1.

Introductory Concepts

1.1. Syntactic categories

All words in a sentence belong to a particular part of speech or, in more modern parlance,
category or syntactic category, like those in (1).
(1) Category Abbreviation Example
a. noun N computer, city, stupidity, event,John, London
b. verb V hear, think, disagree, shorten, eavesdrop, exist
c. adjective A good, obscene, demented, lovely, schoolmasterly
d. preposition P off, by, in, with, from, to, at, inside, despite
e. adverb Adv slowly, often, now, mostly
f. determiner D (or Det) a, the, this, those 1

Here we will not try to give a set of completely failsafe criteria for determining the category a
word, but will describe some issuesarising in defining the various categories. One kind of
criterion is semantic, i.e. based on meaning. Such criteria take the form of statements such as
‘a noun denotes a person, place or thing’, ‘a verb denotes an activity or state’ or ‘an adjective
denotes a property’. Such semantic generalisations are of limited use because they are only
tendencies, not absolute rules. Thus, there are nouns which denote activities (the hammering),
events (recital), states (their love) and properties (silliness).
More reliable evidence for determining the category of a word comes from morphological
and syntactic criteria. Examples of morphological criteria would be that only nouns can take a
plural affix (tables, intervals) and that most verbs change their morphological form according
to the requirements of tense and agreement (I talk, she talks, I talked). If you can add -ly to a
word to form an adverb, you know that word is an adjective (slow>slowly). Examples of
syntactic criteria for various categories are given below. In each case, assume that the gap is
filled by a single word.
(2) a. They have no [N ].
b. The [N ] is very [A ].
c. They can [V ]

1.2. Constituents
Identifying the syntactic category of each word in a sentence is crucial, but is only the first
step in describing the mental processes which allow native speakers to create a sentence.
Suppose we tried to do this for the sentence in (3). An analysis only based on categories
would take the form in (4).
(3) That man likes that woman.
(4) S  D+N+V+D+N (Translation: A sentence can consist of the sequence determiner
+ noun + verb + determiner + noun.)
It is easy to show that the human braindoes not use rules like (4) when it creates sentences.
Suppose we want to give more information about the man spoken of in (3) and/or to say that
he likes someone or something other than that woman. We could then replace that man and
that woman with different, more complex expressions. A small selection of the infinite
number of possible replacements is given in (5) and (6).
(5) a. that old man
b. that old man with the bottle of beer
c. that extremely old and decrepit man with a nearly empty bottle of cheap beer
1 Determiners (some of which are often called 'articles') are discussed in more detail in sectionsError: Reference
source not found and Error: Reference source not found.
 d. that man over there near the window
(6) a. second-rate music by unknown bands
b. people with a flair for the unusual
c. paintings by certain fairly weird and decadent artists
d. his collection of photographs of Victorian guesthouses in Tasmania
The possibility of replacing that man in (3) with any expression in (5) and that woman in (3)
with any expression in (6) gives us twenty-five sentences. We would thus need twenty-five
different rules of the type in (4). Once we start adding further material to the sentence (say, I
believe that at the beginning of the sentence, obviously before likes,and/or and plus any
appropriate string of words you can think of at the end of the sentence), we would need an
infinite number of rules of the type in (4). No scientist would be satisfied with the assumption
that native speakers of a language create sentences using an infinite set of rules. It would be
physically impossible for humans to learn all these rules. Also, such rules are purely
descriptive: they just state observed empirical facts without explaining them.
A way out of this impasse emerges when we realise that what has been lacking in our
analysis of sentences is the idea that words can combine with other words to form larger
groups of words, called constituents. Constituents combine with other words or constituents
to form yet larger constituents, until we eventually have a full sentence. The expressions listed
in (5) and (6) were examples of constituents called noun phrases (NPs), expressions which
include a noun and some material giving additional information about it. NPs can typically be
replaced by pronouns: each NP in (5) and (6) can be replaced by he, her, it,them etc. as
appropriate. We will define NPs and other types of constituents more precisely later. Our
purpose now is merely to show how recognising constituents greatly helps us in analysing
sentences. Now consider (7), which will be rejected later and should not be memorised, but is
far better than (4):
(7) S  NP V NP (Translation: A sentence can consist of the sequence NP+V+NP.)
Even if we are only interested in describing the twenty-five possible sentences consisting of a
NP from (5), a verb and a NP from (6), the benefits of recognising constituent structure
should now be apparent. If we use rules of the type in (4), we would require twenty-five rules
to describe these sentences, whereas (7) describes all twenty-five sentences with just one rule.
We emphasise again that the rule in (7) is being used only as a way of showing the need for
constituent structure. We will later show how this rule can be improved upon.
As another argument for the need for constituent structure, consider the following sentences
containing the possessive ‘s morpheme:
(8) a. [That lady]’s husband left.
b. [That lady over there]’s husband left. (=the husband of that lady over there...)
c. [That lady near the door]’s husband left. (=the husband of that lady near the door...)
d. [That lady you talked to]’s husband left. (=the husband of that lady you talked to...)
e. [That lady you saw]’s husband left. (=the husband of that lady you saw...)
We cannot describe the behaviour of possessive 's in terms of the category of the words it
attaches to: 's can appear immediately after a word of any category. Moreover, 's does not
characterise the word to its immediate left as a possessor: the door in (8)c) does not have a
husband. Rather, possessive 's attaches to a certain type of constituent (marked by square
brackets in (8)), namely a NP. We cannot describe the behaviour of possessive 's without
using the notion of NP. Thus, we cannot describe sentence structure without constituents.

1.3. Tests for identifying constituents

In all sciences, linguistics included, one should be able to assess the truth or falsehood of a
claim by means of objective tests. We now introduce some tests for establishing whether a
string (i.e. group of words) is a constituent or not.

a) Proform test. Proforms are expressions like she, them, somewhere, do so, there which
have the function of representing a constituent which has already been mentioned, so that one
need not pronounce/write the constituent twice. The best-known type of proform is a so-called
pronoun, which replaces a NP, e.g. she/him/they. If you can replace a string with a proform,
the string is a constituent. (9) illustrates the use of the proform test in finding constituents in
(9)a).
(9) a. The lady running the group handed in her resignation on Friday at noon.
b. She handed in her resignation on Friday at noon. [Thus, The lady running the group is
a constituent]
c. The lady running it handed in her resignation on Friday at noon. [Thus, the group is a
constituent]
d. The lady running the group did so on Friday at noon. [Thus, handed in her resignation
is a constituent]
e. The lady running the group handed in her resignation then. [Thus, on Friday at noon is
a constituent]

b) Question test. If you can convert a sentence into a question using a wh-expression (e.g.
where/how/when/why/what/who(m), and phrases like with whom?, at what time?, in whose
house?), the string that the wh-expression replaces is a constituent. (Wh-expressions are
proforms.) The answer to the question is also a constituent. (10) illustrates this with reference
to (9)a). In each case, A and B refer to different speakers, and B’s answer is a constituent.
(10) a. A: What did the lady running the group hand in on Friday at noon?
B: Her resignation.
b. A: Who handed in her resignation on Friday at noon?
B: The lady running the group.
c. A: When did the lady running the group hand in her resignation?
B: On Friday at noon.

c) Movement test. If a string can be moved to some other position in the sentence, it is very
likely to be a constituent. The following examples apply this test to identify constituents in the
respective (a) sentences.
(11) a. Egbert was reading a thick book about formal logic on the balconyon Sunday.
b. On Sunday, Egbert was reading a thick book about formal logic on the balcony.
c. On the balcony,Egbert was reading a thick book about formal logic on Sunday.
d. Egbert was reading on the balconyon Sunday a thick book about formal logic.
(12) a. Rover ran out of the house. b. Out of the house Rover ran.
(13) a. Ann is not a fan of mindless techno music.
b. A fan of mindless techno music, Ann is not.
(14) a. Gertrude wasn’t interested in art.
b. Interested in art, Gertrude wasn’t.
(15) a. Hortense didn’t win the race.
b. Win the race, Hortense didn’t.
In these examples, the movement does not change the number of words in the sentence, but in
more complex cases movement is combined with other operations like passivisation:
(16) a. The people next doorbought a large, tasteful statue of Elvis Presley.
b. A large, tasteful statue of Elvis Presleywas bought by the people next door.
d) Cleft test. Error: Reference source not founda) can be changed into the sentences in (b-d).
These are cleft sentences. (Cleft is from cleave meaning ‘divide’; cleft sentences are ‘divided
in two’.) The general form of cleft sentences is Error: Reference source not founde).In cleft
sentences the material between be and that,