Syntax

In linguistics, 8yntax (Irom Ancient Greek ouvtui "arrangement" Irom ouv syn, "together",
and tui taxis, "an ordering") is the study oI the principles and rules Ior constructing phrases
and sentences in natural languages.
In addition to reIerring to the overarching discipline, the term syntax is also used to reIer directly
to the rules and principles that govern the sentence structure oI any individual language, as in
"the syntax oI Modern Irish."
Modern research in syntax attempts to describe languages in terms oI such rules. Many
proIessionals in this discipline attempt to Iind general rules that apply to all natural languages.
The term syntax is also used to reIer to the rules governing the behavior oI mathematical
systems, such as Iormal languages used in logic. See Syntax (logic);Computer-
programming languages; Syntax (programming languages).
Though there has been an interplay in the development oI the modern theoretical Irameworks Ior
the syntax oI Iormal languages and natural languages, this article surveys only the latter.
Early history
Works on grammar were written long beIore modern syntax came about;
the A/yyi oI Pini is oIten cited as an example oI a premodern work that approaches
the sophistication oI a modern syntactic theory.
|1|
In the West, the school oI thought that came to
be known as "traditional grammar" began with the work oI Dionysius Thrax.
For centuries, work in syntax was dominated by a Iramework known as grammaire generale,
Iirst expounded in 1660 by Antoine Arnauld in a book oI the same title. This system took as its
basic premise the assumption that language is a direct reIlection oI thought processes and
thereIore there is a single, most natural way to express a thought. That way, coincidentally, was
exactly the way it was expressed in French.
However, in the 19th century, with the development oI historical-comparative linguistics,
linguists began to realize the sheer diversity oI human language, and to question Iundamental
assumptions about the relationship between language and logic. It became apparent that there
was no such thing as the most natural way to express a thought, and thereIore logic could no
longer be relied upon as a basis Ior studying the structure oI language.
The Port-Royal grammar modeled the study oI syntax upon that oI logic (indeed, large parts oI
the Port-Royal Logic were copied or adapted Irom the Grammaire generale
|2|
). Syntactic
categories were identiIied with logical ones, and all sentences were analyzed in terms oI "Subject

Copula Predicate". Initially, this view was adopted even by the early comparative linguists
such as Franz Bopp.
The central role oI syntax within theoretical linguistics became clear only in the 20th century,
which could reasonably be called the "century oI syntactic theory" as Iar as linguistics is
concerned. For a detailed and critical survey oI the history oI syntax in the last two centuries, see
the monumental work by GraIIi (2001).
Modern theories
There are a number oI theoretical approaches to the discipline oI syntax. One school oI thought,
Iounded in the works oI Derek Bickerton,
|3|
sees syntax as a branch oI biology, since it conceives
oI syntax as the study oI linguistic knowledge as embodied in the human mind. Other linguists
(e.g. Gerald Gazdar) take a more Platonistic view, since they regard syntax to be the study oI an
abstract Iormal system.
|4|
Yet others (e.g. Joseph Greenberg) consider grammar a taxonomical
device to reach broad generalizations across languages. Andrey Korsakov's school oI thought
suggests philosophic understanding oI morphological andsyntactic phenomena. At Ioundations
oI their linguistic ideas, lies classical philosophy which treats reality as consisting oI things,
their qualities and relationships. From here the Iollowers oI Korsakov's school assert the
subdivision oI words by the parts oI speech.
|5|
Syntactic problems also get their enlightenment in
the terms oI philosophic processes.
|6|
Some more approaches to the discipline are listed below.
0n07ativ0 g7amma7
The hypothesis oI generative grammar is that language is a structure oI the human mind. The
goal oI generative grammar is to make a complete model oI this inner language (known as i-
language). This model could be used to describe all human language and to predict
the grammaticality oI any given utterance (that is, to predict whether the utterance would sound
correct to native speakers oI the language). This approach to language was pioneered by Noam
Chomsky. Most generative theories (although not all oI them) assume that syntax is based upon
the constituent structure oI sentences. Generative grammars are among the theories that Iocus
primarily on the Iorm oI a sentence, rather than its communicative Iunction.
Among the many generative theories oI linguistics, the Chomskyan theories are:
TransIormational Grammar (TG) (Original theory oI generative syntax laid out by Chomsky
in $yntactic $tructures in 1957)
|7|

Government and binding theory (GB) (revised theory in the tradition oI TG developed
mainly by Chomsky in the 1970s and 1980s)
|8|

Minimalist program (MP) (a reworking oI the theory out oI the GB Iramework published by
Chomsky in 1995)
|9|

Other theories that Iind their origin in the generative paradigm are:
Generative semantics (now largely out oI date)
Relational grammar (RG) (now largely out oI date)
Arc Pair grammar
Generalized phrase structure grammar (GPSG; now largely out oI date)
Head-driven phrase structure grammar (HPSG)
Lexical-Iunctional grammar (LFG)
Nanosyntax
at0go7ial g7amma7
Categorial grammar is an approach that attributes the syntactic structure not to rules oI grammar,
but to the properties oI the syntactic categories themselves. For example, rather than asserting
that sentences are constructed by a rule that combines a noun phrase (NP) and a verb phrase (VP)
(e.g. the phrase structure rule S NP VP), in categorial grammar, such principles are embedded
in the category oI the head word itselI. So the syntactic category Ior an intransitive verb is a
complex Iormula representing the Iact that the verb acts as a Iunctor which requires an NP as an
input and produces a sentence level structure as an output. This complex category is notated as
(NP\S) instead oI V. NP\S is read as " a category that searches to the leIt (indicated by \) Ior a
NP (the element on the leIt) and outputs a sentence (the element on the right)". The category
oI transitive verb is deIined as an element that requires two NPs (its subject and its direct object)
to Iorm a sentence. This is notated as (NP/(NP\S)) which means "a category that searches to the
right (indicated by /) Ior an NP (the object), and generates a Iunction (equivalent to the VP)
which is (NP\S), which in turn represents a Iunction that searches to the leIt Ior an NP and
produces a sentence).
Tree-adjoining grammar is a categorial grammar that adds in partial tree structures to the
categories.
D050nd0ncy g7amma7
Dependency grammar is a diIIerent type oI approach in which structure is determined by the
relations (such as grammatical relations) between a word (a ea/) and its dependents, rather than
being based in constituent structure. For example, syntactic structure is described in terms oI
whether a particular noun is the subject or agent oI the verb, rather than describing the relations
in terms oI phrases.

Some dependency-based theories oI syntax:

Algebraic syntax
Word grammar
Operator Grammar
Meaning-Text Theory
Stocha8tic/57obabili8tic g7amma78/n0two7 th0o7i08
Theoretical approaches to syntax that are based upon probability theory are known as stochastic
grammars. One common implementation oI such an approach makes use oI a neural
network or connectionism. Some theories based within this approach are:
Optimality theory
Stochastic context-Iree grammar
:nctionali8t g7amma78
Functionalist theories, although Iocused upon Iorm, are driven by explanation based upon the
Iunction oI a sentence (i.e. its communicative Iunction). Some typical Iunctionalist theories
include:
Functional discourse grammar (Dik)
Prague Linguistic Circle
Systemic Iunctional grammar
Cognitive grammar
Construction grammar (CxG)
Role and reIerence grammar (RRG)
Emergent grammar

A SHORT OVERVIEW OF ENGLISH SYNTAX
Based on %e Cambri/ge Grammar of te Englis Language

Rodney Huddleston
The University oI Queensland

This paper presents a brieI account oI English syntax based on %e Cambri/ge Grammar
of te Englis Language,
|1|
providing an overview oI the main constructions and
categories in the language. The present version is intended primarily Ior members oI the
English Teachers' Association oI Queensland (ETAQ), oIIering an alternative approach
to that presented in the 2007 volume oI their journal Wor/s`Wort by Lenore Ferguson
under the title `Grammar at the CoalIace' - in particular the articles `The structural basics'
(March 2007) and `Functional elements in a clause' (June 2007). I make use oI concepts
discussed in my own Wor/sWort paper `Aspects oI grammar: Iunctions, complements
and inIlection' (March 2008), and take over Functional Grammar's useIul convention oI
distinguishing between Iunctions and classes by using an initial capital letter Ior the
Iormer: thus Subject is the name oI a Iunction, noun phrase the name oI a class.

1 SENTENCE AND CLAUSE
We distinguish two main types oI sentence: a cla:8al 80nt0nc0, which has the Iorm oI a
single clause, and a com5o:nd 80nt0nc0, which has the Iorm oI two or
more coo7dinat0d clauses, usually joined by a coo7dinato7 (such as an/, or,but):

|1| i $ue went to Lon/on last week. |clausal
sentence|
ii $ue went to Lon/on last week an/ er fater went wit er. |compound
sentence|

Note that such an example as We staye/ at te otel wic you recommen/e/ is also a
clausal sentence even though it contains two clauses. This is because one clause, wic
you recommen/e/, is 5a7t of the other, rather than separate Irom it (more speciIically,
the wic you recommen/e/ is part oI the noun phrase te otel wic you
recommen/e/); the larger clause is thus We staye/ at te otel wic you recommen/e/,
and this does constitute the whole sentence, like that in |i|.
The Iact that the two types oI sentence are distinguished in terms oI clauses implies that
we take the clause to be a more basic unit than the sentence, which reIlects the Iact that in
speech it tends to be more diIIicult to determine the boundaries between sentences than
the boundaries between clauses. For most oI this overview we will Iocus on clauses: we
return to coordination in Section14.

2 CANONICAL AND NON-CANONICAL CLAUSES
We can describe the structure oI clauses most economically iI we distinguish between the
most basic and elementary kinds oI clause, which we call canonical cla:808, and the rest.
The idea is that we can present the analysis more clearly iI we begin with canonical
clauses, describing them directly, and then deal in turn with the various kinds oI non-

canonical clause, describing these indirectly, in terms oI how they diIIer Irom canonical
clauses.
The Iollowing paired examples will give some idea oI what is involved in this distinction:



|2| anonical Non-canonical
i a. $e as rea/ your article. b. $e asnt rea/ your article.
ii a. $ue is coming for /inner. b. Is $ue coming for /inner?
iii a. %ey knew te victim. b. $e sai/ tat tey knew te victim.
iv a. e misse/ te train. b. Eiter e misse/ te train or it is late.
v a. %e secretary took te key. b. %e key was taken by te secretary.

These illustrate the Iollowing properties oI canonical clauses:
o They are 5o8itiv0; n0gativ0 clauses like |ib| are non-canonical.
o They are d0cla7ativ0; int077ogativ08 like |iib| are non-canonical, as are the other
clause types: im507ativ08 (e.g. !lease stan/ up) and 0xclamativ08 (Wat a fool Ive
been).
o They are main cla:808; the underlined clause in |iiib| is 8:bo7dinat0 and hence
non-canonical.
o They are non-coo7dinat0; the two underlined clauses in |ivb| are coo7dinat0d and
hence each oI them is non-canonical.
o They are activ0; 5a88iv0 clauses like |vb| are non-canonical. This is a matter oI
inIormation packaging and we can say, more generally, that canonical clauses package
the inIormation in the grammatically most basic way. Thus I ave now rea/ most of
tem is canonical but Most of tem I ave now rea/ is not.

There are two Iurther points that should be made at this point.

(a) In all the above examples the non-canonical clauses diIIer in their structure Irom
canonical clauses, but this is not always so. In |iiib| the subordinate clause is introduced
by tat but we could omit this, giving $e sai/ tey knew te victim, where the underlined
clause is identical with |iiia|; nevertheless it is still subordinate and hence non-canonical.
It is subordinate by virtue oI being Complement oI the verb sai/, but the subordination
happens not to be marked in the internal grammatical structure oI the clause itselI.

(b) A clause is non-canonical iI it lacks at least one oI the above properties. It may oI
course lack more than one oI them. Thus Wasnt te key taken by te secretary? has three
non-canonical properties: it is negative, interrogative and passive. In the discussion below
we will take the non-canonical properties in turn with the understanding that they can
combine.

3 INITIAL LISTING OF THE PARTS OF SPEECH
We distinguish nine primary word classes, or parts oI speech, to use the traditional term.
In this overview we needn't worry about interjections (wow, a, ello, and the like),
which leaves us with eight classes. They are named and exempliIied in |3|:

|3| i V07b e is ill. $e left early.
We want to elp.
ii No:n %e /og barke/. $ue won easily. I love you.
iii Adj0ctiv0 es very young. Ive got a sore knee. It looks easy.
iv Adv07b $e spoke clearly. Youre extremely fit. e works very ar/
v D0t07minativ0 %e /og barke/. Ive got a sore knee. We nee/ some milk.
vi P705o8ition es in te gar/en. Its from your uncle. We went to !aris.
vii oo7dinato7 We saw Kim an/ !at. urry or well be late. Its
ceap but goo/.
viii S:bo7dinato7 I know tat its true. Ask weter its true. I won/er if its
true.



Note that we use `determinative' as the name oI a class and `Determiner' as the name oI a
Iunction;
|2|
we need to invoke the class vs Iunction distinction here to cater Ior the
construction illustrated in te /octors car. Here te /octors has the same Iunction,
Determiner, as te in te car, but it is not a word and hence not a determinative: as Iar as
its class is concerned it's a noun phrase.
The above scheme diIIers Irom that oI traditional grammar in three respects:
o We take pronouns to be a subclass oI nouns, not a distinct primary class.
o Traditional grammars generally take our determinatives to be a subclass oI
adjectives, though some recognise a class oI articles consisting oI te and a. Our
determinative class is much larger, containing not just te and a, but also words
like some, any, all, eac, every, no, etc.; these are very diIIerent Irom words like those
underlined in |iii|.
o We have coordinator and subordinator as distinct primary classes, whereas
traditional grammar has a primary class oI conjunctions subdivided into coordinating and
subordinating conjunctions.

4 PHRASES
For each oI the Iirst six oI the word classes in |3| there is a corresponding class oI phrases
whose Head belongs to that class. In the Iollowing examples, the phrase is enclosed in
brackets and the Head underlined:

|4| i V07b 5h7a80 $e |wrote some letters|. e |is still in Lon/on|.
ii No:n 5h7a80 |%e new lo/ger| is ere. |%e boss| wants to
see |you|.
iii Adj0ctiv0 5h7a80 Its getting |rater late|. Im |gla/ you coul/ come|.
iv Adv07b 5h7a80 I spoke |too soon|.
Its |quite extraor/inarily| goo/.
v D0t07minativ0 5h7a80 I saw |almost every| car/. Weve |very little| money
left.
vi P705o8ition 5h7a80 %eyre |in te gar/en|. e wrote a book |on sarks|.

5 THE STRUCTURE OF CANONICAL CLAUSES

5.1 S:bj0ct and P70dicat0
A canonical clause consists oI a Subject Iollowed by a Predicate. The Predicate is
realised by a verb phrase; the Subject is mostly realised by a noun phrase, but there are
other possibilities too, most importantly a subordinate clause:

|5| S:bj0ct P70dicat0
i ne of is frien/s , calle/ a /octor. |noun phrase as
Subject|
ii %at e was lying , was obvious. |subordinate clause as Subject|

In canonical clauses describing an action the Subject will be associated with the semantic
role oI actor, or agent, as in |5i|. But many clauses don't express actions: we ear/ an
explosion, Ior example, describes a sensory experience, and here the Subject is associated
with the role oI experiencer. There are numerous diIIerent kinds oI semantic role that can
be associated with the Subject: what the role is in a particular instance will depend on the
meaning oI the clause, especially oI the verb.
Meaning thereIore does not provide a reliable way oI identiIying the Subject. But this
Iunction has a good Iew distinctive grammatical properties which together generally
make it easy to identiIy. Here are some oI them.

(a) Position. Its deIault position - the one it occupies unless there are special reasons Ior
placing it elsewhere - is beIore the Predicate.

(b) Formation oI interrogatives. You can generally change a declarative clause into an
interrogative by inverting the Subject with the Iirst auxiliary verb; iI there is no auxiliary
in the declarative you need to insert the appropriate Iorm oI do.
|3|
In either case the
Subject ends up Iollowing the auxiliary verb:



|6| D0cla7ativ0 Int077ogativ0
i a. %e boss is in er office. b. Is te boss in er office?
ii a. Everyone signe/ te petition. b. Di/ everyone sign te petition?

(c) Interrogative tags. To seek conIirmation oI a statement you can add an interrogative
tag, consisting oI an auxiliary verb and a personal pronoun Subject which relates back to
the Subject oI the clause to which the tag is attached: %e boss is in er office,
isnt se?; Everyone signe/ te petition, /i/nt tey?

(d) Subject-verb agreement, Where the verb has person-number properties (in the present
tense and the past tense oI be), they are normally determined by agreement with the
Subject:

|7| a. er son plays te piano. b. er sons play te piano.

5.2 P70dicato7 om5l0m0nt8 and Adj:nct8

At the next layer oI structure below the Predicate we distinguish three Iunctions. The
Predicator is the Iunction Iilled by the verb. The verb is the Head oI the verb phrase, and
Predicator is the special term used Ior the Head oI the verb phrase Iorming the Predicate
oI a clause. Thus in |7b| play te piano is a verb phrase Iunctioning as Predicate
while play is a verb Iunctioning as Predicator.
Complement and Adjunct are diIIerent kinds oI Dependent, distinguished by
the lic0n8ing condition. Complements can occur only iI they are licensed by the Head
verb: the verb must belong to a subclass that permits (or requires) a Complement oI the
type in question. Adjuncts are not subject to this restriction. Compare:

|8| I mowe/ te lawn before it starte/ to rain.

Here te lawn is admissible because the verb 2ow (unlike disappear, Ior example) allows
a Dependent oI this kind, so te lawn is a Complement. But a Dependent indicating time
can occur with any verb, so before it starte/ to rain is an Adjunct.
We will look Iurther at Complements in the next subsection. As Ior Adjuncts, they are
usually realised by adverb phrases, preposition phrases, subordinate clauses, or a very
narrow range oI noun phrases. They can be divided into various semantic subtypes, such
as Adjuncts oI time, place, manner, etc., as illustrated in |9|:

|9| i $e spoke very clearly. |adverb phrase as Adjunct oI
manner|
ii As a result of is action, e was fire/. |prep phrase as Adjunct oI
reason|
iii We cycle to work to save te busfare. |subordinate clause as Adjunct oI
purpose|
iv %ey left te country last week. |noun phrase as Adjunct oI
time|

5.3 Obj0ct and P70dicativ0 om5l0m0nt
Two important subtypes oI Complement are the Object and the Predicative Complement:

|10| a. Obj0ct: E/ blame/ te minister. b. P70dicativ0 om5: E/ was a
minister..

While thousands oI verbs license an Object, only a Iairly small number license a
Predicative Complement, and oI these be is by Iar the most common: others
include beco2e, re2ain, appear, see2, etc. The term `Predicative Complement' is most
easily understood by reIerence to the construction with be: the verb has little meaning
here (it is oIten called just a `linking verb'), so that the main semantic content oI the
Predicate is expressed by the Complement.
There are several grammatical properties that distinguish Objects Irom Predicative
Complements, oI which the two most important ones are illustrated in |11|:

|11| i a. E/ blame/ te minister. |Object| b. %e minister was blame/ by
E/.
ii a. E/ was a minister. |Pred Comp| b. *A minister was been by E/.
iii a. E/ was innocent. |Pred Comp| b. *E/ blame/ innocent.

o The Object oI an active clause can usually become the Subject oI a corresponding
passive clause, but a Predicative Complement never can. Thus the Object oI active |ia|
corresponds to the Subject oI passive |ib|, whereas |iib| is not a possible passive version
oI |iia|. (Here and below the asterisk indicates that what Iollows is ungrammatical.)
o A Predicative Complement can be realised not only by a noun phrase, as in |iia|, but
also by an adjective phrase, as in |iiia|, whereas an Object cannot be realised by an
adjective phrase, as evident Irom the ungrammaticality oI |iiib|.

5.4 Di70ct and Indi70ct Obj0ct
A clause may contain two Objects, distinguished as Direct and Indirect. In canonical
clauses, the Indirect Object always precedes the Direct Object, and typically (but not
invariably) is associated with the semantic role oI recipient or beneIiciary:

|12| i e gave te prisoner some water. |Indirect Object (recipient)
Direct Object|
ii $e bake/ me a cake. |Indirect Object (beneIiciary) Direct
Object|

5.5 S:bj0ctiv0 and Obj0ctiv0 P70dicativ0 om5l0m0nt8
The Predicative Complements in |10b| and |11iia/iiia| are related to the Subject, but it is
also possible Ior a Predicative Complement to be related to the Object: we accordingly
distinguish two subtypes, Subjective and Objective. Compare:

|13| S:bj0ctiv0 P70d om5 Obj0ctiv0 P70d om5
i a. e became angry. b. %is ma/e im angry.
ii a. e was a carlatan. b. %ey consi/ere/ im a carlatan.

5.6 iv0 canonical cla:80 8t7:ct:708
On the basis oI the presence or absence oI the Complement types considered so Iar we
can distinguish the Iollowing canonical clause structures:

|14| xam5l0 St7:ct:70 Nam0
i %ey /isappeare/. S-P O7dina7y) int7an8itiv0
ii %ey were ecstatic. S-P-PC
s
om5l0x-int7an8itiv0
iii %ey bougt a ouse. S-P-O
d
O7dina7y) monot7an8itiv0
iv %ey kept it warm. S-P-O
d
-PC
o
om5l0x-t7an8itiv0
v %ey sent er some flowers. S-P-O
i
-O
d
Dit7an8itiv0

In the representations oI the structures, S stands Ior Subject, P Ior Predicator, PC

s
Ior
Subjective Predicative Complement, O
d
Ior Direct Object, PC
o
Ior Objective Predicative
Complement, and O
i
Ior Indirect Object. The names reIlect the Iact that there are two
dimensions oI contrast:
o One has to do with Objects: an intransitive clause has no Object, a monotransitive
clause has a single Object, and a ditransitive clause has two Objects.
o The other has to with Predicative Complements: iI a clause contains a Predicative
Complement it is complex, otherwise ordinary, though the latter term is oIten omitted (as
it is in |v|, since there is no possibility oI adding a Predicative Complement to a
ditransitive clause).

The names apply in the Iirst instance to the clause constructions, and then
derivatively to the verbs that appear in these constructions. Thus disappear is an
(ordinary) intransitive verb, be a complex-intransitive one, and so on. But it must be
borne in mind that the majority oI verbs can appear in more than one oI them, and hence
belong to more than one class. Find, Ior example, commonly appears in |iii| (We foun/
te key), |iv| (We foun/ er co-operative), and |v| (We foun/ er a fob).



5.7 P705o8itional and cla:8al om5l0m0nt8.
The Complements considered so Iar have been noun phrases or adjective phrases, but
these are not the only possibilities. Complements oIten have the Iorm oI preposition
phrases or subordinate clauses:

|15| i a. $e went to !aris. b. $e took im to !aris.
ii a. $e relie/ on er instinct. b. e congratulate/ er on er
promotion.
iii a. e sai/ e was sorry. b. e tol/ er e was sorry.
iv a. We inten/ to leave on %ues/ay. b. I a/vise you to leave on %ues/ay.

In the |a| examples here the underlined preposition phrase (|i-ii|) or subordinate clause
(|iii-iv|) is the only Complement, while in the |b| ones it Iollows an Object. We look at
diIIerent kinds oI subordinate clause in Section13, but there is one point to be made here
about the prepositional constructions. In |i| to contrasts with other prepositions such
as over, from, via, beyon/, etc., but in |ii| on is selected by the verb: any adequate
dictionary will tell you (iI only by example) that rely takes a Complement
with on, consist with of, refer with to, and so on. Verbs like these that take as
Complement a preposition phrase headed by some speciIied preposition are called
`prepositional verbs'. Most ditransitive verbs also belong to this latter class by virtue oI
licensing a preposition phrase with to or for instead oI the Indirect Object: compare e
gave some water to te prisoner and $e bake/ a cake for me with |12| above.

6 VERBS
6.1 V07b infl0ction

The most distinctive property oI verbs is their inIlection: they have a number oI
inIlectional Iorms that are permitted or required in various grammatical constructions.
The present tense Iorm takes, Ior example, can occur as the verb oI a canonical clause,
whereas the past participle taken cannot: $e takes care, but not *$e taken care.
The great majority oI verb lexemes have six inIlectional Iorms, as illustrated in |16|:

|16| i P70t07it0 cecke/ $e cecke/ te figures erself.
ii 37d 8ing:la7 57080nt cecks $e cecks te figures erself.
iii Plain 57080nt ceck %ey ceck te figures temselves.
iv Plain fo7m ceck $e may ceck te figures erself.
v 07:nd-5a7tici5l0 cecking $e is cecking te figures erself.
vi Pa8t 5a7tici5l0 cecke/ $e a/ cecke/ te figures erself.

It will be noticed that although we have distinguished six diIIerent infl0ctional fo7m8,
there are only Iour diIIerent 8ha508: cecke/, cecks, ceck and cecking. By `shape' we
mean the spelling or pronunciation. Thus the preterite and past participle oI the
lexeme check have the same shape, as do the plain present tense and the plain Iorm. The
same applies to all other 70g:la7 verbs, i.e. verbs whose inIlectional Iorms are
determined by general rules. But there are a good number oI irregular verbs where the
preterite and past participle do not have the same shape: 9ake, Ior example, has took as its
preterite and taken as its past participle.
This means that it is very easy to decide whether any particular instance oI the
shape ceck is a preterite Iorm or a past participle. What you need to do is ask which
Iorm oI a verb like 9ake would be needed in the construction in question. Consider, then,
the Iollowing examples:

|17| i $e may ave cecke/ te figures erself.
ii Im not sure weter se cecke/ te figures erself or not.



II we substitute 9ake Ior check in |i| the Iorm we need is the past participle taken: $e
may ave taken a break. So this cecke/ is likewise a past participle. And iI we make the
substitution in |ii| we need the preterite Iorm took: Im not sure weter se took a break
or not. So the cecke/ oI |ii| is the preterite Iorm. Note that when making the substitution
you need to keep constant what precedes the verb (e.g. $e may ave in |i|) since this is
what determines the inIlection that is required: what Iollows the verb is irrelevant and
hence can be changed to suit the verb you are substituting.
Let us now brieIly review the six Iorms.

(a) Preterite. This is a type oI past tense: the type where the past tense is marked
inIlectionally rather than by means oI an auxiliary verb. Many grammars use the more
general term `past tense': we preIer the more speciIic term to distinguish it Irom the
construction where the auxiliary have marks the other kind oI past tense, as in $e as
cecke/ te proofs.

(b)-(c) The present tense Iorms. There are two present tense Iorms, one which occurs
with a 3rd person singular subject, and one which occurs with any other subject: 1st
person (I ceck), 2nd person (you ceck) or plural (tey ceck). We could call this latter
Iorm `non-3rd person singular', but `plain present' is simpler. `Plain' indicates that it is
identical with the morphological ba80 oI the lexeme, i.e. the starting-point Ior the rules
that produce the various inIlectional Iorms by adding a suIIix, changing the vowel, and so
on.

(d) The plain Iorm. This is also identical with the base, but it is not a present tense Iorm.
It is used in three constructions:

|18| i Im507ativ0 Ceck te figures yourself'
ii Infinitival Its better to ceck te figures oneself. I will ceck tem myself.
iii S:bj:nctiv0 Its essential tat se ceck te figures erself.

The inIinitival construction is very oIten marked by to, but it is also Iound
without to aIter such verbs as can, 2ay, will, do ($e /i/nt ceck te figures
erself), 2ake (%ey ma/e me ceck te figures myself), etc. The subjunctive is much the
least Irequent oI the three constructions and belongs to somewhat Iormal style.
There are two major Iactors that distinguish the plain Iorm Irom the plain present:
o The verb be is highly exceptional in its inIlection in that it has three present tense
Iorms instead oI the usual two (is, am, are) and all oI these are diIIerent in shape Irom the
plain Iorm be. It's the latter Iorm that appears in the three constructions shown in
|18|: Be quiet (imperative); Its better to be safe tan sorry, I will be rea/y in
time (inIinitival); Its essential tat se be tol/ (subjunctive). So we can tell whether a
given instance oI ceck, say, is the plain present or the plain Iorm by using the
substitution test illustrated above, but this time substituting the verb be. Thus
the ceck oI We must ceck te figures is a plain Iorm, not a plain present tense because
we need the plain Iorm oI be in this position: We must be careful.
o The plain present doesn't occur with 3rd person singular Subjects, but the plain
Iorm does. Compare $e cecks te figures erself (not *$e ceck te figures erself)
and $e will ceck te figures erself (not *$e will cecks te figures erself).

(e) The gerund-participle. This Iorm always ends with the suIIix ing. Traditional
grammar distinguishes two Iorms with this suIIix, the gerund and the present participle:

|19| i Cecking te figures can be onerous |traditional
gerund|
ii !eople cecking te figures must be alert |traditional present
participle|



The idea was that a gerund is comparable to a noun, while a participle is comparable to
an adjective. Thus in |i| cecking te figures is comparable to suc cecks,
where cecks is a noun; in |ii| cecking te figures is ModiIier to people and was

thereIore considered adjective-like since the most common type oI ModiIier to a noun is
an adjective.
|4|
There is, however, no verb in English that has distinct Iorms Ior the
constructions in |19|, and so there is no basis Ior making any inIlectional distinction here
in Present-day English: we thus have a single Iorm and the name `gerund-participle'
indicates that it covers both traditional categories.

(I) The past participle. This is used in two main constructions, the perIect and the passive:

|20| i P07f0ct $e as cecke/ te figures.
ii Pa88iv0 %e figures must be cecke/ by te boss.

The perIect is a past tense marked by the auxiliary verb have, while the most
straightIorward cases oI the passive involve the auxiliary verb be. We retain the
traditional term `past participle', though the `past' component oI meaning applies just to
the perIect construction.

6.2 Th0 infl0ctional t0n80 8y8t0m
We have seen that there are two inIlectional tenses in English: preterite and present; we
review now the major uses oI these tenses.

(a) Preterite. Three uses can be distinguished, as illustrated in |21|:

|21| i Pa8t tim0 a. e arrive/ yester/ay. b. $e knew im well.
ii ac8hift a. E/ sai/ e was ill. b. I tougt it starte/ tomorrow.
iii odal 70mot0n088 a. I wis I knew te answer. b. I/ /o it if you pai/ me.

o In |i| we see the basic use, indicating past time. The event oI his arriving took place
in the past, and the state oI her knowing him well obtained in the past (it may still obtain
now, but I'm talking about some time in the past). This is much the most Irequent use, but
it's important to be aware that the preterite doesn't always have this meaning.
o Example |iia| could be used to report Ed's saying `I am ill': present tense am is
shiIted back to preterite was under the inIluence oI the preterite reporting verb sai/. In
|iib| my original thought was `It starts tomorrow': again present tensestarts is shiIted
back to preterite starte/. This example shows very clearly that the backshiIt use is not the
same as the past time use, Ior clearly the starting is not in the past.
o In |iii| the preterite has a modal rather than temporal meaning: it has to do with
Iactuality, not time. In |iiia| the subordinate clause has a counterIactual meaning under
the inIluence oI wis: you understand that I don't know the answer. The time is present,
not past: I don't know it now. The conditional |iiib| is not counterIactual (it doesn't rule
out the possibility oI your paying me), but it envisages your paying me as a somewhat
remote possibility - rather less likely than with the present tense counterpart Ill /o it if
you pay me. Note that the time oI your possibly paying me is in the Iuture. We use the
term `modal remoteness' to cover both these interpretations (as well as others mentioned
brieIly in Section6.5).

(b) The present tense. The two most important uses are seen in |22|:

|22| i P7080nt tim0 a. I promise Ill elp you. b. $e lives in $y/ney.
ii :t:70 tim0 a. Exams start next week. b. Ill go ome wen it gets /ark.

o In |i| we again have the basic and much the most common use: to indicate present
time. In |ia| the event oI my promising is actually simultaneous with the utterance, Ior I
perIorm the act oI promising by saying this sentence. In |ib| we have a state, and the
present tense indicates that the state obtains at the time oI speaking.


o In |ii| the time is Iuture. In main clauses this is possible only when the event is in
some way already scheduled, as in |iia|. But this constraint does not apply in various
kinds oI subordinate clause such as we have in |iib|.

6.3 A:xilia7y v07b8
We turn now to the important subclass oI verbs called a:xilia7y v07b8, or a:xilia7i08:
they are quite markedly diIIerent in their grammatical behaviour Irom other verbs, which
are called l0xical v07b8.

.3.1 Membersip of te class
The main members oI the auxiliary class are shown in |23|, where they are divided into
two subclasses, modal and non-modal:

|23| i odal a:xilia7i08 can, 2ay, 2:s9, will, shall, o:h9, need, dare
ii Non-modal a:xilia7i08 be, have, do

(Coul/, migt, woul/ and soul/ are the preterite Iorms
oI can, 2ay, will and shall respectively, though they diIIer considerably Irom other
preterites, as we shall see.)

.3.2 Distinctive properties
There are several constructions which require the presence oI an auxiliary verb, the two
most Irequent oI which involve Subject-auxiliary inversion and negation.

(a) Subject-auxiliary inversion. We have seen that in canonical clauses the Subject
precedes the verb whereas in most interrogative main clauses the Subject Iollows the
(Iirst) verb. The verb that precedes the Subject, however, must be an auxiliary verb: only
auxiliaries can invert with the Subject. Compare:

|24| A:xilia7y v07b L0xical v07b
i a. $e as taken te car. b. $e took te car.
|declarative|
ii a. as se taken te car? b. *%ook se te car?
|interrogative|

II the declarative doesn't contain an auxiliary, as in |ib|, it is necessary to insert the

auxiliary do so that inversion can apply: Di/ se take te car? This do has no meaning: it
is simply inserted to satisIy the grammatical rule requiring an auxiliary.

(b) Negation. The construction where not is used to negate the verb likewise requires that
the verb be an auxiliary:

|25| A:xilia7y v07b L0xical v07b
i a. $e as taken te car. b. $e took te car.
|positive|
ii a. $e as not taken te car. b. *$e took not te car. |negative|

Again, iI there is no auxiliary in the positive, do must be inserted to Iorm the
negative: $e /i/ not take te car.
A Iurther, related, point is that auxiliaries, but not lexical verbs, have negative Iorms
ending in the suIIix nt: a more inIormal variant oI |25iia| is $e asnt taken te car.

.3.3 Auxiliaries as ea/s
It is important to emphasise that auxiliaries contrast with lexical verbs, not with what
some grammars call `main verbs'. Auxiliaries Iunction as Head, not Dependent, in verb
phrase structure. They mostly take non-Iinite clauses as Complement, like many lexical
verbs. Compare the examples in |26|, where the verb phrase is enclosed in brackets, the
Head is in capitals and underlining marks the non-Iinite clause Iunctioning as its
Complement:



|26| A:xilia7y v07b a8 H0ad L0xical v07b a8 H0ad
i a. %ey |&G% to accept te offer|. b. %ey |IN%END to accept te offer|.
ii a. We |CAN answer teir queries|. b. We |EL! answer teir queries|.
iii a. $e |WA$ cecking te figures|. b. $e |BEGAN cecking te figures|.
iv a. e |WA$ attacke/ by a /og|. b. e |G% attacke/ by a /og|.

The particular type oI non-Iinite clause that is used depends on the Head verb, whether
auxiliary or lexical. :h9 and in9end license inIinitivals
with to, can and help inIinitivals without to; be, in one oI its uses, and bein license a
non-Iinite clause with a gerund-participle Iorm oI the verb; be, in a second use,
and e9 license one with a past participle Iorm oI the verb.
Note, then, that the verb phrase in |iiia|, say, is divided into was cecking te figures,
not was cecking te figures, just as that in |iiib| is divided into began cecking te
figures, not began cecking te figures. And similarly with the other examples.

6.4 Th0 non-modal a:xilia7i08 be have do
Little Iurther need be said about do: it is used in constructions like Subject-auxiliary
inversion and negation when required to satisIy the requirement that the construction
contain an auxiliary. There is also a lexical verb do used in clauses like$e /i/ er

best, I /i/ im an infustice, etc.; here, then, auxiliary do must be added to Iorm
interrogatives and negatives: Di/ se /o er best?, I /i/nt /o im an infustice.

(a) e. Three uses oI be can be distinguished, illustrated in :

|27| i P7og7088iv0 ma707 a. %ey are watcing %'. b. Ive been working all
morning.
ii Pa88iv0 ma707 a. It was taken by Jill. b. e may be arreste/.
iii o5:la a. $e was a frien/ of is. b. %at is very likely.

o In |i|, where be is Iollowed by a verb in the gerund-participle Iorm, it is a marker oI
progressive aspect. It generally serves to indicate that the situation - the action, event,
state, or whatever - was, is or will be in progress at the time in question.
o The clauses in |ii| are passive; |iia| is the passive counterpart oI active Jill took it,
the presence oI be being one oI the major diIIerences between the two Iorms. There is no
active counterpart oI |iib| because the latter has no by phrase (cI. Section15).
o In |iii| be is the only verb, but it still behaves as an auxiliary. Thus the interrogative
oI |a| is Was se a frien/ of is? and the negative oI |b| is %at isnt very likely. In these
examples the auxiliary has as its Complement not a non-Iinite clause but a noun phrase (a
frien/ of is) and an adjective phrase (very likely).

(b) ave. This verb belongs to both lexical and auxiliary classes. In $e a/ a swim it is a
lexical verb, Ior the interrogative and negative counterparts are Di/ se ave a
swim? and $e /i/nt ave a swim. The auxiliary uses are seen in |28|:

|28| i P07f0ct ma707 a. e as broken is leg. b. e may ave taken it
yester/ay.
ii Static have a. $e as enoug cre/it. b. We ave to invite tem all.



o The perIect is marked by auxiliary have a past participle. It is best regarded as a
secondary past tense - the primary past tense being the inIlectional preterite. Note, Ior
example, that the preterite is Iound only in Iinite constructions such as e took it
yester/ay, so it can't occur aIter 2ay (cI. *e may took it yester/ay: 2ay takes an
inIinitival clause as Complement), and perIect have is then used instead, as in |ib|.
Since have itselI can inIlect Ior tense, |ia| is doubly marked Ior tense: it is `past in
present', the past being marked by the lexeme have and the present by the inIlection
on have. This reIlects the Iact that while the event oI his breaking his leg is located in
past time it is seen as having relevance to the present. The most likely scenario is that his
leg has not yet healed, so that he is at present incapacitated. The present tense component
also explains why it is not normally possible to add an Adjunct like yester/ay: *e as
broken is leg yester/ay.
o ave in |ii| denotes a state, unlike that oI the above $e a/ a swim, which is
dynamic, denoting an event. Usage is divided as to whether static have is an auxiliary or
a lexical verb. Those who say $e asnt enoug cre/it or ave we to invite tem all? and

the like are treating it as an auxiliary, while those who say $e /oesnt ave enoug
cre/it or Do we ave to invite tem all? are treating it as a lexical verb. Many people use
both constructions, though the lexical verb treatment has been gaining ground Ior some
time. Note that in |iia| have, like be in |27|, doesn't have a non-Iinite clause as
Complement.

6.5 Th0 modal a:xilia7i08
In this section we Iirst note that need and dare, like do and have above, belong to both
auxiliary and lexical verb classes; we next set out the main grammatical properties that
deIine the class oI modal auxiliaries, then consider the preterite Iorms, and Iinally look at
the kinds oI meaning they express.

(a) eed and dare. These are auxiliaries only when Iollowed by an inIinitival
construction without to, as in Nee/ I boter? and I /arent tell tem, etc. Thus in I nee/ a
aircut, I nee/ to get my air cut, I /are you to repeat tat, etc. we have lexical verbs.

(b) Distinctive grammatical properties oI modal auxiliaries
o They have only tensed Iorms: no plain Iorm, no gerund-participle, no past
participle. Hence the impossibility oI *I/ like to may go wit you; *Were musting invite
tem all, *$e as coul/ speak Frenc since se was a cil/.
o They are invariable in the present tense instead oI having a distinct Iorm in s used
with 3rd person singular Subjects: $e can swim, not *$e cans swim, etc.
o With one exception they license a Iollowing inIinitival Complement without to: $e
can swim, not *$e can to swim. The exception is o:h9: %ey ougt to accept te
offer (|26ia|).

Note that although We ave to invite tem all has essentially the same meaning as We
must invite tem all, this have is not a modal auxiliary: it has none oI the above three
grammatical properties. It is a special case oI the static haveillustrated in |28ii|, and as
such it is Ior many speakers not an auxiliary at all, but a lexical verb.

(c) The preterite Iorms. Coul/, migt, woul/ and soul/ are the preterite Iorms
oI can, 2ay, will and shall respectively, but the use oI these preterites diIIers Irom that oI
other preterite Iorms in Present-day English.
o Only coul/ and woul/ have the basic preterite use oI indicating past time: I coul/ /o
it easily wen I was younger; I aske/ im to elp but e woul/nt.
o The status oI migt and soul/ as preterites is established by their use in certain
conditional constructions and in those cases oI reported speech or thought where present
tense Iorms are excluded. Thus though we can have may in If you come back tomorrow
you may fin/ im in, we need migt in If you came back tomorrow you migt fin/ im
in.
|5|
And iI at some time in the past I had the thought `I shall easily Iinish beIore she
returns' I would report this with soul/, as inI knew I soul/ easily finis before se
returne/ (not *sall).
o The major diIIerence is that while with other verbs the modal remoteness use oI the
preterite is restricted to certain kinds oI subordinate clause, with the modal auxiliaries it
occurs in main clauses and with a wider range oI interpretation; with migt and soul/ it

is overwhelmingly the most Irequent use. The preterites tend to be weaker, more tentative
or polite than the present tense Iorms.



(d) Types oI modal meaning. The modal auxiliaries express a considerable variety oI
meanings, but they can be grouped into three major types.
o 5i8t0mic modality. Here we are concerned with what is necessary, likely or
possible: e must ave overslept; Dinner soul/ be rea/y in a few minutes; $e may be
ill.
o D0ontic modality. Here it is a matter oI what is required or
permitted: You must work ar/er; You soul/ be stu/ying for your exam; You can/may go
wit tem if you like.
o Dynamic modality. Here it is a question oI properties or dispositions oI persons or
other entities involved in the situation: $e can speak very persuasively (ability), Will you
elp me? (willingness). This kind oI meaning is mainly Iound with
just can, will and dare.

In some cases there is a clear ambiguity as to which type oI meaning is intended. You
must be very tactful, Ior example, can be interpreted epistemically (I'm inIerring Irom
evidence that you are very tactIul) or deontically (I'm telling you to be very
tactIul). $e cant be serious may be understood epistemically (She is obviously not being
serious) or dynamically (She is unable to be serious).

7 NOUNS AND NOUN PHRASES
7.1 Di8tinctiv0 57o507ti08 of no:n8
Nouns Iorm much the largest word class. It contains all words that denote physical
entities, but also great numbers oI words that do not have this semantic property: in order
to be able to identiIy nouns we thereIore need to examine their grammatical properties.
We consider them under three headings: inIlection, Iunction and dependents.

(a) InIlection. Nouns generally exhibit inIlectional contrasts oI n:mb07 and ca80:

|29| N:mb07 a80
Plain Genitive
i Singular stu/ent stu/ents
ii Plural stu/ents stu/ents

School grammars commonly use the term `possessive' instead oI `genitive', but that term
is Iar too speciIic Ior the wide range oI relationships covered by this case: compare, Ior
example, Kims parents, te boys beaviour, te trains arrival, te mayors obituary, te
suns rays, to/ays news.

(b) Function. Nouns can Iunction as Head in noun phrases that in turn Iunction as Subject
or Complement in clause structure, or Complement oI a preposition, as illustrated in |30|,
where nouns are underlined and noun phrases bracketed:

|30| i S:bj0ct in cla:80 |ne stu/ent| was arreste/.
ii om5l0m0nt in cla:80 %ey interviewe/ |all te stu/ents|.
iii om5l0m0nt in 5705 5h7a80 %e talk was given by |a stu/ent|.

(c) Dependents. There are some kinds oI Dependent that occur exclusively (or almost
exclusively) with a noun as Head:

|31| i 07tain d0t07minativ08 te stu/ent, a scool, every book, wic exam
ii P70-h0ad adj0ctiv08. mature stu/ents, a new book, an easy exam
iii #0lativ0 cla:808 te stu/ent wo /irecte/ te play, a book Im rea/ing

7.2 Th0 8t7:ct:70 of no:n 5h7a808
Noun phrases typically consist oI a Head noun alone or accompanied by one or more
Dependents. The Dependents are oI three main types: Determiners, Complements and
ModiIiers.



(a) Determiners. These are Iound uniquely in the structure oI noun phrases. They have the
Iorm oI determinatives (or determinative phrases, as in almost all stu/ents, not
many people, too few volunteers) or genitive noun phrases (te girls voice,some
peoples beaviour, my book).

Determiners serve to mark the noun phrase as d0finit0 or ind0finit0.

|32| i D0finit0 te !remier of N$W, te key, tis book, bot copies, te
mans /eat
ii Ind0finit0 a politician, some keys, any serious book, enoug copies, tree /ogs

We use a deIinite noun phrase when we assume that its content is suIIicient, in th0
cont0xt, to identiIy the reIerent. There's only one (current) Premier oI NSW, so the
deIiniteness in the Iirst example is unproblematic, but with the second example there is oI
course very heavy reliance on context to make the reIerent clear. %e is a pure marker oI
deIiniteness, known as th0 d0finit0 a7ticl0. Its use eIIectively pre-empts
a wic question: iI I say Weres te key? I assume you won't need to ask Wic
key? Note that a genitive Determiner conIers deIiniteness on the noun phrase: te mans
/eat means `the death oI the man', and a mans /eat likewise means `the death oI a
man'. Noun phrases like black coffee andfrien/s, which have a common noun as Head
and no Determiner are normally indeIinite.

(b) Complements. The clearest cases oI Complements involve preposition phrases where
the preposition is speciIied by the Head noun, and certain types oI subordinate clause:

|33| i P705o8ition 5h7a808 er review of te play, a ban on alcool, is
marriage to $ue

ii S:bo7dinat0 cla:808 te i/ea tat e migt be ill, an opportunity to make frien/s

Note that nouns, unlike verbs, do not take Objects: we say $e reviewe/ te play, but not
*er review te play; instead we need of te play. With ban and 2arriae the
prepositions required are on and to. The subordinate clauses in |ii| clearly satisIy the
licensing test: only a Iairly narrow range oI nouns can take Complements like these.

(c) ModiIiers. The typical pre-Head ModiIier is an adjective or adjective
phrase: a goo/ book, a very serious matter. But those are not the only possibilities. In
particular, nouns can also Iunction as ModiIier to a Head
noun: a scool play, teunemployment situation, etc. Post-Head ModiIiers are typically
preposition phrases and subordinate clauses that occur more Ireely than Complements in
that they do not have to be licensed by the Head noun: a man of onour, te
ouseopposite te post office, te play tat se wrote, te guy wo spoke first.
It is also possible to have ModiIiers that precede the Determiner: all te books, bot tese
plays, too small a car for our nee/s. Note that adverbs can occur in this position, but not
aIter the Determiner: absolutely te best solution, but not *an absolutely success. Instead
oI the latter we need an adjective, an absolute success.

7.3 N:mb07 and co:ntability
(a) Nouns with Iixed number. Although most nouns have an inIlectional contrast between
singular and plural, there are a good Iew that do not - that have only singular or only
plural Iorms:

|34| i Sing:la7-only
no:n8 crockery, /ross, arm, nonsense; news, mumps, pysics, ...
ii Pl:7al-only no:n8 belongings, clotes, genitals, scissors; cattle, police, ...

Note that the last three items in |i| end in s but are nevertheless singular, as evident, Ior
example, Irom the agreement in %is news is goo/. Conversely, the last two items in |ii|
don't end in s, but are nevertheless plural: cI. %ese cattle arein goo/ ealt.

(b) Count and non-count nouns. Related to the distinction between nouns with variable
number and nouns with Iixed number is that between co:nt and non-co:nt nouns. Count
nouns can take cardinal numerals (one, two, tree, etc.) as Dependent, while non-count
nouns cannot. Compare count s9:den9 (one stu/ent, two stu/ents) and non-
count har2 and clo9hes (*one arm/clotes, *two arms/clotes).


However, most nouns can occur with either a count or a non-count int07570tation:

|35| o:nt int07570tation Non-co:nt int07570tation
i a. e pulle/ out a wite air. b. e as wite air.
ii a. ave anoter cake. b. ave some more cake.
iii a. Can I borrow your football. b. Lets play football.

The interpretations in |a| allow Ior a contrast between one and more than one (cI., Ior
example, e pulle/ out two wite airs), but those in |b| do not. When we speak oI count
and non-count nouns, thereIore, we are reIerring to nouns as used with a count and non-
count interpretation. Thus air is a count noun in |ia|, a non-count noun in |ib|, and so
on.

(c) Subject-verb agreement. We noted in Section5.1 that where a verb has person-number
properties they normally agree with those oI the Subject noun phrase, more particularly
with those oI the Head noun oI that noun
phrase: %e /og isbarking vs %e /ogs are barking. There are, however, certain
semantically-motivated types oI departure Irom this pattern, as illustrated in |36|:

|36| i 0a8:70 0x57088ion8 %wo ours isnt long enoug for suc a fob.
ii ":antificational no:n8 A lot of people like it.
iii oll0ctiv0 no:n8 %e fury avent yet reace/ a /ecision.

o In |i| the hours aren't thought oI individually but as making up a single period, so
the Subject is treated as singular.
o In |ii| the verb-Iorm is determined not by the Head noun lot but by people, which is
embedded within the Subject noun phrase.
o With collective nouns like fury in |iii| there is divided usage, with
singular asnt also used.

7.4 S:bcla8808 of no:n
There are three main subclasses oI noun: common no:n, 57o507 no:n and 57ono:n.
Common noun is the deIault subclass and needs no Iurther comment here.

(a) Proper nouns. This subclass includes nouns such
as Jon, Mary, $mit, Beetoven, $y/ney, Egypt, Nile, Easter, Fri/ay, etc. They
characteristically Iunction as Head oI noun phrases serving as 57o507 nam08, names
individually assigned to particular people, places, Iestivals, days oI the week, and so on.
Note, however, that they also occur, derivatively, in other kinds oI noun phrase: %ats
not te $mit I was referring to, Lets listen to some Beetoven. Conversely, not all
proper names contain proper nouns: cI. Central Avenue, New Years Day, and so on. And
some proper names contain more than just a proper noun: te Nile, Mt Everest, King
Jon.

(b) Pronouns. The grammatically distinctive property oI pronouns is that they do not
normally combine with Determiners: e arrive/, not *%e e arrive/. There are several
subtypes oI pronoun, including:

|37| i P078onal 57ono:n8 , we, yo:, he, she, i9, 9hey, one
ii #0ci57ocal 57ono:n8 each o9her, one ano9her
iii Int077ogativ0 o7 70lativ0 57ono:n8 who, wha9, which, whoever, etc.

We will comment here on only the Iirst oI these categories. Personal pronouns are those
where we Iind contrasts oI 5078on. and we are Iirst person, used to reIer to the speaker
or a group containing the speaker. (`Speaker' is to be understood as covering the writer in
written texts.) Yo: is second person, used to reIer to the addressee or a group containing
one or more addressees. The others are third person: this doesn't encode reIerence to
speaker or addressee and thereIore usually reIers to entities other than the speaker or
addressee. But I can reIer to myselI or to you in the third person: %e writer as notice/
...; %e rea/er may recall ...
The personal pronouns have Iive inIlectional Iorms:


|38| i Nominativ0 I, we, you, ... I /i/ it. It was I wo /i/ it.
ii Acc:8ativ0 me, us, you, ... It bit me. It was me wo /i/ it.
iii D050nd0nt g0nitiv0 my, our, your, ... My son is ere. I saw your car.
iv Ind050nd0nt g0nitiv0 mine, ours, yours, ... Mine was broken. %ats mine.
v #0fl0xiv0 myself, ourselves, ... I urt myself. We talk to ourselves.

Nominatives occur mostly as Head oI a Subject noun phrase. In Iormal style they can also
occur in certain types oI Predicative Complement, with the accusative as a less Iormal
variant: It was I/me wo /i/ it. In other types, however, only the accusative is
possible: %e victim was me, not *%e victim was I, and the like. Dependent genitives
occur when there is a Iollowing Head in the noun phrase, independent ones when there
isn't. ReIlexives usually relate back to the Subject noun phrase, as in the above examples.

8 ADJECTIVES AND ADJECTIVE PHRASES
.1 Two majo7 f:nction8 of adj0ctiv08
Most adjectives can be either att7ib:tiv0 or 570dicativ0:

|39| i Att7ib:tiv0 a ot /ay, some new D'Ds, tis excellent play, lonely people
ii P70dicativ0 Its ot. %ese look new. I foun/ it excellent. %ey seem lonely.

Attributive adjectives are pre-head ModiIiers in noun phrase structure; predicative
adjectives are Predicative Complements in clause structure (see Section5.5).
|6|

There are, however, some adjectives that are restricted to one or other oI these Iunctions:

|40| i Att7ib:tiv0-
only te main speaker, a mere cil/, te only problem, my own car
ii N0v07-att7ib:tiv0 Im afrai/. $es asleep. e looks content. Its liable to floo/.

.2 7adability and g7ad0
The most central adjectives are g7adabl0: they denote properties that can apply in
varying degrees. As such, they can be modiIied by adverbs oI degree and (under
conditions relating to length and Iorm) be inIlected Ior com5a7ativ0 (e.g. taller)
and 8:507lativ0 (e.g. tallest) g7ad0:

|41| i D0g700 modification very goo/, quite ot, rater young, too
ol/, incre/ibly ba/
ii Infl0ction fo7
g7ad0 otter, younger, ol/er, better; ottest, youngest, ol/est, best

Gradable adjectives that don't inIlect mark comparative and superlative degree by means
oI the adverbs more and most respectively: more intelligent, most intelligent.
There are also a good number oI adjectives that denote non-scalar properties and hence
are non-
g7adabl0: alpabetical or/er, te cief /ifficulty, te fe/eral government, er rigt eye, t
ir/ place. Some adjectives, moreover, can be used in two diIIerent senses, one gradable,
the other non-gradable (and usually the more basic). In %e /oor is open, Ior
example, open is non-gradable, but in You soul/ be more open wit us it is gradable.

.3 Th0 8t7:ct:70 of adj0ctiv0 5h7a808
Adjective phrases consist oI an adjective as Head, alone or accompanied by one or more
Dependents, which may be Complements or ModiIiers:

|42| i om5l0m0nt8 goo/ at cess, grateful for your elp, fon/ of
animals, keen on golf,
gla/ tat you like/ it, unsure wat a/ appene/, eager to elp
ii odifi078 very ba/, morally wrong, tis goo/, most useful, muc better, two
/ays long, a bit ol/, cautious to excess, /angerous in te extreme



The Complements are preposition phrases or subordinate clauses; in the Iormer case the
adjective selects a particular preposition to head the
Complement: 1ond takes of, keen takes on, and so on. The ModiIiers are adverbs
(e.g. very), determinatives (tis), noun phrases (two /ays) or post-Head prepositional
phrases. Adjective phrases containing post-Head Dependents cannot normally be used
attributively: es goo/ at cess, but not *a goo/ at cess scoolboy.

9 ADVERBS AND ADVERB PHRASES.
.1 Adv07b8 in 70lation to adj0ctiv08
The majority oI adverbs are derived Irom adjectives by adding the suIIix ly: common -
commonly, rare - rarely, etc. There are a good number oI adverbs not Iormed in this
way, some oI them very common
(e.g. almost, always, often, quite,rater, soon, too, very), but these are normally
recognisable as adverbs by virtue oI being replaceable by ones with the ly suIIix:
compare Its very goo/ and Its extremely goo/; $e always wins and $e frequently wins;
Itll be over soon and Itll be over sortly, and so on.
The major diIIerence between adverbs and adjectives has to do with their Iunctions. We
have seen that adjectives Iunction attributively or predicatively, but adverbs do not
normally occur in these Iunctions: compare attributive a successfulmeeting, not
*a successfully meeting, and predicative %e meeting was successful, not *%e meeting

was successfully. Adverbs Iunction as ModiIier to a wide range oI word or phrase classes,
as illustrated in |43|, where underlining marks the modiIying adverb and capitals what it
modiIies:

|43| Adv07b modifying:
i V07b $e $!KE clearly. $e !LAYED well.
ii Adj0ctiv0 Its a remarkably GD play It looks very GD
iii Adv07b e spoke virtually INA&DIBLY. %ey almost NE'ER
reply.
iv D0t07minativ0 Nearly ALL copies were sol/. %oo FEW copies were
printe/.
v P705 5h7a80 $e is completely IN CN%RL. Its quite BEYND
BELIEF.
vi #08t of cla:80 $urprisingly E'ERYNE AGREED Frankly, I%$ &$ELE$$.

In general adverbs that can modiIy adjectives and other adverbs can also modiIy verbs,
but there are some exceptions, most notably very and too (in the sense `excessively').
Compare es very FND of er and *e very L'E$ er (we neede loves
er very M&C).
A Iew adverbs inIlect Ior grade (soon, sooner, soonest), but Ior the most part
comparatives and superlatives are marked by more and most: more carefully, most
carefully.

.2 Th0 8t7:ct:70 of adv07b 5h7a808
The structure oI adverb phrases is broadly similar to that oI adjective phrases, but
simpler: in particular, very Iew adverbs license complements.

|44| i om5l0m0nt8 Luckily for me, it raine/. We an/le/ it similarly to te
oters.
ii odifi078 $e sang very well. It wont en/ tat soon. We left a bit late.

10 PREPOSITIONS AND PREPOSITION PHRASES.
The most central members oI the preposition class have meanings concerned with
relations in time or space: after lunc, at scool, before te en/, in te gar/en, off te
bri/ge, on te /esk, etc. In this section we look at the Iunction oI prepositions and then at
their Complements, and Iinally consider the phenomenon oI preposition stranding.

(a) Function oI prepositions. Prepositions Iunction as Head in preposition phrases, and
these in turn Iunction as Dependent (Complement or ModiIier) to any oI the Iour major
parts oI speech:



|45| P705 5h7a80 d050nd0nt on:
i V07b $e WEN% to Lon/on. %ey ARE in te gar/en.
ii No:n es a MAN of principle. Its on te WAY to !aris.

iii Adj0ctiv0 $es IN%ERE$%ED in politics. Im RE$!N$IBLE for

tem
iv Adv07b L&CKILY for me, no-one knew. I saw er LA%ER in te
/ay.

(b) Complements oI prepositions. Usually (as in all the examples in |45|) prepositions
take a noun phrase as Complement. There are, however, other possibilities:

|46| i P705o8ition 5h7a80 e emerge/ |from un/er te be/. Ill stay |until after
lunc|.
ii Adj0ctiv0 5h7a80 %at strikes me |as unfair|. I took im |for /ea/|.
iii Adv07b 5h7a80 I /i/nt know |until recently |. I cant stay |for long|.
iv la:80 It /epen/s |on wat se says|. I tol/ er |before se
left|.
|7|


(c) Preposition stranding. In a number oI clause constructions the Complement oI a
preposition is placed at the Iront oI the clause or omitted altogether, leaving the
preposition `stranded':

|47| i a. Wat are you looking at? b. Its someting |wic I can
/o witout|.
ii a. %is is te book |I was referring to|. b. e went to te same scool as |I
went to|.

The construction is characteristic oI relatively inIormal style, but it is a serious mistake to
say that it is grammatically incorrect.

11 NEGATION
(a) Clausal vs subclausal negation. Negation is marked by individual words such
as not, no, never, or by aIIixes such as we have in uncommon, non-
compliant, infrequent, careless, isnt, wont, etc. We need to distinguish, however,
between cases where the negative aIIects the whole clause (cla:8al n0gation) and those
where it aIIects just a part oI it (8:bcla:8al n0gation):

|48| i la:8al n0gation a. e is not well. b. $urprisingly, e wasnt
ill.
ii S:bcla:8al n0gation a. e is unwell. b. Not surprisingly, e was ill.

The clauses in |i| are negative, but those in |ii| are positive even though they contain a
negative element within them. We say this because they behave like obviously positive
clauses with respect to the constructions shown in |49|:

|49| Int077ogativ0 tag8 nd so v8 and nor
i Po8itiv0 a. e is well, isnt e? b. $urprisingly, e was ill an/ so was se.
ii N0gativ0 a. e is not well, is e? b. $urprisingly, e wasnt ill an/ nor was
se.

iii Po8itiv0 a. e is unwell, isnt e? b. Not surprisingly, e was ill an/ so was
se.

o In |a| we have a clause Iollowed by an interrogative `tag' used to seek conIirmation
oI what has been said. The usual type oI tag reverses the `polarity' oI the clause to which
it is attached - that is, it is negative iI attached to a positive clause, as in |ia|, and positive
iI attached to a negative clause, as in |iia|. And we see Irom |iiia|, thereIore, that e is
unwell counts as positive since the tag is negative: the clause is no more negative than e
is sick.


o In the |b| examples we have added a truncated clause introduced by an/ so or an/
nor. We get an/ so aIter a positive clause and an/ nor aIter a negative one. And Not
surprisingly, e was ill is shown to be a positive clause because it takes an/ so.

(b) Non-aIIirmative items. There are a number oI words or expressions that occur readily
in negative or interrogative clauses but generally not in positive declaratives. Compare:

|50| D0cla7ativ0 Int077ogativ0
i N0gativ0 a. e /i/nt fin/ any cracks. b. Di/nt e fin/ any cracks?
ii Po8itiv0 a. *e foun/ any cracks. b. Di/ e fin/ any cracks?

Instead oI |iia| we say e foun/ some cracks. Such items as any in |50| are called non-
affi7mativ0 (with `aIIirmative' understood as combining declarative and positive). They
include compounds with any, such as anybo/y, anyone, anyting, etc., at
all, eiter, ever, yet, bu/ge, can bear, can stan/, ive a /amn, li19 a finger, etc. More
precisely, these are non-aIIirmative in at least one oI their senses: some oI them also have
senses in which they can occur in aIIirmative constructions. The any series oI words, Ior
example, can occur in aIIirmative constructions when the meaning is close to `every', as
in Anyone can /o tat.

12 CLAUSE TYPE AND SPEECH ACTS
We use sentences to make statements, ask questions, make requests, give orders, and so
on: these are diIIerent kinds oI 8500ch act (a term understood, like `speaker', to cover
writing as well as speech). The grammatical counterpart iscla:80 ty50, where we
distinguish declarative, interrogative, and so on. The main categories we recognise here
are illustrated in |51|:

|51| i D0cla7ativ0 You are very tactful.
ii lo80d int077ogativ0 Are you very tactful?
iii O50n int077ogativ0 ow tactful are you?
iv xclamativ0 ow tactful you are'
v Im507ativ0 Be very tactful.

We use diIIerent terms Ior the clause types than Ior the speech acts because the relation
between the two sets oI categories is by no means one-to-one. Consider such examples as
|52|:

|52| i Youre leaving alrea/y?
ii I ask you again were you were on te evening of 14 July.
iii I promise to elp you.
iv Woul/ you min/ opening te /oor for me?

Grammatically, |i| is declarative, but it would be used as a question: a question can be
marked by rising intonation (or by punctuation) rather than by the grammatical structure.
Example |ii| is likewise declarative but again it would be used as a question (perhaps in a
court cross-examination): the question Iorce this time comes Irom the verb ask, in the
present tense with a 1st person Subject. Pro2ise in |iii| works in the same way: this
example would generally be used to make a promise. This illustrates the point that
although we have just a handIul oI diIIerent clause types there are a great many diIIerent
kinds oI speech act: one can apologise, oIIer, congratulate, beseech, declare a meeting
open, and so on. Finally, |iv| is a closed interrogative but would characteristically be used
to make a request. In this use it is what is called an indi70ct 8500ch act: although it is
literally a question it actually conveys something else, a polite request.
All canonical clauses are declarative and we need say no more about this type, but a Iew
comments are in order Ior the remaining Iour types.



(a) Closed interrogatives. These are so called because they are typically used to ask
questions with a closed set oI answers. Usually these are Yes and No (or their
equivalents), but in examples like Is it a boy or a girl? they derive Irom the terms joined
by or: Its a boy and Its a girl. Grammatically they are marked by Subject-auxiliary
inversion (though such inversion is not restricted to interrogatives: in the
declarative Never a/ I felt so embarrasse/ it is triggered by the initial placement oI the
negative never).

(b) Open interrogatives. These are typically used to ask questions with an open set oI
answers (e.g. very, quite, sligtly, etc. in the case oI |51iii|). They are marked by the
presence oI an interrogative phrase consisting oI or containing a so-called `w-
word': wo, wat, wen, were, ow, etc. This phrase may be Subject (Wo sai/ tat?),
Complement (Wat /o you want?) or Adjunct (Wen /i/ e leave?). II it is Complement
or Adjunct it normally occurs at the beginning oI the clause, which has Subject-auxiliary
inversion, as in the last two examples. It is possible, however, Ior it to remain in post-
verbal position, as in An/ after tat you went were? (a construction most likely to be
Iound in a context oI sustained questioning).

(c) Exclamatives. These have, at the Iront oI the clause, an exclamative phrase containing
either ow, as in |51iv|, or wat, as in Wat a fool Ive been'

(d) Imperatives. The most common type oI imperative has you understood, as in |51v|, or
expressed as Subject (as in You be careful; Dont you speak to me like tat). The verb is
in the plain Iorm, but do is used in the negative: Dont move. We also have 3rd person
imperatives like $omebo/y open te win/ow, distinguished Irom the declarative precisely
by the plain Iorm verb. 1st person plural imperatives are marked by lets: Lets go', Dont
lets boter.

13 SUBORDINATE CLAUSES
Subordinate clauses normally Iunction in the structure oI a phrase or a larger clause.
Whereas main clauses are almost invariably Iinite, subordinate clauses may be Iinite or
non-Iinite.

13.1 init0 8:bo7dinat0 cla:808.
The most central type oI finit0 clause is t0n80d, i.e. contains a verb inIlected Ior tense
(preterite or present tense), and most Iinite subordinate clauses are oI this type. There is,
however, one construction containing a plain Iorm oI the verb that belongs in the Iinite
class, the 8:bj:nctiv0:

|53| i $e says tat e is kept well-informe/ |tensed: is is present
tense verb|
ii $e insists tat e be kept well-informe/ |subjunctive: be is plain
Iorm|

Subjunctive is thus the name oI a syntactic construction, not an inIlectional category, as
in traditional grammar. It has a plain Iorm verb and when the Subject is a personal
pronoun it appears in nominative case.
We distinguish three main types oI Iinite subordinate clause: cont0nt cla:808, 70lativ0
cla:808 and com5a7ativ0 cla:808.

13.1.1 Content clauses
These usually Iunction as Subject or else Complement oI a verb, noun, adjective or
preposition:

|54| i %at tey accepte/ te offer is very fortunate.
|Subject|
ii I KNW se likes it. |Complement oI
verb|
iii %e FAC% tat its so ceap makes me suspicious. |Complement oI
noun|
iv We staye/ in BECA&$E it was raining. |Complement oI
preposition|

Like main clauses they select Ior clause type, except that there are no subordinate
imperatives:

|55| i D0cla7ativ0 e /i/nt know tat everybo/y supporte/ te

proposal.
ii lo80d int077ogativ0 e /i/nt know weter everybo/y supporte/ te
proposal.
iii O50n int077ogativ0 e /i/nt know wic proposal everybo/y supporte/.
iv xclamativ0 e /i/nt know wat a lot of tem supporte/ te
proposal..

o Declaratives are oIten marked by the subordinator tat; and since tat occurs in
both the tensed clause and the subjunctive in |53| we include both in the declarative class.


o Closed interrogatives have weter or if instead oI the Subject-auxiliary inversion
Iound in main clauses (compare the main clause counterpart oI the subordinate clause in
|ii|: Di/ everybo/y support te proposal?).
o Open interrogatives have the interrogative phrase in initial position and normally no
Subject-auxiliary inversion (again compare the main clause counterpart oI that in
|iii|: Wic proposal /i/ everybo/y support?).
o Exclamatives mostly have the same Iorm as their main clause counterparts, as with
|iv|.

13.1.2 Relative clauses
(a) Relative clauses as ModiIier. The most central kind oI relative clauses Iunctions as
ModiIier in noun phrase structure:

|56| i a. I agree wit |te guy wo spoke last|. b. I agree wit |te guy tat
spoke last|.
ii a. e lost |te key wic I lent im|. b. e lost |te key I lent im|.

Such clauses contain an overt or covert element which relates back to the Head noun, so
we understand in |i| that some guy spoke last and in |ii| that I lent him a key. This
`relativised element' is overt in |ia| (the relative pronoun wo) and |iia| (wic), but
covert in the |b| examples. This is obvious in the case oI |iib|, and in |ib| tat, although
traditionally classiIied as a relative pronoun, is better regarded as a subordinator, the
same one as is Iound in declarative content clauses like |55i|; on this analysis there is no
overt relativised element in |ib| any more than in |iib|.
The relativised element can have a variety oI Iunctions in the relative clause: in |56i| it is
Subject, in |56ii| Object, and so on.

(b) Supplementary relative clauses. The relative clauses in |56| are
tightly int0g7at0d into the structure oI the sentence, but it is also possible Ior relative
clauses to be set oII by punctuation or intonation, so that they have the status oI more
loosely attached S:55l0m0nt8, as in:

|57| i Ive lent te car to my broter, wo as fust come over from New Zealan/.
ii e overslept again, wic ma/e im miss te train.

In this type the relativised element is almost always overt, and doesn't relate back to a
noun but to a larger unit, a whole noun phrase in |i| (my broter) and a clause in |ii|,
where wic is understood as `(the Iact) that he overslept again'.

(c) The Iused relative construction. This is structurally more complex than the above
constructions:

|58| i a. Woever wrote tis must be very naive. b. You can invite wo you like.
ii a. e quickly spent wat se gave im. b. Wat books e as are in te attic.

The underlined sequences here are not themselves clauses but noun phrases: clauses don't
denote entities that can be naive or be invited or spent or located in the attic. Note,
moreover, that are in |iib| agrees with a plural noun phrase Subject, whereas Subjects
with the Iorm oI clauses take 3rd person singular verbs, as in |54i|. Woever in |58ia| is
equivalent to te person wo and wat in |iia| to tat wic, and so on. This is why we
call this construction `Iused': the Head oI the noun phrase and the relativised element are
Iused together, instead oI being separate, as in |56ia/iia|.
These constructions may look superIicially like open interrogative content clauses.
Compare |58iib|, Ior example, with I aske/ er wat se gave im. The meaning is quite
diIIerent: the latter, where the underlined clause is interrogative, can be glossed as `I
asked her the answer to the question, `What did she give him?'', but there is no such
question meaning in |58iia|. Similarly compare |58iib|, meaning `The (Iew) books he has
are in the attic', with Wat books e as is unknown, where the underlined clause is
interrogative and the meaning is `The answer to the question `What books does he have?'
is unknown'; note that this time the main clause verb is singular is, agreeing with the
clausal Subject.



13.1.3 Comparative clauses
Comparative clauses generally Iunction as Complement to the prepositions as and tan:

|59| i a. Im as rea/y as I ever will be. b. As was expecte/, $ue won
easily.
ii a. More people came tan I/ expecte/. b. e as more vices tan e as virtues.

The distinctive property oI such clauses is that they are structurally incomplete relative to
main clauses: there are elements understood but not overtly expressed. In |ia| and |iia|
there's a missing Complement and in |ib| a missing Subject. Even in |iib| there's a
missing Dependent in the Object noun phrase, Ior the comparison is between how many
vices he has and how many virtues he has. The Iact that there's some kind oI understood
quantiIier here is reIlected in the Iact that we can't insert an overt one: *e as more
vices tan e as ten virtues.

13.2 Non-finit0 8:bo7dinat0 cla:808

There are three major kinds oI non-Iinite clause:

|60| i Infinitival a. e wants to see you. b. I cant elp you.
ii 07:nd-5a7tici5ial a. Buying a car was a mistake. b. es te guy stan/ing
up.
iii Pa8t-5a7tici5ial a. All tings consi/ere/, its K. b. We got tol/ off.

InIinitivals contain a plain Iorm oI the verb, with or without the special marker to;
gerund-participials and past-participials have verbs in the gerund-participle and past
participle Iorms; Ior Iurther examples, see |26| above.
Most non-Iinite clauses have no overt Subject, but all three kinds allow one under certain
conditions.
o In inIinitivals, it occurs in the to-variant with initial for as subordinator: For tem to
be so late is very unusual.
o In gerund-participials a personal pronoun Subject usually appears in accusative
case, but genitives are Iound in relatively Iormal style: We obfecte/ to tem/teir being
given extra privileges.
o Example |iiia| is a past-participial with an overt Subject.

InIinitivals are much the most Irequent oI the three classes oI non-Iinite clause, and
appear in a very wide range oI Iunctions. These include Subject (%o err is uman),
Complement oI a verb (as in |60ia/b|: the Head verb determines whetherto is included),
Complement oI a noun (I applau/ |er willingness to compromise|), Complement oI an
adjective ($es |willing to compromise|), Adjunct ($e walks to work to keep fit),
ModiIier oI a noun (I nee/ |an album to keep te potos in|). In general, prepositions take
gerund-participials rather than inIinitivals as Complement (e left |witout saying goo/-
bye|), but the compound in or/er and so as are exceptions ($e staye/ at ome |in
or/er to stu/y for te exam|).

14 COORDINATION
Coordination is a relation between two or more items oI equal syntactic status,
the coo7dinat08. They are oI equal status in the sense that one is not a Dependent oI
another.

(a) The marking oI coordination. Coordination is usually but not invariably marked by
the presence oI a coordinator, such as an/, or, nor, but; the Iirst three oI these may also
be paired with a determinative, bot, eiter and neiter respectively. The main patterns
are seen in |61|:

|61| i We ave no milk an/ te sops arent open yet.
ii er broter came too, but /i/nt stay long.
iii We can meet on Mon/ay, on %urs/ay or at te week-en/.
iv We can meet on Mon/ay, or on %urs/ay or at te week-en/.
v Bot Jill an/ er usban/ atten/e/ te meeting.
vi e was self-confi/ent, /etermine/, egotistical.

Examples |i|-|iii| illustrate the most usual case: a coordinator in the last coordinate. In
|iv| there is a coordinator in all non-initial coordinates, in |v| a determinative in the Iirst,
and in |vi| no overt marking oI coordination at all.

(b) Functional likeness required between coordinates. Coordination can appear at more or
less any place in the structure oI sentences. You can have coordination between main
clauses (giving a compound sentence, as in |61i|), between subordinate clauses, between
phrases, between words (e.g. ave you seen my fater an/ moter?). But the coordinates
need to be grammatically alike. Usually they belong to the same class, as in all the
examples in |61|. They do not have to be, however: the crucial constraint is that they be
alike in Iunction. Compare, then:

|62| i $e is very brigt an/ a goo/
lea/er.
ii I /ont know te cause of te acci/ent or ow muc /amage was /one.
iii Were leaving Rome an/ next week.

o In |i| we have coordination between an adjective phrase and a noun phrase, and in
|ii| between a noun phrase and a subordinate clause (an open interrogative content
clause). These are acceptable because each coordinate could stand on its own with the
same Iunction: in $e is very brigt and $e is a goo/ lea/er the underlined units are
both Predicative Complements, and in I /ont know te cause of te acci/ent and I /ont
know ow muc /amage was /one they are both Complements.
o But |iii| is unacceptable, even though the coordinates are oI the same class, noun
phrase, because the Iunctional likeness condition is not met. The Iunction
oI Rome in Were leaving Rome is Complement, whereas that oI next week inWere
leaving next week is Adjunct.

(c) Joint coordination. One special type oI coordination is seen in |63|:

|63| a. $am an/ !at are a appy couple. b. $am !at an/ Alex like eac
oter.

What is distinctive about this type is that the properties concerned, being a happy couple
and liking each other, apply to the coordinates jointly rather than separately. So we can't
say *$am is a appy couple or *!at likes eac oter. The Iunctional likeness in this type
is that the coordinates denote members oI a set to which the relevant property applies.
The construction is more restricted than the type illustrated in |61| in that it excludes
determinatives (*Bot $am an/ !at are a appy couple), doesn't allow but as coordinator,
and does require likeness oI class between the ccoordinates.

15 INFORMATION PACKAGING
The grammar oI the clause makes available a number oI constructions that enable us to
express a given core meaning in diIIerent ways depending on how we wish to to present

or `package' the inIormation. For example, Kim broke te vase,%e vase was broken by
Kim, %e vase Kim broke, It was Kim wo broke te vase, Wat Kim broke was te
vase all have the same core meaning in the sense that there is no situation or context in
which one oI them would be true and another Ialse (assuming oI course that we are
talking oI the same Kim and the same vase). The Iirst oI them, Kim broke te vase, is the
syntactically most basic, while the others belong to various info7mation-5acaging
con8t7:ction8. The most important oI these constructions are illustrated by the
underlined examples in |64|:



|64| Nam0 xam5l0 a8ic co:nt075a7t
i P705o8ing a. %is one you can keep. b. You can keep tis one.
ii Po8t5o8ing a. Ive lent to Jill te only copy b. Ive lent te only copy tat
tat as been correcte/. as been correcte/ to Jill.
iii Inv078ion a. In te bag was a gol/ watc. b. A gol/ watc was in te bag.
iv Pa88iv0 a. %e car was /riven by $ue. b. $ue /rove te car.
v xi8t0ntial a. %ere was a /octor on boar/. b. A /octor was on boar/.
vi xt7a5o8ition a. Its clear tat se is ill. b. %at se is ill is clear.
vii l0ft a. It was Kim tat suggeste/ it. b. Kim suggeste/ it.
viii P80:do-cl0ft a. Wat I nee/ is a col/ /rink. b. I nee/ a col/ /rink.
ix Di8location a. Its excellent, tis curry. b. %is curry is excellent.

In the Iirst three we are concerned simply with the order oI elements, while the
others involve more radical changes.
o The basic position Ior the Complement tis one in |i| is aIter the verb, but in |a| it is
preposed, placed at the Iront oI the clause.
o In |ii| the basic position Ior the Object, te only copy tat as been correcte/, is just
aIter the verb but long or complex elements like this can be postposed, placed at the end.
o In |iii| the positions oI the Subject and Complement oI the basic version |b| are
reversed in the inversion construction |a|. (More precisely, this is Subject-Dependent
inversion, in contrast to the Subject-auxiliary inversion construction discussed earlier.
The Dependent is usually a Complement but can also be an Adjunct, as in %ree /ays
later came news of er /eat.)
o In |iv| (the only one where the basic version has a distinct name, `active') the Object
becomes Subject, the Subject becomes Complement oI by and the auxiliary be is added.
o The existential construction applies mainly with the verb be: the basic Subject is
displaced to Iollow the verb and the semantically empty pronoun tere takes over the
Subject Iunction.
o In |vib| the Subject is a subordinate clause (tat se is ill); in |a| this is extraposed,
placed aIter the verb phrase and this time the Subject Iunction is taken over by the
pronoun it.
o In |vii| the cleIt clause is Iormed by dividing the basic version into two parts: one
(Kim) is highlighted by making it Complement oI a clause with it as Subject and be as
verb, while the other is backgrounded by relegating it to a subordinate clause (a distinct
subtype oI relative clause).

o The pseudo-cleIt construction is similar, but this time the subordinated part is put in
a Iused relative (wat I nee/) Iunctioning as Subject oI be.
o Dislocation belongs to Iairly inIormal style. It diIIers Irom the basic version in
having an extra noun phrase, set apart intonationally and related to a pronoun in the main
Subject-Predicate part oI the clause. In the l0ft di8location variant the pronoun occurs to
the leIt oI the noun phrase; in 7ight di8location it is the other way round, as in is fater,
se cant stan/ im.

There are two Iurther comments that should be made about these constructions.

(a) Basic counterpart need not be canonical. For convenience we have chosen examples
in |64| where the basic counterparts are all canonical clauses, but oI course they do not
need to be. The basic (active) counterpart oI passive Was te car /riven by Kim? is Di/
Kim /rive te car?, which is non-canonical by virtue oI being interrogative. Likewise the
non-cleIt counterpart oI It was $ue wo a/ been interviewe/ by te police is $ue a/
been interviewe/ by te police, which is non-canonical by virtue oI being passive: note
then that certain combinations oI the inIormation-packaging constructions are possible.



(b) The inIormation-packaging construction may be the only option. The second point is
that under certain circumstances what one would expect to be the basic counterpart is in
Iact ungrammatical. Thus we can say %ere was an acci/ent, but not *An acci/ent was:
here the existential construction is the only option. One diIIerence between actives and
passives is that the by phrase oI the passive is an optional element whereas the element
that corresponds to it in the active, namely the Subject, is generally obligatory in Iinite
clauses. Compare, then:

|65| i Pa88iv0 a. $ome mistakes were ma/e by E/. b. $ome mistakes were
ma/e.
ii Activ0 a. E/ ma/e some mistakes. b. *Ma/e some mistakes.

Passives like |ib| - called 8ho7t 5a88iv08 - thus have no active counterpart. They are in
Iact the more common type oI passive, allowing inIormation to be omitted that would
have to be expressed in the active construction.