

Then do insertion applies:

The complete rule for the yes/no transformation can be stated as

follows:
Aux1 (not) Aux1 (not) X
Q, + X + tense be (not) Y =,> tense be (not) X
(not) V _ (not) A + V

Not has been included in parentheses to allow for such sentences as Isn’t
she going with us?
According to our rules, we can have a deep structure like Q not she
is going with us. We apply the transformations in the order (1) negative,
(2) yes/no, (3) do. By negative we get the intermediate structure Q she
isn’t going with us. A structure like this that has had one or more trans¬
formations applied to it but which still is not a surface structure is called an
intermediate structure. Then the yes/no transformation applies to produce
the structure Isn’t she going with us? Since tense can be attached to be,
the do transformation is inapplicable.
Now let us examine the other kind of question, the WH question, as in
What is he saying? This surface structure is derived from Q he is saying
something, or preferably Q he is saying NP-WH. After the application of
the yes/no transformation, we have the intermediate structure Is he saying
NP-WH? The WH transformation substitutes the interrogative what for
the noun phrase and shifts it to the beginning of the sentence: What is he
saying? These processes are illustrated by the following trees. Here is the deep
structure:

Q he present be ing say WH

56 chapter six

The yes/no transformation applies:

The WH transformation then shifts the NP with WH attached to it to the

beginning of the sentence and substitutes what:

ing V

What present be he ing say

Notice the process in the following sentences. The ones on the left are
intermediate structures that have undergone the yes/no transformation;
those on the right have had the WH transformation applied to them:

1. are you reading NP-WH What are you reading ?

2. has she torn NP-WH What has she torn?
3. were you giving it to NP-WH Who(m) were you giving it to?

A noun phrase that has WH attached to it is replaced by what or who and

moved to the beginning of the sentence.
If the noun phrase with WH-attachment is the object of a preposition,
either the whole prepositional phrase or just the noun phrase may be moved.
The choice is entirely stylistic, What are you writing with? being less formal
than With what are you writing? We can express this transformation this
way:

who
X + (Prep) NP-WH + Y => (Prep) X x Y
what
 question transformations 57

In the structure are you reading NP-WH now, X is everything before the
noun phrase (are you reading) and Y is everything after it (now). Either X
or Y may be nothing, as in Are you reading NP-WH, Y is nothing.
As we have stated the rule, there is no way of knowing whether Q you
saw NP-WH will become What did you see? or Whom did you see? This
choice between who and what depends on whether the NP has the feature
[ -f- human ] or not. The deep structure will attach to the NP not just WH,
but also [ + human] or [ — human]. The lexicon will give who as
( -)- human ] and what as [ — human ].
But noun phrases are not the only structures that may have WH
attachment. We may find it on an adverbial of place, as in Where are you
going? which derives from are you going Adv-p-WH. Various adverbials
may have WH attachment; the interrogatives that replace them are listed
below, along with a possible answer:

Adverbial Replacement Answer

place (Adv-p) where there; in the yard
time (Adv-t) when then; tomorrow
manner (Adv-m) how carefully; with care
reason (Adv-r) why because of her; for me

In addition to noun phrases and adverbials, determiners may have

WH attachment. Whose replaces a possessive (my, his, John’s, etc.);
which (and sometimes what) replaces possessives or demonstratives. The
WH transformational rule may be summarized as follows:

Adv-p-WH where
Adv-t-WH when
Adv-m-WH how
Adv-r-WH why
Y => ( whose ) X A- Y
Det-WH + N j which 1 + N
( what J

(Prep) NP-WH
_ <-p) (:i:«) _
exercises

A. Transform the following deep structures into surface structures:

1. Qyou present be going to the movie
2. Q^the student past see the flag
3. QI present may leave
4. Q^you present be afraid
5. they past know you
 58 chapter six

B. Give the deep structure from which each of the following sentences was trans¬
formed :
1. Are you watching the clouds ?
2. Have you eaten today?
3. Does she know his address?
4. Were they in the drawer?
5. Did Bill plant the tree?
C. Our rule for the yes/no transformation changes the deep structure Q you present
have a dime to Do you have a dime? It will not produce Have you a dime?
Similarly, the negative transformation of not you present have a dime produces
You don’t have a dime, but not You haven’t a dime. Have you a dime? and You
haven’t a dime are still common in British English, although they are dying out
in American English. Can you revise our rule or add a new one to take care of
these structures ?
D. Perform the WH and do transformations, where applicable, on the following
intermediate structures:
1. were they singing NP-WH
2. present you live Adv-p-WH
3. was she answering Adv-m-WH
4. have they answered Det-WH questions
5. past you see NP-WH
E. Transform the following deep structures into surface structures, performing the
transformations in the order (1) negative, (2) yes/no, (3) WH, (4) do, and only one
at a time:
1. Q, Bob will speak Adv-p-WH
2. Q,not he is going Adv-r-WH
3. Q,he wrote with NP-WH
4. Q, not you have found Det-WH book
5. Q she wanted NP-WH

F. Give the deep structure from which this sentence was derived: What could the
man have been doing?
1. From the deep structure give the words that are represented by the following:
Aux1, V, M, be, Aux, MV, VP.
2. Which tense is found in this sentence? Which word shows the tense?
3. In this sentence why is the ing form of give used instead of gives or given?
Why is the en form of be necessary? (Answer these questions from the
standpoint of form, not meaning.)
4. Perform the yes/no transformation on your deep structure. From the rewrite
rule for the WH transformation, which words in this intermediate structure
are represented by X? by Y?
G. Why do modals not have en or ing forms?
H. In the yes/no rule for sentences containing auxiliaries, why do we specify Aux1
instead of just Aux?
I. The verb have has an ing form (having), but the auxiliary have does not. Why?
 chapter seven
TRANSFORMATIONAL
PROCESSES

We have seen that the phrase structure section of a transformational grammar

enables us to produce a limited number of simple structures such as the
following:

Q you past be ing sleep

We call this an underlying or deep structure. Since this structure contains

the SM Q, it meets the conditions for the yes/no transformation, which
rearranges the structure:
S

59
60 chapter seven

Since this structure does not contain not or WH, it does not meet the
conditions for the other transformations we have discussed so far. This is,
then, a surface structure. The rules in the phonological component of the
grammar will give the structure a pronounceable form: Were you sleeping?
A transformation may involve any of four processes: addition, deletion,
rearrangement, or substitution.

1. By addition we place something in the tree'that was not in the deep

structure; thus, we may add do to

he past not disappear

to give:

he past do not disappear

Since the deep structure must contain the full meaning of the sentence,
only words which are relatively empty in meaning, such as do, may be
added transformationally.
2. By deletion we remove something from the structure. We have not yet
encountered deletion, but you can readily see how it operates on the
following structure:

Bill couldn’t hear you, but I could hear you.

Bill couldn’t hear you, but I could.

In the second sentence the MV underlying hear you has been deleted,
since it is repetitious. Only elements that cause no loss in meaning may
be deleted.
3. Rearrangement changes the ordering of the structure. We have seen this
process in the negative, yes/no, and WH transformations. As with the
other transformations, rearrangement produces a change in structure
and is not just a shifting of words.
4. Substitution involves replacing an element of the deep structure with
another element, as the substitution of where for Adv-p-WH. The WH
transformation involves a combination of rearrangement and substitution.

In this chapter we will examine several transformations that illustrate these

four processes.

ADVERBIAL MOVEMENT

Our phrase-structure rules introduce all adverbials after the MV, as

in sentences la, 2a, and 3a:

la. I saw her at the bank yesterday.

 transformational processes 61

2a. He found some luggage in the closet.

3a. We didn’t stay long because of the rain.

The following are also grammatical sentences of English:

lb. Yesterday I saw her at the bank.

2b. In the closet he found some luggage.
3b. Because of the rain we didn’t stay long.

Since these sentences mean the same thing as the corresponding sentences
la, 2a, and 3a, we would like to account for this in our grammar. The
structure underlying I saw her at the bank yesterday is as follows:

S
i
I
^Nuc

NP

I past see her at the bank yesterday

It will be possible to derive Yesterday I saw her at the bank from this same
deep structure if we rearrange the elements:

yesterday past see her at the bank

Many adverbials may undergo this rearrangement:

la. I eat there sometimes,

b. Sometimes I eat there.
2a. I read about the accident in the newspaper,
b. In the newspaper I read about the accident.
62 chapter seven

Most adverbials of manner do not shift except for emphasis: *Carefully he

checked the papers. The adverbial-movement transformation, like the
negative and yes/no, rearranges the structure underlying a sentence.

INDIRECT OBJECTS

Now let us turn to another group of sentences which share meaning:

la. They sent a note to me.

b. They sent me a note.
2a. Mary has mailed a letter to her friend,
b. Mary has mailed her friend a letter.
3a. She cooked a meal for me.
b. She cooked me a meal.

Our phrase-structure rules will account for la, 2a, and 3a, but not for lb,
2b, and 3b, in which the words me (lb), her friend (2b), and me (3b) are
said to function as indirect objects. Since sentences la and lb mean the
same thing, we would like to derive them from the same deep structure:

Nuc

they past send a note to me

This is the deep structure for They sent a note to me. If we rearrange the
structure, we can derive sentence lb:
 transformational processes 63

This transformation rearranges; it also deletes the preposition to.

It is not enough to say that a transformation rearranges a structure.
We must know what kind of structure the transformation may be applied to.
For example, the negative transformation applies only to those deep,
structures containing the SM not, the yes/no only to those containing Q.
Also, we must know specifically how the structure is rearranged. We write
a rule in which the left side defines the structure to which the transformation
is to be applied and in which the right side describes the structure after
the change is made. In our other rules we included an element to trigger the
transformation: not, WH, Q. These transformations are different from the
indirect object transformation. In the others, the transformed sentence has
a different meaning from the related untransformed sentence without
not, Q, etc. She was going to town, a sentence without the optional SM not,
does not mean the same thing as She was not going to town, in which not
has been selected. But look at the sentences involving indirect objects.
They sent a note to me and They sent me a note are exact paraphrases of
each other. The indirect object transformation is optional; there is, therefore,
no marker in the deep structure to indicate that it must be performed. We
might state our rule as follows:

X + V + NP1 + { j+
f°r Np2 => x + v + Np2 + Npl
The NPs are numbered so that we can distinguish them. We may illustrate
the sentence She cooked a meal for me as follows:

X V NP1 for NP2 => X V NP2 NP1

she past cook a meal for me => she past cook me a meal

We want to state our rules precisely enough so that they will permit
all grammatical sentences of English; at the same time, we want to prevent
all ungrammatical ones. This rule will account for They sent me a note
and the other sentences we have given, but it will also permit the following:

1. *They sent me it.

2. *Mary has mailed her friend it.
3. *The driver must have given John it.

From these sentences we see that in the deep structure the first NP following
the verb must contain a common noun, but the NP after the preposition
may contain any kind of nominal. Hence, the structure of They sent the
book to me fulfills this requirement since book is a common noun. They
sent it to me does not since it is not a common noun. We are speaking only
of American English, since there are slight differences found in British usage.
We should now revise the rule:
64 chapter seven

+ N to
X + V + (Det) + NP
-f- common for
+ N
X + V + NP + (Det)
+ common

This rule is optional. Whether we apply it or not depends upon stylistic

preferences. It will transform We shipped the fruit to Tom to We shipped
Tom the fruit and We made a bird house for Sam to We made Sam a bird
house. In American English it will not apply to The/ gave it to me or We
threw it to John.

IMPERATIVES

Now that we have examined several transformations involving re¬

arrangement, let us look at one of the other processes. An imperative
sentence such as Close the door or Be good has obviously undergone a
transformation, since there is no noun phrase before the verb. This NP
must have been deleted. Traditional grammarians said that these sentences
are derived from You close the door and You be good. At first glance,
You close the door appears to pose no problem, but the string you -f-
tense + be + good yields You are good or You were good, not You be
good. If we inquire about tense in You close the door, we see that this
cannot be turned into past tense and retain the same meaning. The
traditional explanation is not valid.
We may gain insight into what deep structures underlie Close the
door and Be good by examining tag questions like the following:

1. He was ready, wasn’t he ?

2. They had been sleeping, hadn’t they?
3. You can ski, can’t you?
4. She had finished the book, hadn’t she?
5. You will go, won’t you?

The first part of these sentences before the comma is generated by the
phrase-structure rules that we have already outlined. Following the comma
is a repetition of tense, Aux1, and noun phrase. We could write a rule for
this kind of tag question:

NP + tense + Aux1 -j- X => NP + tense + Aux1 -f X + tense

-f- Aux1 -j- not + NP

 transformational processes 65

Notice that the tense remains the same, that the first occurring auxiliary is
repeated, and that there is no change in the noun phrase: He was going,
wasn’t he? but not *He was going, isn’t she?
Now let us see what tag questions are necessary for imperatives:

1. Close the door, won’t you?

2. Be good, won’t you?

✓

3. Answer my question, won’t you?

These examples show that in the deep structure the noun phrase is you,
the tense is present, and the first auxiliary is will. These three sentences
must go back to structures similar to You will close the door, You will be
good, and You will answer my question. These structures have no mor¬
pheme in them that requires a transformation; we, therefore, add the
sentence modifier Imp (imperative). We now can generate a deep structure
like this: Imp you present will go home. Notice that this differs from the
deep structure of you will go home by the inclusion of Imp, which means
that the sentence is an imperative.
Here is the rule for the imperative transformation:

Imp (not) + you -f- present -f- will X => (not) X

This rule is an example of the process of deletion. Notice that it is much

more restrictive than the others we have encountered. In the other trans¬
formations we were able to use NP, since they apply for all kinds of noun
phrases in the same way. The imperative transformation, on the other hand,
applies only to sentences containing you as the first noun phrase. One
principle of our grammar is that the listener must always be able to under¬
stand unambiguously which words are deleted. If just any NP were deleted,
he would not be able to do this; but if there is a rule saying that you in
imperatives may be deleted, both speaker and listener have no problem
deciding what has been left out. Similarly, we must specify present tense,
not just tense, since sentences with past tense cannot undergo this trans¬
formation. Finally, the first auxiliary must be will, not just any modal.
When you hear someone say Open the door or Be on time, you are able to
recover the deleted information unambiguously. Deleted information is
always understood by both speaker and listener.
We have now examined in some detail the four transformational
processes: rearrangement, substitution, addition, and deletion. All trans¬
formations involve one or more of these processes in converting deep
66 chapter seven

structures into intermediate or surface structures. Transformations alter the

structure of the sentence, but they do not affect the meaning given by the
deep structure.

exercises

A. Perform the indirect object transformation:

1. She has mailed a letter to me.
2. The farmer gave a watermelon to Fred.
3. Those people threw pennies to the dancer.
4. Someone was telling a story to us.
5. Don must have handed the eraser to Bill.
B. Transform the following deep structures into surface structures. Take the
transformations one at a time and in the order (1) imperative, (2) negative,
(3) yes/no, (4) WH, (5) do.
1. Imp you will wash the car now
2. Imp not you will open a window
3. Q,he can do NP-WH
4. Imp you will tell the answer
5. Qnot she arrived on time
C. For each of the following sentences give the deep structure and take it through the
relevant transformations to derive the given surface structure:
1. He didn’t tell me his plans.
2. Where are you singing tonight?
3. Listen to me.
4. Yesterday he didn’t know her address.
5. Why aren’t you ready?
 chapter eight
THE PASSIVE «

TRANSFORMATION y

Because transformations affect form, two surface structures may be different

but share the same deep structure. On the surface, She read me a story and
She read a story to me are different, but the native speaker of English
understands them to mean the same thing. Our grammar shows how the
sentences are related by saying that they have the same deep structure but
that the optional indirect object transformation has been applied to the
structure of the first sentence but not to that of the second. Here are some
more pairs of sentences that share meaning:

la. Walter saw me.

b. I was seen by Walter.
2a. The boys had eaten the cake,
b. The cake had been eaten by the boys.
3a. The waiter is clearing the table,
b. The table is being cleared by the waiter.
In each of these pairs of sentences the two noun phrases have been inter¬
changed :

Walter saw me
I was seen by Walter

In addition, the auxiliary has been expanded. This expansion of the auxiliary
should be examined carefully. Below only the sequence Aux + V has been
given:

1. past + see saw

past + be -f en -f see was seen
2. past + have + en + eat had eaten
past -j- have -f- en -f- be + en + eat had been eaten
3. present + be + ing + clear is clearing
present + be + ing -j- be -f- en -f- clear is being cleared

It should be obvious now that be -f en has been added between the auxiliary
and the verb.

67
68 chapter eight

You remember that en is the abbreviation for “past participle of.”

We earlier saw this form after the auxiliary have: have seen, have eaten,
have torn, have slammed, have opened, have set, have hit. In passive
sentences the same past participle occurs, but it follows a form of be: was
seen, was eaten, was torn, was slammed, was opened, was set, was hit.
We may state the passive rule as follows:

(SM) NP1 + Aux + V + NP2 + A=> (SM) NP2 + Aux + be

-p en -f- V -f- by + NP1 -p X

Both Walter saw me and I was seen by Walter have the following deep
structure:

Nuc

NP

Walter past see me

Since the passive transformation is optional, we may skip it and go directly

to the phonological rules, giving the sentence Walter saw me. But if we
apply the passive transformation, we obtain the following derived structure:

Nuc

I past be en see by Walter

Phonological rules will turn this into I was seen by Walter.

 the passive transformation 69

There is no difference in meaning between passive sentences and their

corresponding actives. This transformation, like the one that produces
indirect objects, makes stylistic variations but not changes in meaning. The
passive is often less direct than the active and serves a useful function when
a person wants to emphasize the object or when the actor is unimportant
or unknown:

1. My house was destroyed by the storm.

2. I’ve been stung.
3. He was wounded in battle.

Not all verbs that have noun phrases following them may undergo the
passive transformation:

1. Bill has a cold. *A cold is had by Bill.

2. The book cost a dollar. *A dollar was cost by the book.
3. The box contained a secret.
4. That dress suits her personality perfectly.
5. She resembles her mother slightly.
6. He married his childhood sweetheart.
7. The room measures twelve feet across.

There is one sentence in which have occurs in the passive: A good time was
had by all. This is, of course, an exception. There are also two exceptional
verbs that have passives but no corresponding actives: to be born and to be
had (“to be taken advantage of”). For some of these verbs, such as have,
there will be a feature listed in the lexical entries to block the passive trans¬
formation. Others can probably be shown to result from deep structures in
which there is no NP after the verb.
Sometimes the deep structure subject is vague, as in these sentences:

la. Someone signed the paper.

2a. Someone has found the car.
3a. Something stung me.

By the passive transformation these can become

lb. The paper was signed by someone.

2b. The car has been found by someone.
3b. I was stung by something.
70 chapter eight

We may apply the deletion transformation to the prepositional phrase con¬

taining the indefinite someone or something:

lc. The paper was signed.

2c. The car has been found.
3c. I was stung.
* \
V

Only indefinites or other obvious NPs, such as The defendant was found
guilty (by the jury) may be deleted. The listener must always be able to
understand what has been deleted.
It is possible with the selection or rejection of various optional trans¬
formations to have several different surface structures all deriving from the
same deep structure:

1. Deep structure: Someone past hand a note to Fred

Surface structure: Someone handed a note to Fred.
2. Deep structure: Someone past hand a note to Fred
T-indirect object: Someone past hand Fred a note
Surface structure: Someone handed Fred a note.
3. Deep structure: Someone past hand a note to Fred
T-passive: A note past be en hand to Fred by someone
Surface structure: A note was handed to Fred by someone.
4. Deep structure: Someone past hand a note to Fred
T-passive: A note past be en hand to Fred by someone
T-deletion: A note past be en hand to Fred
Surface structure: A note was handed to Fred.
5. Deep structure: Someone past hand a note to Fred
T-indirect object: Someone past hand Fred a note
T-passive: Fred past be en hand a note by someone
Surface structure: Fred was handed a note by someone.
6. Deep structure: Someone past hand a note to Fred
T-indirect object: Someone past hand Fred a note
T-passive: Fred past be en hand a note by someone
T-deletion: Fred past be en hand a note
Surface structure: Fred was handed a note.

The native speaker recognizes all six sentences as meaning the same thing.
Our grammar accounts for this.
Grammarians in the past were often concerned with the meaning of
the terms subject of and object of In the sentence Her husband washed the
dishes, there is no problem. Her husband, the actor, is in subject position
before the verb; the dishes, the receiver, is in object position after the verb.
 the passive transformation 71

A passive, however, such as The dishes were washed by her husband,

presents a problem, since the actor is not in subject position nor the receiver
in object position. Some grammarians spoke of the dishes as being the
grammatical subject but logical object. In transformational grammar there
is no problem. Such functions as subject and object have meaning only at a
specific level, and these functions do not necessarily correspond among
levels. Hence, in the deep structure her husband is the subject and the
dishes is the object. In the surface structure, however, the dishes is the
subject and her husband the'"object of a preposition. Since meaning is
reflected in the deep structure and form in the surface structure, there is no
problem of “logical subject” and “grammatical subject.”
Our grammar also accounts for certain feature restrictions. In Chapter
Four we saw that the following sentences are ungrammatical because they
violate feature restrictions:

la. *Honesty admired the man.

2a. *Sue amazed his honesty.

The corresponding passives are also ungrammatical:

lb. *The man was admired by honesty.

2b. *His honesty was amazed by Sue.

If we were limited to surface structures in explaining feature restrictions,

we could not merely say that honesty may not be a subject of admire, for
the sentence Honesty is admired by the man is perfectly grammatical. One
rule based solely on surface structures would have to contain the following
points:

1. Admire may not have an abstract subject in active sentences:

*Honesty admired the man.
2. Admire may have an abstract subject in passive sentences:
Honesty is admired by the man.
3. In passive sentences abstract nouns may not occur in by phrases following
admire: *The man is admired by honesty.

Such a statement of facts, although true, has two serious weaknesses: it is

uneconomical in wording, and it fails to show that the three points have
something in common. Yet the native speaker feels that there is some
relationship among the three points, that it is precisely the fact that
*Honesty admired the man is ungrammatical which makes *The man was
admired by honesty ungrammatical. A grammar that recognizes deep
72 chapter eight

structure can both shorten the rule and, more importantly, show the
relationships among the parts: “Admire may not have a deep structure
subject that is abstract; it has no restrictions on the deep structure object.”
Hence, neither *Honesty admired the man nor *The man was admired by
honesty will be possible since they would share the same deep structure
which has been blocked by the feature constraints. Both The man admired
honesty and Honesty was admired by the man are possible, since their
common deep structure violates no constraints. '

exercises

A. Perform the passive transformation on the following structures:

1. The maid was sweeping the floor.
2. Jane sang a song in the theater.
3. Tom must have milked the cow yesterday.
4. The man was throwing the pillow.
5. The sailor could have sunk the boat.

B. Write the deep structure from which each of the following sentences was derived:
1. The rocks are being hurled by that boy.
2. Treasure was found by the pirate.
3. The note could have been planted by the suspect.
4. John was being persuaded by them.
5. Your lesson could be prepared by Sam.

C. Below are five pairs of surface structures that differ. In each pair if the sentences
mean the same thing, account for their differences in form by giving their deep
structure and showing which transformations have been applied. If the two
sentences do not mean the same thing, show how their deep structures account
for this:
la. Sally has been sent a present,
b. Someone has sent a present to Sally.
2a. What has he discovered in the drawer?
b. Has he discovered something in the drawer?
3a. Didn’t he tell the secret to his wife?
b. Didn’t he tell his wife the secret?
4a. Bill wasn’t fired by the superintendent,
b. The superintendent didn’t fire Bill.
5a. You won’t be staying there long,
b. Don’t stay there long.
part three

TRANSFORMATIONS 11
 chapter nine
PRINCIPLES OF
TRANSFORMATIONAL GRAMMAR

Now that we have discussed the phrase structure, the lexical features, and
the process of transformation, you are in a better position to understand the
principles of transformational grammar than when we first mentioned them
in Chapter One. It will be useful now for us to restate those principles and
illustrate them with material from the grammar that you have studied.
During the first years of a child’s life, he hears many sentences spoken.
Eventually he begins experimenting with sentences as he tries to speak. He
learns to associate objects and actions with specific words, and he learns to
combine these words with one another to produce larger, meaningful
structures. The English-speaking child learns that determiners precede
nouns and that prepositions precede noun phrases. He learns that if he has
more than one book, he speaks of his books; similarly, the plural of toy is
toys and of dish is dishes. The plural ending is pronounced differently in
each of these words. When he learns a new noun, he automatically applies
his rule for plural formation to it and produces cats, ducks, dogs, foots, and
mans. The last two he later learns are exceptions and memorizes their
plurals. When the child says foots or mans, we know that he has mastered
the rule for plural formation in English. By the time he is five or six, he has
mastered many other rules, such as word order and past-tense formation.
We do not know what form these rules have taken in his mind, since their
application becomes routine with him. The rules he has developed which
permit him to understand and produce the sentences of his language are
collectively known as a grammar. The child’s grammar constitutes his
knowledge of the language.
Each person develops a grammar of his language. Furthermore, he
develops it largely by himself with very little technical help except an
occasional correction such as “Don’t say singed; say sang.” It should not be
surprising, then, that no two people develop exactly the same grammar.
People’s speech patterns are as distinctive as their fingerprints. This
uniqueness is made obvious by our ability to recognize many people’s voices

75
76 chapter nine

on the telephone, radio, or television. Less obvious but no less distinctive

are the choices people make in selecting words and syntactic patterns.
Although each person develops his own unique grammar, it has much
in common with the grammars developed by other people with whom he
comes in contact; otherwise, they would not be able to communicate with
him or would think he sounded odd. In fact, most English-speaking people
can understand one another. As we might expect,, a person’s grammar will be
quite similar to those of the people he is in frequent contact with. He will
sound more like people in his own family than people in another part of
town, more like people from his own region of the country than people from
other regions, more like people from his own social class than people from
other classes.
We refer to an expression as grammatical for this person if it conforms
to the rules of the grammar that he has developed. When his rules are the
same as the rules of other speakers of his language, all will agree that certain
expressions are grammatical and others ungrammatical. The following are
ungrammatical for all speakers of English:

1. *Book the on the table is.

2. *The girl seemed.
3. *The man elapsed.

Other structures are grammatical for some speakers but not for others:

1. She looked a fool.

2. We shan’t be long.
3. We might should do it.
4. We seen it.
5. We have saw it.
6. We have seen it.

For the speaker of sentence 4, the principal parts of see are probably see,
seen, seen; for the speaker of sentence 5, see, saw, saw; for the speaker of
sentence 6, see, saw, seen. These differences exist because these three
speakers probably were not in contact with the same people as they were
learning to talk. Each one developed a rule based on the raw material that
he heard. Notice that the grammars which produced sentences 4 and 5 have
only two distinct principal parts for see; the grammar that produced
sentence 6 has probably made a similar simplification for such verbs as
cling, slink, stink, and shrink, all of which originally had three distinct
forms. Each of the above six sentences is grammatical for some speakers of
 principles of transformational grammar 77

English; none is grammatical for all. Notice that our use of grammatical
means nothing more than that the structure under consideration follows
the rules of whatever grammar we are discussing. We are not using this
term to appraise the grammar for its effectiveness or its ability to conceal
one’s regional or social origin.
The rules comprising this grammar that each person creates are limited
in number; also, the number of words each person uses and understands is
finite. With a limited number of words and rules for combining them into
larger structures, a person is able to create and to understand an infinite
number of sentences. You would see that this statement is not exaggerated
if you looked for repetitions of sentences in a large book. And if you taped
your own conversations for a week, you would find that such repeated
structures as Good morning, How are you? and What time is it? make up
a very small part of what you say; there is very little repetition in your
other sentences. Although you are not conscious of it, most of the sentences
you encounter each day are totally new to you; on the other hand, the
rules that formed these sentences are very well known to you. This is one of
the ways your speech differs from that of a talking bird. The bird says only
sentences it has memorized. You have memorized almost no sentences, and
it would be impossible for you to memorize all of the ones you need or can
use. You are able to make use of a finite set of elements and rules to create
an infinite number of sentences.
If there were some way to discover the form these rules assume in a
person’s mind, the linguist’s task of describing the language would be easy.
As it is, he must be content with trying to describe the regular patterns that
he observes. His description is expressed in the form of rules, and we call his
collection of rules a grammar. We, therefore, use grammar with two meanings.
The first is the rules a person has developed which permit him to understand
and to create sentences in his language. The second is a theory about these
rules and is an artificial, less efficient way of expressing them. It tries to
describe the native speaker’s knowledge of his language; no one claims that
it succeeds in duplicating this knowledge, nor is it likely that it ever will.
Subject, object, noun phrase, etc. are terms the linguist has created to describe
certain relations. Although the concepts that these terms name must exist in
the individual’s grammar, the terms belong strictly to the grammar of the
linguist. We avoid further confusion over the word grammar by referring to
such matters as choice between don’t and doesn’t, who and whom as usage.
Notice that this kind of grammar is intended not as a model for learning one’s
language. Such a grammar attempts the much more difficult task of de¬
scribing the grammar that the individual has developed.
The central component of transformational grammars is syntax,
which consists of two divisions: the phrase structure and the transformations.
78 chapter nine

The phrase structure generates abstract deep structures such as

SM + Det + N + past -f- be + ing + V + N + PI

and it also gives a structural description to this string of elements, telling us

that Det + N is a noun phrase, that past -f- be + ing is an auxiliary, etc.
The elements N and V are assigned features (T common, + human,
4-_NP, etc.) which are then matched with 'entries in a lexicon to give
the following:

Q -f- the -f- woman + past + be + ing -f- sell -f- apple + PI

With this string and its structural description we momentarily leave the
syntactic component.
The second division of the grammar is the semantic component,
which operates on the deep structure. The morpheme Q, contains the
meaning of “question” in the deep structure of the above sentence and gives
it a different meaning from the related positive statement, in which Q, has
not been selected. The semantic component shows how the meanings of the
lexical items combine to give the meaning of the sentence as determined by
the syntactic structure. As an example of the dependence of the semantic
component upon syntax, look at the following group of words: old several
buckets rusty metal. Unless you rearrange the words in your mind, you
see no meaning beyond that of each word in isolation. If you rearrange the
words to read several old rusty metal buckets, you find more meaning than
is contained in the individual words. You are able to apply the rules in the
semantic component because you perceive a syntactic structure.
We now return to the syntactic component and apply transformational
rules to our structure to produce

past + be + the -f- woman + ing -f- sell + apple -f PI

This transformation has rearranged elements, but it has not affected the
meaning. Other transformations may add or delete elements; some sub¬
stitute one element for another.
The third division is the phonological component. It tells us, among
other things, to pronounce past + be as was or were, ing -f- sell as selling,
and apple + pi as apples. It gives us an intonation pattern that differs from
that of a group of words read as a list. The phonological component operates
on the surface structure and gives the sentence its final form. We will
describe the phonological component in Part Four of this text.
Transformationalists hope that their grammars will eventually specify
all of the grammatical sentences of the language and no ungrammatical ones;
 principles of transformational grammar 79

this is the proficiency that a native speaker has. No such grammar has yet
been written.
Earlier, structural linguists felt that the ideal way to collect data for
their observations about language was to gather large samples of speech,
preferably recorded without the speaker’s knowledge, and to analyze them.
The transformationalists argue that if the linguist does nothing more than
analyze such a corpus of material he is studying the speaker’s performance
of his language, rather than his competence. If a speaker changes his
train of thought in the middle of a sentence, this structure will be analyzed
along with the “normal” ones, as will false starts and words that are out of
place. Furthermore, the corpus may lack certain constructions, such as
passives or indirect objects. The structural approach analyzes only observed
utterances and makes no comments about those that are not observed. The
transformational approach, however, is as concerned with potential utterances
as it is with those actually spoken at a given time. It attempts to state
whether a structure that has not actually been observed is possible or not.
The transformationalist is interested in the speaker’s competence, the
grammar that he has developed. He has to study the speaker’s performance,
or application of the grammar, but it is merely as a means of approaching
the underlying system. In addition, he makes use of the speaker’s comments
about his language, while realizing that these comments may be inaccurate.
We can see the difference between competence and performance if we
look at an example. In speaking about competence, we will use the terms
grammatical and ungrammatical; for performance we will use acceptable and
unacceptable. In a moment of excitement a person might scream, House is
on fire! If we are speaking of performance, the sentence is acceptable; it
will communicate the idea perfectly well and get the desired results. From
the standpoint of competence, the sentence is ungrammatical: house is a
singular count noun and may not exist without a determiner. Studying the
sentence will add nothing to our understanding of the underlying system of
English determiners; in fact, it could easily lead to the false generalization
that determiners which are normally obligatory are omitted in certain
circumstances. We would be basing this assertion on only one sentence, and
we would have a long wait before we heard another spontaneous utterance
like it. On the other hand, if we ask a native speaker about House is on fire!
he will confirm our original impression that this is a nonce occurrence. It is
also possible for a structure to be grammatical but unacceptable because
it is too long, too conversational, too formal, etc.
The transformationalist in studying the system of language says that
every sentence has a deep structure and a surface structure. Traditional
grammarians at times made similar statements when they said that Go
home comes from You go home; that There is a man at the door should be
converted from its “inverted order” to its “natural order,” A man is at the
80 chapter nine

door, before it can be analyzed, that He is as tall as John comes from He is

as tall as John is tall, etc. Unfortunately, the traditional grammars that
have been the most popular during the last two hundred years made no
systematic use of this concept.
The distinction between deep and surface structure gives us much
insight into the structure of sentences. Most people probably more or less
convert a sentence to its deep structure when they point out the subjects and
*

objects in the following sentences:

1. There is too much noise in the room.

2. There were some people leaning against the wall.
3. Were you driving the car?
4. What has she done with the screen?
5. The man you saw was he.
6. There’s a girl in my class taller than Mary.

At times the surface structure greatly obscures relationships. On the surface,

these two sentences look alike:

1. I found the room mentioned by the students.

2. I found the room cluttered by the students.

The second sentence can mean I found the room cluttered up by the presence
of many students, but usually it means I found the room messy. Although
the first sentence can also have two parallel meanings, the more usual is I
found the room that the students had mentioned. The differences lie not so
much in semantics as in the arrangement of the deep structure. Cl uttered by
the students can be a reduced relative clause, but in sentence 2, it would
usually be an objective complement. It is only through an analysis of
underlying or deep structures that this difference can be brought out.
Without the use of deep structure, it is impossible to show the relation¬
ship between the following sentences:

1. I looked the number up.

2. I looked up the number.

With deep structure we can show that these sentences originated from a
common source; without it, we need two sets of rules for adverbial particles
immediately after the verb and at the end of the sentence.
Or look at these sentences:

1. The smiling girl sat down.

2. The charming girl sat down.
 principles of transformational grammar 81

Without using deep structure, a persomhas no way of showing that smiling

is a present participle but that charming is an adjective. Nor is there any
formal difference in the structures The girl was smiling and The girl was
charming. We can account for this difference if we can show that charming is
derived transformationally from the deep structure The girl charms people.
Deep structure is a subject that is currently undergoing much discussion
and investigation. Many linguists believe that all languages share the same
deep structure and that it is much more abstract than any published account
has yet indicated. Since humans" are all very much alike physically, it does
seem reasonable that there are biologically determined universals in language.
For example, the tongue is shaped for all people so that the tip cannot be in
the front and the back of the mouth at the same time. Research in deep
structures of various languages is giving linguists material for comparison
and is rapidly adding support for language universals.
People in all cultures of the world have speech, yet no one is ever
really taught to talk. The human child hears people speaking and learns to
talk largely on his own. Some animals hear as much speech as the human
child does, and some of them are given far more attention; yet only the
human learns to talk, although some pets have the physical capacity for
speech. It is true that some birds learn to repeat a few memorized phrases,
but only man learns to create new sentences and to communicate his ideas
with speech. Man’s brain gives him an aptitude for speech that is lacking
in other animals.
As the child is listening to people talk, he hears many mistakes (false
starts, stammers, words out of order, etc.) mixed in with “normal” utterances.
Occasionally these mistakes are corrected, but usually they are not marked
as being different from other utterances. In other words, children encounter
only performance, yet all of them develop competence from this exposure
to language, even though conditions are usually poor. They learn to create
and to understand new sentences, no small accomplishment when we realize
how complex human languages are. It seems almost certain that each
child has some innate ability for working with the raw language materials
he finds in his environment and for developing them into his grammar.
The child must bring something more than a blank slate to his language
learning. The systematic study of child language learning has only begun,
but it will no doubt shed much light on our understanding of our language
as more research is performed.

exercises

A. Reserved is an adjective in Those are reserved men but a past participle in

Those are reserved seats. Is this difference entirely semantic, or can you use
deep structure to account for it?
82 chapter nine

B. What is a grammar? What is it intended to provide an account of?

C. It has been said that every normal six-year old knows a grammar of his mother
tongue. Explain. How does this meaning of grammar differ from that used in
question B ?
D. Does grammatical mean the same thing to the transformationalist as it does to the
traditionalist? Explain. Do they both mean the same thing by rule?
E. What are some of the objections linguists have to traditional grammar?
F. What is the purpose of studying the grammar of a language ?
G. What effect should our understanding of a child’s competence in his language
have on the teaching of reading in the first few grades ?

H. Considering the scope of grammar, how useful do you think is much repetition
of words in first-grade readers of the type “Look! Look! Look! See Jack run!
Run! Run! Run!” What reactions do you have to the syntax in sentences like
these ?
 chapter ten
COMPOUNDING ,
DELETION, AND PRO FORMS

So far we have discussed sentences that derive from only one S. It is possible
to join two sentences with a co-ordinating conjunction (and, or, nor, but,
yet, for) to produce a compound sentence: Alice wrapped the package,
and Susan addressed the card.

We will restrict the term compound sentence to two sentences joined by a

co-ordinating conjunction. Other pairs of sentences that may be joined by
a semicolon will be treated as two separate sentences: We listened atten¬
tively to the entire speech; we did not, however, agree with what he
said. Our reason for making this distinction is that two sentences joined by
a conjunction may have deletion of identical elements, whereas those
without a conjunction may not.

1. We went to the exhibition, but (we) did not stay long. (Deletion of the
second we is possible.)
2. We went to the exhibition; however, we did not stay long. (No deletion
is possible.)

To account for sentences joined together in this fashion, we need to

revise our phrase-structure rule for the sentence.

S" ]
PI: S
(SM) Nuc J

83
84 chapter ten

With this revision we may still select (SM) Nuc as the rewrite for S and gen¬
erate the sentences we have discussed in the preceding chapters. But we
now have another possibility: S may be rewritten as two or more Ss without
limit (Sn). In the preceding paragraph we saw a sentence in which only two
instances of S were selected. It would be possible to extend this indefinitely:
Alice wrapped the package, and Susan addressed the cards, and Beth
bought the stamps, and Frances mailed the package. .•■ . . Such a sentence
would be monotonous and stylistically awkward; if we made it very long,
no one would be able to remember it all. But these factors do not affect
grammaticalness. Because of this rule it is impossible for us to point to any
sentence and say that it is the longest in the language. Any sentence can
always have another S joined to it.
Our revised rule for S will permit trees such as the following:

s
/i\
s s

A rule will add a conjunction, such as and, before each S except the first one:

If there are three or more Ss, a further rule will optionally delete all con¬
junctions except the last one:

The sentences are accounted for by the phrase-structure and transformational

rules. Conjunctions are added and deleted by the conjunction addition and
deletion transformations.
There are certain constraints on the compound sentence that we will
not be able to explain satisfactorily until more research is performed on how
sentences interact with one another. The semantic component, for example,
in some way will have to require similarity of meaning: We like Mozart,
but we find Hindemith a bore; *We like Mozart, but that chair looks
 compounding, deletion, and pro forms 85

comfortable. The syntactic component will block certain transformations

in compound sentences: *1 didn’t know you had read Beth’s letter, and
tell me what she said. Normally if one sentence is a question or an imper¬
ative, the other must be also if they are joined by and. This restriction does
not always hold with but: I have already heard the story, but tell it
anyway. The grammar internalized by the native speaker handles these
constraints with ease; we hope that transformational grammar will eventually
be able to state them precisely.
When identical words />ccur in both parts of a compound sentence,
they are often deleted in one of the sentences:

1. Dave went to the carnival and (Dave) rode the Ferris wheel.
2. Bill sold his car and (Bill sold) his motorcycle.
3. Sue (visited us last night) and her husband visited us last night.
4. I was frightened, and Tony was (frightened), too.
5. I enjoyed the concert, but Sally didn’t (enjoy the concert).
6. I could see him, and so could Bob (see him).

Elements deleted by transformations are always clearly understood by

both speaker and listener. For example, in the sentence I enjoyed the
concert, but Sally didn’t, any native speaker of English understands that
Sally did n’t means Sally didn’t enjoy the concert, not Sally didn’t go to
the concert or Sally didn’t wash her hands. This understanding is possible
because deletion can occur only under conditions that are very precisely
specified. We may delete certain repeated elements, as in the sentences
in the preceding paragraph, or we may delete vague or otherwise understood
NPs, as we have seen earlier:

1. Janice ate (something).

2. The car has been found (by someone).

Deletion, like other transformations, alters syntactic structures:

86 chapter ten

The repeated VP is deleted to give the following derived structure:

Another transformation will attach the NP Don and the conjunction and
to the first Nuc:

Usually two or more structures may be conjoined if they are alike (i.e.,
both Ss, both NPs, both VPs, etc.):

1. She is eager and industrious. (Two adjectives)

2. They worked fast but carefully. (Two adverbials of manner)
3. They went to bed late but got up early. (Two VPs)

If two different kinds of structure are conjoined, the result is ungrammatical:

1. *She is friendly and a pleasant woman. (Adjective and NP)

2. *The ball rolled under the car and then. (Adverbial of place and
adverbial of time)

Instead of deleting repeated words, we may substitute a generalized

word known as a Pro form:

1. Do for a VP or a verb
a. I saw the accident, and Bill saw the accident, too.
I saw the accident, and Bill did, too.
 compounding, deletion, and pro forms 87

b. He won't answer the telephone, but I will answer it.

He won't answer the telephone, but I will do it.
2. There for an adverbial of place
W e were sitting in the balcony, and they were sitting in the balcony, too.
We were sitting in the balcony, and they were sitting there, too.
3. Then for an adverbial of time
He saw Tom yesterday, and we saw Fred yesterday, too.
He saw Tom yesterday, and we saw Fred then, too.
4. One for a noun or a noun phrase
a. I have a sister, and you have a sister, too.
I have a sister, and you have one, too.
b. I don't have a new car, but I do have an old car.
I don't have a new car, but I do have an old one.
5. Some, any, every, and no for determiners, added to various Pro forms:
someone, anybody, everything, nowhere, sometime, etc.
6. He, she, it, they for NPs
a. John said that John would go with us.
John said that he would go with us.
b. When I saw Jane, Jane looked worried.
When I saw Jane, she looked worried.
c. I bought a new car and wrecked a new car the next day.
I bought a new car and wrecked it the next day.
d. I don’t know the Smiths or the Smiths’ neighbors.
I don't know the Smiths or their neighbors.

Although we did not label them as such at the time, the WH words which
substitute in the WH transformation are Pro forms. The use of Pro forms,
especially pronouns, is not restricted to repetition within a sentence: I didn’t
know that Bill lived next door to you. Does he visit you often? Some
Pro forms, such as one and someone, are often used because the person they
represent is not known or is generalized. We used these forms in explaining
the passive transformation when we said that Someone has found the ring
underlies The ring has been found.
Transformations show how some sentences that are paraphrases of
each other are related:

la. The project was completed by Fred,

b. Fred completed the project.
2a. Yesterday I found the error.
b. I found the error yesterday.
3a. Susan received my letter and answered it the same day.
b. Susan received my letter, and she answered my letter the same day.
88 chapter ten

4a. Ann and Susan were discouraged,

b. Ann was discouraged, and Susan was discouraged.

It would appear that all conjoined structures—NPs, VPs, adjectives, etc.—

are derived from compound sentences.
Now examine these sentences:

la. Bill’s paper and Sam’s speech were similar,

b. ^Bill’s paper was similar, and Sam’s speech was similar.
2a. He sat between Tom and me.
b. *He sat between Tom, and he sat between me.

Certain adjectives, prepositions, and verbs—similar, different, alike,

between, to combine, etc.—require compound or plural nouns; hence, the
compound sentences lb and 2b cannot underlie sentences la and 2a. There
must be a second source for compound NPs. Such a source can be found in
an expanded version of the phrase-structure rule for the NP:

(Det) N (PI)
NPW

This rule is parallel to the one that rewrites S as Sn. It produces compound
NPs like these:

NP NP NP NP NP NP NP NP NP

The same rule that adds conjunctions between compound sentences will do
the same thing here:

N N

NP and NP NPandNPandNP NPand NPand NPand NP

And a rule will optionally delete all but the last conjunction

NP NP NP

NP and NP NP NP and NP NP NP NP and NP

We now have two sources for compound NPs in surface structures

1. From compound sentences via deletion and rearrangement of structure.

2. From compound NPs in the deep structure.

It is thus possible to have two surface structures identical in form, but

 compounding, deletion, and pro forms 89

derived from different deep structures as shown in the following trees for
Ann and Tom are married.

But upon examination we see that this is an ambiguous sentence and that
the two meanings correspond to the different derivations. It can mean:

1. Both Ann and Tom are married. (But not to each other)
2. Ann and Tom are married to each other.

Ann and To m are married with the meaning of the first paraphrase is
derived from the first tree, with the meaning of the second paraphrase from
the second tree. Other examples of ambiguous surface NPs are seen in the
following sentences:

1. He drank vodka and orange juice.

2. Bill and Jim earned $200 last week.
3. Mr. Adams and his wife have written several books.

With the meaning of separate actions, paraphrased by structures with

both . . . and (He drank both vodka and orange juice), the sentences are
derived from compound sentences. With the meaning of a combination,
paraphrased by structures with together (He drank vodka and orange
juice together), they derive from compound NPs.
In this chapter we have seen that it is possible to join two or more
sentences together; if this conjunction of sentences results in a repetition of
words, either the deletion or the Pro-form transformation applies. Some
compound NPs are derived through this process, but others are compound
in the deep structure.
90 chapter ten

exercises

A. For each of the following sentences give the deep structure and show which
transformations have been applied:
1. The figurines will remain in the living room, but the flowers will be moved
to the den.
2. The Smiths dropped by after work but didn’t stay long.
3. I couldn’t hear the speech, but Esther could.
4. We went to the lake and to the park.
5. Won’t Henry or someone be at home in the morning?
6. He didn’t write or telephone us.
7. She is pretty but very dumb.
8. The Smiths left early, and I did, too.
9. Bill arrived early, but he didn’t stay long.
10. We were sitting on the front row, and they were sitting there, too.
11. Bach and Handel sound alike to Homer.
12. Susan and Agnes met each other in Seattle.
B. Show how the ambiguity of the following sentences can be accounted for by their
deep structures:
1. Herb and Louise got divorced last week.
2. I like hot dogs and ice cream.
3. Patsy and Sam have two children.
4. She invited Pam and Frank.
5. A dog and a cat followed me yesterday.
C. Classify the following sentences as grammatical or ungrammatical and decide
what constraints on compounding account for the ungrammatical sentences:
1. The burglar raised the window carefully, and he listened to see whether
anyone had heard him, and then he entered the room.
2. He raised the window, but he thought someone might have heard him, but
then he entered the room.
3. Why did you quit your job, and when will you find a new one?
4. Why did you leave the door open, and close it.
5. He can sing well, and she can too.
6. He can, and she can sing well too.
7. He can sing well, or she can too.
8. I read this book and that book.
9. I read a book and the book.
10. I heard and saw you enter.
11. I heard or saw you enter.
12. I heard but saw you enter.
13. I didn’t know your name, and there’s a fly on my nose.
14. He turned in the report, and he turned in his tracks.
15. She is tall but graceful.
 chapter eleven
RELATIVE CONSTRUCTIONS

The following sentences cannot be accounted for by the rules we have given
so far:

a. The boat that he is building is large.

b. Janice picked the flowers which are in the vase.

c. The man whom you met yesterday has painted the garage.

These sentences seem to contain NPs that consist of more elements than Det,
N, and PI; they are not compounds like Jane and Mary. First of all, we should
make certain that the italicized elements really are NPs. Here are three
kinds of evidence we can use:

1. Pronouns may substitute for them:

a. The boat that he is building is large.

It is large.

b. Janice picked the flowers which are in the vase.

Janice picked them.

c. The man whom you met yesterday has painted the garage.
He has painted the garage.

2. When they occur in subject position, tense and Aux1 move before them
to produce questions:

a. Is the boat that he is building large?

b. Has the man whom you met yesterday painted the garage?

91
92 chapter eleven

3. Under the passive transformation, they fill the positions of NPs:

b. The flowers which are in the vase were picked by Jane.

c. The garage has been painted by the man whom you met yesterday.

Although additional evidence could readily be found, this is enough to

establish the italicized groups of words as NPs.
We can easily break these NPs into three or'four elements each:

Det N PI unknown element

the boat that he is building
the flower PI which are in the vase
the man whom you met yesterday

The unknown elements look like transformed sentences with underlying

structures like the following:

a. he is building the boat

b. the flowers are in the vase

c. you met the man yesterday

We say that these sentences have been embedded into the noun phrases.
The italicized NPs in the deep structure have been moved to the
beginning of the embedded sentence by the relative transformation:

1. The boat he is building the boat is large

T-rel The boat the boat he is building is large
2. Jane picked the flowers the flowers are in the vase
T-rel Jane picked the flowers the flowers are in the vase
3. The man you met the man yesterday has painted
the garage
T-rel The man the man you met yesterday has painted
the garage

If the NP is already at the beginning of the sentence, as in the second

example, there is no change in word order; we say, nevertheless, that the
relative transformation has been applied.
As we saw in the last chapter, repeated words are normally not per¬
mitted in a sentence. We either delete one of them, or we substitute a Pro
 relative constructions 93

form for it. Notice how the following structures are transformed by T-rel
and T-pro:

1. Deep: The cup you are washing the cup is cracked

T-rel: The cup the cup you are washing is cracked
T-pro: The cup that you are washing is cracked

2. Deep: The car he is riding in the car is safe

T-rel: The car the car he is riding in is safe
T-pro: The car which he is riding in is safe

We call who, which, and that relative pronouns and the clauses they
introduce relative clauses.
To account for relative clauses, we must expand our rewrite rule for
the NP:

(Det) N (PI)
NP NP + S
NPn

In the preceding chapters we have seen many NPs generated by the choices
of (Det) N (PI) and of NPn; here is a diagram illustrating the other choice:
The embedded sentence you saw the
S
man will become the relative clause
whom you saw, and the entire sen¬
tence will be The man whom you
saw waved. One constraint placed
on this structure is that the em¬
bedded sentence must contain an
NP identical to the one preceding the man you saw the man waved
the sentence, such as the man in
this case. The deep structure The man (you saw the duck) waved cannot
be transformed into a grammatical surface structure becuase of this con¬
straint.
If the NP has the feature [ -f- human], either who or that may be
selected:

1. He is the man who spoke.

2. He is the man that spoke.

3. *He is the man which spoke.

94 chapter eleven

If the NP has the feature [ — human], either that or which may be

selected:

1. This is the house that he bought.

2. This is the house which he bought.
3. *This is the house whom he bought. «.

» %
\

If the NP is the object of a preposition, the preposition may be shifted with

the NP or not: She is the girl who(m) I danced with; She is the girl with
whom I danced. If the preposition is shifted, whom must be used instead
of who, and that may not be selected:

1. The chair that I sat in was broken.

2. *The chair in that I sat was broken.
3. The chair in which I sat was broken.
4. She is the girl that I talked to.
5. *She is the girl to that I talked.
6. She is the girl to whom I talked.
7. *She is the girl to who I talked.

The relative pronoun may be derived from any NP in the deep

structure:

1. Subject the girls arrived late

who arrived late
2. Direct object we saw the girls
whom we saw
3. Object of a preposition they talked about the cars
about which they talked
4. Subjective complement she had become a witch
that she had become
5. Objective complement we called him a name
which we called him

Another kind of introductory word found in relative clauses is the

relative adverb:

1. This is the town where I was born.

2. We argued about the time when he left.
 relative constructions 95

These are easily accounted for by our grammar:

1. I was born in the town where I was born

2. he left at that time when he left

In Chapter Nine we said that the semantic component assigns a

meaning to the deep structure and that transformations do not affect this
meaning. Since He gave me a dollar and He gave a dollar to me share the
same lexical items (except to) and the two sentences mean the same thing,
we expect them to be derived from the same deep structure, as is indeed the
case. Now look at this pair of sentences:

1. The man whom you saw looks like Jim.

2. The man you saw looks like Jim.

Since these sentences mean the same thing, they must share a common deep
structure: The man you saw the man looks like Jim. By T-rel this becomes
The man the man you saw looks like Jim. If we apply T-pro, we get The
man whom you saw looks like Jim. In Chapter Ten we saw that there are
two possibilities for avoiding repetition: replacement by a Pro form and
deletion. Instead of applying T-pro, we may apply T-del and get The man
you saw looks like Jim. The difference in the two surface structures results
from a different selection of transformations. Here are some more pos¬
sibilities :

1. This is the coat (which) I bought.

2. The boat (that) I found had been abandoned.
3. I am frightened by the kind of person (that) he has become.
4. He is the man (whom) we talked with.
5. He is the man with whom we talked.
6. *He is the man with we talked.
7. Students who ask questions learn quickly.
8. *Students ask questions learn quickly.

As you can see from the last two sentences, if the NP derives from a subject
in its deep structure, it may not be deleted; the same restriction applies to
an object of a preposition if the preposition is shifted with the NP (sentences
5 and 6). In all other cases the NP is freely deletable. Even with deletion,
these are still called relative clauses.
Any time that we have two surface structures with shared vocabulary
and the same meaning, we expect them to have the same deep structure.
96 chapter eleven

These two sentences mean the same thing:

1. The woman who is waiting for John is his wife.

2. The woman waiting for John is his wife.

The second sentence has undergone a different deletion transformation.

This optional transformation deletes relative pronoun, tense, and be:

X -j- j which > -j- tense +be + T=>X+T

that

Here are some more examples:

1. The book (that is) lying on the floor belongs to me.

2. The man (who was) dragged through the streets was no criminal.
3. The chair (which is) behind the desk is black.
4. The apartment (that is) downstairs is for rent.
5. John, (who is) my brother, is visiting me.
6. I saw a man (who was) sound asleep.
7. The man (who is) to answer the question has arrived.

As with other deletions, the rule allows us to recover the deleted material for
each sentence. Transformations do not affect meaning, but they do alter
the form of the sentence. Without deletion each of the above sentences
contains a relative clause. If deletion occurs, the relative clause in sentence
1 becomes lying on the floor, a present participial phrase. Each of the
reduced structures is named by its first word: dragged through the streets
is a past participial phrase; behind the desk is a prepositional phrase;
downstairs is an uninflected word used as an adverbial of place; my brother
is a noun phrase used as an appositive; sound asleep is an adjective phrase;
to answer the question is an infinitive phrase. We speak of all these
structures collectively as adjectiuals, since they all modify nouns.
Now look at the following deep structure: Susan bought a car a car
was red. By the relative and Pro transformations this becomes Susan
bought a car which was red. We may stop at this point, since any further
transformations are optional. However, since the relative pronoun is followed
by be, let us make the optional relative deletion transformation: *Susan
bought a car red. Unlike the sentences in the earlier paragraphs, this one is
ungrammatical. Another transformation, however, will move the adjective
 relative constructions 97

red in front of car and give Susan bought a red car, which is grammatical
and means the same thing as Susan bought a car which was red. This
transformation, which we call the noun-modifier transformation (NM),
moves an inflected single-word modifier to the position immediately in front
of the noun:

X + N + NM + Y => X + NM + N + Y

This rule works for participles as well as for adjectives:

The boy who was yawning looked bored.

T-rel Del: The boy yawning looked bored.

T-NM: The yawning boy looked bored.

As you can see from this example, rule T-NM is optional for most participles,
although it is obligatory for adjectives. Many uninflected words do not
undergo the noun-modifier transformation:

1. The people here are friendly.

2. *The here people are friendly.

But some of them do:

1. The paragraph above is redundant.

2. The above paragraph is redundant.

Generally, adjectives with intensifies may undergo this transformation:

1. He was a pitcher (who was) very good.

2. He was a very good pitcher.

Normally indefinite pronouns (someone, no one, everyone, everybody,

etc.) block the noun-modifier transformation:

1. He found something (that was) unusual in the room.

2. *He found unusual something in the room.

In the rare sentence He found an unusual something in the room, some¬

thing is functioning as a noun, not as an indefinite pronoun; notice the
98 chapter eleven

determiner an. Also, a few noun phrases that have been translated or
borrowed from French do not permit this transformation: the devil
incarnate, court martial, etc. For these cases the entire phrase will be
entered in the lexicon as one unit.
Since a sentence may contain more than one NP, it may contain more
than one relative clause or structure derived from a relative clause:

The pretty girl told a story that was amusing to the children who were
listening attentively.

In this sentence three noun phrases have relative clauses embedded in them.
It is also possible to embed a relative clause inside another relative
clause, as this tree shows:

the woman owned the cat

the cat had kittens

Transformed this will yield the man who met the woman who owned the
cat that had kittens.
It would be theoretically possible for S to recur an infinite number of
times. Our rules have made S a recursive element, like the rules that
rewrite S and NP as Sn and NPn. A recursive element allows infinite re¬
production of itself. Let us look at a sentence that has only limited
repetition of S:

I met a man who had a son whose wife sold cookies that she had
baked in her kitchen that was fully equipped with electrical appliances
that were new.

As you can see from this sentence with only five relative clauses, a sentence
that is extremely long is not pleasing from a stylistic point of view, and the
reader or listener loses track of the meaning. But if we exclude such matters
of performance as style and memory limitation and consider only the system,
a sentence may be indefinitely long.
 relative constructions 99

Now consider this sentence: The little girl in the yard who steals
peaches is a brat. Our rule handles the NP easily:

the girl the girl is little

At times deletion can produce an ambiguous sentence in the surface

structure: I built the boat in the yard. In the yard can tell where I built the
boat, or it can tell which boat I built. The surface structure could be
derived from either of the following:

Nuc

I past build the boat in the yard

Nuc

NP

tense V"" JMR

NP^

I past build
zz
the boat the boat is in the yard

Surface structures are often structurally ambiguous; deep structures never

are.
100 chapter eleven

In this chapter we have considered several structures that result from

the relative transformation:

1. The relative clause

Deep structure: The speech you gave the speech was good
T-rel: The speech the speech you gave was good
T-pro: The speech that you gave was good

2. The relative clause with relative pronoun deleted

Deep structure: The speech you gave the speech was good
T-rel: The speech the speech you gave was good
T-del: The speech you gave was good

3. The reduced relative clause: participial phrase, prepositional phrase, etc.

Deep structure: The man the man was waiting for you disappeared
T-rel: The man the man was waiting for you disappeared
T-pro: The man who was waiting for you disappeared
T-rel Del: The man waiting for you disappeared

4. The adjective and other pre-nominal modifiers

Deep structure: The car the car is old drives well

T-rel: The car the car is old drives well
T-pro: The car that is old drives well
T-rel Del: The car old drives well
T-NM The old car drives well

These five transformations—Rel, Pro, Del, Rel Del, and NM—enable us to

produce a large number of structures.

exercises

A. In the following sentences classify each adjectival as to structure (relative clause,

present participial phrase, etc.) and trace the sentence from its deep structure
to the surface structure:

1. The actor who made the speech looked nervous.

2. He gave it to the usher seating the patrons.
3. I read the book you recommended.
4. He knows my cousin who lives in South Dakota.
5. The announcement read in assembly was a surprise.
6. He is the man we were talking about.
7. The problem bothering me now is minor.
8. The man at the door seemed worried.
9. The stack counted by Jane was wrong.
10. The yellow bird that was in the cage flew away.
 relative constructions 101

B. In this chapter as in the rest of the book, we have made much use of your knowledge
of what is grammatical in leading you into generalizations about the language.
Would this approach be effective if you were studying English as a foreign
language? Explain.
C. Most single-word modifiers in English precede the noun, whereas phrases and
clauses follow it. Is this true because of logic or convention ?
D. Through the use of recursion, we can make a sentence infinitely long. Explain
what limitations performance plays in sentence length. Do we have the same
performance limitations in writing as in speech ?
 chapter twelve
SENTENCES
AS NOUN PHRASES

In the last two chapters we have seen two different ways in which simple
sentences may be expanded to make larger, more complex sentences:
compounding and embedding. In compounding we add two or more Ss or
NPs together, but they are distinct from each other; that is, in the deep
structure we can always show on a tree where one S or NP ends and the
other begins. Embedding, on the other hand, makes an S a part of another
structure, such as an NP. In the deep structure underlying a relative
clause, the S is one of the constituents of an NP. Through the use of em¬
bedding and compounding we can make sentences of any length or complexity
that we choose.
Now examine the following sentences:

1. We know who made the announcement.

2. What he said was wrong.

In these sentences who made the announcement and what he said must be
NPs, for we can substitute pronouns for them:

la. We know it.

2a. It was wrong.

It is obvious that these NPs are sentences derived by the familiar WH

transformation:

lb. NP-WH made the announcement who made the announcement

2b. he said NP-WH what he said

102
 sentences as noun phrases 103

Since our past rewrite rule for the NP will not permit these structures, let
us revise it:

f (Det) N (PI)
NP + S

Instead of the other three choices, we may now select S:

✓

Nuc

By the WH transformation this becomes

what he said was wrong

With the WH transformation we can derive other sentences:

1. We know she did NP-WH

We know what she did
2. We know she went Adv-m-WH
We know how she went
3. We know she went Adv-r-WH
We know why she went
4. We know she went Adv-t-WH
We know when she went
5. We know she went Adv-p-WH
We know where she went
6. We know she went with NP-WH
We know who she went with
104 chapter twelve

This NP can function almost any way a single-word nominal can:

1. subject: Whatever he said was good.

2. direct object: We know who opened the package.
3. object of a preposition: They talked about what she did.
4. objective complement: She made him what he is today.
5. indirect object: He will give whoever finished first a prize.
6. subjective complement: The caller was whorh' we had expected.

The choice of WH words is the same for adverbials as it is for the WH

transformation that produces questions. Only two possibilities exist for NPs,
however: who for [ -f- human] and what for [ — human]. Any of the
WH words except why may be followed by ever: whoever, whatever, etc.
Now look at the sentence We know that she went. That does not
replace anything in the deep structure; she went is complete by itself. The
word that has no meaning; rather, it is merely added to the nucleus as-a
sentence modifier (cf. The book that you bought is missing, in which that
means “the book” and is the direct object of bought). The deep structure
for We know that she went is as follows:

Nuc

tense V NP

we present know she went

The WH transformation cannot apply, since no element in the deep

structure has WH attached to it. We can apply the that-insertion
transformation, however, to a sentence that is embedded as a noun phrase:

NP

that she went

This gives We know that she went.

 sentences as noun phrases 105

Any NP may be written as S, as the following tree shows with a

subject NP:
The that -insertion transformation will
yield That he arrived so early was
surprising.
There is another structure that
means the same thing: For him to
arrive so early was surprising. The
transformation that produces for him
to arrive so early does not merely
add a subordinator to an otherwise
unchanged nucleus, as does the one
that attaches that. Previously we have gained insight into other structures
by writing underlying structures above surface structures and deciding which
transformational processes have been used: addition, deletion, rearrange¬
ment, substitution. Let us try this with the new structure:

Deep structure: he past arrive so early

Surface structure: for him to arrive so early

The surface structure has two words added: for and to. Instead of arrived
(past + arrive), it has arrive; hence, tense has been deleted. We call this
transformation the infinitive transformation and for him to arrive so early
an infinitive phrase.
There is another structure that is similar to that he arrived so early and
for him to arrive so early: his arriving so early, as in the sentence His
arriving so early was surprising. We call this structure a gerund phrase.
Let us follow the same process as we did with the infinitive phrase to see how
it is formed:

Deep structure: he past arrive so early

Surface structure: he + poss ing arrive so early

We say that his is a case of the pronoun he plus the possessive morpheme;
similarly, John’s is John -f poss, their is they -f poss, etc. As in the infinitive
transformation, tense has been deleted and something has been added; this
time it is poss and ing.
We have seen three possible transformations for he arrived so early
and other sentences embedded as NPs:

1. that-insertion: That he arrived so early was surprising.

2. infinitive: For him to arrive so early was surprising.
3. gerund: His arriving so early was surprising.
106 chapter twelve

Other structures such as His early arrival was surprising could no doubt be
derived by similar processes, but we will limit ourselves to these three.
In the sentence The idea that he would leave soon disturbed Sarah,
that has obviously been added by the that-insertion transformation. Since
that he would leave soon is not a relative clause, it cannot be derived from the
S in the rule NP —> NP + S. A final revision of the NP rule is required:

(Det) N (PI) (S) N

NP + S
NP
NPW

Th e idea that he would leave soon is derived from NP —> (Det) N (PI) (S)
as follows:

Compare this with the relative clause in The idea that he mentioned
disturbed Sarah:

This new source will account for the following structures:

la. The thought that I would be on time pleased me.

2a. The realization that he couldn’t remember her name embarrassed Joe.

Instead of these surface structures, we could have the following:

lb. The thought pleased me that I would be on time.

2b. The realization embarrassed Joe that he couldn’t remember her name.
 sentences as noun phrases 107

Since these two sentences mean the same thing as their counterparts above
and since they contain the same lexical items, they must have common deep
structures. We say that sentences lb and 2b have undergone an additional
transformation, the extraposition transformation. This transformation
moves an embedded sentence such as that I would be on time to the end of
the sentence. We may illustrate extraposition by means of a tree. After
that insertion we have the following intermediate structure:

Nuc

The extraposition transformation moves the embedded sentence to the end:

Nuc

Now look at the following pairs of sentences:

3a. That he is leaving is unfortunate,

b. It is unfortunate that he is leaving.
4a. That you are alone seems sad.
b. It seems sad that you are alone.

These sentences seem to share the same relationships as those in the preceding
paragraph did. Sentences 3a and 3b share the same deep structure; the
extraposition transformation has been applied to 3b, but not to 3a. The
same relationship is true for 4a and 4b.
108 chapter twelve

After that insertion applies, we have the following intermediate

structure:

Nuc

Extraposition rearranges the structure as follows:

Nuc

But *ls unfortunate that he is leaving is not a grammatical sentence.

English insists upon something functioning as an NP at the beginning of
most sentences, except in such obvious cases as imperatives. We, therefore,
apply the it-insertion transformation:

Nuc

It has no lexical meaning and cannot, therefore, be in the deep structure.

It is merely a filler, like do.
Often infinitive phrases may also undergo the extraposition trans¬
formation :

Deep structure: he arrived so early was surprising

T-inf: for him to arrive so early was surprising
T-extrap: was surprising for him to arrive so early
T-it: it was surprising for him to arrive so early
 sentences as noun phrases 109

Sometimes relative clauses may undergo extraposition. Hence, The idea

that he suggested was exciting may become The idea was exciting that he
suggested.
In discussing compound sentences and relative clauses, we said that
repeated words are normally deleted or replaced by Pro forms. Deletion
applies also to infinitive and gerund phrases:

Deep structure: John’s ambition is John becomes president

✓

T-inf: John’s ambition is for John to become president

T-del: John’s ambition is to become president

Deep structure: Jane’s worst fault is Jane arrives late

T-gerund: Jane’s worst fault is Jane’s arriving late
T-del: Jane’s worst fault is arriving late

In addition to repeated words, indefinite pronouns may also be deleted:

For anyone to like him is impossible becomes To like him is impos¬
sible.
Since structures resulting from the that, infinitive, and gerund trans¬
formations are dominated by the NP node, we expect them to behave as
other NPs do. The structures we have considered so far do so, as the passive
transformation shows:

la. The judges ignored what he said,

b. What he said was ignored by the judges.

2a. Everyone recognized that he was telling the truth,

b. That he was telling the truth was recognized by everyone.

Now look at these sentences:

3a. Fred tends to trust everyone,

b. *To trust everyone is tended by Fred.
4a. He needs to stop the car.
b. *To stop the car is needed by him.

To trust everyone and to stop the car must be derived from Fred trusts
everyone and he stops the car by the infinitive and deletion transformations,
but they cannot be NPs or the sentences could undergo the passive trans¬
formation. Fred tends to trust everyone must come from a deep structure
no chapter twelve

like this:

Nuc

NP VP

N Aux MV

tense

Fred present te

S is not classified as an NP. The rewrite rule for the MV must be expanded
for this type of structure:

MV
be (NP 1
AP
{ place J
V (NP) (S)

We may select V rather than be and only S after V, since NP is in parentheses.

This expansion of the MV rule will also account for an embedded S
following an NP:

Nuc

NP VP

past ask John John opened the package

Deep structure: We asked John John opened the package

T-inf: We asked John for John to open the package
T-del: We asked John to open the package
 sentences as noun phrases III

This origin of to open the package correctly shows that it is not part of the
NP containing John. Hence, the passive transformation moves only John
to the beginning of the sentence: John was asked by us to open the package.
If to open the package were part of the NP that contains John, the passive
transformation would move the entire NP: *John to open the package was
asked by us. This kind of embedding is directly linked to verb features,
since only certain verbs permit it: ask, persuade, caution, etc.
Now examine the following structure:
✓

_ Nuc_

NP

tense V NP 'S

we present consider her she is lucky

Deep structure: We consider her she is lucky

T-inf: We consider her for her to be lucky
T-del: We consider her to be lucky

There is another kind of deletion which we encountered with the relative

transformation. This is the transformation that deletes relative pronoun,
tense, and be in a structure such as who was waiting for you to produce
waiting for you. This transformation can be extended to sentences that have
undergone the infinitive transformation, so that the following derivation
is possible:

Deep structure: We consider her she is lucky

T-inf: We consider her for her to be lucky
T-del: We consider her lucky

Since We consider her to be lucky and We consider her lucky mean the
same thing, this transformation seems to be correct.
The same procedures will produce such structures as She seems to be
happy and She seems happy.

Deep structure: She seems she is happy

T-inf: She seems for her to be happy
T-del: She seems to be happy
112 chapter twelve

Or by the other deletion transformation:

Deep structure: She seems she is happy

T-inf: She seems for her to be happy
T-del: She seems happy

With this chapter we are ending our consideration of syntax. We have

seen that a small number of phrase-structure rules provide the deep structures
for the sentences of English. Some of these rules include recursive elements:

1. Compounding: S —> Sn

NP -> NPn

2. Verb Phrase Embedding: MV —> V (NP) (S)

3. Noun Phrase Embedding: NP —► (Det) N (PI) (S)

NP —> S

NP -> NP + S

Lexical items from the dictionary are attached to the deep structure, and
the semantic component assigns a meaning to it. This deep structure next
undergoes one or more transformations, which perform any of four processes:
addition, deletion, substitution, or rearrangement. The result is a surface
structure. All that is needed now is a section of the grammar that tells us
how to pronounce the surface structure. This is the phonological component,
which will be the subject of Part Four of this book.

exercises

A. Give the deep structure from which each of the following sentences was derived
and show which transformations have been applied:
1. We didn’t understand why you disagreed with us.
2. We admired his telling the truth.
3. Bill found a rare coin and sold it.
4. The suggestion that you made was interesting, but the proposal that we
should leave wasn’t.
5. It is good that you memorized their address.
6. Ann wanted to drive the car.
7. It is hard for me to remember names.
8. We persuaded Frank to answer the letter.
9. They considered your report to be very original.
10. Sue appears uneasy.
 sentences as noun phrases 113

B. The transformations we have presented are representative rather than exhaustive

of all of those necessary for English. With the knowledge you now have, you
should be able to provide rules to account for the structures in sans serif in the
following sentences:
1. We left after the game was over.
2. They left because they were bored.
3. I found the car where you had parked it.
4. After the game was over, we left.
5. Because they were bored, they left.

C. As a review of syntax, trace the'steps by which the following sentence was derived:
Why didn’t you tell those boys to be quiet? Draw a tree to show the deep
structure, show how lexical items are added, and perform the relevant trans¬
formations. How did you know which transformations to perform ?
 *

'

*
part four

PHONOLOGY
 chapter thirteen
TRANSCRIPTION *

We have studied in some detail the syntactic component of English trans¬

formational grammar. Although there still needs to be much further
research performed on English syntax, it will probably result in additions
to the grammar, rather than major revisions of the material that we have
presented. The syntactic component first generates a deep structure to
which the semantic component assigns meaning; this deep structure then
is converted into a surface structure by means of transformations. This
gives us a meaningful structure with its syntactic structure mapped out for
us; but it still is not in its final form, since we do not know how to pronounce
the sentence and since we still have such structures as post -j- drop, bird +
PI, etc. We now turn to the phonological component, which gives the
sentence its final form and which turns an abstraction into a physical
actualization.
Anyone trying to discuss the sounds of English has problems if he tries
to use the conventional English alphabet. One letter may represent different
sounds in various words: the letter c does not indicate the same sound in
core as it does in cent; ch does not sound the same in champion, champagne,
or choir; i does not indicate the same sound in mine as it does in machine.
Also, one sound may be represented by different letters in various words;
we can see this by examining the vowel sound in the following words: meat,
mete, meet, field, receive, Caesar, amoeba, people, key, quay, pity,
machine. Since there are only 26 letters in our alphabet but 35 distinct
sounds generally heard in English, the conventional alphabet is inadequate
for precise description.
Some people have tried to alleviate this difficulty by using such
descriptions as “broad a” or “flat a.” Unfortunately, there has never been
universal agreement as to which sounds are meant by these descriptions.
Other attempts have been made to identify a particular sound by the use of
key words. Problems arise when a sound is described as the “vowel heard
in root” or as the “final sound heard in father,” since there is much variation
I 18 chapter thirteen

in the pronunciation of these words. Even when words with relatively little
variation in pronunciation are used, it is vexatious to have to say something
like “the vowel heard in bit” if many sounds are to be described. A much
simpler system has been devised: a new alphabet. In this alphabet there is
a unique symbol for each of the 35 sounds heard in English. Conventional
spelling is ignored. The symbol /k/, for example, is used for the initial
consonant heard in both cat and keep, as well as the same sound heard in
account, back, chiropractor, acquaint, sacque, biscuit, and liquor (spelled
cc, ck, ch, cq, cque, cu, and qu respectively). Furthermore, anyone trained
in the use of this alphabet understands immediately the meaning of each
symbol, whether he speaks the variety of English common in Boston,
Houston, Kansas City, San Francisco, Melbourne, or London. We say
that there is a perfect “fit” between this alphabet and the sounds it rep¬
resents: each symbol invariably represents only one sound, and each sound
has only one symbol that represents it. The alphabet used in this book is
similar to the various American modifications that have been made of the
International Phonetic Alphabet. We avoid using the pronunciation
systems used by most dictionaries, since these systems vary from one dictionary
to another. The account of phonology given in this chapter is overly
simplified and in many ways outdated; however, the background provided
is useful for an understanding of the transformational account that begins
with Chapter Fourteen.
In this alphabet, which we call the phonemic alphabet, symbols
for individual phonemes are placed between slanted lines, as /p/, to distin¬
guish them from letters of the conventional alphabet, which are itali¬
cized or written in sans serif type. The symbol /p/ is used to indicate the
consonant sound heard in pea and pay. Similarly, each of the following
symbols is used to represent the first consonant heard in the words
following it:

/t/ to, top, tie, town, tot

/k/ kit, keep, cot, cab, come
/b/ boy, bib, big, bat, by
/d/ down, do, did, dog, Dan
/g/ get, goose, give, gun, guess

Notice that the sound heard at the beginning of gem and general although
spelled with a g has a different sound from the first of got or give; the sound
heard in gem and general will, therefore, not be represented by /g/, but by
another symbol that will be introduced later.
The symbols for the vowel sounds may present a slight problem, since
 transcription I 19

the English spelling system uses considerable variation in representing each

vowel. At the beginning of this chapter, we gave twelve different spellings
for the vowel sound heard in me and key. Regardless of how this sound is
spelled, in the phonemic alphabet it is always presented as /i/. With this
system me is written /mi/, key /ki/.
The vowel sound heard in did, hit, and kick has less variation in the
ways it is normally spelled; the symbol for it is /i/. The three words given
as examples in the preceding sentence are written as follows: /did/, /hit/,
and /kik/. ✓

exercises

A. With these symbols for six consonants and two vowels, you are now able to
transcribe the following words into phonemic notation:
1. pit 12. keep 23. bead 34. peep
2. Pete 13. kit 24. big 35. pip
3. pick 14. keyed 25. dip 36 it
4. peak 15. kid 26. deep 37. eat
5. peek 16. bip 27. Dick 38. eke
6. pig 17. beep 28. dig 39. pea
7. tip 18. bit 29. gig 40. tea
8. tick 19. beet 30. did 41. tee
9. teak 20. beat 31. deed 42. key
10. Teague 21. beak 32. bib 43. be
11. Kip 22. bid 33. kick 44. bee

B. Write the following in conventional English spelling:

1. /tik/ 6. /pik/ 11. /kid / 16. /pip/
2. /dip/ 7. /pit/ 12. /kid/ 17. /did /
3. M 8. /pig/ 13. /bit / 18. /did /
4. /it/ 9. /tip/ 14. /dip/ 19. /big /
5. /bip/ 10. /kip/ 15. / bib/ 20. /bid/

C. Write each italicized vowel sound in phonemic transcription; do not be con¬

cerned with the rest of each word:

1. me 4. clean 7. women 10. poople

2. behove 5. guild 8. Phoonix
3. bzzsy 6. party 9. Caesar

Most of the consonant symbols are easy to learn, since there is little
120 chapter thirteen

variation in the spelling of consonants in English and since the phonemic

symbols for 17 of the 24 consonant sounds are identical with the printed
letters that normally spell these sounds. The following phonemic symbols
represent the initial sounds heard in the words given after them:

/f/ fat, fig, fudge, phone, photo

/v/ vat, vain, vast, void, vice
/s/ sip, soup, sap, soap, seat
jzj zoo, zany, zip, zebra, xylophone
/s/ shall, sheep, shoe, sugar, sure

Only the last symbol will be new. A few seconds of experimentation should
convince you that this is indeed one sound, although we normally spell it
with two letters: sh. Notice the position of your tongue when you pronounce
/s/ as in sip and then /§/ as in ship. Then try pronouncing /s/ and immediately
after it /h/, the sound heard at the beginning of how. The tongue position
for fsf is different from that for /s/ or /h/.
We are now ready to present three new vowel sounds with key words
to show how they are pronounced:

/e/ late, may, steak, bait, rave

/e/ pet, set, sell, neck, wet
/ae/ cat, bad, rag, lack, fast

In transcribing a word into phonemic script, be careful not to let the

spelling confuse you. Pay special attention to the vowels. Notice that you
are recording only the sound that you hear, not the conventional spelling.
Notice the words be and bee, for example. Although the second of these
words is spelled with two e’s, there is no difference in pronunciation between
it and the first word, be; both are, therefore, transcribed /bi/. If we had
included the girl’s name Bea for transcription, it would have been recorded
the same: /bi/. Notice that capitalization is merely an orthographic device
and is not indicated in our transcription since it is not heard. If a person
says /bi/, it could be any of the three: be, bee, or Bea, as well as the name
given in English to the letter of the alphabet, b. Similarly, we do not
transcribe letters that are silent, as the g in gnat or the final e in Pete.
Since Pete and peat are pronounced alike, they should both be transcribed
/pit/. Finally, notice that only one /s/ sound occurs at the end of a word like
pass, which is transcribed /paes/.
 transcription 121

exercises

A. Transcribe the following words into phonemic notation:

1. fit 14. sit 27. Shick 40. fizz

2. feet 15. seat 28. sheik 41. sheaf
3. feat 16. sick 29. shake 42. chef
4. fate 17. seek 30. shack 43. steak
5. fact 18. sack ' 31. fish 44. said
6. fake 19. sake 32. bag 45. sad
7. fib 20. Sid 33. face 46. shave
8- egg 21. gave 34. says 47. zigzag
9. vet 22. ship 35. seize 48. vast
10. bad 23. sheep 36. cess 49. vest
11. vague 24. Shep 37. cease 50. best
12. sip 25. shape 38. see
13. seep 26. sheet 39. sea

Write the following words in conventional spelling:

1. /fek/ 6. /sep/ 11. /kaet/ 16. /pev/

2. /fib/ 7. /fis/ 12. /siz/ 17. /kev/
3. /ti/ 8. /fez/ 13. /sis/ 18. /pes/
4. /sip/ 9. /fes/ 14. /siv/ 19. /kaest/
5. /sip/ 10. /gaesp/ 15. /iz/ 20. /pef

Write the italicized vowel sounds in phonemic notation:

1. lather 4. hfffer 7. bz/ry 10. srtfd

2. veil 5. ffopard 8. gauge
3. sterzk 6. pain 9. m«y

The next group of new phonemic symbols is similar to the last group
in that only one unusual consonant symbol is introduced:

/r/ red, ring, rust

/1/ let, list, lamb
/m/ man, money, moon
/n/ now, noon, nurse
/p/ sing, rang, hung (the last sound in these words)

The sound /p/, although frequently spelled with the two letters ng, is in
reality only one sound, not a combination of /n/ and /g/. Try pronouncing
run and then rung and notice the tongue position at the end of each of these
words. After you have established this difference, pronounce run and add
a /g/ after the /n/. Now notice the pronunciation of finger and singer.
122 chapter thirteen

Finger has two consonant sounds in the middle: /gg/; singer has only one:
/g/. Try pronouncing finger with only the /g/ and singer with /g/ after the /g/.
Another potential source of difficulty is found in such words as sink
and pink. Pronounce each of these words slowly and notice the movements
of the tongue. As you pronounce sin and then sing, you will notice /n/ at
the end of sin and /g/ at the end of sing. For some speakers the vowel heard
in sing is that of seen, not sin. Now try adding the sound /k/ at the end of
sin and sing. The sound /k/ added to sin gives /sink/, which is not a recogniz¬
able English word; but if this sound is added to the end of sing, the result is
sink, which we write phonemically /sigk/. A comparison of pin /pin/,
ping /pig/, and pink /pigk/ reveals the same sound contrasts.
Below are three new vowel sounds with key words to show how they
are pronounced:

I a/ cut, shirt, putt, mother

/u/ full, put, could, hook
/u/ fool, Sue, true, moon

In mother there are two vowels. Although they sound alike to you, for
reasons that will be made clear in a later chapter, we will transcribe this
sound /a/ when it is in a stressed syllable and /o/ when it is unstressed:
/ mAiSur/.

exercises

Transcribe the following words

1. mean 14. sink 27. but 40. dirt
2. man 15. pink 28. boot 41. bank
3. tin 16. ping 29. cake 42. flood
4. tan 17. null 30. blood 43. two
5. ton 18. lose 31. book 44. move
6. run 19. loose 32. tongue 45. zest
7. rung 20. rude 33. put 46. sure
8. rug 21. fruit 34. putt 47. plaid
9. loom 22. took 35. pert 48. pled
10. lung 23. bury 36. group 49. laugh
11. seen 24. should 37. shove 50. does
12. sin 25. fool 38. pain
13. sing 26. full 39. pang

B. When a word contains more than one syllable, one of the syllables is pronounced
louder and more forcefully than the others. We call this phenomenon stress (some
books use the term accent). Although there are various ways of marking stress,
in this book we shall use an acute mark over the vowel of the syllable containing
 transcription 123

the strongest stress: /kaebin/. Transcribe the following words into phonemic
symbols and mark the syllable with the strongest stress; remember to distinguish
between stressed /a/ and unstressed /o/:
1. above 7. (to) permit 13. canoe 19. stinger
2. capital 8. (a) permit 14. pressure 20. malinger
3. capitol 9. (to) survey 15. padding 21. anger
4. singing 10. pruning 16. pudding 22. tinkle
5. (to) suspect 11. easily 17. precious 23. tingle
6. (a) suspect 12. shouldn’t 18. finger 24. parliament
C. Which vowel is found more often than any other in unstressed position ? Which
of the following consonant sounds is never found at the beginning of a word in
English: /m n p r 1/? Do all of them occur at the end?

D. Among speakers of English there is considerable variation in the pronunciation

of such words as root, coop, and hoof. On separate paper draw three columns:
(1) for words with /u/ as in food, (2) for words with /u/ as in took, (3) for words
with /a/ as in cut. Determine which vowel you use in each of the following
words and write the word in the appropriate column: hoof, tool, cook, root,
look, coop, kook, love, hood, mood, mud, Cooper, trooper, stood, move,
hoop, proof, roof, shove, soot.
E. Write the following in conventional English spelling:
1. Iml 6. /kid/ 11. /sprig/ 16. /sip/
2. /maed/ 7. /kid/ 12. /blum/ 17. /lipk/
3. /kud / 8. /sael/ 13. /plAm/ 18. /kipk/
4. /kud/ 9. /maes/ 14. /pruf/ 19. /skraem/
5. /kAd / 10. /ruf/ 15. /strAij/ 20. /stipk/

You should have no more trouble recognizing the remaining sounds

than you had with the preceding ones; but since several of them are rep¬
resented by symbols that will be new to you, we shall take them a little
more slowly than we did those in the other exercises.
The initial sound heard in hit, her, and house is represented as /h/
and the initial sound in wet, wear, and west as /w/. These sounds by them¬
selves will give you no trouble, but when the two come together you should be
careful. The transcription of hot /hat/ and watt /wat/ are as you would
probably expect. Now pay close attention to what. Ignoring the spelling
and noticing only the shape of your mouth as you pronounce the word
slowly several times, you will see that /h/ precedes /w/; hence, what is
written /hwat/. For some speakers of English, the /h/ has been lost so that
what and watt sound alike, as do witch and which, where and wear.
Notice these sounds carefully as you transcribe the following words, using
your usual pronunciation:
1. wet 4. where 7. heel 10. whiff
2. hail 5. wear 8. wheel
3. whale 6. hash 9. who
 124 chapter thirteen

The initial sound heard in yet, you, and young is represented as /y/.
Notice this sound at the beginning of you, and then pronounce the verb use
carefully. The only difference in pronunciation of these two words is that
use has a \z\ at the end of it. We write you as /yu/ and use as /yuz/. Do
not let spelling confuse you; notice that there is no difference in pronunciation
of use and the plural of you that is sometimes heard in the northeastern
part of the United States: youse. We also call the letter of the alphabet u
/yu/, not just /u/. The pronunciation /uz/ gives the word ooze. The three
sounds /y, w, h/ are called glides. Transcribe the following words, giving
your normal pronunciation:

1. few 4. yes 7. suit 10. use (N)

2. beauty 5. moo 8. do

3. booty 6. music 9. unit

The next sound is the consonant heard in the middle of pleasure,

measure, and seizure. This sound is represented as jzf. For many speakers
of English it is the final sound heard in beige, loge, rouge, and garage.
Transcribe the following words:

1. vision 4. racer 7. dilution 10. composure

2. fishing 5. razor 8. delusion

3. measure 6. mirage 9. composer

If you pronounce /t/ and /s/ together rapidly, you will have the initial
sound in child and chew. For English this combination functions as one
phoneme, which we write /c/. Chew is written /cu/ and chum /cAm/. In a
word like clutch be careful not to record a separate /t/, since this sound is
included in the /c/. This word is written /Mac/. A similar combination is
that of /d/ and Jz/, the sound heard both at the beginning and at the end of
judge, which we write j]j. Judge, then, is written /jaj/. Be careful not to
write /d/ after the vowel, since this sound is included in the /j/. The two
sounds /c/ and /j/ are called affricates. Transcribe the following words:

1. church 6. fudge 11. pleasure 16. virgin

2. chap 7. hedge 12. pledger 17. version

3. chafe 8. fetch 13. cheek 18. badge

4- gyp 9. Dutch 14. witch 19. chunk

5. Jeff 10. peach 15. which 20. age

 transcription 125

The next two consonants are both spelled th. That they are distinct
sounds can be clearly seen by comparing thy and thigh, either and ether,
mouth (V) and mouth (N). The sound heard in thy, either, and mouth (V)
is represented /3/; the sound heard in thigh, ether, and mouth (N) is
represented /0/. When you pronounce /S/, listen for the “buzzing” sound
and hold your fingers on your throat to feel the vibration. This sound and
vibration are absent in /0/. Transcribe the following words:

1. teeth (N) 6. the 11. other 16. north

2. teethe (V) 7. bath 12. thanks 17. northern

3. faith 8. bathe 13. thug 18. that

4. thatch 9. this 14. thief 19. thus

5. these 10. thistle 15. with 20. thesis

The remaining vowel sounds can be easily handled in a list that includes
key words:

/a/ father, hot, sock, top

/o/ hope, loaf, soak, grope

/d/ ought, caught, fought

Some people pronounce ought, caught, and fought with the sound /a/ and have
/o/ in only a few words if they have it at all. One test to tell whether you use
/o/ is to see whether caught and cot, taught and tot sound alike or not. If
they are different, then you have this sound; if they are alike, then you
probably do not have it.
Finally, there are three diphthongs:

/au/ house, mouth, plow, about

/ai/ mice, right, kite, rise

/oi/ boy, oil, coy, point

There is much variation among speakers of English in the actual pro¬

nunciation of these sounds. Each person actually gives them varying
pronunciations at different times. Whatever your exact pronunciation is,
use the symbols we have given.
126 chapter thirteen

exercises

A. Transcribe the following words:

1. pull 14. think 27. pot 40. box

2. pool 15. ache 28. pout 41. Mark’s
3. pole 16. daze 29. pit > 42. Marx
4. pile 17. days 30. pat 43. sox
5. pal 18. queen 31. put 44. mule
6. peel 19. quit 32. Pete 45. jazz
7. pale 20. shout 33. pet 46. white
8. pill 21. coil 34. shout 47. wrong
9. Paul 22. thigh 35. ouch 48. purr
10. extra 23. thy 36. joy 49. loath
11. exact 24. choice 37. rye 50. loathe
12. exaggerate 25. choose 38. work
13. excite 26. mutt 39. hurt

Write the names for the letters of the alphabet in phonemics: /e/, /bi/, /si/,

Write the following in conventional English spelling:

1. /sop/ 5. /si p/ 9. /bred/ 13. /brid/
2. /saep/ 6. /sup/ 10. /brud/ 14. /bred/
3. /sip/ 7. /sofa/ 11. /brasd/ 15. /brad/
4. /sAp/ 8. /wimin/ 12. /braid/ 16. /braud/

Transcribe the following words:

1. breathe 14. white 27. foundling 40. soiled

2. nature 15. surely 28. collision 41. fought
3. plowed 16. pleasure 29. exaggerate 42. sliced
4. residential 17. those 30. voice 43. creature
5. book 18. exact 31. quota 44. moth
6. thigh 19. loathes 32. mighty 45. explosion
7. should 20. applause 33. said 46. attacked
8. musician 21. wrong 34. sugar 47. pouch
9. champagne 22. appoints 35. suggest 48. pledger
10. leisure 23. badgers 36. of 49. pleasure
11. plumber 24. fox 37. wharf 50. joining
12. view 25. finger 38. English
13. announced 26, humiliation 39. myth
 chapter fourteen
PHONOLOGICAL FEATURES I'

Up to this point we have been concerned solely with learning a new alphabet
for the transcription of sounds. For the rules which we will be making in the
following chapters, we need more exact descriptions of the sounds. For our
purposes the most effective descriptions are those of the functioning of
various parts of the body in making sounds.
In English all speech sounds are made by modifying the flow of air as
it is exhaled from the lungs. Although we make certain gasps and clicking
sounds with air being inhaled, these are not part of our phonemic system.
There are languages that do have phonemes formed by modifying air that
is inhaled, but English does not. The first point at which we can modify the
outflowing stream of air is in the larynx, which is popularly called the Adam’s
apple. In the larynx are two muscular bands which are wide apart during
normal breathing. These bands, which we call the vocal cords, can be
brought together to block the flow of air completely. We often do this when
we are straining for some purpose, such as lifting a heavy load. Or we can
bring them close together so that a vibration is set up when the air flows
through them. This is done when we produce many of our speech sounds.
These sounds are said to be voiced. If the vocal cords are wide enough
apart so that no vibration occurs when air passes through them, the sound
is voiceless. There are two different tests you can make to determine
whether a sound is voiced or voiceless. You can hold your hand over your
throat and pronounce a prolonged /z/. This should then be contrasted with
a prolonged /s/. The other test is to pronounce each of these sounds while
holding your hands lightly over your ears. Either test should reveal a
definite vibration for /z/, but none for /s/; we, therefore, say that \z\ is voiced
and /s/ voiceless. We described nouns and verbs by saying that they had or
lacked certain features. We use a similar system with sounds; hence, \z\
has the feature [ + voice ] and /s/ has [ — voice ].

f This chapter is based on ideas found in Chapter Seven of The Sound Pattern of English by
Noam Chomsky and Morris Halle (Harper & Row, Publishers, Incorporated, 1968).

127
128 chapter fourteen

Before going further, use either of the tests we have given to determine
which of the following sounds have the feature [ + voice ] and which ones
have [ — voice ]:

1. /z/ 3. /v/ 5. /i/ 7. /S/ 9. /m/

2. /s/ 4. /f/ 6. /s/ 8. /0/ 10. /h/
* \
V

You should have found that the odd-numbered sounds are voiced and the
even-numbered ones voiceless. Now try pronouncing each of these sounds
again, but this time do not use your hand on your throat or over your ears.
Listen for the difference in sound and see if you can feel the slight tension
formed in the larynx for the voiced sounds.
The following sounds will be slightly harder, since they cannot be
pronounced in isolation but always have a vowel (which is voiced) accom¬
panying them. We normally call the letter of the alphabet p by the name
/pi/. Since vowels are always voiced, you will hear the vibration when you
say /pi/. If you notice carefully, however, you will find that voicing does not
begin on the consonant /p/, but rather starts only with the vowel /if. Contrast
this word with /hi/; here voicing begins on the fhf. Classify the following
sounds as [ T voice ] or [ — voice ]:

1. IpI 3. /t/ 5. /k/ 7. I cl

2. /b/ 4. /d/ 6. /g/ 8. HI

All even-numbered sounds in this exercise have the feature [ -f- voice ], all
odd-numbered ones [ — voice ].
For most sounds in English, the flow of air comes from the lungs and
is released through the mouth after being altered by variations in the shape
and positioning of the parts of the mouth. In the production of these
sounds, the velum (the soft palate) is raised so that air will not go into the
nasal passage, as it does in normal breathing. You can easily see the action
of the velum if you look in a mirror as you breathe with your mouth open
and then contrast this position with the one for the production of /a/. It is
also possible to lower the velum and at the same time block the oral cavity
with the lips or tongue, thereby forcing the air through the nasal passage.
This is the way /m, n, g/ are produced. You can tell that /m/ is a nasal
consonant by pronouncing it protractedly and using your fingers to pinch
the nose closed. As soon as you block the nose, the sound stops. Now try
pronouncing /a/ or /s/ and pinch your nostrils together. The sound continues.
For the production of /m/ the lips close the oral cavity; for /n/ the tip of the
tongue blocks it by touching the gum ridge behind the upper teeth; for /g/
the body of the tongue touches the velum. These three sounds have the
 phonological features I 129

feature [ + nasal]; all other sounds in English are [ — nasal ]. All nasal
sounds in English also have the feature [ + voice ].
For all non-nasal sounds the flow of air passes through the mouth and
is modified by the narrowing of two opposite parts (upper and lower lips,
lower lip and upper teeth, tongue and velum, etc.). The degree of narrowing
may be slight (cf. /e/ and /a/) or more extreme, sometimes to the point of
complete closure. When you pronounce a consonant sound, you make an
obstruction in the mouth. This obstruction may be complete closure, as in
the production of /b, p, d, t, g, k/. We call these sounds stops., since the flow
of air coming from the lungs is stopped at some point in the mouth and then
released: between the two lips for jbj and /p/, between the tongue tip and
gum ridge for /d/ and /t/, between the body of the tongue and velum for /g/
and /k/. If there is a complete closure of the oral cavity, as in the production
of the stops and the nasals, the sound is said to have the feature [ — contin¬
uant ]. Since /c/ and /j/ are composed in part by /t/ and /d/, they also have
this feature. All other sounds of English have the feature [ + continuant ],
since the flow of air is not completely cut off in the oral cavity, but rather
continues through it.
When you were in the first grade, you probably learned that the vowels
are “a, e, i, o, and u and sometimes w and y.” All other letters are consonants.
This is a listing of letters, not sounds. English obviously has many more
vowel sounds than this. Since this listing is rarely used for anything more
than a memorization exercise that is an end in itself, it has normally not
been challenged, even the use of w as a vowel. In the chapters that follow,
we will need to make a more precise distinction between vowels and
consonants, and we will concentrate on sounds rather than letters. We
define a consonant sound as one with a radical obstruction in the oral
cavity. The sounds we listed in the preceding paragraph as being [ — con¬
tinuant ] all meet this condition. But this radical obstruction does not have
to be so extreme as complete closure. The condition may also include a very
narrow opening, such as that used in the production of /f, v, 9, S, s, z, s, z/.
In the production of /r/, the tongue comes close enough to the palate to
meet this condition. In the production of /1 /, there is a radical obstruction
with the tip of the tongue against the gum ridge, although the flow of air is
not stopped as in the case of the noncontinuants but passes out over the sides
of the tongue. All sounds with a radical obstruction have the feature
+ consonantal], usually abbreviated [ + cons]. All others are
[ — cons ]. Notice that the glides /y, w, h/ do not have an obstruction in
their production and are, therefore, [ — cons].
According to our system, a sound that is [ — cons ] is not necessarily
a vowel. If you pronounce the following vowels in order, you will notice that
your mouth is open relatively wide for /ae/ and that it becomes gradually
narrower as you progress toward /i/: /ae, e, e, i, i/. You can see a similar
130 chapter fourteen

progression with /a, d, o, u, u/. A sound is said to be [ + voc ] (i.e., vocalic)

if it has a narrowing in the oral cavity no closer than that for /i/ and /u/.
To give the rather general term narrowing more precision, we generally
refer to this less extreme form of narrowing that does not exceed that for /i/
and /u/ as constriction. A narrowing that is extreme enough for the features
that are [ -f- cons ] is called an obstruction. Therefore, all sounds that have
a narrowing more severe than that of constriction are [ — voc]. Notice
that /y, w, h/ are [ — cons ] since there is no obstruction; similarly, they are
[ — voc ] since they are pronounced with too severe a narrowing for
constriction.
By using combinations of the features [ T cons ], [ — cons ], [ + voc ],
and [ — voc ], we can divide the sounds in English into four kinds:

1. [ -f- cons ] and [ — voc ]: /p, b, t, d, k, g, c, j, f, v, 0, 5, s, z, s, z, m, n, p/

2. [ — cons ] and [ + voc ]: /i, i, e, e, ae, a, a, u, u, o, o/
3. [ — cons ] and [ — voc ]: /y, w, h/
4. [ -J- cons ] and [ + voc ]: /r, 1 /

The sounds /r/ and /1 / are obviously [ -f- cons ] because of the obstruction
in the oral cavity. They are also classified as [ -f- voc ] because of certain
operations of the vocal cords that are too technical for the scope of this book.
Sometimes we add another feature, [ + sonorant] , which includes
all sounds that are [ + voc ] as well as the glides /y, w, h/ and the nasals
/m, n, r)/. All other sounds are [ — sonorant ]. Another name for [ — sono¬
rant ] is obstruent.
We have said that a sound has the feature [ -f- cons ] if there is an
obstruction in the oral cavity. If this obstruction is at the gum ridge or
farther forward, the sound has the feature [ + anterior ]. The following
[ + cons ] sounds have this feature: /p, b, t, d, s, z, f, v, 0, 5, m, n, 1/.
These are [ — anterior ]: /k, g, rj, s, z, c, j, r/. Since there is no obstruction
in the production of the vowels and glides, they are [ — anterior ].
Another feature of consonants is determined by the position of the
tongue. The part of the tongue that in a relaxed position lies opposite the
alveolum (the gum ridge) is called the blade; the blade includes the tip of
the tongue. If the blade of the tongue is raised above this relaxed position
on the floor of the mouth, the sound has the feature [ T coronal ]. The
sounds /t/ and /d/ are [ -j- coronal ], since the blade of the tongue touches
the alveolum; /p/ and /b/, however, are [ — coronal], since it is the lips
that form the closure and the blade of the tongue is not affected. The velar
sounds /k/ and /g/ are [ — coronal ], since it is the body of the tongue rather
than the blade that is raised. Vowels are, of course, [ — coronal ], since
their production does not involve the blade of the tongue.
 phonological features I 131

Some consonant sounds are produced with an obstruction that is not

complete closure but which is narrow enough to cause considerable noisiness
as the air is forced out a narrow opening. These sounds are [ — strident ]
and include the following: /f, v, s, z, s, z, c, j/.
We have classified sounds as [ + coronal ] or [ — coronal ] by the
position of the blade of the tongue. We assign features also by the position
of the body of the tongue, that part behind the blade. For normal breathing
the body, like the blade, is relaxed and on the floor of the mouth. For the
production of /e/, the blade of^the tongue does not move, but the body is
raised slightly. We use this position as a starting point in describing the
feature of tongue position.

1. A sound produced with the body of the tongue raised above the position
for /e/ is [ — high ] ; all others are [ — high ]. If you notice the position
of the body of your tongue as you pronounce the following sounds, you
will see that they are [ + high ]: /i, i, u, u, Y, w, k, g, rj, s, z, c, j/. All
other sounds in English are [ — high ].
2. A sound produced with the body of the tongue lower than the position
for /e/ is [ — low ] : /as, a, o, h/. All others are [ — low ] in English.
3. A sound produced with the body of the tongue farther back than the
position of /e/ is [+ back] : /u, u, o, o, a, a, w, k, g, r)/. All others are
— back ].
All vowels have the features [ — cons ] and [ + voc ]; these features
set them off from all other sounds. The three features high, low, and back
help us to distinguish them from one another:

i i e e ae o a u u O A
high + + - - - - —
+ + — —

low - - - - + + + — — — —

back _____ _j_ + + + + +

These features differentiate some of the vowels, but not all. Notice the
following groups:

1. ' + high ' /1 i/ 4. — high /o, A/

— low — low
— back -j- back

2. — high /e, e/ 5. ~ — high “ /a, o/

— low T low
— back + back

3. ' + high ‘ /u, u/ 6. — high

— low low
back — back
132 chapter fourteen

Some vowels are produced with more tension and effort than others.
The following are said to be [ + tense ]: /i, e, u, o, o/. The others are
[ — tense], or lax. Some dialects have a lax /o/, and others have tense
variants of /ae/ and /a/. The glides /y/ and /w/ are [ — tense]. The tense
vs. lax distinction is normally not extended to other sounds in English.

exercises

A. List the features that the following pairs of sounds share:

l- /Kg 1 6. /y, j/ 11. /i, u/ 16. /y, i/

2. /f, v/ 7. /s, z/ 12. /u, u/ 17. /k, a/

3. /m, b/ 8. /p, t/ 13. /a, o/ 18. /u, a/

4. /S, z/ 9. /d, c/ 14. /e, o/ 19. /ae, h/
5. /w, k/ 10. /sv, z/ 15. /o, a/ 20. /u, w/

List the ways in which the following pairs of sounds differ:

1. /m, nj 6. /f, 6/ 11. /v, s/ 16. /a, o/

2. /y, w/ 7. /I, n/ 12. /i, i/ 17. /ae, e/
3. /r, 1/ 8. /k, g/ 13. /1, u/ 18. /u, w/
4. /p, t/ 9. /c, 5/ 14. /r, as/ 19. /e, m/
5. /m, b/ 10. /d, g/ 15. /S, e/ 20. /s, z/

C. Our grammar contains three components: syntax, semantics, and phonology.

In which of these will performance differ the most from competence? Name
several distorting factors on performance.
 chapter fifteen
PHONOLOGICAL FEATURES II
y

As you saw in the last chapter, the production of speech involves the inter¬
action of many parts of the vocal apparatus, the most prominent being the
vocal cords, the velum, and the tongue. The features give an account of
these actions. You are now able to understand that symbols such as /p, g, a/
are not indivisible units but rather abbreviations for feature complexes.
Instead of writing [ + cons, — voc, — continuant, -j- anterior, — coronal,
etc. ], we can write /p/ and let this symbol stand for this complex of features.
In the last chapter we discussed the thirteen features that are most
relevant for a study of English. These features are part of a listing of thirty
or forty that underlie all human languages. To give you some idea of the
other features, we will use two examples that can be readily observed.
When you pronounce /u, o, w/, you round your lips. Contrast these sounds
with /i, e, y/, for which there is no lip rounding. We could have given a
feature [ 4 round ], but we did not because in English all sounds that are
-

-f round ] are also [ + back ]. The feature would not have been of any
real value in showing differences or similarities among sounds. But for some
languages, such as French, German, and Swedish, there are vowels that are
[ — back ] and [ -f- round ]; for these languages rounding is a significant
feature. As a second example, hold your hand in front of your mouth as you
pronounce /paen/, /taen/, and /kaen/. You should feel a puff of air, which we
call aspiration, accompanying /p, t, k/ in these words. This puff of air is
missing if /s/ precedes the consonant: /spaen/, /staen/, and /skaen/. Since
aspiration is predictable at the beginning of a word after /p, t, k/ if /s/ does
not precede, the feature [ + aspirated ] is not significant for English. For
some other languages, such as Hindi, aspiration is significant, since it alone
can distinguish one sound from another.
The feature system can be applied to all languages. When we are
speaking of universal phonology as opposed to the phonological system of a
particular language, it is not adequate to list features merely plus or minus;
instead, they are given numbers on a scale: 1, 2, 3, etc. For example, we

133
134 chapter fifteen

will want to show that when aspiration is present, /p/ is more strongly
aspirated than /k/ or that a sound that is [ —j— back ] may be farther back than
another [ T back ] sound. In working with universal phonology, then, we
will use all thirty or forty features and use numbers on a scale rather than
plus or minus. In working with the phonology of a specific language,
however, we will limit ourselves to the features that are significant for that
language and use only plus and minus values.
Here are the features for the vowels of English as we presented them
in the last chapter:

1 i e 8 ae u u o A a 0
cons
voc + + + + + + + + + + +
sonorant + + + + + + + + + + +
continuant + + + + + + + + + + +
nasal
anterior
coronal
high + + — — —
+ + — — — —

low — — — —
T — — — —
+ +
back — — — — —
+ + + + + +
tense + —
+ — —
+ —
T~ — —
+
voice + + + + + + + + + + +
strident — —

Our matrix is redundant, since for English if we know that a segment is

— cons ] and [ + voc ], we can predict certain other features: [ + sono-
rant, + continuant, — nasal, — anterior, — coronal, + voice, — strident].
We can make our matrix much more economical in space if we are allowed
to omit these features, but first we need a general rule that will reinsert them
when they are needed:

+ voc
— cons
+ sonorant
-j- continuant
-f- voc
— nasal
— cons
— anterior
— coronal
-f- voice
— strident

This is the form that redundancy rules are stated in. The rule says that
we rewrite all vowels with the features [ -f sonorant, -f- continuant, etc. ].
 phonological features II 135

The rule allows us to add these features to any vowel when we wish to
include a full phonetic representation, including features redundant for
English. Some of these features, such as [ + continuant, — anterior,
— coronal, — strident], are universal; others, such as [ — nasal], apply
only to English. Universal and language-specific rules should be kept
distinct and in separate grammars. Because of space limitations, we have
combined them.
Even with these redundancies removed, our matrix will still contain
redundant information. All segments that are [ + high ] are necessarily
[ — low ] because of the shape of the human tongue, and those that are
[ -f- low ] are [ — high]. Two rules will permit us to make further
simplifications:

+ high
[ + high ]
— low
+ low
[ -f low ]
— high

By leaving out the information provided for by the redundancy rules, we

retain the following:

1 i e 8 ae u u o A a 0
cons
voc + + + + + + + + + "T +
high + + — —
+ + — —

low — —
+ — —
+ +
back — — — — —
+ + + + + +
tense + —
+ — —
+ —
T — —
+

omitting redundant phonological features, we are following the same

process that we used with lexical features. A complete lexicon would contain
a redundancy rule such as the following for nouns:

+ animate
T human ] T concrete
T human

Future developments in transformational grammar will probably provide

many more parallels between different components of the grammar.
Now let us review the other sounds and see what redundancies we can
omit:
136 chapter fifteen

£ 1 1 + 1 1 + 1 1 +
^ 1 1 + I 1 + 1 i + 1 +
JG | 1 + 111 + i 1 1 +
- + +1 4-1 ++ 1 1 i + 1 +
+ +
^ + + + +
* + + +
g + + +
>N
+
>C/3 _|_
+ +
N +
+
W + + +
*0 -f +
o + +
> + + +
^ + + + +

>U ~f +
bo + + ++
^ + +
T3 + +
- +

n +
+
£
3 Jh ti c
.2 <3 C c3
• rH S-H c <u (U 5-h
C/3
o C
4-> o3 V o r4
u
CJ ^3 O
o o o
C/3 5h
o bO £
a
• i-H
•£ C
o > O
o3
a o
O -M O
£ 4 ° 43 > -m in
 phonological features II 137

The redundancies here are less" obvious than those for the vocalic
segments. Closer inspection will nevertheless reveal them. First of all, we
gave rules stating that segments that are [ + high ] are automatically
[ — low ] and those that are [ + low ] are [ — high ]. Now look at the
segments that are [ + nasal ]. All of these segments are also [ — continuant,
— low, + voice, — strident, -f sonorant, + cons, — voc ] in English. A
rule will permit us to omit this information:

-j- cons ✓
— voc
+ nasal
+ nasal ] — continuant
— low
T voice
— strident
+sonorant

Notice that this is a rule specifically for English, not a rule for all languages
as is the one that states that [ -j- high ] is also [ — low ]. Next, notice that
/r/ and /l/ are distinctively identified by the combination [ + cons ] and
+ voc ]; all that is needed for them is one further feature to distinguish
them from each other: [ -fi anterior ] or [ — anterior ]. Similarly, segments
that are [ — cons ] and [ — voc ] are [ — nasal, — anterior, — coronal,
— strident, + sonorant]. For English all segments that are [ + cons,
— voc, — nasal ] are [ — low ] and [ — sonorant ]. These restrictions and
several others will be stated in rules like the ones we have given. Omitting
redundancies, we have the following matrix:
V V

P b t d k g c J f v e 5 s z s z
continuant + + + 4-
nasal — — — — — —

anterior + + + + — — — .—
+ + + + + + — —

coronal I -f-
—
- + + — —
+ + — — + + + +
voice —
+ - + —
+ —
+ —
+ —
+ —
4- —
+
strident + + + + — —
+ + 4- +

rn n 0 r 1 y w h
cons + + — — —

voc + + — — —

nasal + + +
anterior + + — — +
coronal — +
high + + —

back —
+

Features that are not redundant are called distinctive. The distinctive
features of a sound are those needed to identify it unambiguously from other
138 chapter fifteen

sounds. Two sounds are different if they differ by at least one feature. For
example, /p/ and /b/ share all except one feature: /p/ is [ — voice ] and /b/
is [ -f- voice ]; hence, only voice is distinctive in separating these sounds.
Similarly, /t/ and /p/ differ only in that /t/ is [ -f- coronal ] and /p/ is
— coronal ]. Voicing is distinctive for all nonnasal consonants in English;
for nasals it is not.
The feature system reveals similarities among sounds more interesting
than their differences. Features can be used to''group sounds into classes.
For example, only /r/ and /l/ have both [ + cons] and [ -f- voc]. These
two features set them off from all other sounds in what we call a natural class
of sounds. We need fewer features to specify a natural class than we do to
specify any member of the class. To specify either /r/ or /l/ by itself, we need
three features: [ -j- cons, -f- voc, — anterior ] and [ -f- cons, + voc,
+ anterior], respectively.
Contrast this natural class with another pair of sounds: /d/ and /v/.
These sounds have much in common: [ -f- cons, — voc, — nasal, + anterior,
— high, -j- voice ]. But this is not sufficient to set them off from /b/, /6/,
and /z/, which share the same features. The only way we can talk about
/d/ and /v/ together is to list all of the features for each; they are not a natural
class. Notice that a mere listing of the symbols does not show the difference
between the pairs /d, v/ and /r, 1/; the feature system reveals the major
differences.
We find motivation for natural classes outside of a discussion of features.
As we will see in Chapter Eighteen, nouns ending in a sound with the features
[ + strident, + coronal ] (e.g., church, dish, judge, etc.) take the plural
/iz/. Verbs with the features [ — continuant, — nasal, -f- anterior,
-f- coronal ] in their last segment take the past tense /id/. There is also
historical motivation for natural classes. During the fifteenth century,
English segments with the features [ V voc, — cons, -f- tense ] underwent
a major change in tongue height. Natural classes defined by distinctive
features provide meaningful, linguistically significant generalizations about
the sound system.
Phonological entries in a lexicon should be given as nonredundant
matrices rather than as alphabetic abbreviations. Instead of listing mail as
/mel/, we would list the following:

+ nasal — cons + cons

-f- anterior -j- voc + voc
— coronal — high + anterior
— low
— back
+ tense

With nonredundant matrices we are not losing the other features: /l/
is still [ — nasal], /m/ is still [ -j-voice], etc. The matrix found in the
 phonological features II 139

lexicon will look like 1 below; redundancy rules will convert it into a fully
specified form like 2 before any natural classes can be defined or any
phonological rules, such as plural formation, can be applied:

1 2
cons +
voc —

continuant —

nasal ' + +
anterior + +
coronal — —

high —■

low —

back —

voice +
strident —

sonorant +

The redundancy rules we have considered so far apply to any phoneme

in English, regardless of which other phonemes surround it. English also
has limitations on the sequences of phonemes that may occur. For example,
there are no words in English that begin with consonant clusters such as
/pt, fk, lz/, nor is this just an accidental omission that might soon be filled
with new words. English permits only a few combinations of consonants
at the beginning of a word. If a word begins with two segments that are
both [ + cons ] and [ — voc ], only five possibilities exist: /sp, st, sk, sm, sn/.
We can, therefore, state the following redundancy rule for /s/, using the
symbol ^ outside the matrix to indicate initial position:

+ cons
T anterior + cons
-f- cons /~v -v-v n

# [ + cons ] _
-j- coronal — voc
V Ub
— voice — continuant
-T strident

In the position for /s/, all features except [ + cons ] have become neutral¬
ized; that is, they are not needed to distinguish /s/ from other sounds.
Another rule in English says that of the nasals, only /m/ and /n/ may occur
at the beginning of a word; this nasal must be followed by a vowel:

+ nasal + voc
# [ + nasal ] [ ] #
T anterior — cons

There are many constraints on the sequences of sounds that are permitted
in English. A complete lexicon would contain a full listing of redundancy
rules, both those that are context free and those that are contextually
140 chapter fifteen

determined. The lexical entries would be free of all redundancies and would
contain only the minimum number of features needed to identify morphemes
unambiguously. The few words which do not observe these constraints,
such as sphere, svelte, and tsetse, would be treated as exceptions; they
would have a comment that certain redundancy rules are not to be applied.
Rules will block all impossible words; they will not block any that are
possible, even though they may not exist at present. Hence, rules will block
/fnaet/ but not /spaed/, although neither is currently an English word.
A complete set of redundancy rules for English will do more than
conserve space in the lexicon. It will state all the significant generalizations
about the phonological patterns in the language and will define all possible
morphemes, including potential ones that have not yet been formed.
We have observed that some phonological constraints are universal
and others are peculiar to English. The study of universals in phonology
has led some linguists to a different treatment of features. Some feature
combinations, such as [ T high, -j- low ], and some sequences of segments,
such as /wgb/, are inherently impossible to pronounce; others are impossible
to perceive. Also, there are other feature combinations and segment
sequences that are possible but very difficult to make or perceive. Some
feature combinations and sequences are inherently more “natural” than
others. All this can be embodied in universal rules applying to all languages.
If any lexical item in a language obeys these rules, it is unmarked; if it disobeys
these universal constraints, it is marked (i.e., it will take on the opposite
specification from what the universal rule would predict). Since most
feature specifications in most segments in most lexical items are predictable
by general, universal rules, most feature specifications will be unmarked.
This procedure will distinguish language-specific elements from universal
elements in a particular language. Research on which feature combinations
are natural and which unnatural is just in its beginning stages, but the
future holds much promise.

exercises

If the sounds grouped together constitute a natural class, give the features that
distinguish them; if they do not constitute a natural class, do nothing with them:
1. /m, n, T)l 9. /k,g, c,j, s, z,r)/
2. Ip, b, m, f, v/ 10. /r, 1, h, y, w/
3. /s, z, s, z, f, v, c, j / 11. /i, i, u, u/
4. /y, w, h/ 12. /u, u, o, a, a, o/
5. /p, b, t, d, r, s/ 13. /i, e, u, o/
6- /k, g, r), w/ 14. /i, e, u, o, o/
7. /©, s, c, y/ 15. /i, i, e, e, ae/
8. 11, d, c, j, 0, 8/
 chapter sixteen
SYNTAX AND STRESSt

In English as in many other languages, certain syllables are pronounced

with more force than others. In a word one syllable will have the strongest
force; in a phrase a syllable of one word will be the strongest. This extra
force on a syllable is commonly called accent, a term which is often used for
other meanings as well: pitch, strangeness of speech, etc. To avoid confusion
with these other meanings, linguists usually use the word stress to name this
added force that is given to a syllable or a word.
Lexical entries for most words will not include stress, since English
stress patterns are normally predictable by rules such as those we will present
in this chapter. Only words which do not follow these rules will have their
stress patterns marked in the lexicon. This practice parallels that for noun
plurals and past-tense forms of verbs, in which only exceptions such as men
and caught are given in the lexicon.
Earlier, we spoke of certain vowels as having the feature [ + tense ]
and others as [ — tense ]. This distinction of tense and lax is important for
determining the placement of stress. Notice the position of the primary
stress in the following words:

1 2

surrender relate
hinder cajole
polish repeat
whistle escape
astonish maintain
relish survive

Each of the verbs in the first column has a lax vowel in the ultimate (i.e.,
last) syllable; stress occurs not on this syllable, but on the penultimate

f This chapter is based on ideas found in Chapter Three of The Sound Pattern of English by
Noam Chomsky and Morris Halle (Harper & Row, Publishers, Incorporated, 1968).

141
142 chapter sixteen

(next to last) syllable. The words in the second column all end in a syllable
containing a tense vowel or diphthong; this tense vowel receives primary
stress.
Now look at the adjectives in the two columns below:

1 2
fragile sublime \
regal secure
rigid polite
explicit severe
purple serene

Adjectives obviously are governed by the same rules as verbs are: primary
stress is placed on an ultimate tense vowel or diphthong; otherwise it goes
on the penultimate syllable.
We can begin formulating a rule for placing stress in a word. In
doing this we adopt certain notational conventions. A capital C will be our
abbreviation for consonant or [ + cons, — voc ]; Vw will abbreviate
vowel, [ — cons, + voc ] (we cannot use just V since that is our abbreviation
for verb). The symbol C2 will mean a cluster of at least two but not more
than five consonants. The lower number means the minimum, the upper the
maximum. If we do not want to state a maximum, we do not give an upper
number; hence, C2 means two or more consonants, no upper limit given.
If we do not want to set a minimum, we use the notation C0; this means no
consonant, one consonant, or any number. Similarly, means no consonant,
one consonant, or two consonants; C* means exactly one consonant,
two consonants, etc. Instead of the last two symbols for an exact number of
consonants, we normally omit the number and list C the number of times
it is used: C (exactly one consonant), GG (two consonants), CCC (three
consonants), etc. This convention saves a great deal of space. The symbol
VwCq is an abbreviation for the following:

Vw
VwG
VwCC
VwCCC

We can now state our rule for stress placement in verbs and adjectives as
follows:

Vw —> [ 1 stress ] /
+ tense
 syntax and stress 143

This says to add primary stress on a vowel when it has the feature [ -f tense ]
and is followed by zero or more consonants and is at the end of a segment
that is a verb or an adjective. For ultimate syllables containing lax vowels,
we need the following rule:

+ voc
Vw —> [ 1 stress ] /_C0 — cons
— tense
s

We will later need to add to these rules, but you should study how they
operate now. The underline indicates the position in which the vowel with
primary stress is found.
Before going further, apply these rules to the words listed below and
check the results with your knowledge of English pronunciation:

maroon handsome valid corrode

compute supreme prohibit imagine

Now look at some more verbs and adjectives:

defend complex
remark rotund
resolve abrupt
regard correct
subtract robust

According to our rules, all of these words should be stressed on the penulti¬
mate syllable, since they all have lax vowels in the ultimate syllables. If you
look carefully at these words, however, you will notice that all end in two
consonants. The adjective complex, which ends in /ks/, has primary stress
on the penultimate syllable for some speakers; for others it is regular. None
of the words we listed earlier ended in more than one consonant. We need
to revise our rule:

1. Vw —> [ 1 stress ] /_C0 [ — tense ] CJ ] VAdj

2. Vw —> [ 1 stress ] /
tense
C0] VAdj
3. Vw —> [ 1 stress ] / C2 ] yjA.dj

If the conditions of rule 1 are present, stress goes on the penultimate syllable;
in all other cases it goes on the ultimate syllable.
144 chapter sixteen

In our syntactic rules we were often able to condense information.

For example, we used braces to indicate a choice; if one element within the
braces was chosen, the others were automatically excluded. In phonological
rules we set an order on elements within braces. For

we select a if it meets our requirements and pass over b and c. If a does not
meet our requirements, we select b and omit a and c. If neither a nor b fits,
we select c. Our convention says that we must proceed from the top down¬
ward until we find a situation that applies; we then disregard other parts
of the rule even though other parts lower down may also apply. With this con¬
vention, we can compress our three rules listed above into two: (a) rule 1
and (b) all other cases:

+ voc
(C0 — cons CJ)
Vw —> [ 1 stress ] / __I _ - tense J ] VAdj
l G0
Study the notation carefully. If a verb or an adjective ending in not more
than one consonant has a lax vowel in the final syllable, primary stress is
placed on the penultimate syllable. If this condition is met, we are not
allowed to apply the second part of the rule. But if it is not met, primary
stress is placed on the final syllable. Notice that verbs and adjectives of one
syllable (run, eat, blue, sad, etc.) will never meet the conditions of the first
rule, since they do not have penultimate syllables. The second rule will
y y y

apply, then, assigning primary stress to their only syllables: run, eat, blue,
sad.
We have used no words with derivational suffixes, such as management
or realize; they will be treated later. Since most adverbs are formed by
derivation, we will not discuss them here. Now let us look at some nouns:

12 3
syllable amoeba surrender
elephant moment complexion
metropolis halitosis lantern

If we omit the final syllable, these words follow the same pattern as do
verbs and adjectives. In column 1, if we drop the final syllable of metropolis,
we have metropol- with a lax vowel in the last syllable. Rule 1, therefore,
 syntax and stress 145

assigns stress to the preceding syllable: metropol-. Similarly, after dropping

the last syllable in the words of column 2, we have a tense vowel in the new
ultimate syllable: halitos-; in column 3 there will be two consonants ending
the new final syllable: surrend-.
We can augment our rule to allow for stress placement in nouns that
have lax vowels in their final syllables:

+ voc
| C0 — cons i
Ml : Vw —> [ 1 stress ] / — tense
l C0 ii

{ + voc \
— cons C0 ]n a
/ — tense
V J b

This revision gives four possible patterns:

a i If the final syllable of a noun contains a lax vowel, this syllable is not
used in determining stress placement. If the remainder of the word
meets condition i, primary stress is placed on the penultimate syllable
of the remainder (the antepenultimate syllable of the entire word):
metropol(is).
a ii If the final syllable of a noun contains a lax vowel, this syllable is not
used in determining stress placement. If the remainder of the word ends
in a syllable with a tense vowel or two or more consonants at the end,
then stress is placed on the ultimate syllable of this remainder (the
penultimate syllable of the entire word): amoe(ba).
b i If the word is not a noun ending in a syllable with a lax vowel, the final
syllable must be considered. If the final syllable contains a lax vowel
and ends in no more than one consonant, stress is placed on the
penultimate syllable: explicit.
b ii If none of the above conditions is met, stress is placed on the final
syllable: defend, sublime.

These rules apply in order; we must use the first one that is applicable and
are not allowed to use any other.
Notice that a noun ending in a tense vowel does not meet condition a i
/ / /

or a ii; condition b ii is met for these words: balloon, marine, brocade, etc.
Words of three or more syllables are often subject to another rule that
adds a new stress. Anecdote receives primary stress on the last syllable since
it contains a tense vowel: anecdote (b ii). A new rule adds primary stress
146 chapter sixteen

back two syllables: anecdote. By a special convention of phonology, any

time we add a primary stress, all previously existing stresses are reduced by
one. By this rule the last syllable of anecdote will reduce to secondary
stress: anecdote. Here is the rule:

+ VOC
+ voc
M2: Vw —► [ 1 stress ] / C, Ci — cons C0 JxVAdj
— cons »
1 stress

Let us follow this rule through with meditate. By Ml, primary stress is
placed on the final tense syllable: meditate (b ii). By M2, primary stress is
added two syllables back, and the already existing stress is reduced by one:
meditate. Notice how these rules apply to the following words:

Farenheit isosceles pontificate

formaldehyde escalate corrugate
satellite chloroform pedigree

So far we have seen only how stress is added within words; let us now
see what happens when words are combined into larger structures. The
transformational rules can leave us with a surface structure like the following:

NP

Det Adj N

a pretty valentine

The words of this NP should be expressed in generalized phonological

representations, but such abstract representations would be only confusing
now. You can apply the stress placement rules with your knowledge of how
the sounds in these words are pronounced in their final forms.
We can express the structural information given in the above tree in
another form that is more convenient for placing stress:

[np a [xdj pretty ]Adj [N valentine ]N ]NP

This labeled bracketing, like the tree, says that pretty is an adjective,
valentine is a noun, and a pretty valentine is a noun phrase. We will
establish a convention that determiners, auxiliaries, and prepositions are left
unbracketed, since they are not stressed except for emphasis.
 syntax and stress 147

In its underlying form, the final vowel of pretty is lax. Rules Ml and
M2 apply as follows:

M1 [np a [acLJ pretty ]Adj [N valentine ]N ]NP

M2 [Np a Ldj pretty ]Adj [N valentine ]N ]NP

After we have applied all the rules in the cycle, we erase innermost brackets
and start over: '

[np a pretty valentine ]xp

Since this is a structure larger than a word, Ml and M2 do not apply.

We are now ready for a new rule:

M3: Vw —> [ 1 stress ] / [ X

1 stress

Condition 1: Y contains no occurrences of [ 1 stress ].

Condition 2: The bracketed group is a phrase larger than a lexical

category (N, V, Adj, etc.).

This rule tells us that within a bracketed phrase to add primary stress to the
last vowel that already has primary stress; this rule has no noticeable effect
on the vowel with primary stress, but it reduces all other stresses by one.
By M3, a pretty valentine will become a pretty valentme. Unstressed
vowels are not marked. We now have three degrees of stress, listed in
decreasing order of strength:

1. primary:
2. secondary: a
3. tertiary: ^

You probably do not hear the difference between secondary and tertiary
stress in a pretty valentine, although you should have no difficulty in
recognizing primary and weak (i.e., unstressed). Eventually, with enough
practice, you will hear all of these differences. Until then you can rely on
the rules for marking stress.
Notice that rule M3 applies to any bracketed group, not just the noun
phrase. It does not apply to lexical categories, such as N, V, etc. The VP
bought a pencil will receive stress as follows:

Ml [vp past [MV [v buy ]v [NP a [N pencil ] N JnP ImV ] VP

148 chapter sixteen

Rule M2 does not apply since its conditions are not met, and M3 applies
only to structures larger than the word. We now erase the innermost
brackets and start the cycle of rules over again:

[yp Past [mv [v buy ]v [NP a pencil ]NP ]MV ] VP

Notice that we are following a process similar to that which we used with
transformations in embedded structures. Elements within innermost
brackets are parallel to the most deeply embedded S. We now go through
our cycle of transformations again. Again Ml applies, but it does not alter
the stress pattern, since there is only one stressed vowel in each bracketed
structure. Since M2 and M3 do not apply, we again erase brackets and
start over on our cycle:

[vp past []y[y buy a pencil ] v ]vp

Ml and M2 are inapplicable now, but M3 assigns primary stress to the first
syllable of pencil and reduces all other stresses by one:

M3 [yP past []\jy buy a pencil jyjy ]yp

Since tense is not subject to stress, we erase the MV brackets and do not
apply the cycle further:

[yP past buy a pencil ]VP

If we have an entire sentence, such as Janice bought a pencil, we would

follow the same steps as we did for bought a pencil:

Is [np Janice ]NP [VP past buy a pencil ]VP ]s

Removing innermost brackets, we have

[g Janice past buy a pencil ]g

M3 [s Janice past buy a pencil ]s

After converting past buy to bought and erasing the final brackets, we
have Janice bought a pencil. Notice that rule M3 applies to any bracketed
group larger than a lexical category: AP, VP, NP, S, etc.
If a word is pronounced in isolation, it functions as a sentence:

[s [np cavalcade ]NP ]s

All [g [jvj'p cavalcade ]np J s
■^2 [s [NP cavalcade ]NP ]s
M3 [s cavalcade ]s

Hence, no word pronounced in isolation will ever have secondary stress.

 syntax and stress 149

For any new sentence that we create, we are automatically able to

assign the correct stress pattern; stress placement in the individual’s internal¬
ized grammar is, therefore, determined by rules. For new words we normally
have the same proficiency. Rules Ml, M2, and M3 are an elementary
attempt at accounting for this ability.

exercises ✓

A. Use rules Ml and M2 to determine the stress placement for the following words:

1. adapt 7. synopsis 13. achieve 19. hypotenuse

2. sturdy 8. antelope 14. arsenal 20. devote
3. analysis 9. arena 15. matador 21. neophyte
4. anecdote 10. complete 16. appendix 22. determine
5. remote 11. baritone 17. certain 23. horizon
6. elect 12. cancel 18. robust 24. mediate

B. Apply the cycle of transformations Ml, M2, and M3 as many times as needed to
give the stress placement on the following sentences:
1. Terry stayed sound asleep.
2. The cook baked a cake.
3. Gwendolyn has frightened Frankenstein.
4. An ugly woman drank the lethal gasoline.
5. Elizabeth spoke in a monotone.
 chapter seventeen
STRESS AND PITCH

In the last chapter we avoided stress placement in words like writer,

lovely, familiarize, and management. Each of these is derived from another
word: write, love, familiar, manage. We speak of a suffix (-er, -ly, -ize,
-ment) that is used to derive one word from another as a derivational
suffix. In most cases the derived word belongs to a different part of speech
from the original (write is a verb, writer a noun), but not always (both
host and hostess are nouns).
Most dictionaries list derived words as separate entries from the words
they are based on. For the kind of lexicon proposed by transformationalists,
this procedure is uneconomical and ignores significant generalizations about
English. For example, the agentive suffix -er (also spelled -or) means one
who performs an act, regardless of which verb it is attached to. Column 1
below gives the kind of listing normally found in dictionaries; column 2
shows the kind we are suggesting. Imagine each word with a full lexical
entry: definition, phonological form, and idiosyncratic features.

1 2
act act
actor -er
govern govern
governor play
play write
player
write
writer

This limited listing should be enough to suggest the tremendous saving that
will be achieved by not listing derived forms that are predictable by rule,
especially when we consider the thousands of verbs in the language to which
this suffix may be added. More important than the saving of space are the

150
 stress and pitch 151

generalizations that rules such as one for adding agentive suffixes will
provide. A native speaker of English who learns a new verb, such as the
possible flate, will understand the noun flator the first time he encounters it,
or he might introduce it himself without realizing that he is creating anything
new. The procedure of listing derivatives as separate entries does not show
how this process is possible; the rule we are suggesting does. Since an adequate
grammar should approach as closely as possible the native speaker’s
knowledge of his language, this is an important generalization. Our process
✓

shows the relationship between actor and other words with agentive suffixes;
these relationships are unquestionably accounted for by the individual’s
internalized grammar.
The procedure we are suggesting will be used with all derivational
suffixes, as well as with derivational prefixes (enjoy, devalue, unreasonable,
etc.). Prefixes and suffixes together are known as affixes. The lexical entry
for each affix will include full information on which words it may be added
to. This information whenever possible will be given in terms of features.
Because languages are systematic, we frequently see repetition of
patterns, such as the WH rule for questions, relative clauses, and noun
clauses. Whenever a generalization in one part of the grammar leads to
simplifications of other parts, we know that we have formulated a significant
generalization. In the last chapter we saw that our treatment of syntax
enables us to give a correct placement of stress patterns on words and
sentences. Such classifications as N, V, Adj, NP, VP, etc., enable us to give
stress placement by rule rather than by individual markings on each word
in the lexicon. This process is parallel to the individual’s knowledge of how
to place stresses on a sentence he is creating for the first time and his knowl¬
edge of how most new words he reads should be pronounced. The concept
of ordered transformational rules that operate in a cycle from the most deeply
embedded element upward is found again in the rules for stress.
This cyclic principle applies to derived words as well as to noun
phrases and sentences. Personal will be given on a tree diagram as

or with a labeled bracketing as

[Adj [n person ]N al ]Adj

152 chapter seventeen

We begin with the word within the innermost brackets, the noun person,
and use all applicable rules:

Ml [N person ]N

Since this is a noun, the final syllable with a lax vowel is passed over for
stress placement purposes; since there is only one other syllable, the stress
must fall on it. This word is too short for M2 to apply, so we erase the
innermost brackets and begin the cycle again for the adjective personal.
An addition to rule Ml would direct us to treat derivational adjectival
affixes with lax vowels the same as we do final syllables in nouns: disregard
them for stress purposes.

Ml [Adj personal ]Adj

After marking off the derivational affix, we have a lax vowel followed by
only one consonant in the last syllable of the remainder of the word:
person(al); hence, stress goes on the first syllable.
Look now at personality:

[n [Adi [n person ]N al ]Adj ity ]N

The first two cycles will apply as we have already stated:

Ml person (noun)

Ml personal (adjective)

Now we erase the next set of brackets and have the noun personality. The
last syllable is dropped from consideration : the penultimate contains a lax
vowel followed by a single consonant. The stress falls, then, on the ante¬
penultimate syllable, and the previously existing stress is weakened by one:

Ml personality

In a sentence further application of the rules will, of course, reduce the

secondary stress, as in Her personality is pleasing:

[s [NP her personality ]NP [VP is pleasing ]VP ]s

Erasing innermost brackets and applying rule M3, we get

[s her personality is pleasing ]s

 stress and pitch 153

You will notice that most determiners, auxiliaries, prepositions, and pro¬
nouns are not stressed.
For most words such as telegraph and telescope, if we remove the
prefix, we are left with an element that does not fit easily into any of our
lexical categories (N, V, Adj). Although we do have the words graph and
scope as nouns in English, these do not seem to be the stem to which tele-
has been added, although they are etymologically related. We will refer to
graph and scope in telegraph and telescope as stems. Stress is placed on
telescope as follows:

[n tele [stem scope Igtemlx

Ml [N tele [stem scope ]stcm]N
Ml [N telescope ]N
M2 [N telescope ]N

Words such as telegraph and telescope receive stress placement by the*same

cyclic principles as other words with prefixes or suffixes.
Our treatment of stress placement has been illustrative and far from
exhaustive. A complete discussion of stress in English would require a book
at least the length of this one, and there are still many unanswered questions.
Our purpose has been to show you how stress rules operate and to relate
them to the rest of the grammar.
Interacting with stress is pitch, the relative highness or lowness of a
sound. Each person has a normal pitch level from which he varies upward
or downward to provide variation. If he speaks in falsetto or some other
manner not usual for him, he sets up a new “normal” pitch level. In
speaking of pitch in language, we are talking about relative pitch—the
variation from the norm—not absolute pitch as in music. A person’s normal
level is given the number 2; variation below that level is given the number 1,
and anything above it the number 3. Most phonologists give a fourth level
for extra high pitch; this level is found only in rare cases, such as excitement,
and will not be included in the discussion that follows.
The rules for stress placement will assign the following patterns:

1. Bill went fishing.

2. Fran rode to town.
3. Ray saw Freddy.
4. Ray saw him.

For normal speech that does not give special emphasis to any word or show
contrast, you begin each of these sentences at the level that is normal for you.
You continue with this pitch level until you reach the word with primary
154 chapter seventeen

stress; then you rise to the third level and immediately afterward drop below
your normal level to level 1 and fade out. Pitch numbers are placed before
the syllables to which they apply:

1. 2Bill went 3fish1ing

2. 2Fran rode to 3town1 * %

\

3. 2Ray saw 3Fred1dy

4. 2Ray saw Freddy 3Ad1ams

5. 2Ray 3saw Jhim

In sentence 1 since fishing contains two syllables and is stressed on the first,
the change from the third to the first pitch level occurs between syllables and
causes no distortion of the word. In sentence 2, however, this change occurs
on the monosyllabic word town, producing a “drawling” effect. Contrast
this pronunciation of town with that in Fran rode to town yesterday.
Sentences 3 and 4 illustrate the combination of primary stress and third
level pitch on the last syllable of the sentence.
The yes/no question has a different pitch pattern from the affirmative
sentence:

1. 2Did she 3an3swer

2. 2Are you 3go3ing

3. 2Was she 3sure3

As with the affirmative sentence, you start off on the normal second pitch
level and rise to level 3 on the syllable with primary stress. Instead of
falling to level 1 and fading off, you fade off from level 3.
Notice that WH questions do not have the same pitch pattern as
yes/no questions:

1. 2Where are you 3go1ing

2. 2What did you 3see1

3. 2When’ll you be 3rea1dy

WH questions follow the /2 3 1/ pattern of the affirmative sentence. This

difference in pitch pattern gives us further support for distinguishing yes/no
and WH questions in addition to their differences in structure.
 stress and pitch 155

Deviations from the normal stress or pitch pattern came a change in the
meaning of the sentence:

1. John said not to go

2. John said not to go
3. John said not to go
4. 2John said not to 3go3
5. 2Where am I 3go3ing '
6. 2This cake is 3good3

The first three sentences have primary stress on words that would normally
not receive it, thereby giving special emphasis to certain words and additional
implied meaning. Sentences 4 and 5 are paraphrases of Did John say not
to go? and Did you ask where I am going? Sentence 6 is a typical kind of
sarcastic question. Since these deviations from normal stress and pitch
patterns cause changes in meaning, they must be indicated in the deep
structure with morphemes similar to C) so that the semantic component can
give the sentences their correct meaning.
The sentences we have illustrated so far have been short ones that are
normally uttered fairly rapidly with no pauses for breath. Examine a
longer sentence:

2Those old trunks in the 3at2tic 2contain many su^pris^s

Notice that there is a primary stress on attic with a corresponding rise in

pitch; the pitch level does not fall to 1, but returns to 2 and remains until
surprises, where it again rises with the primary stress and falls to level 1.
The pattern /2 3 2/ in those old trunks in the attic shows that the structure
is not complete; the /2 3 1/ pattern of contain many surprises does indicate
completion. We call a structure containing one primary stress accompanied
by a rise in pitch a phonological phrase. The sentence we illustrated
contains two phonological phrases; for short sentences, the entire sentence
is one phonological phrase. Phonological phrases are partially determined
by breathing habits, partially by syntactic patterns. In the sentence above,
the break between the two phrases comes between the NP and the VP, a
normal breaking point when the NP is long. Nonrestrictive modifiers occur
as phonological phrases separate from the rest of the sentence. Any long
sentence contains several phonological phrases.
There is probably some well-defined relationship between phonological
phrases and syntactic patterns. This is an area of phonology that has hardly
been examined in the past. Rules for phonological phrases and pitch patterns
will probably be placed in the phonological component of future grammars,
156 chapter seventeen

near rules for stress. All of the transformational rules must have been applied
and the pattern of the surface structure been defined before any of the
phonological rules can operate. Future research will probably tell us
something about junctures, or pauses. Current beliefs about junctures have
been severely challenged, but there has been nothing yet to replace them.

* *
V

exercises

A. Use rules Ml and M2 to mark the stress patterns in the following words:

1. inert 7. beautiful 13. universal 19. telephone

2. lurid 8. morose 14. incidental 20. telegraphic
3. habit 9. escape 15. designate 21. pomposity
4. habitual 10. pompadour 16. circumstance 22. traumatic
5. public 11. blanket 17. character 23. Palestine
6. serene 12. adjectival 18. momentous 24. equatorial

Use rules Ml, M2, and M3 to place stress on the following phrases and sentenc

1. an insipid circumstance
2. an academic teacher
3. a superb holiday
4. The chicken crossed the road.
5. The operator separated the customers.

C. Mark primary stress and pitch levels:

1. The hunters shot a rhinoceros.

2. Where did you see it?
3. Weren’t you frightened?
4. Those deplorable performances caused the theater to be closed.
5. Frank Ibsen, who won first prize last year, isn’t entering the contest.
6. My idea is agreeable?
7. Because we couldn’t see for the heavy fog, we stopped driving.
8. When did you get married?
9. Wasn’t the old woman on the back row about to fall asleep?
10. The troops seemed restless.

D. It would be possible to define a sentence as an utterance with one pitch pattern

that is /2 3 1/ or /2 3 3/. Discuss the effectiveness of this definition.

E. Give several reasons for making the division indicated by I rather than 2:
1. All the people in the audience / heard what he said.
2. All the people in the audience heard / what he said.
 chapter eighteen
PHONOLOGICAL RULES

In the last chapter we saw that for most words in English, stress is predictable
by rules. Since the lexicon lists only idiosyncratic features, stress will be
given only for those lexical entries that are exceptions. The stress-placement
rules permit us to see a system in our language that would be totally obscured
if we merely specified the stress for each word in the lexicon.
Another regular phonological process is the formation of plurals for
most English nouns. Since the same rules apply for both noun plurals and
for third-person singular present-tense verbs, we group the rules together:

-f- anterior
M4: present T V —> V T T coronal
+ strident

Rule M4 rewrites present plus a verb as that verb plus a segment that is
+ anterior, -f coronal, + strident]. You will notice that this describes
the natural class which includes /s/ and jzj. Another name for a class such
as this is archisegment. The lexical entries for have, be, do, and say, as
well as those for the modals, will block the application of this rule for them.
We next consider the morpheme we have indicated merely as PI:

-f- anterior
M5: PI -f- coronal
+ strident

Up to this point the grammar has kept the verb singular ending s (eats, sits,
etc.) separate from the plural ending because they are different morphemes
with different meanings. Now they can both be stated as the same archi¬
segment.
-T voc
-f- anterior
+ high T strident
M6:
— back // -f- coronal
T coronal
-f- strident
— tense

157
158 chapter eighteen

Rule M6 adds /i/ if the archisegment follows a segment that is [ —{— strident ]
and [ -f coronal ].
We now need a rule that will convert the archisegment specifically into
/s/ or /z/:

-j- anterior
M7: -j- coronal a voice ] / [ a voice ]
+ strident

Since rules M4 through M7 are numbered, they apply in order. Rule M7

now makes the feature [ voice ] on the archisegment agree with this feature
on the preceding segment. The symbol alpha means that the feature can
be plus or minus, but that all alphas in one rule will be consistently the same.
If the segment before the archisegment has the feature [ -f- voice ], then the
archisegment becomes /z/; if the preceding segment has [ — voice], then
the archisegment becomes /s/.
With these rules, the following derivations are possible. We are using
alphabetic abbreviations for all segments except those that are being added
so that you can concentrate on the application of the rules more easily.

baend + PI present -j- paes

-f- anterior " + anterior
M5 baend T coronal M4 paes + coronal
T strident _ + strident
M6 does not apply M5 does not apply
-f- anterior + voc
-f- coronal + high -j- anterior
M7 baend M6 paes
+ strident — back + coronal
+ voice — tense + strident
baendz + voc -j- anterior
+ high + coronal
M7 paes
— back -j- strident
— tense + voice
paesiz

Another pair of regular formations are past tense and past participle
in words such as drop, dropped, dropped. All other formations will be
listed as exceptions in the lexicon and will block the application of the
following rules:
— continuant
[ past ) — nasal
M8: + V +
[ en J -j- anterior
T coronal
 phonological rules 159

Rule M8 replaces past and en with an archisegment that includes /t/ and
/d/ and moves it after the verb.

+ VOC — continuant — continuant

+ high — nasal — nasal
M9 /
— back — anterior — anterior
— tense -f- coronal + coronal

Rule M9 adds /i/ if the archisegment follows /t/ or /d/.

— continuant
— nasal
M10 [ a voice ] / [ a voice ]
+ anterior
— coronal

Rule M10 makes the archisegment agree in voice with the segment that
precedes it.
The next rule can be simplified by the use of alphabetic abbreviations
for the complexes of features:

Mil: ing - V-^V + irj

This, of course, is the rule that gives the present participial symbol ing a
phonological form and attaches it after the verb. This rule is very simple
since there is no variation in its application.
We need a rule for vowels that have not received stress by the stress
rules Ml through M3:

+ VOC

— cons
Ml 2 9
— stress
— tense

Any vowel that has not received stress by our rules has the feature [ — stress ].
All unstressed lax vowels reduce to /a/, the sound of the vowel in the last
syllable of batter, Cuba, pleasure, etc. This is a generalized form of all
vowels that are lax and unstressed and was, therefore, not listed in our
inventory of vowels in earlier chapters. There is much dialectal variation
in the actual performance of this vowel, since it also may occur as a higher
160 chapter eighteen

vowel similar to /i/. More sophisticated rules would, of course, take care of
these variations.
So far we have said little about the way morphemes will be entered in
the lexicon except that they will be stated in features with all redundancies
omitted. We also said that derived forms such as actor and personality will
not be included, since they can be predicted by rules and the entries for
derivational affixes. We run into a problem on citation forms, since many
words vary in pronunciation according to the environment. For example,
we could enter photograph as /fotagraef/ and encounter no difficulties with
the word in such combinations as photograph equipment or in the derivative
photographic. When we try to derive photographer, however, we have a
problem:

citation form: fotograef

derivation: fotagrafor

We can easily arrive at /o/ in the unstressed vowels of the derivation by

rule M12, but there is no way we can predict /a/ if we have /a/ in the citation
form, since /a/ is the reduced form for all vowels. We will say that /a/ cannot
appear in any citation forms and that its only source is rule Ml2. Instead
of /fotagraef/ as the citation form, we will use /fotagraef/ and call it a deep
phonological form. Since deep syntactic structures are often radically
different from surface structures, it should not be surprising that deep
phonological forms are different from those on the surface.
The study of deep phonology has barely begun. Some linguists are
advancing convincing reasons for using deep forms that are very similar to
regular English spelling. These forms will be the same for all dialects of
English, all differences being accounted for by phonological rules. The
phonological rules, in turn, can be applied only after the structure of the
sentence, its basic syntax, has been established. Because so little is known
about deep phonology, we have limited our goal in this last chapter to giving
you some understanding of the role that rules play in transforming deep
phonological forms into surface forms.
We have studied the interrelatedness of syntax, semantics, and
phonology. It is the syntactic component that is central to transformational
grammar, since the other two are dependent on it. Only the syntactic
component can create new structures without limit. Semantics and phonol¬
ogy merely assign meaning and form to an abstract structure. The general
principles we apply in transforming deep syntactic structures into surface
structures hold true in the phonological component as well, and we suspect
that they will also prove applicable to the semantic component when research
in that field is more advanced.
 phonological rules 161

exercises

A. Apply rules M4-M7 to the following:

1. joke + PI 4. pig — PI 7. present + pitch 10. present + grab
2. judge + PI 5. wish — PI 8. present + laugh
3. bat + PI 6. present + rip 9. present + please
B. Apply rules M8-M11 to the following:
1. past + stop 5. past + close 9. ing + eat
2. en + stab 6. en — seed 10. past + open
3. en — test 7. past + marry
4. ing + go 8. past + wait
C. Why are the and a normally pronounced /Sa/ and /a/ instead of /Si/ and /e/?
What pronunciation do you give was in He was going?
D. Prepare rules to account for the phonological changes found in the following
words:
1. advocate, advocacy; pirate, piracy; complacent, complacency (/t/ and /s/)
2. soft, soften; haste, hasten; moist, moisten (/t/)
E. Show how a transformational grammar produces the sentence The little girls
cried. Apply all relevant rules in each of the components of the grammar.
 BIBLIOGRAPHY

This bibliography is restricted to those works which will be of greatest

interest to the reader of this book in giving fuller accounts of the grammar
and in presenting viewpoints different from those found in this book. Most
of the full-length works contain more detailed bibliographies.

Bach, Emmon, An Introduction to Transformational Grammars. New York:

Holt, Rinehart and Winston, 1964.
Bach, Emmon, and Robert T. Harms, eds., Universals in Linguistic Theory. New
York: Holt, Rinehart and Winston, 1968.
Bolinger, Dwight, “The Atomization of Meaning,” Language, XLI (1965),
555-73.
Chomsky, Noam, Aspects of the Theory of Syntax. Cambridge, Mass.: MIT
Press, 1965.
-, “Current Issues in Linguistic Theory,” in The Structure of Language,
ed. Jerry A. Fodor and Jerrold J. Katz. Englewood Cliffs, N. J.: Prentice-
Hall, Inc., 1964.
-, “Some General Properties of Phonological Rules,” Language,
XLI 11 (1967), 102-28.
-, “Some Methodological Remarks on Generative Grammar,” Word,
XVII (1961), 219-39.
-, Syntactic Structures. ’s-Gravenhage: Mouton and Co., 1957.
-, “A Transformational Approach to Syntax,” in Third Texas Con¬
ference on Problems of Linguistic Analysis in English, 1958, ed. Archibald A.
Hill. Austin: University of Texas Press, 1962.
Chomsky, Noam, and Morris Halle, The Sound Pattern of English. New York:
Harper and Row, 1968.
Fillmore, Charles J., “The Position of Embedding Transformations in a
Grammar,” Word, XIX (1963), 208-31.

162
 bibliography 163

Gleason, H. A., Jr., Linguistics and English Grammar. New York: Holt,
Rinehart and Winston, 1965.
Halle, Morris, “On the Bases of Phonology,” II Nuovo Cimento, XIII, Series X
(supplement) (1958), 494-517. Repr. in The Structure of Language, ed.
Jerry A. Fodor and Jerrold J. Katz. Englewood Cliffs, N.J.: Prentice-
Hall, Inc., 1964.
Halle, Morris, “Phonology in Generative Grammar,” Word, XVIII (1962),
54-72. Repr. in The Structure' of Language, ed. Jerry A. Fodor and Jerrold
J. Katz. Englewood Cliffs, N. J.: Prentice-Hall, Inc., 1964.
Harms, Robert T., Introduction to Phonological Theory. Englewood Cliffs, N.J.:
Prentice-Hall, Inc., 1968.
Jacobs, Roderick A., and Peter S. Rosenbaum, English Transformational
Grammar. Waltham, Mass.: Blaisdell Publishing Company, 1968.
Katz, Jerrold J., and Jerry A. Fodor, “The Structure of a Semantic Theory,”
Language XXXIX (1963), 170-210. Repr. in The Structure of Language, ed.
Jerry A. Fodor and Jerrold J. Katz. Englewood Cliffs, N. J.: Prentice-
Hall, Inc., 1964.
Katz, Jerrold J., and Paul M. Postal, An Integrated Theory of Linguistic
Descriptions. Cambridge, Mass.: MIT Press, 1964.
Klima, Edward S., “Negation in English,” in The Structure of Language, ed.
Jerry A. Fodor and Jerrold J. Katz. Englewood Cliffs, N. J.: Prentice-
Hall, Inc., 1964.
Langacker, Ronald W., Language and Its Structure. New York: Harcourt,
Brace, and World, Inc., 1968.
Langendoen, D. Terence, The Study of Syntax. New York: Holt, Rinehart
and Winston, Inc., 1969.
Lees, Robert B., The Grammar of English Nominalizations. Research Center in
Anthropology, Folklore, and Linguistics, Publication No. 12. Bloomington,
Indiana: Indiana University Press, 1960.
—*-, “Grammatical Analysis of the English Comparative Construction,”
Word, XVII (1961), 171-85.
Lees, Robert B., and Edward S. Klima, “Rules for English Pronominaliza-
tion,” Language, XXXIX (1963), 17-28.
Postal, Paul M., Aspects of Phonological Theory. New York: Harper and Row,
1968.
Reibel, David A., and Sanford A. Schane, Modern Studies in English. Engle¬
wood Cliffs, N.J.: Prentice-Hall, Inc., 1969.
Smith, Carlota S., “A Class of Complex Modifiers in English,” Language,
XXXVII (1961), 342-365.
-, “Determiners and Relative Clauses in a Generative Grammar of
English,” Language, XL (1964), 37-52.
164 bibliography

Stanley, Richard, “Redundancy Rules in Phonology,” Language, XLIII

(1967), 393-436.
Stockwell, Robert P., “The Place of Intonation in a Generative Grammar of
English,” Language, XXXVI (1960), 360-67.
Thomas, Owen, Transformational Grammar and the Teacher of English. New
York: Holt, Rinehart and Winston, Inc., 1965.
 INDEX

Abstract noun, 35 Conjunction addition transformation, 84,

Accent, 141 88
Adjectival, 96 Conjunction deletion transformation, 84,
Adjective, 15 88
Adjective phrase, 15 Consonantal, 129
Adverbial, 14, 57 Consonants, 129-30
Adverbial movement, 60-62 Constriction, 130
Affix, 151 Continuants, 129
Affricate, 124 Coronal, 130
Alpha, 158 Count noun, 37-38
Alphabet, 117-18
Alveolum, 130 Deep phonological form, 160
Animate noun, 34 Deep structure, 8, 44, 52, 59, 71-72, 79-81,
Anterior, 130 95
Archisegment, 157 Deletion, 31,60, 65, 70, 85, 96, 111
Arrow, 11 Derivational suffix, 150
Aspiration, 133 Determiner, 13, 57
Asterisk, 7 Determiner and noun restrictions, 36-38
Auxi, 45 Distinctive features, 137
Auxiliary, 14, 19-26 Do transformation, 48

Back, 131 En, 20-21,68, 158

Be, 15, 19 Extraposition, 107
Blade of the tongue, 130
Bloomfield, Leonard, 6 Features, lexical, 27-40
Body of the tongue, 131 Features, phonological, 127-40
Brackets, 50 Francis, W. Nelson, 6
Fries, Charles C., 6
Chomsky, Noam, 6, 127n, 141 n
Common noun, 36-37 Generative grammar, 6
Competence, 8, 79 Generative-transformational grammar, 6
Compound NP, 88 Gerund phrase, 105
Compound sentence, 83 Glides, 124
Concrete noun, 35 Grammar, 30, 75-77
Conjunction, co-ordinating, 83 Grammatical structure, 7, 76-77, 84

165
166 index

Halle, Morris, 127n, 141n Performance, 8, 79

Have, 20-21 Phoneme, 118
Hierarchical grouping, 10, 12 Phonemic alphabet, 118
High, 131 Phonological component, 29, 78
Hill, Archibald A., 6 Phonological phrase, 155
Human noun, 33-34, 57, 93-94, 104 Phrase structure, 13, 45, 78
Pitch, 153-56
lmp, 65 Place, adverbial of, 14, 57
Imperative, 64-65 Plural, 13, 15.7
Indirect object, 62-64 Possessive, 57, 105
Infinitive phrase, 105 Poutsma, H. A., 5
Infinitive transformation, 105 Present participle, 20, 159
lng, 19-20, 105, 159 Pro form, 86-87, 92-93
Intensifier, 15 Proper noun, 36-37
Intermediate structure, 55 P-terminal string, 28
International Phonetic Alphabet, 118 P-terminal symbol, 28
Intransitive verb, 32
It-insertion transformation, 108 Q, 52
Question transformations, 51-58
Jespersen, Otto, 5
Reason, adverbial of, 14, 57
Larynx, 127 Recursive element, 98
Lax vowel, 132 Redundancy, 134-37
Lexical features, 27-40 Relative adverb, 94
Lexicon, 29, 39-40 Relative clause, 93, 91-100
Linguist, 6 Relative pronoun, 93-94
Linguistics, 6 Relative transformation, 92
Low, 131 Round, 133
Rule, 11-12, 45, 63, 75
Main verb, 14 Rides, transformational
Manner, adverbial of, 14, 57 do, 48
Markedness, 140 imperative, 65
Modal, 22 indirect object, 63
negative, 50
Narrowing, 130 noun modifier, 97
Nasal sounds, 128-29 passive, 68
Natural class, 138 relative deletion, 96
Negative transformation, 43-50 WH, 57
Neutralization, 139 yes/no, 55
Node, 17
Nominal, 13 Semantic component, 29, 78
Noun modifier, 97 Sentence modifier, 11
Noun phrase, 11 Sequential constraints, 139-40
Nucleus, 11 Shall and will, 24-25
Should and would, 24-25
Obstruction, 130 Sonorant, 130
Obstruent, 130 Stageberg, Norman C., 6
Stem, 153
Parentheses, 11 Stops, 129
Passive, 67-72 Stress, 141-53
Past participle, 20-21, 68, 158 Strident, 131
 index 167

Structural grammar, 5-6, 79 Transformations (continued)

Structural linguistics, 5 passive, 67-72
Subject-verb restrictions, 33-36, 71 pro-form, 86-87, 92-93
Surface structure, 8, 44, 52, 59, 71-72, 79- relative, 92
81, 95 relative deletion, 96
Syntax, 77-78 that-insertion, 104
WH, 55-57, 102-103
Tense, 14, 22, 23-24, 157-58 yes/no, 51-55
Tense vowel, 132 Transitive and intransitive restrictions
Terminal string, 28 30-33
Terminal symbol, 28 Transitive verbs, 32
That-insertion, 104-5 Tree, 12
Time, adverbial of, 14, 57 Triangle, 13
Traditional grammar, 3-5, 8, 23-24
Transformation, 8, 29, 45, 51,59-60, 78
Ungrammatical structure, 7
Transformational-generative grammar, 6
Universal deep structure, 8, 81
Transformational grammar, 6-8, 27-30,
Usage, 7, 77
75-81
Transformations
Velum, 128
adverbial movement, 60-62
Verb phrase, 11
conjunction addition, 84, 88
Vocal cords, 127
conjunction deletion, 84, 88
Vocalic, 130
deletion, 70, 85, 95, 109
Voiced sound, 127-28
do, 48—49
Voiceless sound, 127-28
extraposition, 107
Vowels, 129-30
gerund, 105
imperative, 64-65
indirect object, 62-64 WH questions, 51, 55-57
infinitive, 105 Will and shall, 24-25
^-insertion, 108 Would and should, 24-25
negative, 43-50
noun modifier, 97 Yes/no questions, 51-55
 AN
INTRODUCTORY
TRANSFORMATIONAL
GRAMMAR
BRUCE L. LILES

During the past ten years the rapid progress of research in all phases
of linguistics has produced extensive developments and growth. This new
book concentrates on the discoveries in transformational syntax and
phonology. It introduces the reader to the many recent theories and
provides an intensive study of specific sentences.

AN INTRODUCTORY TRANSFORMATIONAL GRAMMAR is

restricted to the English language and is based entirely on post-1965
scholarship. It is the most up-to-date book of its kind, and covers phonology
in simplified, fully descriptive terminology.

An understanding of transformational grammar requires both theory and

application. This book fuses the two and contains a series of exercises
at the conclusion of each chapter. These permit the reader to work
immediately with the material that has been discussed and to explore
original ideas. The explanations illustrate that there are still
many aspects of the English language that are poorly understood.
Professor Liles provides a fascinating glimpse of the varied problems
and theories uncovered.

The syntax covered goes beyond the study of simple sentences to include
compounding, relative constructions, and other embedded sentences.
Lexical features of nouns and verbs are treated in detail.

PRENTICE-HALL, INC., Englewood Cliffs, New Jersey

13-502286-X