Introduction to Syntax∗

Manuela Schönenberger and Cornelia Hamann, revised by Michael Treichler

1 Introduction
Linguistic theory has the following three main objectives:

1. to determine the set of rules (knowledge) which generate (i.e. enumer-

ate explicitly by means of rules) the grammatical utterances of a native
speaker,

2. to determine the common properties pertaining to the different languages

of the world, and

3. to gain some insight into how the human brain works by studying lan-
guage

The concepts of competence and performance are central to linguistic theory:

1. Competence refers to the implicit knowledge a speaker has of his mother

tongue. This knowledge is mostly unconscious, and more or less com-
plete by the age of 6. The concept of competence is similar to Saussure’s
concept of langue.

2. Performance refers to the actual use of this linguistic knowledge. Lan-

guage production may be affected by one’s psychological state, e.g. tired-
ness, drunkenness, stress. Performance - in contrast to competence - is

∗
This introduction to syntax is aimed at 1st-year students in English linguistics at the
University of Oldenburg. Various sources were used, in particular the syntax manuscript
by Genoveva Puskás (2002), the book ’English Grammar: A generative perspective’ by
Liliane Haegeman and Jacqueline Guéron (1999), and Cornelia Hamann’s lecture notes.
We are grateful to Solveig Bosse and Danica Grade for helpful comments.

1
 not perfect and errors can occur, as, e.g. in slips of the tongue or in-
complete sentences. In other words, these errors belong to the domain
of linguistic performance, and are independent of the linguistic compe-
tence. The concept of performance more or less corresponds to Saussure’s
concept of parole.

Syntax is one of the core domains of linguistics. Other core domains are mor-
phology, semantics, pragmatics, phonetics, and phonology. While morphology
investigates the internal structure of words, syntax concentrates on how words
are arranged in a sentence. It is concerned with the structure of sentences and
the smaller parts which make up sentences. It studies the rules which govern
word order and sentence structure. Some of the aspects discussed in syntax
are:
Word order: there are variations in word order between languages spoken
nowadays (e.g. English vs. German) as well as within a language from a
diachronic point of view (e.g. Old English vs. present-day English).

(1) a. Hedwig smiled gleefully.

b. Harry noticed that [Hedwig smiled gleefully].
(2) a. Hedwig lachte überglücklich.
b. Harry beobachtete, dass [Hedwig überglücklich lachte].
(3) a. God him worhte Ta reaf of fellum
God them wrought then garments of skins
’Then God made garments of skin for them.’
b. Dæt he Saul ne dorste ofslean
that he Saul not dared murder
’that he didn’t dare to murder Saul’

Hierarchies: there are hierarchical relations between the elements of a sen-

tence. A sentence is not simply a string of words; it comes with a hierarchical
structure.

(4) a. Harry and Ron think that they get on very well with each other.
b. They think that Harry and Ron get on very well with each other.
c. That Harry and Ron get on very well with each other is what they
think.
(5) a. Dobby washes his new sock [with colourful stripes] [with care].
b. *Dobby washes his new sock [with care] [with colourful stripes].

2
Ambiguities: some sentences have more than one meaning. If the ambiguity
is structural - as opposed to lexical1 - it can be related to the syntactic structure
of the sentence.

(6) Hermione saw the witch with binoculars.

a. Hermione saw [the witch with binoculars]. (the witch has binocu-
lars)
b. Hermione saw [the witch] [with binoculars]. (Hermione has binoc-
ulars)

To visually represent the structure of sentences phrase markers (tree dia-

grams or labelled bracketing) are used. A detailed representation of the
sentence in (7) is given in (8) and (9):

(7) Hermione has often smiled at her best friends after a quidditch game.

Labelled Bracketing:

(8) [ IP [ DP Hermione] [ I has] [ AdvP often] [ VP [ VP smiled [ PP at [ DP her [ NP

[ AP best] friends]]] [ PP after [ DP a [ NP quidditch game]]]]]].
(9) Tree diagram:

1
Example (i) is a case of lexical ambiguity. The ambiguity is due to the different meanings
of match, which can either refer to a sporting encounter or a thin piece of wood used to
light a fire.
(i) Harry has lost the match.

3
2 Word Categories
2.1 Lexical Categories
A basic distinction is drawn between lexical categories and functional cat-
egories. Section 2.1 discusses lexical categories, and Section 2.2 functional
categories. In English there are five lexical categories: N(ouns), Adj(ectives),
Adv(erbs), V(erbs) and P(repositions)2 . Lexical categories are open-class
categories, (i.e. new members can easily be added to the list of nouns, verbs,
etc.), and they have a rich semantic content.

2.1.1 Nouns (N)

The words in (10) are nouns. You may simply know this, but how can you
actually support this intuition? There are basically two tests we can use to
show that the words in (10) are nouns. The first concerns morphology, the
second distribution.

(10) cat, frog, lynx, ox

1. Morphology
In English, nouns can usually be inflected for plural, i.e., they bear an
overt morphological mark for plural. The regular form is -s (pronounced
as [r], [z], or [Iz], which are allomorphs of the morpheme plural), as in
(11-a),(11-b),(11-c), but there are also less productive forms, as in (11-d)
and (11-e).

(11) a. a cat - cats

b. a frog - frogs
c. a lynx - lynxes
d. an ox - oxen
e. a mouse - mice

There is hardly any overt case distinction in modern English. The phrase
the wizard is a subject in (12-a) and an object in (12-c). Although the
wizard has different functions in (12-a) and (12-c) (subject vs. object)
and occupies different positions, no difference in case is visible. When the
wizard is replaced by a pronoun, as in (12-b) and (12-d), a case distinction

2
Adverbs are sometimes not regarded as a separate lexical category, but subsumed under
the category of Adjectives.

4
 between nominative and accusative becomes visible. The same holds for
the phrase the witch. Nouns still show genitive case marking, as in (13),
a difference which is also visible in pronouns.

(12) a. The wizard kissed the witch.

b. He kissed her.
c. The witch kissed the wizard.
d. She kissed him.

(13) a. the cat’s tail - its tail

b. the wizard’s broomstick - his broomstick

2. Distribution
Distribution considers in what environment a word can appear. As shown
in (14), nouns can be preceded by e.g. articles (14-a), possessive pronouns
(14-b), or demonstratives (14-c):

(14) a. the cat, the frog, the lynx

b. my cat, my frog, my lynx
c. this cat, this frog, this lynx

2.1.2 Adjectives (A)

The words in (15) are adjectives. We can again use the two criteria discussed
above, i.e. morphology and distribution, to show that adjectives are different
from nouns, etc.

(15) red, happy, handsome

English adjectives do not inflect for plural - as opposed to German: das alte
Haus vs. die alten Häuser - but there is morphological marking for comparative
and superlative forms. Comparative and superlative may be realised as bound
morphs (er, est), as in (16-a) and (16-b), or appear as free morphs (more, the
most), as in (16-c).

(16) a. red, redder, the reddest

b. happy, happier, the happiest
c. handsome, more handsome, the most handsome

Concerning their distribution, adjectives can be modified by a degree adverb,

as in (17-c). They can appear with a noun when they are attributive, i.e.

5
they modify the noun, as in (18-c), but they can also be predicative, i.e.,
they appear in a sentence with e.g. (copula) be, as in (19-c).

(17) a. too red

b. very happy
c. rather handsome
(18) a. the red cat
b. a (very) happy frog
c. this handsome wizard
(19) a. The cat is (too) red.
b. The frog is (very) happy.
c. The wizard is handsome.

2.1.3 Adverbs (Adv)

Adverbs are often (but not always) derived from adjectives by suffixation of
ly, as you can see in (20). They are basically invariant, but some may occur
with comparative morphology (er, est).

(20) happily, softly, carefully, quickly, mildly, fast, often

As to their distribution, like adjectives they can be modified by degree adverbs

(see (17-c) and (21-c)). Unlike adjectives, however, they cannot appear in the
environment of a noun (22-b). They can, however, appear with verbs, as in
(23-c).

(21) a. very happily

b. rather fast
c. quite cunningly
(22) a. *a happily cat
b. *the cunningly answer
(23) a. purr happily
b. run fast
c. hunt cunningly

Some types of adverbs, e.g. usually, frequently, often, can modify a whole
sentence:

(24) Usually, the Cheshire Cat purrs happily.

6
2.1.4 Verbs (V)
Some examples of verbs are given in (25).

(25) bite, sing, sleep, play, tickle

Verbs can show inflection, as in (26), where -s stands for 3rd person singular
present tense (26-a), and -ed for past tense (26-c),(26-d).

(26) a. Shawn jumps, kicks, and baas.

b. Shawn and his woolly friends jump, kick, and baa.
c. Shawn jumped, kicked, and baaed.
d. Shawn and his woolly friends jumped, kicked, and baaed.

Verbal inflection in English is very poor. In the present tense only two forms
are distinguished (sleep vs. sleeps). A comparison with German and Italian in
(27) shows that German has five (or four (e.g. machen)) different forms, and
Italian has six different forms. In the past tense English only has one form
(slept (irregular), jumped, kicked ). Verbal inflection is also visible in the suffix
– ing in present participles (sleeping, jumping, kicking).

(27) English German Italian

I sleep ich schlaf(e) dormo
you sleep du schläfst dormi
he,she,it sleeps er,sie,es schläft dorme
we sleep wir schlafen dormiamo
you sleep ihr schlaft dormite
they sleep sie schlafen dormono

Verbs can co-occur with auxiliaries (28-a) and modals (28-b), which, as we
shall see in Section 2.2.2 have to be distinguished from verbs:

(28) a. Shawn and his woolly friends have finally escaped.

b. Wallace and Gromit can celebrate now.

2.1.5 Prepositions (P)

Some examples of prepositions are given in (29).

(29) in, on, after

7
Prepositions in English are invariant, i.e. they never show any inflection.
They are usually followed by nominal constituents, as in (30-c). Some prepo-
sitions can also be followed by a sentential constituent (31-b).

(30) a. in [the cottage]

b. on [the roof]
c. after [the play]
(31) a. after [the Cheshire Cat vanished]
b. before [Humpty Dumpty wolfed down his breakfast]

Nouns, adjectives, adverbs, verbs, and prepositions are usually classified as lex-
ical categories. Typically lexical categories have rich semantic content. If you
replace ’dog’ by ’wolf’ or ’before’ by ’after’ the meaning changes drastically.
Lexical categories are open-class categories, which means that speakers regu-
larly add new members. In this respect, prepositions are atypical. Although
they have rich semantic content, it is hard to create new prepositions.

2.2 Functional Categories

In contrast to lexical categories, functional categories have little, or rather
abstract, semantic content. Functional categories are closed-class categories,
which means that no new members can be added. In English three basic
functional categories are distinguished. These are determiners, auxiliaries and
modals–later subsumed under the category Inflection (cf. Section 5.2.2)–and
complementizers.

2.2.1 Determiners (D)

Articles belong to the class of determiners. In English, the singular definite
article and the plural definite article take the same form the, as in (32-a) and
(32-b). The indefinite article (32-c) is realised in the singular by a or an,
depending on whether the following word starts with a phonetic consonant or
vowel. Is there a plural indefinite article, as for example in French (les maisons
’the houses’ vs. des maisons ’houses’)? To the extent that the information
([indefinite] and [plural]) seems to be conveyed despite the absence of an overt
determiner, it is a reasonable assumption that there is an indefinite plural
article, but that it is not phonologically realised: it is phonologically null.
Similarly, the indefinite article which precedes mass nouns in English (33) is
null:

(32) a. the cat, the ox

8
 b. the cats, the oxen
c. a cat, an ox
d.  cats,  oxen
(33)  water,  rice

We assume, for simplicity’s sake, that demonstratives and possessive pronouns

also belong to this category, since they are in complementary distribution with
the article in modern English:

(34) a. this cat, that dog

b. *this the cat, *the this cat, *that the dog, *the that dog
(35) a. my cat, his dog
b. *the my cat, *my the cat, *his the dog, *the his dog

2.2.2 Auxiliaries and Modals

The primary function of auxiliaries is to mark the temporality in a sentence.
Generally, auxiliaries cannot occur on their own: out of context, examples of
the type in (36-c) and (36-e) are ungrammatical. Auxiliaries do not have rich
semantic content; the verb which they precede contains the lexical information.
We therefore draw a formal distinction between lexical verbs, which have lexical
content, and auxiliaries, which do not.

(36) a. Humpty Dumpty eats breakfast.

b. Humpty Dumpty is eating.
c. *Humpty Dumpty is.
d. Humpty Dumpty has eaten.
e. *Humpty Dumpty has.

English also has a set of modals (e.g. can, may, must, shall ), which do not
have rich semantic content. Their role is to express the modality of the sen-
tence, i.e. necessity or possibility, and other similar concepts. Like auxiliaries,
modals need to co-occur with a lexical verb:

(37) a. The Cheshire Cat can smile.

b. *The Cheshire Cat can.
c. The Cheshire Cat should smile.
d. *The Cheshire Cat should.

In contrast to modals, auxiliaries can be inflected. There are two different

forms for have in the present tense (have, has), and three for be (am, are, is).

9
The auxiliaries have and be also have an infinitival form. The infinitival form
have follows the modal may in (38-a) and the infinitival form be follows the
modal might in (38-b) . Note that if the modals had been omitted from these
sentences, the finite forms of these auxiliaries would have surfaced, i.e. has in
(38-a) and is in (38-b).

(38) a. Humpty Dumpty may have eaten too much.

b. Alice might be upset.

Modals do not carry an –s in 3rd person singular present tense: she can/*cans,
nor do they have an infinitival form: *he should can eat more (cf. Er sollte
mehr essen können). The forms may and might, for example, do not necessarily
express a difference in tense, i.e. present vs. past tense (other modals do, cf.
can/could ), but a difference in probability. Compare (39-a) with (39-b), in
which might suggests a somewhat lower probability than may.

(39) a. He may have a tummy ache tomorrow.

b. He might have a tummy ache tomorrow.

2.2.3 Complementizers (C)

Besides determiners, auxiliaries and modals, complementizers also belong
to the class of functional elements. Complementizers are elements which in-
troduce embedded clauses, shown in (40). (The parentheses around that in
(40-a) indicate that the complementizer that is optional.)

(40) a. I believe [(that) Alice and Humpty Dumpty are good friends].
b. Alice regrets [that Humpty Dumpty eats too much].
c. She wonders [whether he will ever get tired of eating marshmal-
lows].
d. He will of course ask her again [if he can have some more].
e. [For Humpty Dumpty to go on a diet] would be quite pointless.

10
3 Phrase Structure
3.1 Constituents
The words of a language can be grouped into different categories. In a sentence,
these words combine to form larger units, called constituents. A sentence is
not simply a string of linearly ordered words: words combine to form larger
units, which in turn combine to form a sentence. Consider the sentence in
(41):

(41) Harry will meet his closest friends after dark.

There are groups of words in this sentence which seem to belong together, as
indicated by the bracketing in (42). Most of us share the intuition that the
bracketing in (42-a) is natural, whereas that in (42-b) is not.

(42) a. [Harry] [will] [meet [his closest friends] [after dark]].

b. *[Harry will] [meet his] [closest friends after] [dark].

Each of the bracketed units in (42-a) is a constituent. A constituent consists

of one or more words which behave as a single entity from a syntactic point of
view. Thus a string of words (and sometimes just a single word) forms a con-
stituent if it can be subject to syntactic manipulations as a unit, such as being
moved around, being replaced by another word, etc. These manipulations can
serve as constituency tests. Applying such test reders possible variants of a
sentence which are recognised as grammatical by native speakers. We consider
five of these tests, which can help us determine whether a given unit is indeed
a constituent.

3.1.1 Substitution
Substitution is a manipulation in which a unit is replaced by a shorter form,
called a pro-form. Units containing a noun can usually be replaced by pro-
nouns, temporal adverbials by then

(43) a. HE will meet his closest friends after dark.

b. Harry will meet THEM after dark.
c. Harry will meet his closest friends THEN.

Substitution operates on constituents. As shown in (43-b) his closest friends

can be substituted for by them, while neither his closest (44-a) nor his closest
friends after (44-b) can be substituted for by them.

11
(44) a. *Harry will meet THEM friends after dark.
b. *Harry will meet THEM dark.

Based on substitution we can conclude that his closest friends is a constituent

and so is Harry and after dark.

3.1.2 Clefting
Clefting isolates a constituent by moving it into the slot ( ) in the frame it is
that/who, which results in putting emphasis on the constituent:

(45) a. It is HARRY who will meet his closest friends after dark.
b. It is HIS CLOSEST FRIENDS that Harry will meet after dark.
c. It is AFTER DARK that Harry will meet his closest friends.

In (45) Harry, his closest friends, and after dark are again identified as
constituents, while his closest and after do not pass the clefting test:

(46) a. *It is HIS CLOSEST that Harry will meet friends after dark.
b. *It is AFTER that Harry will meet his closest friends dark.

3.1.3 Questions
A constituent can be questioned. This means that a constituent can be re-
placed by a question word, as in (47):

(47) a. WHO will meet his closest friends after dark? (who = Harry)
b. WHO(M) will Harry meet after dark? (who(m) = his closest
friends)
c. WHEN will Harry meet his closest friends? (when = after dark)

We can show that the target of the question can only be a constituent:

(48) a. *His closest WHO will Harry meet after dark?

b. *After WHEN will Harry meet his closest friends?

Once more we have identified Harry, his closest friends, and after dark as
constituents.

3.1.4 Deletion
Deletion is another constituency test. It has to be handled with care, however.
Only certain/optional units can be omitted/deleted from a sentence and can

12
thus be identified as constituents, as shown in (49). Both after dark and closest
pass this test. Note that deleting closest and after dark at the same time, as
in (49-c), does not show that closest after dark is a constituent. This might
be quite obvious, since after all closest and after dark are not adjacent to each
other; friends intervenes. In (50-a), on the other hand, nothing intervenes
between in the belfry and after dark and still in the belfry after dark is not a
constituent. Deletion cannot identify all constituents: Although we know that
his closest friends is a constituent, it does not pass the deletion test (50-b) (cf.
Section 4.2).

(49) a. Harry will meet his closest friends after dark.

b. Harry will meet his closest friends after dark.
c. Harry will meet his closest friends after dark.
(50) a. Harry will meet his closest friends in the belfry after dark.
b. *Harry will meet his closest friends in the belfry after dark.

3.1.5 Movement
A unit which forms a constituent can be moved, unless this movement violates
other constraints3 . (The symbol marks the site from which the constituent
has been moved.)

(51) a. After dark, Harry will meet his closest friends .

b. His closest friends, Harry will meet after dark.

As shown for other tests, movement only affects strings of words which form
a constituent:

(52) a. *After, Harry will meet his closest friends dark.

b. *His closest, Harry will meet friends after dark.

To summarize, the constituency tests do not apply in the same way to all
constituents. This is why they should be considered as indications rather
than as acid tests. Certain constituents cannot be identified as such in the
substitution test, for example, not because they do not form constituents,
but because there is no pro-form which could function as a proper substitute.
Although dark is nominal, the pro-form it cannot substitute for it.

(53) *Harry will meet his closest friends after it.

3
The Clefting-test in Section 3.1.2 and the Question-test in Section 3.1.3 also involve move-
ment.

13
If a string of words can be identified as a constituent by one of the constituency
tests that is enough. Although we have shown that his closest friends is a
constituent using various tests, any one test would have been sufficient.

3.2 Ambiguity
Constituency tests can be used to disambiguate structurally ambiguous
sentences. The sentence in (54) is ambiguous, since it can have more than one
interpretation. It can either mean that Hermione used binoculars to see the
witch, or that the witch herself had binoculars (and Hermione might just have
good eye-sight).

(54) Hermione saw the witch with binoculars.

Depending on how the constituent with binoculars is analysed, we either obtain

the first reading (Hermione has binoculars) or the second reading (the witch
has binoculars). Substitution, clefting, and question formation in (55-c) show
that with binoculars does not belong to the witch. The same tests in 2 show
that these two constituents together can form a larger constituent (the witch
with binoculars). We arrive at these seemingly conflicting results because the
sentence is structurally ambiguous.

(55) a. Hermione saw her with binoculars. (her = the witch)

b. It is [with binoculars] that Hermione saw the witch.
c. How did Hermione see the witch? (how = with binoculars)
(56) a. Hermione saw her. (her = the witch with binoculars)
b. It is [the witch with binoculars] that Hermione saw.
c. Who did Hermione see? (who = the witch with binoculars)

These tests show that the sentence in (54) is structurally ambiguous. None
of the words in (54) gives rise to this ambiguity, but depending on how we
organise the elements of the sentence we obtain two different interpretations.

3.3 The verbal constituent

We have seen that the subject of the sentence is a constituent, and that other
units which follow the verb are also constituents, but what about the verb
itself? Does it form a constituent? In our example (41), the verbal form is
composed of the modal will and the lexical verb meet. First we address the
question of whether the modal and the verb together form a constituent. The
answer is no. There are several tests which show that the verb and the modal

14
do not form a constituent. In fact, the modal seems to be a constituent by
itself. This can be shown by the yes/no question test. The verb does not
usually form a constituent by itself, which can be shown by pseudoclefting.

3.3.1 yes/no questions

The answer to a yes/no question is usually yes or no or maybe. Applied to
example (41), we derive the grammatical question (57-a) if will by itself is
moved to the beginning of the sentence. Moving the modal and the lexical
verb together results in an ungrammatical sentence (57-b), as does moving the
modal and the verbal unit (57-c). On the one hand, this test shows that will
meet and will meet his closest friends are not constituents, on the other, that
will should be treated as being separate from the verbal unit.

(57) a. [Will] Harry meet his closest friends after dark?

b. *[Will meet] Harry his closest friends after dark?
c. *[Will meet his closest friends] Harry after dark?

3.3.2 Pseudo-Clefting
Pseudo-clefting is a process similar to clefting. Part of the sentence (i.e. the
subject and the modal/auxiliary and the appropriate form of do) are preposed
and inserted in the frame what is [...] .4 The string of words in the frame
[...] identifies the verbal unit. This test again shows that will does not belong
to the verbal unit.

(58) a. What Harry will do after dark is [meet his closest friends].
b. What Harry will do is [meet his closest friends after dark].
c. *What Harry does is [will meet his closest friends after dark].

4
To illustrate this point consider the example in (i) containing an auxiliary rather than a
modal. The corresponding pseudo-cleft is shown in (ii). Since the auxiliary has selects
a past participle, and not an infinitive, the appropriate form of do is done.
(i) Harry has met his closest friends after dark.
(ii) What Harry has done is [meet his closest friends after dark].
In (iii) the verb meet occurs in the 3rd person singular present tense form meets. In
the corresponding pseudo-cleft in (iv) the inflected form does has to be used.
(iii) Harry meets his closest friends after dark.
(iv) What Harry does is [meet his closest friends after dark].
In all these pseudo-clefts the verbal unit is identified as [meet his closest friends after
dark]. In other words, neither the modal will, nor the auxiliary has, nor the inflection -s
are part of the verbal unit.

15
Both pseudo-clefting and the yes/no question test show that the modal and
the verb belong to two different constituents.

16
4 Argument Structure
4.1 Subcategorization
We have shown that words often combine with other words to form larger
constituents. Let us look at some of these words and at verbs in particular and
see what kinds of constraints they impose on the composition of constituents
and the sentence. Consider the examples in (59):

(59) a. Hedwig sleeps.

b. *Hedwig sleeps the bed.
c. Hedwig sleeps peacefully.
d. Hedwig sleeps peacefully at night.

The verb sleep in (59) is intransitive, which means that it cannot co-occur
with an object (59-b). On the other hand sleep can be accompanied by other
types of constituents, which tell us more about the sleeping event. Peacefully
(59-c) tells us more about the manner of sleeping and at night (59-d) tells
us more about the time of sleeping. These constituents are not obligatory,
i.e. they can be left out without rendering the sentence ungrammatical (see
(59-a)). Such optional constituents are called adjuncts.
Besides intransitive verbs there are also transitive verbs and ditransitive
verbs. Transitive verbs require one object, as shown in (60). The verb deliver
is transitive. It must be followed by a direct object (mail ). Leaving out the
object, as in (60-b) and (60-d), renders the sentence ungrammatical. The
constituent during meals is optional. Replacing the obligatory constituent
mail by the optional constituent during meals does not improve the sentence
(60-d). We say that the transitive verb deliver selects a nominal object or
subcategorizes for a nominal object. The direct object which is selected
by the transitive verb or which is subcategorized for by the transitive verb is
called a complement.

(60) a. Owls deliver mail.

b. *Owls deliver.
c. Owls deliver mail during meals.
d. *Owls deliver during meals.

Other verbs can select other types of complements. The verb live in (61)
selects a prepositional complement, the verb wonder in (62) selects a sentential
complement.

(61) a. Unicorns live in forests.

17
 b. *Unicorns live.
(62) a. Ron wonders [whether Harry will be attacked again].
b. *Ron wonders.
c. *Ron wonders [another attack on Harry].

Some verbs take different types of complements, as want in (63):

(63) a. Hermione does not want [another pet].

b. Hermione wants [her cat to recover].
c. Hermione wants [her milk hot].

There are also verbs which require two complements. These verbs are ditran-
sitive:

(64) a. Harry put his magic wand under the bed.

b. *Harry put his magic wand.
c. *Harry put under the bed.

The selectional properties of words are captured by subcategorization frames.

The subcategorization frame of some of the verbs mentioned above are shown
in (65):

(65) a. sleep, V[ ] (sleep is a verb (V), and it takes no complement)

b. deliver, V[ DP] (deliver is a verb (V), and it takes one comple-
ment, which is a nominal constituent (DP))5
c. put, V[ DP PP] (put is a verb (V), and it takes two complements:
the first complement is a nominal constituent (DP), and the sec-
ond complement is a prepositional constituent (PP))

Crucially only complements are mentioned in the subcategorization frame.

Neither the subject nor adjuncts are mentioned, since in English all verbs
co-occur with a subject and all verbs can be modified by adjuncts. What
distinguishes the different types of verbs is whether they can take one comple-
ment, two complements, or none, and which syntactic category the complement
belongs to. In Section 4.2 and Chapter 5 we will show that other lexical cate-
gories besides verbs can take complements, which implies that other categories
besides verbs have subcategorization frames.

5
The abbreviations DP, PP, etc. are discussed in Chapter 5.

18
4.2 Complements and Adjuncts
As mentioned above, a distinction must be drawn between obligatory con-
stituents and optional constituents. The subcategorization frame of a verb
lists the constituents which are selected by the verb. Each verb has its own
subcategorization frame, which is part of the lexical entry of a verb. Sen-
tences contain other constituents besides those subcategorized for by the verb.
Consider the examples in (66).

(66) a. Hermione solved the puzzle.

b. *Solved the puzzle.
c. Hermione solved the puzzle in the library.
d. *Solved the puzzle in the library.

Although the subject Hermione is not subcategorized for by the verb solve, it
is an obligatory constituent of the sentence. There is a constraint which says
that every English sentence must have a subject which is actually prnounced.
The fact that the sentences in (66) have to have a subject is not specific to
these sentences, but a general property of English sentences.
The verb solve is transitive and selects a nominal complement (the puzzle).
The constituent in the library is an adjunct; it is not selected by the verb solve.
In contrast to the nominal complement the puzzle in (67-a), in the library in
(67-b) can easily be left out without rendering the sentence ungrammatical.

(67) a. *Hermione solved.

b. Hermione solved the puzzle.

A selected constituent cannot be omitted (except under special circumstances).

The non-selected constituents which function as adjuncts can be deleted. The
main property of adjuncts is to provide information concerning time, place,
and manner. Note that the order of complements and adjuncts is strict:

(68) a. Hermione is knitting woolly bonnets near the fireplace at mid-

night.
b. *Hermione is knitting near the fireplace woolly bonnets at mid-
night.
c. *Hermione is knitting at midnight woolly bonnets near the fire-
place.
d. *Hermione is knitting near the fireplace at midnight woolly bon-
nets.

19
The noun solution is closely related to the verb solve. Although solution like
solve takes a complement, as in (69-a), it is possible to omit the complement of
solution, as in (69-b), while it is not possible to omit the complement of solve
(cf. (67-a)). The fact that the complement of the noun solution can be left
out does not mean that of the puzzle is an adjunct. In general, complements
of nouns, as opposed to those of verbs, can be left unexpressed.

(69) a. Hermione immediately came up with the solution of the puzzle.

b. Hermione immediately came up with the solution.

4.3 Thematic Roles

A verb plays an important role in the sentence, as its selectional properties
determine the minimal composition of the sentence. In addition to these se-
lectional properties, there are other constraints. Consider the examples in
(70):

(70) a. #Wallace startled the corpse.

b. #The crispy bone thanked Gromit.

From a syntactic point of view these sentences are grammatical, because the
transitive verbs startle and thank are followed by a direct object (DP). From
a semantic point of view, however, these sentences are anomalous, which is
indicated by the diacritic #.
The function of a verb is to relate ’participants’ in the sentence. These par-
ticipants are the arguments of the verb. In (70-a), for example, the verb startle
takes two arguments: the argument in the subject position, which causes the
startling, and the argument in the object position, which is being startled.
The roles are formally expressed as thematic roles, and are assigned by the
verb. The type of thematic role depends on the meaning of the verb. It is
crucial to know what the verb means in order to be able to decide what kind
of role the arguments will play. The notion of thematic role is distinct from
the syntactic notion of subcategorization. In (71-a) the transitive verb frighten
subcategorizes for a complement, which is realised by the Cheshire Cat. The
subcategorization frame of frighten in (71-b) expresses the fact that frighten
takes a DP complement, but does not say anything about the thematic role
borne by this complement. Moreover, the fact that frighten needs a subject is
not even expressed in its subcategorization frame, because this is a property
common to all verbs and therefore does not need to be explicitly mentioned
in a verb’s subcategorization frame. Thus frighten takes two arguments: the
subject argument and the object argument.

20
(71) a. The little mouse frightened the Cheshire Cat.
b. frighten, V [ DP]

We now need the process of thematic role assignment, which ensures that the
arguments which are realised are compatible with the meaning of the verb.
In other words, from a semantic point of view, a verb like frighten assigns a
thematic role which expresses the fact that someone/something (animate) is
doing the frightening, and another thematic role which expresses the object
which is frightened. These two thematic roles are assigned to the two obligatory
arguments of frighten. In example (72) the verb frighten assigns the thematic
role of agent to the argument in the subject position (the little mouse) and the
thematic role of patient to the argument in the object position (the Cheshire
Cat).

(72) The little mouse frightened the Cheshire Cat.

Verbs assign thematic roles to their arguments, that is to their selected com-
plements and the subject. Do all verbs assign thematic roles? Consider the
examples in (73):

(73) a. Hedwig snores loudly.

b. Hedwig is snoring loudly.
c. Hedwig shouldn’t snore loudly.

All the examples in (73) contain the verb snore. In (73-a), the subject is the
only argument: it is assigned a thematic role. In (73-b) the verb snore is
accompanied by an auxiliary verb, and in (73-c) by a modal. But in all these
sentences we only have one argument (Hedwig). We conclude that auxiliaries
and modals do not assign thematic roles. Remember that neither auxiliaries
nor modals have a rich semantic content. It is therefore not really surprising
that they do not assign thematic roles: they do not express any kind of action,
or state. Lexical verbs, on the other hand, do assign thematic roles.

21
5 Syntactic Representations
It should be evident by now that a sentence is not just a string of words put
next to each other: words combine to form units which in turn combine to
form larger units. In other words, there is a hierarchy in the ordering of words
within a sentence. In this chapter, we are going to examine the different levels
of hierarchy we can identify within a sentence. The difference between positions
occupied by complements and adjuncts will play a major role. Ultimately we
want to derive an abstract schema which is common to all the word categories
introduced in the previous sections. This abstract schema is called X-bar
schema, represented by means of a tree diagram in (74).

(74) XPE
yyy EEE
y EE
yy EE
yy
Specifier X’
E
yyy EEE
yy EE
y EE
yy
X Complement

5.1 Phrases
Consider example (75), which we will analyse in detail.

(75) The old wizard will concoct the poisonous potion in the classroom.

The main constituents of (75) can be bracketed as in (76):

(76) [the old wizard] [will] [concoct the poisonous potion in the classroom]

5.1.1 The Verb Phrase (VP)

We will first concentrate on the verbal constituent. As pointed out in Section
3.3, the modal does not belong to the verbal constituent (see also Section 5.2.2
below). The verbal constituent is not ’flat’ since it has a hierarchical struc-
ture because the words inside it are grouped into smaller constituents. Each
constituent contains a word which ’characterizes’ it. The verbal constituent in
(77) is characterized by the verb (V), which is the head of the constituent –
the Verb Phrase (VP). The other words are organised in sub-groups around
this head:

(77) [VP concoct [ the poisonous potion] [ in the classroom]]

22
There is a variant of the substitution test called the do-so test, which only
works for VPs, shown in (78):

(78) a. The old wizard will concoct the poisonous potion in the classroom
and Malfoy will do so too.
b. The old wizard will concoct the poisonous potion in the classroom
and Malfoy will do so at home.
c. *The old wizard will concoct the poisonous potion in the classroom
and Malfoy will do so a sleeping draught at home.

In (78-a) do so substitutes for the whole VP [concoct the poisonous potion in

the classroom], which contains the verb concoct, its complement the poisonous
potion, and an adjunct in the classroom. In (78-b) do so substitutes for the
constituent [concoct the poisonous potion], which consists of the verb and its
complement. Example (78-c) is ungrammatical: do so would have substituted
only for the verb concoct, which is not possible.
Based on the do-so test we conclude that concoct the poisonous potion in
the classroom is a constituent, and so is concoct the poisonous potion. The
latter forms a constituent called V’ (V-bar). The central elements of a VP,
i.e. the verb and its complement(s), are contained in VP. The adjunct in
the classroom is optional. It provides additional information on where the
poisonous potion will be concocted. Adjuncts are adjoined to the projection
they modify, here the place-adjunct modifies the VP and is therefore adjoined
to VP. Our tentative analysis of the VP in (77) is depicted in (79):

(79)

A VP can be modified by more than one adjunct, as in (80). Since after

the next full moon is an adjunct, it also has to be adjoined at the VP-level. It
is adjoined at a higher VP-level than the adjunct in the classroom, because it
linearly follows the adjunct in the classroom, as in (81). Note that the linear

23
order of words can be read off the tree diagram.

(80) The old wizard will concoct the poisonous potion in the classroom
after the next full moon.
(81)

To summarize so far, a VP always contains a verb, which is the head of the

VP, the verb’s complement(s) - if the verb is transitive -, and possibly one
or more adjuncts. The verb and its complement(s) form a constituent called
V’. Adjuncts are adjoined at the VP-level. By assigning different positions to
the complement(s) and adjunct(s) we capture the fact that complements and
adjuncts do not entertain the same relationship with the head: complements
are obligatory, i.e. they are selected by the head, whereas adjuncts are optional.
Besides transitive verbs there are also intransitive verbs and ditransitive
verbs, which we will turn to now. Example (82) contains an intransitive verb.

(82) Neville and Harry [VP snore at night].

Using the do-so test we can identify snore at night as a constituent in (83-a).
We can also identify snore as a constituent in (83-b):

(83) a. Neville and Harry snore at night and Ron does so too.
b. Neville and Ron snore at night and Ron does so during classes.

Since snore is intransitive it does not select a complement. Like other verbs,
snore can be modified by an adjunct, here the time-adjunct at night, which is
adjoined at the VP-level:

24
(84)

Turning to the example with the ditransitive verb put in (85) we can show
that put a crispy bone under the sofa on Sunday is a constituent (86-a), as is
put a crispy bone under the sofa (86-b). However, neither put a crispy bone
(86-c) nor put (86-d) are constituents. We conclude that a crispy bone and
under the sofa are complements of put and occupy complement positions, while
on Sunday is a time-adjunct, which is adjoined to VP. The structure of the
VP is depicted in (87).

(85) Snowy [ VP put a crispy bone under the sofa on Sunday].

(86) a. Snowy put a crispy bone under the sofa on Sunday and Gromit
did so too.
b. Snowy put a crispy bone under the sofa on Sunday and Gromit
did so on Monday.
c. *Snowy put a crispy bone under the sofa on Sunday and Gromit
did so on the kitchen table on Monday.
d. *Snowy put a crispy bone under the sofa on Sunday and Gromit
did so a dead bird on the kitchen table on Monday.
(87)

25
5.1.2 The Noun Phrase (NP)
In this section we will look at the nominal constituents in (88), which do not
have a determiner6 . The essential word in these nominal constituents is the
noun: wizards in (88-a) and teachers in (88-b). The Noun (N) is the head of
the Noun Phrase (NP).

(88) a. [ NP wise wizards]

b. [ NP old teachers of alchemy from Rome]

In the discussion of the structure of VP we distinguished between comple-

ment positions and adjunct positions and made use of the intermediate level
V’ and VP-adjoined positions. Extending this reasoning to NPs, N and its
complement(s) form an N’ (N-bar), and adjunction occurs at the NP-level.
Our ultimate goal is to show that all word categories show the structure in
(74), where X in the tree diagram can stand for V, N, P, D, etc.
Since the constituents wise and old are adjuncts in (88) they must be ad-
joined to the NP-level. The deverbal noun teachers is related to the transi-
tive verb teach, and hence teachers of alchemy are people who teach alchemy.
Since alchemy is a complement of the verb teach, of alchemy is a complement
of the noun teachers. From Rome in (88-b) can be compared to in Rome, which
in the VP teach alchemy in Rome functions as an adjunct. A crucial difference
between complements selected by nouns, as opposed to complements selected
by verbs, is that the former can be left out (see also Section 4.2). Thus omit-
ting a constituent in an NP does not tell us whether the omitted constituent
is a complement or an adjunct. The structure of the NPs in (88) is shown in
(89):

6
The strings wise wizards and old teachers of alchemy from Rome could be preceded by a
determiner, as in (i):
(i) a. [DP these [NP wise wizards]] b. [DP the [NP old teachers of alchemy from
Rome]]
The constituents in (i) are called Determiner Phrases (DP), which will be discussed in
Section 5.2.1. In fact, wise wizards and old teachers of alchemy from Romeare also DPs,
in which the Determiner, which heads the DP, is nonovert. Remember that English does
not have an indefinite plural article, as opposed to French des, for example.

26
(89)

5.1.3 The Adjective Phrase (AP)

By using the deletion test (see Section 3.1.4) we can show that old and wise
in (88) are constituents. The head of these constituents is the adjective (A),
which heads the maximal projection of the same name: Adjective Phrase or
Adjectival Phrase (AP). Pursuing our line of reasoning we want to establish
whether APs have the same type of hierarchical structure as VPs and NPs.
The answer is yes. Some adjectives select a complement, as in (90). Note that
the adjective envious is related to the transitive verb envy. The sentence Ron
is envious of Harry can be compared to Ron envies Harry. We conclude that of
Harry is a complement of envious. There is no verb which is directly related
to the adjective proud. Still, of his pet dragon functions as a complement
of proud. Although it is possible to leave out the complements in (90), the
meaning of the adjective changes slightly.

(90) a. Ron is [ AP envious of Harry].

b. Hagrid is [ AP proud of his pet dragon].

We assign the following representations to the APs:

(91)

27
5.1.4 The Prepositional Phrase (PP)
The structure applied to VP, NP, and AP can also be extended to Preposi-
tional Phrases (PPs), which are headed by prepositions (Ps), as in (92):

(92) a. Harry, Hermione, and Ron are hiding [ PP in Hagrid’s cottage].

b. Hagrid is boiling tea [ PP over an open fire].

The phrase Hagrid’s cottage in (92-a) is a complement of the preposition in,

and an open fire in (92-b) is a complement of the preposition over, depicted
in (93):

(93)

5.1.5 The X-Bar Schema

All the constituents we have examined so far are organised in the same way.
They include a head X, which may or may not select a complement.7 The
complement and the head form a sub-constituent, called X-bar. An element
which we have not introduced yet (see Sections 5.2.2 and 5.2.6) – the Specifier
– and X’ project up to XP, the maximal projection (96-a). Adjuncts are
taken to be adjoined to XP (96-b).

(94)

7
The variable X can stand for V, N, A, etc.

28
5.2 Functional Categories
5.2.1 The Determiner Phrase (DP)
The determiner (D) is not part of the NP. However, a special relationship holds
between the determiner and the noun heading the NP. The determiner and the
noun have to agree in number (singular vs. plural) in English. The determiner
this is singular and the noun heading the NP must be singular as well (95-a).
The determiner these is plural and the noun heading the NP must be plural as
well (95-b). A reasonable assumption is that the determiner – the head of the
Determiner Phrase (DP) – selects the maximal projection of N, namely
NP, as its complement. The detailed structure of the DP these tiny dragons
and the DP Hagrid is given in (96). It is customary not to show the detailed
structure of a constituent if it only contains a head (and no complement(s) nor
any adjuncts), as in (96-b’) (and also (96-a)), where the triangle means that
no detailed structure is shown.

(95) a. Hagrid admires this tiny dragon / *this tiny dragons.

b. Hagrid admires *these tiny dragon / these tiny dragons.
(96)

5.2.2 The Inflectional Phrase (IP)

As shown in the previous section the X-bar schema can also be applied to
functional projections (DPs). Let us now examine whether the X-bar schema
can be extended to the sentence itself.
Consider example (97):

(97) The old witch might buy a new broomstick.

The subject DP the old witch can be identified as a constituent by e.g. substi-

29
tution (she) and the VP buy a new broomstick by e.g. pseudo-clefting (What
the old witch may do is [buy a new broomstick]). The modal might is not
contained in the VP. This can be shown by the yes/no question test, in which
might is moved to the beginning of the sentence, and the VP itself does not
move:

(98) a. Might the old witch buy a new broomstick?

b. *Might buy a new broomstick the old witch?

We observed in Section 5.1 that the head X of a maximal projection XP consists

of a single word. In example (97) the head of the subject NP is witch and the
head of the VP is buy. Since both witch and buy head their own projection
they cannot possibly be the head of the sentence. On the other hand, might is
also a head, but has not yet been associated with a maximal projection. We
therefore consider that might is the head of the sentence. Before revealing what
kind of functional element might is, we will discuss other functional elements
which behave in the same way.
If the yes/no question test is applied to the examples in (99), the auxiliary
has is fronted in (100-a), and the inflection -s supported by the auxiliary verb
do in (100-b). Fronting the inflection –s by itself leads to ungrammaticality
(100-c), as does fronting of the inflected lexical verb buys (100-d). The former
is due to a morphological constraint: inflectional morphs are affixes, which
must be bound and can therefore not survive on their own. The latter is due
to a peculiarity of present-day English. In Old English and Middle English
it was possible to front lexical verbs in questions, a possibility still allowed in
Shakespearian times. In present-day English lexical verbs can no longer be
fronted (except for possessive have (in British English) and copula be).

(99) a. The old witch has bought a new broomstick.

b. The old witch buys a new broomstick.
(100) a. Has the old witch bought a new broomstick?
b. Does the old witch buy a new broomstick?
c. *S the old witch buy a new broomstick?
d. *Buys the old witch a new broomstick?

Without entering the intricacies of this hypothesis, the Inflection (I) is taken to
be the head of the sentence (IP). In sentences without modals and auxiliaries
inflection is marked on the lexical verb. However, both pseudo-clefting and
yes/no questions show that the inflection is not part of the VP. The same holds
for auxiliaries, which can carry inflection. Although modals do not inflect,
they behave like auxiliaries and are therefore treated in the same way, i.e. as

30
elements in I.
The phrase marker of example (97) is given in (101). The head of the
sentence – the Inflectional Phrase (IP) – is I, which is occupied by the
modal might. I selects VP. VP and I project up to the intermediate level
I’ (I-bar). The subject, a DP, sits in the specifier of IP (Spec,IP). One of
the differences between a specifier position and an adjoined position is that
there is only one specifier position in a maximal projection, while there is no
limit on the number of adjoined positions. Thus there is only one [Spec,IP] in
a (simple) sentence.

(101)

5.2.3 Embedded sentences

So far we have only looked at simple sentences. Complex sentences consist of
a matrix clause and one or more embedded clauses. The embedded clauses in
(102) are introduced by a complementizer. A complementizer is a functional
element in C. (C used to be called ’comp’, which is short for ’complementizer’).
The complementizer is the head of the complemenizer phrase (CP) and
selects a clause.

(102) a. Wallace is wondering [ CP if Gromit has read this novel].

31
 b. Gromit may believe [ CP that a lost sheep should feel unhappy].

The phrase markers for these sentences are shown in (103).

(103)

32
 Since complex sentences consist of more than one clause there is more than
one Spec,IP position, i.e. one per clause (IP).

5.2.4 Levels of Representation

As you may have noticed, all the sentences discussed so far either contained a
modal or an auxiliary. This was done on purpose to avoid the problem posed
by inflected lexical verbs. We will turn to this problem now. Consider the
examples in (104), which contain the inflected verb forms smiles and smiled :

(104) a. The Cheshire Cat smiles contentedly.

b. The Cheshire Cat smiled contentedly.

In these examples, the inflectional affixes –s and –ed are bound to the verbal
stem smile. If we form yes/no questions based on these sentences we obtain the
sentences in (105), in which the inflectional affixes are no longer attached to
the verbal stem smile, but to the stem of do. Since inflectional affixes must be
bound, the dummy verb do is used to host them. Note that do does not have
any semantic meaning in these sentences, and that is why it is often referred
to as a dummy verb.

(105) a. Does the Cheshire Cat smile contentedly?

b. Did the Cheshire Cat smile contentedly?
(106) a. *Smiles the Cheshire Cat contentedly?
b. *Smiled the Cheshire Cat contentedly?

In fact, the dummy verb do is also used in other contexts, e.g. with sentential
negation (i.e. not) unless a modal or auxiliary is present:

(107) a. Gromit doesn’t read detective stories.

b. *Gromit reads not detective stories.
c. *Gromit doesn’t may read detective stories.
d. Gromit may not read detective stories.
e. *Gromit doesn’t have read any detective stories.
f. Gromit hasn’t read any detective stories.

Modals and auxiliaries behave alike: they can be fronted in questions and they
are incompatible with do. Lexical verbs, on the other hand, cannot be fronted
in questions and cannot precede negation. They require the presence of do
in these contexts, which is sometimes referred to as do-support. Do–like
modals and auxiliaries can be fronted in questions and can precede sentential
negation.

33
 The fact that the inflection surfaces on the lexical verb in (104), but on
do in (105) and (107-a), suggests that the inflection itself is dissociated from
the lexical verb at some abstract level. In the theory of generative syntax we
embrace here, (at least) two levels of representation are distinguished: one
level, called Deep Structure, and another level, called Surface Structure,
which more or less coincides with the actual sentence we perceive (in written
or spoken form).
Let us examine the sentences in (108), in which the inflection occurs on the
auxiliary (108-a), and on the lexical verb (108-b):

(108) a. Hagrid has freed his pet dragon.

b. Hagrid freed his pet dragon.

D-Structure encodes information relating to lexical and functional properties,

as e.g. that free is a verb, which selects a DP-complement, or that there is
past tense inflection –ed. Since free is a verb, it appears under the V-node,
and –ed being inflectional, appears under the I-node. The D-Structure of the
sentence in (108-b) can be diagrammed as in (109):

(109) Deep Strucuture

The Deep Structure ’Hagrid –ed free his pet dragon’ does not correspond
to an actual sentence. At Surface Structure the words are arranged so they
correspond to an actual sentence. Thus the inflectional affix –ed must appear
bound to a verbal stem. In theory, –ed and free could combine by either
raising the verb free up to –ed in I, or by lowering –ed onto the verb in V.
To determine which option is actually chosen in English, we can look at the

34
position of adverbs, modifying the VP, such as quickly.

(110) a. Hagrid quickly freed his pet dragon.

b. *Hagrid freed quickly his pet dragon.

The adjunct quickly modifies the freeing event and is therefore adjoined at
the VP-level. Since the inflected verb freed linearly follows the adverb in the
grammatical example in (110-a), we conclude that the inflection is lowered onto
V, which is called affix-hopping. Affix-hopping is a type of movement.
A trace (abbreviated as ’t’) under I indicates the original position of the
inflection, i.e. the position occupied by the inflection at Deep Structure. The
trace t and the moved element –ed are coindexed. Note that affix-hopping
involves movement from a head-position (I) to another head position (V).

(111) Surface Strucuture

In example (112-a) the modal will precedes the manner-adverbial adamantly,

and in (112-b) the auxiliary has precedes the manner-adverbial slowly. Manner-
adverbials are adjuncts which modify VP and which are therefore adjoined to
VP. Since both the auxiliary in (112-a) and the modal in (112-b) precede the
manner-adverbial, we conclude that they occupy I. Since no movement has

35
occurred, the D-Structure and the S-Structure of these sentences coincide.8

(112) a. They will adamantly guard the secret.

b. Harry may slowly regain consciousness.
(113) a. IP
E
yyy EEE
y EE
yy EE
yy
DP I’E

444 yyy EEE

444 yy
y EE
EE

yy
They I VP
yEE
yyy EEE
yy EE
yy E
will AdvP E VP
yyy EEE
y EE
yy EE
yy
adamantly V’
yEE
yy EEE
y
yy EE
yy E
V DP

guard D’
E
yy EEE
yyy EE
y EE
yy
D NP 6
 666
 66

the secret

8
From now on, we will use tree diagrams in which the words are not on one line. They may
be more difficult to read but nevertheless you should try to become acquainted with the
diagram style introduced in (113-a)

36
 b.

IP
yEE
yy EEE
y
yy EE
yy E
DP
7 I’E
 77 yyy EEE
 777 yy
y EE
EE
 7 yy
Harry I VPE
yyy EEE
y y EE
y EE
yy
may AdvP VP
777
 77
 77

slowly V’
E
yyy EEE
y EE
yy EE
yy
V DPI
uuu III
uu II
uu II
uu I
regain consciousness
To summarize, in this section two levels of representation were introduced:
D-Structure and S-Structure. D-Structure is an abstract level encoding in-
formation relating to selectional and functional properties of words. Between
D-Structure and S-Structure movement can take place, which may influence
the order of words. Affix-hopping is a type of movement which does not change
the linear order of words. A case where movement does have an impact on
word order is discussed in the next section.

5.2.5 Movement in yes/no Questions

The two sentences in (114) are clearly related to each other: one is in the
declarative form, the other in the interrogative form. The D-Structure and the
S-Structure of example (114-a) coincide. In other words, no movement has ap-
plied (see (113-a) above). There is a difference in meaning between (114-a) and
(114-b) – statement vs. question – which is due to a difference in word order
(and intonation). In (114-a) the modal will follows the subject DP Gromit’s
friends, whereas in (114-b) the modal will precedes it. Since will surfaces
in different positions in these two sentences, we conclude that movement has
taken place in (114-b). The D-Structure of both these examples is the same,
shown in (115). But while no movement applies in (114-a) – the D-Structure

37
and S-Structure coincide – there is movement of will before S-Structure in
(114-b), resulting in a different word order. What kind of position could will
have moved to? Recall from 5.2.3 that there is a higher projection than IP,
namely CP, which we needed to account for embedded sentences: embedded
sentences are CPs introduced by a complementizer in C. A complementizer
linearly precedes the subject, and so does will in (114-b). We deduce that
will occupies the same position as the complementizer, i.e. C, and that is has
moved from I to C (from a head-position to another head-position). The trace
in I indicates the original position of will and is co-indexed with will in C.

(114) a. Gromit’s friends will love the presents.

b. Will Gromit’s friends love the presents?
(115) a. D-Structure of (114-a) and (114-b):
IP
R
yy RRRRRR
yyy RRR
y RRR
yy
DP I’E
yyy EEE
y EE
yy EE
yy
D’
E I VP
yyy EEE
y EE
yy EE
yy
D NP will V’
::: yyEEE
 ::: yy EE
 : yy EE
 yy E
Gromit’s friends V DP

love D’E
yy EEE
yyy EE
y EE
yy
D NP
> >>
>>
>>
the presents

38
 b. S-Structure of (114-b):
CP

C’
yRRRRR
yyy RRR
yy RRR
yy RR
C IP
y yRRRRR
yy RRR
yy RRR
yy RR
Willi DP I’E
yyy EEE
y EE
yy EE
yy
D’
E I VP
yy EEE
yyy EE
y EE
yy
D NP ti V’
::: yEE

 :: yy EEE
y
 :: yy EE
 yy E
Gromit’s friends V DP

love D’
yEE
yy EEE
y
yy EE
yy E
D NP
> >>
>>
>>
the presents

5.2.6 Movement in Wh-Questions

We showed that yes/no questions involve movement of a modal (or auxiliary)
from I to C. There are sentences where more than one movement has taken
place, as e.g. in wh-questions. In the wh-question in (116) two movements
have occurred: will has moved from I to C, and the wh-constituent what,
which linearly precedes the modal will in C, has moved to the specifier of
CP. The latter involves movement of a maximal projection (DP) to a position
specified for maximal projections (Spec,CP).

(116) What will Gromit’s friends love?

39
The D-Structure of (116) is shown in (117-a); the S-Structure in (117-b). We
co-index the moved elements with their respective traces. Since there are two
movements we choose two different indices.

(117) a. D-Structure:
IPR
yy RRRRRR
yyy RRR
y RRR
yy
DP I’E
yyy EEE
y EE
yy EE
yy
D’
E I VP
yy EEE
yyy EE
y EE
yy
D NP will V’
::: yEE

 :: yy EEE
y
 :: yy EE
 yy E
Gromit’s friends V DP
3 3
 333
 3
love what
b. S-Structure:
CP
E
yy EEE
yyy EE
y EE
yy
DP 8 C’R
 888 yyy RRRRRR
88 y RRR
 8 yyyy RRR

Whatk C IP
y yRRRRR
yy RRR
yy RRR
yy RR
Willi DP I’E
yyy EEE
y EE
yy EE
yy
D’
E I VP
yy EEE
yyy EE
y EE
yy
D NP ti V’
::: E
yy EEE
 :: yy EE
 :: yy EE
 yy
Gromit’s friends V DP

love tk

40
Let us briefly summarize this chapter:

• Except for affix-hopping, the Surface Structure and the Deep Structure
of declarative sentences in English is the same. Affix-hopping is a type
of movement which involves lowering the inflection in I onto the verb in
V. The lexical verb does not move.

• Interrogative sentences have the same word order as declarative sentences

at Deep Structure. The surface order of interrogative sentences is derived
by movement. In yes/no questions, the modal (or auxiliary) in I is moved
to C. Thus it ends up in a position where it linearly precedes the subject.
Wh-questions involve two movements: movement of the modal (or aux-
iliary) from I to C, and movement of the wh-constituent into Spec,CP.
Both the wh-constituent and the preposed modal (or auxiliary) precede
the subject.

• In this system heads move to head-positions (I-to-V, and I-to-C) and

maximal projections move to positions reserved for maximal projections
(e.g. wh-constituents move to the Specifier of CP). It is conventional to
co-index the moved constituents with their respective traces.

6 References
Haegeman, Liliane (2006), Thinking syntactically: A guide to argumentation
and analysis. Oxford: Blackwell.
Haegeman, Liliane and J. Guéron (1999), English grammar: A generative
perspective. Oxford: Blackwell.
Haegeman, Liliane (1994), Introduction to government and binding theory
(2nd edition). Oxford: Blackwell.
Ouhalla, Jamal (1994), Introducing transformational grammar: From prin-
ciples and parameters to minimalism. London: Arnold.
Puskás, Genoveva (2002), English linguistics 1rst year handout: Introduction
to syntax. Ms., University of Geneva.
Roberts, Ian (1997), Comparative syntax. London: Arnold.

41