Professional Documents
Culture Documents
Andrew Spencer
Room: 4.334
email: spena
privatewww.essex.ac.uk/∼spena
2
1 Basics of morphosyntax 11
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2 Lexical vs functional categories . . . . . . . . . . . . . . . . . . . . 11
1.3 Content words . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.1 Nouns, verbs, adjectives . . . . . . . . . . . . . . . . . . . . 12
1.3.2 Homonymy between classes . . . . . . . . . . . . . . . . . 13
1.3.3 Prepositions, adverbs . . . . . . . . . . . . . . . . . . . . . 13
1.4 Function words and functional categories . . . . . . . . . . . . . . 15
1.4.1 Conjunctions . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.4.2 Determiners . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.4.3 Pronouns . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.5 Lexemes and word forms . . . . . . . . . . . . . . . . . . . . . . . 16
1.6 Functional categories and features . . . . . . . . . . . . . . . . . . 17
2 Inflection 19
2.1 Noun inflection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.1.1 Plurals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.1.2 Count vs. mass nouns . . . . . . . . . . . . . . . . . . . . . 19
2.1.3 Possessor . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.1.4 Nominal function words . . . . . . . . . . . . . . . . . . . . 20
2.1.5 Possessive pronouns . . . . . . . . . . . . . . . . . . . . . . 21
2.1.6 Reflexives . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.1.7 Demonstratives . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.2 Verb inflection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.2.1 Analytic vs. synthetic forms: the tense/aspect system . . 23
2.2.2 Participles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2.3 -ing form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3
4 CONTENTS
2.3 Adjectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.3.1 Comparative and superlative . . . . . . . . . . . . . . . . . 26
2.3.2 Comparative and superlative - inflection or derivation? . 26
3 Words 29
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2 Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3 Clitics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.4 Types of word . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4 Dictionaries 35
4.1 Introduction - What is a dictionary? . . . . . . . . . . . . . . . . . 35
4.2 Describing words . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.2.1 Lexical entry . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.2.2 The dictionary entry . . . . . . . . . . . . . . . . . . . . . . 37
4.2.3 Polysemy/homonymy . . . . . . . . . . . . . . . . . . . . . 38
4.3 Types of dictionary . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.4 Dictionary entries and English . . . . . . . . . . . . . . . . . . . . 41
6 Grammatical functions 53
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
6.2 Subjects and complements . . . . . . . . . . . . . . . . . . . . . . . 53
6.3 Modification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.3.1 Modification of nouns . . . . . . . . . . . . . . . . . . . . . 59
6.3.2 Modification of verbs and sentences . . . . . . . . . . . . . 60
6.3.3 Modification of other categories . . . . . . . . . . . . . . . 61
6.4 Agreement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
6.5 Basic parsing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
7.2 Passive . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
7.3 Middle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
7.4 Object Omission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
7.5 Double Object constructions . . . . . . . . . . . . . . . . . . . . . . 71
7.6 Locatum-location alternations - spray∼load . . . . . . . . . . . . . 72
7.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
17 Conversation 179
17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
17.2 Conversational (discourse) particles . . . . . . . . . . . . . . . . . 180
17.3 Sentence accent in discourse structure . . . . . . . . . . . . . . . . 181
17.3.1 Rhythm and Intonation . . . . . . . . . . . . . . . . . . . . 181
17.3.2 De-accenting . . . . . . . . . . . . . . . . . . . . . . . . . . 181
17.4 Focus, accent and information structure . . . . . . . . . . . . . . . 182
17.5 Tone units in English . . . . . . . . . . . . . . . . . . . . . . . . . 184
17.6 Intonation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
8 CONTENTS
9
Chapter 1
Basics of morphosyntax
1.1 Introduction
We distinguish lexical (content) words and function words. Lexical and func-
tion words are grouped into categories (‘parts of speech’) depending on their
distribution in phrases.
‘Lexical’ means ‘to do with words’ (as opposed to phrases, sentences, etc).
‘Category’ means words are grouped together into classes because they have
similar linguistic properties and hence behave in similar ways in sentences.
We can define lexical categories in terms of
• the typical meanings they have (semantics)
• the forms the words take (inflectional morphology)
• where they occur in sentences (syntactic distribution)
11
12 1. BASICS OF MORPHOSYNTAX
Basic meanings
Noun
• typically has special possessive form: the boy’s name/the boys’ names, the
children’s parents
(but notice also the girl in the corner’s parents)
Adjective
• many adjectives don’t have these forms, specifically those with more than
two syllables: *intelligenter/*intelligentest
Verb
A regular verb has three forms with a special ending (suffix) and one base
form:
Present tense she sleeps, walks, . . .
Past tense she slept, walked, ate . . .
-ing participle she is sleeping
base form (no endings) they sleep, to sleep, she may sleep
Perfect participle she has slept/eaten the apple
Passive participle the apple was eaten
Distribution of words
Noun
A noun typically co-occurs with:
1.3. CONTENT WORDS 13
• Determiners
• Various other phrases: the girl in the picture, the knife which Bill has sharpened,
...
Adjective
An adjective co-occurs with:
Verb
A verb co-occurs with:
In English things are complicated by the fact that many words can belong to
more than one class:
A preposition (very nearly) always comes before a noun phrase and serves to
relate the meaning of that noun phrase to some other element, e.g. (The money
is) in the box.
14 1. BASICS OF MORPHOSYNTAX
An adverb is free-standing.
Typical meanings for prepositions: place relating one thing to another
The box is) under the table
(The vase standing) on the table . . .
The prepositions in, on, under are associated with clear spatial meanings.
This isn’t always the case: in a way, on purpose, under these circumstances.
The easiest way to understand what kinds of meanings prepositions express
is to see typical examples (note that many place prepositions also double as
time prepositions: on the table, on Tuesday).
place:: in, on, off, at, under, over, within, inside, outside, to, towards, from,
around, by, between, against, . . .
time:: after, before, to, till/until, from, within, by, during, throughout, . . .
Adverbs: typically modify verbs (in the same way that adjectives modify
nouns)
1.4.1 Conjunctions
1.4.2 Determiners
Determiners: These are function words which help determine the kind of
thing a noun refers to.
• Numerals: self-explanatory
• Quantifiers: Like numerals except that they don’t refer to a precise num-
ber:
every/each girl, all children (under five), all the children, many people, most cats,
few dogs, a few dogs, little water, a little water, less food, more wine
1.4.3 Pronouns
Given the notion of inflection we can distinguish two more senses of ‘word’.
Consider the verb swim:
The forms {swim, swims, swam, swimming, swum} are inflected word forms of
the lexeme swim. The lexeme itself combines information about its word class,
inflections, and meaning:
1.6. FUNCTIONAL CATEGORIES AND FEATURES 17
We can give two (or more) distinct morphosyntactic descriptions to the form
/sheep/, /put/. Thus, we have one word form but two (or more) morphosyn-
tactic words (sometimes called grammatical words).
Summary:
We must distinguish the following ways of describing ‘words’
Grammatical relations
subject (or Nominative Case)
object (or Accusative/Objective Case)
adjunct (or adverbial)
possessor
modifier (e.g. attributive adjective)
Nominal features
definite
number
Pronominal features
person:
number:
gender:
18 1. BASICS OF MORPHOSYNTAX
Adjectival features
comparative:
superlative:
Verbal features
tense: {past, present, future}
aspect: {simple, perfect, progressive}
voice: {active, passive}
mood: {indicative, imperative, interrogative}
modality: {various modal auxiliary verbs}
polarity: {positive, negative}
Agreement
SubjAgr: {3sg}
Number:{singular, plural}
Tense:{Past, NonPast}
SubjAgr:{Person:3, Number:Sg}
1 You might have noticed that that I’ve distinguished here between those features that govern
syntactic structures, written as [syntactic feature: value] and those features that tell us how to
form an inflected word, written as [Morphological feature:value]. But quite often we’ll ignore
that distinction.
Chapter 2
Inflection
2.1.1 Plurals
Some words only have a plural form: scissors, trousers, shears, glasses/spectacles
(shades)
Usually these are single artefacts with paired parts.
The technical name for such words is: plurale tantum noun (plural: pluralia
tantum).
There are also nouns which only occur in the singular, though most of these
are mass/uncountable nouns (see next section).
One or two nouns have the form of singular nouns but are plural: police
A few nouns have irregular plurals that are the same as the singular form:
sheep, deer.
19
20 2. INFLECTION
Only count nouns can take a plural form. However, many nouns seem to be
in both categories:
This is common behaviour for other nouns, where a count or mass interpreta-
tion is forced on a mass/count noun - a process often known as coercion.
2.1.3 Possessor
BT voicemail message:
You were called today at 9.00. The caller didn’t leave their number
Also: Someone called earlier but they didn’t say who they were
<Why is the plural form (they/their) used here?>
Other types of pronoun:
-one pronouns: anyone, someone, everyone
-body pronouns: anybody, somebody, everybody
<Is there any difference between these?>
-self pronouns: see below
(This parallels demonstratives v. inf.)
2.1.6 Reflexives
Work out what the basic rule is for using reflexive pronouns.
2.1.7 Demonstratives
The only modifiers which have special plural agreement form are this/these,
that/those (see also Chapter 6).
Can be used as either a modifier or as the head of the nominal phrase (like
possessives):
Tense
On the other hand, verbs only have two tense inflections. In fact, the so-called
future tense is really just one use of a pair of modal auxiliary verbs (will/shall
and not a tense in any recognizable sense.
The term ‘past tense’ is slightly misleading since the past tense verb forms
are often used in contexts with have nothing to do with past time reference.
For instance later we will see conditional clauses such as:
(14) If we left now we’d catch the train with plenty of time to spare
To emphasise the mismatch between form and past time reference linguists
often use a more neutral term as a substitute for ‘past tense’: preterite. In that
case we say that English verbs distinguish present and preterite forms.
Aspect
These are all expressed analytically, i.e. by means of a phrase which combines
the lexical verb with an auxiliary verb:
HAVE + -en Perfect
BE + -ing Progressive
We will discuss these in detail in Chapter 10.
Agreement
In verbs other than be, agreement is only found with the 3sg present tense
forms (but even then not for modal auxiliaries). The verb be retains an earlier
pattern, supernumerary agreement in present and preterite forms:
24 2. INFLECTION
Present Preterite
1sg am 1/3sg was
2nd; pl are 2sg; pl were
2sg (art)
3sg is
2.2.2 Participles
BUT with most irregular verbs we find different forms for the preterite and
past participle.
Past participle as adjective. The past participle can be used in the manner of
an adjective, even though it retains the basic meaning of the verb (this is to
be distinguished from those adjectives which derived historically from past
participles, but which are now proper adjectives, and not verb forms at all,
such as I’m very surprised you should think that
Participles are much more common as post-modifiers than as pre-modifiers:
What patterns are there in the inflection of irregular (strong) verbs? How
representative are the following verbs: keep, send, make, teach, cut, hold,
take, speak, write, run, sing?
These uses are all dealt with in greater detail in Chapter 10.
2.3 Adjectives
The semantics of the comparative and superlative is rather complex. E.g. longer
means ‘long to a greater extent than some reference point’
semantic point of view one might wish to say that they’re different lexemes
because of the significant meaning change. It’s not obvious that we want to
say that this meaning (roughly MORE and MOST) is actually grammaticalized
and hence is represented as a grammatical feature. Moreover, the compar-
ative/superlative forms have different complementation properties, because
they take a than-phrase, whereas the positive form doesn’t permit this.
28 2. INFLECTION
Chapter 3
Words
3.1 Introduction
In this chapter we examine the different sense we have to assign to the term
‘word’.
We know that ‘word’ can refer to the notion ‘lexeme’, ‘word form’ or ‘gram-
matical/morphosyntactic word’.
The term ‘word’ also denotes the objects that get distributed by rules of
syntax, that is, phrases and sentences consist of words. In Cats chase mice there
are three words, but what about The cat’s sleeping - how many words does that
consist of?
3.2 Compounds
Compounds are words which consist of more than one lexeme. Examples are:
Typically in English they have a head, the rightmost element. The head de-
fines the morphosyntactic category and also the meaning of the word. Thus,
the words in (1) are all respectively (a, b) nouns, and (c, d) adjectives. More-
over, a houseboat is a (kind of) boat while a boathouse is a (kind of) house.
Typically we assign constituent structures to compounds.
29
30 3. WORDS
(2) a. N b. N
N N N N
(3) N
N N
N N cover
N N book
coffee table
Finally, there are examples such as (give someone a good) send-off, (offend some-
one with a cruel) put-down etc. These aren’t really compounds but are rather
nominalizations of particle verbs. (See later for particle verbs.)
3.3 Clitics
The kind of clitic we are mainly concerned with is one which surfaces in the
same place in the linear syntactic string of words that the full form word would
appear in. Such clitics are called simple clitics. An example would be the re-
duced form of the pronoun them /@m/ as in ‘I haven’t finished’em yet’. This
has exactly the same distribution as the full form of the pronoun but it’s phono-
logically attached to the previous word.
(5) a. it is b. it’s
a. could have b. could’ve
a. she shall/will b. she’ll
a. we had/would b. we’d
The full forms are found (in ordinary spoken English) when the auxiliary is
stressed, to emphasise the polarity of the sentence: We HAD locked the door
32 3. WORDS
(after all). The reduced forms are cliticized to the word to the immediate left.
The clitic appears in the same syntactic position as the full form auxiliary,
which means that in principle it can attach to a word of any category. In (6) we
see this ‘promiscuity’ exemplified by its attachment to a noun (6a), a pronoun
(6b), an adverb (6c), and a preposition (6d).
Similar examples can be constructed for other clitics (see also Chapter 9 on
auxiliaries).
Consider again the question of how many words there are in a phrase such as
The cat’s sleeping.
The full form of the sentence is shown in (7).
But the form with the clitic auxiliary is a paraphrase of this and the auxiliary
clitic fulfils the same function as the full form of the auxiliary. This means that
the syntactic structure of The cat’s sleeping is that shown in (8).
Clearly, the form cat’s is a single syllable phonologically, even though it corre-
sponds to two words syntactically. We say that cat’s is a single phonological
word but two syntactic words. It’s a feature of clitics that they generally repre-
sent an independent word in syntax but form a single phonological word with
their host.
Now consider a sentence containing a compound noun such as (9), with
simplified syntactic representation (10).
The top line of (10) is intended to show the level of syntactic terminals (syn-
tactic words). There we see that there are two syntactic nouns, N1 and N2 .
However, these are compounds, consisting of two nouns themselves. That
structure isn’t, however, visible to the syntax, which treats catfood and kitchen
table as single nouns.
Now, each of the members of the compound can be thought of as a phono-
logical word in its own right. Each word bears some degree of stress, in par-
ticular. On the other hand, the determiners and the verb is are unstressed, and
effectively behave as clitics, attaching to the following stressed word as host.
This is illustrated in (11), in which the phonological words are indicated by ω
(ω . . . ).
We can now illustrate the various categories for the two sentences The cat’s
sleeping and The catfood is on the kitchen table. We will analyse the examples in
terms of the following categories:
clitic:: {’s}
phonological word:: {{D@kats}, {sli:pIN}}
(The notation <the, –> indicates that this lexeme, the definite article, has no
inflected forms.)
Dictionaries
The linguistic description of a word is the lexical entry for a lexeme. This com-
bines all those pieces of information about a lexeme that are not predictable
from more general principles, such as the rules of morphology of the language.
A typical lexical entry would be that for cat shown in (1).
(1) FORM /kat/
SYNTAX Noun
SEMANTICS ‘furry quadruped that miaows’
For other types of word we may need to include more information. For
instance, for words with irregular morphology we have to record the irregu-
larities under the heading of ‘FORM’, as in (2).
35
36 4. DICTIONARIES
(2) FORM /Si:p/
Number:plural /Si:p/
SYNTAX
Noun
SEMANTICS ‘Ovis aries’
Other details of grammatical usage may also need to be included. For instance,
the verb like can be followed by a clause of the form to do or doing, while enjoy
can only be followed by a clause of the form doing:
A dictionary entry will typically take the lexical entry for a lexeme and pro-
vide additional information relating to the lexeme’s use, its history or other
properties which are not part of the lexical entry proper.
There are other types of information that are generally recorded in a good
dictionary. Some of this information is particularly important from the point
of view of language learning.
First, a dictionary will often tell us what kinds of idiomatic combinations
a word might be used in (though this information is presented separately in
specialist dictionaries of idioms or phraseological dictionaries).
Second, a dictionary will often provide the typical sorts of expressions that
a word is found in, i.e. its collocations. This is particularly important in dic-
tionaries specifically aimed at learners, because learning words through fre-
quent collocations is a very good way of remembering vocabulary and in any
38 4. DICTIONARIES
case, using standard collocations makes ones speech sound more natural and
native-like.
Third, a good dictionary should provide information about the connotations
associated with a word. These can be extremely varied and difficult to char-
acterize. Many of them boil down to information about the register of the
word:
Finally, a dictionary will often give information about the historical origins
of a word, i.e. its etymology.
Project: Take several different kinds of dictionary (monolingual aimed
at native speakers, e.g. Concise OED, monolingual aimed at advanced
learners, multilingual, e.g. Collins-Robert French/English) and note the
sets of descriptors used for different word types. How much agree-
ment/overlap is there between dictionaries of comparable size? Do dic-
tionaries produced by the same publishers adopt the same set of de-
scriptors for each type of dictionary?
4.2.3 Polysemy/homonymy
Two words (lexemes) are homonyms (also called homophones) if they have the
same form but entirely different meanings. If they are also spelled the same
way they are homographs, too.
The simplest type of homophony/homonymy is illustrated by a verb such
as draw. This means (i) to make a pictorial image (e.g. with a pencil) or (ii)
to pull, extract, attract. The verb is irregular in exactly the same way in both
meanings: {draw, drew, drawn}.
A slightly more complex example is illustrated by the verb ring. This has
two sets of meanings. The first, which we can call ring1, is to do with making
or causing to make a ringing sound (ring a bell, the bell rings). The second,
which we can call ring2, is to do with making a ring around something (The
teacher rings mistakes in red ink) or fitting a ring onto something. However, the
homophony only applies to certain forms of the verb because ring2 is regular,
while ring1 is irregular {ring, rang, rung}.
For ring1 we can discern a number of distinct uses.
These meanings seem to be related to each other (at least at some intuitive
level), though it’s not entirely obvious what the link is. They are not, however,
related to the meanings of ring2.
The Shorter OED treats ring1, ring2 as two distinct lexical entries. This
means that ring1, ring2 are homonyms. However, it treats the separate mean-
ings listed in (11, 12) as variant meanings of a single entry, i.e. as polysemous
readings of the two lexemes. In addition, there is a further dictionary entry,
the noun ring. The Shorter OED explains that ring2 is derived from the noun
ring.
For the noun ring there are also various polysemous meanings:
Under the heading ’Noun∼Verb’ are included two sorts of difficulty. First, a
very large number of nouns can be used as verbs without any change in their
form. In morphology this is called ‘conversion’. In general, we can readily see
a relationship between the noun meaning and the verb meaning, but we can’t
actually predict what that relationship will be. For instance, the various mean-
ings of ring2 are clearly related to one or other of the meanings of ring2, but
we can’t predict which noun meanings will give rise to a verb, nor can we pre-
dict precisely what that verb will mean. To see this, consider the noun circle.
This has almost the same meaning as the noun ring and it gives rise to verbs
with various meanings. In some cases, the verb is more or less synonymous
with a corresponding meaning for ring2: The teacher ringed/circled the spelling
errors. In other cases, however, this is not true: Vultures circled/*ringed overhead;
The police ringed/*circled the demonstrators (though you can say The police encircled
the demonstrators).
We’ll look in detail at prepositional verbs and particle verbs in Chapter 8.
5.1 Introduction
This chapter revises basic notions and develops them further. We look at the
structure of phrases within sentences (phrase structure, constituent structure,
or c-structure) . An important aspect of this is the linear order of words, the fact
that we say the black cat rather than *cat black the. However, phrase/sentence
structure is much more than just linearizing individual words. A phrase such
as incredibly fascinating can itself be part of a larger phrase (e.g. an incredibly
fascinating lecture), so phrase structure is hierarchical. We then look at the
grammatical functions of subject, object, and so on (GF structure).
Adverbial clauses:
43
44 5. SIMPLE CLAUSE STRUCTURE
Relative clauses
The basic clause type is the declarative clause whose main function is to ex-
press a statement.
The interrogative clause type takes two forms, corresponding to yes-no and
wh-questions.
Constituents
Noun Phrases:
These are built around a lexical head which is a noun. (There are interesting
exceptions we’ll look at later)
cats
black cats
Verb Phrases:
fight
have been fighting each other all night behind the dustbins
Each of these phrases behaves as a single unit in certain respects. We can see
this by comparing a somewhat elaborated sentence with a much simpler one.
The two examples in (11, 12) have essentially the same structure: S = NP VP
(13) a. S
NP VP
Cats fight
b. S
NP VP
NP, VP are the immediate constituents of the sentence S (or constituents for
short).
See below for examples of adjective phrases and prepositional phrases.
Constituents of constituents:
To see how a phrase can itself contain one or more phrases we’ll build up some
simple examples.
(15) a. AP b. NP c. NP
Degree A Num N
very hungry
(16) [The black cats that live behind our house] [fight each other all night
behind the dustbins]
(17) [NP [NumP two] [AP very hungry] [N cats]]
Constituency tests
(20) a. The children put their toys into this box before bedtime
5.3. PHRASE STRUCTURE OF MAIN PHRASE TYPES IN ENGLISH 47
3) (Echo)-question test:
You can often question a constituent (but you need to use an echo question
usually; see Tallerman):
4) Focus test:
You can often focus a constituent by means of a cleft or pseudo-cleft.1
(22) a. It was [two very fat cats] that I kicked out of my garden [it-cleft]
b. What I kicked out of my garden was [two very fat cats] [pseudo-
cleft]
(23) [Two very fat cats] got kicked out of the garden (by me)
Here are some basic templates showing typical, fairly complex constructions.
Remember that only the head is obligatory in these phrases.
Noun phrases
We’ll look at NP structure in much more detail in a separate chapter. For the
present note that the basic shape of the noun phrase is (25):
Deg A
Adjective Phrase
Prepositional Phrase
The basic type consists of just Preposition + Noun Phrase, though the prepo-
sition can often be modified by a word such as right, just, . . . :
Verb Phrase
There are a great many ways in which we can add complements to a verb, and
the overall patterns are made even more complex when we consider adjuncts.
The basic shapes are: VERB - NP - NP - PP, . . . , AdvP, . . . , . A small sample
of the possibilities is shown in (31). Notice that these are all simple clauses -
we’re ignoring complex sentences, with clausal complements to the verb.
Another popular way of analysing phrases (e.g. (Radford et al., 2009: Part 3))
takes a functional element as the head. For noun phrases this is the determiner
element D, for verb phrases it is I (essentially an auxiliary verb). For adjective
phrases this is presumably Deg. (It isn’t clear what the head of a prepositional
phrase would be, other than a preposition).
This mode of analysis has certain theoretical advantages but it isn’t very
helpful for describing more complex phrase types. Also, it isn’t obvious how to
use it to describe languages which lack the appropriate functional categories,
such as determiners or auxiliary verbs.
5.4 Coordination
We can coordinate any string of words or phrases (usually of the same cate-
gory).
(32) Conjoined NPs:: [Tom] and [Jerry], [the cat] and [the mouse]
Conjoined APs:: [old] and [young], [too small] and [very ex-
pensive]
a Noun Phrase: a man for all seasons, a cup of tea, the girl with the flaxen hair,
or
a Verb Phrase: live in a burrow, read for two hours, leave before dawn
Exercise: Try out the tests to show that both the c-structures in (36) are
possible:
(36) a. S
NP VP
They
V NP PP
NP VP
They
V NP
observed
the N PP
Grammatical functions
6.1 Introduction
Sentences consist of more than subjects and predicates. The predicate itself
generally be split up into components with express grammatical functions.
These functions are often intimately connected with the expression meaning
but they are grammatical, syntactic notions and not semantic ones (this will
become clearer in the next sections).
The other main grammatical function within the predicate is that of object.
Some verbs express a relationship just between a subject and a predicate, e.g.
Dogs bark, but other verbs express a relationship between a subject and some
other entity expressed by a noun phrase. That noun phrase has the grammati-
cal function of object (OBJ). Here are some examples:
53
54 6. GRAMMATICAL FUNCTIONS
These three verbs have roughly the same meaning but different grammatical
function properties. Thus, devour is obligatorily transitive, while dine is obliga-
torily intransitive (though it can take a phrase such as on salmon). Eat is differ-
ent again, in that it is basically transitive (11a), but the object can be omitted,
(11b). It’s not obvious how we could derived these different properties from
meaning differences between the three verbs so they are generally treated as
purely grammatical differences in the types of objects which the verb takes (or
doesn’t take).
The subjects and objects a verb takes are an intrinsic part of the structure
of the verb and these are (usually) obligatory components of clauses contain-
ing those verbs (though there are various ways in which grammars can allow
subjects or objects to be omitted, or at least appear to be omitted).
The next grammatical function is almost always optional, in that it signals a
circumstance (time, place, manner, reason, etc.) which is not a central compo-
nent of the meaning of the individual verb in the way that subjects and objects
are. This function is that of adjunct (or adverbial). An adjunct can be thought
of as an optional modifier of the verb or clause.
These are the main types of case. In addition, there are rarer types such as (19,
20):
Subcategorizing verbs
We will see later that this list can be extended once we look at more complex
sentence types. In particular, there are a great many verbs that take entire
clauses as complements.
One special type of predicate is that formed using verbs such as be, seem,
become as in (22–24):
A verb such as these is traditionally called a copular verb (or copula). Cop-
ular verbs link the subject to a noun phrase, adjective phrase or prepositional
phrase which then conveys the main content of the predication. This is the
predicative grammatical function of noun, adjective or prepositional phrases.
We see how verbs take complements of various kinds. Other lexical categories
can take complements too. Prepositions almost always take a single comple-
ment: on the table, over the house. This is often called the object of the preposi-
tion, by analogy with the object of a verb. Some prepositions take a conjoined
complement, because of their meaning: between London and New York. Some
prepositions can be used intransitively, that is, without their complement. Such
1 You’ll sometimes see other terms used for subject/object complement.
58 6. GRAMMATICAL FUNCTIONS
prepositions are not really intransitive, rather, they are similar to verbs such as
eat which imply an object, but whose object can be left implicit.
Nouns and adjectives can also have complements. In (28) we see nouns with
a complement expressed as a preposition phrase.
Given these basic types of grammatical function we can ascribe to any given
sentence a grammatical function structure (GF-structure) together with labels
showing which phrases express which functions.
(30) a. The children put their toys in the box before suppertime
b. NP V NP PP PP
The children put their toys in the box before bedtime
SUBJ VERB OBJ1 OBL ADJ
In (30b) we’ve just recorded the grammatical functions associated with the
verb. We’ve seen, however, that we can say that prepositions (and other parts
of speech) can take complements. Thus, the two prepositions in, before have a
complementation structure. We can illustrate this in (31).
6.3. MODIFICATION 59
(31) NP V NP PP PP
P NP P NP
The children put their toys in the box before bedtime
OBJ-in OBJ-before
OBJ-of
COMP-proud
6.3 Modification
The final GF to discuss is one which relates to the internal structure of separate
phrases (noun phrase, adjective/adverb phrase, prepositional phrase) as well
as the whole clause. All the main types of word can be modified in various
ways, but we can also modify whole phrases and clauses.
Adverbials of various sorts answer questions such as when? where? how? why?
to what extent?
Many adverbs are related to adjectives. An adverb such as quickly is essen-
tially a form of the adjective quick used to modify a non-noun.
VP-oriented adverbials
(33) Manner:
a. Tom ran quickly
b. Harriet drew the picture carefully
c. Dick gave Fido a bone unwillingly
(34) a. Tom ran at great speed
b. Harriet drew the picture with care
c. Dick gave Fido a bone without any enthusiasm
(35) Place:
Harriet read the book in the garden
(36) Time:
a. They left late/at 10.00 pm/shortly afterwards/last week
b. Do you come here often
c. They visit her every day
d. She arrived four days ago
(37) Others:
a. They went for a walk despite the rain
b. They went home on account of the rain
Clause-oriented adverbials
Many adverbials don’t modify the way, time, place etc. an event occurs but
modify the meaning of the whole clause, often modifying its role or force in
the discourse (examples from CGEL: 576):
Cf.
Many of these adverbs are polysemous between VP- and clause-oriented uses:
We can also have complex modifiers of the form slightly too hot, a little hotter,
...
In more complex cases we can even find modifiers of modifers:
MOD
Notice that the comparative and superlative words more, most are effectively
adjective/adverb modifiers.
Modification of prepositions is possible, usually when the preposition has a
spatial meaning:
6.4 Agreement
The word police looks like a singular noun but it always takes plural agree-
ments.
Sentences which introduce the existence of an indefinite noun phrase and
which begin with there pose interesting problems of agreement. The noun
phrase after the verb has most of the properties of the subject:
However, we’ll see later that in many ways it’s better to treat there as the subject.
When the NP is plural we get two possibilities:
The version with ‘wrong’ singular agreement is more informal, but increas-
ingly accepted in ordinary speech.
(51) The big fat cat sat on the cosy mat behind the sofa
In order to figure out the c-structure it is helpful to know the overall gf-
structure, but in order to figure out the GF-structure we sometimes need to
know the c-structure. Parsing therefore in practice is a joint operation. How-
ever, the first step is always the same: We begin by labelling the categories of
all the individual words:
The verb is the head of the predicate (VP). The subject is generally the NP
immediately preceding the V (the big fat cat). This gives us (53):
Within the predicate verb phrase the verb sat is followed by two preposi-
tional phrases. We can interpret the second of these, behind the sofa, in one of
two ways. On one reading the PP modifies the verb and tells us where the
cat is sitting (54). Thus, together with on the mat, it serves as an OBLIQUE
64 6. GRAMMATICAL FUNCTIONS
(notice that we can also say The cat sat behind the sofa). On the other reading,
the PP behind the sofa tells us where the mat is and hence must be taken as a
postmodifier to the noun mat, and hence must be a constituent of the noun
phrase headed by mat. That means that there is just a single PP OBL function
(55).
(54) NP V PP PP
The big fat cat sat on the cosy mat behind the sofa
SUBJ VB OBL OBL
(55) NP V PP
The big fat cat sat on the cosy mat behind the sofa
SUBJ VB OBL
The adjective phrases big and fat (each consisting of one word!) modify cat.
On the reading illustrated in (55) the PP behind the sofa also has the MODIFIER
function. That reading is illustrated in (56).
V P Det A N P Det N
sat on the cosy mat behind the sofa
MOD | MOD |
VB | OBL |
| PREDICATE |
Analyses such as these represent what I shall call the basic parsing of a
simple sentence. We can enrich such a parsing in various ways. In many
analyses we would wish to provide a more detailed description of the phrase
structure (constituent structure or c-structure) of the clause.
Finally, for completeness, we can add an indication of semantic roles for
each of the grammatical functions (SUBJ, OBL) in the sentence. In simplified
form this can be seen in (57):
(57) NP V PP PP
The big fat cat sat on the cosy mat behind the sofa
SUBJ VB OBL OBL
Theme Location Location
Chapter 7
In school grammars (in the days when grammar was taught at school) the
grammatical functions used to be defined in semantic terms, roughly:
There are various reasons why this won’t do and in fact we will see that we
need a further level of analysis, semantic role structure or sr-structure, and
that this level is distinct from both c-structure (that is, phrase structure) and
gf-structure.
Consider the following examples:
65
66 7. ARGUMENT STRUCTURE ALTERNATIONS
In (2) the SUBJ function is realized by Tom (2a–2d), then the crystal, trees, this
key in (2e–2g) and in (2h) the rat. In examples (2a, 2e, 2g) we can imagine a
paraphrase which would make the subject the ‘doer’ of an action.
However, for the other examples such a paraphrase is odd or utterly bizzare.
The construction with ‘do’ implies that the subject is some sort of (conscious)
agent in bringing about the action, or an instrument that can be used by an
agent, but this is not compatible with the meanings of verbs such as see, fear
or receive. (We’ll discuss the interesting examples in (4e) in the next section).
This shows that the semantic role of the subject can be quite varied and is not
limited to that which we can call Agent. Indeed, in (2h) the Agent is expressed
but with an ADJUNCT function (see below).
You will often see labels attached to the semantic roles expressed by the
subjects in (2a–2h). The conventional names are:
Several of these roles are more common with OBJ, OBL or ADJUNCT func-
tions. Some examples:
(12) Experiencer:
Rats frighten Tom
7.1. SEMANTIC ROLE STRUCTURE 67
(13) Recipient/Benefactive:
a. Tom baked Harriet a cake
b. Tom baked a cake for Harriet
(14) (Spatial) Goal:
Tom went to London
(15) Source:
a. The light comes from the crystal
b. Tom arrived from London
(16) Location:
The house is in the forest
(17) Instrument: We opened your door with my key
There are further cases where it is not at all clear how to label the semantic
role of a noun phrase. These occur particularly often with predicative NPs:
Given that the garden probably bears the role location, what are the roles of
Tom and a linguist in (18)? A general role for such cases is that of theme.
(People often use the term ‘theme’ as a synonym for ‘patient’ though it’s more
useful to separate them.) The theme role can then be employed to describe the
semantics of such noun phrases as rats in (31), (the) light in (2e/9) or the house
in (2f/10, 16).
These semantic roles can be broadly divided into three groups, those which
are broadly speaking ‘active’ or ‘agentive’ and those which are broadly speak-
ing ‘passive’ or ‘patientive’ and those which are neutral. Where a transitive
verb has two arguments with two distinct roles the one with the more ‘agent-
like’ meaning will tend to be the subject and the one with the more ‘patient-
like’ meaning will tend to be the object. You will sometimes see these described
as the ‘Proto-Agent’ role (P-A) and the ‘Proto-Patient’(P-P) role. I shall use
other more-or-less synonymous terms: actor (A) and undergoer (U). A more
sophisticated version of school grammar definitions of subject and object will
then be that the subject has the Proto-Agent role and the object has the Proto-
Patient role. (Interestingly, even this rather vague characterization encounters
difficulties with more or less synonymous pairs like (2h) and (12)).
The division into actor/undergoer vs. more fine grained distinctions (agent,
theme, experiencer etc.) allows us to capture the fact that a NP can sometimes
appear to have two roles simultaneously. For instance, in (19) the subject NP
Harriet is the source, but she’s also acting voluntarily and intentionally as a
agent.
We can capture this by saying that Harriet is the actor and secondarily also the
source. This explains why (19) is synonymous both with (20a) and (20b).
The dual characterisation of Harriet as agent and source also explains why (21)
is a possible paraphrase of (19).
7.2 Passive
In the previous section I said that the SUBJ function is associated with Proto-
Agent roles while the OBJ function is associated with Proto-Patient roles. But
in that case, how come we have a patient SUBJ, the rat, in (2h) when Tom
realizes the agent role?
There are two points here. First, example (2h) has the following gf- and
semantic role structure:
There is no OBJ function in (22), showing that the verb is (in some sense)
intransitive.
The other point is that was killed is not the basic form of the verb kill. Rather
it is in a derived form, the passive (or more accurately, the passive voice). The
passive voice is to be contrasted with the normal form of the verb, the active
voice. We can picture the active∼passive alternation as a process in which
the OBJ of the active clause is promoted to the position of SUBJ in the passive
clause. The old SUBJ is then demoted to the position of an optional ADJUNCT.
Notice that we don’t need to mention the agent Tom in (2h): that constituent is
optional.
Thus, to understand examples such as (2h) we need to start with the basic
active form of the verb, illustrated in (2). That sentence is a prime example of
the generalization about Proto-roles and gf-structure.
The passive is constructed out of a special morphological form of the verb,
the passive participle, introduced in Chapter 2. For regular verbs this ends in
-ed and is identical to the past tense. For other verbs there may be a special
7.3. MIDDLE 69
form (e.g. write ∼ written, sing ∼ sung). In addition, the construction requires
an auxiliary verb ‘be’ (auxiliaries are discussed in more detail in Chapter 9).
The passive is essentially a device for altering the argument structure of the
verb without altering the meaning. Thus, the situations described by (2, 2h)
are essentially identical. The main difference is in Topic-Comment or Informa-
tion Structure: (2) tells us something about what Tom did, while (2h) tells us
something about what happened to the rat. In addition, (23), which is essen-
tially a variant of (2h), allows us to be non-committal about the identity of the
agent.
There is another form of the passive, which uses a different auxiliary verb,
get (this construction is often called the ‘get-passive’):
This form of the passive has very subtle modal nuances distinguishing it from
the ordinary ‘be-passive’. One of the nuances which is sometimes discernible
is that the ‘get-passive’ tends to ascribe some degree of volitionality, respon-
sibility or even agency to the subject. Thus, one way of understanding (24)
(though not the only one) is that the rat (deliberately?) behaved recklessly and
brought about its own end, as is unambiguously the case in the variant (25):
These types of example illustrate how subtle is the relationship between gram-
matical structure and semantic interpretation, providing yet more justification
for setting up a separate level of semantic structure to describe these subtleties.
7.3 Middle
These examples are all intransitive, yet they are based on what seem to be
transitive verbs:
This alternation is reminiscent of the Passive in that the SUBJ of (24) is demoted
and the OBJ is promoted to SUBJ. However, this construction doesn’t require
a special participle form and it doesn’t require an auxiliary. Moreover, it isn’t
possible to express the agent.
This construction is known as the Middle (Voice). Its function is to take a verb
which normally expresses an event (of reading, steering, assembling) and to
use it to express a property of the derived subject along the lines ‘one can
VERB NP (easily, with difficulty,. . . )’.
Both the Passive and Middle have the effect of creating an intransitive verb
out of a transitive one. It may seem as though this is true of the alternation
illustrated in (11a, 11b) in Chapter 6 and repeated here as (29):
It can be used when no reading matter has been mentioned in the earlier
context in which case it will mean ‘some book or other’ or ‘the newspaper’
(Harriet sat in her office all afternoon and read). Alternatively, we can use
Object Omission verbs when the object can be recovered from the context, but
even here the object has to be a canonical object for that verb:
7.5. DOUBLE OBJECT CONSTRUCTIONS 71
(31) a. The baby grabbed the spoonful of stewed apple and ate
b. ?? The baby picked up a handful of earthworms and ate
The precise semantic conditions on object omissibility aren’t very well under-
stood, nor is it clear why some verbs allow and others don’t.
As you can easily verify, the alternation consists in taking a verb which has one
OBJ and one OBL and turning the OBL into what appears to be another OBJ.
We thus end up with the double object verbs mentioned earlier this chapter.
This alternation is therefore called the Double Object alternation (though for
historical reasons it is also often referred to as the Dative Shift alternation).
In fact, the label ‘Double Object’ is misleading, because in most respects the
second ‘object’ doesn’t behave like an object at all.
There are other ways in which a verb can appear to have two objects. In (35)
we see a construction with a small number of verbs of judgement:
These examples differ very subtly in meaning. In (37b, 38b) there’s a strong
implication that the wagon or the wall were completely filled or painted over,
while in (37a, 38a) there is no such assumption.
Not all verbs with meanings similar to spray, load permit this alternation:
The reasons for these differences are rather complex and require a detailed
analysis of the semantics of these verbs which would take us too far afield.
7.7 Conclusions
Complements to P
The basic type consists of just preposition + noun phrase, though the preposi-
tion can often be modified by a word such as right, just, . . . (see Chapter 6):
73
74 8. PREPOSITIONS AND PARTICLES
These words are all temporal prepositions that can also take clauses as com-
plements (cf. after dinner, since Tuesday, until noon, before breakfast). In
some books these are analysed as complementizers (like that), though CGEL
has good reasons for rejecting that analysis.
This means that the term ‘preposition’ is wider in its scope than in some
descriptions. In fact, some prepositions don’t even take NP complements (usu-
ally):
The word although takes AdjP, PP and clausal complements, but not nor-
mally NP complements. When it does take NP the NP is interpreted as part of
a predicate (5), as it is in (4).
Intransitive prepositions
“Compound prepositions”
Prepositions are generally single words but there are many expressions that
look like prepositions that are multiword expressions, out of, away from, in
front of. Other expressions look like compound/complex/composite preposi-
tions of this sort but they are probably better treated as types of noun phrase
(CGEL:617-623), what CGEL calls prepositional idioms.
(8) at variance with, by means of, for lack of, in front of, on top of, under
the auspices of, with a view to,
There are two types of prepositional idiom. On type retains some of the nom-
inal properties of the noun head. For instance, the noun means in by means of
can be modified just like another noun, as can effect in with effect from:
However, it should be noted that the possibilities are not very broad.
The second type is what CGEL refers to as the ‘fossilized’ prepositional
idiom (p. 619), illustrated by in front of :
(10) in front of
*in which front? (in front of what?)
*in the immediate front of the house (immediately in front of the house/to
the immediate rear of the house)
the house it’s clear that of forms a phrase with house, not with in front. We can
see this when we use the prepositional complex intransitively (ie. without the
object house): She stood in front of the house ∼ She stood in front (*She stood in front
of ).
There are several properties of nouns that can be used to test
whether the noun component of a fossilized prepositional ex-
pression is still a noun. These include
i. singular/plural number
English is notorious for its phrasal verbs, in which a single verbal lexeme is
expressed by two separate words, one of which is often homophonous with a
preposition. Sometimes the overall meaning is a compositional function of the
meaning of the verb and particle (11), but more often than not the combination
is partially (12) or completely (13) idiomatic.
The particle has the unusual property that it can intervene between the verb
and its direct object, as seen in (11–13). Compare (14–16):
Many of the particles are homophonous with prepositions and may even
retain prepositional meaning. Which prepositions can serve as particles isn’t
easy to predict. For instance, we don’t see:
Other particles are related to other parts of speech. Some seem to be adverbs
and may retain the meaning of those adverbs.
There is one further type of particle which differs from the others in that it
can’t appear after the direct object, but instead appears in the position of an
ordinary preposition. An example is to look áafter (NP). Notice that in this verb,
it’s the preposition/particle that is usually stressed, not the verb:
(20) V + PP combinations
watch TV in the living room
walk in the park
Some verbs have both a particle and a selected specific preposition: put up with
Apply the tests below to demonstrate that up is a particle and
with is a selected preposition.
(23) V+PP
In the living room we watched TV
This is also (usually) true of prepositional verbs but it’s not found with particle
verbs:
8.2. PREPOSITIONAL VERBS 79
2. Pied piping
This refers to the property that a preposition can be associated with a wh-word
at the beginning of a clause rather than being left ‘stranded’ at the end of the
clause:1
3. Wh-fronting
A related construction is that in which we have a wh-phrase at the beginning
of the clause associated with a preposition. A true preposition allows this (at
least in the written or more formal language).
(27) V+PP
In which room did you watch TV?
b. Particle verb
*Up which answer did they look?
*After which children were they looking?
4. XP interpolation
A prepositional phrase used as an adjunct can be easily separated from a verb’s
direct object.
(29) V+PP
We watched TV quietly in the living room
Genuine particle verbs do not allow interpolation, though look after does allow
it:
(31) V+PP
*The living room was watched TV in
*The park was played in by the children
(Some) Prepositional verbs and particle verbs (usually) do allow the passive.
However, there are various factors that make the passive difficult, so not all
verbs work well in this construction:
Results:
V+PP PrepV PartV
PP fronting Y X X
Pied-piping Y Y X
wh-fronting Y Y X
XP in terpolation Y Y X
Passive N (Y) Y
Sometimes such verbs may undergo alternations similar to the spray-load type:
9.1 Overview
We begin with a reminder of the main role played by auxiliary verbs (from
Chapter 2).
Auxiliary verbs are used for signalling grammatical categories of mood, aspect,
and voice, as well as negation.
Aspect refers to the progressive and perfect constructions is writing, has written
and so on. These are discussed in more detail in Chapter 10.
Auxiliary verbs differ from lexical verbs in a number of ways, specifically their
syntax (cf NICE properties below).
9.2.1 Passive
[For more on the syntax of passives see Chapter ?? and Chapter 10]
Formation: BE + V-ed
83
84 9. AUXILIARY VERBS: PROPERTIES AND TYPES
We’ve seen that the passive construction is also possible with get as the auxil-
iary, with a somewhat different modal meaning.
9.2.2 Aspect
There are two marked aspects, progressive and perfect. Both can occur in past
or non-past tenses .
Progressive. Denotes on-going activity (as opposed to completed or timeless
activity).
Formation: BE + V-ing
Perfect. This aspect denotes event which occurred prior to time of utterance
but which has relevance for present. We’ll look at this in more detail in the
next chapter.
Formation: HAVE + V-ed
9.2.3 Mood
may/might
shall/should
(5) can/could leave/have left/be leaving
will/would
must
9.2.4 Infinitive
marker is a separate word and can be separated from the base form of the verb
by, for instance, adverbs.
We can combine auxiliaries (including the infinitival to) in various ways, but
with restrictions. For instance, the two aspects combine, but only in the order
progressive + perfect:
The perfect auxiliary have selects the -en participle form of the next verb,
whether lexical or auxiliary:
(8) HAVE+-en
BE+-ing
SING
has been singing
We can combined aspect and passive voice. Either of the two aspectual
auxiliaries can precede the passive auxiliary (though the combination of both
aspects and passive sounds rather strained).
(11) a. * Harriet has willed sing (vs. Harriet will have sung)
b. * Tom is maying read (vs. Tom may be reading)
Auxiliaries are the elements which realize the following meanings or gram-
matical relationships:
• Negation
• Interrogatives (Inversion)
• Code
• Emphasis
Lexical verbs do not exhibit the NICE properties. Despite the acronym it’s
easier to understand the system if we start with Emphasis.
Since we’re only talking about finite clauses here, the NICE properties don’t
apply to the infinitival auxiliary element to.
9.4.1 Emphasis
We can accent the polarity of a statement without emphasising any other word
or phrase. This is most readily seen when we answer a question with ‘yes/no’:
It’s helpful to think of the emphasised form of a simple clause as the basic
form, from which all other construction types can be derived.
Polarity is a property with two basic values: affirmative (‘yes’) and negative
(‘no’).
The ‘free form’ stressable not occurs after leftmost auxiliary, or it occurs as
an inflectional affix on the leftmost auxiliary in the clause. The latter construc-
tion is the only one found in spoken English unless we want to emphasis the
negation.
In combinations:
9.4.3 Inversion
English has a construction in which the subject of the finite clause swaps places
with the leftmost auxiliary verb. This is mainly (though not exclusively) found
in the formation of questions:
As usual, in the absence of any other auxiliary we use the dummy aux.:
Again, lexical verbs don’t show this behaviour (hence, the need for the
dummy aux.):
(29) Tom will have been reading a book and so will have been reading Har-
riet
(30) Will Harriet have been reading that book?
Yes, she will (have (been (reading that book)))
(32) They’re leaving early, are they (we’ll see about that!)
Cf:
(33) They’re leaving early, aren’t they (I think those were the arrangements)
As usual, a clause without an auxiliary in it has to call upon the services of the
dummy auxiliary:
Again, this behaviour isn’t seen with lexical verbs, especially the tag ques-
tion construction:
Auxiliary verbs differ from ‘true’ lexical verbs in a number of important re-
spects. In each case we can link the differences to the fact that an auxiliary
verb is a verb which has changed its function over time from being a meaning-
ful lexical item to expressing a grammatical property or category of some sort.
This historical process of change is known as grammaticalization.
The first and most obvious distinction between auxiliaries and lexical verbs
is that auxiliaries don’t have an obvious meaning. This is particularly clear in
the case of the dummy auxiliary do. The aspectual/voice auxiliaries can only
express these categories in conjunction with a special form of the lexical verb
(a participle form).
The modal auxiliaries retain a certain semantic force, expressing notions of
obligation, necessity, desire, intention, and so on. In many cases there exists a
lexical verb expressing the same kind of meaning:
and so on. The reason we regard the modal words will, must, can, should,
may etc as special is because of their form and their syntax.
The modal auxiliaries lack forms which other verbs (including other aux-
iliaries) have, notably participle forms. There is no -ing/en participle and no
(genuine) past tense for the modals: *canning, *musted, etc. The 3sg form
is unique in that it doesn’t take the -s suffix: Harriet can/*cans speak Russian.
We’ve seen above that several modals, along with some other auxiliaries, have
special negative forms can’t, won’t, shan’t etc.
Most significantly, the modals share the same unusual syntax as the other
auxiliaries, notably the NICE properties. They also enter into the special com-
binations with other auxiliaries that we’ve seen above.
However, grammaticalization isn’t always an all-or-nothing phenomenon
and in fact is an on-going process. For this reason we will sometimes find
words which have some of the properties of the ‘standard’ auxiliaries but not
all. First, this relates to the infinitival marker to. This is historically derived
from the preposition to rather than from a verb like the other auxiliaries and
for that reason it doesn’t have special negative forms (not to, to not but there’s
no form ton’t). In addition, the inversion property isn’t found because the
function subserved by inversion doesn’t apply to non-finite clauses. Similarly,
9.6. DEFECTIVE FORMS AND UNUSUAL FORMS: MODAL AUXILIARIES 91
(40) a. * Tom is musting open the door with his credit card
cf Tom is having to open the door with his credit card
b. * The door is musted open
cf The door is needed open; the door must be opened
c. * Tom has musted open the door with his credit card
cf Tom has had to open the door with his credit card
As we’ve seen, modals lack a special 3sg form, the default form being used
instead.
Expressions like ‘to be able to, to have to, to be obliged to, to be about to are often
used to fill in the lacunae in the analytic paradigms with auxiliaries.
The main, lexical verb of a clause expresses the main idea of that clause (usu-
ally), and the auxiliary verbs express a variety of grammatical or functional
meanings and relations (i.e. the TAMV meanings/relations). The lexical verb
and the auxiliaries therefore enter into a single functional complex and effec-
tively behave as a single element at clause structure, like the Subject NP or an
Adjunct phrase. We can therefore refer to the lexical verb and the auxiliary
satellites around it as a single (functional) element, the Verb Group. A simple
example of this is shown in (43).
S VbGp A
However, at phrase structure this Verb Group can be split by adverbs and
even adjunct phrases of various kinds. These can appear at the very beginning
of the Verb Group, or possibly after any of the auxiliaries that make up the
Verb Group (depending on subtle factors to do with meaning and emphasis).
For instance, the adverb apparently can appear immediately before or after the
Verb Group, or after on of the auxiliaries (as well as at the beginning or end of
the clause).
Notice that in (44f), where the adverb is in absolute clause-final position, the
adverb is separated from the rest of the clause intonationally (indicated in
writing by the comma).
Other adverbs are more selective. Where an adverb refers to the overall truth
or likelihood of a clause it tends to have greatest freedom of placement. Thus,
the adverb apparently in (44) is equivalent to something like It is apparently
the case that . . . . Where an adverb modifies the lexical verb specifically, as with
9.7. AUXILIARIES AND THE VERB GROUP 93
Notice that in (45f) the adverb isn’t separated intonationally in the way that
apparently is in (44f).
Sometimes the semantic contribution of an adverb will change depending
on the position in which it’s placed, especially in negative clauses.
Because it comes immediately after the lexical verb it’s difficult to interpret
obviously in (46b) as modifying the meaning of the whole clause rather than
having some kind of manner meaning (The cat was sitting on the mat in an
obvious manner).
In (47a) we’re saying that it’s obvious that the event hasn’t taken place (It’s
obvious that the cat hasn’t been on the mat), whereas in (47b) we’re denying that
it’s obvious that it has taken place (It’s not obvious that the cat has been on the
mat).
The adverbials that intervene inside the Verb Group have their own clause
level functions as adjuncts, and so in a complete analysis it would be neces-
sary to indicate their presence as in (48).
S VbGp A VbGp A
94 9. AUXILIARY VERBS: PROPERTIES AND TYPES
Chapter 10
10.1 Introduction
10.2 Passive
The subject of the active form can be expressed optionally by means of a by-
phrase (‘long-passive’) or the by-phrase can be omitted (‘short’ passive’).
The short form allows us to defocus the active subject.
(1) (Someone or something) killed the rat ⇒ The rat was killed
In this way we don’t have to mention who, if anyone, was responsible for the
killing. (It might be a person, or a cat, or it might be a natural event like
lightening or a flood). The passive form also tends to put accent/emphasis
on the predicate, emphasing the fact that killing took place, and putting less
emphasis on the fact that it was done to the rat.
The construction BE + passive participle is very close to the predicate adjec-
tive construction.
Verbs typically denote events, things that happen at some point in time. The
passive form of a verb can denote an event. For instance, we would normally
interpret (2a) to mean that at some point in time a killing event took place.
This is particularly true if we mention the agency by which the rat died, as
95
96 10. TENSE, ASPECT, MOOD, VOICE
in The rat was killed by Tom/the cat/lightening. However, the short passive can
often denote a state rather than an event. For instance, a sentence such as (3)
is ambiguous between two slightly different readings, (4a) and (4b).
This could mean either that someone or something had earlier broken the vase
(the eventive reading) or it could be telling us what condition the vase was in
(the stative reading). Adjectives denote states, for instance, the state of being
tall (a permanent state) or the state of being tired (a temporary state).
This process can go further, so that some adjectives have the form of a pas-
sive participle but are usually used as adjectives:
The progressive aspect isn’t found with verbs which denote States (as opposed
to ‘dynamic’ events which evolve through time such as Activities, Processes
and so on)
10.3. VERBAL ASPECT 97
Perfect. The perfect aspect denotes event which occurred prior to time of
utterance but which has relevance for present (current relevance).
Formation: HAVE + V-ed
E R/S (=present)
(12) (When Harriet got home) Tom had washed the dishes:
E R S
Tom washed the dishes at E and they were clean at the Reference time, R
(Harriet’s arrival).
The meaning/use of the perfect is rather subtle and context-dependent. In
some cases there’s virtually no difference between simple past and present
perfect:
98 10. TENSE, ASPECT, MOOD, VOICE
(14) a. I washed the dishes five minutes ago (so we can start cooking at
once)
b. ?? I’ve washed the dishes five minutes ago . . .
On the other hand in British English it’s much less natural to use already/yet
with a simple past:
i.e.
Perf Prog Verb
HAVE -ed
BE -ing
SING
Combining tenses/aspects with passive:
10.4 Tense
There are two tenses, preterite (past) and present (non-preterite, non-past).
There is no future tense.
It is sometimes difficult to distinguish tense from aspect.
The present tense form has a variety of meanings/functions. First, we
should distinguish between occurrences/happenings (sometimes called ‘dy-
namic events’) and states. For instance, in (24) the verb is in the present tense
because it denotes an event that is happening right now.
This reference to the time of utterance is actually not the usual meaning of
the present tense, however (which distinguishes English from a great many
languages). The present tense frequently refers to a state.
Other examples of the present tense being used to denote dynamic (non-state)
events are seen in (28).
Preterite forms generally denote an event which occurred before the time of
utterance. There is no other implication, usually, though with verbs denoting
a completed action the preterite/past tense usually implies that the action is
over and done with: Harriet ate the apple.
English (sometimes) exhibits the phenomenon of sequence of tenses or ‘back-
shifting’. In this construction we have a main clause verb with a meaning such
as ‘say’, taking a subordinate clause. When the main clause verb is in the past
tense, the subordinate clause verb is put in the past too, whatever the meaning:
Note in particular that in (33c) ‘tomorrow’ can refer to the day after the day
when the speaker utters the sentence, i.e. it can have genuinely future reference
despite the fact that ‘left’ is in the past tense (Cf. Oh, she’s gone already! But I
thought she was leaving tomorrow).
Strictly speaking the nominalization of a verb ought to come under the heading
of ‘Derivation’, but in English (as in many languages) productive nominaliza-
tion is actually closer to inflection. It makes sense, therefore, to think of the
nominalized form of a verb, just as we can think of the passive form or the
preterite form.
10.5. NOMINALIZED FORMS OF VERBS 101
Verbal properties
When the OBJ is realized in the manner of the OBJ of a verb, the Gerund/VN
is modified by adverbs (like a verb) not by adjectives.
The perfect aspect has a verbal noun form derived from the -ing form of the
auxiliary have.
In (45a, 46a) we see the basic meanings of can = is (physically) able to, must = is
obliged (morally or by circumstances) to. In (45b, 46b) however we see meanings
which involve a higher degree of abstraction, can = permission (is ‘morally’
able to) and must = speaker inference, the speaker infers that Harriet has left
(say, from the fact that her coat isn’t on the coathanger).
We can easily construct genuinely ambiguous examples:
A very interesting aspect of the use of modals relates to the interaction between
modal meanings and negation. This is not as transparent as it might be (and
104 10. TENSE, ASPECT, MOOD, VOICE
root epistemic
can ability possibility
permission
may permission possibility
will willingness prediction
intention
insistence
shall intention prediction
insistence
(obligation)
could past ability possibility (hypothetical)
past permission
might permission possibility
would (past) willingness past prediction
(past) insistence probability
volition with preference characteristic past activity
should obligation logical necessity
real conditions
must obligation logical necessity
<ought to> obligation logical necessity
<have to> obligation
usage in English differs from that in some other languages, for instance). As
a preliminary, CGEL:175 makes a useful distinction in the types of meanings
expressed by modals, in terms of strength, kind, degree.
• Strong - necessity, obligation
• Weak - possibility
We can then relate strength and negation (CGEL:176).
The relation between possibility and necessity
Notice that can’t doesn’t mean CAN [NOT [LEAVE]] (despite the order of can
and n’t). One way to express the logical order can [NOT [. . . ]] is to use a
non-modal construction.
However, we can separate not from can and pronounce them as two distinct
words, preferably with stress on not (i.e. distinct from the stress pattern on
can’t, cannot).
Modal auxiliaries show only a rather residual tense opposition. We can think
of could as the past tense/preterite of can in some cases.
Example (61b) is simply the future prediction expressed in (60a) projected back
into the past. But (61a) means something rather different than (60a) in the past;
it has a conditional meaning completely lacking in (61b). Other modals lack
any obvious past tense (preterite) form.
However, could/would/should do behave like past tense forms in one con-
struction, backshifting. How do auxiliaries behave? In (63) we see the progres-
sive/perfect auxiliaries and also can/will/shall have past tense forms when the
main verb is past tense, suggesting that could/would/should in those examples
are formally speaking past tenses of can/will/shall in (62).
We can express modal meanings in a variety of ways that don’t involve the
standard modal auxiliaries.
108 10. TENSE, ASPECT, MOOD, VOICE
Pedagogic grammars (and some academic articles) sometimes refer to the ‘fu-
ture tense’ in English. But does English have a future tense?
Clearly, there’s no morphological future tense form.
In terms of meaning the will/shall auxiliaries for the most part express the
same property that is expressed by a future tense in other languages:
Given that auxiliaries can be used to (partially) express the two aspects, why
shouldn’t we say that will/shall express a future tense (even though we nor-
mally reserve the term ‘tense’ for a category that is expressed morphologi-
cally)? This is certainly a possible analysis, in which case we would be saying
that English has a periphrastic future tense.
However, there really isn’t any very good reason for treating the will/shall
constructions as a tense as such (cf. CGEL:208–210). Instead, the future time
reference can be thought of as just a very general and non-concrete type of
modal meaning. (It’s interesting to note that historically the auxiliary devel-
oped from a verb meaning ‘want’).
There are a good many other ways of referring to future time. We can use
the present form of verb (in simple or progressive aspect):
Each of these has slightly different modal and temporal nuances. For instance,
the simple present tends to be interpreted as expressing some prearranged
scheduled event (The bus leaves at 9.00 tomorrow morning). On the other hand,
some of the constructions, notably about to, tend to imply immediate future
compared with the other constructions (and, indeed, about to tends not to be
used with an explicit temporal adverbial because it’s understood that the ac-
tion will take place soon).
111
Chapter 11
Subordinate clauses I
• Verbless clauses 62
113
114 11. SUBORDINATE CLAUSES I
(6) Go away!
(
to go away
a. Harriet told Dick
that he should/must/ought to go away
b. Harriet asked Dick to leave
The finite clause in (7a) is introduced by the function word that, which is often
called a complementizer (‘COMP’, ‘C’). We will see several different sorts of
complementizer. In some types of subordinate clause a complementizer of
some kind is obligatory, while in others, for instance, in the non-finite clauses
seen in (1b, c), a complementizer is impossible (though we’ll see a variant on
(7b) in which another type of complementizer is obligatory).
The traditional term for ‘complementizer’ is ‘subordinating conjunction’
though this is not a very accurate term, since these complementizers tend to
have different properties from other conjunctions, i.e. coordinating conjunc-
tions, and because other elements can be complementizers without serving as
‘conjunctions’ in any obvious sense. I shall use the simpler term ‘subordinator’
for these elements.
The non-finite clauses typically lack an overt subject. ‘Overt’ means a word
or phrase that’s actually pronounced rather than ‘covert’, a phrase that the
linguist imagines is there. However, it’s usually obvious what the subject cor-
responds to. Traditionally, we refer to the ‘understood subject’ of the clause.
We’ll examine this notion later under the heading of ‘subject gap’ when we
come to look at the internal structure of subordinate clauses in more detail.
In (11) it’s clear that the word it is referential (it refers to this chapter) and not
an dummy/expletive. However, the boundary between the two constructions
can sometimes be difficult to discern.
Note that some predicates only permit the extraposed form, e.g. seem, be
likely.
)
It seems
(12) a. that the earth is flat
It is likely
(
seems
b. *That the earth is flat
is likely
In (3) we’ve seen a finite adjunct clause (adverbial clause of reason). Other
examples include:
These are traditional descriptive labels, though they aren’t always particu-
larly useful. Some types of adjunct clause don’t have widely accepted names
anyway. The conditional clause is particularly important in English so we’ll
discuss it separately.
There are also subordinators that can or must take infinitival or ing-clauses.
(17) They left early (in order) to beat the rush hour
(18) They stopped for ten minutes (in order) for the engine to cool down
(19) They beat the rush hour by leaving early
(20) They criticized us for leaving early
(21) Despite (our) leaving early we arrived late
<Some of these examples are from Quirk et al 1972; others from CGEL>
Real conditions
Non-past
(24) If you water this plant regularly, it produces beautiful red flowers
(25) a. If you water this plant regularly, it will produce beautiful red flow-
ers
b. If you have watered this plant regularly, it will produce beautiful
red flowers
Past
(29) a. If you had watered this plant regularly, it would produce beautiful
red flowers
b. If you had watered this plant regularly, it would have produced
beautiful red flowers
(30) I you watered this plant regularly, it would produce beautiful red flow-
ers
(31) I you were to water this plant regularly, it would produce beautiful red
flowers
118 11. SUBORDINATE CLAUSES I
(34) I don’t know whether he broke it or not, but I doubt it; if he had done
he would probably have told her about it.
Here, the whole point is that the speaker doesn’t know whether the possible
world they are postulating is unreal or is the actual world.
For learners of English conditionals often pose serious problems simply be-
cause of interference from grammars which express such constructions in dif-
ferent ways. Thus, in many languages there’s a conditional or subjunctive
mood and both clauses have to be in that mood. For this reason, even very
advance learners often say things like:
(35) *I you would water this plant regularly, it would produce beautiful red
flowers
(36) *I you would have watered this plant regularly, it would have produced
beautiful red flowers
Unless
(38) If you hadn’t watered this plant regularly it would have died ,
Unless you watered this plant regularly it would have died
11.7. TYPES OF CLAUSAL COMPLEMENTS SELECTED BY VERBS AND OTHER WORDS119
Most but not all examples of clausal complements so far have been comple-
ments to verbs such as say, claim, believe, expect, . . . . Such verbs are often used
for reported speech (indirect speech) as opposed to direct speech. Reported
speech refers to a clause type which reports what could have been said di-
rectly. Simple examples are:
There are very many verbs that take such sentential/clausal complements.
However, a number of nouns and adjectives also take finite and non-finite
clausal complements.
Nouns:
With verbs, nouns and adjectives we sometimes find that different types of
complement have different meanings.
In (1) we have the recollection of an abstract event - note that for (1) to be
true it doesn’t necessarily have to be the case that Tom remembers the actually
event of closing the door, cf:
cf
With the infinitive the construction means ‘didn’t forget to carry out intended
act’:
(49) Tom obviously remembered to close the door (though he has no recol-
lection of having done so)
Comparative words such as more and the comparative form of an adjective can
often take a subordinate clause complement introduced by the complementizer
than.
(55) Tom likes reading novels more than Dick likes reading plays
11.8. COMPARATIVE CLAUSES 121
(56) a. Tom likes reading novels more than Dick likes reading them
b. Tom likes reading novels more than Dick does
c. Tom likes reading novels more than Dick
Notice that we usually elide the repeated part of the VP in the first clause (as
in (56b, c)).
With adjectives the typical construction treats than as a preposition whose
complement is the thing contrasted.
However, than can often introduce a clause (usually elliptical, as in the exam-
ples above).
(58) The tree is taller than the church tower was (before it got demolished)
(59) The TV is wider than the cupboard is tall (slightly artificial example)
Subordinate clauses II
In (1) the two types of clause have slightly different meanings: (1a) means
essentially ‘didn’t forget’ while (1b) means ‘had a recollection of a past event’.
In other cases it’s difficult or impossible to discern any semantic difference
between the two types of clause, for instance, (2, 3).
Many verbs take only one of the two types of non-finite clause.
123
124 12. SUBORDINATE CLAUSES II
It’s often difficult to discern any good semantic reason for the complementa-
tion pattern shown by any particular verb. For instance, why do cease and
like take both types of non-finite complement while stop and enjoy take only
the ing-clause?
We will see that there are several types of infinitival complement, depending
on the main (matrix) verb. The principal patterns are those seen in (??).
Huddleston and Pullum (2002: Ch14, §2.1, 2005: Ch13, §4]) refer to the con-
structions with infinitival complements to verbs like try, begin, seem as cate-
native constructions. This term comes from the Latin word catena ‘chain’, and
alludes to the fact that we can have chains of infinitival complements. Actu-
ally, there’s nothing special about this since it’s just an instance of recursion.
Recursion refers to the process of embedding a construction inside a larger
construction, which in turn can be embedded inside a larger construction, and
so on, like a never-ending series of Russian dolls. What’s special about the
catenative construction is that the infinitival verb phrases can be joined to each
other without needing a subject of their own:
(It doesn’t matter that such sentences get completely uninterpretable – the
point is that they obey the rules of English grammar.)
Although these verbs are used mainly with a gap for a subject, it’s often
possible to express an overt subject as well. In case the verb is in the ing form
the subject is expressed directly. In case the verb is in the infinitive form the
subject is introduced by the subordinator for:
With ing-clauses the SUBJ function can also be expressed by a ‘Saxon genitive’:
We’ll return briefly to the case of overt subjects with infinitival clauses later.
For the present we’ll mainly be considering the situation when there is no overt
subject.
Make a list of verbs which take non-finite clausal complements.
Which take just the infinitival complement and which take just
the ing-complement? Which can take both types of comple-
ment? Of those which take both the infinitival and the ing-
complement, which show some kind of meaning difference be-
tween the two types of complement? (Be careful to consider
just examples in which the clause really is a complement to the
verb and not an adjunct. You can usually identify an adjunct
clause because it’s usually possible to place it at the beginning
of the sentence, and in the case of infinitivals, you can usually
replace to with in order to: Tom left to catch a train ∼ Tom left in
order to catch a train.)
As should be clear from the examples given so far the subordinate clause has a
missing subject. However, the identity of that subject is identical to the subject
of the main clause. We say that the subject of the main clause controls the
subject gap in the subordinate, complement clause.
We usually indicate control using subscripts.
I’ve indicated the gap by means of a line: . However, you’ll often see the
gap indicated in other ways. In certain respects the gap is sometimes thought
to be a little like a pronoun (cf Dick hoped [to meet Harriet]∼Dick hoped [that he
would meet Harriet]) and a conventional representation is PRO: Dick hoped [PRO
to meet Harriet].
work, even though they are not as numerous as the more standard types of
verb.
There are two broad classes, illustrated by (13), (14).
The difference between the two is essentially to do with the meanings of the
subjects. In (14) Harriet is (actively) making an attempt to bring about some
situation, and that situation is one in which she, Harriet, is calm. Another
way of thinking of this is that Harriet and the empty SUBJ function have two
distinct semantic roles, a ‘trier’ role (roughly an Agent) and a ‘being calm’ role
(roughly a Theme).
In fact, the SUBJ function of try is always associated with an agent, while
the missing subject of the complement clause can be associated with a distinc
semantic role, as seen in (16).
By contrast the verb seem imposes no semantic role on its subject at all. That
role is inherited from the subject gap in the complement clause.
Here, we can see that the semantic role of the subject of seem has to be what-
ever role is assigned to the subject of the subordinate clause by its verb. Cru-
cially, the verb try doesn’t permit a paraphrase along the lines of (21).
We often say that a verb such as seem has a non-thematic subject (from the
term ‘thematic role’, a synonym of ‘semantic role’).
Verbs such as try behave in a more or less normal fashion in that they have
their own subject to which they assign a specific semantic role. Verbs such as
seem are unusual – they effectively inherit their subject from the complement
clause. It is as though the subject starts out life as the subject of the comple-
ment and then ‘raises’ to the higher position of subject of seem. The ‘raising’
metaphor has a long history in modern linguistics and it’s a useful way of pic-
turing the behaviour of seem-type verbs. For this reason we often refer to such
verbs as raising verbs.1
Compare (23a–23c):
Sentences (23a, 23b) mean the same thing, and each entails (23c). The analysis
of (23a) presents no problem: Dick is the object of persuade and there is an
additional clausal complement whose subject happens to be identical to Dick.
It doesn’t have to be this way, since the subject of the complement clause could
be different.
complement clause is related to the subject of the main clause. We will see that there are other
verbs in which the subject of the complement clause is related to the object of the main clauses,
and those verbs are usually referred to as raising-to-object verbs.
128 12. SUBORDINATE CLAUSES II
Clearly, the complement clause has a missing subject. Moreover, that missing
subject in (23b) has to be identified with Dick, in that it’s Dick that is supposed
to arrive early, not anyone else. In particular (23b) doesn’t mean the same as
(26).
Using the notational conventions adopted above we can represent (23b) as (27):
In cases such as (23b/27) we can see, then, that it is the main clause OBJ NP
Dick that controls the empty subject in the complement clause.
We’ve effectively conflated two types of representation in cases such as (27),
the actual string of words and the labels for the various grammatical functions.
Obviously, it isn’t the case that the string of words in (23b) contains a gap (or
the word PRO or whatever) in a literal sense. We can separate these out,
however, by means of the notational conventions of our basic parsing:
There are two variants on the pattern shown in (23b/27). The first is exempli-
fied by the verb expect.
Clearly, (29, 30) mean the same and express a relationship of expectation be-
tween Harriet and the eventuality denoted by ‘Dick will be late’. In (29) this
leads to a simple syntactic structure in which that Dick would be late is a clausal
(sentential) complement to the verb.
Originally, this prompted some linguists to adopt the structure shown in
(31) for (30), in which Dick is the subject of the subordinate clause and expect
doesn’t take a direct object:
One reason for adopting such an analysis is the fact that we can’t say that (30)
entails (32) in the way that (23b) entails (23c):
12.1. SENTENTIAL COMPLEMENTS WITH MISSING SUBJECTS 129
If we were to analyse (30) as in (31) we wouldn’t have any empty subjects and
therefore there’d be no control.
Nonetheless, it turns out that in certain ways the NP Dick in (30) does, after
all behave in some ways like the object of expect. For instance, we can put
sentence (30) into the passive voice, in which case Dick behaves like the direct
object of expect and becomes the subject of the passive alternant.
Moreover, when a pronoun occupies the position of Dick in (30) that pronoun
is in the object form not the subject form.
The reason that Dick is a special kind of object is that Dick doesn’t have the
usual semantic properties you’d expect of an object. In particular, Dick isn’t the
semantic object of Harriet’s expectation in (30). In this respect it contrasts with
the object Dick in (23b), which is the semantic object of Harriet’s persuasion
(this is what the contrast between (23) and (32) shows). In other words, we
can say that in (23b) Dick is the object of persuade both semantically and
syntactically, while in (30) Dick is the object of expect only in the syntax, not
semantically. This mismatch between semantics and syntax is not especially
uncommon in English.
In (37) we see yet another variation on the theme of the missing subject.
The kind of analysis we want for (37) would seem to be that given in (40):
In the intransitive use of expect, the distinct semantic properties of the main
clause and subordinate clause subjects are if anything even more clear than in
the case of seem. Since expecting something isn’t an action as such, but rather
a mental state, let’s assume that the semantic role of its subject is experiencer.
We can now ask what semantic roles are assigned to the subject gap in the
complement clause to expect. It turns out that, as with try, the complement
clause subject can have any semantic role.
There are a number of common adjectives which behave rather like seem when
used as predicates. Consider the following examples.
( )
certain
(42) a. Harriet is to catch a fish/like the film/receive a present/. . .
likely
( )
anxious
b. Harriet is to catch a fish/like the film/receive a present/. . .
eager
There are two sorts of verbs which take non-finite complements with missing
subjects. The first sort, illustrated by try, assigns its own semantic role to
its subject, and this can be different from the semantic role of the missing
subject of the complement clause. The second sort, exemplified by seem, fails
to assign any semantic role at all to its subject, which inherits the semantic role
assigned by the complement clause verb. The verb expect behaves like try in
this regard.
The properties of verbs such as seem or adjectives such as certain, anxious
are not widespread in the English lexicon. Most verbs that take non-finite
complements belong to the try, expect class. Such verbs are also found widely
in other languages, so that it is often easy to translate literally from English
into such languages and vice versa. However, many languages lack seem-type
12.2. MISSING SUBJECTS IN ADJUNCT CLAUSES 131
probable
(43) *Harriet is possible to catch a fish/like the film/receive a present/. . .
definite
(45) [Despite i leaving in good time], wei arrived late for the concert
(46) Tomi went into his study [(in order) i to check his email]
(47) [By i watering this plant regularly] youi can get it to flower
(48) [For i stealing Harriet’s sandwich] Dicki got ten years hard labour
(51) Lurking motionless in the middle of the river we saw a large crocodile
(Presumably it’s the crocodile that’s lurking)
(52) Driving our four-by-four along the Limpopo river, we saw a large crocodile
132 12. SUBORDINATE CLAUSES II
Missing subjects appear in other contexts, in which they don’t have a controller
elsewhere in the sentence. Consider the examples in (53):
Here we have a missing subject in the clauses running/to run ten miles (who is
running?) but no controller. In cases such as this it seems that the missing
subject is always interpreted as ‘people in general’, i.e. as some unidentified or
generic human. An empty element with the interpretation of ‘generic person’
is often referred to as having the ‘arbitrary’ interpretation (meaning ‘arbitrary
person’) or ‘arb’ for short. This is then often used as a label or subscript for
the grammatical function which has this interpretation, as in (54):
Although missing subjects are by far the most common and important type
of systematically absent grammatical function/clause element, there are also
constructions in which it is the direct object which is missing. One class of
such cases has become famous in the linguistics literature. It’s a construction
which is of wider interest, however, because it’s characteristic of English, but
is frequently absent in other languages. It’s thus another special construction
which confronts any L2 learner of English.
12.4. CLAUSES WITH MISSING OBJECTS 133
In example (61), the predicate is (be) eager which means essentially the same as
wants:
The second thing to note is that the subject of please in (62) is effectively arbi-
trary, with the meaning of ‘people in general’. We can see this by considering
the near paraphrases in (66).
A further difference between (62, 61) is evident when we compare (64) with the
counterpart construction based on the predicate easy. Whereas it’s perfectly
possible for us to specify an object for please in (64) this is completely excluded
for (62).
This is not surprising: the object function of please in (62) has been taken over
by the subject Tom in (62). We can illustrate the grammatical relationships by
means of the diagrammatic representation in (68).
134 12. SUBORDINATE CLAUSES II
The difference between the minimal pair in (61, 62) is a locus classicus in
modern theoretical linguistics (the distinction was noted for the first time by
Chomsky (1965), though the analysis he gave there for it was rather different
from the analyses that would be proposed now, including the one above). The
construction is often known by the informal name of tough-construction, be-
cause the predicate tough exhibits the same structure: This problem is tough to
solve. It is rather restricted in scope because it is basically found with pred-
icates which mean ‘easy/difficult’ and a few others (such as pleasant, nice,
good and their antonyms: Her voice is nice/unpleasant to listen to).
The tough-construction is of great interest precisely because it shows how
systematic analysis brings out considerable differences in syntactic which are
not reflected at all in the surface arrangement of words. This shows that syntax
is much more complex than just putting words in the right order.
Implications
While it’s generally easy to find the translation equivalent of eager examples
it’s often impossible to translate ‘tough-constructions’ literally into other lan-
guages. Moreover, theconstruction is limited to a small number of predicates,
mainly expressing meanings of ease or difficulty in doing something.
Chapter 13
13.1 Introduction
In this chapter we look at two constructions which have very different func-
tions but which share certain features. Wh-questions obviously involve wh-
words, but they also involve a particular kind of gap, corresponding to the
wh-phrase itself. A topicalized sentence is one in which a phrase is displaced
from its normal place in the clause, leaving behind another type of gap. In
the next chapter we’ll see another construction which involves wh-words and
a gap, the relative clause.
13.2 Wh-questions
We’ve seen that a simple (yes/no) question requires inversion of subject and
first auxiliary. This is also a component of the syntax of special questions, i.e.
those formed using question words such as who, what, why, when, . . . , how.
In addition to polarity or ‘yes-no’ questions, we can question specific con-
stituents using special function words, most beginning wh-, the so-called ‘wh-
words’. Structurally speaking, the simplest type of wh-question is the echo-
question, illustrated in (1–3) (the punctuation indicates a high rising intonation
on the question word):
135
136 13. WH-CLAUSES AND TOPICALIZED CLAUSES
There are two ways in which the wh-word figures in a wh-phrase. In (7a, b)
we see which and whose used as modifiers of the head noun film, car (with what
being a stylistic variant of which). In example (7c) we see a wh-word which
serves as the complement of a preposition. In (7d, e) we see both types, where
the wh-phrase modifies a noun which itself heads a noun phrase complement
to a preposition.
13.2.2 Wh-gaps
We’ve already seen that non-finite clauses often have a missing subject. Now
we turn to gaps which can occur in any part of the sentence.
13.2. WH-QUESTIONS 137
If we consider the cases with who/what in the position of subjects and objects
we see that the fronted wh-phrase is associated with a gap in the syntactic
structure:
Recall that the lexical representation of a verb includes a list of the comple-
ments that it can or must take (its subcategorization information or subcatego-
rization frame). The verb see is transitive, and so is subcategorized to take an
object:
I’ll label the grammatical role of the wh-word as simply ‘WH’ because it
doesn’t have a grammatical role of its own. Instead, it’s associated with a
gap elsewhere in the sentence and it inherits the grammatical role of that gap.
In other words, the wh-phrase is associated with the grammatical role it would
have had if it had been left in situ.
Examples (12a, b) illustrate further types of wh-gap.
In these examples we see that the wh-word together with the whole phrase
which contains that wh-word appears in clause initial position. However, in
many cases, we can ‘strand’ the preposition of a prepositional wh-phrase in its
‘original’ position (preposition stranding).
Where the preposition is taken along with the wh-word (i.e. where it is not
stranded) we sometimes say that it has undergone Pied Piping and say the
preposition has been pied-piped.1
(15) The suitcase we can put in the boot, but the hold-all we’ll have to put
on the back seat
To my grandmother we gave a photo album and to my grandfather we
gave a packet of pipe tobacco
There’s a clear intuition with such sentences that the topicalized phrase a book
of that sort, the suitcase, to my grandmother and so on is ‘really’ the object or
oblique complement of read, put, gave. This suggests an analysis in which the
topicalized phrase corresponds to a gap in the position of the object/oblique.
term illustrates in an intriguing way how language can become ‘illogical’: it’s not the pipe or the
piping that was ‘pied’, but the piper (or rat-catcher) himself. English is full of such mismatches
between form and function. For instance, an electrical engineer isn’t actually electrical . . . .
13.4. LONG-DISTANCE DEPENDENCIES 139
This means that there are actually two topicalization constructions. The one
illustrated in (19a) is often called Left Dislocation. Conversely, we might find
Right Dislocation, as in (20).
(Interestingly, there are languages in which structures such as (19b) are nor-
mal, and in which it would be ungrammatical to leave out the pronoun. See
also the discussion of resumptive pronouns in relative clauses in Chapter 14.)
One very important respect in which the Topic Construction and wh-questions
are similar is that the TOPIC/wh-phrase can be separated from its ‘gap’ by an
arbitrary number of clause boundaries:
It is through examples like (21a–22d) that these constructions earn the name
‘long-distance dependencies’. In English such long-distance effects are usually
only possible with non-finite clauses, what CGEL/SIEG call ‘catenative con-
structions’. In such cases the image of a ‘chain’ is particularly appropriate,
because the wh-phrase can be thought of as lying at the end of that chain.
It is possible to have more than one wh-word in a single sentence. When
this happens only one can appear in initial position, the rest have to remain ‘in
situ.2
Cf:
(Languages differ with respect to this so that examples similar to (24a–24c) are
grammatical in languages such as, say, Bulgarian).
These are indirect ways of asking the questions shown as direct questions in
(28).
Verbs of knowing can take either indirect statements or indirect questions (in
which case we treat the subordinate clause as the answer to an imagined ques-
tion).
The wh-words which introduce the indirect questions are the same as those
found in direct wh-questions. To introduce an indirect yes-no question we
use the words if, whether. These are largely interchangeable in contemporary
English. However, whether tends to occur in the construction whether . . . or not
more often than if, and if has to be separated from or not.
(30) a. Harriet can’t figure out whether Tom has left or not
b. Harriet doesn’t know whether or not Tom has left
(31) a. Harriet can’t figure out if Tom has left or not
b. * Harriet doesn’t know if or not Tom has left
A salient aspect of the structure of NPs is the way that the head noun of the
NP can be modified by various elements. A very important type of modifier is
the relative clause. This is an embedded sentence whose function is to modify
the head noun. Some examples are shown here:
In each of the NPs in (1) there is a clause modifying the head noun (un-
derlined). We can tell that it’s a clause because it’s headed by a verb, whether
finite (1a, b, c, d, e) or non-finite (infinitival) (1f, g). The structure of the relative
clause is somewhat complex, but we now have most of the machinery needed
to describe these constructions.
The relative clause is built up in two ways, using either the all purpose
complementizer that or using an appropriate wh-word.
Notice that the verb built in (2) is transitive and should therefore be followed
by a direct object. However, this is impossible.
143
144 14. RELATIVE CLAUSES AND THEIR KIN
This is of wider interest because in this type of clause there is no gap. The
position where we would expect to find a gap is occupied by an appropriate
pronoun.
Such a pronoun, which takes the place of a gap, is called a resumptive pro-
noun. In some languages (Welsh, Arabic, colloquial Czech, and many others)
relative clauses don’t have any gaps and instead use the resumptive pronoun
strategy. In English ordinary relative clauses don’t usually have a resumptive
pronoun, but in very complicated examples we may find just such a pronoun
creeping in.
(9) the house that Jack had to promise the Council that he wouldn’t build
it using asbestos in the roof
(10) What really knocks me out is a book that, when you’re all done reading
it, you wish the author that wrote it was a terrific friend of yours and
you could call him up on the phone whenever you felt like it. [J. D.
Salinger, writer (1919-2010 )]
Explain the exactly how the two gaps come about in (12).
Relative clauses are used with two slightly different functions, illustrated in
(13):
(13) a. Teenagers who spend all their time text-messaging have larger-
than-average thumbs
b. Teenagers, who spend all their time text-messaging, have larger-
than-average thumbs
In (13a) we’re talking about a subset of all teenagers, namely, those that txt
msg a great deal, while in (13b) we’re talking about all teenagers and adding
the additional (explanatory) information that all teenagers spend all their time
text messaging. In example (13a) we’re restricting the referent of the whole NP
by means of the relative clause. This is the usual function of a relative clause
(and in some languages the only function). Cf the man who we met yesterday -
this tells us about not just any old man but a specific man. We can paraphrase
(13a) as in (14).
(14) Of all teenagers, those who spend all their time text-messaging have
larger-than-average thumbs
Relative clauses of this type merely provide additional information about the
noun they modify. They are rather like appelations such as Mr. Bun, the Baker
or the Prime Minister, David Cameron. The second expression in examples like
these is said to stand in apposition to the first expression. The relative clause
exemplified in (13b) is of this kind and is therefore often called an apposi-
tional relative clause. (Other terms are also used, such as descriptive relative
clause).
In (16) we see a determiner those in the N head position and in (17) we see a
modifier second in head position.
This kind of construction is called the fused head construction in CGEL (410f).
In other reference works on syntax you’ll often see these constructions referred
to as headless phrases or null head phrases.
Other examples of fused heads:
Example (20) has the explicit form of a partitive NP, while (21) is implicitly a
partitive NP (cf [few of them] were taken up).
Representing headless/fused head constructions. If we take those in (16b) to be a
headless phrase we could represent it in one of two ways.
14.3. HEADLESS NPS 147
(22) a. NP b. NP
Det N Det
those ∅ those
You’ll often see one or other of the two representations in (22) in books and
articles on the syntax of such constructions. However, in many ways its more
intuitive to adopt the fused head notion proposed in CGEL:412, under which
we can represent the word as being simultaneously a Det and N.
(23) NP
Det N
those
The tree diagram in (23) wouldn’t be licit in many theoretical models, but it’s
a very useful descriptive device for representing the fact that the determiner
has dual functions.
These are NPs with relative clauses but no lexical head noun. Some examples:
See if you can see the difference in structure between (28) and (29):
(28) [When you fix the car] remember to check the oil
(29) [When you fix the car] is your business
cf
<These are often lumped together with other types of subordinate clause and
called ‘nominal’ or ‘noun clauses’ in ELT literature (cf. Berry p.59) though this
is misleading terminology.>
The fused head type of relative pronoun often corresponds to an ordinary
relative clause modifying a personal pronoun in a more formal register.
These are comparable to relative clauses modifying pronouns such as the one
or very general nouns such as person.
(34) it-clefts
a. The Knave of Hearts stole the tarts
b. It was the Knave of Hearts who stole the tarts
c. It was the tarts that the Knave of Hearts stole
(35) c. [The thing that/which the Knave of Hearts stole] was [the tarts]
14.5. PARTICIPIAL PHRASES AS POSTMODIFIERS 149
Some phrases can’t be (easily) clefted. For instance, we can’t normally focus
sentential complements to verbs using a cleft construction.
There is another way in which a verb phrase can serve as the attributive
modifier of a noun in a noun phrase. Recall that there are two participle
forms for a typical verb, the ing-form (gerund-participle in the terminology of
CGEL/SIEG) and the past participle (perfect/passive participle). Both of these
can head phrases which come after a noun (or pronoun) and modify it.
In the past four chapters we’ve seen a variety of gaps. Here is a brief summary:
150 14. RELATIVE CLAUSES AND THEIR KIN
It can be quite hard at times keeping track of all these different types of gap.
One of the factors that makes it particularly difficult, and which can pose
problems for all kinds of language users and learners, is the fact that we can
easily have more than one gap in a given clause, each with different functions.
In some cases the different types of gap are easy to distinguish. For instance,
consider a sentence which has an infinitival clause with missing subject and
inside that clause a noun modified by a relative clause. The two gaps are
usually fairly easy to spot. Consider a simple case in point. Take the complex
sentence in (41) and expand the direct object the book in the embedded clause
with a relative clause.
However, when the relative clause is itself in the form of an infinitival, it may
have a subject gap in addition to the relative clause gap.
Here there’s a subject gap with arbitrary reference and a direct object relative
clause gap. In (43) we see an unusual case in which the infinitival subject gap
is also the relative clause gap.
15.1 Introduction
In this chapter we look in more detail at the the nominal phrase, including
adjective phrases. We’ll fill out various details of structure and at the same time
look in more detail at the meanings expressed by function words associated
with nouns and noun phrases, and especially at the function of determiners.
15.2 NP structure
Overview
In Chapter 5 example (25) we saw the basic structure for the phrase the two very
big black cats in the garden. Here is the structure for a slightly more elaborated
version of that phrase, showing one way of representing how all the parts
(constituents) fit together. This is by no means the only way such phrases
can be analysed. Different theoretical models provide rather different ways
of representing even relatively simple structures such as these. In particular,
some theoretical models will take the determiner the to be the head of the entire
phrase, thereby making the phrase into a Determiner Phrase or DP. Under such
an analysis the head of the whole phrase is a function word, while the lexical
head, that is, the most important content (lexical) word serves as a kind of
complement to that function word. However, the diagrams for DPs tend to
get rather complex and by no means all descriptions adopt the DP analysis,
so we’ll keep things relatively simple and assume that the head of the whole
phrase is a noun.
The head of the noun phrase is the noun which (almost always!) determines
the basic meaning of the phrase (I say ‘almost always’ because of the existence
153
154 15. NOMINALS - STRUCTURE AND FUNCTIONS
(2) NP
PreDet Det QP AP AP N PP
Q Num Deg A A
Both the other two very big black cats in the garden
Nouns can often take complements of their own, especially if they are derived
from verbs. We’ll see examples later in the chapter.
In Chapter 14 we saw how relative clauses serve as modifiers of N.
A noun can be modified by any number of pronominal APs and postnominal
PPs. However, the functional categories of Quantifier and Determiner can
only appear once in a NP. Similar, only a single (restrictive) relative clause
and a single complement clause are permitted for a given noun. To indicate
‘indefinite number of phrases of this type’ we use the * notation, so that AP*,
PP* means ‘any number of APs, PPs’. The general NP schema can now be
written as (3):
Determiners
(4) Articles
Count Mass
Singular Plural
the cat the cats the wine
a cat cats, sm cats wine, sm wine
The form ‘sm’ represents unstressed ‘some’ (/sm/). As we can see from the
"
table in (4) a mass term like wine takes the same set of articles as the plural
form of a count noun.
The semantics of the articles can get extremely complex. The definite article
basically indicates definiteness. The indefinite article a indicates that the noun
referent is countable and singular. The form sm functioning as a plural form
of a) indicates that the noun referent is countable and plural.
‘Definiteness’ in the context of the definite article means essentially one of
two things. Either it means that an individual or set of individuals can be
identified from the context, or it means that there is an individual/set of in-
dividuals which is unique. Most often the definite article means both these
things. To illustrate the uniqueness property consider an example in which a
person, say, Harriet, owns three cats. Because we know that Harriet is a cat-
owner, the set of cats can be identified from the context and so it makes sense
to say Harriet came home and fed the cats. However, in that case it would be odd
to say Harriet came home and fed the cat because we know there are three and so
the cat doesn’t uniquely identify any one of them.
Now consider the contextual property. The simplest way of seeing the role of
context and expressions with the definite article is to see what happens when
we introduce participants into a discourse and then discuss them.
(5) There were three children in the room, a girl and two boys. The girl
was slightly older than the boys.
ways, referring to the standard parts of the house using the definite article,
even if we are mentioning those things for the first time:
(6) We’ve just visited Harriet in her new house. The kitchen is very large
and the bedrooms have great views of the river. But the chimney needs
repairing and the garden is rather small.
(7) We’ve just visited Harriet in her new house. The kitchen is very large
and the bedrooms have great views of the river. But the funnel needs
repairing and the dungeon is rather small.
One of the reasons why the definite article is one of the trickiest aspects
of English grammar for learners is that there are many situations in which
our contextual knowledge overrides the strict interpretation of uniqueness or
identifiability. For instance, in (6) we can use the definite article for the phrase
the river even if our audience doesn’t actually know that there’s a river near
Harriet’s house, because then the hearer will ‘reverse engineer’ the semantics
of the definite article). and conclude that there is (exactly one) such river.
Similarly, suppose some goes into a room with five windows, and finds that
it is stuffy. They can ask someone to open the window even without specifying
which one (it doesn’t matter which). This is presumably because we normally
only ventilate a room by opening just one of the windows, not all of them.
Similarly, if one of five windows is open and the speaker asks ‘Why is it cold
in here?’ a possible answer would be ‘Because someone’s left the window
open’.
Another common use for the definite article is with generic expressions.
These don’t refer to any specific or uniquely identifiable individual but rather
to a kind, type or species.
The determiner function of some has to be distinguished from its stressed coun-
terpart, /s2m/, which means something like ‘a part of a mass or group (but by
implication not all)’.
Again, we get a different reading when we put stress on any, because then
the word becomes a kind of quantifier with a meaning roughly ‘an arbitrarily
chosen individual or set’.
(12) You can buy any bread/apples (it doesn’t matter, the choice is yours)
However, we often use some or a null determiner with plurals and mass terms
even in questions and negative clauses, usually with a meaning which differs
in very subtle ways from the corresponding expression with any.
(13) a. Have you bought some bread/apples (like I asked you to)?
b. I didn’t buy some bread/apples - that’s what I forgot!
Demonstratives
Possessive pronouns
Notice that this means that possessor expressions are effectively a kind of de-
terminer.
However, there are other determiner-like elements that come before or after
the true determiners. These elements co-occur with determiners, i.e. they
form syntagmatic relations with determiners. These are the predeterminers
and postdeterminers.
Predeterminers:
A small number of words with quantificational meaning can come before the
determiners.
(16) all/both/half
multiplicatives: double/three times/two thirds the effort
other fractions don’t occur here when they have a partitive meaning with count
nouns: *two thirds the students vs. two thirds of the students.
Postdeterminers/Quantifiers (QP)
Numerals and other quantifiers can follow immediately after the determiners.
These quantifiers always come before any attributive adjectives: the three black
cats; *the black three cats.
These elements are sometimes treated as a separate category entirely (Quan-
tifier Phrase, QP). Sometimes they are just called ’postdeterminers’
The postdeterminer quantifiers include expressions such as a few/a good many/little.
Note the difference in meaning between a few (= a small number) and few (=not
a large number, not many, fewer than we would want).
These quantifiers are often used in partitive constructions with of, however.
The predeterminers can also be used in this partitive construction (see below).
Possessive constructions
There are two main ways of expressing a ‘possessive’ construction, with the
so-called ‘Saxon genitive’ and with the preposition of.
In most cases one or the other construction will sound better depending on
the precise meaning relation between the components, emphasis, the structure
of the NP as a whole and various other factors. In general, an inanimate
possessor tends to take the of -construction while animate/human possessors
take the POSS-S construction.
Recursion in PossPs
A possessor phrase is a kind of noun phrase, which itself can take a possessor.
This leads to recursive structures.
(23) a. the name of the brother of the neighbour of the . . . of the director of
the school
b. my neighbour’s friend’s sister’s . . . neighbour’s house
160 15. NOMINALS - STRUCTURE AND FUNCTIONS
In the ‘Saxon genitive’ or ‘POSS-S’ construction the /z/ suffix attaches to the
right edge of a noun phrase and turns head of that phrase into a ‘possessor’.1
Tom’s
a man called Tom’s
(24) the girl who used to know Tom’s house
the guy in the blue jacket’s
the guy you were talking to’s
Only one -s suffix is allowed even where the sense predicts combination of
plural and possessive:
(26) the girl in the corner’s three favourite Italian renaissance painters
Cf also:
Expressions such as *the this cat, *that the cat are ungrammatical because demon-
stratives are inherently definite and so the would only add the component
‘definite’ redundantly. However, in some languages demonstratives and/or
possessives must be accompanied by an ‘agreeing’ definite article (e.g. Greek,
Italian, . . . ).
Partitives. These are constructions that express a part of some quantity.
1I put ‘square quotes’ around the term ‘possessor’ because possession as normally under-
stood is only one of the many meanings conveyed by this construction.
15.2. NP STRUCTURE 161
Complements to noun
Nouns can take PPs as well as finite and infinitival clauses which correspond
semantically to complements to related verbs:
( )
to earn
(36) a good opportunity [S some money]
for earning
The V-ing form seems to be too close to being a noun, so we get such comple-
ments introduced by a preposition:
Expletives
In (39) it has a concrete referent (namely, the dinner). Out of context we assume
the speaker has some specific (inanimate) object in mind.
In (42) there is a locative adverb meaning ‘in/at that place’ (e.g. ‘next to
the pool’). Notice that in (there1a) the adverb there seems to have the SUBJ
function (cf: Down the hill puffed the little train, Out of the hat jumped a rabbit etc.)
However, in (40) it has no referent, and this is even more clear in the case
of (41) (just try asking the questions What’s raining?, or What’s obvious that Dick
15.3. ADJECTIVE PHRASE 163
will be late?). What is the function of it in these examples? It’s simply there
to fill the position of the SUBJ, which would otherwise remain empty (as in
many languages). In English the SUBJ position (in a main clause) has to be
filled by some pronounced (overt) element, so in constructions which lack a
proper subject we add a meaningless element.
An element of this kind that has no referent and is only placed there to meet
syntactic requirements is called an expletive. It’s the pronominal equivalent of
the dummy auxiliary do. In (42, 43) we see that there can be used in the same
fashion. Note that there in these examples does NOT mean ‘in that place’. We
can easily see that by comparing (44) with (45):
Example (44) is odd because it seems that the tree has to be in two places
at once. Example (45) is perfectly fine, because there is meaningless and hence
can’t clash in meaning with here.
In the literary language we distinguish between there is and there are. The
verb therefore agrees with the NP which comes after the copular verb, as in
(43c) there’s is becoming an invariable marker of this kind of construction (in
the present tense, at least), and hence doesn’t undergo agreement.
Adjectives can also take finite, (48), and infinitival complements, (49):
An -ing verb form complement is usually only possible as part of a PP, when
the V-ing form is treated as a kind of noun:
A final point to make about adjectives is that they only show their full range
of complements when they are used predicatively (that is as predicates, and
hence as complements to a verb such as be, seem, become etc.). When used as
attributes (i.e. before the head N of a NP) they cannot take any complements
at all:
However, these expanded phrases are possible if they are postposed to the
right of the head noun:
(56) a. I have yet to meet [a man [AP more stupid than Dick]]
b. * I have yet to meet a more stupid than Dick man]
(58) the girl in the corner, with a red sweater, next to Harriet, . . .
Some adjectives are defective in that they can only be used as attributive
(prenominal) modifiers (Huddleston and Pullum, 2002: 553-559):
(60) a. the house was ablaze ∼ *the ablaze house [and many other adjec-
tives beginning in a-]
b. the child was ill ∼ *an ill child [OK - a sick child]
c. several students were present ∼ *the present students
(61) Several of the students present at the meeting voted against the pro-
posal
3 Note that present has another meaning, roughly ‘current’, which can be used attributively
The present students are less satisfied than last year’s.
166 15. NOMINALS - STRUCTURE AND FUNCTIONS
Chapter 16
16.1 Introduction
This chapter deals with the way that texts are organized. ‘Text’ forour purposes
means ‘two or more sentences, linked to each other and uttered as part of a
single communication’. A text can therefore be something as simple as a pair
of greetings as in (1) or a complete paragraph or this entire booklet or the
complete works of Shakespeare . . . .
To begin with we’ll distinguish two related notions cohesion and coherence.
Coherence refers to the logical organization of a discourse. A coherent text
is one which is organized in such a way that it makes good logical sense, how-
ever it might be expressed. Usually, we guarantee that our texts are coherent
by the use of cohesive devices, to ensure that the parts of the text (e.g. individ-
ual sentences) are linked to each other in a logical fashion. For instance, using
pronouns to refer back to previously mentioned noun phrases is a cohesive de-
vice. Using conjunctions and expressions such as moreover, on the other hand, . . .
and many others also helps the hearer keep track of the logical organization.
16.2 Cohesion
There are five main linguistic devices that we can identify which serve to
achieve cohesion between sentences/utterances.
• Reference
• Ellipsis
• Substitution
• Lexical cohesion
• Conjunction
167
168 16. TEXT ORGANIZATION AND COHESION
The most obvious way to achieve cohesion is to repeat words exactly. How-
ever, this is only normally used for emphasis or to avoid ambiguity, or where
we haven’t established what the topic of the discourse is.
(2) Jack and Jill went up the hill. Jack fell down.
<We wouldn’t normally say ‘He fell down’ here because the previous sentence
isn’t specifically ‘about’ Jack as such>
(3) Today’s lecture is about Noam Chomsky. Chomsky has been described
as ‘America’s leading dissident’
Here we repeat the proper name for emphasis (we could easily have said ‘He
has been described as . . . ’)
16.2.1 Reference
By ‘Reference’ we mean mainly the way that pronouns are used to track partic-
ipants in discourse. Pronouns typically refer to some entity (or sometimes to a
situation) already mentioned in the discourse. This is backwards reference or
anaphora.
(6) a. Goldilocks tried the second bowl of porridge. That was too cold.
b. She tried the third bowl of porridge. This was just the right tem-
perature.
Another example of a cohesive device that may come under this heading in
certain cases is the use of definite descriptions, specfically those introduced by
the definite article the. The definite article is used to pick out a single, unique
object or group of objects that is relevant or salient AND which we assume the
hearer can identify.
16.2. COHESION 169
(9) John and Mary have two children, Jack and Jill. The girl is the elder of
the two.
(10) (In a room with five windows, exactly one of which is open):
It’s drafty in here - could you close the window?
(11) The University has introduced new IT equipment in the cafés. The
students figured out how it worked almost at once, but the teachers are
still struggling to learn how to use it.
Here ‘the students’ refers to ‘one unique identifiable group consisting of all the
students (at Essex)’ and ‘the teachers’ refers to ‘one unique identifiable group
consisting of all the teachers (at Essex)’.
In these examples the definite phrase with the refers back to some en-
tity/entities that have been mentioned already. But we can sometimes ‘reverse
engineer’ the utterance and figure out that the speaker must have been refer-
ring to some unique, identifiable referent:
(12) John and Mary have two children. The girl is the elder of the two.
Here we just assume that exactly one of the children is a girl, otherwise the
discourse wouldn’t make sense It would be incoherent because the cohesion
device of using the definite article would have been wrongly used. Other
examples:
(13) Harriet was late for work this morning and she left home without feed-
ing the cat.
16.2.2 Ellipsis
We frequently omit words and phrases that have already been mentioned or
which are otherwise obvious.
Notice that we still have to observe certain grammatical niceties, e.g. we still
need an indefinite article on the singular noun:
(16) a. Harriet has arrived already, but Tom hasn’t <arrived yet>
b. Harriet has been correcting her assignment, and Dick might have
<been correcting his/her assignment>, too
c. Harriet has been correcting her assignment, and Dick might have
been <correcting his/her assignment>, too
(17) *Harriet has been correcting her assignment, and Dick might have been
correcting <his/her assignment>, too
Similarly, an infinitival verb phrase can be elided under identity with a previ-
ously mentioned verb phrase, but the infinitive marker to has to be retained.
Without the right context elliptical expressions can be hard to process. What
does the following mean?
16.2.3 Substitution
We frequently substitute indefinite NPs and VPs with special expressions which
behave very much like pronouns. These are:-
• one - indefinite NP
• do so - VP
(20) Tom would like a sandwich but Harriet doesn’t want one
(22) Tom drives a red car and Harriet drives a green *(one)
16.2. COHESION 171
<The symbol *(. . . ) means ‘it’s ungrammatical if you miss out . . . ’>
The equivalent substitution for verbs and verb phrases is ‘do so’:
(23) a. Harriet is happy to walk to work but Dick refuses to <walk to work>
Ellipsis
b. Harriet is happy to walk to work but Dick refuses to do so Substi-
tution
A common error with second language learners of English is to replace ‘do so’
with ‘do it’:
We only use ‘do it’ for specific, identifiable events/actions, not general ones
like walking to work:
(25) We want Dick to move his motorbike from the driveway but he refuses
to do it.
We can make a text cohesive by using words that are related to each other
in meaning. The usual way in which we describe such meaning relations is in
terms of (i) semantic fields and (ii) sense relations especially hyponymy/hypernymy,
meronymy. Hyponyms are subordinate terms while hypernyms are superor-
dinate terms, as in
Words are grouped into networks of related meanings. For the purposes of
cohesion we have subordinate/superordinate terms:
16.2.5 Conjunction
We can explicitly mark the logical (cohesive) relations between one sentence/utterance
and another by means of linkers or conjunctions. These include words such
as and, moreover, furthermore, but, nevertheless, however, yet, . . . . We can identify,
broadly speaking, four logical and linguistic relationships that are typically
expressed or realized by conjunctions.
The conjunction and often has implications of temporal sequence rather than
just addition.
Although the notions of cohesion and coherence are clearly related they are not
identical. We can see this because we can construct texts that make good use of
cohesive devices but which still aren’t coherent (at least not in any reasonable
context of utterance). Similarly, we can construct texts that are coherent (given
some appropriate and reasonable context) in the absence of cohesive devices.
Cohesion without coherence:
I’m going to try a new recipe using olives. These are an important export for
many Mediterranean countries such as Greece. That country is often regarded as
the birthplace of democracy. As Churchill observed, that may not be ideal form of
government, but all the others are much worse. ‘Churchill’ is also the name of an
insurance company.
Exercise: identify all the cohesive devices in this passage.
Coherence without (very much) cohesion:
Mr. Blair declared that Saddam Hussein was able to deploy weapons of mass de-
struction at 45 minutes’ notice. In the past 50 years the United States has invaded the
16.3. COHESION VS. COHERENCE 173
following countries: Grenada , Panama, and Haiti. From 1995-2000 Dick Cheney was
CEO of the Halliburton Corporation. Draw your own conclusions.
<Adapted from an actual conversation(!)>
Explanation (the context is discussion of the build up to the US invasion of
Iraq in 2003).
(1) Mr. Blair declared that Saddam Hussein was able to deploy weapons of
mass destruction at 45 minutes’ notice.
(2) In the past 50 years the United States has invaded the following countries:
Grenada , Panama, and Haiti.
(3) From 1995-2000 Dick Cheney was CEO of the Halliburton Corporation.
(4) Draw your own conclusions.
From (2): these countries are all extremely small and have virtually no armed
forces (Panama is banned by its constitution from having a standing army).
From (1) we conclude that the US only invades countries that are almost en-
tirely defenceless. If (1) is true it follows from (2) that the US would never
invade Iraq. But the US did invade Iraq therefore (1) must be false. Addi-
tionally, Dick Cheney was Vice-President to George W. Bush from 2000-2008.
The Halliburton Corp. was given very substantial US government contracts
to rebuild Iraqi infrastructure after the invasion (using Iraqi petrodollars). (4)
The implications are that (i) Blair lied, that the US invaded Iraq knowing it had
no effective (conventional) means of defence and that Cheney encouraged the
invasion knowing it would profit a company he was closely connected with
(e.g. as a major private shareholder).
None of this text makes any sense, of course, unless you have access to all
of the relevant contextual information.
We’ve seen how the context of utterance is important for determining the
meaning of deictic words such as personal pronouns. Another instance is
illustrated by the following mini-dialogue which took place between the author
(A) and a colleague (D):
(28) D: Hmph! If I’d known it was going to be fish I’d have put in my
contact lenses.
A: You don’t like kippers, then.
In (28) A has clearly interpreted D’s utterance in exactly the manner intended
by D, because the context provided ample clues. Actually, A then continued
‘Do you realize how incomprehensible your last statement would sound out
of context?’ I have yet to find anyone who can provide the correct context for
D’s utterance without a lot of hints.1
1 Oh, alright then, here’s the background: A and D were at a conference. A was finishing
breakfast, and there was only a short time before the first talk. D had just arrived, somewhat
late, wearing his spectacles, only to discover that the food being served was something he didn’t
like.
174 16. TEXT ORGANIZATION AND COHESION
This discourse seems hopelessly incoherent until we learn that Mrs. Smith is
trying to rent a flat and Mr. Jones is the landlord.
Context is essential for understanding how a text is coherent, and can often be
important for understanding how a text is cohesive.
There’s another sense in which context is important for computing the con-
tribution of cohesive devices, namely in the articulation of utterances into topic
and focus and into given (old) information and new information.
Consider the mini-dialogue in (30):
The capital letters on MARY indicate that this word is pronounced with more
emphasis: it’s slightly louder and relatively longer than the other words and
it starts at high pitch and falls rapidly to low pitch (see Chapter 17). Actually,
the rest of Tracy’s utterance in (30) is completely redundant and could easily
be omitted completely. Now consider (31), a slight variant on (30) in which
Tracy emphasises a different word:
The dialogue in (31) sounds very odd, as though Tracy had misheard or mis-
understood what Joan said.
The reason for the oddness of Tracy’s utterance in (31) is intuitively clear:
Tracy and Joan both know that the conversation is about two books on lin-
guistics and Joan wants to know the author. But Tracy’s reply in (31) makes it
sound as though Joan needs to know the precise number of books on linguis-
tics that Mary has written. On the other hand, Tracy’s utterance in (31) would
be a perfectly fine answer to (32):
16.4. TOPIC-FOCUS ORGANIZATION 175
Phenomena of this sort are often studied under the heading of information
structure. We say that Tracy’s utterance in (30) is articulated into two com-
ponents, MARY, which is new information and the rest of the sentence which
is old information or given information. We can make this more explicit by
‘translating’ (30) into the representation in (33):
In (33) we use a variable ‘x’ to represent unknown information, and Tracy’s re-
ply has to provide a value for that unknown. Clearly, the ‘given’ component in
Tracy’s utterance is the part that can be safely omitted. Similarly, the question
in (32) can be represented as in (34):
We can now see what goes wrong with (31). The answer provided by Tracy
is attempting to fill in a value for the wrong variable. (Check that you under-
stand exactly how this works by translating the ill-formed dialogue in (31) and
comparing it with a possible dialogue based on (32)).
We’ve illustrated given/old and new information using the traditional de-
vice question-answer pairs, sometimes called the ‘commutation test’ for given/new
information. However, the question part can remain implicit or presupposed.
Consider (35):
(35) Hey, I’ve just heard that Mary has written ANOTHER book on linguis-
tics.
An utterance of (35), in which the word ‘another’ bears the main emphasis, is
only felicitous if the speaker thinks the addressee already knows that Mary has
written at least one book on linguistics. For this reason, the ‘given’ portions of
the representations in (33, 34) are often known as the presupposition (though
this term has a number of other, slightly different, uses, as we will see later).
By varying the position of emphasis in a sentence such as (35) we can vary
the articulation into given/new information. In principle, any of the content
words of a sentence can be emphasised in this way and turned into new infor-
mation. However, we can emphasise more than just individual words. Con-
sider the mini-dialogue in (36):
(36) Joan: What topic has Mary written a new book on?
Tracy: (Mary has written a new book on) LINGUISTICS
The structure of (36) is essentially the same as that of (30) or (35), except that
this time x = linguistics. But now consider (37):
176 16. TEXT ORGANIZATION AND COHESION
Here the new information is the whole phrase a new book on linguistics. The
placement of emphasis in Tracy’s utterance in (37) is exactly the same as in
Tracy’s utterance in (36) but the scope of the new information is the whole
phrase, not just a single word. In fact, the scope can extend even further than
this. In (38) the whole utterance can be interpreted as new information:
Closely related to the notion of given information is the notion of topic. Broadly
speaking the topic of a sentence(or utterance of a sentence) is what the sentence
is about. In English identity of the topic tends to be implicit, though we can
sometimes explicitly announce a topic, as in (39):
We can divide a sentence such as (39) into two parts, the topic, Mary, and what
we say about Mary, the comment:
From the examples we’ve seen so far it looks rather as though the topic of
a sentence is the same as the old or given information while the comment
conveys new information. In fact, matters are more complex than this, and
one of the important current research questions in this area of linguistics is
pinning down precisely what a language conveys when it codes a word or
phrase as topic or comment, or what exactly constitutes old/new information.
To give a flavour of some of the complexities, consider sentences (41a, b):
(41) a. Sue called Mary a linguist and then she INSULTED her.
b. Sue called Mary a linguist and then SHE insulted HER.
In (41b) the two pronouns she and her are accented, which implies that they
must be new information. On the other hand, they are also pronouns, which
implies that they are old information. What is new information here is actually
the identity of the people to whom the pronouns refer. Presuppositions
Consider the examples in (42):
16.6. INFORMATION STRUCTURE AND COHESION 177
If sentence (42a) is true then (42b) must also be true. But the opposite of (42b),
namely (42c) must also be true if (42a) is true. If it were the case that Tom
hadn’t been reading a book then it would be very odd to say (42a).
What if Tom wasn’t reading anything earlier? Would Tom stopped reading the
book be false? That seems odd - rather, we would say that Tom stopped reading
the book was just inappropriate. So, in order for us to be able to decide whether
a sentence like (42a) is true or false we need to make the prior assumption that
(42b) is true, otherwise, (42a) will be inappropriate.
The proposition expressed by (42b) which expresses this prior assumption is
called a presupposition of (42a). A presupposition of a sentence remains true
even if that sentence is negated. That is, if we say Tom didn’t stop reading the
book, that still presupposes that Tom was reading the book earlier.
It is not just words which bear presuppositions. Function words and even
entire constructions may do so too.
Example (43) would be odd (infelicitous, inappropriate, . . . ) if Harriet owned
more than one cat (though we couldn’t really say that (43) was false as such).
Likewise, (44) would be infelicitous if Harriet didn’t actually feed (any of) the
cats, even if she had several cats and one of them was a ginger tom.
(44) One of the cats that Harriet fed was a ginger tom
This is because the expression one of the cats that Harriet fed in (44) presupposes
that Harriet fed at least two cats. This presupposition remains even if we
negate the original sentence: One of the cats that Harriet fed wasn’t a ginger tom.
This type of presupposition is independent of information structure and
instead is dependent on the meaning of individual words and constructions.
Nonetheless, in each case the presupposition is a proposition which must be
assumed to be true in order for a given utterance to be felicitous.
emphasize the distinction between old information and focussed, new infor-
mation. The limiting case of this is ellipsis, in which the old information is
simply not mentioned.
Extra reading: for a summary of topic-focus organization (‘information pack-
aging’) see Huddleston and Pullum (2005: Ch15)
Chapter 17
Conversation
17.1 Introduction
Most of what we have learnt so far is true of written and spoken language
equally.1 However, there are various ways in which spoken language differs
from written language, some obvious, some not so obvious. The most obvious
difference is that spoken language has phonological properties (largely) lack-
ing in written language. In particular, we can express a good deal by means of
intonation, accent, rhythm and speed of speaking. Another difference is that
spoken language generally involves two or more speakers interacting with
each other (‘talk-in-interaction’). One branch of linguistics that studies the
way conversations are structured is Conversation Analysis (CA). An impor-
tant aspect of CA is the way that speakers/hearers negotiate who is to speak
when.
In this chapter we’ll look at the organization of intonation and accent and
how it can express various meanings and functions, including the topic-focus
(information structure) organization mentioned briefly in the previous chapter.
We’ll conclude with a brief summary of the structure of typical conversations.
A slightly less obvious difference between speaking a writing concerns typi-
cally modes of expression: spoken language tends to use simpler clause struc-
ture than written language, and some vocabulary is basically part of spoken
and not written language while other vocabulary is basically written rather
than spoken. However, the boundaries are difficult to draw, and neither writ-
ten nor spoken language is a single homogeneous type (or genre). The lan-
guage of a church sermon may well be essentially that of the written genre,
1 The terms ‘spoken’ and ‘written’ are somewhat opaque nowadays because a great many
texts are the result of electronic media in which you can mimic conversation but in the form of
a written/typed text. Thus, with instant messaging and similar technologies we can see all the
properties of spoken conversation, except that it happens to be written down (and therefore the
intonational characteristics have to be fudged, e.g. by using capital letters, emoticons, and far
too many exclamation marks!!!)
179
180 17. CONVERSATION
Probably all languages have conversational particles which guide the hearer in
interpreting utterances. One such particle in English is after all (ignore the fact
that this looks like two particles - it’s common for such particles to have the
form of a short phrase). What does this particle mean in example (1)?
(1) Natasha can do the Russian interpreting. After all, she’s from Moscow.
A speaker would normally only use after all in a sentence such as (1) if they
believed that the hearer already knew the content of the proposition which
after all introduces. This is clearly seen in (2):
(2) We MUST go out somewhere nice tonight, after all it IS your birthday
It’s hardly likely that the speaker would use (2) to inform the hearer that it
was her birthday (or even to remind her of this fact).
But why should anyone tell the hearer something they already obviously
know? In particular, how can such an utterance ever be relevant to the hearer?
The answer is that after all serves to tell the hearer that the speaker believes
that this (shared) fact provides crucial evidence to back up what the speaker
has just claimed. The fact that the proposition which is introduced by after
all is presented as unequivocal shared knowledge therefore makes it difficult
for the hearer to disagree. As a result, even an apparently wholly redundant
utterance can be relevant.
There are probably about a hundred or so conversational particles in com-
mon use in English. Most of them are ignored in language teaching curricula
even though they are a vital part of spoken English. Their meaning is ex-
tremely difficult to pin down and the only real way to learn their use is by
seeing numerous examples of them in the right context. (This is how corre-
sponding conversational particles are taught in the German as a Foreign Lan-
guage pedagogic tradition, for instance.) Another construction which fulfils
17.3. SENTENCE ACCENT IN DISCOURSE STRUCTURE 181
the same role as conversational particles is the tag question (haven’t they, won’t
she, aren’t you, isn’t it etc.).
Here is a list of some of the more common particles:
actually after all alright (all right)
but kind of like
now oh so
sort of then though
you know you see well
Rhythm
Intonation
17.3.2 De-accenting
Full lexical words have to be deaccented if they have low information content
(‘old information’), e.g. anaphorically used nouns:
(5) They chased the mugger down the street, but they didn’t manage to
cátch the chàp/blìghter/bàstard/. . .
Also:
(6) I always promise my friends that I’ll send them lots of letters but I often
don’t find the tíme to wrìte much
Commutation test:
17.4. FOCUS, ACCENT AND INFORMATION STRUCTURE 183
Accented elements supply new information, hence can be the focus of a wh-
question.
In the absence of special context, phrasal accent tends to fall on the last
lexical word of the utterance, 1a. NB: this could also answer questions 4.:
Contrast
Even unstressed syllables (!) can be accented in order to contrast two words
(‘metalinguistic contrast’):
Focussing particles
Tone units in English are usually treated as having the strong accent first
(e.g. trochaic feet). Apparent cases of initial weak syllables are often analysed
as involving a ‘silent beat’ (cf. musical upbeat/anacrusis). This can be notated
using the symbol , as in:
ˆ
Stress timing
(21) a. Speed / kills
b. Speeding / kills
c. Speeding can / kill
d. Speed / kills you
e. Speeding can / kill someone
Two tone units in each sentence, each takes roughly the same time to deliver
(given a constant speaking rate). Number of syllables is less important than
number of accents. Hence, English is stress-timed language.
This is not true of languages such as Japanese, Polish etc., where number of
syllables remains constant irrespective of stresses: syllable-timed language.
17.6. INTONATION 185
17.6 Intonation
yes
She made one
terday
terday
She made one
yes
yes
She made one
terday
ter
She made one yes
day
The tone extends for the whole of the tail, e.g. where the first syllable of the
tone unit is the Nucleus:
Tune 1, high fall
Harr
iet wrote a letter to Tom
m
To
to
letter
wrote a
Harriet
Tune 4, fall-rise:
Harr
m
iet wrote a letter to To
Tune 5, rise-fall:
i
Harr
et wrote a letter to Tom
17.7.1 Turn-taking
So far our discussions have involved only very short stretches of speech and
in general we’ve been able to make our points using constructed examples.
However, there is another aspect to the act of talking which isn’t covered by
the perspective we gain from pragmatics. When we listen to a group of people
in conversation we generally find that the talk is organized in a rather effi-
cient fashion. And yet conversation usually involves at least two people who
may each want to speak, and who don’t necessarily want to listen. How then
does people negotiate who is to ‘have the floor’ and when that privilege can
be ceded to another participant in the conversation? In other words, how is
turn-taking organized? It turns out that there is a host of more or less subtle
linguistic signals that we use for this purpose. In addition, talkers often need
to convey their attitude to the conversation without explicitly discussing it. For
instance, there may be topics which a talker doesn’t want to discuss in detail,
or alternatively there may be topics which the hearer wishes to know more
about. Languages have a variety of means to allow talkers to give each other
information of this kind. The study of these various devices is Conversation
Analysis (CA). CA originated in the work of social psychologists and sociolo-
gists and for a long time was poorly integrated into the kinds of mainstream
linguistics we have been discussing in this book. Even studies of pragmatics
tended to ignore CA. However, recently specialist in a variety of areas of lin-
guistics, including phonetics, pragmatics and language disorders have been
looking in detail at the way talk is managed (or mismanaged).
The prototypical, and in some sense simplest, kinds of interchange are paired
utterances, such as pairs consisting of question-answer or offer-acceptance, but
including more or less formulaic pairs such as or greeting-greeting. Such pairs
are called adjacency pairs. The key fact about such pairs is that the first ut-
terance virtually demands a response. In other kinds of talk, however, it may
not always be so obvious who should talk and for how long. To manage the
188 17. CONVERSATION
On the other hand in (34) Emma uses a different conversational particle, that’s
right, and thereby is effectively signalling that she knew this all along, and
hence implying that she didn’t need to be told. In CA this implication is
known as a claim to epistemic priority (roughly, ‘I got there first, actually’):
17.7. THE STRUCTURE OF TALK (CONVERSATION ANALYSIS) 189
By using conversational particles such as that’s right (rather than oh) speakers
can try to manipulate their position in the conversation and make it less easy
for others to disagree with them.
1 Zoe Mum?
2 Lyn hello (pause)
3 Lyn I’m here (pause)
4 Zoe okay- (pause)
5 Lyn ((coughs/clears throat)) (pause)
6 Zoe hello
7 Lyn hi
8 Zoe where’s the cigarettes (shorter pause)
9 Lyn in the kitchen (long pause)
10 Zoe the camera’s on
11 Lyn yes (slight pause)
12 Zoe are you talking to it while you WORK?
13 Lyn no (slight pause) [heh heh-
14 Zoe [what you DOING then
15 Lyn hahh hahh hahh (pause)
16 Zoe what’s the point (slight pause)
17 Zoe oh god (slight pause) look what I’m wearing
...
31 Zoe ↑oh ↑god (.) look what ↑I’m wearing
33 Lyn eh hah hah [HAh
34 Zoe [hah
35 Lyn you look like (slight pause) ↑Fa:gin
37 Zoe eh HAh HAh h[ah
38 Lyn [hah hah
40 Zoe w’ maybe I am.
41 Lyn y’ just need th’ little gloves, with th’ ↑fingers out.
44 Zoe v’ funny
Chomsky, N. (1965). Aspects of the Theory of Syntax. The MIT Press, Cambridge,
MA.
Leech, G., Deuchar, M., and Hoogenraad, R. (2006). English Grammar for Today.
A New Introduction. Palgrave Macmillan, Basingstoke.
Radford, A., Atkinson, M., Britain, D., Clahsen, H., and Spencer, A. (2009).
Linguistics: An Introduction. Cambridge University Press, Cambridge, 2nd
edition. First edition published 1999.
191