A

Grammar
An
Introduction
to Systemic
Grammar
James Muir
With the more widespread inclusion of
linguistics in English language courses, there
has been a growing need for a working
grammar between the elementary textbook and
the advanced study. The present book has been
written in response to requests from practising
teachers, undergraduates and college of
education students for an extended presentation
of the author's approach.
This is an essentially practical introduction to
English grammar, based on the model
developed by Professor M A K Halliday,
known as ‘Scale and Category" or (more
recently) 'Systemic' linguistics. The book's
coverage is both comprehensive and detailed:
Part 1 sketches an overall picture of the
grammar; Part 2 provides a comprehensive
description of ‘Surface Grammar' and Part 3
introduces the reader to ‘Deep Grammar’. The
author's writing is as clear as the book's
organisation; all technical terms are fully
explained, and notes have been kept to the
essential minimum.
James Muir writes from many years' experience
of teaching linguistics. He is currently Lecturer
in English at the Jordanhill College of
Education, Glasgow.

£2.20 net
A Modern Approach to English Grammar
A Modern Approach
to English Grammar
An Introduction to Systemic Grammar

James Muir

B. T. Batsford Ltd
London
 First published 1972
© James Muir 1972

Printed and bound in Great Britain by

Richard Clay (The Chaucer Press) Ltd,
Bungay, Suffolk
for the publishers
B. T. BATSFORD LTD
4 Fitzhardinge Street, London W1H OAH

ISBN O 7134 2008 1

£ i>ec i a. I
Cell.
PE
I \0(r
MS
To the memory of O.K.S.
Contents

Preface ix
1 General Survey 1
2 Surface Grammar 14
2.1 Morpheme and Word 14
2.1.1 Inflexion 18
2.2 Word and Group 25
2.2.1 Nominal Group 25
2.2.2 Verbal Group 41
2.2.3 Adverbial Group 49
2.3 Group and Clause 51
2.3.1 Complements in Clause Structure 57
2.3.2 Predicators in Clause Structure 59
2.3.3 Adjuncts in Clause Structure 63
2.4 Clause and Sentence 65
3 Deep Grammar 89
3.1 Systems at Clause Rank 93
3.1.1 Mood 93
3.1.2 Theme 97
3.1.3 Information 101
3.1.4 Transitivity 104
3.2 Systems at Group Rank 120
3.2.1 Nominal Group: Number, Case, Gender 120
3.2.2 Systems at d in Nominal Group Structure 128
3.2.3 Systems at Verbal Group: Finiteness, Modality,
Tense, Polarity, Aspect, Voice, Contrast, Focus
(of Contrast) 131
Notes 143

Bibliography 145

Index 147
Preface

This book is intended to present a description of English gram¬

mar which is based on a particular theory or ‘model’ of lin¬
guistics. This model has been known as ‘Scale and Category’
linguistics, or, more recently (and more appropriately), as
‘Systemic’ linguistics. Such theoretical foundation is discussed
in this book only where necessary, and no discussion of the
advantages and disadvantages of this particular model is
offered. For those interested in such questions, a highly technical
account of the theory will be found in the article ‘ Categories of
the Theory of Grammar’ by M. A. K. Halliday, in Word (Vol.
17, No. 3, 1961); a fuller and more up-to-date account will be
found in Systemic Linguistics by H. M. Berry, to be published
by B. T. Batsford Ltd. in 1973, which may, in fact, be considered
as a theoretical counterpart to this book and which (like this
book) is intended for undergraduates, training college students
and interested teachers.
This book makes no claim to originality in anything other
than in a few places. It provides a reasonably comprehensive
description of ‘Surface Grammar’ (Part 2) and a somewhat less
detailed description of ‘Deep Grammar’ (Part 3). The General
Survey (Part 1) is a summation of the other two parts and pro¬
vides a general (an ‘overall’) picture of the grammar, which can
be enlarged and deepened throughout Parts 2 and 3. Essentially,
 \

X PREFACE

this book is intended to provide a grammar which is neither an

elementary textbook nor an advanced linguistic dissertation.
There is a space between these extremes which needs to be filled,
both for those who will teach from elementary textbooks and
for those who will proceed to more advanced and diverse
linguistic study.
Since little is original, I obviously owe much to teachers and
colleagues over a number of years. It is perhaps appropriate to
mention here particularly the teaching and later the writings of
M. A. K. Halliday and J. M. Sinclair, on which much of this
book is based. More specific acknowledgement is made else¬
where. I also owe particular thanks, of a somewhat different
nature, to Mr. A. R. Thompson, editor of the Scottish Secondary
Schoolteachers’ Magazine, and Mr. J. McGrath of the Language
Sub-committee of the Glasgow Education Authority Committee
for English, for reading through an early draft and offering
helpful comment and advice.

JM
General Survey

It is important to be aware that language is always linearly mani¬

fested; that is, serial in presentation. In speaking or writing, items
follow each other in temporal or spatial succession. But the extra-
linguistic phenomena to which language refers may not be sequen¬
tial; the most commonly observed events are complex, although their
various aspects—the interaction of perhaps a number of com¬
ponents which constitutes the whole event—may be perceived simul¬
taneously.
Professor Quirk1 gives the simple example of a boy patting a dog,
and comments:

We could not readily assign an order to the boy, the patting, the
dog, the boy’s stoop, the dog’s tail. On the other hand, as soon
as you try to report what you have seen, you find that you not
only can, but must, assign an order to it and must break up your
impression into pieces of your own choosing and present them,
not simultaneously, but one by one.

Music and mathematics show us the advantages of simultaneous

over serial expression. Professor Quirk invites us to compare

with a linguistic form of this expression: ‘the cube root of a hundred
and twenty five divided by three’. As soon as we do this, however,
we become aware that there is, in fact, an order in the serial expression
of language. We must, in using the linguistic form of the expression,
know how to punctuate it (in writing) or how to modify our voice
2 GENERAL SURVEY

(in speaking) in order to ensure that it will be interpreted as

and not ^125 -f- 3 or even '^/T00 +

In other words, language—despite its necessarily linear presenta¬
tion—is patterned activity; it has structure. We must, in the above
example, know the linguistic grammar in order to interpret the
linguistic expression correctly, just as we must know the mathe¬
matical grammar in order to interpret the mathematical expression
correctly. A linguistic expression is not just the additive value of
words (or of any other unit), but the complex patternings formed by
the operation of units within each other. Let us consider two lin¬
guistic expressions and their possible patterning:

(i) twenty two month old babies

(a) twenty (two month old) babies
i.e. twenty babies who are two months old
(b) twenty two (month old) babies
i.e. twenty two babies who are one month old
(c) (twenty two month old) babies
i.e. babies who are twenty two months old

(ii) new car salesman

(a) new (car salesman)
i.e. a car salesman who is new
(b) new car (salesman)
i.e. a salesman who sells new cars

It is obvious that these expressions cannot be simply represented as

(i) twenty two year old babies; (ii) new car salesman. The completion
of Professor Quirk’s analogy aptly illustrates this:

A linguistic expression is less like

4 + 8 + 7 + 9 + 8 + 6 + 5-f-9 + 4
than like
[(4 + 2)3] + [3(5 + 7) + 6].

Native English speakers will agree that an expression such as

dog bites man
has three parts or elements. We do not say this because the expression
contains three words, but because we recognise a pattern. The same
pattern is evident in the expression
 GENERAL SURVEY 3
(the great big heavy dog) (bites) (the little thin man).
We do not say that this expression has ten elements of structure
because it has ten words, but rather that it also has three elements.
Language, to repeat, is patterned activity. This means that certain
regularities are exhibited over certain stretches of language activity.
The grammatical category set up to account for these ‘stretches of
language’ is the unit. A unit may be defined as any stretch of
language which exhibits a recognised pattern. The units of a language
are of varying extent, and may occur sequentially or be included
within each other. The patterns which these units carry take the form
of like events, and the category which accounts for likeness of events
is the structure. Each unit operates at an element of structure in the
unit next above, and the particular configuration of units operating
as elements of structure in a given example of a higher unit form the
structure of that unit. Thus the expression
1 2 3
(the great big heavy dog) (bites) (the little thin man)
is a single unit, the structure of which is formed by the elements of
structure realised by the three smaller units.
There are, however, restrictions on which units can operate at
which elements of structure in the next highest unit; it is not the
case that any unit can occur at any place in structure. In English,
not all members of the unit ‘phrase’ can operate at ‘predicator’
in clause structure, and those which can do so cannot operate else¬
where, say at ‘subject’. According to its potentiality of operation in
the structure of the unit next above, each unit is assigned a class name.
The two preceding paragraphs have been rather abstract, and
perhaps somewhat confusing. Their intent should become clearer
in the following pages. The important point to be made at this
juncture is that these three categories of unit, class and structure
(there is one more to be mentioned a little later) are theoretical
categories generally applicable to language and languages. In con¬
cerning ourselves henceforth with English, we shall be seeking to
discover, define and illustrate the descriptive instances of these
categories.
It may be advantageous to break into the description of English
by discussing the word word, which is a well-known and well-used
term for a particular grammatical unit. Traditionally it has been
regarded as a basic or central unit; whether or not it should be so
regarded (and there are weighty arguments on both sides) it is
4 GENERAL SURVEY

demonstrably not the lowest unit. Native speakers of English will

agree that if we examine the word unfriendliness we shall find it is
composed of four elements: un + friend + li + ness. These elements
of structure of the word are realised by members of the unit below the
word, and this unit is called the morpheme. A rough working defini¬
tion of the morpheme is that it is the smallest meaningful unit in
language; somewhat more accurately, that it is the smallest unit
pertinent to grammatical description. The morpheme is not, however,
defined by reference to its meaning; it is identified and defined by
the same formal criteria as are applied to other units. If we arrange
the units of English on a scale of rank, then the morpheme will be
the lowest unit.
A cursory consideration will reveal that there are two main classes
or types of morpheme: those which operate (or may operate)
without being in association with other morphemes; and those which
only occur in association with other morphemes. The former are called
free morphemes and the latter are called bound morphemes. Thus:
the word book consists of one free morpheme
the word books consists of two morphemes; the free morpheme
book and the bound morpheme s
the word bookish consists of two morphemes; the free morpheme
book and the bound morpheme ish
the word bookcase consists of two morphemes; both are free
morphemes and they form a dithematic compound word.
Bound morphemes may be inflexional, as s in books, or derivational,
as ish in bookish.
Morphemes make up words; words are made up by morphemes.
That is, morphemes operate as elements of structure in the next
highest unit on the rankscale, which is the word. The structure of any
given word will be the configuration of elements of structure realised
by morphemes. Each word contains one obligatory element of
structure, which is called the base element. Further elements of
structure of the word are optional and are dependent on the base;
these are called affixes, and further designated prefix, if preceding the
base, or postfix, if following the base. We can thus analyse un¬
friendliness as:
Unit: word
Elements of structure: prefix (un), base (friend), postfix (li-ness)
Structure: pbf
 GENERAL SURVEY 5

At the elements p (prefix) and / (postfix) operate bound morphemes;

at the element b (base) operates a free morpheme.
This consideration of morphemes has entailed looking downward
from the word and examining how it is made up of smaller con¬
stituents. We must now look upward and examine how the word
operates in the structure of the next highest unit on the rank-
scale.
It has already been observed that the value of a linguistic expression
is not just the additive value of the words, but that words enter into
groupings:
{the big dog) {bites) {the little man) {deeplyjin the arm)
The words form into, operate as elements of structure in, these
groups. It is convenient to call this next highest unit the group,
but the term phrase is also used widely in this meaning; we shall
use both terms interchangeably. There are three classes of group in
English: nominal group, verbal group and adverbial group; e.g.

nominal group: the great big dog; the little man; the boy; table
verbal group: bites; will bite; will be biting-, has bitten-, bit
adverbial group: deeply’, quickly-, fast', in the field-, in the arm.

The nominal group has one obligatory element of structure. This is

called the headword or head, as boy in the groups boy, the boy,
the big boy, the big boy with red hair. It is evident from these examples
that an element of structure may precede the head element: this is
called the modifier; and an element of structure may follow the head
element: this is called the qualifier. Thus:

boy is a nominal group with one element of structure, an h element,

giving the simple structure: h
the big boy is a nominal group with two elements of structure, an
m element {the big) and an h element {boy) giving the structure:
mh
the big boy with red hair is a nominal group with three elements of
structure, an m element {the big), an h element {boy) and a q
element {with red hair) giving the structure mhq.

At h in the nominal group operates the word-class substantive;

at m in the nominal group various word-classes operate, which may
for the present be designated pre-noun; it is not common for single
words to operate at q in the nominal group.
6 GENERAL SURVEY

The verbal group also has one obligatory element of structure;

this is called the lexical element, and in the verbal group this is
always the final element. The lexical element may be preceded by an
element which is called the auxiliar element, and/or by an element
which is called the negative element. Thus:
comes is a verbal group with only an / element of structure, giving
the structure: /
will come is a verbal group with two elements of structure, an a
element (will), and an / element (come), giving the structure: al
will not come is a verbal group with three elements of structure, an
a element (will), an n element (not) and an / element (come),
giving the structure: anl.
At / in the verbal group operates the word-class lexical verb; at a
in the verbal group operates the word-class auxiliary verb; at n in the
verbal group operates the word-class negator.
There are two distinct types of adverbial group in English. In the
first type there is again an obligatory element, called the apex. There
is an element which may precede this, which is called the temperer,
and there is an element which may follow the apex, and this is
called the limiter. Thus:
sweetly is an adverbial group with one element of structure, an a
element, giving the structure: a
very sweetly is an adverbial group with two elements of structure,
a t element (very) and an a (sweetly), giving the structure: ta
very sweetly indeed is an adverbial group with three elements of
structure, a t element (very), an a element (sweetly) and an /
element (indeed), giving the structure: tal.

At t in the adverbial group operates the word-class sub-modifier;

at a in the adverbial group operates the word-class adverb; single
word items are not characteristic of l in the adverbial group.
The second type of adverbial group has two elements of structure,
both of which must be present. These are called prepend element and
completive element. Thus:

on the hill is an adverbial group with two elements of structure,

ap element (on) and a c element (the hill), giving the structure: pc.

At the elementp operates the word-class preposition; at c operates

a complete nominal group.
 GENERAL SURVEY 7
Groups operate as elements of structure in the next highest unit,
which is called the clause. Thus the expression, the big dog bit the
little man on the leg is one clause with four constituent groups:
12 3 4
(the big dog) (bit) (the little man) (on the leg).
To these four elements of structure of the clause we give the tradi¬
tional names subject, predicator, complement, adjunct. Thus, the
clause
the big dog bit the little man on the leg
has four elements of structure, an s element (the big dog), a p element
(bit), a c element (the little man) and an a element (on the leg),
giving a structure: spca.
It is important to note that, unlike the elements of structure of the
units word and group, the sequence of elements is not fixed in clause
structure. Any combination of the four elements may occur; there
are clauses with the structures: spac, aspc, pca, etc. Nominal
groups operate at s and c in clause structure; verbal groups operate
at p in clause structure; adverbial groups operate at a in clause
structure.
There is in English one unit higher than the clause, and this is
called the sentence (;>). The clause operates as an element of struc¬
ture in the sentence. Thus the sentences
(i) the big dog bit the little man and he kicked it
(ii) the big dog bit the little man because he kicked it
are each composed of two clauses. The clauses which make up the
structure of a sentence may be either independent as in (i) or depen¬
dent as because he kicked it in (ii).
It is evident from this brief survey that the description of English
requires five units, which, arranged on a rankscale of descending
magnitude, are:
Sentence
Clause
Group/Phrase
Word
Morpheme.
The relationship which holds between the units on the rankscale
is crucial to the grammar here presented: moving from highest to
lowest, each unit consists of one or more than one of the next lowest
unit. Thus, a sentence consists of one or more than one clause, a
B
8 GENERAL SURVEY

clause consists of one or more than one group, a group consists of

one or more than one word, a word consists of one or more than one
morpheme. Since the structure of each unit is formed by the operation
of the next lowest unit, it follows that the lowest unit on the rank-
scale, the morpheme, has no (elements of) structure; if it had it
would not be the lowest unit.
There are two further points to be made about the rankscale.
First, it is obvious that there are two ways of looking at any unit on
the rankscale, upward or downward. If we look upward we are
looking at how the unit operates in the structure of the unit next
above; if we look downward we are looking at how the unit is made
up by members of the unit next below. The former is called syntax
and the latter is called morphology. Both syntax and morphology
extend over the whole rankscale. The syntax of (for example) the
nominal group is that it operates at s and c in clause structure; the

FIGURE 1
Theoretical Categories
->
Unit Class Structure (System)
Simple alpha
t Sentence Compound alpha (beta)
_t
Ind. at alpha (s), P, (c), (a)
Clause Dep. at beta
8 -—
•c Nom. at s/c
o (m), h, (q)
60
<L>
eS Group Verb, at p (a), («), /
U ->->
Adv. at A (0, a, (/)/pc
>
o, _t
•c
o pre-noun at m
co
o subs, at h
Q
u neg. at n __
8 Word aux. verb at a (j>), b, (/)
1cG lex. verb at l
Pi prep, at p
adv. at a

J- bound at p//
Morpheme free at b
Y
 GENERAL SURVEY 9
morphology of the nominal group is that it has elements of structure
m, h, q, at which operate members of the unit word. Second, and
consequent on the first point, we never, in fact, discuss only one
unit on the rankscale; any discussion necessarily involves the unit
next above or the unit next below.
Figure 1 is a diagrammatic presentation of what has been dis¬
cussed; and (though perhaps somewhat forbidding at first) has
proved useful as a mnemonic and as an illustration of the grammar
here presented.
The following points may be noted:
(i) The arrowed lines connecting elements of structure and units
indicate the syntactic (‘upward’) and the morphological (‘down¬
ward’) relationships discussed in the preceding paragraph.
(ii) Brackets round elements of structure indicate that these elements
are optional, and are dependent on the presence of the obligatory
element.
(iii) The arrowed lines above elements of structure indicate that the
given sequence of elements is obligatory.
It will perhaps be advantageous at this point to introduce a short¬
hand for marking boundaries of units. The following symbols are
used in this grammar:
Unit Symbol Example
Sentence /// ///The boys reminded Tom that he had
hidden the books///
Clause II ///The boys reminded Tom//that he
had hidden the books///
Group / / //The boys/reminded/Tom//that/he/
had hidden/the books///
Word HI The boys/reminded/Tom//that/he/
had hidden/the books///
Morpheme + / //The boy+s / remind+ed/T om//that/
he/had hid+(d)en/the book+s///

It follows from what has been said that the boundary of any unit
will also be the boundary of all lower units. Thus, a sentence boun¬
dary is also a clause boundary, a group boundary, a word boundary
and a morpheme boundary. The categories so far discussed will
provide:
10 GENERAL SURVEY
(i) a description of the units on the chain of linguistic events; this
means naming patterned stretches on the linear succession of lan¬
guage activity;
(ii) class names for each of these units based on their place of
operation in the chain;
(iii) a functional description based on the relations between classes
of unit: e.g. mh, sp.
Thus we can analyse the sentence the big dog will bite the little man in
the arm (stopping at group rank) as follows:

Ind
s p c A

N V N Ad
m h a / m h P c

m h
The big dog will bite the little man in the arm

It should be noticed that each place on the chain provides an en¬

vironment for further relations. Thus, s provides an environment for
an mh relation, p provides an environment for an al relation, and so on.
Places on the chain also provide another sort of environment; not
for further chain relations, but for choice relations. Choice here
means the selection of one particular term at one particular place on
the chain in preference to another term or other terms which are also
possible at that place. Thus, at s we must have a nominal group as
already discussed, but we have then to choose between singular and
plural number; at p we must have a verbal group, but we have then
to choose between present, past and future tense, and between
positive and negative polarity, and between active and passive voice.
Sets of choices are in this way associated with places on the chain.
Such a set of choices is called a system (the fourth of the theoretical
categories), and the possible choices are terms or features of the
system. Thus, there is a system of number, which has the terms sin¬
gular and plural; and a system of tense, with the terms present, past
and future. Diagrammatically:
 GENERAL SURVEY 11
System Terms System Terms
Present
Number-JSingu'ar Tense— > Past
|_Plural
Future
A network of such systems may be associated with a place on the
chain. Some of these systems may depend on prior selection from
other systems. In other words, some choices are more delicate
choices than others, and the application of a scale of delicacy is
important in the developing description of systems. The example of
the number system is, in fact, an over-simplification. It is clear that
at s in clause structure we may have a nominal group which chooses
for number, or a nominal group which does not do so. Only if the
former is chosen is there then the possibility of choosing between
singular and plural. So these choices may be arranged according to
delicacy in this way—

Number-
t Countable—
Uncountable
Singular
Plural

(See p. 120#.)
It is evident that grammar extends along two dimensions or axes.
These have been informally designated ‘chain’ axis and ‘choice’
axis: a class denotes the potentiality of operation of a unit on the
chain axis; a systemic term (or feature) denotes the selection made
from a set of mutually exclusive choices at a place on the chain. In
this book the expression ‘Surface Grammar’ denotes all that is
implied by the former, and the expression ‘Deep Grammar’ all that
is implied by the latter.
Chain and choice are presented diagrammatically in Figure 2.

figure 2
(Chain axis: classes of units)
Subject Predicator
Ih
<D
r Singular r Finite
O '[^Plural ’I Non-finite
a
o
" Masculine Present
Feminine Past
X Neuter Future
a
<L>
O
o
X!
r Active
[_ Passive
U *
12 GENERAL SURVEY

It is by way of systems that different syntagms (i.e. stretches of

language which would be described differently on the chain axis)
can be shown to be related, and similar syntagms (i.e. stretches of
language which would be similarly described on the chain axis) can
be shown to be unrelated. The two expressions (i) Tom hit Bill
and (ii) Bill was hit by Tom will entail different surface descriptions.
Thus:

(ii) f
1 1
Cl. Cl.
/\ /\
/ \ / \
/ \ / \
S P c s :P A
1 1
l 1
N V N n v Ad.
1 1 1 /\ /\
h l h h a Ip C

Conversely, the two expressions (iii) The boy sings well and (iv) The
material wears well will entail similar surface descriptions. Thus:

(iii) and (iv)

f
Cl.
/\
/ \
/ \
s P A
1 | 1
1 1 1
N v Ad.
/\ 1 1
m

Yet native English speakers will agree that (i) and (ii) are gram¬
matically related sentences, but (iii) and (iv) are not. If a grammar
does not show that structurally dissimilar items may be closely
related and that structurally similar items may be quite unrelated,
then the grammar is inadequate as a description of how language
works. The close grammatical relation of clauses (i) and (ii) is
evident only in their deep grammar. The surface differences of these
clauses can be attributed to the selection of different systemic
terms. Thus, clause (i) has selected ‘operative’ and clause (ii) has
 GENERAL SURVEY 13
selected ‘receptive’ from the system of transitivity.! It is clear that
the surface grammar alone gives no direct indication of the grammati¬
cal relatedness of clauses. It can be further shown that the original
expression
Tom hit Bill
is related to each of the expressions:
Bill was hit by Tom
Did Tom hit Bill?
It was Tom who hit Bill
Hit Bill, Tom.
by the selection of different systemic terms.
Some systems in English are discussed at length in Chapter 3. In
the view of language presented here, the description of a language
and items in a language requires two components—a surface com¬
ponent and a deep component. It is perhaps also obvious that as
the description of grammar becomes more delicate we are approach¬
ing the problem of meaning; but, properly viewed, this is the direc¬
tion from the beginning—it is not a question in language of ‘ grammar
and meaning’ but a question of ‘grammatical meaning’.

f The significance of these terms, and their relation to surface structures is

explained in section 3.1.4.
2
Surface Grammar

2.1 Morpheme and Word

It was stated in the General Survey that word was not the lowest
unit in English grammar, but that a lower unit, the morpheme,
operated at elements of structure of the word. The morpheme is
formally established by comparing words and discovering recurrent
patterns. Native English speakers will presumably agree that
hunting is composed of two elements, hunt + ing, but that stout is not
composed of two elements, st + out. This is so because the native
speaker recognises the occurrence of patterns already evident in the
language. He knows that ing appears after many other elements:
singing, shooting, swimming, turning, hoping etc., and that hunt
appears without other elements (i.e. as a word): tigers hunt their
prey; the hunt was successful; and that ‘hunt’ also occurs before
other endings: hunts, hunted, hunter. On the other hand, he knows
that although out occurs alone and before other endings, st does not
regularly occur with other items. He also knows that when out
occurs alone it has a meaning that it does not have when it occurs
as part of stout. Recognition of the various types of patterning
requires a formulation and extended application of this intuitive pro¬
cess.
If we compare:
eats eating
with walks walking
or
boy boys
with lad lads
 MORPHEME AND WORD 15
we can recognise identical stretches vertically and horizontally. We
have, in fact, a ‘square’2 of the form

AC AD
BC BD

where a is eat, b is walk, c is s and d is ing. Such a procedure will

provide a first step in the identification of morphemic structure of
words.
There are some initial problems, however. First, it is obvious
(since we can have structurally simple words) that there will be
squares where one of the elements is represented by zero (0):

eat eating hunt hunter

walk walking swim swimmer

horse horses tiger tigress

cow cows lion lioness

In the interest of methodological consistency we can say that a

word such as eat is composed of an element expounded by eat and
an element expounded by 0: but this is only a methodological con¬
venience which enables us to utilise the principle of the square. The
convention will be dropped in the final description of word struc¬
ture.
A second problem is that we intuitively want to include such pairs
as man \men goose:geese sheep'.sheep. The intuition is consolidated
by formal considerations. We are concerned in this chapter primarily
with the morpheme, but also with the word since (as discussed in the
General Survey) we are always concerned with two units on the rank-
scale. If we look at the syntax of the word we find that the man: men
type pattern in the same way as boy: boys. This means that they will
conform to the criteria for ‘noun’ (p. 27) and carry distinctions of
number and case (pp. 120, 123). It may also be said that we know
(or can elicit from a native speaker) that men means ‘more than
one man’ just as boys means ‘more than one boy’. Such semantic
evidence may be accepted as corroboration, but it is the overall
syntactic patterning which is criterial. In short, the following
frames—
16 SURFACE GRAMMAR

The ... is coming The . . . are coming

boy boys
man men
goose geese
sheep sheep
will accommodate only one member of each pair. In terms of our
square we can formalise this by stating that if we have a pair of words
which are syntactically and semantically like other pair(s), as
man:men is to boy:boys, then the first pair is acceptable if the
second pair enters into a perfect square, as

man men

boy boys

lad lads

A third problem is encountered by the occurrence of squares such as

detain contain
deceive conceive.

In such squares our analysis will yield the items de, tain, ceive, con
as elements of structure of words. None of these items, however,
occur alone as do the items boy, lad, man, walk, etc. If we recognise
such squares we must qualify this by saying that the elements of such
words are different from the elements of other words; in fact, we
have a small number of base elements which are bound (do not occur
alone) in contrast to the vast majority of base elements which are
free (which do occur alone). Again we find semantic corroboration
in that we cannot attach meanings to the elements tain, ceive, etc.
as we can with s (plural) or ed (past) or er (agent).
We can, then, identify a word as simple or complex in structure,
as composed of one morpheme (if it has 0 in the square) or more than
one morpheme, if it enters into a square of one of the following
types:

boy boys man men receive deceive

(ii) (iii)
lad lads boy boys retain detain
 MORPHEME AND WORD 17
The element which is common horizontally we will call the base
element of structure—and dispense with the 0 conventions, so that
boy does not analyse as base + 0 but rather as base; the element
which is common vertically we will call affix, and further distin¬
guish as prefix (p) and postfix (/) according to whether they precede
or follow the base.
This yields the elements of structure of the word pbf\ since it
can be seen that b may occur alone, but that p and / only occur in
association with a b, we can more delicately represent elements of
structure of the word

(/>)» b> (/)

where brackets indicate optional elements of structure; the arrow
indicates that these elements, when occurring, must occur in this
sequence, and the commas indicate that this is a list of elements of
structure (not a structure; we can only have a structure of a given
unit).
The vast majority of base elements in English are realised by one
free morpheme, e.g. boy, lad, horse, cow, mountain, sky. There are,
however, bases such as bookcase, blackbird, which enter into
squares—
bookcase bookcases
boy boys

That is, we shall have structures such as bf where b is realised by an

item such as bookcase. But book and case are themselves established
as free morphemes operating as base elements in word structure.
We may call this b2 as distinguished from the mono-morphemic
base (bi). Still another type of base must be accounted for: the type
exemplified by tain and ceive above. We may call this b3. We thus
have:

bx a structurally simple base realised by one free morpheme.

b2 a structurally compound base realised by two free morphemes
(bookcase, sandstone).
b3 a structurally compound base realised by two bound mor¬
phemes (deceive, detain).

We have morphemes which can occur alone, classed as free; and

morphemes which cannot occur alone, classed as bound. Free
18 SURFACE GRAMMAR

morphemes operate at b1 and b2; bound morphemes operate at b3

and at p and /.
A further consideration of word structure will show that there are
two main classes of bound morphemes. One class will form into
morphological paradigms or sets; e.g.
(i) eat, eats, ate, eating, eaten
(ii) boy, boy's, boys, boys'
(iii) green, greener, greenest.
The morphemes occurring in such sets are called inflexional mor¬
phemes, and are distinguished from bound morphemes which do not
enter into such sets; e.g.
-ish boyish
-ist ■ as in ■ novelist
-ise normalise.
This second class of morphemes are called derivational morphemes.
In addition to forming morphological sets, inflexional morphemes

(i) are not recursive; only one inflexional morpheme may occur
in the structure of any one word,
(ii) such an inflexional morpheme always occurs finally in word
structure, and therefore
(iii) if both a derivational and an inflexional morpheme occur in the
structure of a word, then the derivational morpheme must
precede the inflexional morpheme, e.g. novel + ist + s.

Derivational morphemes, on the other hand,

(i) may be recursive, e.g. boy + ish + ness (but sequences are fixed,
we cannot have *boy + ness + ish);
(ii) do not form morphological sets; and so cannot be fully accounted
for in grammar; they may be considered as on the border of
grammar and lexis;
(iii) if a derivational and an inflexional morpheme are present in the
structure of a word, then the derivational morpheme must be
non-final as noted above.

2.1.1 Inflexion
It is obvious that the three types of morphological sets above corres¬
pond to the traditional word-classes of verb, noun, adjective. It
must be stressed again that classes of units are determined syntactic¬
ally, not morphologically; thus, boy will be classed as a noun
 MORPHEME AND WORD 19
because it patterns in a certain way in syntax (to be described in the
next chapter), not because it enters into the morphological pattern
discussed above. It is because of this that we class sheep also as a
noun which enters into singular and plural contrast though it is
structurally dissimilar to boy. This, of course, is why we included the
man:men/child:children type in the squares. Nevertheless, though
syntactic criteria take precedence over morphological criteria, if
there is some congruence between syntactic and morphological
criteria in a language then this is helpful and can be utilised in descrip¬
tion. In English there is considerable congruence between syntactic¬
ally defined classes of noun, verb, adjective and the morphological
sets discussed above. Thus we can say that:
boy boy's boys boys'
eat eats ate eating eaten
green greener greenest
form a typical noun, verb and adjective paradigm respectively.
Before discussing English inflexions in more detail, it is necessary
to discuss the relations between the morpheme, or more properly
the morpheme class, and its realisation in substance. Anticipating the
material of a future chapter (p. 120), we can say that the word-class
noun is involved in selection from the system of number, which has
the terms singular and plural. If we say that boy is a typical example
of noun singular and that boys is a typical example of noun plural,
it might appear that we are saying that the written <s> or the spoken
fzj is the morpheme plural. But formal units are not the same as the
substance by which they are realised. The <s> and fzj are examples of
how the morpheme plural may be realised, but they are not the
morpheme plural. The morpheme plural, in fact, has no ‘ shape ’; it,
like other units, is an abstraction from the linguistic data. Grammar
is concerned with classes of items, not with particular realisations
in substance. The grammatical analysis of boys is noun: plural and
of men is noun:plural. That the former is formed by adding <s> or
fzj to the singular but the latter is not formed in this way does not
affect the grammatical classification.
The morpheme class plural has various realisations, among which
are vowel alternation man:men, and no overt change from the
singular form sheep‘.sheep (i.e. no change morphologically; we
could say plurality is realised by concord with determiners and
verbs: this sheep is‘.those sheep are', see further p. 120 ff.).
20 SURFACE GRAMMAR

NOUN INFLEXION

The vast majority of English nouns do form the plural by adding a

substantial realisation of the plural morpheme to the singular form.
In spelling, this means the addition of <(e)s> (though syllabic <y>
is replaced by <i> lady’.ladies, and <f> may alternate with <v»; in
pronunciation the realisation is various.
1. After the fricatives /s, z/ and the affricates /t, d/ the addition
is /Iz/:
Singular Plural
price <s> /Iz/ prices
judge <s> /Iz/ judges
church <es> /Iz/ churches
maze <s> /Iz/ mazes

2. After /p, f, t/ the addition is /s/:

Singular Plural
cap <s> /s/ caps
laugh <S> /s/ laughs
cut <S> Isl cuts
3. After /n, mi, b, d/ the addition is /z/:
Singular Plural
bun <s> /z/ buns
sum <s> /z/ sums
ring <S> /Z/ rings
cub <S> /z/ cubs
wood <s> /z/ woods

4. In some nouns the plural form involves the change of a voiceless

fricative to a voiced fricative and the addition of /z/ or in (c) below
/Iz/:
(a) /f/ becomes /v/ and /z/ is added; in spelling this requires the
replacing of <f> by <v> and the addition of <es>:
Singular Plural
calf <f->-v> + <es> /f/—>/v/ + /z / calves
eU (f-^v) + <es> /f/~>-/v/ + /z/ elves
wolf <f->v> + <es> /f/^/v/ + /z / wolves
Thus: half knife, leaf, life, loaf, scarf sheaf, shelf, thief, wife.
(b) /0/ becomes /S/ and /z/ is added; in spelling this requires only
the addition of <s>:
 MORPHEME AND WORD 21
Singular Plural
path /0/->/8/ + /z/ paths
(c) Jsf becomes /z/ and /Iz/ is added; in spelling only the addition
of <s>.
Singular Plural
house <s> /s/->/z/ -f /Iz/ houses
5. In some nouns no addition is involved, but there is an associated
vowel change; the sound change requires the corresponding spelling
change:
Singular Plural
man men
foot feet
tooth teeth
goose geese
louse lice
mouse mice
woman women
penny pence

6. Plurals of foreign nouns show considerable variation, and nearly

all such will, in some varieties of speech, occur as ‘Anglicised’
plurals:
Singular Plural
nebula nebulae
datum data
criterion criteria
phenomenon phenomena
crisis crises

7. The ‘double plural’ children involves vowel change from the

singular and the addition of /ran/; and the plural oxen involves the
addition of /an/:
Singular Plural
child children
ox oxen

English nouns are also inflected for marked case (p. 123#.). This
requires the addition of /s, z, Iz/ to the singular and plural forms;
which particular phoneme occurs depends on the preceding phoneme
(as in plural formation). In spelling, this requires the addition of
<’s> to the singular and <s’> to the plural.
22 SURFACE GRAMMAR

ADJECTIVE INFLEXION

There is a theoretical problem in classing adjective morphemes as

inflexional or derivational; they partake of the characteristics of
both. Though they are here being classed as inflexional, which is
perhaps the conventional view, this is only a convenience of descrip¬
tion. The problem should be appreciated.
There is no problem concerning the structure of adjectives: the
positive form consists of a base, the comparative and superlative
of base + postfix (bf):
b b p b p
green green + er green + est.
They appear to be inflexions, and perhaps most of us would class
them as such intuitively. This is because, like noun and verb in¬
flexions, they form a restricted set; they do not have the variety
exhibited by derivational endings; they help to identify a major word-
class. Against this, however, there are three arguments:
(i) They do not enter into concord relations, a relationship relevant
to nouns and verbs. In the expressions
the big boy walks'.the big boys walk
the noun and verb show morphological change, but the adjective
does not.
(ii) Although they help to identify a major word-class, they in fact
only identify a number of members of this class. Many adjectives
form their comparative and superlative periphrastically by pre¬
posing more and most.
(iii) Some of their members can be followed by inflectional mor¬
phemes : our elders, our betters.
The adjectives which regularly form the comparative and superlative
by adding -er, -est are of three types:
(i) practically all monosyllabic adjectives:
green, bright, fast, soft, fine, etc.
(ii) disyllabic adjective which end in -ly, -le, -er, -ow:
holy, noble, clever, shallow
(iii) a few other disyllabic adjectives:
common, concise, cruel, obscure, profound, quiet.
Longer adjectives usually prepose more and most’, but it should be
noted that adjectives which regularly have -er, -est can also prepose
 0

MORPHEME AND WORD 23

more, most. There are a few irregular adjectives in English, forming
their comparative and superlative by suppletion:
good better best
bad worse worst
little less least
much more most

VERB INFLEXION

The verb is that word-class which operates in the verbal group;

this primary class is subdivided into two secondary classes, auxiliary
verbs and lexical verbs (pp. 41 ff.). Lexical verbs have from three to
five different forms; because of syntactic patterning it is necessary to
recognise five forms as ‘ forms of the verb ’: these may be illustrated by:
eat eats ate eating eaten.
For convenience, these forms will be referred to in this book as the
base -s -ed -ing -en
form of the verb respectively. Not all verbs distinguish five forms
morphologically, and it is only because some members of the class
do have five forms that invariable forms in other verbs can in fact be
recognised as different syntactic forms. The variant forms of the
verb can be realised by linear addition, as always with the -s form
and the -ing form:
base -s -ing
walk walks walking
run runs running.
The -d form may be realised by addition of a plosive consonant to
the base form, or by vowel alternation in the base form, or by both
vowel change in the base form and addition of a plosive consonant:
base -d (change/addition)
walk walked +<ed> +/t/
take took <a>-Koo> /ei/->/u/.
say said <a)->-<ai> /ei/->/s/.
The verb may be what is usually called ‘invariable’, i.e. may not
morphologically distinguish between the base and the -d form—
base -d -en
cut cut cut
hit hit hit.
It is evident that these ‘invariable’ verbs do not distinguish the -en
form morphologically from the base and -d forms; most English
c
24 SURFACE GRAMMAR

verbs, in fact, do not morphologically distinguish between the -d

and -en forms:
-d -en
loved loved
walked walked
said said.
The -en form is sometimes marked by addition of <en>, /en/;
sometimes the -en form has the same vowel as the base and sometimes
the same vowel as the -d form, but sometimes it has a different vowel
from either.

PERSONAL PRONOUNS

Personal pronouns have separate forms for distinguishing terms in a

system of person (1st, 2nd, 3rd person) and, for the 3rd person
singular, terms in a system of gender (masculine, feminine, neuter;
see p. 127); they also distinguish terms in a system of case (s-case,
non s-case; and terms in a system of number (singular, plural; see
see pp. 120 ff). This is shown in tabular form in Figure 3.

FIGURE 3

Singular Plural

Non Non
s-case Gen. s-case Gen.
s-case s-case

1st Person I me my/mine we us our(s)

2nd Person you you your(s) you you your(s)

m. he him his

3rd Person f. she her her(s) they them their(s)

n. it it its

The alternative forms of the genitives indicate that the form without
the <s> (and my) occurs when the item operates at d in nominal
group structure, and the form with the <s> (and mine) occurs when
the item operates at h in nominal group'structure (at c). Thus:
It is her book This book is hers.
It is my book This book is mine.
 WORD AND GROUP 25

2.2 Word and Group

The previous chapter was concerned with the syntax of the mor¬
pheme and the morphology of the word. This chapter takes one
step up the rankscale, and is concerned with the syntax of the word
and the morphology of the group. Any discussion of group structure
must take account of the different classes of group in English. It is
the case that the units word, clause and sentence have classes
which are similar enough for these units to be assigned basic struc¬
tures irrespective of their various classes; thus in the previous chapter
the structure of the unit word (irrespective of classes such as noun,
verb, etc.) could be generalised as (p), b, (/). The different classes of
the unit group do not have enough in common to admit of such a
generalisation in respect of structure.
In the clause, the big dog will bite the little man in the arm I deeply
there are four elements of structure:
12 3 4
(the big dog) (will bite) (the little man) (in the arm/deeply)
To these elements of structure we can give the well-established names
subject (s), predicator (p), complement (c), adjunct (a). It is evident
that in clause structure the patterns at the elements 1 and 3 above
(i.e. s and c) are alike. Thus, though there are four elements of
structure, we require only three classes of group defined syntactically,
since one class of group operates at two places in clause structure.
Accordingly, the following three classes of group are established:
nominal group (operating at s/c in clause structure)
verbal group (operating at p in clause structure)
adverbial group (operating at a in clause structure).
The elements of structure of these groups, and the word-classes
operating at these elements are discussed in the following sections.

2.2.1 Nominal Group3

In the structure of the nominal group there is an obligatory element;
this is called the head or headword. The simplest type of nominal
group is therefore one which consists of an h element which is
realised by one word, as the italicised items in the following:
Cows eat grass. Fish swim.
He was pale. John shot Jim.
Mountains are magnificent. Cowboys chase horses.
26 SURFACE GRAMMAR

The h element may be preceded by another element; this is called

the modifier (m), and though this element may be realised by a
number of words (as any element may be) they are primarily grouped
together as realising the element m. The following groups have the
structure mh:
m h
the boy
m h
the big boy
m h
the very big boy
m h
all the other big American boys
There is also an element of structure which may follow the h ele¬
ment; this is called the qualifier (q). The following groups have the
structure hq\
h
Boys with red hair (are nice)
h q
Houses in the country (were fashionable then)
h q
Cars that break down (are a nuisance)
All three elements may, of course, be present; the following groups
have the structure mhq:
m h
The big boy with red hair (scored the first goal)
m h q
The old car that I sold (is still going strong)
m h q
(He bought) a nice house in the country
The elements m, h, q, are the elements of structure whose particular
configuration (h; mh; hq; mhq) will form the structure of any given
nominal group. Since h is an obligatory element, but m and q are
optional and dependent, we may list the primary elements of
structure of the nominal group as:
-v
(m), h, (,q)
where, as previously, the arrow indicates that sequence is invariable,
the brackets indicate optional elements, and the commas indicate
that this is a list of elements of structure, not a structure.
 WORD AND GROUP 27
It is perhaps worth repeating that primary structure brings to¬
gether all those items which share a potentiality of occurrence. Thus
a complex nominal group such as:
all the other ten very worn school books in the library
and a relatively simple group such as:
the man himself
are both analysed primarily as mhq:

m h q
all the other ten very worn school books in the library

m h q
the man himself.

A headword can thus have a modifier preceding it and/or a qualifier

following it. Some headwords occur with a considerable variety of
modifiers and qualifiers, but others do not. The words which operate
at h in nominal group structure may be classified according to the
patternings of modifier and qualifier with which they occur.
Since it is possible to have simple nominal groups, i.e. groups
with only a headword, it is obvious that we can first of all class
together all those words which can appear at h. We can call this
class substantive, and the words operating at h in the minimal nominal
groups in the following are therefore substantives :
Horses run races. I live in Glasgow. Running is fun.
Training is good for you. He sings songs.
If headwords in more complex groups are considered it becomes
obvious that different members of the class substantive occur in
different patterns. There are four main subclasses.

(i)
h h
These boys are superb examples,
h
The bath I had was very refreshing.
h h
All the books that I read on holiday belonged to my brother,
h
The spot where the river bends is tricky to negotiate.
The words operating at h in the italicised groups are nouns. The
positive characteristics of the word-class noun are that its members
28 SURFACE GRAMMAR

occur with a large variety of modifiers and qualifiers (demonstra¬

tives, adjectives and, in particular, the definite and indefinite articles),
and they exhibit the morphology discussed in the preceding chapter.
Negatively, they are characterised by the fact that they cannot be
modified by the submodifiers very, rather, quite.

(ii)
h h
Nobody in the room would admit to it.
h h
She danced me off my feet.
h h
Anyone who is an artist will tell you.
h
They all went up the hill.
h
They both play very well.
The words which operate at h in the italicised groups are pronouns.
Pronouns are not usually modified, though they can be qualified.
The qualifiers of pronouns are themselves complete clauses or groups
of somewhat restricted types, as in:
h q
Anyone [who is an artist}
h q
Nobody [in the room].
Some pronouns occur uniquely with other qualifiers:
(a) the so-called ‘ indefinite pronouns’ occur with the item ‘ else ’-
h q
everyone
no one
someone else
somebody
nobody
(b) the ‘plural personal pronouns’ occur with the item ‘both’—
h q
theyl
both
we
Pronouns, then, take virtually no modifiers, and restricted qualifiers;
in this they differ from nouns. In particular, they cannot be modified
by the definite or indefinite article.
 WORD AND GROUP 29
(iii)
h
London is the capital.
h
Come to Canada.
h h
Matthews won the cup for Blackpool.
The words operating at h in the italicised groups are proper nouns.
Proper nouns cannot be modified or qualified, and so differ from
other types of headword. The occurrence of groups such as Bonnie
Scotland, Sunny Sussex, does not invalidate this statement. The
adjectives in such groups clearly do not fulfill the usual function of
adjectives in the nominal group. They do not identify the headword
as a particular member of its class: thus, dark night means ‘of the
class night that member which is characterised by darkness', or
sunny day means ‘ a day which is sunny ’; but Sunny Sussex does not
mean ‘of the class Sussex that member which is sunny’, a Sussex
which is srnny'.

(iv)
h
He was old.
h
His attempt was rather feeble,
h
She turned quite pale at this.
The words operating at h in the italicised groups are adjectives.
They have a restricted range of modifiers and qualifiers; they can be
qualified by the items enough, indeed, or by complete clauses or
groups of restricted types:
h q
He is older [than Harry]
h q
He is keener Ithan I am}
m h q
He kept himself as fit [as possible].
Unlike nouns, adjectives cannot be modified by articles or demon¬
stratives ; but they can be modified by the submodifiers very, rather,
quite, etc.
These various characteristics are intended as formal definitions of
30 SURFACE GRAMMAR

the word-classes noun, pronoun, proper noun, adjective, as opposed

to the semantic definitions of the type ‘ a noun is the name of a place,
person or thing, living or dead’.
The structure of nominal groups which have a noun at h are the
most varied and complex, which is what we should expect since
noun is defined as that type of headword occurring with more types
of m and q than other types of headword. So that, although the
complex group all the other ten very worn American school books in
the house may be primarily analysed as mhq, it is apparent that the
occurrence of the various items is not random, that there are, in
fact, discernible secondary structures at m. Sequence of items is
important, since we can have:
all houses
all the houses
all the same houses
but not:
*the all houses/same all the houses
*all houses the/ten the houses
*stone the five all houses same.
The more we disturb the sequence the more unacceptable the group
becomes. On the basis of this obligatory sequence we can establish
secondary elements of structure at m. Thus:
d o e n Qi)
all the other ten very worn American school (books)
The names for these secondary elements of structure are:

d deictic (at which operates the word-class determiner)

n numerative (at which operates the word-class numeral)
e epithet (at which operates the word-class adjective)
c classifier (at which operates the same word-class as at h).

Looking more closely at the element d, we find that there are further
sequence restrictions within this:
fall the other
we may have
(both the same
'the all other
the same both
but not *
other the all
same the both.
We have, in fact, three distinct places at d:
 WORD AND GROUP 31
d1 (predeictic) all, both, half
d2 (deictic) the, this, that; his, its, John’s, mother’s, my; a, any,
another, no, neither, every, several
d3 (post-deictic) other, same.

The occurrence of items at d is not entirely independent of other

items at m, as might appear from the segmentation of m given above.
The occurrence of items at d is related to the occurrence of items at o.
At o the two types of numerals operate (cardinal and ordinal).
When we use cardinal numerals they can be selected independently
of determiners. Groups such as—
one horse five horses the five horses the same five horses
show that such numerals may occur with or without a determiner.
It should be noted, however, that number concord is required
between numerals and headwords; i.e. the cardinal one requires a
singular headword, and other cardinals require a plural headword.
When we use ordinal numerals we require a determiner for ‘identi¬
fying’. Thus:
his first opportunity his first attempts
John’s third bicycle her third trials
my ninth birthday their ninth birthdays
These examples show that ordinal numerals require a determiner,
but also that they do not require number concord with the head¬
word, since the same numerals occur in the second column with
plural headwords as occurred in the first column with singular
headwords.
At the element e the word-class adjective operates (it has already
been shown that this class can operate at h also), and the number of
items in this class is very great. At this element of structure in the
nominal group recursion becomes marked; that is, there are often
two or more exponents of e present in one group. The position at e
is somewhat more complicated than at d, and the place orderings,
the permissible sequences, have not been fully worked out. The se¬
quences are not invariable, they are preferred rather than prescribed,
and even the usual sequences can be altered if accompanied by an
intonation break. There is, however, a general pattern which gives
fairly clear guidance. Adjectives which identify the headword by
‘comparison’ or ‘degree’ seem to occur in first position; this means
adjectives which are regularly compared either by preposing
more‘.most or taking the -er:-est inflexions, and adjectives which are
32 SURFACE GRAMMAR

submodified by items such as rather, quite, very, terribly, unbearably,

etc. We may call this position e1. Thus:
e1 e1
the ten nicest men a very lovely landscape
e1 e1
the cleverer brown dog the rather elegant man
e1 e1
the more graceful blue car the oldest wooden houses
el
the fifteen best American sprinters
It can be seen from these examples that such adjectives precede
other adjectives such as blue, American, wooden. It can also be seen
that there are two points to note when superlative adjectives occur
at e1: i) the presence of a superlative in this position requires a
determiner at d, a relation similar to that obtaining between ordinal
numerals and determiners. Thus:
the best chair his most polished performance
the sheerest silk
and ii) a superlative almost always points forward to a q in the
nominal group. Thus:
e1 h q e1 h q
the best chair [in the room] the most graceful animal [in the world]
Place orderings within e1 itself can be distinguished. In first place
occurs the ‘size-shape’ group, in second place the ‘quality’ group,
in third place the ‘ age ’ group. Thus:
size shape quality age
big fat fine young
tall thin graceful old
large slim scraggy new
huge square precious year-old
etc. etc. etc. etc.
We rarely have groups with a member from each group present, but
smaller groupings suggest this sequence. That is, we find:
the tall scraggy man rather than the scraggy tall man
a large square object „ „ a square large object
a tall young man „ „ a young tall man
a precious young thing „ „ a young precious thing.
These adjectives are followed (at e2) by the ‘colour’ adjectives. The
colour adjectives are less readily submodified, and have a range of
submodifiers with which they typically occur. Thus:
 WORD AND GROUP 33
salmon pink emerald green royal blue dark blue
bottle green buttercup yellow
These adjectives are followed in turn (at e3) by what may be called
the derived adjectives such as wooden, silken, American, strategic,
etc. That is, adjectives formally derived from other word-classes, so
they are usually compound in structure. These adjectives do not
regularly take comparison. Thus:
e1 e2 e3 el e1 e3
the large blue American carpet the huge old wooden trunk
At the element n operates the same word-class as operates at h in
the nominal group. It is often said that a noun in this position is
‘acting as an adjective’, and should presumably be considered as an
exponent of e. There are, however, both formal and semantic
differences between adjectives at m and nouns at m. In the most
general terms, items which operate at e denote accidental properties
of the headword, but items which operate at n denote inherent
properties of the headword: thus, a stony path is not the same as
a stone path; a stony path is presumably a path which has stones on
or in it, but this is not an essential feature and does not con¬
tribute to the inherent nature of the path; a stone path, on the other
hand, is presumably a path made of stone, and the ‘stoneness’
is an inherent property of the path. Formally, stony may be com¬
pared like other adjectives, but stone may not; it is possible (if un¬
likely) that a stony stone path may occur, but not a stone stony
path.
Since both n and h are expounded by the same word-class, and
since both may be compound, there is potential ambiguity in identi¬
fying structures. This type of nominal group, frequently with several
exponents of n, is now well-developed in English and is a feature of
the present-day language in some registers. Intonation is important
in the spoken language for identifying such structures, and hyphena¬
tion in the written language, though there is no great consistency in
the written language in this respect.
The essential characteristic of h in a nominal group is that it
contains the final stressed syllable, as in (stressed syllable italic¬
ised) :
the white house
the black bird structure: deh
the old book
Compound headwords can therefore be identified by the fact that
34 SURFACE GRAMMAR

their first base element will contain a stressed syllable, final in the
group, but their second element will not, as in:
the Whitehousz
the black bird structure: dh (h compound)
his pocket-book
Nominal groups which have a structure nh will normally have a
stressed syllable on the n element and on the h element. This means,
of course, that h is still identifiable as the final stressed syllable.
Thus:
the coin boxes
the hotel prices structure: dnh
the gold mine
the pa-wenger vehicles
Difficulty of interpretation may arise either when there is more
than one exponent of n, or when there is a structure nh in which one or
the other exponent is compound. The main types are listed below, and
hyphenation is consistently used to indicate compound items though
in some of the (occurring) examples hyphenation was not present.

0)
time-table commitments
bank-rate rise
man-hole cover
The structure of these groups is nh (n compound). Normally such
groups will contain a stressed syllable on the first base element of n
and have an intonation break (//) between n and h. Thus:
time-inb\e/ /commitments
bank-rate//rise
man-hole/[cover

(ii)
school sum-book
London road-map
roadside telephone-boxes
The structure of these groups is nh (h compound). Normally such
groups will contain a stressed syllable on n and a stressed syllable
on the first base element of h, and have an intonation break between
n and h. Thus:
school/1sum-book
London//road-map
roadside/1 telephone-boxes
 WORD AND GROUP 35
(iii)
subscriber trunk dialling
Fife County Council
London briar pipes
The structure of these groups is nnh. This structure is perhaps less
common than the preceding two. They are by no means unusual,
however; and they may have several exponents of n: consider—
subscriber trunk dialling
subscriber trunk dialling system
subscriber trunk dialling all-number system
subscriber trunk dialling all-number system fault
subscriber trunk dialling all-number system fault engineer
Structural ambiguity seems highly possible, but in practice the con¬
text of the utterance is usually such that only one interpretation is
possible. (Cf a head buyer (i) in Glasgow, (ii) in Borneo: suggest
different structures!) Nevertheless, the decision between assigning
the structure nnh and nh (with one or the other compound) may on
occasions be difficult and perhaps arbitrary.

In general, a different type of item occurs at q than that which occurs

at m. The element q is typically realised by a complete group or
clause. That is, another complete unit operates as an element of
structure in the nominal group, a unit of the same or higher rank.
For example, in the clause, the man I visited in town has gone away
the subject is the man I visited in town. It has already been said that
the element subject is realised by a nominal group. It is obvious,
however, that this example consists of not just one group, but several;
and indeed the organisation of only part of this sequence is what we
shall find in clause structure. We can, in other words, morphologically
identify I visited in town as a clause: it has a structure spa.
A clause normally operates higher on the rankscale, in the struc¬
ture of the sentence; indeed, this is crucial in the definition of clause.
In the present example, therefore, this clause seems to have moved
down the rankscale in order to operate in the less usual position of
q in a nominal group. Since the grammar is concerned with the
arrangement of units on a rankscale, a clause operating in the struc¬
ture of a lower unit is said to be rankshifted. The use of this term is a
descriptive convenience; the structure could easily be called some¬
thing else (the traditional terms ‘restricted/non-restricted’, ‘defining/
non-defining’ would describe relative clauses so involved, but there
36 SURFACE GRAMMAR

are a few other rankshifts in English, and a generic term is needed).

The main point is that such structures have to be accounted for.
It is important to distinguish between clauses which are operating
in sentence structure (the norm; not rankshifted) and clauses which
are operating at q in the nominal group (rankshifted). In the clauses:
the man who came to dinner stayed a month
nobody who calls himself an artist would paint like that
the nominal groups operating at s have rankshifted clauses at q:

The American runner, who was getting on a bit, failed to finish.

the clauses are operating in sentence structure. In the written lan¬
guage such clauses are marked off by commas, and in the spoken
language they are marked off by intonation breaks.
Also at q, as is evident from earlier examples, other complete
groups operate. The normal place of operation of groups is in clause
structure, so groups operating within the structure of another group
are also called rankshifted. In the clauses:
a car that price is beyond me
shoes this size are usually dearer
the nominal groups operating at s have rankshifted nominal groups
at q. Thus:

m h q
A car [that price] is beyond me
 WORD AND GROUP 37
s

h q
Shoes [this size] are usually dearer.

It is very common for adverbial groups to operate at q in nominal

group structure, rather than their normal operation at a in clause
structure. In the clauses:
I bought a house in the city
the snow on the hill was deep
he chose the books with leather bindings
the nominal groups: a house in the city, the snow on the hill, the books
with leather bindings have adverbial groups operating at q. Thus:

m h q
The snow [on the hill] was deep

m h q
He chose the books [with leather bindings].

Complexity at q is occasioned mainly by recursion. In most instances

this is a relation of successive rankshifting: there is one q in the
group but this q itself has a structure which contains further rank-
shifts. Thus, I found him in the house in the main street does not mean
I found him (a) in the house, and (b) in the main street, but rather /
found him in the house (i.e. the house which) is in the main street.
The analysis is therefore:

s p c A

N v N Ad

h / h P c= M
m h q
I found him in [the house [in the main street.]]
38 SURFACE GRAMMAR

Such sequences can become quite complex. If we add in the town

to the above example we will have a further rankshift. And, although
most examples can be successively analysed as above, there are
occasions when there is ambiguity between successive adjuncts in
clause structure and rankshifted adverbial groups at q in nominal
group structure. For example, if, he decided on the house in the country
means he decided on the house which was in the country, then the
clause has a structure spc, and c is realised by a nominal group
which has a rankshifted adverbial group at q:

s p c
N v N

h / m h q
He decided on the house [in the country.]

If, however, it means while in the country he made up his mind about
the house then there is an adjunct in the clause structure:

s p c A
N v N Ad
h / m h P c = [N]

m h
He decided on the house in the country.

The foregoing sections were intended to illustrate the general

patterns of nominal group structure in English. There are a number
of problems, however, which were not touched upon.
A constant problem in nominal group structure is the flexibility
of members of the word-classes which realise the elements of struc¬
ture. It is this which motivates questions of ‘noun acting as adjec¬
tive’, ‘adjective acting as noun’ and so on. These questions are
connected with problems of derivation, and with problems of lexis
—since the word, though a grammatical unit, is the closest corres¬
pondent to a lexical item. A general answer to such problems,
however, is perhaps that the question* ‘what class of word is redl’
is an unreal question (or at least requires several answers): the
question should really be, ‘what class of word is red in the groups
the reds of autumn: the red bookV
 *

WORD AND GROUP 39

There are a few problems connected with initial place in nominal
group structure. It was implied that deictics initiated nominal group
structure, as in all the other boys. But there is another element which
may precede such items:
almost
nearly ■ all the boys,
very nearly
Such items only initiate the group when there is a pre-deictic present;
we do not find:

rhe boys.
*almostj
Since they thus enter the nominal group structure in association
with another element, it seems they ought to be treated as sub¬
modifiers, similar to other sub-modifiers, such as those with adjectives:

m h

d e

s-m adj.
the very old houses

A special instance of sub-modification occurs when there is a super¬

lative adjective in the group, and the items much, quite, can initiate
group structure:
much
quite ■ the best house,
very nearly
A further problem is raised by certain groups containing the item
‘of’:
several of the men five of the men both of the men
a lot of the men all of the men.
If we analyse such examples as:
m h h
a lot [of the men] both [of the men]
the results are surely counter-intuitive. That is, lexical words such
as men will occur merely as part of a q to headwords such as lot,
both. Such examples are not, of course, to be confused with groups
which do have lexical items at h and of qualifiers:
m h q m h q
the hearts [of men] a basinful [ofjoy].
D
40 SURFACE GRAMMAR

Further comparison, and particularly a study of systems at d (pp.

128 ff) will show that the items a lot of, several of, etc., operate like
deictics, and it would seem reasonable to include such compound
items in the class which realises pre-deictic. This will give a more
satisfactory analysis of the groups above:

m h m h

d d

di d% *k d2
a lot of the men both of the men

It was stated earlier that items which occur at m do not usually

occur at q. Adjectives at e precede their noun at h: but there are a
number of instances where the adjective follows a noun. This is
sometimes for conventional reasons, sometimes for stylistic reasons
and frequently to give an appositional relationship:

(a) with certain constructions historically modelled on French:

fee simple
court martial
body politic
Postmaster General
(b) when there is more than one adjective:
a leer menacing and horrible
thoughts dear and tender
(c) when the adjective is sub-modified:
a talent so great
a joy too divine
a laugh so infectious
(d) adjectives placed after the h in apposition:
a man, cruel beyond belief
the woman, beautiful and clever.

Proper nouns present some problems. The type Sunny Spain

has already been touched on. It should be added that proper nouns
may be plural: the Alps, the Antipodes, and that homonyms of unique
names may be used with the ordinary noun contrasts and patterns:
the Joneses, a John Smith, the young Matthews.
 WORD AND GROUP 41
2.2.2 Verbal Group4
The verbal group is one of the most complex areas of English gram¬
mar. This complexity is brought about by the fact that a relatively
large number of systemic choices are made here, and there is not a
one-to-one correspondence between a term in a system and an
element of structure realising this choice. The same element(s) of
structure may be the realisation of a number of systemic choices.
Thus, a verbal group such as is coming will be systemically de¬
scribed as finite, non-modal, active, positive, present in present, im-
perfective. But we cannot say which ‘bits of structure’ realise
which terms. The elements of structure which realise even one
systemic term may not be discrete. Thus, the ‘tense’ of the verbal
group:
(He) will have been going to be doing (it for a month now),
will ultimately be analysed as:
present in future in past in future,
but we do not have neatly separated elements of structure realising
each component of this compound tense.
It is obvious, however, that this complexity is primarily a systemic,
a deep grammar complexity. Initially the verbal group structure may
be described in relatively simple terms. But when consideration is
given to the various systems (pp. 131 ff.) a more complex description
which relates the various structures to the terms they realise will be
necessary.
In the structure of the verbal group there is again an obligatory
element; in the verbal group this is always the final element of
structure. This is called the lexical element, because (apart from the
very restricted list of auxiliary verbs discussed later) any verb in
the language may operate here. The simplest type of verbal group
will therefore be one having only a lexical element of structure.
Thus:
comes
walks
fought J. structure: /
swindled
fire
There is an element of structure which may precede the lexical
element. This is called the auxiliar element, and the verbs operating
at this element are called auxiliary verbs. The structure of such
groups will therefore be auxiliar + lexical element. Thus:
42 SURFACE GRAMMAR

will come'
can come
is coming ■ structure: al
will eat
has eaten
There are only eleven members of the class of auxiliary verb, so
they can easily be listed:

be, have, do, will I would, shall\should, canjcould, mayjmight, must,

ought, dare, need.

Auxiliary verbs are differentiated from lexical verbs by a number

of criteria, morphological and syntactic. The morphology of
auxiliary verbs differs from that of lexical verbs (pp. 45-46) in that
the auxiliary verbs have either more or less forms than the lexical
verbs. In syntactic function there are four main criteria by which
auxiliary verbs are classified:

(1) Negation: when they operate in negative verbal groups, the

auxiliary verbs regularly take the enclitic form of the negator, n't;
the lexical verbs never take the enclitic form of the negator:
I can't come *7 jumpedn't it
I mustn't go but not *1 hopen't so
Tom didn't go */ liken't it.

(2) Inversion: auxiliary verbs are obligatory in clauses which have

‘inversion’, i.e. clauses which have the s included within the P.
Inversion is characteristic of clauses which are interrogative, and
clauses which contain semi-negative adverbs in initial position:
Is the boy coming? Rarely had they seen such things
Isn't the boy coming? Scarcely had the noise subsided . . .
Will he come ? Never have I said such a thing.

(3) Substitution: the function of the auxiliary verb ‘do’ is to sub¬

stitute for a lexical verb; it may substitute for any lexical verb in the
language. For this reason the auxiliary verb ‘do’ is often said to be a
‘dummy’ or ‘lexically empty’ verb:
I like it and so do they
I scored and so did John.
In a linked sequence of clauses in which the verbal groups contain
an auxiliary verb, only the auxiliary verb need be repeated after the
 WORD AND GROUP 43
first clause. The auxiliary verb is thus a substitute for the whole
group:
He can come and so can John
He will play on Thursday and I think Harry will too.
Given an initial lexical verb as the ‘key’, such a ‘code’ of auxiliaries
may extend across sentence boundaries in dialogue.

(4) Marked Positive Element: This is closely related to criterion (2),

and may be considered here. It is possible to stress the auxiliary in
a verbal group and thus stress the polarity (the ‘yes-ness’ or ‘no-
ness’, p. 137) or the modality (p. 132) without stressing the Texi-
cality’ of the verb at /; an auxiliary verb is therefore necessary for
this purpose:
He did hit the boy
You must see him
He has been asked
I can come.

When a verbal group contains an auxiliar and a lexical element, the

lexical verb operating at / will take a non-finite form, and the auxiliary
verb operating at a may take a non-finite form (must if it is a
modal verb). It is an important distinction whether a verbal group
is finite or non-finite and so—although properly this is a systemic
distinction and is discussed later (p. 131)—some discussion and
illustration is perhaps useful at this point. The realisation of finite¬
ness occurs always at the initial element of structure of the verbal
group: when there is only a lexical element present the lexical
verb will have either a finite or non-finite form; when there is an
auxiliar element present then the auxiliary verb will occur in initial
position and will determine the finiteness of the group. The forms
are:
Finite Non-finite
eats I does eat Infinitive: to eat/eat
ate/did eat to have eaten
will, shall eat to be about to eat

beginning with:
'is, has, was, had Participle: taking
shall!should, canjcould taken/having taken
j may j might, must, ought being about to take
V.
dare, need
44 SURFACE GRAMMAR

We can sum this up by saying that there is a system of finiteness

which has the terms finite and non-finite. If finite is selected then this
will be realised by the occurrence at initial position in group struc¬
ture of an -s, -d or base form of the verb, or a modal: if non-finite
is selected then this will be realised at initial position in group
structure by the occurrence of the t-inf, base-inf, -ing, -en form of the
auxiliary if present, or the lexical verb if there is no auxiliary.

System Terms Realisation {at initial element)

'finite-v-s, -d, base form; or a modal
Finiteness
non-finite->-ing, -en, t-inf., base-inf.

There are four basic types of compound verbal group involving

the occurrence of an auxiliar and a lexical element. These are given
below, and are discussed here in terms of structure only, but some
brief notes are made indicating the nature of the systemic features
which the various structures realise, and which are discussed in more
detail later.

(i) auxiliary verb + -ing form of the lexical verb

The auxiliary verb in this structure is realised by a form of the be,
as in:
is coming
was coming
were coming.
This structure is particularly associated with ‘tense’ and ‘aspect’;
(‘continuous present’, etc.) in English; in the later discussion an
attempt is made to show that these are different systems, and should
be treated separately in description.

(ii) auxiliary verb + -en form of the lexical verb

The auxiliary verb in this structure is realised either by (a) a form
of the verb have, as in:
has bitten
had bitten,
or (b) a form of the verb be, as in:
is bitten
was bitten.
We clearly have two different sub-structures here, and—as would
be expected—they realise different systemic features. The structure
with have is like (i) in that it is associated with features of tense and
aspect.
 WORD AND GROUP 45
The structure with be occurs as a realisation of the term passive in
the system of voice.

(iii) auxiliary verb + the base form of the lexical verb

The auxiliary verb in this structure is realised by a member of a
subset of auxiliary verbs called modal verbs, as in:
can swim
must go
will play.
The features which are realised by the modal verbs in English are
complicated; but the various shades of meaning can be discussed in
general terms of ‘ obligation, duty ’ and the like. A problem is posed
by the auxiliary verbs will and shall: these verbs are usually said to
be primarily concerned with tense, in particular future tense. But
they certainly do not only express tense, and by their occurrence in
this structure may correctly be classed as modal verbs.

(iv) auxiliary verb + t-inf form of the lexical verb

The auxiliary verb in this structure is realised by a restricted set of
modals—ought, have, be (the have and be in this structure are different
in function from the have and be in (i) and (ii) above), as in:
he ought to go
he has to go
he is to go.
This structure should not be confused with examples such as:
he wanted to go
he hoped to go,
in which, it is argued below (pp. 59ff), we have two predicators in
clause structure, and not a compound verbal group realising one
predicator in clause structure.

We may return to the auxiliary verbs, and in particular to their

subclassification which was mentioned briefly above. There are
morphological and syntactic criteria by which the auxiliary verbs
may be subclassified. In terms of morphology, auxiliary verbs form
into five distinct classes:

(1) Be: this is the only verb in English which has five finite forms—
am, are, is; was were: it also has a base-inf., a t-inf., an -ing and an
-en form.
(2) Have: this verb has three finite forms (have, has, had); it also has
a base-inf., a t-inf. and an -ing form; it has no -en form.
46 SURFACE GRAMMAR

(3) Do: this verb has three finite forms {do, does, did), but no non-
finite forms (though the full verb do has non-finite forms).
(4) Will, shall, can, may: these verbs have two finite forms {will/
would, shalljshould, canjcould, mayjmight), but no non-finite
forms, and no -s forms.
(5) Must, ought, dare, need: these verbs have only one finite form,
and no non-finite forms.

If we take as a criterion the presence or absence of -s forms, these

auxiliary verbs fall into two classes:
with -s form: be, have, do (call these non-modal)
without-s forms: will, shall, can, must, ought, dare, need (call
these modal).
A similar classification is derived from the syntax of the auxiliaries
in relation to the four basic structures discussed above. These were:
(i) aux. + -ing
(ii) aux. + -en
(iii) aux. + base
(iv) aux. + t-inf.
In operation in these structures, the verbs be, have occur with -ing
and -en forms, i.e. structures (i) and (ii), but not with the base or t-inf.
forms, i.e. structures (iii) and (iv); the verbs will, shall, may, must,
dare, need, can, ought {be, have) occur with either the base or t-inf.
forms, i.e. structures (iii) and (iv), but not with the -ing or -en forms,
i.e. structures (i) and (ii). The verb do occurs with structure (iii), but
its overall patterning is unlike the other modals; it is unlike be and
have in occurring in structure (iii), but it is more like them than it is
like the modals and it is already classed with them morphologically.
So we may class the auxiliary verbs as follows:

Non-modal—be' have
*-b) do

Auxiliary verb—>

_ will, shall, can, may, must,

ought, dare, need, {be, have).

These are some problems connected with the identification of

some auxiliary verbs which also have lexical verb forms. The verb
be is exceptional both in morphology and syntax; it is already
evident that there are (at least) three be verbs in English, as in:
 WORD AND GROUP 47
(i) The world is round
(ii) a He is eating
b He is eaten
(iii) He is to leave immediately.
In (i) the form is is the only realisation of the verbal group at p
in clause structure. It must therefore realise the / element of the
group, so must be a lexical verb. In (ii) a and b the auxiliary realises
the auxiliar element of structures (i) and (ii) above, and so is an
auxiliary non-modal verb concerned with the realisation of features
of aspect and tense and voice. In (iii) the form is occurs with the
pattern of structure (iv) above, and is an auxiliary modal verb
concerned with modal function of ‘obligation, duty’.
Dare and need also present some problems, and perhaps remind
us that, as was said regarding word-classes in the nominal group,
we cannot properly identify items in isolation, but only as they
occur in syntagms. The verbs dare and need are characterised as
being auxiliaries by taking the enclitic negator and by inversion:
He daren’t go Dare we go?
You needn't ask Need they suffer like this?
and are further characterised as auxiliaries by having no -s forms:
we do not find— *daresn’the? *needsn’the?
On the other hand, we shall find:
He doesn't dare to go Do they dare us?
You don't need to ask Do they need to ask?
where the presence of do in the negative and inverted structures
shows that do is the auxiliary verb and dare and need are lexical
verbs. It is also evident that in:
He dares to call me a coward You dare to say that
He needs to be examined They need to have their tea
they are lexical verbs, because they have distinct -s forms, and be¬
cause they occur in structure (iv), not structure (iii).
It is obvious from what has been said that there is a negative (n)
element in verbal group structure in addition to a and /. This,
indeed, is one element of structure in the verbal group which does
stand in a one-to-one relation with the choice of a term in a system:
the choice of negative (rather than positive) in the system of polarity
is always realised by the presence of the negator {not, n't) in the
verbal group. The list of primary elements of structure of the
verbal group is therefore:

(a), (n), l
48 SURFACE GRAMMAR
where, as before, brackets indicate optional elements of structure,
the arrow indicates obligatory sequence, and the commas indicate
this is a list of elements of structure, not a structure. Thus:
structure: l al anl
comes will come will not come
taken will take won't take
votes did vote didn’t vote.
Longer verbal groups can be described as combinations of the
four basic types. This means that when two or more auxiliary verbs
occur in the verbal group they will combine with each other
and finally with the lexical verb in sequences of the four types.
We can primarily give the structure of such groups, and then de¬
scribe the combinations involved in terms of the four basic types.
Thus:
will be taking structure: aal; (iii), (i).
And so
will have taken structure: aal; (iii), (ii)3,
will be taken structure: aal; (iii), (ii)b-
will have been taken structure: aaal; (iii), (ii)a•, (ii)b-
will have been being taken structure: aaaal; (iii), (ii)a-,(i),(ii)b-
ought to have been taken structure: aaal; (iv), (ii)a,5 (ii)b‘
There is one further point of verbal group structure to be con¬
sidered. If the clause, he decided on the ship means ‘he made up his
mind while he was on the ship’ then the structure is spa:
s p A
he decided on the ship
but if it means ‘he chose the ship’ then the structure is spc and the
p is realised by a phrasal verb:
s p c
he decided on the ship.
There are many instances in English where verb + particle (preposi¬
tion or adverb) operates as a single unit. We can often replace the
verb + particle with a simple verb, as in:
He put down the rebellion: He suppressed the rebellion
She looked after her mother: She tended her mother.
Also, in passive constructions the particle remains with the verb:
The ship was decided on
Her mother was looked after
He was looked up to by his fellows.
It is also the case that which particles go with which verbs (and
 WORD AND GROUP 49
with which meaning) is restricted: look up to but not look down to
(but—look down on); take to but not take from.
Such criteria are usually cited to show that verb + particle form a
single unit in such structures, and realise the element p (c) in clause
structure; and are therefore different from verb + adverb realising
the elements p (a) in clause structure.
The concept of the phrasal verb, as this structure is usually
called, is well established in grammatical description, and is a valid
concept. But it must be said that the identification of phrasal verbs
in particular structures often relies heavily on semantic criteria, and
there is often room for disagreement; the line between phrasal verb
+ nominal group (pc) and verb + adverb or preposition (+ nomi¬
nal group) is very difficult to draw.
Nevertheless, there are many examples where the difference is
clear, and many long-established phrasal verbs can be easily recog¬
nised, as in:
s p A s p c
We went to the place: We took to the place.

2.2.3 Adverbial Group

There are two distinct types of adverbial group in English, but they
are initially classed together because they both realise the element
A(djunct) in clause structure. They may be secondarily classified
(on morphological grounds) as ‘adverb-headed’ adverbial group,
and ‘preposition-headed’ adverbial group.
As the name suggests, the first type has the traditional ‘adverb’
operating at an element of its structure. This is the obligatory
element of structure in the first type of adverbial group; groups with
only this element are common, particularly as realisations of
‘grammatical’ adjuncts (p. 63): if, that, because, though and the
like. Other typical realisations of this element are:
beautifully swiftly fast well
nicely now quickly adverbially
comfortably then there soon.
This element of structure is called the apex,5 and the simplest struc¬
ture of this type of group is therefore one with only an a element, as in:
I run fast I’ll do it now she sings beautifully.
There is an element of structure which may precede the a element;
this is called the temperer (t), and at this element operate submodi¬
fiers such as very, more, as, right, nearly, just, exactly, over. Thus:
50 SURFACE GRAMMAR

very nicely'
right there
structure: ta
over here
just before

There is also an element of structure which may follow the a

element; this is called the limiter, and a restricted range of realisa¬
tions occur here, some of which correlate with items occurring at t.
A unique item, enough, occurs at / when there is no t element, as in:
nicely enough "1 , . ,
. ,, structure: al
quickly enoughj
Another single word item, indeed, correlates very often with very at t,
as in:
very quickly indeed
very nicely indeed structure: tal
very fast indeed
When certain items occur at t, typically more and as, the / element is
realised by than/as + a nominal group or a clause, as in:
(He ran) more quickly than last time
(He runs) more quickly than Tom can run
structure: tal
(He ran) as fast as a hare
(He ran) as fast as his legs would carry him.
The list of elements for this type of adverbial group is therefore:
(t), a, (/).
The second (‘preposition-headed’) type of adverbial group has
two elements of structure, both of which must be present. These are
called the prepend, at which operates the word-class preposition,
and completive, at which operates a rankshifted nominal group.
The list of elements of structure for this type of group is therefore:
pc.
The similarity of symbols to the pc of clause structure is intended to
reflect the similarity of the ‘transitivity’ of prepositions and verbs;
this type of group is often described as a ‘ minor predication ’ in the
clause. Thus:
up the hill
in the road
on the rooj r
under the counter
J

A number of items occur which seem best treated as compound

 GROUP AND CLAUSE 51
prepositions operating at p in an adverbial group. Indeed, the dif¬
ference between some ‘multi-word’ prepend realisations and single
word prepend realisations is often no more than spelling convention.
Compounds are frequent with the item of (recalling some problems of
nominal group structure). Thus:
p c P c P c
out of the river in spite of the wind apart from him
p c P c P c
because of the snow in front of the house due to the rain.
There are two modifications to be made to this general picture.
First, it was stated that the element c was always realised by a rank-
shifted nominal group, but there are some structures which seem
best described as pc adverbial groups in which c is realised by an
adverb. Thus:
until recently'
until then
> structure: pc
from now
before now
Second, there are a number of items which can operate before the
p element of a pc adverbial group. We may denote this element (quite
arbitrarily) as o. A restricted, but fairly wide, range of items realise
this element. Items such as very, nearly, quite typically precede
the prepositions near, like, as in:
very near the house
structure: opc
very like the last onej
Other items, which occur with numerous prepositions, are:
just, right, straight, almost, half right to, just for,
as in:
just under the surface'
straight from town
> structure: opc
almost over the top
right from under him

2.3 Group and Clause

As we move up the rankscale, while no unit is in any sense more

important than any other, the potentiality of structural complexity
is increased. This is structurally obvious in three ways:

(i) Sequence is less determined. Elements of structure are not effec-

52 SURFACE GRAMMAR

tively defined by position alone. At the lower ranks we had elements

of structure such as pbf where p and / were defined by position, as
occurring before and after the b element, so we could generalise and

symbolise as pbf. Similarly, in group structure we had mhq, tal, pc.

In clause structure we will have four elements of structure—spca—
but actual occurring structures will exhibit various combinations
and sequences of these—sap, pca, aspc and so on.

(ii) Elements occur recursively, or in depth. This was already evident

in word and group structure:
f1 f1
kind + li + ness
e2 e1
lush green grass
n3 n2 n1
subscriber trunk dialling system.
But at the ranks of clause and sentence this becomes more marked
and complex.

(iii) The occurrence of inclusion or discontinuity. That is, one unit

occurring within the lineal representation of, but not rankshifted
into the structure of, another unit; or (looking at it the other way)
a unit being discontinuous in that its elements of structure are in¬
terrupted by another unit. Thus, s included in p in clause structure
(discontinuous p):

p
Is John coming ?
Clause included within the structure of another clause (discontinuous
element of sentence structure):

John,11 ifyou give him a ring,//will come over.

As mentioned already, there are four elements of clause structure

initially, to which are given the names s(ubject), p(redicator),
c(omplement), A(djunct). There is no obligatory element of structure
in the English clause. It is often said that p is obligatory (and some¬
times s) but many clauses occur without a p, and many clauses occur
without an s (and it is not necessary to claim that they must be
‘understood’). The elements of structure of the clause must therefore
all be listed with brackets if the most generalised statement is to be
 GROUP AND CLAUSE 53
given. It is, however, true that the distinction between clauses
containing a p and clauses not containing a p is a very important one.
Clauses containing a p are called major, and clauses not containing
a p are called minor. Major clauses have thus a list of elements of
structure:
(s), p, (c), (a)
The problem of identifying these elements of structure is some¬
what different from that at lower ranks, partly because of the
variability of sequence. It is usually said that the main criterion for
identification of s (if we ignore semantic definitions such as ‘ the thing
being talked about ’) is concord with p, as in the bird sings but the birds
sing. This means that a nominal group which can commute with a
gender bearing pronoun {he, she, it), or one of these pronouns at h,
requires the -s form of the verb at p when the tense is ‘non-past’.
In past tense verbal forms this concord is not marked. So that, for
identifying s, concord means that a gender bearing nominal group
requires the -s form of the verb where that form is possible. Thus,
in the clause, that man makes the suits there are two nominal groups;
that man can commute with he—a gender bearing pronoun, but the
suits can only be replaced by the genderless they/them (we cannot say
the suits can be replaced only by them, the complement form of
they, until we have in fact decided that the suits is the complement).
In this clause, therefore, it is that man which correlates with the -s
form of the verb, and since the -s form does occur in the clause then
that man is identified as s. Similarly, in the clause those men make the
suit it is the suit which can be replaced by a gender bearing pronoun
{it), and would require the -s form of the verb if it were s; since the
-s form of the verb does not occur, then it must be those men which
is s. But the clauses in which the s can be identified in this way are
restricted; in any of the following it is not possible:
that man makes the suit
those men make the suits
that man made the suit
that man made the suits
those men made the suit
those men made the suits.
Thus concord, though important in English, will only identify s in a
small number of possible structures. We normally, in fact, identify
by sequence alone. We take the s to be that nominal group which
immediately precedes the verbal group; so in either of the clauses
54 SURFACE GRAMMAR

he saw the film

the film he saw (but not the play)
he is identified as s.
The element p is identified by concord and sequence. We have
already morphologically identified p as one or more verbal group(s),
i.e. as having a different pattern from the nominal group. The clause
the boy will bite the dog contains two examples of the nominal group
pattern and one example of the verbal group pattern. When we
consider these groups syntactically we find that the sequence
nominal-verbal-nominal is very common, and that certain members
of the verbal group are realised by the -s form of the verb when
preceded by certain members of the nominal group; and further, the
sequence nominal-verbal often occurs alone, and is invariable,
though the second nominal, when present, may alter position, as in
the clauses
he saw the film
the film he saw (but not the play).
Concord relations between what we have called s and the verbal
group, which we can now say is at p, are mutually defining, as is
sequence.
The element c is like s in that it is realised by one or more nominal
group(s). Identification of c is already made, since we can say it is
that nominal group which is not the s of the clause.
Adverbial groups realise the element a in clause structure. In
theory any number of adjuncts may occur in a clause, but there are
some restrictions which will be considered later.
This four-element picture of English clause structure is, however,
complicated by the fact that there are situations where we can
morphologically identify a nominal group operating in the clause, but
we cannot identify it as s or c of the clause syntactically. In the clauses
they want the boys to do it
if wet, (the show will be cancelled)
Bill, shootjshoot, Bill
the items the boys, wet. Bill can be identified as nominal groups but
not as s or c. Such nominal groups are said to realise a z element of
clause structure. The reason for inability to identify the groups as
operating at s or c is not the same in each case; we have, in fact,
three different types of z element:

(i) In the first of these clauses the nominal group the boys is the
 GROUP AND CLAUSE 55
c of want and the s of to do. This is called a z positive element
(zpos), since it is positively identified as a conflation of s and c.
(See further, pp. 59 ff.)
(ii) In the second clause, if wet, the nominal group cannot be identi¬
fied as s or c because identification of these elements depends on
concord and sequence relationships with P, and this is a minor clause.
In distinction from the previous type, this is called a z negative (zne9).
(iii) The third clause is not a minor clause, nor is the nominal group,
Bill, a conflation of s and c. Though it may immediately precede s,
it is marked off from the rest of the clause by a comma (an intonation
break in the spoken language), and is variable in position—it
follows the p in the second form of the clause. This element is called
a z vocative (zvoc). (See further, pp. 94-95.)
The general picture of English clauses (and clause is perhaps one
of the areas to teach early, and certainly the best area for breaking
into the analysis of texts) is that the structure of the clause is
composed of some configuration of these five elements:
predicator (realised by one or more than one verbal group)
subject
complement • (realised by one or more than one nominal group)
z element
Adjunct (realised by one or more than one adverbial group)

The relations holding between these elements of clause structure

are in some ways different from those holding between elements of
structure of other units. There is perhaps no automatic procedure
for recognising a clause in English; a working guide is that any
structure whose elements stand in any of the possible spcaz relations
is a clause. This means that there are no restrictions such as ‘ a clause
must contain a finite verb’, ‘a clause must contain a subject’. In
each of the following sentences there are two clauses:
Having read the book, I won't go to the film
For John to take advantage of the offer, all his friends contributed.
Each of the first clauses exhibits pc relations, and the first clause of
the second sentence exhibits sp relation also; they can also commute
with clauses which do contain finite verbal groups:
Now that I have read the book, I won't go to the film
So that John could take advantage of the offer, all his friends
contributed.
56 SURFACE GRAMMAR
There is, however, a crucial difference between the clauses in each of
the sentences. The second clause in each can occur alone, as the
realisation of a sentence; the first clause in each cannot. Such clauses
are called independent and dependent clauses respectively. The main
criteria for the identification of independent and dependent clauses
are:
Independent Dependent

(i) absence of ‘ binding ’ presence of ‘ binding ’

adjunct adjunct: who, which, what, when,
where, how, that, if, because, although,
provided that, in case, in case that,
etc.
(ii) presence of finite presence of non-finite verbal group
verbal group at P at p:
obligatory a) infinitive:
To take part in the game, he came
home.
John was the one, the man to carry it.
b) participle:
The windows being open, the room
was cool.
Their king captured, the enemy
surrendered.
Dependent clauses may be secondarily classified according to
what is called the ‘vector of dependence’ into additioning, con¬
ditioning and reported clauses.
The dependent additioning clause corresponds to the non-defining
relative clause of traditional grammar. Such a clause cannot occur
before the independent clause to which it is bound, though it can
occur within (‘be included in’) it, or can follow it. Thus:
My young brother, who is very tall, plays the game well
I am shorter than my brother, who plays the game well.
Such clauses have a relative binding adjunct at initial place in
structure—who, which, when, where, how; in whomlwhich, by whom/
which, etc. It should be noted that the item that does not introduce
such clauses, though it does introduce rankshifted (restricted)
clauses.
The dependent conditioning clause corresponds approximately
 GROUP AND CLAUSE 57
to the adverbial clause of traditional grammar. It can occur before,
within or after the independent clause to which it is bound. Thus:
If you bring them, John will help you
John will help you, if you bring them
John, if you bring them, will help you.
Such clauses have either a non-finite verbal group at p, as in
John being in no mood to delay, the trip was started early
The empire conquered by the barbarians, the last of the Romans
took his own life
or have a binding adjunct such as after, as, as if, as soon as, because,
if, though, etc.
The dependent reported clause corresponds approximately to the
traditional noun clause of indirect speech. Such a clause follows the
independent clause to which it is bound, as in
He said that it was true
He declared it was a fact.
Such a clause may have an initial binding adjunct such as that,
whether, if; but it may occur without such an adjunct. Reported
clauses follow ‘reporting’ clauses which typically have a ‘reporting’
verb at p: expect, feel, find out, hear, imagine, say, tell, inform, claim,
shout, etc.

2.3.1 Complements in Clause Structure

The element c in clause structure is identified as that nominal group
(or those groups) which is (are) not in s relation to p. Not all English
clauses contain a c in their structure; some clauses contain one c,
and some contain two c’s. Clauses which do contain a c element are
called transitive, and clauses may be transitive single (containing
one c) or transitive double (containing two c’s); clauses which do
not contain a c element are called intransitive.

intransitive (no c)
Clause- >
[ transitive-►
single (one c)
double (two c’s)
As in:

structure: sp

he called her
structure: spc
he left a note
58 SURFACE GRAMMAR

he called her a beauty

>transitive (double) structure: spcc
he called her a cab

If, however, we consider the clauses

(i) Bill bought a new car or he called her a cab
(ii) Bill felt fine or he called her a beauty
we recognise intuitively that although both (i) and (ii) contain com¬
plements, there are different types of c involved. The distinction
between the c’s is in part a reflection of the difference between
noun-headed nominal groups and adjective-headed nominal groups.
Those nominal groups which have an adjective operating at h may
realise the element c in clause structure, but not the element s.
Nominal groups which have a noun operating at h may realise
either c or s in clause structure. Adjective-headed nominal groups
cannot normally operate at c in a pc type adverbial group, but noun¬
headed groups can. That is, we can have in the building but not
*in happy. Noun-headed groups can commute with pronimals, as:
the man->he/him
the trees->they[them
but not adjective-headed groups,
very happy-►*—.
The different types of complement are called extensive (noun¬
headed) and intensive (adjective-headed). But, though it is true that
adjective-headed complements are always intensive, it is not true
that all intensive complements are adjective-headed groups; the
clause Bill felt fit could be rendered Bill felt a fit man where a fit man
is an intensive complement, but is a noun-headed nominal group.
So the relation is only one way: given an adjective-headed group
at c we can say the c is intensive; but given an intensive complement,
we cannot say it will be realised by an adjective-headed group.
The main point to be made is that clauses containing an extensive
complement will have a passive transform; that is, any grammatical
clause which has an extensive complement will have a related clause
which has the p in the passive voice and whose s is the extensive
complement of the original.
Bill bought a new car->A new car was bought by Bill.
This is not so with clauses containing only an intensive complement,
but not
Bill felt fine->*Fine was felt by Bill.
Contextually, intensive means having the same referent as some
 GROUP AND CLAUSE 59
previous element of clause structure (Jine refers to Bill); extensive
means not having the same referent as something previously men¬
tioned. Superscript can be used to denote the different types:
extensive complement—cE, intensive complement—c1.
Clauses which contain double c’s will have different possibilities
of transform: if both c’s are extensive there will be two possible
transforms. Thus:
He gave her a shilling structure: spcece
She was given a shilling (by him) structure: spce(a)
A shilling was given her (by him) structure: spce(a)
If one of the c’s is intensive there will only be one possible transform.
Thus:
He called her a beauty structure: spceci
She was called a beauty (by him) structure: SPCr(A)
not *A beauty was called her (by him).
A further point which may be observed from the above examples
is that an intensive complement may be intensive to the s (necessarily
so if there is only one c in the clause), or to another c (if there are
two c’s in the clause). In the clause:
He called her a beauty structure: SPCECI
the second complement is intensive to the first; but in the clause:
He made her a good husband structure: spceci
the second complement is intensive to the subject. So, for a full
marking we ought to write:
He called her a beauty structure: spcecI(C)
He made her a good husband structure: spcecI(S).
Certain verbs will operate more readily with the more complex
types of structure, e.g. buy, call, consider, make, get, leave, like.
The possible variations, using make are:
She made a good cake structure: spce
She made a good wife structure: spc1
She made him a good cake structure: SPCECE
She made him a good husband structure: spcecI(C)
She made him a good wife structure: spcecI(S)

2.3.2 Predicators in Clause Structure6

In discussing verbal group structure (p. 45) it was noted that
sequences such as started to bite, wanted to bite did not fit the
auxiliary + lexical pattern under consideration there. The first items
are clearly not auxiliary verbs; they do not take the enclitic negator;
60 SURFACE GRAMMAR
we do not find *startedn’t to bite, *wantedn,t to bite. Nor do such
structures occur in interrogative clauses; they require an auxiliary
verb:
Will he start to do it?
Does he want to bite?
It is evident that we have two lexical elements in these structures,
and since only one lexical element occurs in each verbal group we
must have two verbal groups; this in turn means that we must have
two predicators in the clause structure. Predicators operating in
this way are called phased predicators. Thus:
He started to bite structure: spp
He wanted to eat structure: spp
He hoped to win the trophy structure: sppc
He refused to sign the paper structure: sppc
The last two examples show the occurrence of a complement after
the second predicator in the phase sequence. Such phased structures
also occur with a nominal group intervening between the phased
predicators, with or without a complement after the second predi¬
cator. Thus:
He wanted Tom to do it
He asked the skipper to turn the ship
He begged the other to find the jewel
He told the boys to go.
In these clauses the nominal groups Tom, the skipper, the other,
the boys, are complements of the first predicator and s of the second
predicator. This is frequently illustrated in a transformational way;
given the pairs of sentences:
(i) He asked John. John does it.
(ii) He told his elder brother. His elder brother sold books.
(iii) He begged her. She returned the book.
The complement of the first member of each pair can be conflated
with the subject of the second member of each pair, and the second
predicator given a non-finite form, to produce the grammatical
sentences:
(i) He asked John to do it
(ii) He told his elder brother to sell books
(iii) He begged her to return the book.
Nominal groups which conflate c and s in this way are called z
elements of clause structure (p. 55), more precisely zpos elements.
The analysis of these clauses is therefore:
 GROUP AND CLAUSE 61
(i) He asked John to do it structure: spzpc
(ii) He told his elder brother to sell books structure: spzpc
(iii) He begged her to return the book structure: spzpc
When the intervening nominal group is the realisation of an
intensive complement, however, the position is different. The clauses
He was eager to swim
He seemed content to remain
He became anxious to find her
have phased predicators and intervening nominal groups. The
nominal groups are realisations of intensive complements, and are
in no sense the subject of the second predicators. Since intensive
complements cannot operate at s in non-phased constructions this
is not surprising. They are like the first type of phase structure in
that they involve only one s, (we might say one ‘actor’ who both
‘wants’ and ‘eats’ in I want to eat, and one ‘actor’ who ‘is eager’
and ‘goes’ in lam eager to go). The intensive complement, therefore,
remains an intensive complement, and the analysis is:
He was eager to swim structure: spc{p
Perhaps the obvious question is whether there are clauses with
two ‘actors’ which contain intensive complements, as there are for
one ‘actor’ clauses. The clauses
I am eager for John to do it
I was anxious for Mary to come
would seem to fill this slot. In English we do not have two nominal
groups juxtaposed to introduce a second ‘ actor-action ’ sequence in
a phased construction, i.e. we do not find
*1 am eager John to do it
*/ am anxious Mary to come
We must insert the preposition for before the nominal group which
is s of the second predicator in the phased sequence. But such clauses
are the fourth type of phase structure, and it seems best to analyse
the ‘for + nominal group’ (following a c1 in a phased sequence)
as the realisation of a z element of clause structure. This gives us
four types of phased structures, two with one ‘actor’, and two with
two ‘actors’. Thus:
I hope to attend structure: spp
I am keen to attend structure: spc'p
I asked John to attend structure: spzp
lam keen for John to attend structure: spc!zp
Phased structures involve what are often called ‘linking’ or
62 SURFACE GRAMMAR

‘catenative’ verbs, and the structures are recursive. The analysis of

clauses containing more than two predicators follows naturally from
the analysis of two predicator clauses. Thus:
. f I asked Bill to come early structure: spzpa
2 ps <
[ I'd like to see you for a moment structure: sppca
I wanted John to invite Mary to come structure:
SPZPZP
3 p’s . I began to feel reluctant to play structure: sppc'p
I was eager for John to get Tom to play structure:
spdzpzp
4 p’s I told John to ask Mary to tell her brother to come
structure spzpzpzp
We have thus far only considered the t-inf. as the non-initial
component of a phased sequence. The -ing form of the verb also
occurs regularly, and—more rarely—the base form. Thus:
I detested asking him structure: sppc
I remember John doing it structure: spzpc
I kept them working structure: spzp
I kept working them structure: sppc
The shock made her faint structure: spzp
I saw them swim four lengths structure: spzpc
Phased predicators which are non-initial are always one of these
non-finite forms, t-inf., base, -ing. Which form occurs will depend
on the initiating verb, and verbs can be classified according to the
non-finite form they take: some verbs will occur only with the t-inf.
form, e.g. seem, offer, expect; some verbs will occur only with the
-ing form, e.g. relish, enjoy, deny, some will occur with both these
forms, e.g. stand, feel, start, and there may be a difference m meaning
in the two structures, as in:
He stood watching the race: He stood to win a packet
A few verbs will occur with a succeeding base form, e.g. make, feel,
help.
Since dependent clauses can occur with non-finite predicators
(p. 56) it is important to distinguish between sentences which con¬
sist of two clauses (an independent and a dependent clause) and a
sentence consisting of a single clause with phased predicators. The
sentences
I refused, to let him know my stand in the matter
John continued, to give them the chance
I agreed, to avoid a clash in the office
 GROUP AND CLAUSE 63
have two clauses. But the sentences
I refused to let him know my stand in the matter
John continued to give them the chance
I agreed to avoid a clash in the office
have one clause with phased predicators.
The difference between the two-clause and one-clause sentence is
that sequence is variable in the former but not in the latter; we can
have
To let him know my stand in the matter, I refused
To give them the chance, John continued
but not
*To do it he asked them
*To give them the chance he continued;
the non-finite part of a phased structure must follow the finite part,
and there is no possibility of an intonation break in the spoken
language or comma in the written language. Sentences with two
clauses regularly have an intonation break or comma between the
clauses. It is also the case that if a binding adverbial group can be
inserted without any change in meaning then we must have two
clauses and not a phased structure. Thus:
He began, asking for their help: After asking for their help, he
began
with which compare
He began asking for their help.

2.3.3 Adjuncts in Clause Structure

The number and position of subject, predicator, complement, in
clause structure are fairly restricted (allowing for phase); but there
is no theoretical limit to the number of adjuncts which can occur in
clause structure, and there is more variability of position. The
positions are, however, restricted by the class of adjunct.

(i) ‘Grammatical’ or ‘binding’ adjuncts are those adjuncts which

bind dependent clauses to independent clauses, and these occur at
initial position; their position is fixed. We can call this position 1:
a sp c
1
(ii) Position 2 for adjuncts is between s and p; at this position operate
the ‘semi-negative’ adverbs particularly. These adjuncts are perhaps
halfway between ‘grammatical’ and lexical adjuncts. Typical
64 SURFACE GRAMMAR

realisations are: never, usually, scarcely, seldom, hardly, nearly, even,

just, merely, only, quite, always. Thus:
He never managed to do it
I usually go to the Christmas service
I seldom find much of interest in his books.
The structure of the verbal group at p in these clauses is /; if the
verbal group at p has the structure al then the Adjunct occurs be¬
tween these elements. Thus:
He did manage to do it
I don't usually go to the Christmas service
I would seldom find much of interest in his books.
There is no difference in position of the adjunct involved here, it
depends entirely on the structure of the verbal group. We may call
this position 2:
sapc
2
(iii) Other adjuncts, and in this case the adjuncts are ‘lexical’, occur
in clause final position. Adjuncts which are realised by a pc type
adverbial group occur here, and many adverbial groups in -ly.
Thus:
He crashed the car into the wall
He crashed the car violently.
We can call this position 3:
A P C A
3

Further problems concerning adjuncts are in identifying different

types of adjuncts, usually into adjuncts of time, place, manner, and
deciding which sequence of these occurs if more than one is present
in clause structure. That is, if the two adjuncts to the shop (place)
and immediately (time) are to occur in position 3 in the clause
beginning He sent her . . . will the sequence be time, place or vice
versa? The preferred sequence in English seems to be place, time,
manner. That is,
He sent her to the shop immediately rather than He sent her
immediately to the shop
He went to the block with a smile rather than He went with a
smile to the block.
But the sequences are preferred and by no means fixed; different
varieties of English seem to follow different principles; and it is
 CLAUSE AND SENTENCE 65
certainly the case that the mobility of adjuncts is utilised ‘ stylistic¬
ally’, i.e.—the occurrence of adjuncts in ‘marked’ or ‘thematic’
positions (see pp. 97 ff.). A large and objective survey is required
to determine the main lines.

2.4 Clause and Sentence

There are well over two hundred definitions of the unit sentence.
These fall into various groups or types of definition; each group
usually adds to, or contradicts, the others. Most grammarians seem
to have found it necessary to correct and improve part or all of the
previous definitions of sentence, before each produced the definition
(which it never was, of course). Two things are immediately obvious
from this: i) grammarians are unanimous in recognising the
sentence, but ii) definition of this unit is very difficult.
Many attempts at definition were found to be unsatisfactory or
inadequate because they relied solely on meaning, and attempted to
delimit the sentence as being ‘a complete thought/meaning/idea’, or
even ‘the number of ideas that can be grasped in their relation in
one act of attention’. But the necessary pre-definition of ‘thought/
idea’ and so on (whether or not ‘complete’) makes such definitions
untenable and unworkable.
A second major approach attempts to delimit the sentence by
reference to its constituent elements. In this type of definition we
are told that the ‘secret’ of the sentence is to ‘name some object or
place or person or thing’ and then to ‘say something about that
object or place or person or thing’. Such definitions usually say that
we need a subject and a predicator to make a sentence. This means
that most commands, exclamations and requests would not be con¬
sidered sentences. In such instances, of course, the s is often said to
be ‘understood’. But how much (or little) can be ‘understood’? If
we are allowed to ‘supply’ elements then any expression can be
made to conform to the definition. And it should be noted that
such definitions usually state not only that a s must be present, but
that the s is ‘the thing being talked about’. Even in the simplest of
sentences, say John shot Jim are we talking about John or Jim (after
all, it may be his last chance to be talked about), or shootingl Or,
as seems more reasonable, about John shooting Jim ?
It is not a question of whether such definitions are ‘true’ or ‘false’;
only that they do not provide reliable criteria for grammatical
66 SURFACE GRAMMAR

description. There are obvious correlations between many of the

definitions and occurring English sentence types. There is good reason
for the recurring insistence on completeness; but it is not necessarily
a ‘thought/idea’ which is complete. There is also good reason for
trying to find a universal semantic component in sentences; sentence
is typically the unit with which English works in contexts of situation:
each language will have such a unit and perhaps this may be called
sentence; but the grammatical description of sentence will be re¬
defined for each language.
There was, in fact, confusion of levels in many definitions: the
sentence can be described contextually (in terms of meaning),
grammatically (in terms of form), phonologically (in terms of in¬
tonation) or orthographically (in terms of punctuation). But these
are different descriptions, and only after describing at these separate
levels should we describe their interrelations.
The orthographic definition is frequently used; a sentence is
defined as a stretch of language which begins with a capital letter
and ends with a full stop, question mark or exclamation mark. This
is, in fact, a valid working identification of a sentence, and even
many of those who advocate a theoretical definition of sentence
work with this definition in practice. In most varieties of present-day
English orthographic sentences and grammatical sentences do tend
to coincide. But this cannot be applied absolutely; in older texts
we may have absence of punctuation or an alien system of punctua¬
tion, and numerous present-day writers have idiosyncratic systems
of punctuation.
There are also certain correlations in the spoken language with
grammatical sentences, certain stress patterns, intonation contours
and breaks. Again this is a fair working guide, but the correlations
are not always realised in practice, and it is very often impossible to
tell what grammatical division is indicated by intonation patterns.
The completeness of the sentence has been increasingly recognised
as a formal completeness. Most linguists have tended to follow
Bloomfield’s definition:

‘. . . an independent linguistic form [unit] not included by virtue

of any grammatical construction in any longer linguistic form
[unit].’

This definition does not seem inadequate, and for the present model
of grammar may perhaps be paraphrased ‘A sentence is that unit
 CLAUSE AND SENTENCE 67
which does not operate in the structure of any higher (grammatical)
unit.’ The whole difficulty of defining the sentence stems from the
fact that it is the highest unit on the rankscale. Other units are identi¬
fied and defined by their operation at elements of structure in the
unit next above; thus, group is that unit which operates in clause
structure, morpheme is that unit which operates in word structure,
and so on. But, in the case of sentence, there is no higher grammatical
unit, so that sentence cannot be defined syntactically.
It should be noted that sentences may, of course, be part of some
larger linguistic context. The completeness or independence of the
sentence is relative to this, because sentences which contain pro-
nominals or anaphoric determiners will necessarily be related to the
antecedents of these pronominals or anaphoric determiners. Thus:
He became headmaster of Rugby
The boys turned the corner, ran down the lane and vanished
That house has stood up well to the gale
are sentences in which the referents of He, the (boys), the (corner),
the (lane)', that (house) must be given by cotext or context if the
sentences are to be understood correctly. But since there is no
grammatical unit above the sentence, this inter-sentence connection
cannot be completely accounted for grammatically. It is, of course, a
linguistic matter, and will be of importance for stylistic studies: indeed,
one linguist has defined stylistics as ‘grammar above the sentence’.
We cannot, therefore, classify sentences according to their opera¬
tion in the structure of the unit next above; the classification must
be effected in a different way from the classification of other units.
The most obvious way is by morphology; we have taken as a prin¬
ciple that morphology is subsidiary to syntax in grammatical classi¬
fication, but in the case of sentence there is no syntax. It seems
important to stress that the morphology of the sentence is not the
morphology of the clause; numerous grammars talk of the subject
(etc.) of the sentence, and exemplify with simple (i.e. one-clause)
sentences. But compound sentences may contain a number of clauses,
and therefore a number of subjects (etc.), no one of which can be
called the subject of the sentence, s, p, c, a are elements of clause
structure: it is complete clauses which operate at elements of sen¬
tence structure.
Classification in terms of morphology will yield a primary divi¬
sion into simple sentences (containing one clause) and compound
sentences (containing more than one clause). Compound sentences
68 SURFACE GRAMMAR

may contain a succession of independent clauses, or may contain

an independent and a dependent clause, or may contain a number
of independent and a number of dependent clauses. We can name
these compound, complex and compound-complex respectively.
Thus:
Sentence

Simple Compound
One independent clause except One independent clause plus
in answers, which may have one or more independent
one dependent clause. clauses and/or one or more
dependent clauses.
Ind + Ind (Ind ...) (Dep ...)
or
Ind + Dep (Ind ...) (Dep ...)
or, more delicately:
,-Simple (one constituent clause)
^—»- .-Compound (more than one Ind clause)
LNon-simple—> -Complex (one Ind and one Dep clause or more)
•-Compound-Complex (more than one Ind and Dep)
It is also usual to classify sentences contextually into statement,
question and command. There are certain correlations between these
contextual classes and punctuation marks:
. statement
? question
! command/exclamation
and with certain intonation contours. They also correlate to a con¬
siderable extent with types of response, in that statement is often
accompanied by ‘continuation signals’ (a-huh, go-on, ye-es), command
is often followed by action response (‘Shut the door’) and question
by oral response or action response; it is also true that in the case
of the simple sentence there are correlations with clause classes to a
large extent:
statement declarative
question interrogative
command imperative.
But the correlations are not one-to-one; there are sentences realised
by a declarative clause which may be contextually classified as
commands, and so on; in particular minor (and therefore moodless)
 CLAUSE AND SENTENCE 69
clauses can usually be assigned to a contextual class of sentence
(see pp. 75-76#.).
Since the simple sentence is by definition coterminous with the
one constituent clause there is little more to say about its structure.
Compound sentences consist of more than one clause, and this
raises an immediate question of contextual class: which clause deter¬
mines the contextual class of the sentences. In the sentences:
I came and I saw and I conquered
Do this or do that
Will you stay here or will you go to Canada?
we find, in fact, that there are no problems. The linker and conjoins
like items, and therefore the contextual class of any of the clauses
(considered as a simple sentence) will be the contextual class of the
sentence. Thus:
I came and I saw and I conquered statement
Do this or do that command
Will you stay here or will you go away? question.
In the sentences:
Tell him to come home when the game's over
John cut the rope after Bill had shouted
Keeping to the old route, Bill got home safely
both independent and dependent clauses are involved. It is only the
independent clauses which affect the contextual class of sentence.
If we alter the first sentence to Will you tell him to come home when
the game's over? the contextual class changes from command to
question. If we alter the dependent clause this does not change the
contextual class of sentence:
Tell him to come home when the game's over command
Tell him to come home if he's finished command.
It can be seen at this point that there are two elements of sentence
structure, one of which is obligatory. The obligatory element is
distinguished by two criteria: i) it can stand alone as a simple
sentence; ii) it decides the contextual class of the sentence.
The obligatory element of structure is called alpha (a) and the
optional element of structure is called beta (p). Thus:
When he comes, tell him to eat his lunch structure: Pa
He'll be in tomorrow, if he's free structure: ap
I didn't come because he told me structure: ap
Harry, who had never been before, didn't know structure:
*«P»
70 SURFACE GRAMMAR

It is obvious that independent clauses operate at a and dependent

clauses at p; and that the possible sequence or inclusion of elements
of sentence structure is as stated above for conditioning dependent
clauses, additioning dependent clauses and reported dependent
clauses. Thus, if we symbolise, p1 = conditioning, p2 = additioning,
p3 = reported, p1 may precede, follow or be included in a:
When in doubt, shout out structure: p*a
Shout out when in doubt structure: ap1
John, if you ask him nicely, will help structure: a«p1»
If you ask him nicely, John will help structure: p’a
John will help, if you ask him nicely structure: ap1
p2 may follow or be included in a; it may not precede a:
My uncle, who is a sailor, brings presents home structure:
««P2»
Central Africa, which is very hot, is not for me structure:
a«P2»
I like summer, when the days are long structure: ap2
I went to Aberdeen, which is in the North structure: ap2
p3 with an introductory adjunct cannot precede a; it may precede a if
it does not have an introductory adjunct; it cannot be included in a:
She said that she would come structure: ap3
She said she would come structure: a p3
She would come, she said structure: p3a
The sentences so far considered have all been two-clause sentences,
and it may seem that a and p are merely alternative names for inde¬
pendent and dependent clauses, since each a element has so far
consisted of one independent clause, and each p element has con¬
sisted of one dependent clause. But this is the position at any rank
when an element of structure is realised minimally; if (say) a p
element in clause structure is realised by a verbal group with only an
l element then verbal group and verb will be coterminous, but we
could have other elements of structure in the verbal group, and we
could have a number of linked verbal groups at p, as in
N V & V N
He washed and polished the car
0Cf He washed,//and polished the car)
or linked nominal groups at s, as in
s
N N & N
Tom, Mary and Bill went out.
 *

CLAUSE AND SENTENCE 71

If we consider sentences with more than two clauses, such as
(i) I stayed in and read a book because I had nowhere to go and
no one to see
(ii) If Tom comes and I’m not back, tell him to wait
we find the same position in sentence structure. Sentence (i) contains
four clauses, independent + independent + dependent + depen¬
dent. But the structure of the sentence is not merely a concatenation
of clauses. The two independent clauses are linked by and and by
omission of s in the second. What we have, in fact, is one a element
consisting of two clauses, just as in:
N & N
John and Jim went out
we have one s consisting of two linked nominal groups. The analysis
should therefore be:
a

Ind & Ind

I stayed in and read a book structure: a (Ind & Ind)

The two dependent clauses are also linked, and there is omission of
the binding adverbial group and s in the second. The analysis should
be:
P
Dep & Dep
because I had nothing else to do and no one to see

structure: p (Dep & Dep)

The analysis of the sentence is therefore
3
a

Ind & Ind

I stayed in and read a book

Dep & Dep

because I had nothing else to do and no one to see

We are, however, already involved in problems of linkage and

ellipsis, and it may be profitable to digress at this point in order to
discuss these briefly.
F
72 SURFACE GRAMMAR

Linkage (i.e. ‘coordination’) poses two questions for the analyst:

what items are linked? and what is the structural status of the linker?
There are two types of linkage in English. In the sentences:
I don't know, in fact, what he is going to do
We could, therefore, keep it going for quite a time
We all sent presents, nevertheless
the items in fact, therefore, nevertheless link the sentences to what has
gone before. These are called sentence linkers. Typical realisations
are consequently, therefore, finally, however, nevertheless, so, thus,
then. This type of linker does not present many problems. There are
two important points to be made about them: (i) their position is
not fixed in clause structure, thus:
However, I don’t think we should go on with it.
I don't think, however, that we should go on with it
I don't think we should go on with it, however;
(ii) these linkers do not allow what is usually called ‘branching’;
this means the formation of a sentence with linked clauses. Thus:
He therefore disagreed ^
I
Cl
Consequently, he disagreed ^

Cl
Such items can be regarded as realising an adjunct element in clause
structure, and if desired can be superscripted ALink to distinguish
them from other adjuncts.
The second type of linkage, realised by the linkers and, or, but,
is quite different. The position of these linkers is fixed and they
permit ‘branching’, thus:
£
I came in and he went out Cl Cl

John played well but Mary was off form Cl Cl

It is with this type of linker that problems are encountered. First,
their status as constituents: if two clauses are linked in the above
sentences, they are both linked. It is quite arbitrary to say that the
and belongs to either the first or the second—any two concatenated
items stand in the same relation to the concatenating item. But if we
decide on a ternary division of such sentences, e.g.
 CLAUSE AND SENTENCE 73

Ind Ind
I came in and John went out

what is the status of andl One of the requirements of the theory on

which systemic grammar is based is that all items should be accounted
for at all ranks, with nothing left over. Thus the description
John shot Jim -> one sentence, one clause, spc, nvm, hlh, noun
verb noun, base base base,
accounts for every item at all ranks with nothing left over. If we
apply this to I came in and John went out we have clauses I came in,
John went out; but we have and left over. If we are required to account
for this item at this (sentence) rank, then it will need to be a clause.
This is a consequence which is surely counter-intuitive and undesir¬
able. There is, in addition, a further problem: such linkers may link
items at any rank; and this means that we should be required to treat
linkers as realisations of units at each rank. Thus:
I play chess and he plays draughts and — clause
I play good chess and poor draughts and — group
I play good and interesting chess and = word
Fish and chips are good and = morpheme.

And in all of this we would have said nothing general about the
function of and, which is surely the same in all examples.
It seems, in fact, that linkers should not be regarded as constituents
in structure at all, but as realisation of a rank-free system of linkage.
It means in this case that where linkage occurs this is marked (&)
but not accorded constituent status. Thus:
I came in and John went out structure: Ind Cl & Ind Cl
big fish and little fish structure: nom grp & nom grp
fresh and bright structure: epithet & epithet
fish ’n chips base & base
If we now consider a sentence such as:
I stayed in and read a book because I had nothing to do and no
one to see but John went into town to see a friend.
the analysis is
74 SURFACE GRAMMAR

3
a

Ind & Ind

/ stayed in and read a book

P
Dep & Dep
because I had nothing to do and no one to see

& a P
Ind Dep
but John went into town to see a friend.

Two points are evident from such an analysis: (i) clauses are not
merely strings of coordinated or subordinated clauses; clauses operate
at elements of sentence structure, the analysis
^-> a(3 & a(3
reflects the structure of the sentence better than
^ —>■ Ind & Ind Dep & Dep & Ind Dep.
(ii) it is evident that within one alpha element ellipsis of s is pos¬
sible:
I stayed and read a book
and within one beta element ellipsis of binding adjunct and subject
is possible:
because I had nothing to do and no one to see.
And it can be seen from such examples that a clause cannot always
be identified as operating at alpha or beta by its own structure. The
clause
I had no one to see
is a dependent clause at beta in a finked series of dependent clauses
at beta in:
because I had nothing to do and I had no one to see,
but is an independent clause operating at alpha in:
(i) I had nothing to do that day, and no one to see (finked series)
(ii) I had no one to see (simple sentence)
We might say that the s of the first clause in a finked series at
alpha or beta is the s of the alpha or beta element, and the binder in
the first beta clause is the binder of the beta element. When we
change s or binder we have a new element of sentence structure.
Thus:
 CLAUSE AND SENTENCE 75

3
a

Ind & Ind

I came in and read a book.
but
3
a & a

Ind Ind
I came in and John went out.

3
P a

Dep & Dep Ind

IfVm out and he phones, give him a message.

but

P P a

Dep Dep Ind

If Vm out when he phones, give him a message.

It might seem unnecessary to separate the dependent clauses in this

last example (if all conditioning, all additioning, all reported clauses
constitute elements of sentence structure); but it is not the case that
if and when are exactly the same conditioning. Since the depth of the
clauses is important (we cannot reverse these without an intonation
break), this suggests that we have two conditioning elements, two
beta elements, rather than two clauses at one beta element.

MINOR SENTENCES

Sentences which do not contain an independent clause which con¬

tains a P element in their structure may be classed as minor sentences;
such sentences will be realised by moodless or dependent clauses.
Minor sentences are sometimes added, as a class, to the ‘statement,
question, command’ contextual classification of sentences. It seems,
however, that a morphological classification of minor as a sentence
which does not contain an independent clause with a p element is
more valid, easier to apply and allows for the fact that most minor
76 SURFACE GRAMMAR

sentences can be assigned to a contextual class, though this necessi¬

tates the addition of an ‘exclamatory’ class. Thus:
Shoot! command
On Saturday? question
Blast it! exclamation
Next week, statement.
There are three main types of minor sentence in English, for which
the usual names are:
(i) completive sentence
(ii) exclamatory sentence
(iii) aphoristic sentence.
(i) The first is called completive in the sense that it completes the
cotext (the surrounding language) or the context (the relevant situa¬
tion). Such completive sentences occur typically as answers to
questions and additional comment to previous statements, or as
situational comments such as introductions, directions, instructions.
Such sentences are like major sentences in that they form an open
set, which means that new instances can be produced at will, and
they have intonation patterns like major sentences. They exhibit
relationships with other sentences in the discourse. Thus:
(Who did it?) John.
(Is it true?) Yes I no)perhaps.
(When is it?) On Tuesday. Next week. After dinner.
(Introduction) Mr. Black. Mr. White. The Jones boys.
(Titles) West Side Story. East of Eden.
(Directions) Straight down the street and across the bridge.
(ii) Exclamatory sentences also have the characteristics of an open
class, but do not incur relationships with other parts of the discourse.
Such sentences are usually divided into primary interjections such as
Ouch! Gee! Whew! Blast! Hi.
Gosh. Golly. Hello.
and secondary interjections in which there is more than one element,
such as
dear me. bless you. thank you.
blast it. good heavens, by thunder.
A third type of exclamatory sentence is the verbless equational
sentence such as:
John a manager!
Thompson a first team man?
 CLAUSE AND SENTENCE 77
(iii) Aphoristic sentences differ from the previous types in that they
form a virtually closed class. They are traditional expressions, and
the native speaker is not normally at liberty to coin new examples.
They operate, however, as full sentences. Thus
The more the merrier.
Least said soonest mended.
One man one vote.

ANALYSIS OF MINOR SENTENCES

Minor sentences which are realised by moodless clauses have either

nominal groups or adverbial groups as constituents; since there is no
p element by which s/c relations may be established the nominal
groups can be regarded as realising the element z in clause structure,
and all adverbial groups can be regarded as realising the element a
in clause structure. This means that we are identifying the constituents
morphologically, and correlating as far as possible with their normal
syntactic operation in clause structure. Thus:
z z z z
John the captain! The more the merrier.
z z z
e e h h h
West Side Story. Joe Smith. Gosh!

A A K

P c P c P c
On [Tuesday] About [nine.] For [a minute.]

Minor sentences which have dependent clause(s) as constituents

will, of course, have normal clause analysis. Thus:
ASP
{When are we going?) When John returns.
Minor sentences are frequent in some types of literary work. These
will obviously be considered as completive sentences, but this does
not mean that they should be analysed by ‘supplying’ (finite) verbal
or (subject) nominal elements. They occur in the text as minor
sentences, and the fact that they do has obvious stylistic rami¬
fications. It is therefore not proper to treat them as ‘elliptic/frag¬
mented’ sentences and suggest that the ‘full’ form must be accounted
for.
78 SURFACE GRAMMAR

Two short extracts may illustrate the point:

(i) Fog everywhere. Fog up the river, where it flows among green
aits and meadows; fog down the river, where it rolls defiled
among the tiers of shipping, and the waterside pollutions of a
great (and dirty) city. Fog on the Essex marshes, fog on the
Kentish heights. Fog creeping into the cabooses of collier-brigs...
Verdict accordingly. Accidental death. No doubt. Good after¬
noon.
(C. Dickens: Bleak House)
(ii) It looked like a miniature bank. A short counter. A grilled
window protected by burglar proof plastic.
. . . and there was a travel office! A sure out.
(A. Bester: The Demolished Man.)
The various minor sentences realised by moodless and dependent
clauses must be analysed as discussed above, e.g.
z A
Fog up the river.

s p A
Fog creeping into the cabooses of collier-brigs.
z z
e h d e h
Accidental death. A sure out.

RANKSHIFT

Some discussion and illustration of rankshifting have already been

entered into in the previous pages. It may be of value to consider
rankshift as a separate aspect, and to illustrate again the main types
of rankshift which occur.
The theory on which the model of description is based allows for
a downward rankshift. This means that a unit which normally
operates in the structure of the next highest unit (and this is a crucial
criterion in the definition of units) may also operate in the structure
of a lower unit or in the structure of a unit of the same rank. A unit
so operating is said to be rankshifted.
Clause is that unit which operates at elements of sentence struc-
 CLAUSE AND SENTENCE 79
ture; but it has been shown that clause may also operate at q in
nominal group structure. Group is that unit which operates at
elements of clause structure; but it has been shown that groups
may also operate at q in other groups, and at d in other groups.
There are, in fact, five main types of rankshift which occur in
English, and it may be useful to illustrate these separately.
1 Clause rankshifted to operate at q in nominal group structure:
(i) With finite verbal group at p:
(a) who, which
The man ['who came to dinner] stayed a month
The box [which he thought most suitable] was already taken
(b) where, when, why
The place [where John lived] was Paris
The day [when they were married1 was Tuesday
The reason [why they left} was the lateness of the hour
(c) that
The colour [that I like] is red
(d) -
The boy {you like] is here
(ii) With non-finite verbal group at p :
(a) infinitival: The person [to ask} is John
(b) participial: The room \ facing South} is cold.
2 Clause rankshifted to operate as whole nominal group (nominal
clause):
(i) With finite verbal group at p:
(a) who, which
[ Who the man was} is a mystery
The results showed which theory was right
(b) where, when, why, how
[ Where the river bends} is a nice spot
Tuesday was [[when his plan worked}
{How (why) John's plan works} is important
(c) that
[That he should sing} is an outrage
(ii) With non-finite verbal group at p :
(a) infinitival: I like {to be left alone}
(b) participial: | Driving fast} is what I like
{Running a race} is good fun
John is la man driven crazy}.
80 SURFACE GRAMMAR

3 Adverbial group rankshifted to operate at q in nominal group

structure:
The houses [by the river] will not last long.
In the identification of this rankshift, care must be taken to distin¬
guish between rankshifted adverbial group at q and non-rankshifted
adverbial group at a in clause structure, thus:
p
bought the house [by the river]
p c
left my clothes by the river.

4 Nominal group rankshifted to operate at d in nominal group

structure:
[The town council]'s houses
[The house on the hill]’s four green doors.

5 Nominal group rankshifted to operate at c in an adverbial

group:
P c
by [the river]
P c
up [the hill]
In an example such as
The houses by the river are to be sold
the nominal group the river is rankshifted to c in the adverbial group
by the river, and the adverbial group by the river is rankshifted to
q in the nominal group The houses by the river. We obviously have
double rankshift, thus:
<
___

m h
II

P c=[ N]
The houses [by [the river]] . . .

If the nominal group the river had itself a further q element this
would involve multiple rankshift, a feature of some English registers.

SENTENCE ANALYSIS

It is evident from previous illustration's that the analysis of a sen¬

tence (or of any other unit) may be presented in the form of a
‘box’ analysis as in Figure 3 or a ‘tree’ analysis as in Figure 4.
 CLAUSE AND SENTENCE 81
There are, of course, various ways of presenting analyses; the chief
virtues of the two methods mentioned are that they show the step-
by-step description down the rankscale, and that they preserve
the sequence of the text. This means that we do not get (as in some
‘column’ analysis) a sentence analysis presented as
the on sharpened
soldiers impaled stakes prisoners their
heedless of
their pain
for The soldiers, heedless of their pain, impaled their prisoners on
sharpened stakes.
The ‘tree’ type of presentation also facilitates comparison of
sentences, and measurement of the relative complexity of sentences,
since the number of branching ‘nodes’ will be a reflection of the
complexity of the sentence and in particular of the amount of rank-
shifting involved. It is also possible to curtail the analysis at any
point on the rankscale, and the analysis will be valid and com¬
prehensive to that point; we may for example wish to analyse to
clause rank, thus:

$
a P
Ind Dep

s p A s p
The place where I was born was rebuilt after it had been bombed

Some illustration of these types of presentation of analysis is given

in Figures 5-9, using the following sentences as examples:
(i) He struck a blow.
(ii) All the boys called him a good player.
Simple <
(iii) In the evening, Tom and Dick will visit the man who
lives there.
82 SURFACE GRAMMAR

(iv) When he comes, give me a ring.

(v) The company, who now employ seventy persons,
Compound <
hope that work will begin at once and that the
factory will be ready in time.

figure 3

3
a
Ind
s p c A

N V N Ad
d h / d h P c
[N]
det noun verb det noun prep d h
det noun
The dog bit the boy in the arm

FIGURE 4

noun

dog bit arm

 *

CLAUSE AND SENTENCE 83

FIGURE 5

(i) He struck a blow.

a
Ind
s p cE
N V N
h / d h
pron verb det noun
He struck a blow.

a
1
1
Ind
1 ^
1
S p cE
| 1 1
1 1 1
N V N
1
h
1
/ d
/\ h
1 I I 1
1 1 1
pron verb det noun
I 1 1
1
He struck a blow
84 SURFACE GRAMMAR

FIGURE 6

(ii) All the boys called him a good player.

3
a
Ind
s p cE c1
N V N N
dx d2 h / h d e h
det det noun verb pron det adj noun
All the boys called him a good player.

det det noun verb pron det adj noun

1
1
all the boys called him a good player
 CLAUSE AND SENTENCE 85

FIGURE 7
(iii) In the evening, Tom and Dick will visit the man who lives there.

a
Ind
A s P cE
Ad N & N V N

P c h h a / d h q
prep pr pr aux verb det noun ttcm
noun noun
s p A

[N] N V Ad
d h h / a
det noun rel verb adv
In the evening Tom and Dick will visit the man who lives there

a
 86 SURFACE GRAMMAR

FIGURE 8

(iv) When he comes, give me a ring.

0 a

Dep Ind

A s p p cE cE

Ad N V V N N

a h / / h d h
adv pron verb verb pron det noun
When he comes give me a ring.

p-
I
Dep

ASP cE
I
Ad
I
N
I
V V
I
N
I
N
I
a
I
h
I
l /
I
h d
/ '
'h

adv pron verb

I I I
verb pron det noun
I I
When he comes
I .1
give me a ring
 *

o;
<u
v
.*0
t*.

as
w
e5
O
O
£
G

*
 *«■ • » Lilli

The company, who now employ seventy persons, hope that work will begin at once and that the factory will be ready in time
 3
Deep Grammar

Part 2 was intended to provide a description of the occurring pat¬

terns of English sentences. Two points regarding the inadequacy of
this description were made in the General Survey:

(i) similar descriptions will be given to unrelated sentences, as

This coat buttons easily structure: spa
This knife cuts easily structure: spa
and dissimilar descriptions will be given to related sentences, as
John shot Jim structure: spc
Jim was shot by John structure: spa;

(ii) the description will not show what else may have occurred at
any particular place, i.e. from what set of possibilities the particular
occurrence was selected; for example: that a nominal group the
boy represents a choice of singular from the set of possibilities
singular/plural, that a verbal group bites represents the choice
present from the set of possibilities present/past/future.

A further point may be made. Sentences such as It was John who read
the book present some difficulties for analysis. The simplest analysis
would seem to be:
s
It was John who read the book.
But this involves awkwardness in the further analysis of c: John is a
proper noun and as such would not be expected to have a q element,
which would seem to be the only possibility for who read the book.
Usually It is said to be a ‘dummy’ subject and the ‘real’ subject is
90 DEEP GRAMMAR
John. This explanation is well motivated; it certainly seems that a
description which assigns John to subject function is desired. The
explanation of ‘dummy’ and ‘real’ subject goes part of the way to¬
wards explaining such structures, but does not show the relatedness
of this sentence to other sentences, such as:
John read the book.
The book John read (but not the play).
The book was read by John.
It must be shown that the original sentence was a choice from a
number of possibilities, a set called ‘ theme ’ below.
The set of possibilities available at any place specified by the
surface grammar is called a system, and the individual choices are
called terms or features in the system. Thus:
System Terms Environment ('place on the chain ’)
(“Singular
Number- Subject/Complement
-*t Plural
“Present
Tense- -Past Predicator
-Future
(“Indicative
Mood- Major Clause
Hi-Imperative
There are two points to be noted about systems:

(i) We need to specify the place of operation, the environment on

the chain, before we specify the choices available there: the assign¬
ment of systems presupposes the surface grammar already discussed.
Nevertheless, it may prove desirable to give a systemic description,
a feature specification, of expressions from which structure may be
derived rather than the other way round. This is because structurally
similar descriptions may be the realisation of different combinations
of systemic features, but systemic descriptions are unique. This point
is illustrated in the following sections.

(ii) Some systems are dependent on selection from other systems;

only verbal groups which have already chosen finite in the system of
finiteness can make a further choice from the system of modality;
selection of a term in the system of theme at clause rank is only
available after selection from the system of mood. There is, therefore,
a hierarchy of systems. The principle of‘deep’ and ‘surface’ gram¬
mar is evident in traditional formulations; when it is said that a
 DEEP GRAMMAR 91
clause is ‘indicative, finite’ etc., or a verb is ‘finite, present, active’
etc., this means that a place ‘clause/verb’ is the point of origin for
the choices ‘indicative, finite/present, active’. This is not, however,
made explicit; neither is the dependency of choices on prior choices.
We need to know two things: (i) what other choices were possible at
the same place, and (ii) what the ‘entry requirement’ was, i.e. what
previous selection or selections was or were necessary before this
choice was possible.

A full systemic description of English is not yet available. Clause and

group are ranks where systems are numerous and important. A
description of the most prominent systems in these environments is
given below.
It is perhaps appropriate to mention here the further implications
of the term ‘Introduction’ in the title of this book. The following
brief discussion explains what this book is intended to introduce.
It becomes increasingly evident that the grammar of a language,
at its deepest level, is concerned with the organisation of systems of
choices which contrast with each other, and which are therefore
mutually exclusive and mutually defining. At this level the question
of choice is a question of choice between different meanings.
These meaningful choices are related to the classes and structures
of surface grammar by a scale of realisation. Each systemic term
specifies a structural component: thus, the feature ‘indicative’
specifies that the element s will occur in clause structure; the term
‘declarative’ will further specify the sequence sp in clause structure;
the term ‘ operative ’ will specify the presence of a c element in clause
structure.
Thus we can say that the feature selection: ‘indicative; declarative;
operative’ is realised in clause structure as spc.
The structures of language are themselves functional; they occur
as they do because that is how they achieve effective communica¬
tion. These functions have been designated experiential, interper-
personal and textual.
The functions of language may perhaps be most broadly and
generally expressed as ‘meaning’, ‘mood’ and ‘message’.
Meaning in this sense is concerned with matters of experience
(whether ‘real’ or not), and is most clearly related to the transitivity
systems, to the participants and processes involved. Mood in this
sense is concerned with the speaker’s role (questioning, stating,
92 DEEP GRAMMAR
commanding, etc.), and is clearly related to the mood and modality
systems, but also to the modal operation of certain adjuncts. Message
in this sense is concerned with the distribution of information points
and highlights throughout the discourse, and is most clearly related
to the theme and information systems (though the scope of such
systems is only touched upon in this book). Halliday has designated
these functions of language ‘experiential, interpersonal and textual’.
The components of a grammar must be related to these functions
of language. This is shown in Figure 10.

figure 10

Experiential Interpersonal Textual

Clause Transitivity: Mood: types of Theme:

process, speech function; identification,
participants, modality; tone predication,
circumstantial substitution of
elements, theme
polarity

Verbal Verb classes: Modality; Voice; contrast;

group tense marked polarity substitution and
ellipsis
Nominal Noun and Modification Deixis and
group adjective classes; intensification anaphora;
modifying pronouns;
functions comparison;
—epithet, substitution;
quantifier, ellipsis.
possessive

For further exemplification and description see the following

articles by M. A. K. Halliday: ‘Functional Diversity in Language
as seen from a consideration of Modality and Mood in English’ in
Foundations of Language Vol. 6, 1970; 322-361; and ‘Language
Structure and Language Function’ in New Horizons in Linguistics
ed. John Lyons, 1970; 140-165.
 SYSTEMS AT CLAUSE RANK 93

3.1 Systems at Clause Rank

Any clause which is major and independent must select from the
systems of mood, theme, transitivity and information which are
discussed individually in the following sections.

3.1.1 Mood
There are two points to be noted initially about this system:
(i) it is not to be confused with modality, which is a system whose
environment is the verbal group, and whose features are realised by
modal verbs;

(ii) traditionally, features of mood (or, more precisely, the realisa¬

tion of features of mood) are assigned to a form of the verb, and a
clause containing such a verbal form is said to express a ‘state¬
ment, question, command’ according to which form of the verb
occurs.

It is evident from the earlier discussion of verbal inflexion that a

‘form of the verb’ will not itself indicate features of mood; thus,
take is said to realise the imperative mood of the verb take, but
does not differ from the base form of the verb in any of its various
functions. Semantic definitions which rely on interpretation of
clauses as ‘statement, question, command’ and so on are more
appropriate to sentence rank, and certainly some more formal
indication of features will be expected at clause rank. There are
sentences which might reasonably be interpreted as ‘commands’
but which may not contain what we want to call an imperative
clause. There seems little point in talking about ‘indicative, sub¬
junctive, imperative ’ moods of the verb, and then saying that the one
verbal form take realises all three features. It seems clear that
features of mood are realised at clause rank, not at verb or sentence
rank: the mood features are realised, in fact, by particular arrange¬
ments of (some of) the elements of clause structure; and features of
mood are not in a one-to-one relation with contextual classes of
sentence. If we consider the clauses
(i) John comes
(ii) Is John coming?
(iii) Come
we see that (i) and (ii) have a s element but (iii) does not. Clauses
which contain a s are called indicative, clauses which do not contain
94 DEEP GRAMMAR
a s are called imperative. The first choice in the mood system may
thus be represented:
Mood-^["Indicative
LImperative
This, of course, is to beg the question of subjects in imperative
clauses. Traditionally it is said the second person pronoun (you) is
—overtly or covertly—present in all imperative clauses, and is the
subject of such clauses. The evidence for this point of view can be
presented very convincingly. Referring to this problem, O. Jespersen
(1940; Vol 3: 224) states:

A subject may be added, chiefly to give emotional colouring to the

imperative; often with some emphasis on the pronoun (and with
a pointing gesture).

This type of clause is exemplified by:

You, come here
Come here, you.
Elsewhere, (1940; Yol 5: 467) Jespersen states:

We frequently find the imperative with a third person subject.

Sometimes this may be considered a vocative, as manifested by
the comma which modern editors insert.

This type of clause is exemplified by:

Come here, somebody
Somebody, come here.
Jespersen seems correct in adducing ‘vocative’ and in drawing atten¬
tion to the comma. He seems less correct in stating such elements
may ‘ sometimes ’ be regarded as vocatives (how do we define ‘ some¬
times’?) and in calling them subjects. They are certainly nominal
groups, and may be as complex structurally as any nominal group.
They are clearly not complements, and so may indeed be considered
as subjects at first. But they do not behave like other subjects, in
that:

(i) they are often (and may be always) separated from the rest of the
clause by a comma in the written language and an intonation break
in the spoken language. As in
Boys, stop playing when it gets dark
(cf. Boys stop playing when it gets dark).

(ii) Not only may such an element be separated from the rest of the
 SYSTEMS AT CLAUSE RANK 95
clause in this way, but also it may appear in clause-final position.
As in
Stop playing when it gets dark, boys
(cf. *Stop playing when it gets dark boys).
(iii) Since imperative clauses cannot occur with certain (‘past¬
time’) adverbial groups, there is no possibility of an imperative
interpretation of:
Mother let us go to the pictures last night
(cf. Mother, let us go to the pictures tonight)
And because of restriction on ‘place’ adjuncts, there is no likelihood
of an imperative interpretation of:
Boys stop playing when it gets dark in China.
It seems best to consider such nominal groups as vocative ele¬
ments ; if anything is consistently present in imperative clauses it is the
feature vocative. The realisation of imperative is absence of subject,
and possible presence of a z vocative (zvoc) element in clause structure.
The initial choice in the mood system is thus between indicative
and imperative. If a clause is indicative, there is a further choice to
be made between declarative and interrogative. A declarative clause
has its subject preceding its predicator, as in:
s p
John is coming
p
The centre scored three goals
s c
He was called a saint.
An interrogative clause has either the subject included in the pre¬
dicator, or an initial wh element (i.e. who, which, where, when,
why, how), or both, as in:
p— s —P
Is John coming?
A P-_ s —p c
When will that centre score a goal?
s p c
Who will be the first to dive?
The Mood system may now be represented:
. rDeclarative
Indicative-
Mnterrogative
Mood-
•-Imperative
96 DEEP GRAMMAR

It is obvious that interrogative clauses may be either wh inter-

rogatives or non-WH interrogatives. This distinction might alter¬
natively be labelled (as it often is) ‘yes/no’ or ‘non-yes/no’
interrogatives. In non-WH interrogatives it is the polarity (p. 137)
which is in question, the ‘yes-ness or no-ness’ of the clause, and the
predicted answer is ‘yes’ or ‘no’. In wh interrogatives it is not the
polarity which is in question, and the answer is not predictable in
any generalised sense.
Further, in wh interrogatives it may be the identification of the
subject which is in question or it may be some other element. If it is
the former then the wh element is the subject of the clause, and takes
the initial clause position usual for wh elements (not the included s
position for interrogatives), as in Who will be the first to dive? If it is
not the subject of the clause which is in question the wh element will
take initial position in the clause and the s will be included in the
p, as in When will he score a goal? Where shall we find another one?
The Mood system may now, and finally, be represented:

[r-WH->\
i-Interrogative-H
rSqu

l-Non-Squ
Indicative—m L-Non-WH
•-Declarative
Imperative
A problem in identification of mood features is occasioned by
clauses such as Here come my friends. In this clause the nominal
group my friends would appear to be subject, since a change of
number from plural to singular will necessitate concord with the
verbal group at p: Here comes my friend. Such clauses could be
considered a special type of declarative clause, marked by the presence
of the (unstressed) introductory adjunct. However, such clauses
occur without any introductory adjunct: said John. The main
difference between interrogative clauses and declarative clauses is
that the former have subject included in the predicator, the latter
have subject preceding the predicator; but if there is only one item
realising the verbal group at p then it is not possible to show the dis¬
tinction. There would seem to be possible ambiguity of mood
features. That is, a clause such as has John may, on what has so far
been considered, be either declarative or interrogative. It is ob¬
viously necessary to distinguish between ps declaratives and ps
interrogatives. The mood feature realisation depends on the item
 SYSTEMS AT CLAUSE RANK 97
realising the verbal group: if the item is an auxiliary verb then the
clause is interrogative; if the item is a lexical verb then the clause is
declarative. Since ps interrogatives contain only an auxiliary verb
their occurrence will be restricted in the same way as the occurrence
of non-interrogative clauses with only an auxiliary verb in the
verbal group i.e. will be a case of substitution (pp. 42-43). Wemay
therefore incorporate the description of these ps interrogatives in the
description of auxiliaries, and consider other ps clauses as declarative.
We can now say that the feature declarative is realised by either sp
or ps in clause structure, the feature interrogative is realised by psp
in clause structure, and the feature imperative is realised by P with
no s in clause structure. The following pairs of clauses differ in
only one mood feature, whatever their differences in surface
structure:
A B
John comes John, come
indicative imperative

The centre scored a goal Did the centre score a goal?

indicative: declarative indicative: interrogative

Who broke the window? Is the window broken ?

indicative: interrogative: wh indicative: interrogative:
non-WH

When did he buy the book ? Who bought the book?

indicative: interrogative: indicative: interrogative:
wh: non-Squ wh: Squ.

3.1.2 Theme
The preceding section was concerned with presence or absence of s
in clause structure, and with the various arrangements of s and p
as realisations of terms in the mood system. The position of a in
clause structure was discussed previously (pp. 63 ff.). The presence or
absence of c in clause structure is determined by the transitivity
systems, to be discussed in the next section; for the moment we may
say that if a c is present in clause structure it usually occupies the
position immediately after p. In other words, apart from ‘binding’
adjuncts—if, because, etc., and linkers—and, or—which are all
fixed in position and are not relevant to the theme system, we can
say that the usual order for elements of structure in a declarative
clause is s p c a. It is not the case, however, that this order of elements
98 DEEP GRAMMAR
is fixed in English; only that it is being taken as the most normal. We
can regard this as neutral or unmarked, and contrast other permitted
sequences against this norm. What is called the thematic element in
a clause is the first element which results from choice (i.e. excludes
the binders and linkers mentioned above, since their position is
fixed, and not a matter of choice).
We can, therefore, distinguish between clauses unmarked for
theme, and clauses which are marked for theme. An unmarked
theme will be realised by having the mood exponents in initial posi¬
tion; a declarative clause, for example, which has s in initial position
is unmarked for theme. A clause which has an element other than a
mood exponent in initial position will be marked for theme. The
primary distinction in the theme system is thus:
rMarked
Theme-
■-Unmarked.
We may now proceed to consider what different types of marked
theme occur in English.
The c in clause structure normally follows p, and if no A is present
it is clause-final. But the c may occur in clause-initial position. Such
structures are usually discussed in terms of‘inversion’; this name is
appropriate if it means inversion of the most usual order in English;
but if it is taken to mean (as it sometimes seems to be) that there is
some ‘breaking’ or ‘extending’ of grammatical rules involved it is
clearly inappropriate. It is often said that English word order is
fixed, but when examples are given to illustrate the point it is obvious
that it is not ‘ words ’ that are meant, but elements of clause structure
—spc. The elements of clause structure are variable in sequence,
and it is precisely because they are variable that values can be given
to certain sequences in contrast to other sequences which might have
occurred.
Clauses with initial c are perfectly grammatical in English. Thus:
c s p A
Football /’ ll play anytime
c s p A
Him the almighty power hurled headlong
s
These papers you can have (but leave those).
These clauses have marked theme, marked by initial position of c.
This is called c-theme (thematic c).
The possibility of A-theme is similar. The ‘lexical’ adjuncts which
 SYSTEMS AT CLAUSE RANK 99
usually occur at position 3 (p. 64) can occur in initial position in
clause structure. The clause is marked for theme by this position of
adjunct. Thus:
On a former occasion we had discussed this
In the evening, we went down to the beach to play
Never has so much been owed f ASP
On the occasion of this anniversary, we drink to .. .
The third of these examples shows that position 2 adjuncts can also
be thematic, and when they are the subject of the clause it is in¬
cluded in the predicator; other examples are:
Seldom have I seen so poor a game
Rarely did I find anything interesting there.
This does not cause ambiguity with the included subject interroga¬
tive clause, because of the thematic adjunct marking the clause.
Often a thematic A correlates with an intonation break in the spoken
language, and this is reflected by the commas in the above examples.
In a declarative clause which is unmarked for theme the subject is
usually ‘ given ’ in that it has already been mentioned in the text or is
known in the situation. It is also usual for the tonic syllable (primary
stress) in a clause to fall on the last lexical item in the clause. This
is what we should expect: if the subject is ‘given’, is the theme of the
clause, then some other element is ‘new’ and so receives the high¬
light afforded by the tonic syllable. So in the clause John read the
book (i.e. not the play) (tonic ‘book’), John is already known and
book is the ‘new’ information and takes the tonic syllable. A subject
can, however, be marked as ‘new’ information by taking the tonic
syllable. John read the book (i.e. not Mary or Bill). This, indeed, is
only one of the possibilities of marked tonicity (pp. 101 Jf.); and could
be included in the information systems. It seems in place here as a
foregrounding of an element of clause structure similar to the fore¬
grounding of c and a discussed above. It is also the case that s-new
often correlates with the fourth, and final, term in the marked theme
types: predicated theme. Some discussion was entered into above
regarding the difficulty of a satisfactory analysis of It is John who
read the book. If we consider this to be the selection of predicated
theme as against non-predicated theme (i.e. John read the book),
then we can say that it is realised in surface structure as
It + ‘be’ -f s (of the non-p theme clause) + wh (i.e. appro¬
priate relative) + p (of the non-P theme clause) + c (of the
non-P theme clause).
100 DEEP GRAMMAR

And if we relate these clauses in this way they will be shown as closely
related, as differing in only one systemic feature:
John read the book indicative: declarative: non-p-theme
It was John who read the book indicative: declarative: p-theme.
We have, therefore, four possible types of marked theme in the
clause:
fs ‘new’
p-theme
rMarked- >
•c-theme
Theme- > •A-theme
-Unmarked.
The theme system applies to interrogative clauses also, and
declarative and interrogative clauses can be considered together in
relation to theme. It may, however, be clearer to add a separate
statement of interrogatives. In wh interrogative clauses, whether
Squ or non-Squ, the wh element normally occurs in initial position.
Thus:
Who is coming?
When is he coming?
Initial wh element is therefore unmarked for theme in interrogative
clauses. An interrogative clause may be marked for theme, if either:

(a) the wh element occurs in other than initial position

or
(b) the wh element is initial but takes the tonic syllable; (this is
usually called an ‘echo’ question, since it is a request for informa¬
tion to be repeated).
Thus:
You’re going where ?
Who is coming?
It is obvious that with a Squ only the latter possibility of marking
is possible, since John shot whom? will always be interpreted as
spc. This provides an illustration of the fact that discussion of ‘case’
in English (by reference to the few remaining pronominal inflexions)
as marking clause functions of s and c are misguided; this type of
clause would be possible if even the pronominal cases were effective.
We have the following possibilities of marked theme:
Unmarked Marked
rWhom did he shoot?
Whom did he shoot 1-
t He shot whom ?
 SYSTEMS AT CLAUSE RANK 101
_ rWhere did he go?
Where did he go ?-
•-He went where ?
Whom is presumably ‘accusative’ case, and marks ‘object’ function
in the clause in contrast to its ‘nominative’ form (who) which marks
subject function in the clause. Where is a non-inflexional adjunctival
element which does not affect the point. If the uninflected ‘ nomina¬
tive ’ case of whom did mark ‘ subject ’ then a clause such as * Who shot
John ? ’ ought to have the same thematic possibilities as the previous
examples.
Who shot John-►!"Who shot John?
L(John shot who ?)
That this is not so, that John shot who will always be interpreted as
spc, with John as ‘actor’ and who as ‘goal’, is a reflection of the
fact that order of elements of clause structure is a stronger realisa¬
tion of function in the clause than is ‘ case ’ marking.
We might diagram separately for interrogatives:

[w . ,
Marked-►
twh non-initial
Lwh tome
Unmarked

3.1.3 Information
Connected speech is a sequence of stretches marked by phono¬
logical breaks. A. A. Hill (Introduction to Linguistic Structures,
New York, 1958) gives the example:
He will act roughly, in the same manner
He will act, roughly in the same manner,
and rightly observes that the commas in the written language reflect
the occurrence of intonation breaks in the spoken language. A
stretch of language between intonation breaks is a tone group.
Within any such tone group there will be one tonic syllable (‘pri¬
mary stress ’). Thus:
He will act roughly, in the same manner
He will act, roughly in the same manner.
Reaching over each tone group, and having its distinctive move¬
ment at the tonic syllable, is one of the possible tones in English.
There are thus three factors to be considered:
‘ stretches ’-►tone groups (tonality)
‘ stress ’-atonic syllable (tonicity)
‘ voice inflection ’-►tone (tone)
102 DEEP GRAMMAR
Such features can be established in phonology, without reference
to grammar. Difficulty in linguistic description is occasioned if such
features of intonation are taken to be direct exponents of gram¬
matical units and classes. The most common statement of this kind
is perhaps that one tone group = one clause. If we consider the ex¬
pression I didn’t go because he told me we know this may occur with
an intonation break after go:
I didn’t go 11 be cause he told me
or without such a break:
I didn’t go because he told me I /.
The difference is a meaningful difference; but it is not a difference
between two clauses and one clause. In either case the sentence
contains two clauses, an independent clause followed by a depen¬
dent clause. So with grammatical classes; it is often said that a
‘rising tone’ signals a question, i.e. an interrogative clause. How¬
ever:
(i) Is John doing it?
(ii) When will John do it?
(iii) John is doing it
may all be said with a ‘falling tone’, but (i) and (ii) are interrogative
and (iii) is declarative.
We can consider phonologic patterns from the ‘phonetic’ point
of view, and discover what phonological patterns are in fact avail¬
able in English; or we can consider phonological patterns from the
grammatical point of view, and discover what grammatical dis¬
tinctions are realised by features of intonation. The latter approach is
adopted here. The distinctions marked by intonation features are
taken to be meaningful and to be grammatical; they are just as much
formal contrasts as are those between (say) singular/plural, past/
present, indicative/imperative. A full description of intonationally
expounded systems would assign such distinctions to the appropriate
places in the grammar; some are concerned at group level, with
different classes of group, some at clause level and so on. The
primary distinctions realised by tonality and tonicity are glanced at
below. No attempt is made at a comprehensive description even at
primary delicacy; the intention is to show how such intonationally
realised choices are accommodated in the grammar (but see also
nominal group structure (pp. 33 Jf.), and the systems of contrast and
focus in the verbal group systems (pp. 139 Jf.)).
Tonality is the distribution of tone groups in an utterance; where
 SYSTEMS AT CLAUSE RANK 103
such stretches begin and end. Tonality is obligatory in that speech
will necessarily contain tone groups, but the speaker has the option
as to where they will occur. There is a high correlation between tone
group and clause in English—though this, as was said earlier, should
not be raised to a definition of clause; indeed, it is only after a clause
has been identified that such a statement is possible. This correlation
can be utilised, and one tone group being coterminous with one
clause can be regarded as neutral tonality. Thus:
I didn’t go//because he told me.
There are then two marked possibilities; a tone group may be more
than a clause or less than a clause. The former occurs frequently
when a reporting followed by a reported clause occurs in sentence
structure, as in:
I asked Tom how it was done//
I told him why we did it that way//.
The second possibility, tone group less than a clause, occurs fre¬
quently before adjuncts in clause structure, as in:
He had become a teacher//in defiance of his father
He will act roughly/jin the same manner.
This is also very frequent with thematic adjuncts, as in:
On this great opening day//we must pay tribute to .. .
After a sumptuous meal,//we went back to the . . .
These stretches are called information units; so an information
unit equal to one clause is an unmarked information unit; an
information unit equal to less than or more than a clause is marked.
We can also recognise a neutral or unmarked term in tonicity.
When the tonic syllable is located on the last lexical item in the clause
then the tonicity is unmarked. Thus:
I meet him in the square
I meet him.
Marked tonicity occurs when the tonic falls on some other lexical
item or on a clause-final grammatical item. Thus:
/ meet him in the square
I meet him
It may be alright.
We can summarise this brief consideration of intonation features
by saying that there is a system of information which is concerned
with the distribution of information units (i.e. the location of tone
groups), and with the focusing on a point of information within each
such group (i.e. the location of the tonic syllable). Thus:
104 DEEP GRAMMAR

r Unmarked

r Distribution-*- 0 unit = 1 clause>

■More (than a

Information-

[-Unmarked (tonic on last lexical item).

-Focus
[
LMarked (tonic not on last lexical item).

3.1.4 Transitivity9
On pp. 57 ff. the unit clause was discussed in terms of transitivity;
this related to the presence or absence of the element c in clause
structure. The scale-category definition of transitivity is exemplified
in the following quote from Halliday:
The terms ‘intransitive, transitive’ refer to clauses with no c,
at least one c, one c and two c’s respectively.
The main difference between this and traditional accounts is that
here transitivity is being accounted for at clause rank, but traditionally
it is accounted for as a feature of the verb, i.e. at word rank (or, at
best, at (verbal) group rank). The traditional view, as expressed by
Treble and Vallins (An A.B.C. of English Usage, Oxford, 1936) is:
A verb is said to be used transitively when the action or state that
it denotes is regarded as ‘going over’ to, i.e. as directed towards,
an object. When the action or state is regarded as affecting only
the subject, i.e. as not directed towards an object, the verb is
called intransitive; in ‘birds fly’ the verb is intransitive, in ‘boys
fly kites’ it is transitive. Many verbs can thus be used transitively
or intransitively.
The difference between the approaches is not a trivial one. Consider¬
able research work is currently being carried out on the numerous
problems connected with transitivity in English. It is now recognised
that the surface concept of transitivity is the environment for a net¬
work of systems. The relationships between nominal elements of
clause structure (s and c) and verbal elements of clause structure
(p) are various, and are realisations of relationships between ‘par¬
ticipants’ and ‘processes’.
It is evident that the form of the verb alone will not indicate
whether it is transitive or intransitive in Treble and Vallins’ terms:
 SYSTEMS AT CLAUSE RANK 105
in their examples birds fly, boys fly kites, the identification of fly as
transitive or intransitive is not made by reference to the form of the
verb fly; it is necessary to observe whether or not there is a c present
(an ‘object’ to which the action ‘passes over’). Since the identifica¬
tion of transitivity types thus requires the recognition of clausal
relations (pc or not pc) then it seems more appropriate to consider
transitivity as a feature of the clause rather than a feature of the
verb.
In any case, the many problems connected with what underlies
such clausal relationships are not revealed by such a definition as
that above. It is not, for example, evident that in
Boys fly kites
He threw the ball
He turned out a great player
quite different features are realised by the elements s and c, though
all the clauses are transitive.
Consideration of such problems perhaps depends on what we want
our grammar to account for; if we want a description of the physically
realised events of language, the various ‘ states ’ in which we find the
language, then a surface description of transitivity is perhaps
sufficient. But if we want to (attempt to) account for the relations
between the various ‘states’ of language, for what underlies the sur¬
face realisations, then something further is required.
Not only the problems posed by the different features realised
(the different ‘ roles ’ fulfilled by s and c) in clauses such as
He sank the boat
He eats fish
He became a teacher,
but the fact that He cuts easily is interpreted differently from He
cut dexterously and He shaves well is interpreted differently from
That barber shaves well is involved in the elucidation of problems of
transitivity.
It has been observed that in the clauses:10
(a) John polished the table
(b) i. The table polished well
ii. The table was polished (by John)
iii. The table was highly polished
(a) is the exception in terms of sequence. It seems reasonable to
assume that the relationship between the table and polished is con¬
stant, though in (a) the table is c and in (b) the table is s; and further,
106 DEEP GRAMMAR
that the relationship between John and polished is constant, though
in (a) John is s, and in (b) ii. John is optionally present in an a. It
follows that if the table and John are realisation of ‘participants’
connected with a process (polishing)—realised as the p of the clause—
then neither of these roles is tied to any one element of clause struc¬
ture.
We may summarise the argument by saying that traditional notions
of ‘actor-action-goal’ in English clauses have good foundation, but
it is important to note that we cannot generalise and say that in
English s = actor, c = goal. There are other types of participants
than the two above; and it is not the case with these that ‘actor’ is
always realised as s and ‘goal’ is always realised as c: it is only the
case that s = ‘actor’ and c = ‘goal’ in a particular type of clause,
and in another type of clause the s may realise the ‘ goal ’. What is
required is an account of different English clause types according
to the relationships realised by the nominal and verbal elements in
clause structure.
Such an account is, in part at least, a description of the systems
for which transitivity is the environment. An attempt to discuss these
is given in the following pages, beginning with the more obvious
distinctions and proceeding thereafter to some more delicate
distinctions. The account is not intended to be comprehensive, but
to illustrate and describe some of the principal options available in
the transitivity systems.
The discussion involves the use of nine examples as clause types,
and these may be listed here for convenience, with the numbering
which they retain throughout the discussion:
(i) The barber shaved ten customers
(ii) Ten customers were shaved
(iii) The soldiers marched
(iv) Mary seemed happy
(v) The sergeant marched the soldiers
(vi) The soldiers were marched
(vii) The barber shaved (sc. ‘himself’)
(viii) The barber shaved (sc. ‘ customers ’)
(ix) The customers shaved (‘ easily ’)
The discussion is broken into five sections for ease of presentation.
Section 1 concerns the primary distinctions, and involves clauses (i)
to (iv); section 2 concerns more delicate distinctions, and involves
also clauses (v) to (vii); section 3 concerns the remaining two
 SYSTEMS AT CLAUSE RANK 107
distinctions, and involves clauses (viii) and (ix). Section 4 presents a
summary and discussion of the system network, and section 5 pre¬
sents a brief discussion of verb classification in relation to transitivity
systems.

SECTION 1

Presented with the clauses:

The barber shaved ten customers
Mary seemed happy
native speakers of English will agree that a ‘passive transform’ is
possible with the first, but not with the second. That is,
The barber shaved ten customers->Ten customers were shaved
(by the barber)
but not
Mary seemed happy->*Happy was seemed {by Mary)
This is a long recognised and accepted feature of English grammar,
and is included in even elementary grammar books. But it is im¬
portant to say that surface grammar alone will not account for this
relationship between
(i) The barber shaved ten customers
and
(ii) Ten customers were shaved.
Surface grammar will describe (i) differently from (ii), i.e. as spc
and sp(a) respectively. So sentences which native speakers feel are
closely related will be shown as different, and sentences which are
not felt to be closely related will be shown alike, since (i) will be
analysed in a way similar to
(iv) Mary seemed happy.
At this stage we might postulate two clause types, and call them
extensive and intensive; thus:
rExtensive (i, ii)
Transitivity-
•-Intensive (iv)
This is only giving names to well-known English clause types. We
might then look to see if this distinction is characterised by any overt
marking in surface structure. It may be that s is differently inflected
in the types—since it may be suspected that different types of s are
involved, one which can also occur (optionally) in an a (we may call
this ‘actor’), and one not so (we may call this ‘attribuant’). The s is
not so inflected in English; we can have:
108 DEEP GRAMMAR *

He shaved the customers

He seemed nice
in which the element s is the same in surface structure. In some
languages there would be an inflexional ending on the s in the exten¬
sive, marking it as the realisation of‘actor’.
We might consider a marking on the c; perhaps they are different
types of c, and are marked as such. In (i) above the c is a noun¬
headed nominal group; in (iv) Mary seemed happy it is an adjective¬
headed nominal group. This, then, may be the distinction. In fact,
this is a partial indication; an adjective-headed nominal group at c
will mark a clause as intensive, because adjective-headed groups
cannot operate at s (therefore no ‘passive transform’ is possible).
But it is not the case that this will characterise all intensive clauses,
because noun-headed groups can occur at c in intensive clauses.
Thus:
Mary seemed a beautiful girl
He became a great painter.
We might, finally, consider the p in the clause types. Initially
this would seem the most promising procedure. In the examples
above shave and seem are different lexical verbs, and many native
speakers want to account for the difference of clause types as ‘lexical’
differences in verbs and nothing more. If we consider further clauses
such as
He eats fish (extensive)
She turned pale (intensive)
we will get similar results. It seems, therefore, that different lexical
verbs characterise extensive and intensive clauses; and if we com¬
pile lists of which verbs occur in which types of clause this will give
identifying criteria. Thus:
Extensive Intensive
shave seem
eat turn
shoot become.
But besides She turned pale (which will not transform -> *Pale was
turned) we find She turned a page (which will transform -> A page
was turned). And similarly:
He got cold He felt very fit * She turned out a good girl
He got a present He felt the stone She turned out good work.
It is evident that the lexical verb alone will not identify all clauses
 SYSTEMS AT CLAUSE RANK 109
as extensive or intensive. There is, in fact, no single criterion by which
the distinction is marked. This is similar to what was said previously
about the ‘ singular/pluraT distinction (p. 19)—that there is no
single identifying criterion for the feature ‘plural’.
When we posit related features in a language, such as ‘singular/
plural’, ‘extensive/intensive’, we are making a hypothesis (which is
independent of the surface realisation of such features); this hypo¬
thesis must be tested as extensively as possible and, if validated, be
formulated into an explicit statement. Thus, at word-rank we estab¬
lish paradigms of nouns and so on, and this is a statement that the
various forms in the paradigms are related to each other (whatever
the difference in surface structure). What is being suggested here is
that there are paradigms of clauses also, from which a selection of a
particular form will be made in any instance. What follows may be
viewed as an attempt to build up clause paradigms.
The element c is obligatory in intensive clauses, and so these differ
from a clause such as (iii) The soldiers marched which has no c. So
(iii) would seem to be extensive and (iv) Mary seemed happy is
intensive. Thus:
pExtensive (i), (ii), (iii)
Transitivity- 1 Intensive (iv)
The difference between the extensive clauses (i) and (iii) is that in
(i) there is an active P with a following c, but (iii) has an active p
with no following c. In semantic terms, the process in (i) is involved
with two participants, an ‘actor’ (s) and a ‘goal’ (c); in (iii) there
is only one participant, the ‘actor’, concerned in the process. In (ii)
there are still (optionally) two participants involved, the ‘goal’
(now s, but still, of course, ‘ goal ’) and in a final (optional) adjunct,
the ‘actor’, (now in the a, but still, of course, ‘actor’). So (i) and
(ii) need to be distinguished from (iii). These different types are
called effective and descriptive. Thus:
rEflfective (i), (ii)
rExtensive-
•-Descriptive (iii)
Transitivity-
H., -Intensive (iv)
The difference between the extensive: effective clauses (i) and (ii)
is that (i) has an active p and a following c, but (ii) has a passive p
and optionally a following a, and has at s what was the c in (i). In
semantic terms, clause (i) has s as ‘actor’ and c as ‘goal’; clause (ii)
has s as ‘goal’ and optional a as ‘actor’. These clause types are
called operative and receptive respectively. Thus:
110 DEEP GRAMMAR

rEffective-^r0perat!ve 0)
[-Extensive-► LReceptive (11)
Transitivity—>\ ^-Descriptive (iii)
•-Intensive (iv)

If the distinctions proposed and outlined in the final diagram are

applied to the four example clauses, then each clause will have a
different systemic description, though they will not each have a
different structural description.
Example Structure Systemic description
(i) The barber shaved SPC extensive: effective:
ten customers operative
(ii) Ten customers were SP extensive: effective:
shaved receptive
(iii) The soldiers marched SP extensive: descriptive
(iv) Mary seemed happy SPC intensive
Such a systemic description will not only show similar surface
structures to be realisations of different clause types, as (i) and (iv),
but will also show the relative relationship of clauses of different or
similar surface structure; so the fact that (i) and (ii) share two systemic
features and differ only in a third suggests they are more closely
related to each other than either is to (iii), which shares only one
feature with each of them; but all three are contrasted with (iv),
which shares no features with any of them.

section 2

Indeed, if the clause types discussed in section 1 were the only clause
types in English, then the above description would serve to identify
each type uniquely. But we also find clauses such as:
(v) The sergeant marched the soldiers
(vi) The soldiers were marched.
In surface structure, these resemble clauses (i) and (ii), and are
transforms of each other. They can obviously be labelled operative
and receptive:
Operative Receptive
(i) The barber shaved ten customers (ii) Ten customers were
shaved
(v) The sergeant marched the soldiers (vi) The prisoners were
marched
 SYSTEMS AT CLAUSE RANK 111
There are, however, distinctions which show that while (i) and
(ii) are effective clauses as discussed above, (v) and (vi) are descrip¬
tive clauses, and are related to (iii) The soldiers marched.
Effective:operative clauses, such as (i), occur without an overt c
element, but are interpreted in just the same way as when the c
element is present. That is, clauses such as:
That barber shaves well
Mother washes on a Monday
occur regularly and are interpreted as meaning:
That barber shaves (people) well
Mother washes (the clothes) on a Monday
But clauses such as:
The sergeant marched
He sank
are not interpreted as meaning:
The sergeant marched the soldiers
He sank the boat
Semantically, native speakers agree that in the clause:
(i) The barber shaved ten customers
it is the barber who does the shaving', but in the clause:
(v) The sergeant marched the soldiers
it is the soldiers who do the marching (whether the sergeant does so
or not—the clause does not specify this). In other words, the actor is
realised as s in (i) but as c in (v). It is then evident that the sergeant
(i.e. the s) in (v) is fulfilling a different role, is realising a different
feature, from that in any of the other clauses. It is neither actor nor
goal: this feature is called initiator (or causer, reflecting the fact that
the combination of features (extensive) descriptive operative is
one way of expressing ‘causation’ in English). It would appear,
therefore, that there are three possible participant roles which can
be fulfilled by s in descriptive clauses. Thus:
(v) The sergeant marched the soldiers->s as initiator
(descriptive: operative)
(vi) The soldiers were marched->S as actor
(descriptive: receptive)
(iii) The soldiers marched->S as initiator/actor (i.e.
(descriptive: middle, the term combines the roles)
‘middle’ is used to distinguish
from operative and receptive).
Effective clauses, on the other hand, have:
112 DEEP GRAMMAR
(i) The barber shaved ten customers—>>S as actor
(effective: operative)
(ii) Ten customers were shaved->-s as goal
(effective: receptive).
The obvious question is whether there is a type of effective clause
which combines the roles of actor and goal in a middle form, as the
descriptive combines the roles of actor and initiator in a middle form.
In English we have clauses such as:
The barber shaved himself/He shaved himself
where the reflexive complement (cRe/) is treated as an extensive comple¬
ment in clause structure: the reflexive does not take the tonic syllable
in such clauses since it is a grammatical, not a lexical, item; it thus
differs from the ‘ emphatic ’ reflexive as in She asked the question her-
self (cf She asked herself the question). We also find clauses such as:
He shaved
He washed and dressed
which are interpreted as meaning:
He shaved himself
He washed and dressed himself
so that in:
(vii) He shaved
we have a combination of actor and goal in an effective: middle clause.
We can summarise the discussion thus far by saying that effective and
descriptive are different types of extensive clauses; both of these types
have an operative, a middle and a receptive form. Clauses which are:
effective:operative will have s as actor, c as goal
effective: middle will have s as actor and goal combined
effective deceptive will have s as goal and A as actor (option¬
ally)
descriptive: operative will have s as initiator and c as actor
(‘ causative ’)
descriptive: middle will have s as initiator and actor combined
descriptive: receptive will have s as actor and a as initiator
(optionally).
Diagrammatically:

Operative Middle Receptive

Effective 0) (vh) ' (ii)
Descriptive (v) (iii) (vi)
 SYSTEMS AT CLAUSE RANK 113
And the system network may now be represented as:
Effective
Descriptive
r Extensive—>-
■ Operative
Transitivity—> |_ -Middle
Middle
•Intensive •-Receptive
Receptive
This indicates that a clause which is extensive will choose between
effective and descriptive, and simultaneously between operative,
middle and receptive.

SECTION 3

There are only two further distinctions to be taken into account.

The first of these has already been mentioned. Clauses such as:
He shaved
Mother washes (on a Monday)
will be interpreted as meaning:
He shaved himself
Mother washes the things {on a Monday).
This means that the clause:
(viii) The barber shaves {well)
is extensive:effective and is operative, since it has s as actor. The only
difference between this clause and (i) is that in (i) we have a surface
realisation of goal (at c), but in (viii) we do not. This can be simply
represented as a choice between goal realised/goal unrealised. Thus:

(Extensive: effective: operative}

This choice presumes the prior selection of effective and operative:

it is not open to descriptive: operative, since the c in such clauses is
the actor; nor is it available to receptive clauses, since in such
clauses the s is the goal.
The final clause type involves clauses such as (ix) He shaves easily
meaning ‘ it is easy to shave him ’: such clauses are very common in
present tense, and in certain contexts. Thus:
They wash easily
These books sell well
This material wears well
In clause (ix) the relationship between ‘the customer’ {he) and
shaves is as in the other clauses. That is, ‘the customer’ is still the
114 DEEP GRAMMAR

goal. So we have an effective clause which is also receptive since it

has s as goal, and so has the same features as (ii) The customers
were shaved. There is an obvious surface difference between these
clauses in that (ii) has a passive verbal group at p, but (ix) has an
active verbal group at p. The (ii) type seems directed towards an
actor, as is emphasised by the optional inclusion of actor at a;
The customers were shaved by the barber
and the related interrogative clause is:
(the customers were shaved)->By whom? {Were they?)
The type exemplified by (ix), on the other hand, seems directed
towards the possibility or facility of the process, as is shown by the
almost obligatory manner adverb:
This coat buttons easily
This poem recites beautifully
and the related interrogative clause:
{this coat buttons easily)->Does it?
This distinction has been given the appropriate names of agent-
oriented (ii), and process-oriented (ix). Thus:
Agent-oriented
(Extensive: effective: receptive)->P
Process-oriented
This choice is obviously only open to effective: receptive clauses.

section 4

Thus, with a total of nine clause types, we can represent the whole
network as in Figure 11.

figure 11
^Effective
-Extensive |-- rGoal realised
f-» LDescripti
LGoal unrealised

Transitivity
-Operative
' Process-oriented
-Middle
LAgent-oriented
L Receptive'

L Intensive.
 SYSTEMS AT CLAUSE RANK 115
And the description of the clauses will be:
(i) The barber shaved ten customers spc
extensive: effective/operative: goal realised
(ii) Ten customers were shaved sp
extensive: effective/receptive: agent-oriented
(iii) The soldiers marched SP
extensive: descriptive: middle
(iv) Mary seemed happy spc
intensive
(v) The sergeant marched the soldiers spc
extensive: descriptive: operative
(Vi) The soldiers were marched sp
extensive: descriptive: receptive
(vii) The barber shaved (sc. ‘himself’) sp
extensive: effective: middle
(viii) The barber shaved (‘people’) SP
extensive: effective: operative: goal unrealised
(ix) The customers shaved (‘ easily ’) sp
extensive: effective: receptive: process-oriented.
In the above descriptions, some of the clauses have the same
structural description (three have spc, six have sp) but each has a
unique systemic description. This means that if we consider structure
as the way in which systemic features are realised we can assign
unambiguous descriptions to each clause, but this is not so with
structural description alone. We can, in other words, predict struc¬
ture from system, but not system from structure: given the structural
description spc, it may be any one of three clause types above, and
the structural description sp may represent any one of six clauses
above; but given the feature specification
extensive: descriptive: operative
we can predict the structure spc, and we can further say that it will
be a ‘causative’ clause, with s as initiator and c as actor. From
the hearer’s position, of course, ambiguity is obviously possible
since it is structure he has to interpret. Halliday has pointed out
that:
Children don’t wash easily
may be interpreted as any of the following:
(vii) Children find it difficult to wash themselves
(viii) Children find it difficult to wash things
(ix) It is difficult to wash children.
116 DEEP GRAMMAR

Much of the foregoing discussion has been informal and seman¬

tic. We may note some more formal indications of the distinctions.
The distinction between extensive and intensive is fairly clear:
an intensive clause must have an active p and must have a c1 present,
and the verbs which can occur in such clauses form a fairly well
defined class (though it must be remembered that some will also
enter into other classes), traditionally called ‘copula’ class, e.g.
be, seem, turn, become, grow, appear, turn out, look, get. Thus:
["Intensive—>-verb of‘copula’ class; active p; c1 present
L Extensive->-verb not of ‘copula’ class; cE may be present
An operative clause will have an active verbal group at p; if it is a
descriptive clause then a cE will be present and the clause is ‘ causa¬
tive ’; if it is an effective clause then it may have a cE present (goal
realised) or it may not (goal unrealised).
A middle clause will have an active p whether it is effective or
descriptive; but will have only one nominal element, namely s.
A receptive clause will have a passive p if it is descriptive; if it
is effective it may have a passive p (agent-oriented) or it may have
an active p (process-oriented).
The indication of distinctions within extensive clauses depends,
however, on the distinction between effective and descriptive. This
is the primary distinction within effective clauses, and it is this dis¬
tinction which seems to present the essential difficulty of the transi¬
tivity systems.
The clause paradigms listed underneath are intended to illustrate
the difference, and indeed to illustrate again what is meant by para¬
digmatic axis: given a feature such as ‘ effective ’ there is a choice from
a list of possible clause types; given a feature like ‘descriptive’
there is a choice from a different list of possible clause types. It is
the different possibilities of choice which enables us to posit features
like effective and descriptive in the first place, not the other way
round.

Effective Descriptive
The barber shaved ten customers The soldiers marched
Ten customers were shaved The sergeant marched them
He shaved (sc. ‘himself’) The soldiers were marched
The barber shaved (‘well’)
The customer shaved (‘easily’)

He sells books Corn grows

 SYSTEMS AT CLAUSE RANK 117
Effective Descriptive
Books are sold The farmer grows corn
He sells (sc. ‘himself’): Corn is grown
non-occurring?
He sells (sc. ‘ things ’)
These books sell (‘ well ’)
Mary washed the dishes The boat sank
The dishes were washed He sank the boat
Mary washed (sc. ‘herself’) The boat was sunk
Mary washed (sc. ‘the dishes’)
It washes (‘easily’)
There are three points to note in relation to these paradigms:

(i) The occurrence of types (viii) and (ix) in the effective paradigm,
as against the non-occurrence of such types in the descriptive
paradigm. We can have in English:
she washes the clothes or she washes
she dressed herself or she dressed
where the alternatives are interpreted alike; but not
he sank the boat or he sank
he grows corn or he grows
without a difference in meaning.

(ii) In the effective paradigm, the operative clause may have either
a cE or a cRef, thus:
He kicked the ball: He kicked himself
He washed the dishes:He washed himself
This is not so with the operative in the descriptive paradigm, thus:
He marched the soldiers: *He marched himself
He sank the boat: *He sank himself
The fact that cRef can occur in descriptive: operative clauses with
certain accompanying adverbial elements of clause structure:
He marched himself down the road
He sank himself into the nearest chair
does not invalidate the important point that in effective: operative
clauses the cRef can occur without such adverbials, but this is not so
in descriptive clauses.

(iii) If clauses with double complements are considered (pp. 57 Jf.)

it becomes evident that not all clause types will accept two comple¬
ments. The first complement in the clause:
118 DEEP GRAMMAR

He gave John a present

is usually called ‘indirect object’: it must precede the other c, and
it may be realised as a ‘to/for’ adjunct, as in:
He gave a present to John
This is sometimes referred to as ‘dative case’; but the fact that the
element in question can also occur as s:
John was given a present
(where ‘dative’ is surely inappropriate) argues against this. The
important point is that John realises the same feature in all three
clause positions. This feature has been called ‘beneficiary’ (i.e. one
who or that which benefits from the process of the clause—for good
or ill). Only effective clauses can accept a beneficiary. Thus:
He gave John a book *He marched John the sol¬
diers
A book was given John but not *The soldiers were marched
John
John was given a book *John was marched the sol¬
diers

section 5

It is obviously possible to classify verbs according to their occurrence

in the effective or descriptive (or intensive) paradigm. This should
yield a more satisfactory classification than the traditional ‘transi¬
tive/intransitive’ classification; and we will not have to talk about
‘intransitive verb used transitively’ in:
Boys fly kites
He sank the boat
or ‘transitive verb used intransitively’ in:
These books sell well
This cloth washes beautifully,
to say nothing of ‘ used reflexively, absolutely, absolutely (for refl.) ’.
There will, indeed, still be some overlap; some verbs will enter
into more than one class, which is why we said verb classes will not
identify clause types, for example grow in:
He grew pale (‘ copula ’)
He grew (‘ descriptive ’)
but the overlap should be minimised. *
Not all verbs will realise all possibilities of the paradigms; but it
is difficult to say that any particular verb will not occur in any
 *

SYSTEMS AT CLAUSE RANK 119

FIGURE 12 Systems at Clause Rank

r Declarative rSqu.
rlndicative- i-WH-
-Interrogative- non-Squ.
r-Mood-
Lnon-WH
LImperative

■ s-new
■p-theme
-Marked —
■ c-theme
r-(Decl. theme) - a-theme
-Unmarked

-Theme
•wh non-initial
-Marked —
•wh initial, tonic
^(Interr. theme)-
-Unmarked
Major
Clause -Effective
■Goal-realised

rExtensive- Goal-unrealised

■Process-oriented
Transi¬
tivity ■Agent-oriented

-Intensive
■Greater
Marked—
Distribution- LLesser
Unmarked
Infor-_
mation Marked
F'ocus ■
-Unmarked

I
120 DEEP GRAMMAR

particular clause type. Given a suitable context, it seems that any of

the possibilities of the paradigm will be generated by the grammar.
The most obvious is type (vii) in the effective paradigm; only a few
verbs—wash, shave, dress—occur regularly in such clauses. It might
seem more appropriate, therefore, to consider this as a subclass of
(viii). But other verbs do occur, if less regularly or frequently:
He flung out of the office
Hold still
Behave!
This indicates that it is a genuine clause type, and new examples
can be formed on the model if desired. Other verbs will only occur
in types (viii) and (ix) in highly restricted circumstances: but such
circumstances may obtain, and the clauses will be produced. We
might predict, for example, the verbs demand and encourage would
not occur in such types, but reported examples are:
(viii) He demands all the time (He demands (things) all the time)
I do that because it encourages (Encourages people)
(ix) They encourage too easily (It is too easy to encourage them)
They don’t demand any more (Nobody demands them any
more)
A chart of all systems at clause rank is given in Figure 12.

3.2 Systems at Group Rank

The unit Group provides the environment for further systems. The
following sections provide a first approach to a description of the
more prominent systems at Nominal and Verbal Group.

3.2.1 Nominal Group

The principal systems operating at Nominal Group are those of
Number, Case and Gender.

3.2.1.1 NUMBER
The system of number is usually regarded as a feature of word-
rank, and as a system having the terms singular and plural. That is:
_ [“Singular
Number-
^Plural
It is true that the terms in the number system are to a large extent
realised morphologically at word-rank; thus we can say:
boy = singular: boys = plural.
 SYSTEMS AT GROUP RANK 121
As discussed earlier, however, not all nouns are overtly marked for
number, and those that are can be marked in different ways. There is
not one invariable criterion by which the terms singular and plural
are marked. Rather, there are a number of criteria, one or more of
which may be present in any instance.
Also, it is not the case that only nouns realise terms in the number
system. Words which operate at d and at o in the nominal group also
choose for number. Thus:
Singular Plural
that those
one two (etc.)
There are, therefore, nominal groups which contain a number
realisation at d and/or at o but not at h, and which can be thus
identified as singular or plural groups:
Singular Plural
this sheep those sheep
one sheep two sheep.
Some items which occur at d do not select for number {the, all, my,
etc.), and some items which occur at o do not select for number
(ordinal numerals—-first, second, third, etc.).
Two points emerge from such considerations. First, number
would seem to be more appropriately assigned to group rank, since
it can be realised at any one of three places in nominal group struc¬
ture. Second, since all three places can be filled by words which do not
choose for number, it is possible to have nominal groups which may be
described as uncountable, as opposed to nominal groups which do
choose for number, and which may be described as countable. The
system of number would therefore be more properly represented as:
Singular

Plural
Uncountable
The terms in the system will be realised by certain items at d, at o and
at h. Thus:
at d at o at h
Singular: a, this, that. one any singular noun
each, either, (i.e. any noun mor¬
every phologically marked)
Plural: these, those, two, three any plural noun (i.e.
many, few etc. any noun morpho¬
logically marked)
122 DEEP GRAMMAR *

Uncountable: all, the, such, first, second any unmarked noun

some, no, my etc. (e.g. sheep, deer)

It is the number of the nominal group which determines agree¬

ment with the verbal group at p, and with pronominal groups in
succeeding clauses. It is obvious that pronouns also realise terms in
the number system. Thus:

When the boys come, they will tell us

When the boy comes, he will tell us

A singular nominal group exhibits agreement with p in that the -s
form of the verb will occur where appropriate (i.e. in present tense,
third person). A plural nominal group in the same environment will
not be followed by the -s form of the verb. An uncountable nominal
group in such an environment will have either the -s form or not the
-s form following. Thus:
Singular: The bird sings This sheep walks
Plural: The birds sing Those sheep walk
Uncountable: The sheep grazes The sheep graze.
We can, of course, identify each occurrence of the sheep wasjwere
as singular or plural by the agreement exhibited with the following
verb (or pronominal group), and we can rightly consider such agree¬
ment to be a realisation of a term in the number system. There are
two points to note, however: (i) we are not identifying with reference
only to the nominal group in question, but with reference to a
relation between the nominal group and another verbal group (or
pronominal group); and (ii) we can make no generalised statement,
and grammar is concerned with generalised statements, not with
citation of particular items: the generalised statement for the nominal
group in this case is that it is unmarked for number.
From the previous discussion of inflexions in nouns, it is obvious
that there are four main classes of nouns (which can be variously
subclassified) according to their operation in singular, plural and
unmarked nominal groups; and pronouns can obviously be classified
according to number. In other words, although the system of number
is here being assigned to group rank, it is not irrelevant to word rank,
since words are overtly marked in their own structure for singular
and plural, and, by default, unmarked. Thus:
 SYSTEMS AT GROUP RANK 123

NOUN
1. Either sg. 2. Always sg. 3. Always pi. 4. Unmarked
or pi.
boy beauty clergy sheep
lad courage people deer
bat despair dregs fish
horse mathematics bellows fowl
etc. etc. etc. etc.

PRONOUN
1. sg. 2. pi. 3. Unmarked
I/me we jus mine/our
hejhim theyj them you/your
she/her his/its
it herjtheir

In terms of this classification there is a difficulty in assigning the

traditional ‘collective’ nouns to classes (e.g. army, committee). Such
nouns are regular members of class 1 according to their own struc¬
ture, since they morphologically distinguish singular and plural
forms. They are irregular in that their singular form occurs in
nominal groups which take a following -s form of the verb or not an
-s form of the verb, e.g.
The committee was/were . . .
Semantically, of course, this is a reflection of the referents of the
words being thought of as a corporate body (singular) or a collection
of individuals (plural); it may also relate to different varieties of
English, different ‘ styles ’.

3.2.1.2 case

The system of case in English nouns has two terms, a marked and
an unmarked term. The unmarked term is usually called the common
case and the marked term the possessive or genitive case. For the
marked case neither of the names is really suitable; the function
of the case is to mark relation certainly, but this relation may be one
of a number of types of relation, and certainly not only one of
possession. It would perhaps be better to use terms such as marked/
unmarked, but the names common and genitive are reasonably
neutral and are well established (though ‘possessive’ is inappro¬
priate). Thus:
124 DEEP GRAMMAR
*

Common Genitive
boy boy's
horse horse's
man man's

■men men's
boys boys'
daughters daughters'

Perhaps the only real problem of case in English is whether it is

a feature of word rank or of group rank (or has relevance for both,
as the number system). We usually think of case as applicable to
nouns, and certainly nouns form paradigms according to realisation
of case and number features:
boy boys ox
oxen
boy's boys' ox's
oxen's.
When the noun operates at d in nominal group structure it is the
genitive form which occurs:
the boy's book!the boys' book
the ox's tail/the oxen's tails.
It is clear, however, that the 's is in fact added to a complete nominal
group. That this is so is evidenced by the fact that nominal groups
containing qualifiers can operate at d (and are therefore rank-
shifted) in this way, and when they are the's occurs after the qualifier,
not after the headword. Thus:
the [shop across the road]'s sales have gone up.
Further, it is possible in English to transform a qualifier with the
structure ‘of + nominal group’ into a delement, which is marked
by the ’s. Thus:
The houses of the town council The town council's houses.
We lose one the in such a transform, since we do not say in Eng¬
lish:
* The the town council's houses
We might (quite arbitrarily) consider it to be the second the which
has been excised, and thus the analysis is:
The [town council]'s houses.
The problem with such an analysis is that it seems to treat's as a
word, since it is operating in group structure, or alternatively to
treat town council as a word taking an enclitic 's morpheme. The
choice would seem to be between downgrading town council from
group to word, or upgrading's from morpheme to word. Neither
 SYSTEMS AT GROUP RANK 125
solution is satisfactory, since ’s is unlike any item we want to call a
word, and (the) town council is clearly a group and not a word.
There is perhaps another possibility, if we consider more carefully
the concept of rankshift. A unit does not become another unit when
it is rankshifted, but operates at an element of structure usually
realised by a lower unit. A rankshifted unit therefore retains its own
structure but can realise features usually realised by a unit of lower
rank. Thus, the feature ‘actor’ is usually realised by a nominal
group at s; but if a clause is rankshifted to s then it can realise this
feature. Thus:
s = Actor
l Whoever comes last] will kill him
just as
s = Actor
John will kill him.
Similarly the group (the) town council is rankshifted to operate at
d in another group; it retains its own structure, but realises features
of ‘specification’ (see p. 128) usually realised by a word. The ’s can
be regarded as an overt marker of this rankshift and feature realisa¬
tion only, and its structural value need not be stated. This is perhaps
but another unsatisfactory solution, though it has resemblances to
similar problems elsewhere in the grammar, notably the linker and
and intonation features which are regarded as realisations of systemic
terms, but are not considered as constituents of structure.

3.2.1.3 GENDER

Gender is usually said to be ‘natural’ in English. This means that

the contextually-determined classes of masculine, feminine and
neuter are reflected by the grammatical system of gender; in other
words, nouns referring to males are masculine in gender, nouns
referring to females are feminine in gender and nouns referring to
neither are neuter in gender. This is largely, but by no means com¬
pletely, true for English. The formal indication of gender in English
is agreement between nominal groups and the gender-bearing pro¬
nouns he, she, it. (The only remnant of morphemic indication of
gender which need be noticed is the derivational -ess, which marks
feminine gender in nouns such as authoress, actress. Even this word-
structure marker has decreased greatly in use in recent times, and
periphrases such as male-teacher, woman doctor are now common.)
It follows from what was said above that if gender were com-
126 DEEP GRAMMAR
*

pletely natural in English there would be three gender classes,

as follows:
Masculine, correlating with the pronoun he
Feminine, correlating with the pronoun she
Neuter, correlating with the pronoun it.
There are, indeed, many nouns (or rather, noun-headed groups,
since it is the whole group for which the pronoun can substitute;
this is why I have incorporated gender into the group systems,
though it is obviously relevant at word rank also) which do fulfil
this correlation, e.g.:
Masculine: man, uncle, grandfather, actor, boy, bachelor, William,
son'. Time, Death
Feminine: woman, aunt, girl, actress, mother, daughter, maid,
Avril; Nature, Charity
Neuter: wood, can, tin, tree, holiday, Sunday, bible, force, ab¬
straction, fission
It should be noted that the gender of these items is not, in fact,
determined by the sex of the referent. It is sometimes said that we
need to know the meaning of the noun in order to know its gender.
This is not necessary, and in any case would fail with personifications
such as Death, Charity, all that needs to be known is the pronominal
correlation, e.g.
Death, man he
Charity, girl she
box, mountain it
There are many other nouns, however, which do not stand in a
one-to-one correlation with pronouns. The correlations which they
exhibit are various, and they can be subclassified on this basis:
(i) Nouns which correlate with either he or she. These are largely
nouns denoting occupations, human relationships and some proper
names, e.g.
doctor, teacher, pupil, companion, servant, boss, supervisor,
Terry, person
Such nouns are usually said to have common gender. It is true in this
class that gender determination in any occurring instance is deter¬
mined by sex of the referent. Agreement with he or she is in accor¬
dance with the known or assumed sex of the referent; in a number of
cases there is a bias towards a particular correlation where the
 SYSTEMS AT GROUP RANK 127
referent is not known, thus doctor more frequently collocates with
he, and teacher with she.
(ii) Nouns which correlate with either he or it; mainly animate, non¬
human nouns such as:
insect, spider, beetle, cock, ram, bull
(iii) Nouns which correlate with she or it; these include animate,
non-human nouns, nouns referring to machines or engines, and
place names, such as:
mare, ewe; ship, car, train; country, Britain
(iv) Nouns correlating with he or she or it, mainly names of animals
and non-adult humans, such as:
deer, dog, cat, horse, cow; baby, child, tot
In none of the groups (ii), (iii) or (iv) is sex the determining
factor for occurring correlations (or obviously for many of them
there would not be alternatives). The type of language and speaker,
the register, seem to be the relevant factors.
We can perhaps bring together all the subclasses as unmarked for
gender, and they can be further specified if desired. Thus:
P Masculine {he agreement)
-Feminine {she agreement)
Gender11-> -Neuter {it agreement)
rhelshe agreement
-Unmarked- -he\it agreement
-hs el it agreement
-he/she/it agreement.
Systems at group rank are summarised in Figure 13.
figure 13 Systems at (Nominal) Group Rank
P Singular
r Number-> -Plural
Unmarked

- Case- p Common
Nominal group- L Genitive

r Masculine
-Feminine
L Gender >
-Neuter
-Unmarked
128 DEEP GRAMMAR

3.2.2 Systems at d in Nominal Group Structure

The general word-class which realises the element d in nominal
group structure is called determiner. Determiners may obviously
be subclassified according to their operation in the substructures
d2d2d2 (pp. 30-31). Determiners may also be subclassified along
various dimensions according to their realisation of systemic terms.
Perhaps the primary function of determiners is, as the name
suggests, to determine or identify. They identify the class which the
headword expounds, and may or may not specify a particular
member of that class. Some determiners are therefore specific and
some are non-specific. Thus, in the nominal groups a house, any
house, the determiners do not specify any particular house. But in the
groups my house, this house, the determiners specify that it is ‘that
member of the class “house” which belongs to me’ and ‘that mem¬
ber of the class “house” which is nearer us’ that are being referred
to. We can therefore consider specification to be the primary system
here, with the terms specific and non-specific. Thus:
[“Specific (the, my, your, John's, which, whose, etc.)
L-Non-specific (a, any, all, both, neither, etc.)
The specific determiners may specify by various means, or they
may not specify in themselves but merely indicate that specification
will be given. There is a system of selection with the terms selective
or non-selective from which specific determiners choose. The only
specific: non-selective determiner in English is the: what in fact this
item does in the language is to indicate that something in the cotext
(the surrounding language) or in the context (the situation of the
utterance) will identify the member in question of the class denoted by
the headword. Thus, in the groups
(throw) the ball
(hit) the thing
the identification is likely to be situational: only one ball or thing
will be present (or perhaps pointed to); in the groups
the university of Glasgow
the man who came to dinner
the determiner points forward to the qualifier in the groups which
identifies the member of the class university and the class man. All
other specific determiners are selective; they select from the various
means of identification. Thus:

(Specific) selection-4Se'eCtiVe ('fe’ my’your’ mother's' et«

LNon-selective (the).
 SYSTEMS AT GROUP RANK 129
Determiners which are specific: selective may select a means of
identification from the possibilities available. The selection may be
according to a genitival relation or a non-genitival relation; the first
of these features is obviously realised by the genitive form of pro¬
nouns and by rankshifted nominal groups as discussed in the
previous section. We can call this system selectivity, meaning that it
is the selection of the means of identification of the member of the
class denoted by the headword. Thus:
(Specific, selective)-^ J°hn’°• ete')
selectivity LNon-genitival (this, that, which, etc.)
The determiners which are genitival will make a further choice
between nominal, pronominal and relative; and genitival .•pro¬
nominal determiners will make further selection from the system
of person. Thus:
“Nominal (John's, mother's, etc.)
Specific ^
r 1st (my)
Selective relation’- ■ “Pronominal->person^l 2nd (jw)
GenitivalJ
L Relative (whose) -* 3rd (his, her,
its)
The determiners which are non-genitival are also involved in a
further system, which we may call relation2, according to whether
they are relative or demonstrative. Thus:
Specific (-Relative (which, what)
Selective ^relation2-
Non-genitival J •-Demonstrative (this, that)
Determiners which are non-specific according to the primary
division set up indicate a class but do not specify any particular
member of that class. They may, however, refer to the whole class
or parts of the class; so we can say they choose from a system of
coverage which has the terms total and partial. If total is selected,
then the reference may be to all the members of the class or to none
of the members, so we can say there is a system of elusion with the
terms inclusive and exclusive. Thus:
rlnclusive (all, every, each, both)
Total-^clusion-
Exclusive (no, neither)
Coverage-
-Partial (a, some, any, another, either)
This is not the only way in which determiners may be classified. It
may seem unnecessary to subclassify to such a degree, though further
 130 DEEP GRAMMAR

division is necessary if the items are to be uniquely classified. If

items such as determiners are ‘grammatical’ rather than ‘lexical’
items then the grammar must uniquely specify them. Thus the feature
description:
determiner: specific: selective: genitival: pronominal: 1 st person
means (is realised by) only one item in the language, namely my.
It can be seen that further subclassification is possible, and all
such items can be uniquely specified in this way. The items no,
neither, for example, will—according to what has been considered
so far—both be described as:
determiner: non-specific: total: exclusive
and the application of a choice between dual and non-dual (in a
system of duality) will distinguish them. Thus:
no determiner: non-specific: total: exclusive: non-dual
neither determiner: non-specific: total: exclusive: dual.
Systems at deictic in nominal group structure are summarised in
Figure 14.

figure 14 Systems at deictic in Nominal Group Structure

rNominal John's mother's

rlst my
r Genitival ■ Pronominal—^-Person - -2nd your
.3rd his, her,
r Selective-^- L Relative whose its

Non- r Relative which what

Specific - genitival L Demonstrative this that

-Non-selective the

r Inclusive all every each both

rTotal
Non¬ L Exclusive no neither
specific
Partial a some any another either

my—determiner: specific: selective: genitival: pronominal: 1 st person

his—determiner: specific: selective: genitival: pronominal: 3rd person
(singular: masculine).
 SYSTEMS AT GROUP RANK 131
3.2.3 Verbal Group
It was stated above (p. 41) that the verbal group is one of the most
complex areas of English grammar. This is so because the verbal
group is the environment for a number of systemic choices and the
terms selected are not in a one-to-one relation with their realisation
in structure. There are at least eight systems relevant to the verbal
group; but not all verbal groups can choose from all systems (only
finite groups can choose from all systems); and it is also the case
that most of the systems are binary and non-recursive. That is,
we choose one from two possible terms, and we make the choice
only once in a group. The main complexity, in fact, is occasioned
by the tense system, which is not binary and which is recursive. The
principal systems are:
1. Finiteness
2. Modality
3. Tense
4. Polarity
5. Aspect
6. Voice
7. Contrast
8. Focus (of Contrast).

3.2.3.1 FINITENESS

As already discussed, the system of finiteness is a primary distinction

in the verbal group; further systemic choices depend on selection
here, and the choice here is also relevant to clause classification since
a finite verbal group is obligatory in an independent clause and a
non-finite verbal group is one criterion of a dependent clause. The
system has the terms finite and non-finite, thus:
. . rFinite
Fimteness->\
LNon-fimte
and realisation of the selected term occurs at initial position in the
verbal group. As shown in the diagram on p. 44, the terms and
realisations are:
System Terms Realisation {at initial element)
Finite_>-s, base form of the verb,
Finiteness->\ or modal
^Non-finite-►-ing, -en, t-infi, base, form of
the verb.
132 DEEP GRAMMAR

3.2.3.2 MODALITY

Since modal verbs are always finite it is evident that only finite verbal
groups will select for modality; and further, since finiteness is realised
at initial element in group structure, then if a modal verb is present
it will be the initial element. There is, in practice, no need to mark
a verbal group finite:modal since the marking modal will presume
marking finite. The choice in the system of modality is a choice
between a marked and an unmarked term, the presence of an element
contrasting with its absence, and can therefore be named modal
and non-modal. Thus:
System Terms Realisation {at initial element)
rModal-*any modal verb
Modality-►
LMon-modal—>-any other group

3.2.3.3 TENSE12

Discussion of the system of tense in English tends to be complicated

by three factors:
(i) confusion of tense and aspect
(ii) confusion of tense and time reference
(iii) non-consideration of more complex tense formations.
Though there is obvious reason for the confusion between tense
and aspect in English, it has been maintained that they are different
systems (p. 44), and should be kept apart in description. Accordingly,
names such as ‘continuous present’ ‘past perfect’ are not used in
this discussion. The components ‘continuous/perfect’ refer to
aspectual features, and their combination with tense features serves
only to blur the tense features which are involved.
The confusion of tense with time reference stems largely from the
conception that the tense system is a reflection of some ‘naturally-
given’ time system. This in turn leads to the view that past/present/
future time is necessarily expressed by past/present/future tense in
language. Not only is such correspondence not necessary, but tense,
in fact, need not be a feature of a language—though such a lan¬
guage may perfectly well refer to time. The point is that time reference
can be made by devices other than tense, and in English is often
made by adverbial groups rather than verbal groups. It is not denied
that there is a close relation between tense and time reference in
English; but before this relation is discussed an adequate description
 SYSTEMS AT GROUP RANK 133
of tense itself is required. In what follows the discussion is limited
to discussing tense as a system of the verbal group.
In any attempt to describe the formation of compound tenses in
English it becomes obvious that the regularity of formation pro¬
cesses permits compound tenses which have a number of components.
It is not enough to account for the tense in verbal groups such as:
will come
will be coming
will have come
because considerably more complex forms are found:
will have been coming
will have been going to be coming
has been going to have been coming
It is true that the more complex of the theoretical possibilities
do not occur nearly as frequently as the less complex possibilities.
In spoken English, however, longer verbal groups which involve
complex tense features are by no means uncommon, notwithstanding
the fact that most native speakers are quite unaware of this. It is
perhaps because the participants in a spoken situation already know
the temporal implications of the discourse (they do not have to
‘work them out’) that such complexity is tolerated more in the
spoken than in the written language. Nevertheless, the important
point is that if a particular context requires the formulation of a
particular combination of tense features then the grammar of the
language (within certain stateable restrictions) will generate the
required tense complex.
This last point is one further illustration of the inadequacy of a
grammar which will only describe a given corpus. A grammar should
not only describe what has occurred, but should be capable of
predicting what structures and combinations may occur.
Traditionally, the tense system in English has three terms: present,
past, future. They are retained here without comment for the
primary distinction. Thus:
^Present
Tense- Past
LFuture
The term present is realised by the -s, base forms of the verb (/ eat;
he eats); the term past is realised by the -d form of the verb (he
atejwalked); the term future is realised by will/shall + base form of
the verb (I will eat; they shall eat); the primary distinctions are thus:
134 DEEP GRAMMAR

[Present->--s, base form of verb

Past->-d form of verb
Future->■ will/shall + base form of verb
Complexity is occasioned by the fact that English permits com¬
pound tenses; in other words, tense is recursive—it may be chosen
more than once in the verbal group. This fact in itself need not
occasion complexity; recursion is a feature occurring at various
points in the grammar, and can be described as a depth relation
(counting from the ‘head’ element). Thus:
e3 e2
adjectives at e in nominal group structure: the funny old American
hat
A f»
derivation (f) in word structure: boy + ish + ness
In these examples, however, the elements which are recursive are
realised by discrete items; they can be marked off and numbered.
In the realisation of compound tenses in the verbal group this is not
so; when there is a second tense element in the verbal group, the
first tense element does not remain as above but undergoes a change
in form and requires two items for its realisation. If we have the
simple tense present (e.g. eats) and we wish to add past, we do not
say *was eats but rather was eating. This means that if there is more
than one tense element in a verbal group, the (simple) present will
be realised by the -ing form of the verb preceded by some part of the
verb be\ if the (simple) tense is past then in a compound tense this
requires the -en form of the verb preceded by some part of the verb
have (has I had eaten); if the (simple) tense is future then in a compound
tense this requires going to + base form of the verb preceded by
some part of the verb be. It is more accurate, therefore, to give two
realisations of the simple tenses. Thus:
1 item verbal group more than 1 item verbal group
Present -s form be + -ing form
Past -d form have + -en form
Future will + base form be + going to + base form
The particular form of be and have which occurs in such structures
will realise the second component of the compound tense.
In a verbal group such as is eating, we have placed the present
(be + -ing) in the present (is); but in the group was eating, we have
placed the present (be + -ing) in the'past (was); and in the group
will be eating, we have placed the present (be + -ing) in the future
 SYSTEMS AT GROUP RANK 135
(will -f- base). These compound tenses can then be read as present
in present, present in past, present in future.
We can in similar fashion place the past in the three terms, and
this requires have as the tense auxiliary. In the group has eaten,
we have placed the past (have + -en) in the present (has); in the
group had eaten, we have placed the past (have+-Qn) in the past (had);
and in the group will have eaten, we have placed the past (have + -en)
in the future (will + base). We can then read these compound tenses
as past in present, past in past, past in future.
We can also place the future in the three other terms. In this case
the compound tenses are realised by some part of the verb be + base.
In the group is going to eat, we have placed the future (be going to
+ base) in the present (is); in the group was going to eat, we have
placed the future (be going to + base) in the past (was); in the group
will be going to eat, we have placed the future (be going to + base)
in the future (will + base). These compound tenses can then be
read as future in present, future in past, future in future.
In summary: a verbal group with only one element of structure
realises one of the primary tenses present, past; a verbal group
with the auxiliary willlshall + base form of the verb realises the
primary tense future. Each tense can then be chosen a second time
in combination with itself or either of the others; such compound
tenses are not realised by discrete elements, but require that the
(simple) present be realised by some part of the verb be + the -ing
form of the verb, that the (simple) past be realised by some part of
the verb have + -en form of the verb and that the (simple) future
be realised by some part of the verb be + going to + base form of
the verb. The second component of the compound tense will be
realised by the particular form of the verb be or have in the struc¬
tures.
We have, therefore, three simple tenses and nine compound tenses,
giving a total of twelve structures at this point. Figure 15 illustrates
how the tenses are built up and how they may be read.
This takes tense to two places in the verbal group (counting back
from lexical verb). Since tense may go up to five places, it is evident
that long verbal groups realising quite complex tense formations
are possible. There are however some restrictions: each term can
only follow itself once in a group; the term present can only further
occur at final position; the term future can only occur twice in a
group. It may be useful to extend Figure 15 to three places; because
K
136 DEEP GRAMMAR

FIGURE 15
Present Past Future
Simple eats ate will eat
Present in is eating was eating will be eating
Past in has eaten had eaten will have eaten
Future in is going to eat was going to will be going to
eat eat

of the restrictions just given this means only four additional struc¬
tures, as shown in Figure 16.
In order to analyse the tense of a compound verbal group it is

figure 16

Present Past Future

Simple eats ate will eat
Present in is eating was eating will be eating
Past in has eaten had eaten will have eaten
Future in is going to eat was going to will be going to
eat eat
Present in
Past in has been had been will have been
eating eating eating
Present in
Future in is going to be was going to will be
eating be eating going to be
eating
Past in
Future in is going to was going to will be going to
have eaten have eaten have eaten
Future in
Past in has been going had been will have been
to eat going to eat going to eat

obviously necessary to consider pairs of items in the group. Counting

begins from the lexical verb and moves back. If the final verb is an
-ing form this will indicate present, if it is an -en form this will
 SYSTEMS AT GROUP RANK 137
indicate past and if it is a base form this will indicate future.
Thus:
{be)-ing-►present
(have) -en->past
(be going to) base-►future
If the tense goes to three places the items preceding the lexical verb
need to be considered as a pair, and then the third item singly, to
give the third component of the tense. Reading back from the lexical
verb, items one and two equal the first component of the compound
tense, items two and three equal the second component of the com¬
pound tense, and item three equals the third component of the
compound tense. Thus:
past past pres,
f l-ll-1
had been eating present in past in past
was going to be eating present in future in past
11-ii_i
past fut. pres.

3.2.3.4 POLARITY

The system of polarity is a system with the terms positive and nega¬
tive, and in this system there is a one-to-one correlation realisation,
because the term negative is always realised by the negator (not, n't).
So this system is simply:
System Terms Realisation
rPositive->-any group without negator
Polarity-
•-Negative- ->not, n't

3.2.3.5 aspect

The system of aspect has been confused with that of tense, and in
certain respects the two do merge. But they are different systems
and should be kept apart in description. Aspect is perhaps best
considered primarily as a distinction between imperfective and
perfective. The difference between the clauses
(i) I have eaten the apple
(ii) I am eating the apple
is (apart from tense) that in (i) attention is drawn to the fact that the
action is completed, but in (ii) attention is directed to the fact that
the action is in progress, is continuing. The contrast is not, however,
between complete/incomplete in reference to a non-linguistic event.
138 DEEP GRAMMAR

It is simply that the language directs attention in this way. In the

clause
(iii) I was eating the apple
the action is complete since it is in past time, but attention is still
drawn to the continuance over an unspecified period of time of the
action. It is the tense, and only the tense which differs between (ii)
and (iii). This is why we ought to distinguish between aspect and
tense. The clauses
(i) I have eaten the apple
(ii) I am eating the apple
differ in tense, (i) being past in present, (ii) being present in present;
and they differ in aspect, (i) being perfective, and (ii) being imper-
fective. But the clauses
(ii) I am eating the apple
(iii) I was eating the apple
differ only in tense, (i) being present in present, and (ii) being present
in past; but both are imperfective aspect.
There are a number of problems involved in any comprehensive
discussion of aspect in English, but a primary division such as is
being suggested here may provide a starting point and may help to
illustrate that aspect and tense are different systems. Thus:
System Terms Realisation
rPerfective->have + -en
Aspect-► T _ .
'-Imperfective—>be + -mg

3.2.3.6 voice

The system of voice in English has two terms, active and passive,
the former being unmarked and the latter realised by part of the
verb he + -en form of the verb. The system is simply thus:
System Terms Realisation
_ rActive->-any group not marked as passive
^-Passive-Kpart of) be + -en.
This system is obviously closely related to the transitivity systems,
since the distinction between operative (John washed the dishes)
and receptive (The dishes were washed by John) requires, among
other things, an active verbal group at p in the former and a passive
verbal group at p in the latter. But it was pointed out that the cor¬
relation was not one-to-one, because clauses such as
 SYSTEMS AT GROUP RANK 139
This poem recites well
This coat buttons easily
It shows to advantage here
are receptive clauses (they have s as goal), though they have active
verbal groups. It would, of course, be possible to talk about receptive/
active and receptive/passive clauses; but this would be combining
features of clause and group rank. It seems more appropriate to
assign the features operative and receptive to clause rank, and the
features active and passive to group rank.
Some verbs do not admit of a passive form (see intensive clauses,
pp. 107 jf.). There are no passive equivalents of
He resembled his brother
He had a great set of cards.

3.2.3.7 CONTRAST

The system here being called contrast is one of the possibilities of

marked tonicity (pp. 101 jf.). In the (clause final) verbal group
(John) has been asked
the tonic syllable falls on the lexical verb. The tonic may be moved to
another item in the verbal group, thus contrasting with the neutral
position. We can, therefore, distinguish a contrastive/non-contras-
tive opposition:
System Terms Realisation
_ rNon-contrastive—atonic on lexical verb
Contrast-► _ . . , , . ,
■-Contrastive-►tonic elsewhere m the group

3.2.3.8 focus (of contrast)

If the verbal group is contrastive, one of three elements may pro¬

vide the focus of contrast. If the verbal group contains a modal
verb it may be this item which provides the focus. Thus:
(He) should come
(He) can come
(He) may manage it.
If the verbal group is positive it may be the positive element which
provides the focus, and the fact that it is the positiveness of the
group which is being emphasised and not the lexicality of the item
at /, is shown in that the auxiliary do is required when the non-
contrastive form would be a verbal group with only an / element of
140 DEEP GRAMMAR

structure, and that we usually require the full form of an auxiliary,

not the enclitic form. Thus:
Non-contrastive Contrastive
He’s coming He is coming
He came He did come
He comes He does come.
If the verbal group is negative in polarity then the negative element
may provide the focus; this usually requires the full form of the
negator and not the enclitic form. Thus:
Non-contrastive Contrastive
He isn’t coming He’s not coming
He won’t do it He will not do it.

figure 17 Systems at Verbal Group Rank

rModal
rFinite->Modality-
-Fimteness—► J L-Non-modal
LNon-finite

"Present
-Tense > -Past
-Future

Positive
-Polarity-
Verbal Negative
group'
Perfective
Aspect—
Imperfective

["Active
‘Voice —
L Passive
Modality
rContrastive->Focus-> Tense
Contrast-
^Non-contrastive L Polarity
(John) has been asked—finite, non-modal, past in present, positive,
perfective, passive, contrastive/polarity.

Finally, the focus of contrast may be provided by the tense ele¬

ment. Thus:
Non-contrastive Contrastive
He’s been asked He’s been asked.
 SYSTEMS AT GROUP RANK 141
We may, therefore, though this is simplifying somewhat, show the
system as
System Terms System Terms
T N on-contrastive M odality
Contrast-
L-Contrastive- ->Focus- Tense
Polarity
Systems at verbal group rank are shown in full in Figure 17.
 Notes

1. Randolph Quirk: The Use of English, Longmans Green & Co.;

1962: pp. 176 jf.
2. Although the procedure of comparing word-pairs to identify
morphemic constituents of words is well established, the principle
of ‘squares’ is taken from J. H. Greenberg: ‘The Definition of
Linguistic Units’ in Essays in Linguistics, University of Chicago
Press; 1957.
3. Nominal group: parts of the presentation here derive from the
excellent presentation of nominal group structure given by J. M.
Sinclair: A Course in Spoken English, Part 3: Grammar, Oxford
University Press; 1965 (limited issue).
4. Verbal group: for more detailed and lucid description see F. R.
Palmer: A Linguistic Study of the English Verb, Longmans Green
& Co.; 1965.
5. The terms ‘temperer, apex, limiter’ are ultimately derived from
Mr. G. J. Turner, I believe. I was given the terms at a linguistic
conference; they seemed very apt, and so I have taken the liberty of
retaining them here.
6. Phased predicators: this section owes much to the books by F. R.
Palmer and J. M. Sinclair already mentioned, and to discussions
with various colleagues.
7. It was John who . . ., example from M. A. K. Halliday: ‘Some
Notes on Deep Grammar’, Journal of Linguistics, Vol. 2, No. 1.
8. O. Jespersen: A Modern English Grammar on Historical Principles,
Copenhagen; 1940.
9. The section on transitivity systems is perhaps of doubtful value
as presented here. It follows closely, and in effect is my reading of,
NOTES 144
a series of articles by M. A. K. Halliday in the Journal of Lin¬
guistics, Vol. 3, Nos 1 & 2; and Vol. 4, No. 2. Halliday acknowledges
that what is contained in these articles is inadequate, though not
invalid. He is preparing an extended monograph on the subject
(private communication). I agree with this (and other linguists are
at present concerned with problems of transitivity from differing
viewpoints), but it does seem to me that the ‘actor-action-goal’
view of transitivity is valid as far as it goes, and brings out further
problems and directions for the grammar. The quotation is taken
from the first of these articles.
10. See J. Anderson: ‘Ergative and Nominative in English’, Journal
of Linguistics, Vol. 4, No. 1.
11. This consideration of gender derives ultimately from J. C. Cat-
ford; see B. Strang: The Structure of English, Arnold; 1962 (p. 95
and note of acknowledgement there).
12. The presentation of tense given here is based on lectures by
M. A. K. Halliday. Professor Halliday is at present preparing a
paper ‘On Finiteness, Tense, and Modality in the English Verb’
which will provide a full description of this system of tense. (Private
communication.)
Bibliography

‘A Synopsis of Linguistic Theory’, in Studies in Linguistic

firth, J. r.
Analysis, (Ed.) J. R. Firth, Oxford: Basil Blackwell (special
volume of the Philological Society), 1957
-Papers in Linguistics 1934-1951, London: Oxford U.P., 1957
-Selected Papers of J. R. Firth, (Ed.) F. R. Palmer, London:
Longman’s Linguistics Library, 1968
halliday, m. a. k. ‘Some Aspects of Systematic Description and
Comparison in Grammatical Analysis’, in Studies in Linguistic
Analysis, Oxford: Basil Blackwell, 1957
-‘Categories of the Theory of Grammar’, in Word, Vol. 17, No.
3, 1961
-‘Class in Relation to the Axes of Chain and Choice in
Language’, in Linguistics 2, 1963
-‘Some Notes on Deep Grammar’, in Journal of Linguistics,
Vol. 2, No. 1, 1966
-‘Notes on Transitivity and Theme I—III’, in Journal of Lin¬
guistics, Vol. 3, No. 1, Vol. 3, No. 2, Vol. 4, No. 2, 1967/8
-‘Language Structure and Language Function’, in New Hori¬
zons in Linguistics, (Ed.) J. Lyons, London: Penguin Books,
1970
-‘Functional Diversity in Language, as seen from a considera¬
tion of Modality and Mood in English ’, in Foundations of Language,
6, 1970
Huddleston, r. d. ‘Rank and Depth’, in Language, 41, 1965
Hudson, r. a. ‘Constituency in a Systemic Description of the English
Clause’, in Lingua, Vol. 18, No. 3, 1967
146 BIBLIOGRAPHY
davies, e. ‘Some Notes on English Clause Types’, in Transactions of
the Philological Society, 1967
mcintosh, a. and halliday, m. a. k. Patterns of Language: Papers in
General, Descriptive and Applied Linguistics, London: Longman’s
Linguistics Library, 1966
mcintosh, a., halliday, m. a. k. and strevens, p. The Linguistic
Sciences and Language Teaching, London: Longman’s Linguistics
Library, 1964
palmer, f. r. A Linguistic Study of the English Verb, London:
Longman’s Linguistics Library, 1966
mitchell, t. e. ‘Syntagmatic Relations in Linguistic Analysis’, in
Transactions of the Philological Society, 1958

Other references:
quirk, r. The Use of English, London: Longmans, Green & Co., 1st
edition 1962, 2nd edition 1968
Bloomfield, l. Language, London: Allen & Unwin Ltd., 1935
anderson, j. ‘Ergative and Nominative in English’, in Journal of
Linguistics, Vol. 4, No. 1, 1968
Greenberg, j. h. Essays in Linguistics, University of Chicago Press,
1957
jespersen, o. A Modern English Grammar on Historical Principles,
Copenhagen, 1940
Index

actor, 106 command, 68

additioning clause, 56-57 comparative adjective, 22
adjective, 18, 29, 40 compound sentence, 68-75
adjective inflexion, 22-23 concord, 53-54
adjective sequence, 31-33 conditioning clause, 56-57
adjunct, 7, 54, 63-65 contrast, 139
adverb, 6, 49 countable, 121-122
adverbial group, 5, 6, 49-51, 54 coverage, 129
agent-oriented, 114
alpha element, 69-75 declarative clause, 68, 95-97
apex, 6, 49 deep grammar, 11, 89-141
aphoristic sentence, 76 deictic, 30, 39^10, 128-130
aspect, 137-138 dependent clause, 7, 56-57
auxiliar, 6, 41 derivational morpheme, 4, 18
auxiliary verb, 6, 42-47 descriptive clause, 109
descriptive categories, 8
base, 4, 17 determiner, 30, 128-131
beta element, 69-75
bound morpheme, 4, 17-18 effective clause, 109
element of structure, 3
cardinal numeral, 31 ellipsis, 71-75
case, 21, 123-125 epithet, 30, 31-33
chain axis, 10-11 exclamatory sentence, 76
choice axis, 10-11 extensive clause, 107
class, 3 extensive complement, 58-59
classifier, 30
clause, 7, 51-65, 93-120 feature, 10
complement, 7, 54, 57-59 feminine gender, 126
completive element, 6, 50-51 free morpheme, 4, 17-18
completive sentence, 76 finite, 43
 148 INDEX

finiteness, 44, 131 noun inflexion, 20-21

focus, 139 number, 10, 120-123
numeral, 30
gender, 125-127 numerative, 30
goal, 109
goal-realised, 113 operative clause, 109
goal-unrealised, 113 ordinal numeral, 31
group, 5, 25
personal pronoun inflexion, 24
headword, 5, 25, 33-35 phased predicators, 60-63
phrase, 5
imperative clause, 68, 94-95 phrasal verb, 48-49
independent clause, 7, 56-57 plural, 19-21, 120-127
indicative clause, 93-95 polarity, 10, 137
inflexion, 18-19 postfix, 4, 17
inflexional morpheme, 4, 18 predicator, 7, 54, 59-63
information, 101-104 prefix, 4, 17
initiator, 111 prepend, 6, 50
intensive clause, 107 preposition, 6, 50
intensive complement, 58-59 process-oriented, 114
interrogative clause, 68, 95-97 pronoun, 28
intransitive clause, 57-59 proper noun, 29, 40

lexical element, 6, 41 qualifier, 5, 26

lexical verb, 6, 18 question, 68
limiter, 6, 50
linkage, 71-75 rankscale, 4, 7
rankshift, 35, 78-80
major clause, 53 receptive clause, 109
masculine gender, 126 relation, 129
minor clause, 53 reported clause, 56-57
minor sentence, 75-78
modal verb, 45-46 selection, 128
modality, 132 selectivity, 129
modifier, 5, 26 sentence, 7, 65-78
mood, 93-97 sentence analysis, 80-88
morpheme, 4 sequence of elements, 7, 51-52
morphology, 8 simple sentence, 67-68
singular, 19-21
negative element, 6, 47 statement, 68
negator, 6, 47 structure, 3
neuter gender, 126 specification, 128
nominal group, 5, 25-40, 120-127 subject, 7, 53
non-finite, 43 sub-modifier, 6, 39
noun, 18, 27 substantive, 5, 27
 INDEX 149
superlative adjective, 22, 32 transitive clause, 57-59
surface grammar, 11, 14-89 transitivity, 104-120
syntax, 8
uncountable, 121-122
system, 10, 90-92
unit, 3

temperer, 6, 49 verb, 18, 23, 41

tense, 10, 132-137 verbal group, 5, 6, 41-49, 131-141
term, 10 verb inflexion, 23-24
theme, 97-101 voice, 10, 42, 138-139
theoretical categories, 8
word, 3, 4, 14
tonality, 102-103
tonicity, 102-103 Z-element, 54-55, 61-62
OUR EXPERIENCE OF LANGUAGE
Walter Nash

‘Our Experience of Language was designed in

the first instance for college of education and
university students, and will certainly be
recommended reading for mine.'
Speech and Drama

‘In this well-ordered and attractive book

Walter Nash scrutinizes personal situations
with a view to ascertaining how language
works, not as an abstraction, a structure, or a
system, but as a daily activity.' Times Literary
Supplement

‘ . . . a scholarly and pleasantly written book.'

English Studies

* isbn 0 7134 2008 1