Principles & Parameters Theory - Towards A Contrastive Syntax of English and Arabic-Compressed

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 285

PRINCIPLES AND PARAMETERS THEORY:

TOWARDS A CONTRASTIVE SYNTAX OF ENGLISH AND


ARABIC

Mohamed Khalil Ennassiri

2015
Titre de l'ouvrage: Principles and Parameters
Towards contrastive syntax of English and
a
I'heory:
Arabic

Auteur: Mohamed Khalil Ennassiri

Mention d'édition: 1ere Editioon

Imprimeur: Imprimerie Al-Khalij Al Arabi


152 Av. Hassan II- Tétouan

Tél. 0539710225

N° du dépôt légal: 2015M03719

ISBN: 978-9954-36-412-3
Snmemory ofa goodfpiend and eolleague,
Shdernaggak sshir
TABLE O F CONTENTS

*** Vii
Foreword....

Preface X

A list of phonetic symbols. ****. Xii

A list of abbreviations . .

Chapter 1: Basic Assumptions... *e*****

1.1.Defining syntax ..|

1.2.The notion of sentence as the basic unit of syntax.. . 2

1.3.Grammaticalknowledge
1.4.Language and the problem of knowledge
1.5.Levels of representation .. 13
1.6.Aspects of the lexicon... .. 15
1.7.Modules of Principles and Parameters model .. 21

Chapter 2: X-Bar Theory and Phrase Structure .. 23

2.1.Inadequacies of phrase structure rules 23


2.2.Principles of X-bar theory. 25
2.3.Extending the X-bar format to other categories .33

24.A sketch of Arabic phrase structure 39


2.4.1 The Determiner Phrase in Arabic .
..39

2.4.2 The VP-shell hypothesis .


.. 58

2.5.Clausal Categories 62

Exercises ..
72

Chapter 3: Case Theory ... 73


*****e**s***

3.1.Overt Case vs. abstract Case 73


32.Accusative Case in English 75
3.3. Structural conditions on Case assignment in English ... 77
3.3.1 Adjacencyrequirement.. 78
3.4.Nominative Case in English.... . 81
82
3.5.Exceptional Case Marking (ECM) constructions. .

3.6.Case assignment in Arabic .... 84


90
Exercises....
Chapter 4: Theta Theory 91

4.1.Lexical entries and argument structure .91


93
4.2.Types ofsemantic/thematic roles.. .99
4.3.Grammatical functions: A-positions and A'-positions
.101
4.4.Theta-positions vs. theta-bar positions .

102
Exercises

Chapter 5: Head Movement 104

5.1.Head movement in English. 105


in Arabic 108
5.2.Head movement . .

5.2.1. V-raising Vs. T-lowering.. .. 110


I18
5.2.2. V-movement and Double Object Constructions...
. 131
Exercises

Chapter 6: NP-Movement 132

6.1.Passive structures .
... 132
138
6.2.Raising constructions
6.3.Passivization and raising interaction 140
constructions in Arabic 142
6.4.Raising ... 151
6.5.Topic structures: A cartographic approach
.
164
Exercises

. 165
Chapter 7: Wh-Movement

Non-subject extraction in English and Arabic..166


7.1.Root questions:
7.2.Root questions: Subject extraction in English and rabic
.. I76
7.3.Embedded questions
. . 178
7.3.1. Non subject extraction
out
of complement clauses in
English and Arabic. .
Subject extraction out of complement clauses in
.
184
7.3.2
English and the That-trace Effect
7.3.3. Subject extraction out of complement clauses8
i 6
Arabic ..190
7.4.Standard Arabic vs. Moroccan Arabic: A microparametric
variation? 197
7.5.Argument-adjunct asymmetry 200
Exercises 203

Chapter 8: Binding Theory . . ***o204

8.1.The interpretation of overt nominal expressions in English ...205


8.1.1. Anaphors, pronominals and Binding Conditions... 205
8.1.2. R-expressions and Binding Theory... .216
8.2.The distribution of overt DPs in Arabic. 219
8.2.1. The interpretation of anaphors and pronominals....219
8.2.2 The interpretation of r-expressions.. 225 **

Exercises . .
228

Chapter 9: Control Theory ... ********sce


2 ose2
s 9

9.1.The distribution of PRO in English 229

9.2.The nature of .. 230


PRO...
9.3.Control structures in Arabic.. 237

9.3.1. PRO vs. ..237

pro
9,5.2. The structure of obiect control constructiond****255
243

Exercises

Bibliography . 256
FOREWORD

Noam Chomsky's Principles and Parameters Theory has generated


considerable interest, but the technical nature of his novel description
of language has discouraged many linguists from investigating it in

any thorough way. Ennassiri's book was written to make it easier for
students and those curious about Chomsky's Principles and

Parameters to understand this interesting linguistic theory.

This introduction to Principles and Parameters Theory was

originally a set of courses taught by the author to undergraduate and


graduate students of linguistics in Morocco and Saudi Arabia. It

serves as a background and further reading text for the syntax course.
It focuses on the structure and linguistic function of basic and
embedded sentences, phrase structures, case assignment, argument

structure, semantic and thematic rules, head movement- particularly


V-Movement- NP-Movement, Wh-Movement, nominal expressions,
anaphors, and pronominals, raising and control, among other aspects

of Arabic and English syntax.

It is primarily addressed to anglophone students, drawing on

their experience of English and Arabic. It contains extensive

illustrations from both languages, with plenty of emphasis on the

syntactic similarities and differences between them. At the end of each


chapter, there are exercises in which again English and Arabic figure.
These exercises are meant to help students deepen their knowledge of
the theory concerned in a comparat1ve perspective.
book 1s not ntended to
course
Though this a

analysis of any
One structure or
individual
systematic in-depth
basic material to serve as the co
it contains sutficent
language,
advanced language-specilic
studies in syntax
point for subsequent

may be unfamiliar to tho


A number of technical expressions
that they are used in Principles and
reader, especially in the sense

Parameters Theory. However, Ennassiri defines these expressions


when necessary for the understanding of the underlying theory. He
also illustrates his book with many linguistic examples from Arabic

English, and other languages.

This textbook grew out of Ennassiri's efforts to develop


teaching material for the undergraduate and graduate levels

Introduction to Syntax course that he has taught at Abdelmalek


Essaadi University in Tetouan and at Dammam University in Saudi
Arabia for over twenty years. He used earlier versions of this text for

several years; the final version was revised and expanded many times
to satisfy the needs of students, whose feedback has been of immense
value in shaping the present book.
I feel it is particularly well suited to general readers or os
who work in
disciplines related to linguistics, such as
psycno
philosophy, mathematics, or computer science. All that
an
requ
is
interest in
rigorous approaches to the analysis of the gramil
atical

structure of natural
languages. Ennassiri has tried to strike a
nce

between linguistic analysis (centered a


of
on the developine
contrastive analysis of the syntax of Arabic and English) and data-
oriented problem solving. In addition, he has tried to place the
proposals presented here into historical and comparative perspectives.

Grounded in linguistic research and argumentation, Principles


and Parameters 7heory is written to help students and readers become

independent language analysts capable of critically evaluating claims

about the language structure and meaning. Written in a clear style, it

guides its readers on topics including basic assumptions about Arabic


and English syntax, clause elements and patterns, how clauses may be

combined into sentences, and how clauses and sentences are modified

in neutral language settings.

The book avoids presenting the language as a set of arbitrary

its conclusions in the analytic generative methods


facts by grounding
that have characterized the best grammatical and linguistic practices
based on
for the last five decades. Although its perspectives
are

generative grammar, specifically principles and parameters theory, its


its readers with a broad spectrum of fundamental
goal is to provide
Each chapter contains a
knowledge about English and Arabic syntax.
numerous practical exercises.
of terms, a list of readings, and
glossary

Moha Ennaji, PhD

Professor of Linguistics
International Consullant

IniernationalInstitute for Languages and


President and Co-founder of
Culures (INLAC)
PREFACE

book for
This book
is intended as a
course
ediate
of the material herein is bace
A large porton on
svntas course.
for several years
lectures that I
have prepared
and taught in my
contrastive inguistics at the Faculty of
classes on syntax and
Morocco, and the Faculty of LetterS and
Letters in Tétouan,
Humanities in Dammam, Saudi Arabia. Although the book is
technical in nature, it does not necessarily require the reader to

have a deep knowledge of formalism. However, some

background knowledge of syntax is required in order to


appreciate the line of reasoning pursucd in it. The book has been
designed primarily as a textbook for graduate and advanced
undergraduate students of syntax in Morocco, but it also targets
a wider readership in the field of contrastive syntax.

As the title makes it clear, the aim of the book is twofold.


First, it attempts to provide formal
a
description of English and
Arabic phrase architecture, based on the
parameters (P&P) model.
principles-and-
Second, it attempts to show
fixing a finite array of ho
options (parameters) assoc1atcu
universal
principles yields *
rich and diverse
Cxpressions in
English
linguisstic
and Arabic. This will,
Detter hoperu
appreciation of the extent to which tne
graminars of these two
angages may be typologically and genetically unrelated
compared
with Chomshy's (1995: and
contrasted. This is im eeping
6) contention
o) contention that"luneuose ditfereuces
tha
and typology should be reducible to choice of values of
paramelers.

The book consists of nine chapters. The first chapter is a


sort of general introduction to the book. The other chapters each
deals with a component of Principles and Parameters Theory

and outlines its theoretical underpinnings. The sets of exercises


at the end of each chapter are intended to allow the readers to

further deepen their appreciation of the theory in a more

practical perspective.
A very special thank you goes to Professor Moha Ennaji
for kindly agreeing to write a foreword to this book. Thank you
words also go to my students, both at Abdelmalek Essaadi

University in Tétouan, Morocco, and King Faisal University in


the years in
Dammam, Saudi Arabia, for their patience over

dealing with a number of half-baked ideas concerning the syntax

of Arabic. I very much hope that this version of the book will

and equip them with a tool to


compensate for their patience
explorations in contrastive syntax.
pursue further

Mohamed Khalil Ennassiri

October, 2015

Tétouan, Morocco
SYMBOLS
PHONETIC

OF
A LIST

Consonants

glottal stop (¢)


voiced-palato-alveolar fiicative (E)
J

voiced pharyngeal fricative ()


H
voiceless uvular fricative ()

voiced interdental fricative (5)


dh
voiced alveolar trill ()

sh voiceless palato-alveolar fricative ( )

S voiceless pharyngo-alveolar fricative ( )

D voiced pharyngo-dental stop (J)


T voiceless pharyngo-dental stop (>)
Dh voiced pharyngo-interdental fricative ()
voiced pharyngeal fricative (E)
R voiced uvular fricative (E)

voiceless uvular stop (3)


voiceless laryngeal fricative
(^)
labio-velar semi-vowel (3)
palatal semi-vowel ()

Other consonants are


Double consonants represented by conventional letters.
representstress.
Vowels
short. open vowel

long. open vowel

short, close, back, rounded vowel


long, close, back, rounded vowel
i short, close, front, unrounded vowel
i: long, close, front, unrounded vowel

xii
A B B R E V I A T I O N S

OF
A LIST

1-first person
2- second person

3= third person

A= Adjective
Acc= accusative Case

AP Adjective Phrase

Arb arbitrary reference

Det- Deteminer

DP= Determiner Phrase

ECP Empty Category Principle

EPP- Extended Projection Principle


F-feminine
Gen- genitive Case

LF= Logical Form

M= masculine

N= Noun

Nom= Nominative
NP= Noun Phrase

P-plural
P&P-
Principles and Parameters model
PF-Phonetic Form
PLD- Primary Linguistic Data
PP= Prepositional Phrase

RM- Relativized Minimality

S- singular

Subj= subjunctive mood


UG Universal Grammar

V- Verb

VP= Verb Phrase

A daslh is used to mark the morpheme boundaries if the


morpheme is a prefix or a sufiix.
aims to reduce descTiptive statements to
The P&P approach
and language particular
two categories: language-invariant,
staternents are principles (including
The language-invariant
on a par with a principle of UG): the
the paramneters, each
language-particular ones specifications of particular
are

values of parameters. The notion ol construction, in the


traditional sense, effectively disappears; it is perhaps useful
for descriptive taxonomy but has no theoretical status. [...1
The parametric options available appear to be qujte
restricted. An assumption that seems not unrealistic is that
there is only one computational system that forms
derivations from D-Structure to LF; at some point in the
derivation (S-Structure), the process branches to form PF
by an independent phonological derivation. Options would
then be restricted to two cases: (1) properties of the lexicon,
or (2) the point in the derivation from D-Structure to LF at
which structures are mapped to PF.

-Noam Chomsky

The Minimalist
Program, pp. 25-26
I. BASIC ASSUMPTIONS

This chapter will discuss the following topics

Defining syntax
The notion of sentence as the basic unit of syntax

Grammatical knowledge
The lexicon

Language and the problem of knowledge


Levels of syntactic representation

1.1 Defining syntax

that specifies how words can


Syntax is a branch of linguistics
form sentences. It also
be combined to fom phrases and phrases to

structure to all well-formed sentences.


assigns an appropriate syntactic
restrictions on the order of words
Different languages impose specitic
in sentences. Consider the following examples:

() a. John ate an apple. (English SVO)

b. Pierre a mangé une pomme. (French SVO)

Pierre has eaten an apple

"Peter has eaten an apple.

(2) Watshi ga hako o akemasu (Japanese SOV)

box open

are
and processes by which sentences
Syntax is the study of the principles Structures.
Noam Chomsky, Syntactic
Constructed in particular languages."
box."
"I open the

kita:ban (Arabic/ VSO)


shtara: zaydun
bought Zayd book

book."
"Zayd has bought a

(4) Nahita ny mpianatra ny vehivavay. (Malagasy/ VOS)

saw the student the woman

"The woman saw the student."

5) Dieses Buch kaufte Karl gestern. (German/OVS)


this book bought Karl yesterday
"Charles bought this book yesterday."

1.2 The notion of sentence


as the basic unit of
syntax
The unit of
analysis in syntax is the sentence. But what is a

Sentence We
simplistically define a sentence as a s
can

Sequence of words. However, not


all strings of words form ct
Sentences. In this
co
connection, consider the following examples.

There
different types of sentences in glish and crosslinguistically, of course
are
(ia) below is
declarative, (ib)
English and cro (id) is
exclamative: is
interrogative, (ic) is imperative and
() a. He fed the
cat.
b. Have
you seen John?
c. Put the milk
d. the table.
on
What a day we
had!
(6) a. John frightened Paul.

b. Paul frightened John.

c.
Frightened John Paul.
(6a&b) are correct:
by contrast, (6c) is ungrammatical (1.c. (i.c.
syntactically deviant, ill-formed, unsyntactic or ruled out) because it
doesn't conform to the syntactic rules of English. It follows, then, that
the meaning of sentence is
a
computed from the meaning of the words
of which it is composed and (more importantly, perhaps) from the
arrangement/ linearization of these words within the sentence. Even in
languages where the order of major constituents is comparatively
freer, the merger of words is wild. For the sake of
not
exposition,
consider the following examples in Arabic, a language that exhibits
much word order freedom:

(7) a. shariba 1-?awla:du llabana (VSs0)


drank the-children the-milk

The children have drunk milk."

b. llabana shariba l-?awla:du (OVS)

the-milk drank the-children

It's milk that the children have drunk."

c. -7awla:du sharibu: llabana (SVO)

the-children drank the-milk

By convention, an ungrammatical sentence is preceded by an asterisk (*).


have drunk milk."
"The children

llabana -?awla:du (VOS)


d. shariba
drank the-milk the-children

It's the children who


have drunk milk."

e. 1-Pawla:du llabna sharibu: (SOV)


the-children the-milk drank

f.1labana1-Pawla:du sharibu: (OSV)

the-milk the-children drank

(7e&f) are at best marginal in Arabic, even under a proposed

topicalization analysis. Consider now the following examplesn


English and Arabic, respectively:

(8) a. Darkness frightens Jane.

b. Jane frightens darkness.


(9) a. EaDDa l-kalbu l-walada

bit the-dog the-boy

"The dog bit the


boy."
b. EaDDa l-waladu l-kalba
bit the-boy the-dog

The boy bit the dog."


Sentenees (8b) and (9b) are grammatical, i.e. they are

syntactically well-formed, but they are not acceptable because they are
not consonant with our
conception of the world. Thus, grammaticality
has more to do with whether sentence is
a
'properly put together than
with whether it is
meaningful or true.

1.3 Grammatical knowledge

Native speakers of a language L know the grammar of that


language. For example, they know- among other things- which
sentences are
grammatical and which are not. Any native speaker of
English would be able to tell you that (6c) is bad in this language
because in that particular arrangement, the words in it are not well-

formed. Grammatical knowledge also accounts for structural


ambiguity, i.e. ambiguity due to syntactic structure. An ambiguous
sentence is a sentence that has two meanings. Some sentences are

ambiguous when they contain an ambiguous word. This is a case of

lexical ambiguiy. By contrast, some sentences are ambiguous without

containing any ambiguous words. This is a case of structural

ambiguity, e.g. (10a-b):


(10) a. [Old men and women] were watching the match.

b. Jane hit the thief with the stick.

The bracketed subject NP in (10a) has two different structures

associated with two different interpretations, viz. (1 1a) and (1 lb):


P r i n c i p l e sa n d t

where the
(where adject
di
old
men)
and
women

does tus
a. [old
(11)
quality women)

men and
both men.
both
womenJ
(where womCN are
old
b. old [ menand
name

has two
dillerent
inlerprelations,
(12a) and
Similarly. (10b)

(126):

stick to hit
the thief.
the
a. Jane used
(12)
(12)
holding the stick.
the thief who
was

b. Jane hit

are also
able to determine that ttwo
.

language
Native speakers of a

each other. For instance, native speakere


sentences are
paraphrases of
about the same meaning. Similarly
know that (13a&b) have
of Arabic
know that (14a&b) are
speakers of English
virtually all native
different:
though their forms
are
paraphrases of each other,
even

SSa:khibat-i
la: ya-ru:qu li:
sama:E-u 1-mu:si:qa:
(13) a.

loud-Gen
not appeals to me listening-Nom the-music

music."
"I don't like listening to loud
SSa:khibat-1
b. la: PastamtiEu bi-sama:E-i 1-mu:si:qa:
the-music loud-Gen
not I-enjoy with-listening-Gen
T don't enjoy listening to loud music."

(14) a. The cook fried the fish.

b. The fish was fried by the cook.


to
ability
our

Grammatical knowledge also accounts counts


for
n o w

knot
who
People
recognize grammatical relations in a sentence.
Ce.
roo
English know that John has dilferent functions in
(15a) and (15b), viz.
object and subject, respectively:
(15) a. John is easy to please. (John functions as object, (cf. It is

easy (for anyone) to please John.))


b. John is eager to please. (John functions subject, (cf.
as
John
strongly wants to please other people.))

1.4 Language and the problem of knowledge

The question of how children learn their LI has intrigued

immemorial. This is so because the


philosophers and linguists for time
normal
cognitive abilities of young children are quite limited, yet
the highly complex structures of their native
children seem to master
six years, despite the
short time span, normally five
or
language in a

stimulus with which they are presented. Chomsky


poverty of the
this apparent paradox as "Plato's problem.
(1986b) refers to

nativist approach to language


Before the advent of Chomsky's
the linguistic arena
was dominated by
the fifties,
acquisition in
known from Bloomfield
(1933).
(1933).
structuralism best
American
be accidental in the
to
Bloomfield
considered language acquisition
and
the child in the babbling
accidental sounds produced by
sense that with specitic
associated
not
often than
are m o r e
holophrastic stages this
reactions. Implicit in
outside world
via their parents'
and is
objects in the starts from without
acquisition
that language vein.
View is the idea reinforcement.
In the same

reactions and
adults'
characterised by
for
for language
language
Skinner (1957) rejected
any
explanation
acquisition
ion from
from
within the organism.
Like Bloomfield, (ibid) accounted for
Skinner (ibidh

in terms of outside
factors and/or c o n d :
language acquisition ons.

However. in his review of Skinner's Verbal Behaviour,


Chonnsky dismantled the Skinnerian approach to language acquisition.

pointing out that the assumptions


made by Skinner with regard d to such
t.

vital terms as 'stimulus, 'response and reinforcement' were


were not
tenable. Cook and Newson (1996) argue that Chomsky's dismantline
ing
of the Skinnerian framework has remained his main influence on
psyehology. rather than his later work. Chomsky's alternative innate
approach has come to be known as the Universal Grammar Theory of
language acquisition. The latter theory posits that children are bom
with a built-in faculty for
language acquisition called Universal
Grammar, or UG for short." Chomsky (1981), for
instance, argues that
UG is a system of
principles and rules common to all human
languages and available to all children prior to
experience. The formal
features of UG are innate features of
human beings; they belong to the
characteristics of the species and have thus
been programmed into the
genetic equipment of the mind/ brain.
This accounts for the
With which children rapidity
acquire the rules of their native
matter how language, no
complex they may be. Thus,
explicitly takes language Chomsky's innate hypotnesi
without as argued
acquisition to start from within, not rom
by Skinner and others. i

"The
sortof
arguments for Universal Grammar
preliminary ritual to be
at hand." have by now become almost a mant
Jackendoff (1997:2). performed before plunging into tnE echn.
Of course. UG alone is not
enough to acquire a language, for it
this were indeed the case, all human beings would end up speaking
one and the same language. For the most part, the
speech community
into which we are born decides for us the type of language we will end
up using. That is, we seem to acquire whatever language is dominant
in the environment into which we are thrust. Thus, according to the

Universal Grammar hypothesis, human beings must additionally be


exposed to linguist input, i.e. primary linguistic data (PLD), in order

for them to eventually develop a particular language. Probably, UG

interacts with experience acquired in a linguistic environment and

serves as an interface between the input- characterised by PLD- and

the output, or what Chomsky calls a generative grammar. Chomsky

(2007) confirms this view, arguing that

Evidently, development of language in the individual must involve


three factors: (1) genetic endowment, which sets limits on the
attainable languages, thereby making language acquisition possible
(2) external data, converted to the experience that selects one or
another language within a narrow range, (3) principles not specific to

FL

Grammatical knowledge is therefore untaught; rather, evidence


from the environment allows the child to create a core grammar, or So.

In addition to this core grammar, the child also acquires a large

number of lexical items (words), having each a lexical entry that


specifies idiosyncratic information such as pronunciation, mcaning

The linguistic input/data is itself impoverished in respect to the bewildering


Complexity of syntactic structures.

9
Prnciples und P l u

Chomel

restrictions.
This s q u a r e s
with
nsky's (1982) view
andsyntactic of da
natter of
large part
of language
learning
is a matter
determining from
that a

the elements of the tal dictionary) and


lexicon (mental din.

oresented data
the child needs to learn that cry
that .

iS
For example, a
verb
their properties.
VP with a following
NP object; that .
occur in a send is a
that may not
VP With two following NP comnl..

verb which may


occur in a
lements
an NP and a PP; that tell is verb ich
which requires either
reauire
or, alternatively,
NP objects or an NP and
a following NP object, two folloWing a thas
at
clause. etc. This shows that children and L2 adult learners, for that

matter, continue to learn novel lexical items even after they have

reached a fully developed linguistic competence, referred to by


Chomsky as a steady state, or S. To this effect, Jackendoff (1997:5)
argues that:

Chomsky's next question is, If linguistic knowledge consists of a


mental grammar, how does the mental
grammar get into the speaker's
mind? Clearly a certain amount of
environment is necessary, since
childrenacquire mental grammars appropriate to their communities.
However, the combinatorial principles of mental be grammar cannot
directly perceived in the environmental input: they must De
generalizations constructed
OT the
(unconsciously) in response to
percepu
input. Therefore the
language learner must come to the task o
acquisition equipped with
a
capacity to construct I-linguisu
generalizations on the basis of
E-linguistic input.
Grammatical knowledge estigated indirectly via
may be
vague native speaker investigated nu
intuitions, i.e. how native about

See below for


speakKCi
more on
this.
the structures of their language. Support for this from
Chomsky
comes

(1977). who remarks that *lt is an obvious and uncontroversial fact


that informant judgments and other data do not fall neatly into clear
categories: syntactic, semantic, etc." Language investigators are
therefore required to have at their disposal extra techniques to
circumvent this shortcoming. Chomsky (1986), for example, argues:

Linguistics as a discipline is characterised by attention to certain kinds


of evidence that are, for the moment, readily accessible and
informative: largely, the judgments of native speakers. Each such
judgment is, in fact, the result of an experiment, one that is poorly
designed but rich in the evidence it provides. In practice, we tend to

operate on the assumption, or pretence, that these informant


JudgmentS give us direct evidence as to the structure of the I-

language [F internalised language, i.e. language which the speaker has

internalised], but, of course, this is only a tentative and inexact


working hypothesis, and any skilled practitioner has at his or her

disposal an armoury of techniques to help compensate for the errors

introduced.

Linguists engaged in investigating syntactic knowledge seek to

theory of language structure, or a syntactic theory. In this


develop a

respect, Ouhalla (1999: 7) states:

task is essentially one of reconstruction, in the sense that the


[Their]
linguist tries to reconstruct, via the process of analysing data,
that exists in the mind of native speakers. In other words,
knowledge
to formulate a theory, sometimes called a
the task of the linguist is
model, of language, insofar as theories of natural phenomena in

attempts at reconstructing the mechanisms underlying


general are

those phenomena.
Principles and Paru

framework
ntactic
ofsyntactio heory is one
a). c1
P r i n c i p l e s - a n d - P a r a m e t e r s

The
(198), (1982),
(1986a) (19865). (1988)
Chomsky
(cf.
such model,
The latter framework is
cited therein). construed as a
andreferences aho
Universal
Grammar (UG), which, as noted above, isa
theory of
the initial state (So) of the language facul.
with
characterisation of
UG consists of a cuct
which man is genetically endowed. f

principles and a finite array or parameters having a finite


ite
invariant
most two. As such, UG IS characterised by hein.
number of values, at ng
all natural languages, and (2) a theorv
(1) theory of grammar
across
a

of the innate linguistic endowment. It follows, then, that it must meet

two requirements, VIz. (1) it must be comprehensive enough to

provide grammars of all natural languages, and (2) it must be


constrained enough to allow children to construct a particular

grammar from the impoverished primary linguistic data to which they


are exposed. These seemingly contradictory requirements can only be

reconciled if UG is a theory that has a set of basic (innate) principles


but with a finite number of parameters which may be set one way or

another v1a exposure to a particular language, thus accounting for

cross-linguistic variation. In other words, learning the grammar or a


particular language reduces to fixing UG parameters for that anguag
according to PLD. When all the
parameters are fixed, the mind/brai
reaches a
final steady state for language, i.e. Ss. This is
capturcu o

See, for
example, Van Valin
In this (2001) for a selection of other
sense, UG is the
languages are theories of linguistic theory of So, and modeis.
various
1-(nternal) languages. gramun
figure in (16), (cf. Cook & Newson (1996: 125). See also Ennassi

(2004).

principles LI grammar (principles,

(16) LI Input UG parameter settings,


parameters vocabulary)

1.5 Levels of representation

Parameters theory is to reveal the


The aim of Principles and
consists of
universal structure common to all languages. It
underlying
been
organized in the form of what has
a number of components
literature as a T-Model, usually pictured by an

referred to in the

down T shape as in (17):


upside
D-structure
(17)
(Move a)

S-structure

(Move a)

LF
PF
Logical Form (meaning)
Phonetic Form (sounds)

13
P r i n c i p l e sa n d P a l u

LE-
d LF-
PF and
form the
form the computatiional
The
four
levels-
DS, SS,
of this system, each sentence has a distinct forma system
level of
cach the next
the one
one by
by
At
that is
mapped
onto

face' between the l.


computational
representation

DS is an
internal interfa lexicon and the
operations. lev:.
generated by the projection of
computational
component, ems
theoretic condit.
in accordance With
Ä-bar
tions. DS
from the lexicon
to selectional and thematic nrn.
must contorm es of
representations
then, DS IS a representation of
lexical items. Strictly speaking,
functions". Representations at sLthis
thematically relevant grammatical
level are mapped onto SS via the general rule "Move a', the lattor

that places tho


being an operational rule of the computational system
burden of explanation on general principles rather than specific rules

encoding constructions', (cf. Safir (1985). For example, it will be


seen below that NP-movement is passive and raising structures is
driven by Case theory, and not by the early construction-specifie
transformations of the Standard Theory. Viewed from this
perspective
Move a is a sort of short hand notation for
1.e. a
computational operations,
meta-rule that relates
representations at the various levels of tne
syntactic component. This rule
as well
overgenerates, producing well-TOTmeu
as
ill-formed structures. But
the various principles orI
the
But see, for
principi
minimised to example,
only twoChomsky
msky (1995),
where the com
Conceptual necessity. LF levels: PF and LF. These computational system has been
Conceptual
interfaces with
is the
Intentional system level of
levels are required
representation
Tepresentation
required by virtual
that interfacesith
that with tne
the
the (CI), and PF is the level of inte
generative grammar,
PF, all ron
ammar, all
conditionserceptual
(cf.Chomsky (1995: 219)). system
system Articulatory-Perceptual
level of reprresentation that
(AP). In this ersion
version of
on
syntactic representations hold at LF and0o
yntactic AP versio
M.K. Ennassiri

Principles and Parameters interact to rule out representations whicn

violate the appropriate licensing conditions at each syntactic level.

Well-formedness conditions are thus conditions on representations,


not on Move a itself, (cf. Ross (1967), Chomsky (1973), (1976),

(1977)).

As for SS, it is an intermediate abstract level which contains

empty categories of various types." Representations at this level are

in turn mapped onto the two external interface 1levels of LF and PF via

additional computational processes, such as Wh/Quantifier raising and


stylistic/scrambling rules, respectively. LF is a level of representation
where S-structures are associated with a logic0-semantic interpretation

(meaning), and PF is a level of representation where they are

associated with a phonological representation (sounds)." The objects

restricted to those properties of sentences that are


of LF and PF are

relevant for their meaning and sound, respectively.

1.6 Aspects of the lexicon

The lexicon constitutes the building block of language. The


be conceived of as an inventory
lexicon of a particular language may

level which is derived as a result ofa


SS is "basically a derived level, that is a
and yields S-structure as output.
process which
takes D-structure as input
mapping which derives the
serves as input
to another mapping process
In turn, S-structure
(1991:21).
Logical Form (LF)." (Ouhalla and syntax and semantics may be
The mapping between syntax and phonology
represented as (i) below:
SEMANTICS

() PHONOLoGY - SYNTAX
of all the
lexical items
available in that
Each lexical
language.
e. Fach

that specifies 1diosyncratic


phonolooic item
tem
has a lexical entry cal, syntactie
and semantic
information that cannot
herwise predicted by
be otherwise

lexICOn is a specification of id:.


general rules. As such,
lexical properties
the

of each word in the language. It has been


been
idioargued
syncratinic
the lexicon is the locus of parametrie
generative literature that
We shall
choices
and hence of cross-linguistic variation. see below some
aspects of this variation between English and Arabic. The levi
Icon
determines the thematic properties of those lexical items that funetio
as heads and their thematic and argument structures.* In the P&p
model, lexical categories are defined in terms of the features [EV] and
+N]. Thus, [+V, -N] is a verb, EV, +N] is a noun, [+V, +N] is an

adjective and [-V, -N] is a preposition.

Besides lexical categories, there other


are
categories which are
not definable in terms of [+V] and (+N] features. These are referred to
as functional/inflectional categories, whose class is a close class croSS-
linguistically. These categories include DET, AGR, TENSE, ASP,
NEG, COMP, to cite but a few. Since Pollock
(1989), the working
hypothesis has been that each one of these
allowed to head its own
categories should De

maximal projection, in with -bar


theoretic conditions.
agreement
Clause structure has
therefore become ore
"The selectional properties of
categories they take. lexical items determine the
the theta
theta roles or semantie
This is referred
category is realised as a to roles mntic
as
S-(emantic) selection.
semantic

category via the ruie Eaanonical


Structural Realization certain syntactic
structural realization (CSR), (cf. Chomsky (1986)). For example, anonical
ple, the canonical

the CSR of of the semantic


is NP, and
NP, and
proposition (=CSR category patient (=CSR
(proposition)) is CP, etc. (patie
articulated than before, perhaps too articulated, (cf. latridou (1990),
Ackema & Neeleman (1992). Functional categories play a significant
role in the P&P approach to syntax, as they are 'the flesh and blood of

grammar, (ef. Ouhalla (1991:1). Further, Chomsky (1988:2) points


out that "if substantive elements (verbs, nouns, etc.) are drawn from
an invariant universal vocabulary, then only functional elements wil
be parametrised."

The syntactic category to which a word belongs determines its

distribution, i.e. the context in which it may occur. This amounts to

saying that words of category X are not interchangeable for those of


(17) dismiss in
category Y. For example, we may not replace the verb
below by its corresponding noun dismissal, despite semantic affinity:

(18) a. I will dismiss the trouble maker.

b. I will dismissal the trouble maker.

Any native speaker of English knows that (186) is


not good as an

sentence. So, to insert a certain verb under the node V in a


English
whether this verb may occur under that node.
verb phrase depends on

Instead of dismiss, any of the following verbs may be used: chase,


fire, criticize, blame, arrest, watch, as they are all transitive verbs.

Speakers of English (whether native or non-native) know the

complement structure of different types of verbs. For example, they


in a VP in which there is a following NP
know that go cannot occur

in a VP in which there are two


complement and that give must occur

NP complements, or, alternatively, an NP and a PP. This is part of


Principlesa n d Paramete).
d Arabi

knowledge is
k n o w l e d g e .
This
This
e x i c a l kno
lexical
l
ncoded in the
outlined
below
o t l : . .

manner
the
manner

in the
in
lexical

their verbs
those
of
entries
lexical

V.
(19) a.g0
b. dismiss:
V.--NP
NP- PP]
NP- NPV [(--
v.
C.give

of Arabic know that qa:la (say) rea


Similarly,
speakers requires a
and that zawu
(that-clause), zawwaa
following
Pinna-clause

alternativelu
marry)
complements, or,
requires two
following
NP
and
encoded in the resneo:
e lexical
knowledge is
a PP. This type of
verbs, too:
entries of those

(20) a. qa:la: V, [-- ?inna-clause]


ka?i:ba-tun
a'. qa:la ?inna ssama:?-a
said-he that the-sky-Acc gloomy-Nom
He said that the sky is gloomy.

b. zawwaja: V, [--- NP-NP]/ [--- NP-PP]


b. zawwaja zayd-un Eamr-an ?ibnat-a-hu
married Zayd-Nom Amr-Acc daughter-Acc-his

"Zayd has given Amr his daughter in mariage.


C.
Zawwaja zayd-un ?ibnat-a-hu li-¬amr-in
married Zayd-Nom
daughter-Acc-his to-Amr-Gen
"Zayd married his daughter to Amr."
The frames in
(19a-c) and (20a&b) above identify subcategorie
of
verbs and are thus
called
contained in them is subcategorization frames. The informat1
called that
subcategorization information. We say
M.K. Ennassiri

a verb subcategorizes lor or selects its complement(s). Which


complements are Selected by
particular verb is an arbitrary property
a

of that verb. Lexical properties of lexical items must be accurately


reflected in all syntactic levels and can in no way be altered." This

requirement is stated in terms of a principle called the Projection


Principle.

(21) The Projection Principle


Representations at each syntactic level (i.e. DS, SS and LF) are
projected from the lexicon, in that they observe the
subcategorization properties of lexical items.
It follows that representations that do not observe this condition are

excluded. In addition to the filter stated in (21), Chomsky (1981,


1982, & 1986) proposes the following principle, which he calls the
Extended Projection Principle (EPP):

(22) The Extended Projection Principle (EPP)*


Clauses must have subject

By way of examples,
consider the following expressions:

The police arrested the suspect.


(23) a.

was arrested t.
b. The suspect

C. arrested the suspect.

information: (1)
relevant lexical
syntactically
There thus three types of thematic
are
information and (3)
subcategorisation
categorial information, (2)
information. but it is subject to
applies to all languages,
that EPP while
I t has been assumed require an overt subject
e.g. English,
variation. Some languages,
parametric
null subject.
others, e.g. Arabic, allow
a
P r i n c i p l e sa n dP a r a m e t e .

maqa:lat-an

zayd-un
katab-a

w r o t e - 3 M S Z a y d - N o m a r t i c l e - A c c

a.
(24)

article."

an
wrote

Zayd

zayd-un
t?
katab-a
ma:dha:
b.
what wrote-3MS Zayd-Nom

write?"

"What did
Zayd

c . P a w q a f - a I - m u t t a h a m n - a

arrested-3MS the-suspect-Acc

"Arrested the suspect."

of the verbs arrest and


the lexical properties
In (23a) and (24a),
verbs Select a
complement- are
that both
i.e. the fact
kataba (write)- transformation
movement
D-structure, as no
of
satisfied at the level
these lexical
and (24b), however,
has applied yet at
this level. In (23b)
S-structure via the NP- and wh-
are satisfied at the level of
properties
elements
from which the original
traces left in the respective position
all pass the Projection
have moved. Therefore, (23a&b) and (24a&b)
In contrast, (20)
Principle and the Extended projection Principle.
is
that
violates EPP, as the sentence lacks a subject. Notice
which
correct in Arabic, which is a null-subject language, 1. ntent is
allows the subject of the clause to be dropped. However, its co

recovedrable from the verbal inflection. So unlike tnc Projection

Principle, EPP may then be parametrisable with regard to the syntac


level at which it must
apply in different languages.
Ouhalla (1991:10) assumes that lexical categories and
functional categories belong to two separate modules of the
mind/brain. The former belong to a mental lexicon which exists

independently of UG, i.e. an independent module of the mind/brain

(the conceptual system); the latter belong to a grammatical lexicon,


itself determined by UG, (see Ouhalla (ibid) and Tsimpli and Ouhalla
(1990) for details and discussion). An interesting hypothesis advanced

by Ouhalla (1991: 201, note 2) is that differences in word order

analysis, be attributed to the order


among languages can, in the final
of functional categories which make up the verbal complex, (see
bear
(2014a) for examples from Arabic that
on
Ennassiri (1996) and

this proposal).

Parameters model
1.7 Modules of Principles and

consists of the following


The Principles-and-Parameters model
these
often themselves called sub-theories. Each of
separate modules,
vis-à-vis which natural
modules is associated with parametersS
and contrasted.
languages may be compared

(23) a. X-bar Theory

b. Case Theory
c. Theta Theory
d. Movement Theory
e. Binding Theory
f. Control Theory

21
ud ru
sub-theories operate on syntactic ctic structur
above
All the but
levels oft syntactic vsis- ng
analysis-
DS, SS theshey
may affect
diflerent or
LF In
chapters,
we shall address these
these modules and the
subsequent

the paranmeters
assoCiated with
cach one of them It
will investipate
be seen
the grammatical
model adopted he that
of is more often
the complexity
the interaction of these modules
than not the result of
II. X-BAR THEORY AND PHRASE STRUCTURE

This chapter will discuss the following topics:

Inadequacies of traditional phrase structure rules


Principles of X-bar theory
Analysing Arabic sentences

The Dcterminer Phrase (DP) hypothesis


The VP-shell hypothesis

Clausal categories

Rules
2.1 Inadequacies of Phrase Structure

deals with the structure of phrases and sentences,


This chapter
phrases and sentences are formed in English and
that is with how
associated
Arabic. Recall in this regard that the Phrase Structure Rules

Transformational Grammar recognize only


with the carly stages of
level of representation above
the head node, i.e. the phrasal level
one

structures:
XP. This is illustrated in the following

(1) a. NP b VP

NP
(Det
eat an apple
the story

technique will be used to


and the labelled bracketing
Both labelled tree diagrams
structure of phrases
and clauses.
represent the

23
Towards a Contrastive Svntes of
Phin iplesanad
Purameters Theory:
English and :
d. PP
(AP
PP NP

fond of Mary on the shelf

Accordingly, the structure associated with example (2)) is given in (3


given

below
(2) John will [vp finish his homework on Saturday|

(3)

NP Aux VP

John will V NP PP

finish Det N P NP

his homework on Sat

The
The graph-theoretic representation in (3) provides a visual

representation of the constituent structure of sentence (2), aand


assumed to have been generated by the following pre-P&P rme
Structure rules (PS-rules):

(4) a. S - NP Aux VP
b. VP V NP PP

c. NP- Det N

d.PP P NP

PS-rules such as (4a-d) were justified in a number of ways.

in favour of these syntactic rules is


One of the arguments

nature. If the same chunk of language can appear in a


distributional in
number of positions with the same internal dependencies, then that

chunk is likely to be a constituent. For example, in (3) NP appears as

and prepositional object, but it has the same

subject, direct object


of the slot it fills within the
sentence.
internal irrespective
structure
one of the basic
PS-rules enable us to explain
More importantly,
Within phrase
human viz. recursion/nesting.
language,
properties of
sentences of indefinite
to generate
structure syntax, it is possible
X to be
acategory of type
length cross-linguistically by allowing
a finite rule
of the same kind. Thus,
embedded inside another category
sentence
infinite number
of possible
an
enables us to generate
16

structures in human language.

Theory
2.2 Principles of X-Bar
here. First, the
number of problems
however, a between
There are,
internal hierarchy
i.e. there is
no
VP is flat, sisters to
Structure of the as
c o n s t i t u e n t s appear

verb; rather, all


of the those
ne constituents PS-rules
such a s
by right
is captured
recursion
o n the both
occurs
Structure Syntax, S here-
thin Phrase same
category-

where the
VEn (ia&b) below,
in
arrows:
and left side of the
VP
a. SNP AUX
b.VPVS
25
Principles and
Parameters Theory:
Towards a
Contrastive Syntax ofE.
English and Arabie
three branch
V. Second, this tree untypically involves branches, in
hypothesis, which reaui
contravention of the binary branching quires that
all branching nodes are binary branching, 1.e. a branching node.
can
only have two sub-nodes, not more. Third, it violates
violates the
endocentricity condition, which insists that phrases must be projected

from a head."

To put our discussion on a more concrete footing, consider the


following examples:
(5) John will finish the homework on Saturday and Mary will do
So too.

(6) John will finish the homework on


Saturday and Mary will do
so on Sunday.

In (5), do substitutes for the entire VP, i.e.


so
finish the homework on
Saturday. In (6), however, do so substitutes for
only part of the VP,
i.e. finish the
homework. It has been argued in the
literature that
substitution is
structure-dependent, i.e. only constituents can be
substituted for by an element. It
seems, then, that in (6) do so
Substitutes for constituent that is somewhere between VP and
a
V.
This
binary-branching intermediate projection is labelled V-bar, or V'.
The VP now has the following layered structure, where there are three
different levels of projection, viz. VP, V and V. V is the head

"i.e. for all PS-rules, the


following generalisation must hold:
XP.X...
M.K. E n n a s s i r i

(terminal
inal node) and VP is the root
node, and the lines
connecting the
nodes are
called branches:

VP
(7)

V NP PP

finish his homework on Saturday

In the above structure, the PP is detached from the head V, and the
close relationship between the verb and its object is expressed in terms
of sisterhood.

Thus, X-bar theory is a model of phrase structure that consists


of basic principles whose function is to constrain structural
representations at different levels." These principles are stated in
terms of the following general schemata, (cf. Chomsky (1986b):

(8) a. XX" x* [Head-Complement]


b. X"X" x [Specifier-Head]
(where order is irrelevant, and * stands for any number of)

As Chomsky (1986: 91, fn. 3) has clearly stated:


X-bar system is determined by
For a particular (core) language, the
in
the of X-bar theory (head-first, etc.)
TIXing the values of parameters
them are determined
accordance with whatever dependencies among

attempt has
been made to
18 where an

Dut see Kayne (1994) and Chomsky (1994),


deduce X-Theory from more basic principles.
x
P r i n c i p l e s a n d P a r a m e t e r

0jof English
Engiishsh andand ro
.
choices constitutes the XL
particular set of
by UG, a

of L.
component of
the grammar

Stowell (1981) argues


that the Ä-bar schemata in
8) impose the
well-formedness conditions, to which he
following refers as 'node
admissibility conditions
on DS representations, a term .
m he
McCawley (1968):
attributes t
(9) a. Every phrase is endocentric;

b. Specifiers appear atthe X level; subcategorized


appear within X" nponents
c. The head always appears adjacent to
boundary of X. one

d. The head term is one bar-level lower than


the immediatelv
dominating phrasal node;
e. Only maximal projections may appear as non-head tenms
within a phrase.
According to (8), each X category- e.g. N, V, A, P,
and T-
NEG, AGR
combines with its
complements to form an X'-level, which, In
turn, forms a maximal
projection, X", with its specifier. Therefore, the
basic notions in this
system are X" categories, i.e.
heads, speciiers and
complements, latter being X"
the
categories, i.e. maximal projec ons.
1
Chomsky (1986:4) points
complements is determined byout that specifiers are optional while the choicé of
cifiers are
optional wn
the
Boeckx Projection Principle given
given ini(
in I (21).
() (2006:175) points out that
a.Minimal the OOWing
following relations
A
minimal Projection (X ought to be recognize
relations oug
b.AMaximalprojection
Projection isX)a lexical item selected from the lelexicor
maximal ion is a
further. projection syntactic object that
that does not project any
c.
Intermediate Projection (X')
These notions play a crucial role in X-bar theory as they are involved in
accounting for the
licensing of some positions in phrase structure. The
order at each
phrase level is subject to parametric variation acrosS
languages. The order chosen for
particular language depends on the
a

directionality of Case assignment, thematic marking and government in


this language. The noted
cross-linguistic variation is assumed to be the
result of fixing the head and
specifier parameters stated in (10) and (11)
below, (cf. Koopman (1984)

(10) The Head Parameter


a. X» YP X (head-final)
b. X X YP (head-initial)

(11) The Specifier Parameter


a. XP(YP) X*(specifier-initial)
b. XP X' (YP)(specifier-final)
Arabic.English, French and Spanish, for example, are all head-
initial languages, whereas Japanese-type languages are head-final
languages. This parameter determines the basic order of elements in
sentences. (cf. I: 1-5). And Arabic, English, French and Italian, for
instance, arespecifier-initial languages, while Bulgarian, for example,
is a
specifier-final language. X-bar structures are basically projections
of heads selected from the lexicon. The basic relations are therefore

An intermediate projection is a syntactic object that is neither minimal nor


maximal
Languages differ as to whether they realize their heads to the right or to the left
of complements. Languages also differ with respect to the relative position of the
specifier. This indicates that there is no universal position for heads and specifiers.
Towards a rastive
Contrastive
Syntax of English
Principles and
Parameters Theory:
and A
and Ardhit

Specifier-head relation
and head-complement relation elation, (cf. Chomsky
(1992).
To illustrate further the working of X-bar theorv. let
consider
the following example:
The students will [vp all finish heir homework on
(12) Saturday
The quantifier all occurs to the left of the verb inside the Vp
Its
position is identified as a specifier position, (cf. the VP-intema
subject Hypothesis).4 The structure assoCiated with the VP is
thus
given as (13) below, where the floating quantifier all has been
stranded off the subject the students which it quantifies:

(13) VP

Spec

all V'

NP PP
A

finish their homework on Sat

According to this hypothesis, the


Kuroda (1988), Fukui & Speas thematic subject (cf.
others). In DS, the NP "the (1986) and
originates wi
Koopman & Sportiche
Sporticne (1991),
among

Suosequently moves to students" appears as the all and


of all and
derived as in (i) below: Spec-TP leaving all in Spec-VP complem thus be
position. (12) *
) IT the
students; Ir will lvp
[ve all t v finish on
Saturday) v finish their homework) l
M.K. Ennassiri

Schematically, English VPs are fomed according to the following

format:

(14) a. V » Spec V

b. V V complement

Going back to the PP on Saturday, we can see that it is detached from


the verb finish. This is so because the PP is not obligatory, i.e. it is not

subcategorized for by the verb. Rather, it is an optional element

(referred to in the literature as adjunct) that has been added to the VP

by recursively expanding a V' into another V'. Thus, the VP all


finished their homework on Saturday now has the following
hierarchical structure, where the node labelled with a question mark in

(13) is labelled instead with V':

(15) V''

Spec

all

NP PP

finished t h e i r homework on Saturday

that a verb (or any other


lexical category for
a phrase
Recall that a complement is subcategorization
frame.
its
hence appears in
that matter) selects and

31
Contrastive G.
Theory:
Towards a
Syntax of English
Principles
and
Parameters

and Arabig
NP their
homework merges/ combines
nes with the
with the 1
head V to
In (15), The
on Saturday in tss
turn merges with
and the PP adjunct
form the lower V
another (higher) V'. Adjuncts are. f
the lower V'
to form the most
and shown to be peripheral in X-bar system E:
part- optional y, the
V finished their homework on Saturday merges with the subiet all to

form the maximal projection V7 VP, All that is needed is to add to

the rules given in (14) the adjunct rule (16):

(16) XX' ZP (adjunct)"


A tree like the one given in (15) illustrates two different types

of relations that are crucial for syntactic analysis: (1) hierarchy


relation defined in terms of dominance, and (2) ordering relation
defined in terms of precedence. Another important node-relation is
that of c-command. Chomsky (1986a). for example. defines this
relation as follows:
(17) a C-commands ß iff a does not dominate B and every y that

dominatesa dominates B.
What (17) means in fact is that a node a c-commands its sister node p
and all the nodes
dominated by B, i.e. its nieces, to use a famiy
relation term. So in (15), V
(finished) c-commands NP (eu
homework), and
Spec c-commands V, V, NP and PP. The C
command is relaiou
symmetric, meaning that NP (their
commands V (finished). homework) aso

"The following
terminology ensues
(i)
(i)
A
specifier is daughter offrom (10), (11) and (16):
XP and sister to X.
An
(ii)
adjunct is daughter of X' and
A
complement is daughter of X' sister to X
and sister to X.
Summarizing, then, X-bar theory expresses hierarchical
relations among elements. For example, the relation that a hcad

entertains with a complement is closer than that it entertains with an

adjunct. The schemata in (8a&b) predict that the specifier and


complement of a given category are distinguished from those of other

categories. For example, N complements cannot be immediately

dominated by V, and vice versa. (8a&b) also exclude structures where


a head category is immediately dominated by the projection of another
category. For instance, the category V cannot be dominated by the

maximal projection NP, and similarly, N cannot be directly dominated


by the maximal projection VP. X-bar theory, then, requires that a
lexical head, or a functional head, for that matter, and its maximal
projection belong to the same category. More important, still, is the

fact that in an X-bar structure the head always governs its


complements, but not its specifier. It will be seen below that this
property plays a crucial role in Case assignment, both in English and

Arabic.

2.3 Extending the X-bar format to other categories

X-bar does not apply only to VPs but extends to other


theory
well. So, the schema underlying NPs should be
lexical categories as

that given in (17) below:

(18) a.N" Spec N'


b. N' N complement

c. N' N' adjunct


Contrastive
Theory:
Towards a
Syntax of Englishish and
Principles and Parameters

So, let us considen


issue, though.
This is an
empirical
folowin
examples:
solution to the problem]
satisfied even
(19) a. [NP The body.
solution to the probler was quite cleve
b. [NP Mary's
c. NP Students of linguistics
from Essex] got good grade
des.
d. [NP Students] hate exams.

The bracketed NPs in (19a-d) have the following respectivestructur

(20) ab. N"

Spec N

Mary's N PP

the solution to the problem


C. N"

N
PP

PP from Essex

students of linguistics
d. N"

students

It seems, then, that NPs, too, are analyzable in terms of X-bar


schemata in that they, too, have two internal levels of structure: an N"

which consists of N' and a


possible specifier, and an N' which consists
of the head N and possible complements.
However, it has been argued in the generative literature that
NPs are best analysed in terms of what has been referred to as the

Determiner Phrase (DP) hypothesis, (cf. Abney (1987 ), Chomsky


(1995) and Fukui & Speas (1986), inter alia). This hypothesis
proposes that a nominal expression is headed by a determiner that
takes a noun phrase as its complement. *"What this implies in the case
of bare noun expressions (i.e. noun expressions used without any

modifying determiner), Radford (1997:95) succinctly argues, "is that


such nominals are DPs headed by a mull determiner."Accordingly,

(19a-d) have the following respective alternative structures, irrelevant

details omitted:23

the determiner, but not


host either the possessive morpheme
25 or
The head D can

both.
35
Contro..
Towards a ntrastiveSSyntax
and
P a r a m e t e r s
Theory:
of English and
Principles
hrg
DP
(21) ab.

D
Spec

D NP
Mary

N PP

the solution to the problem

C DP

D NP

N
PP

N PP

students of linguistics from Essex


d. DP

D NP

N*

students

Adjective phrases (APs) and preposition phrases (PPs) are also

diagrammable in terms of X-bar format. Consider the following

examples:

(22) a. Jane is [ap Very fond d of birds]

b. The candy shop is [Pp right around the corner]

In (22a),. fond is the head, the degree adverb very is the

specifier, and the PP of birds is the complement. The structure

associated with very fond of birds is thus given as in (23) below:

Within this assumption, the


APs have recently been reinterpreted as DegPs.
structure associated with the bracketed AP
in (22a) is thus given as (i) below.. But
we will not follow this analysis here.

birds]111
) beg very LaP fond [e of lor IN
37
Contrastive Syne
Principles and
Parameters Theory: Towards a
Syntax of English
and
Arah
AP
(23)

DegP A

Deg A PP

L---
very fond of birds

In (22b), around is the head word, right is the


is the
specifier, and the corner
complement. The prepositional phrase right around the
has the corner
following structure:
(24) PP

Spec

right P

around L
the corner
In
notation forsummary, the X-bar format
ormat
discussed above
a important
constituent is phrase structure discussed above providesa
a

invariant
generalizations. It imposes
constituents is projection ofa head andgeneralizations.
that the
i
that

internal structure of

gorial y. Taken interna


crosscategorially. part
par as part of Universal
Grammar, this constrained view of the format of
phrase structure
contributes towards explaining
the
rapidity and uniformity of the
acquisition of syntactic structures.

2.4 A sketch of Arabic


phrase structure
2.4.1 The Determiner Phrase in
Arabic
In Arabic, too, all
phrases are formed in about the same way as
their English counterparts. To illustrate, consider the
following
examples:
(25) a.
[al-kita:b-u] mufi:d-un
the-book-Nom interesting-Nom
The book is interesting."
b. [ha:dhihi l-madrasat-u] mashhu:rat-un
this the-school-Nom famous-Nom

"This school is famous."

c. [qami:S-u l-walad-i] ja:hiz-un


shirt-Nom the-boy-Gen ready-Nom

"The boy's shirt is ready."

d. [baED-u tilka l-7afla:m-i] radi:?at-un

some-Nom those the-films-Gen bad-Nom

Some of those films are bad."


ntras Ve dyntaax
Towards a LOHI
htax of English anc
and
P a r a m e t e r s
Theory:
English and Ar.
and Arut
Principles

assumption
that all
languages
are underlvin
underlyingly similar, (c
Under the the moment
and leaving
aside for
r moment the s
issue of
Kayne
(1994)) the
eg.,
the head noun, the bracketed
NP in (25a
ND

feature
attached to
Case
phrase ker marker
(tree diagram)
(tree diagram), where al.al-
have the folloWing
should
definite article
the in English:27
to the
corresponds

DP
(26)
D NP

kita:b-u

The graph in (26) is isomorphic to that given for the English noun

phrase in (21b) above, modulo, of course, the PP complement. In

(25b), the determiner al (the) is preceded by a demonstrative adjectve


ha:dhihi (this), which agrees with the head noun in gender and

number. The demonstrative adjective is hooked under Spec-DP. (256)


thus has the following tree diagram:

2
The definite article (a)! (the) is c a i in
not
nature.
it gets incorporated into lexical in
Arabic; rather, t is
syntactically. the modified head noun,
MA. Ennassiri

DP
27)

Spec D

ha:dhihi D NP

(a)-

madrasal-u

demonstratives and articles are in


In English, however,
the other may not, as
complementary distribution, i.e. if one appears,
we see from the following grammaticality
judgement:

(28) a. ('the/a) school


This

b. (The/a) this school

and the same category, viz. D(eterminer),


This shows that they are one

for the same position.


and hence they compete
referred to in the
instance of a special construction,
(25c) in an
Fehri
Benmamoun (2000), Fassi
literature as the construct state, (cf.
Ouhalla (1988) and Ritter (1991), among
(1993), Mohammad (2000),
possessed head noun, qami:su (shirt),is
construction, a
others). In this Genitive
which is assigned
complement,
Tollowed by a possessor
noun cannot
take a
constructions, the head
state
ase. In the construct
This IS
noun may.
possessor/complement
determiner, but the

41
Towardsa
Contrastive Syntax of
Contrastive

Principles and
Parameters
TheorY:
English und hr
grammaticality 2st between (29%&k
contrast .

the following
illustrated by
and (29c&d):
al-walad-i
a. al-qami:S-u
(29)
the-shirt-Nom the-boy-Gen

b. qami:S-un walad-in

shirt-Indefinite article/Nom boy-Gen

C. qami:S-u walad-in

shirt-Nom boy-Gen

"A boy's shirt"

d. qami:S-u l-walad-i

shirt-Nom the-boy-Gen

The boy's shirt"

The question we should ask at this juncture is: what is the structure or

the bracketed DP in (25c)? The simplest assumption- and hence the

most desirable, both theoretically and empirically- is to posit that such

DPs as that in (25c) have the structure (30)


below, where the constr
noun qami:S-u (shirt) moves from under N" and gets relocatcu uunder

D' for reasons to do 28/29


mainly with Genitive Case assignment

In (30),
DP has a subject-like function, and N' is predicatea
that the

P means
something like 'the boy has a O
more on this. The N to-D shirt'. See Benman of Vt
n t o f V-to-T

movement, (cf. verb movement movement portrayed above is


29
The below). ren
strikethrough (i.e.
squares with recent qami:S-u) stands for the copy
ent.
Thi

py of the moved elem


movement to be no
developments in the generative enterp lementtoE
that take

copying, and the


new/highest copy of the n
M.K. Ennassiri

(30) DP

Spec D

Dgen
NP

qami:S-u DP N'

al-walad-i

GamiS#

It has been pointed out above that the construct noun may not
be preceded by the determiner al (the). We can now account for this
impossibility by the fact that the only position that would otherwise
have hosted al is already filled
by the abstract Case assigner Dgen» (cf.
Ritter (1991)). What this implies in fact is that Dgen and the definite
determiner are in complementary distribution. This is indeed
corroborated by (29a), repeated for convenience as (31) below:

(31) al-qami:S-u al-walad-i


the-shirt-Nom the-boy-Gen

In Moroccan Arabic (alias MA), however, the possessed noun


may be definite if it functions like the subject of the DP. In such cases,

the one that gets mapped onto PF. But nothing seems to be at stake here, apart
from terminology.

43
possessive particle. This
is is illhust
the head D is filled
with a
illustrated byby thethe
from MA:
following data

1-ktab d/dyal/ntaE drari


(32)
the-book of children

"the children's book"

We shall assume that (32) has the following structure:

(33) DP

DP D'

1-ktab D NP
d(yal)/ntaE drari

However, MA also has constructions quite akin to Arabic constuct


state structures, as we can see
from the following bracketed
examples:
(34) a. ETat-u [flu:s l-maSru:f
gave-she-him money the-expenses
"She gave him
pocket money.
b. ma-tonsa-sh [ras 1-Hanut
Neg-forget-Neg head the-shop
Don't forget (to
add) spices"
c.
glas-na f [wast d-dar]
M.K. Ennassiri

sat-we in centre the-house

We sat in the lobby"

In these constructions, too, the head noun- i.e. flu:s (money), ras

(head) and wast (centre)- may not be definite. Therefore, (34a), for
example, will have the structure (35) below, which is parallel to (30):

(35) DP

D NP

flu:s DP2 N

lmaSru:f N

However, the construct state in Arabic may still be modified


by a quantifier. This is illustrated by the following example:

(36) [kull-u buyu:t-i l-madi:nat-i] qadi:mat-un


the-city-Gen old-Nom
All-Nom houses-Gen

*All the houses in the city are old."


where the
The structure associated with (36) is given in (37),
quantifier kullu (all) is assumed to head a quantifier phrase (i.e. QP)
hooked under
and thus Case-governs the construct (possessed) noun

D
C o n t r a s t i v e Sur
Parameters
Theory:
Towards a
Syntax
itax of
oJ English
Engusn and
and Argbi
Arabiec
Prneiples and

QP
(37)

DP

D NP
kull-u

buyu:t-i DP N'

-madi:nat-i N

buyu:t

Further evidence in support of the claim that kullu (all) heads a QP

that dominates a DP may be gleaned from such examples as (25d).


where the bracketed nominal expression consists of the head noun

Pafla:m (films) and three preceding elements, namely the determinet

(the), the demonstrative adjective tilka (those) and the quanc


baD-u too.
(some). It is therefore plausible to assume that in (25d),
the quantifier ba£D-u merges with the DP tilka 1-Pafla:m-i to lo
then
maximal projection, i.e. the QP ba&lD-u tilka 1-2afla:m-i. It woul
follow that (25d) will have
the structure (38) below:
M.K. Ennassiri

(38) QP

Q DP

ba&D-u Spec D

rilka D NP

Pafla:m-i
The head noun may be further modified by
adjective phrase. This is
an

illustrated in (39), where the adjective al-jadi:d (new) modifies


qami:S (shirt), witness the fact that it agrees with it in definiteness
as

and, more crucially, in nominative Case:

(39) yuejibu-ni: [gami:S-u Lwalad-i al-jadi:d-u


appeals-me shirt-Nom the-boy-Gen the-new-Nom appeals
I like the boy's new shirt."

The structure associated with (39) is outlined in (40) below, where AP


is adjoined to D'.3

30
Alternatively, the AP al-jadi:d-u may be assumed to adjoin to DP*. But nothing
hinges on this at this stage.

47
a Contrastive
Ive
Theory:
Towards
Syntax of
Principles
and
Parameters

Engish and
Aru
DP
(40)

D'
Spec

D' AP

D NP

qami:S- DP N'

-walad-i N

al-jadi:d-u

Let us now turn to the structure of AP, PP and VP in Arabic.

Consider first the folowing bracketed APs.


(41) a. walad-un [dhakiy-un]
boy-indef-Nom clever-indef-Nom
"A clever boy"

b. bint-un [dhakiyy-at-un]

girl-indef-Nom clever-F-indef-Nom
"A clever gir"

c. al-kita:b-a:ni [al-mufi:d-a:ni

the-book-dual the-useful-dual
M.K. Ennassiri

The two useful books"

(41a-c) indicate that in Arabic adjectives come after the nounthey


modify and agree with the post-modified noun in number, gender,
Case and (in)definiteness. Abstracting away from agreement
phenomena for the moment, the internal structure of a
post-modified
noun such as (41a), for instance, may be diagrammed as follows:
(42) DP

Spec D'

D AP

D NP

N'
4

walad-un
dhakiyy-

But in Arabic, too, adjectives may in turn be modified, as the

following examples illustrate


(43) a. ?ana: [ap masru:r-un jiddan) I-yawma
happy-Nom very today

31
The dual form of an adjective is made by adding (a:ni) or (ata:ni) to the base form
of the adjective.

49
"I am very happy today."

I-yawma
b. Pana: [apjiddu masru:r-in]

I very happy-Gen today


I am very happy today."

c. zayd-un [aP mu:la&-un bi-l-ba:diyyat-i]

Zayd-Nom keen-Nom with-the-countryside-Gen

Zayd is keen on the countryside."

d. imra?at-un [aP jami:l-at-u l-wajh-i]


woman-Nom beautiful-F-Nom the-face-Gen
"A woman with a beautiful face." (Kremers (2005))
It has been
pointed out above that in Arabic, adjectival modifiers
normally come after the
adjective they modify. This is illustrated in
(43a), (43c) and (43d). Sometimes,
the adjective. In
however, they may occur before
(43b), for instance, the intensifier
modifies masru:r jiddu (very) pre-
(happy). We will assume here that
has a structure along the lines of (44) example (43a)
jiddan (very) adjoined to A':
is
below, where the degree advere

Prof.
Abderrahman Boudrae
pointed out
so to me that (from the Arabic Studies
they do not examples such
assume that reflect the
(43b) is derived usual
as
(43b) are Departtment)hic.
(pc) has
not very current in Arab and
and
from (43a) adjective-modifier
via order. Ne shall, therefore,
head-to-head movement.shall
M.K. nuassiri

44) AP

DegP
A Deg
masru:run jiddan

(4 3b) is the paraphrase of (43a), but with the position of the adjective
and the modifier reversed. Given the intriguing fact that the degree

adverb assigns Genitive Case to the adjectival head, we shall assume

that the degree word jidd (very) moves to a higher functional head -

perhaps Gen- from where it would be able to c-command and hence

Case govern the adjective masru:r-in (happy). If these assumptions


are on the right track. (43b) will be derived in the manner indicated in

(45) below:

(45) GenP

Gen AP

jiddu

A DegP
A Deg

masru:r jiddt

51
and Ara
In (43c), the adjective mu:la&un (keen) is
prepositional phrase. This is reminiscent of
of (22a) imin
post-modif with
(22a) English, a
the adjective fond takes a PP complement. (43c) shoul whee
have the structure in (46) below, which parallels that
of course, the pre-adjective material:
in therefore
(23), moduloreo,
(46) AP

A PP

mu:laE-un P DP

bi- Det NP

ba:diyyat-i
The adjective phrase in (43d) is an instance of what has been
referred to in the literature
adjectival construct, (cf.
as
Fassi Fehri
(1993), Kremers (2005), Siloni
(2002), among others). Adjectival
Constructs in Arabic involve
an adjectival head in construction wiun a

genitive complement
"which denotes a property, part or qualniy f the
houn that the
adjective modifies (in attributive use) or is cated of
(n predicative
wIll assume
constructions)", (cf. Al Sharifi & Sadler
(200
here that (43d)
has the following internal strucu where
al-wajh (the face) is the is
subject of the AP, and ami:lat (beautiful)
M.A. Ennassiri

predicated ot it. Subsequent movement of jami:lat (beautitul) to cten

will eventually result the final surface order


in
exhibited in (45d).
(47) GenP

Gen AP

jami:lat DP

-wajh-i A

jetmitet

(47) is parallel to the structure of (29d) given in (30), wherefore it was

assumed that the possessor and the possessed nominals entertain a

predication relationship, (cf. fn. 28).

bracketed examples of
Consider now the following
prepositional phrases in Arabic:

(48) a. bayt-i: qari:b-un [Pp mina 1-ja:mikat-il


near-Nom from the-university-Gen
house-my

is the university."
My home near

min taHt-i r-ruka:m-i]


b. akhraj-u: 1-juththat-a [pp

33 Kremers (2005) for a different analysis


But see, for example,
53
got-they the-body-Acc
from underGen

from under the rubble.


ibble-en
the-rubble.r

They got the body

c. inEaTaf-tu [Pp lamna.man ila: 1-yami:n-1


turned-I completely to the-right-Gen

I turned completely to the right."

nd a prepositicms
(48a) is a simple PP, consisting of a preposition and prepsitiona
DP complement. The structure associated with it 15
pretty much like
that of its English counterpart, viz. (49):

(49) PP

DP

mina D NP

ja:mieat-i
In (48b), the
preposition min (from) takes another PP
itself complemeu
consisting of the preposition taHt-i and
and the
the DP
(under)
complement r-ruka:mi (the rubble). The this
example is thus
structure associated w
given as (50):
M.K. Ennassiri

(50) PP

P PP

min P DP

taHt-i D NP

ruka:m-i

The prepositional phrase in (48c) is exactly the same as that in (22b),


and so its structure should be identical to that in (24):

(51) PP

Spec PP

tama:mnan P DP

ila: D NP

yami:n-i

55
Englishn arand
7 0 W u .

Theory:
Parameters
Arot
Principles and

structure of VP in
in Arabic. For
Arah:
turn to
the
the
Let us now
examples:
sake
following
consider the
ofexposition,
1-baHr-a
yu-Hibbu zayd-un
(52)
he-likes Zayd-Nom the-sea-Acc

sea."
"Zayd likes the
In (52). the verb yuHibbu (like) and its DP complement a (the

a constituent by virtue of the fact thas


sea) do not seem to form the
subject Zayd intervenes between them. However, assuming the

widely-adopted VP-internal subject hypothesis, (52) has the follo ina


ng
structure, where the verb has vacated its initial position for reasons
ns
34
that will become apparent below:"

We will argue below that the VSO order in Arabic is obtained by head-to-head
movement. The verb moves out of the head V position in VP into the head T
position in TP, forming the inflected VT (cf. Ennassiri (20146). In negative
ntences such as (ia-c) below, the tense feature/ morphology is realized on the
negative particles lam, lan and la: (not). Mohammad (2000:fn. 19) argues that "lam
carries past tense and places the verb in the
jussive; lan carries future tense and
places the verb in the subjunctive; la: carries
what Arab grammarians have
present tense and places the verb in
termed as nominative."
() a. lam na-ntahi ba£du
not-past we-finish yet
"We are not yet done."
b. lan
yu-sa:fir-a 7ab-i:
not-future he-travel dad-my
"My dad will not be
travelling."
C. la:
Pu-Hibbu S-Sayd-a
not -like the-hunting-Acc
"I don't like
hunting"
M.K. Ennassiri

(53) yuliibb-u VP

DP

zayd-un V DP

ty D NP

baHr-a

(53) resembles in relevant respects the structure of the verb phrase in


English, where the verb and the complement do indeed form a
constituent, i.e. a V', (cf. (15) above).
All in all, then, the simple notation of the X-bar schemata
given in (8a&b) and (16) seem to cover all phrase structure in Arabic,
too, at least for the investigated data. More importantly, differences in
linearization between Arabic and English are also accounted for in

(a), for example, will have the derivation (ii) below:

(i) lre lam, INeg t lv» pro lv na-ntahi baEdu]]]

In (i), movement is local in that the Neg head has moved into the head position in
the next higher phrase in the clause structure. Movement of the verb in such
structures would be disallowed by virtue of the fact that the head position in the
immediately containing phrase is already filled with the negative particle lam. And
movement of V to T would cross the intermediate Neg head, in violation of head
movement constraint (HMC), (cf. Travis (1984)). We shall see below that this
constraint is reducible to the Empty Category Principle (ECP).
It thus becomes clear
ar that
that crOSS-linguistic
terms ofX-bar parameters.

direct bearing on syntactic analysis. For


variation has a

functions as a pronominal
example,
al
the
fact that the genitive possesor
to the possessed
English but as a postnominal complement ssed construct
head noun in Arabic has lead us to an approach where the head
noun
c-commands its complement and thus assigns it the oo

state/Gen Case.

2.4.2 The VP- Shell Hypothesis

So far, we have dealt with VPs like (54) and (55) in English
and Arabic, respectively, where the lexical verbs take a
single
complement:

(54) a. The police will open an investigation.

b.Panhayt-u kita:bat-a l-maqa:lat-i


finished-I writing-Acc the-article-Gen
"I finished writing the article."
Such VPs have been
analysed in terms of binary-branching nodes. But
consider now the
following examples of ditransitive verbs:
(55) a.
[DP1The minister] sent [Dpz the ambassador] [DPs irgent
an ug

message|
b.
DPThe minister] Ipp
sent an urgent messagej [Pp *the
ambassador].
MK. Ennassiri

c. DPi The
new salesman] convinced [DP2 quite a few
our products|.
customers] [cp to buy

,S6)3. sallam-a DP1 -qa: id-uJ [DP2 I-junu:d-a] [DPs Tutab-a-hum -


(56)
jadi:dat-a].
gave-he the-chief-Nom the-soldiers-Acc grades-Acc-their the
new-Acc

The chief gave the soldiers their new grades."

b. sallam-a [DPi l-qa:?id-u] [DP2 r-rutab-a l-jadi:dat-a] PP li-l-


junu:d-i].
the-chief-Nom the-grades-Acc the-new-Acc to-the-
gave-he
soldiers-Gen

"The chief gave the new grades to the soldiers."

Eala: T-Tulla:b-i
C. PiqtaraH-a l-mushrif-u ba?D-a tta?di:la:t-i

some-Acc the-modification-
suggested-he the-supervisor-Nom
Gen on-the-students-Gen

Some modifications to the


The supervisor suggested

students.
constructions seem to be
The above instances of double-object
virtue of the
incompatible with the binary-branching hypothesis by
two of which are
1act that the italicised verbs take three arguments,
out of this
constituent. A way
45signed internally, i.e. within the V-bar
VP-shell
would be to adopt the
Seemingly problematic data
in (55a-c)
ditransitive verbs like those
Pesis, according to which
and (50a-c) above coc m Iwo parts n the manne.
manner outln
in
below, (ct. Larson (988) and Chonsky (1995), inter
alia) s7
(57)

SUB

VP

IO V

V DO

The traditional VP structure of constructions such as (55a) and (56a


which contain a verb and two
complements, is now contained within a
higher vP. The small v, which is
referred to as a light verb, carries
strong person and number features.
Apparently, these strong features
trigger movement of the lexical/ lower verb
to v. The
complete
derivation of the verbal
superstructure of (56a), for example, is
outlined as follows.
First, the verb sallam-a (gave) is
direct merged wtn ue
object rutab-a-hum -jadi:dat-a
"their new grades" to fom tne
V-bar sallam-a rutab-a-hum
This
1-jadi:dat-a "gave their new des
V-bar is then merged with the
soldiers" to form the indirect object -junu.u
subordinate VP structure, as shown ow:
beio
(58) VP

Dp

1-junu:d V DP

sallam-a rutab-a-hum I-jadi:da-ta

Notice that in (58) the indirect object (DP") functions as the specifier
of VP, whereas the direct object (DP) functions as the complement of
the V. The resulting VP is then merged as a complement of the light
verb (v). It has been mentioned above that v carries strong/formal
features, which fact causes the lexical verb sallam-a "gave" to raise to
it, thus forming v'. The resulting v is eventually merged with the
subject -qa:?id-u "the chief" to form vP. The complete verbal
architecture is therefore given as (59) below, where the lower V
36
substitutes for the light verb, as shown by the dotted

The
35 VP-Internal Subject Hypothesis should now be interpreted as meaning that
of the outer P shell.
the thematic subject appears as the specifier (55a).
to English examples such as
The same analysis is equally applicable
PrincipIE3

(59) P

DP

-qa Pid-u
VP

sallam-a DP

-junu:d-a DP

saltem-a rutab-a-hum.

The lexical verb sallam-a "gave" assigns Acc Case to DP, while the

light verb, v, assigns Acc Case to indirect DP. This analysis squares

in essence with Vergnaud's proposal that there should in principle be


only one argument for each verb, (cf. Vergnaud's USC (1995) Fall

syntax class lectures). This proposal is captured by the following

schema:

(60) I verb >Iargument"

2.5 Clausal categories

hrase
So far, have been able to extend X-bar schema of
we pa
clausal
structureto all phrasal categories, but can we also exten
given in early
categories? To do this, the sentence rule give
Underlying (60) is the Single Complement Hvpo thesis, (cf. Larson (708))
nsformational generative grammar (TGG) would
have to be recast
in terms of X-bar theoretic rules.
But this
requires us to
determine a
head, a
specitier, and a
conplement for S*. For ease of
ansider again the exposition,
example given in (2), repeated for
convenience as
(61):

(61) [NP John) [aux will] [vp finish his homework on


Saturday]
Obviously, this tri-partite division is inconsistent with the
principles of
X-bar theory discussed above. We therefore need to
reassign (61) an
alternative structure which is consistent with
X-bar theoretic
principles. Notice that in (61), the main verb is preceded by a modal
auxiliary, 1.e. will. However, not all sentences contain a modal
auxiliary. We may equally have sentences such as (62a-b) below,

where the NP subject is apparently directly followed by a VP:

(62) a. My daughter likes Tom and Jerry.

b. He published a new book on syntax.

But we still want to analyse a sentence in terms of NP, Aux, and VP

because although a modal auxiliary may be missing, its position, Aux,

should be available to host tense elements such as simple present -s

These elements are inflectional morphemes, and


and simple past -ed.
work TGG, it was assumed that
hence they need a bearer. In early on

the main verb via an Affix Hopping Rule,


These inflections end up on

38The sen (i) below:


entence rule assumed for clauses
was given as

() S NP AUX VP
63
to lower these
Sulixes and reposition tho
hem on the
whose fiunction was

have found a ad for the


head the
sentenee, i.e
n

verb. It seems, then, that we

Aux, because it is a zero level category. The other twom


main

sentence are plhrasal in nature, and henco


constituents of the they
cannot serve as heads. We will therelore reter to Aux as TENSE
or
would then be its complement
(T). for short. The following VP and
the DP subject would be its speeifier. The senlence rule would
u
eventually have to be reanalysed as (63) below:"

(63) a. TP> Spee T°

b. T T VP

Accordingly, the structure associated with (61) is now given in (64)

below:

Chomsky (1995) abandons the AGR node since it makes no semantic co ha


ion
to the sentence. This squares with hat
Culicover and Jackendoff'
"The most
explanatory syntactic theory is one that s (z tructure
ure
the
necessary to mediate between phonology and imputes But see, minimu thers
thers,
Rizzi (1997,
2004), who proposes a more meaning." d toa
proliferation of functional articulated clause stru
categories in the left periphery.
(6-4)
TP

DP T

John VP

will DP V

CJohn V PP

DP on Sat

finish his homework

difference between the subject and


In a clause structure like (64), the
in terms of X-bar format; the
the object is expressed
sister of V and daughter ofV, while
as a
object/complement appears
as a sister of V° and daughter of V'
the (thematic) subject appears

(VP).
clause which
(64) is the structure of a simple clause, i.e. a

such clauses are referred to


stands on its own. In generative literature,
matrix clauses or main clauses. Most of the
variously as root clauses,
for a
Sentences we have seen so far are of this type. But it is possible

65
clause to be embedded inside another
inside another
clause.0ThThis is
simple
(65a&b) below, where the bracketed clas.
s occur as
illustrated in

complements of
the higher/ matrix verbs think
think and wonder,
respectively:

(65) a. I think [that students should read more]

b. I wonder [whether he has arrived]

In ore-P&P versions of Generative Grammar, it was


suggested that the
embedded clauses in (65a-b) are
generated by the following PS-rules:
(66) a. SComp S

b. SNP Aux VP

c. Comp +Q1

Therefore, the partial structure of (65a), for example, was given as


(67) below:

(67) ve think [s [coMP L-O1 that] [s students should read more]|l


It is
easy to see from (67) that S' is
an exocentric projection, i.e. a
projection not headed
by a head of its own
projectio,
of S' is the category. Rather, tne i d
sentence S, and its
argued above that specifier
pecifier is COMP. However, *
heads of
not
phrasal categories. structures must be zero-level
zero-level categories,
cas
be the We may then
n rectify this by taking COMP to
head and the rectify this by takins
In X-bar following
Dwing sentence
theoretic terms, thissentence (i.e. S) to be is omplement

40
Cf. fn. 16. meea
neans that C heads a
functional
saIDerstructure called
complementizer phrase (CP). The
a
PS-rules
gjven in (66) should, therefore, be revised as (68)
below:4
(68) a. CPSpec C"

b. C C TP

Consequently, the bracketed embedded clause in (65a) should now


have the structure (69) below:2

(69) CP

Spec C

Co TP

L- -

that students should read more

1o
In English, embedded CPs may be headed by a nul or overt

complementizer. By contrast, matrix clauses may not be preceded by

that orany other complementizer,


for that matter. The impossibility of

an overt in English root clauses in clearly indicated


complementizer
by the following grammaticality judgements:

41
(63) and (68) reflect the working hypothesis that clause structure is basically
(2) the inflectional domain,
divided into three domains: (1) the lexical domain, VP,
TP and (3) the left periphery, CP.

42 discuss wh-movement below.


We will see what goes into Spec- CP when we

67
read more.
should
students
a.
That
(70)
arrived.
Whether he has
b.

root clauses may be d hby


preceded
an ov
however,
In Arabic,
data illustrate:43
as the following
complementizer,

lafi: naEi:m
1-Pabra:r-a
Pinna
(71)
(71) a.

the-righteous-Acc
in delight
that
will be in delight."
Verily, the righteous

b. Pinna-hum ja:?-u:
that-they.M arrived-they-M

arrived."
"They have indeed

c. Pinna-hunna saEi:d-a:t-un

that-they.F happy-PIF-Nom

"They are indeed happy.

d. ?inna-hu ja:?-a faSl-u SSayf-i

that-clitic came-he season-Nom the-summer-Gen

"The summer (season) has indeed arrived."

But independently of whether the complementizer ?inna 1s us


root or embedded
clauses, it must be immediately followed y
pro)nominal expression to which it assigns Accusative CasC; (71
\
a-C). In cases where there is no adiacent nominal exprco

Arabic is an archetypal example of a


VSO language.
astic cli
pleonastic clitic appears after inna, as in (71d).44/45 Accordingly, the

clauses given in (72a&b) below both have the canonical structure

(73).

(72) a. John will givea talk tomorrow.


b. inna 1-wazi:r-a qad ya-HDuru l-qimmat-a Radan

that the-minister-Acc may 3MS-attend the-summit-Acc tomorrow


That) the minister may attend the summit tomorrow."

44 ither in root or
Ihe
following configuration is not allowed in Standard Arabic,
embedded contexts:

() Pinna Ir V-S- XP]


45
clitic is taken to be a spell
n (2014: 16/17), this invariant pleonastic
Ennassiri this
but nothing really hinges
on

EXICalisation of the Case feature of Pinna,


view at this
stage.
69
(73) CP

Spee

TP

DP T

inna
John T VP

-wazi:r-a

will DP

qad
V NP
John
pro

DP tomorrow

Radan

8Ive a talk

yaHDur -qimmat-a

The clause structure in the


geometry outlined in (73) will be adopted
will
following sections, both for English and Arabic. Such an analysis

allow us to attain
unitary characterisation of the syntax oofclaus
a

nese two
languages, and to posit that all clauses are CPs whic
hav an overt or a
phoneticaly empty
maV seem, this diiference nas
complementizer. But slight as it
important consequences for
contrastive analysis of English and Arabic. a
syntactic

71
EXERCISES

X-bar model draw a tree diaOr


tree
Exercise 2.1: Using the diagram for the
bracketed VPs:

in the attic]
(a) John [vp worked at the job

(b) He [vr goes to the movies on Sunday]


He will vp Completely read the novel]
(c)
(d) Paul [ve put the book on the shelf on Saturday]

(e) ha:dha: al-falla:H-u [vp ya-zaEumu ?anna l-mazrakat-a tu-ntiju

the-farmer-Nom pretend-3MS that the-farm-Acc 3FS-produce


fawa:kih-a jayyidat-an]

fruit-Acc good-Acc

"This farmer pretends that the farm produces good fruits."

Exercise 2.2: Using the X-bar model draw


bracketed NPs:
a tree diagram for tne

(a) NP Your reply to my letter] was too late.


(b) NP The loss of the ship] baffled
everybody.
(c) A tall, dark, ragged stranger] walked into my office.
(d) They remodelled [NP the shop on the
corner|.
(e) INP The
journalist who the policeman arrestedj
the company.
III. CASE THEORY

This chapter will discuss the following topics:


Overt Case vs. abstract Case
Accusative Case in English
Structural conditions on Case assignment in
English
Nominative Case in English
Exceptional Case Marking (ECM) constructions
Case assignment in Arabic

3.1 Overt Case vs. Abstract Case""


Case Theory is one of the modules/subtheories of
Principles
and Parameters theory which is concerned primarily with the
distribution of overt noun phrases. In this sense, it is a filter that
applies at S-structure. It is related to the traditional syntactic ideas of
case, which saw the relationship between elements in a sentence as
being shown by their morphology as well as by word order. Chomsky
(1995: 110-111) characterizes this module thus:
In some languages (Sanskrit, Latin, Russian, .), case IS

morphologically manifested, while in others, it has little (English,

French,.) or no (Chinese,..) overt realization. In line with our


general approach, we assume that Case is always present abstractly. In

Nominative/accusative languages, the subject of a finite clause is

46 involve. Latin, for


Languages differ vis-à-vis the number of overt cases they
example, has six overt cases, German has four, and Arabic has three, (see below).
a Le lIVe.
tax oj
yntax of Engish
Towards

Theory:
P'rinciples and
Parameters
English aar
the object of a transitive verh
assigned nominative Case;
VUsativ Case: (...) and the object
of a pre- or postpositi n

assigned oblique
Case (...). The basic idea of Case theory ore..
Jt of
the investigation of the distribution of overt NPs. thoese
with
morphological content.

For the sake of exposition, consider the tollowing sentencesin E. lish


and Arabic:

(1) a. katab-a zayd-un kita:b-an


wrote-he Zayd-Nom book-Acc

Zayd has written a book."

b. Fred likes frogs.

In (la), the subject zaydun carries Nominative Case -un while the
object kita:ban (book) carries the Accusative (alias Objective) Case -
an."47 This isn't immediately obvious from a sentence like (1b) in

English, where Fred and frogs aren't overtly inflected for the
Nominative/Accusative case distinction. However, if we replace Fred
by an overtly Case-marked pronoun, we require the nominative fom

he, not the


accusative form him; and conversely, if we
replace frogs by
an
overtly Case-marked pronoun, we require the accusative orm
them, not the nominative form 48

(2) a. Hel Him likes frogs.


41
Case is given
45 capital letter to show its
The English
pronominal system: technical use.
NOM we
you
ACC me
he she it they
us
you him her them
GEN my our your
it
his
her itsj their
b.Fred likes them/ they.

It can easily be seen that Case is


morphologically represented in
Arabic since this language is
morphologically rich. In English,
however, Case is abstract, 1.e. not
morphologically represented.
Following Chomsky (1986:74), shall assume that Case "is
we

assigned in a uniform way whether morphologically realized or not."


So conceived, the Case feature is a
property of the Case assigning
element, and so the NP would receive it only through assignment, in a
manner to be made precise below.4

3.2 Accusative Case in English


Consider the following English examples:

(3) a. John heard the news.

b. Paul wandered in theforest.

In (3a) the NP the news is assigned Accusative Case by the verb heard.

In (3b), however, the NP the forest is assigned Accusative Case by the

preposition in. It follows then that transitive verbs and prepositions are

Case assigners.

A very important question to ask at this stage is how is Case

assigned? To answer this question, consider the P-markers of (3a&b)

given in (4) and (5), respectively:

49
that Case is not so much an independent grammatical
Joseph Aoun (1979) argues
category as a pre-condition for theta-role assignment, (cf. Visibility Condition).

75
TP
(4)

Spec

John VP

+Tense

V DP

heard the news


..

oACC

(5) PP

P DP

in the forest

ACE
It seems that the
syntactic relationship between the Case assigner and
the Case marked NP is
that of government, a structural
relationsip that
IS central
for Case assignment as well as for other
aspects or
Orammar, (see below for more
on this ).
Government is defined ao " (6)
below:
(6) Government
a
governs Biff:
M.K. Ennassiri

. a is a
zero-level category
. a C-commands p

The notion of c-{onstituent)


command is turn
defined as in (7):
(7) C-command (cf. Reinhart (1976)
a c-commands B iff:

1. a does not dominate Band


11. The first branching node
Bdoes dominate a
not
dominating a also dominates f
According to clause (i) of the definition of
government (6), the class of
governors is restricted to X categories, i.e. heads. Thus, only V and P
qualify as govenors in (4) and (5). In (4), V governs DP because (i) V
is a head category, and (1i) V c-commands DP. The same relationship
holds between P and its DP complement in (5). By now, it should be

clear to the reader that government subsumes the relationship of


sisterhood between heads and their complements.

3.3 Structural conditions on Case assignment in English

out at this juncture that there are


It should perhaps be pointed
()) inherent Case and (2) structural
two types of Case-assignment:
D-structure and is
at the level of
ase. The former type is assigned
at S-structure
The latter type is assigned
connected to theta-Imarking.
listed below:
under the Case assignment rules

See for example Woolford (2006:111-130).


77
Nominative
if governed by finit
teTSI
.

marked
a. NP is
(8) (8) Accusative if governed by V
b. NP is marked
is governed by D
Oblique/Accusative
N P is marked

marked Genitive
if governed by D
d. NP is

3.3.1 Adjacency requirement

Consider now the following example in English:

(9) John makes frequently mistakes.

() suggests that English makes use of the condition of adjacency.


in

addition to government, in regulating Case assignment. This means that

a head category has to both govern and be adjacent to an NP to be able


to assign it Case. But what about (10) below?

(10) John knocked repeatedly on the door.

In (10), the complement of the verb is a PP, i.e. a category that is not
subject to the Case requirement. In this context, an adverb can intervene
between the verb and its PP
complement, which fact shows that ouy
NPs require a Case feature. This of

an
requirement is interpretea n

S-structure filter that has erative


come to be known in the g
literature as the Case
Filter, (cf. Chomsky
(1981)):

This would mean


that finite T
assigns
gns Nominative Case to Spec-TP in English
NominativeCase to
>pE
(11) The Case Filter

Every DP must be
assigned abstract Case32
It is clear from this
definition that the Case Filter
is restricted to NPs
which have phonetic content, i.e. overt NPs. Null NPs such
as PR0,
and possibly others, are not
subject to the condition expressed by the
Case Filter.

The Case Filter makes an


interesting prediction, namely, if an
NP occurs in a non-Case
position, it cannot exist as a lexical form in a
phrase marker, (cf. Safir (2007) Syntax Notes, Rutgers
University).
This prediction is indeed borne out, the
as
following set of examples
illustrate:

(12) a. It is not possible [Mary to leave]


b. John tries [himself to win]
The embedded subjects in (12a-b), i.e. Mary and himself, fail to meet
the Case Filter given in (11) because the infinitive marker to is not a

Case marker in English. The embedded subjects cannot get Case from
outside because possible is an adjective, and hence not a Case assigner
in English, and try is not an exceptional Case marking verb, (see

below). However, (12a-b) become grammatical if the overt embedded

infinitival subjects are deleted:

(13) a. It is not possible [PRO to leave]

52 DP will be used interchangeably


.Unless otherwise specified, the terms NP and
here.
Principles and Pu

to win]
John
tries [PRO
b. the latter
PRO, the
are
now
filled with being an
positions e
The subject be
to be Case-marked. Its
Case-marked
presence
not
ire to
require
that does
category
IX for more on thic)
more on this).
empty
(see Chapter
required by
EPP,
is probably
prediction is
that if the subiecte
cts of
no less important,
Another,
With Case somehow, then
n we
assOCiated
can get
infinitival clauses
examples such as (12a-b). Thio
overt NPs in
to find
should expect
structures like (14a-b), where the
borne out, given
prediction is also
is now
Case-marked by the prepositional
prepositional
subject position
complementizerfor:

(14) a. It is not possible [for [Mary to leave]]

b. We very much hope [for [John to win]]]

overt NPs 'Mary' and 'John' are licensed


In both (14a) and (14b), the
by the prepositional complementizer for, which sits in C.
It has been pointed out above that the Case Filter applies at Ss

to filter out constructions where overt NPs appear in caseles

positions. However, we will see below that it is possible for an overt


NP which does not have Case at DS to the
move to a positiol *
phrase marker where it can be Case-marked. Passive structui and
aising constructions are two cases in point. In the former the
as
0Dject NP moves to the
subiect position of the same clause, and
atter case, the
subject of the embedded infinitival clause moves to the
matrix
subject position. We will n ent
movem
see below se two
wo that these
anerations apply for no other reason
than to comply with the Case
Filterstated in (11).

3.4 Nominative Case in English

It has been argued above that Accusative Case is


assigned to
an NP by a verb or preposition under government in terms of c-

command. It has also been alluded to the idea that the subject of a
finite clause is assigned Nominative Case by finite T. However, this

poses a problem for the definition of government given in (6) above.

To see this, consider the structure of (l15) given in (16) below:

(15) The postman left.

(16) TP
DP

the postman T VP

[Tense
NOM V

left

defined in terms of c-command,


Obviously, if government is uniquely
then T does not govern the subject NP
in (16), for although T is an X

81
the postman Th
category, it
does not
c-command the DP
his being the
of government
needs to be revised to aco.
fact, the definition

nominative Case assignment.


A posSible way of oing ththis
doino
is by
used the definis:
revising the definition
of C-command in
efinition of
government by making reference to maximal projections er
than
of c-command, m-commnand will be
branching nodes. So instead used:

(17) M-command

a m-commands ß iff:

1. a does not dominate ß and ßB does not dominate a


11. The first dominating a also dominates
maximal projection B
Under this revised definition, T now governs the subject NP because it
m-commands it. The first maximal projection dominating T, i.e. TP,
does indeed dominate the NP the
postman in the subject position;
therefore, T assigns Nominative Case to the postman.

3.5 Exceptional Case Marking (ECM) constructions


It was seen
above that the subject NP receives Nominative case
from T.
However, this is not the case when the clause is e.

Containing a non-finite verb. For the intnuva


sake of exposition, consiaer he
following examples
(18) a. Bill believes [John to be
b. Bill intelligent)
believes [him to be
c. Bill believes intelligent
[he to be
intelligent
d. John; is believed [tt to be intelligent

Now, how come that the embedded subject is


assigned Accusative
Case, not the usual Nominative Case?
The heart of the matter here is
that the
complement clause does not contain a tensed head;
rather, T is
marked as -Tense], and so it does not qualify as a Case
assigner.
Remember that only when T is marked [+Tense] can it Case mark the
subject. But then given the Case Filter stated
above, all NPs must
receive a Case feature. In the GB literature, it has been assumed that in
such constructions as (18a-c), the matrix verb assigns Accusative Case
to the subject of its clausal complement. This explains why (1 8a-b) are
grammatical while (18c) is ungrammatical. Verbs that assign
Accusative Case to an NP that is not their internal argument are called
Exceptional Case Marking (ECM) verbs. The transitive nature of
these verbs is shown independently by the fact that they can take an
NP object, as shown in (19) below:

(19) Bill believes him/them.

ECM verbs such as believe are said to trigger CP-deletion as a lexical


property. This means that their clausal complement can be canonically

realized as either CP or TP, the choice being dependent on the context.

(18d) shows that the matrix verb can no longer assign


Accusative Case to the embedded subject, its being now passivized,

and hence more adjective-like in nature. The only way out for the
to the matrix
to pass the Case filter is
to move
Cmbedded subject

83
where
indeed be Cas
it will indeed Case-markec by
position,
empty
subjeet
on this).
the
VI for
more

matrix T. (see Chapter

in Arabic
assignment
3.6 Case

forms, namely,
Arabic has three major
Case
Nominative,
Nom

indicated
Accusative and Genitive.
These Case torms are
cated by the
vowel suffixes: -u, - a and -i. Unlike English,
Un
following respective
where Case is only a property of NPs, these Case forms may appear at

the end of nouns and adjectives. We can already see here a locus of

syntactic variation between the two languages.

Consider the following example in Arabic:

(20) a. sha:had-a [Np /1-?awla:d-u] [Np 1-EarD-a] Eabra [Np n-na:fidhat-

watched-3MS the-children-Nom the-parade-Acc through the-

window-Gen
"The children have watched the
parade through the window.

53
We
ignore here the verbal
negative words lam and lan jussive and subjunctive Cases, assigned Dy the
Case in dual and (not), as in (i) and (i), of
plural NPs and respectively, and the To
adjectives.
(i) lam ya-njaH zayd-un
not succeed-Jussive Zayd-Nom
"Zayd has not
succeeded."
(i) lan yantaSer-a l-¬aduww-u
not triumph-Subjunctive the-enemy-Nomm
"The enemy
shall not
triumph.
There are three NPS in
(20), and
every one of them is
Aifferent structural Case by a assigned a
different Case
10:fidhat-i (the window) is
assigner. Thus, the NP n-
assigned Genitive Case by the
Sabra (through), -larD-a
(the parade) is preposition
sha:had-a (watch)- the latter being a assigned Case by the verb
transitive verb in Arabic- and
NP subject 1-?awla:d-u is the
assigned Nominative Case by T.
however, that all these cases are Notice,
assigned under government in terms
of strict c-command. The structure
associated with (20) is thus
as (21): given

(21) TP

Spec T

VP

sha:had-a NP1 V

1-Pawla:d-u.. V PP ****

NP2 NP3

1-EarD-a Eabra n-na:fidhat-i

Once also that Arabic patterns with English with regard to the VP-

nternal Subject Hypothesis; it is underlyingly an SVO language, just

85
Tudic
the verb sits
like English. However, in the S-structure the sits under T verb under T for
reasons we shall outline in detail in Chapter V.

As pointed out above, ArabiC adjectives may also be (aase-


marked. This is illustrated by the pieces of data given in (22), where
the adjectives are inflected with the same Case feature as the
d
noun they modify. We will leave open the question of how precisel,

adjectives get associated with their Case features. Suffice it to indicate


here that this can be done via percolation from the modified head
noun.

(22) a. dakhal-a &alayna: rajul-un Rari:b-un

came-in-3MS upon-us man-Nom strange-Nom


A stranger came upon us.

b.qara?t-u riwa:yat-an jami:lat-an


read-IMS novel-Acc good-Acc
"I have read a good novel."
c. ka:na ya-qu:du bi-surEat-in kabi:rat-in
was-3MS 3MS-drive
with-speed-Gen big-Gen
He was
driving very fast."
Let us now
address instances of ECM
For the sake of
constructions in ArabIC
illustration, consider the following
pieces of data:
(23) a. Dhann-a zayd-un
[?anna Eamr-an naja-Haj
believed-3MS Zayd-Nom that
Amr-Acc succeeded-3M>
Zayd believed that Amr had succeeded."

b. Dhann-a zayd-un [Eamr-an naja-Ha]


believed-3MS Zayd-Nom Amr-Acc succeeded-3MS
Zayd believed Amr to have succeeded."

(24) a. Hasibt-u [?anna l-mushkilat-a


Hulla-t]
thought-I that the-problem-Acc resolved-3FS
"I thought that the problem had been resolved."
b. Hasibt-u
[l-mushkilat-a Hulla-t]
thought-I the-problem-Acc resolved-3FS
I thought the problem to have been resolved."

The examples show that in Arabic, the verb Dhanna (believe) and
Hasiba (think) select two types of sentential
complements. The first
type is introduced by the complementizer Panna (that), and the second
type is a
complementizerless complement clause, (cf. Ennassiri
(2014a)). Obviously, in both (23a) and (24a) the complementizer
Panna (that) assigns Accusative Case to the DPs
Eamr-an and 1-
mushkilat-a, which must be sitting in a non-Case marking position.
According to traditional Arabic grammarians, this is a topic position
where Nominative Case is normally assigned by default in the absence
of a structural Case assigner. So, given that the DPs Eamr-an and /-
mushkilat-a follow the complementizer ?anna, we shall assume- for
the time
being- that they are base-generated in [Spec,TP] and that the
entire clausal complement is a CP, (but see 6.5 below for a different

87
On the other hand, (23b) and (24b) are
analysis of topic structures).
ana

instances of Exceptional Case Marking constructions, where


nere the
preverbal DP in the embedded clause is assigned Case by th.

verb. Culicover (1997:66) points out that ECM "allows t


the clause
associated with a verb like "believe to extend into another al
clause, in
the sense that believe'governs not only within its own clause .
t the
subject of the complement." In (236) and (24b), too, the matriv a

governs the preverbal DP in the embedded clause and


assigns
Accusative Case to it. But here, too, the preverbal DP cannot be sittino
in a Case-marking position, for otherwise a Case filter violation would
arise. So, as in (23a) and (24a), the preverbal DP in the embedded
clause in (23b) and (24b) occupies a topic position, the latter being a
non-Case marking A-position. As the topic position is
assigned Acc
Case by the higher verb, it is tempting to conclude that the
entire
clausal complement is a TP and not a CP. We shall assume that
this is
indeed the case.

We would like to
provide
analysis to ECM constructions in
an
Arabic in terms
ofL-marking à la Chomsky (1986b). So, we shall
assume that in
constructions such as (23b) and
verb L-marks the
(24b), the matrix ECM
preverbal DP inside the
complement clause. L
marking is defined as in (25):
(25) L-marking
Where a is a
lexical category,
head of y that is
a L-marks Biff ß agrees wimthe
0-governed by a.
(25)
ats
amounts to saying that if a
category a
L-marks
another category B. it also L-marks the
(theta-governs)
the matrix verbs in
specifier of B. More
specifically,
(23b) and (24b) L-mark the
inside the embedded clauses because preverbal
they L-mark the respective
m
imal categories or whicn these DPs are the specifiers. More
enecifically still, the verbs Dhanna
spe (believe) and Hasiba (think)
Eamr-an and -mushkilat-a,
govern
respectively, through the
nsnarent TP complement, and therefore assign Accusative Case to
them.

54
for a different analysis of ECM
B e,
e for example, Chomsky (1995)
Constructions.
89
EXERCISES

Explain how the sentences bclow are excud


cluded in the
Exercise 3.1:
above.
context of the theory of Case outlined

(a) Bill smiled John.


b) John to fail to attend the meeting.
(c) He believes to understand the explanation.
(d) T hope they to succced.
(e) John was believed that is a liar.
(f) qa:1-a Pinna qa:bal-a Eamr-an?
said-3MS that met-3MS Amr-Acc
He said that (he) met Amr?"
(g) John to like mathematics would
suprise me.

Exercise 3.2: Draw tree


a
diagram for (c) and (f), using the X-bar
model
IV. THETA THEORY

This chapter will discuss the following topics:


Lexical entries and argument structure
Types of semantic roles

A-positions vs. A-bar positions


Theta-positions Vs. theta-bar positions
The Theta Criterion

In this
chapter, we look at the interaction of syntax and the
lexicon via the Projection Principle and the
aspects of meaning
covered by Theta Theory. Anticipating somewhat, Theta Theory is
Concerned with the assignment of thematic (theta/e-) roles to
arguments in theta-positions. The assumption is that every lexical item

which selects one or more complements entertains a thematic relation


with these complements."

4.1 Lexical entries and argument structure

the place where


Earlier, we saw that the lexicon was

information about individual words is stored, i.e. their pronunciation


and
and meaning, inter alia. But as well as providing phonological

on generative
5 book- and other books
"thematic" is abbreviated in this
Word "thematic roles" or
"thematic relations
by the Greek letter theta (). So
dX -relations
dre referred to as "theta/e-roles" and "theta/e

91
Principles and.

the
lexicon also ludes information abou
includes infor.
information,
semantie
is Noun
whether a word a a
Verb, etc. In
syntax,
for example
how a word behaves
ntactically
es..

deseribes
lexical entry
particular the
and phrases. Take for u.
phrases. T'ake
connection with other
words
example the
in

be used in sentences
such as (1), but not in d
those
verb hate, which
can

in (2a-c)
hate noise.
(1) My children

hate.
(2) a. My children
b. Hate noise.
c. My children hate noise violence.

Our knowledge of the verb hate allows us to make these judgements.

We know for example that the act of hating requires one entity to
experience the feeling of hating and a second entity to be hated, i.e.
the two entities involved in the act/situation that the verb hate denotes.
The entities involved in an action are referred to as arguments, and the

elements which say something about these arguments are d

predicates. Predicates express meaning relationships between


arguments, for instance the meaning relationship between the
predicate hute and the our
arguments my children and noise. Part oi
knowledge of the meaning of hate is hat
It
dependent on the knowic
takes two
arguments. This aspect of meaning 1s called
is argumen.
structure.
ng Canc

Knowing the argument structure of a predicate means


wing
now
many arguments are
of

involved and their types


Arabic, for instance, know that in
(3) below the
(narrate) requires three intristic semantic
predicate Haka:
arguments corresponding to
the entity doing the narrating, i.e.
I-jaddat-u (granny), the entity being
narrated, i.e.
Hika:yatan (a story), and the
entity to whom the story is
told, i.e. aHfa:d-i-ha: (her
grandchildren):56
(3) Haka-t 1- jaddat-u
Hika:yat-an li-aHfa:d-i-ha:
narrated-3FS granny-Nom
story-Acc to-grandchildren-Gen-her
"Granny told her grandchildren a
story."
Each of the three entities has a
different semantic role. These semantic
roles are referred to as thematic
roles or 0-roles, for short. Theta
Theory is thus the module of
the grammar that
regulates the
assignment of thematic roles, i.e. the names of the
participant roles
associated with a predicate, to arguments. $7

4.2 Types of Semantic/Thematic Roles


The doer of the action in (3) is Agent 6-role, the thing told
an

bears a 7Theme 0-role, and the individuals enjoying listening to the


story bear a Beneficiary or Goal 6-role. There is no agreement among
linguists as to how many 6-roles there are and their types, but they do

56
The category of V is divided into the following subcategories based on the
argument structure of each specific verb:
i. Intransitive verbs: 1 argument (the subject)
ii. Monotransitive verbs: 2 arguments (the subject & the object)
Ditransitive
object)
verbs: 3 arguments (the subject, indirect object & direct
$7
onike lexical categories such as V, N and A, functional categories, e.g. C, T, NEG
dnd AGR, do not have thematic grids. For instance T does not theta-mark its VP

complement.
on the notion of Theta Theory
and thematie roles,3Thie
This being
agree
that the lexical entry for any
the case, we can postulate predicate
predio.

includes information about the 0-roles that its arguments bear

information is given in terms of a 0-grid. The lexical entry for the


verb
Haka: (narrate) in Arabic, for example, would thus be as in (4) belouw
ow,
where Theme and Beneliciary are the internal theta-roles, and Agent

is the external theta-role:

(4) Haka: Agent, Theme. Beneficiary/Benefactive>

The external theta-role is determined by the verb and its complement,


i.e. compositionally. Verbs like Haka: (narrate) which take three
arguments are called three-place predicates, those like (1) which take
two arguments called
are
wo-place predicates, and those like
ibtasama (smile) which take
argument are called one-place
one

predicates. Arguments can also be propositions, that is, clauses. For


the sake of exposition, consider the
following examples:
(5) John decided that Bill should take over.

(6) Dhann-a l-muttaham-u ?anna l-Hukm-a


Rayru munSif-in
5
What's important here is not so
much precisely what theta-role
bears as merely the fact that
it bears one.
a given nomnd
59
The
distinction between internal and
external theta-roles
hierarchical structure, as in
(i):
is coded in
syntdectic
() Ive NPeext lv V
NPeiml
ibtasam-a TTefl-u
Smiled-3MS the-baby-Nom
"The baby smiled."
thought-3MS the-culprit-Nom that
The
the-verdict-Acc not fair-Gen
culprit thought that the verdict was
not fair."
The verbs decide in
English and Dhanna
(think) in Arabic each takes
two arguments, one of which is a
proposition. Propositions are
arguments in the sense that
they refer to a state of affairs in a
given
world.

We have alluded
above to the fact that the
definition of
individual theta-roles is much less clear in some cases. While some of
the terms adopted are
self-explanatory, e.g. Agent and Theme, others
are much less so. Below, we will content ourselves with
illustrating
the major theta-roles recognized in the literature.
(7) a. The little girl likes the toy.
b. The little girl (Experiencer), the toy (Theme)
(8) a. My mom prepared the dinner for the guests.
b. My mom (Agent), the dinner (Patient), the guests (Benefactive)

(9) a. He left the car in the garage.

b. He (Agent), the car (Theme), the garage (Locative)

(10) a. Ahmad gave the book to Saad.


b. Ahmad (Agent), the book (Theme), Saad (Goal)

(11) a. Mary stole the money from the drawer.

b. Mary (Agent), the money (Theme), the drawer (Source)

(12) a. Zaynab opened the window with a na

95
(7heme), a nail (lns.
b. Zaynab (Agen?),
the window
nstrumental)
The patient theta-role
is generally understood to
imply a.
imply a
changein
theta-role to imply a change in lo
state, and the theme location or
is not clear sometimes whether the t h .
position. However, it -Tole
involved should be characterized as patient or theme. For examnl. .it

is not clear whether the window in (12) should be characterized


as
patient or theme. This is the reason why the term patient has ber
been
dropped, and the term 1heme refers to a change in position or state ar
or
both. A crucial principle of Theta Theory is the Theta Criterion stated

in (13) below:

(13) Theta Criterion

(i) Each argument must be assigned a theta role.

(ii) Each theta role must be assigned to an argument.


The Theta Criterion defined
so bi-unique
expresses a

relationship between the number of arguments and the number o


available theta roles. When Move-
a applies, the moved element
inherits its
theta-role from the trace left at the extraction site. The
moved element and its
trace form a chain which has a unique theta
role. So given the Theta Criterion ust
stated in every (13), Ca
(13) may be violated
either
in a well-formed syntactic by unassigned e-roles or unlicensea DPs (or CPs).
censed
So

Cf. the Chain representation,


Condition in Chomsky both (i) and (ii) must De d.
(1986a) stated beloW:
The Chain Condition
ifC={,, O,) is a maximal
position, and a, its chain, then a, heta
unique Case-marked occupies un
its

position.
and only one
one
theta-role. The
ement ccan only be to a
movement
theta-bar position;
implication of
this is that
wouldould
receive two in theta-roles, otherwise, the chain
34 Different semantic
criterion.0Differer contravention of the theta-
roles are
sVntactic categories.
syntactic categories5 The expressed by different
Lackendoff following quote from Culicover
(2005:176-177) clarifies this statement: and
To a
great extent, the
syntactic category used to
argument can be express a semantic
predicted from the
Archi-objects (including objects, argument's ontological type.
substances, and aggregates) are
invariably expressed as NPs. Places and
Paths are almost
expressed as PPs. invariably
Properties are expressed as APs,
NPs, and the occasional idiomatic PP predicate nominal
(in luck, out of sorts). Situations
(including Events) and Propositions are
expressed as Ss or NPs
(earthquake, concert). Thus when a verb takes a semantic
argument of
a
particular ontological type, the category of the
corresponding
syntactic argument is fairly restricted on general grounds.

Some of the examples given above contain adjuncts, i.e. constituents which are
not selected by the lexical verb, but which provide
optional, extra information. In
(12), for example, the phrase "with a nail is an adjunct. It takes the form of a PP,
whose head, with, has a theta-grid of its own, being able to assign a theta-role
(instrument) to its complement NP, a nail.
64
We have alluded above to the fact that 6-roles are represented in the thematic
grid of verbs. The Theta Criterion (13) explicitly requires that these -roles be
expressed. However, the following example seems to challenge this requirement by
virtue of the fact that the O-role (Patient) fails to be saturated:

() Have you eaten yet?

may be argued here that there is indeed an implicit argument (something eaten)
to which this -role is assigned.
Cf. Chapter 1, footnote 12.
Principles and Parameters

ues
argues
that there
that there hould in
should in
principle
.
he
(1995:61)
Chomsky O-roles: thn
thus, agent is
Further, representation
oj
structural

theme or
o r patient
patien
withn
a
unifom

associaled
with /Spec, VP)
VP],
Iheme

npicall
on."
complement
to V, amdso
roles to arguments
is contingernt
nt on
thematic
assignment
of
The the notion.
defined below, where of
government
the notion of theta theta-role hu a
refers to the
assignment of a

theta-marking simply
argument(s):
predicate to its

14) Theta-government

a theta-governs ß iff:
i) a is zero-level category
a
theta-marks B°0
i) a

11) a and ß are sisters

repeatedly been challenged. To see why,


However, this scenario has

consider the following assumed structure for double object


constructions, (cf. Chapter II (59):

66
Actually, a is said to
theta-mark B if a theta-marks the cupied byBora
trace of B. position occup
P
(15)

DP

-qa: Pid-u VP

Sallam-a DP
V

-junu:d-a V DP
Settam-a 'rutab-a-hum..

In (15), the indirect object 1-jumu:d-a (the soldiers) is theta-marked by


the verb sallam-a (gave) even though it is not its sister. The noted

discrepancy is the result of the fact that the definition given in (14) is
based on assumptions which have for the most part been abandoned
by the P&P model. However, its intuitive content is still retained.

4.3 Grammatical Functions: A-Positions and A'-positions

Grammatical functions (GFs) refer to notions such as subject


functions do not have
adnd object of the sentence. In P&P theory, these
defined as particular
andependent status; rather, they are

is the
of the sentence. The GF subject
COnIigurations in the structure

object is the NP
mmediately dominated by TP, and the GF
verb.
dominated by V', i.e. the complement of the

99
b. VP
TP
(16) a.

V
(NP

NP)
also extends to the GF obJect of preposition
The concept of object n,
dominated by P', as in (17):
defined as the NP immediately

(17) PP

P NP
Here again, the NP is sister to P. The definition of
a
grammatical
functions in terms of configurations proves to be crucial to several

aspects of the theory. It has to be reminded here that we are talking

about the D-structure positions of these GFs. We will see below thal
these GFs can
change their positions via the operation Move a.
Gts are linked to 0-theorv, i.e. there tions
in a
are a set of NP
clause structure in
which we typically find arguments ad»
ssociated
with GFs. The
specifier of TP, for is Wherc the subject
CCurs,
Similarly, the complements
example,
of verbs S are
typical object and prepu
positions.
A-positions (argument Collectively, these positions a
are referred to as

arguments of a
positions),
s), as they
they are
as
are the usual
usuai location for
the

predicate. NP
positions rguments are no
where aB
qenerated are callednon-A-positions or A-bar positions
symbolized as
A'-positions. An example of
non-A-position is the specifier of CP.
The distinction between A- and
non-A-positions is crucial to the
theory of movement, (see below).

4.4 0-positions and 0-bar positions


Also relevant here is the
concept of 6-position, which may be
defined as a
position in a clause structure to which a 0-role is
assigned. Recall that 6-roles are only assigned to sisters. It's quite
important to be able to distinguish between
6-positions and A-
positions. While all 0-positions are
necessarily A-positions, the
opposite is not true, for there are some
A-positions which receive no
theta role. These always concern subject positions. By way of
examples, consider the following sentences:
(18) a. It seems that the students have already left.
b. There seems to be a fly in your soup.

In (18a-b), it and there indeed structural


are
subjects, but they receive
no 6-roles and hence contribute nothing to the meaning of the
structures.""67

67
tt and there
are mere place holders required by EPP.
101
EXERCISES

Exercise 4.1: Discuss the argument


structure
underlined verbs in
ofthe underlined
in Arabic:
the following examples
1-Pama.nat-a ala: s-sama:wa:t-i wa l-?arD-i wa
(a) Pinna: EaraDn-a:

jiba:1-i
that-we offered-1P the-trust-Acc on the-heavens-Gen and the-earth-

Gen and the-mountains-Gen

"Truly, we did offer the trust to the heavens and the earth and the
mountains.

(b) fa-lamma: ?a:saf-u:-na: ?intaqam-na: min-huum


so-whenangered-3MP-us punished-1P from-them
"So, when they angered us, we punished them."

(c) fa-lamma: qaD-a: zayd-un min-ha: walar-an zawwaj-na:-ka-ha:


s0-when accomplished-3MS Zayd-Nom from-her desire-Acc married
IP-you-her
So, when Zayd had divorced her, we gave her to you in
marriage
(d) wa-?amma: bi-niEmat-i Rabb-i-ka
faHaddith
and
with-grace-Gen lord-Gen-your proclaim-2MS
And proclaim the
grace of your lord."

Exercise 4.2 Discuss ana

the Vero announce


explain how the sentenceargument structure of
in (c) is excluded in
the verb
of the
of
The
context
The

theory discussed in this the


chapter. Why is (e) problematic
(a Mary announced
the news to
John.
announced it to John.
(b) Mary
Mary announced it the news to John.
(c)
(d) Mary announced to John that she would resign from her job.

(e) Mary announced it to John that she would resign from her job.
V. HEAD MOVEMENT

discuss the following topics:


This chapter will
Head movement in English
Arabic
> Head movement in
Verb raising versus tense lowering

Verb movement and double object constructions in Arabic

The notion of movement in P&P theory refers to the fact that


some categories move from one position to another in the phrase

marker, i.e. constituent structure tree. All the transformations in the


pre-GB theory have now been reduced to a single rule Move a, where
a is a variable which ranges over all categories. Move a can be simply
defined as in (1) below:
(1) Move a
Move any category anywhere.
There is a sense in which we can say that (1) is unconstrained and 1s

likely to
produce both grammatical and ungrammatical
sentc Ices.
Therefore, we need to impose certain
conditions on Move a to po
t from
overgenerating. This means that we allow Move
to

Overgenerate in principle, but impose


in practice we

Conditions/constraints on its output which will have tne


the effect of

excluding undesirable
representations.
68

68 " Ea
"Each constraint
relation between determines or licenses
s

conforms to all two small pieces.


a small piece of linguistic
aceptable struct i f itt
overain
A
applicable constraints." linguistic str
Culico
icover & Jackendoff 005: 15)
5.1 Head Movement in English
e start by an instance of Move d, viz. head
Consider the following yes-no question: movement.
(2) a. Can you read this word?
b. Will John attend the
meeting?
c. Are you coming?
d. Have your parents visited you recently?
e. Does your boss pay overtime?
To be noticed in (2a-d) is the fact that an
auxiliary, which normally
follows the subject, precedes it here. This phenomenon is called
inversion. We have seen above that there are four levels of
representations: DS, SS, PF and LF. With regard to (2a), for instance,
we can assume that its DS representation is that given in
(3b) below,
where the modal can is in its "normal" position after the subject.
(3) a. You can read this word.

b. LcP C-o lTP you [r can [vp read this word]]]"

Move a applies to (3b) and repositions can before the subject. Recall

that we have posited that the clause structure in English is the one

given in (4) below:


(4) lcP Spec [c C [TP Spec [r T [v? Spec [v V ...]]

59
force of the clause, i.e. yields
it
The+Q
Q feature
feature C indicates the interrogative
on
the has phoneticC content:
In languages, e.g. Japanese, C.a
a) stion meaning. some

Taroo-ga hannin da-tta ka?


Taroo-Nom criminal be-past a
"Was Taroo a criminal?"

105
on
operation in (2a) as
(2a)
follows. The
movement

the
explain
We can now
to the head C
of the comnlen-

head of TP,
i.e. T, moves

landine
entizer phrase
and the
that both
the extraction
ing sites are
CP. Notice
operation is terme.
head
head-to-head
movement
Therefore, the
Dositions.
The SS represen
movement or simply head movement.
ntation is sho
where the original position of can is now flad
ed with the
in (5) below,
trace of can. This follows from the trace theory of movement. es

described in (6)."
(5) [cP 0 [c cani [Tp you IT ti lvPlv read [NP this word]11

(6) Trace Theory


Moved elements leave behind a trace of themselves in the
position from which they have been extracted."

Notice also that the moved element is co-indexed with its trace. This
convention is used to mark the binding relation between a trace and ts
antecedent. We ill see later the role of indexation in the theory or
govemment. But for the time being, make it a habit to co-index a
70
we shall assume here (as is standardly and-
recognised in the
PrinCipie
Parameters
leave this
model) that lexical verbs in English are base-generated in V anu
ever

position, at least in the overt g . perfective have

be, move to T" (and possibly Auxiliary


ana syntax. verbs, e.8: e
progressive
modals such as c) when finite. On the omove to C
hand,

will, can, may, etc.


interrogative structures and base-generated
ase-generated inin T but may
e-g. in are
Du e h e negative
expression is fronted. structures like (i) below,
(See Chapter VIl for more on
thiS).
Never again will I
raise the issue with him. (2014b:2
71 (cf. Ei nassiri

Traces/empty
OCcupied by any categories are are
not

audible
and semantic properties. or
positions in
constituent
tituent structures
which

visible material, but which nevertheless


synta

which ne
mov element and its trace. In
(2e), there is no Aux in T, but the
tense feature hooked therein cannot by itself
move to C. Tense
features are morphological in
nature, and so they need a bearer. In
ciich cases as this, a form of do
inserted in T, (cf. Do-support), and
is

subsequently, T to C applies in the manner sketched above.

You may wonder at this stage why we need to assume that


moved elements leave behind an invisible trace. The answer partlyy
follows from the Projection Principle, which requires that lexical

structure be represented at every syntactic level, i.e. DS, SS and LF.

Notice also that T-to-C movement is a local operation. This follows


from a principle called Head movement constraint (HMC), (cf. Travis

(1984), Chomsky ( 1995).


(7) Head Movement Constraint (HMC)

Movement of an X category a is restricted to the position of a


head B that governs the maximal projection of a.

must move to the next c-


that a head
(7) informally means
structure. To see the
in the clause
Commanding head position up
consider the following example:
Working of (7),
(cf. Ouhalla (1999:284))
(8) How tall will John be?

[c will [Tp John [r tin lvp


be taow tll J ]
(9) hoytall
eens***sssasrr
****

Condition
Minimal Link
(1995: 296)
2
subsumed under Chomsky's
i C may be
(MLC) given below:
operation
Move B,
Minimal Link Condition legitimate
there is no
K only if
A Canraise to target
K.
where ß is closer to
107
(8). Here, the move
(9) is the
S-structure
associated with

head position
novement
TL
of wil
is legitimate,
as it does not
cross a
Therefore,
:/7
th e
however, moving be across
s will leads to
sentence is ruled in. In (10)

ungrammaticality. That is (10) violates HMC.

(10) How tall be John will?


(11) cP how tall [c be [Tp John |r will lvp tbe thow tall ]]ii

5.2 Head Movement in Arabic


Consider now the following set of examples in Arabic:
(12) a.?akal-a zayd-un tuffa:Hat-an
ate-3SM Zayd-Nom apple-Acc
Zayd has eaten an apple
b. nu-sha:hidu l-muba:ra:t-a

3PS-watching the-match-Acc
We are watching the
match."
(12a-6) point out an important
difference between English and Arabl
namely that the former is an SVO
language while the latter is a VsO
anguage, (see Ennassiri
(1998), (2014a) and (2014b) for
Dearing on this diSCUS
issue). Now, if we assume that at
per level of
analysis languages have a
deep
the same
that (12a), for instance, structure, we have to assun
will have
will
has the
following D-structure
(13) TP

Spec

pro T VP

-a
Spec
Zayd-un V NP
Pakal tuffa:Hat-an
We are adopting here the VP-internal
Subject Hypothesis, according
to which the thematic
subject originates in Spec-VP. The question one
should ask here is: how is (12a) derived from an
underlying structure
like (13)? The answer to this will be via head-movement. The analysis
runs as follows. The verb ?akal (eat)
head-adjoins to T to support the
formal tense features hooked therein. These
stong features are affixal
in nature and so
require a lexical item/bearer to support them. This
being the case, the S-structure of (12a) is given in (14) below:
(14) [TP pro [r ?akal-a [vp Zayd-un [y t, tuffa:hat-an]]]]
*****..
he moved verb leaves behind a verbal trace by convention. Notice
that unlike English, which requires the subject to move to Spec-TP,
Arabic does not require subjects to move from their D-structure
position. This is because Arabic is
so basically
language, a V-initial
. a language where the subject can remain in the post-verbal
POSItion without violating any grammatical rule. But then how is it

Sgned Case? With regard to English, we have assumed that the

hematic subject moves to Spec-TP to get its Nominative Case therein,

109
m-command. In Arahi
inflectional head T under bic, we are
i.e. from the
the subject under s t r i .
that T Case governs strict
forced to assume
this below). This proposed analysis accontts for
on
command, (more
Arabic.
the V-initiality in

5.2.1 V-raising vs. T-lowering

We have looked above at the phenomenon of V-movement in


by S-structure, the verb will
Arabic. We have hitherto assumed that
to T to inflect for tense. This was
was
already have been adjoined
proposed in order to account for word order facts in this language.
moves to T for
Arabic, then, patterns with French, where the verb also
the same reason. To illustrate this, consider the following examples:

(15) a. Les enfants ont déjà mangé.

b. Jean embrasse souvent Marie.

c. Jean souvent embrasse Marie.

(16) a. The children have already eaten.


b. John kisses often Mary.
c. John ofien kisses Mary.
and and
) (16) illustrate a phenomenon dealt with in Pollock(196
Chomsky (1988), namely that the verb tes with
the
tense

amalgamates wiu
inflection in former
different ways in French and English.
eas in the
language, this amalgamation is done
latter
through V-raising, w
language, it is done via
W e r i n g . H o w e v e r , C h o m s k y

T-lowering. HOw
1988:5) points out that it is not so much
V-raising vs. T-lowering that
A:fferentiates French and English as some other
reauires verbs to raise in the
principle of UG which
former language but
which bars this
ontion in the latter, (cf. (156) vs.
(166))." We will follow here the
working assumption that tense morphology is
strong in French, but
weak in English. We will also assume that
the verbal element in both
languages enters the computation with the tense suffix
already
attached to it. However, for checking purposes, the verbal head needs
to move up to T to eliminate the formal features hooked
therein. The
strength of T in French requires that V raise overtly, whereas the
weakness of T in English restricts this movement to LF. This follows
from an economy principle that favours covert/LF movement
over

overt ones. Put differently, overt movement operations apply only


when they have to. (15a), for example, has the following derivation,
where the verbal head is now related to a higher position in the phrase
marker. For surface structure purposes (e.g. word order), the higher

copy in the tree diagram is considered to be primary:

in the
adjoins to V
manne
13 adverbials
assumed here that the italicised and
nave verb-adverb order reflects V-to-T raising,
indid in (i) below. Therefore, the
(cf. Affix Hopping,
Chomsky (1957))
E
adverb-verb order reflects T-to-V lowering,

souvent/often I V OBJ]]]
() TP Spec IT° Ivp SUBJ [y

111
-
, -
*********
.*ss.
s
**************
-
TP
(18)

DP T

the children T VP
les enfants

have DP
tont

Tchildren AdvP V
Ules enfants

already I+Aux) VP
deja
- - thave
V
tont

V
eaten
mangé

ne movement of Aux to T depicted above is triggered by the same

reason as that of V to T, i.e. by the need to check/eliminate the strong

V-feature associated with T.


Consider now the following examples in English:
(19) a. Bill likes not Mary.
b. Bill not likes Mary.
C. Bill does not like Mary.

113
across Neg in the manne
manner
main verb has
raised
outline
In (19a), the barred in English, hence
this possibility is the
but
in (20),
sentence.
ungrammaticality
of the

(20) TP

DP T

Bill T NegP

likes Neg VP

not V DP

ty Mary

In
(195),the tense inflection adjoins to the verb in accordance winthe
analysis sketched above for English. But curiously enough, (
apparent controversy'
also out. How then can we account for this apparent conro
ative

Obviously, the new factor in (196) is the presence or


ead, not, which
prevents the verb from inflecting for tens have

Apparently, in contexts of n,
sentential negation, Engi
English must
upport t h
recourse to an idiosyncratic Do-insertion
mechanD"
affixal tense feature. Should this not
apply
device
language-speci
-specific
affix would remain stranded at the
appropriate interface level
hout interpretation, tnus
resulting in
ungrammatical sentences 74
There is still a
problem with regard to
examples like (16a),
here T lowering to V results in an
improper chain- by dint of the fact
hat the lowered T fails to
antecedent-govern its trace- and hence in
ECP violation. In fact, this is true for all
lowering rules. The strategy
used by Chomsky (1988) to circumvent this ECP violation is to
assume that the complex verbal element [v V+T] raises back to T at
LF, whereupon the newly-created trace would be antecedent-
governed. But if this salvaging device works for English, why is (15c)
in French not analysed in a similar fashion, i.e. in terms of lowering in
the overt syntax and then raising at LF? Surely, the result would be the

raising of the verb. That would indeed


same as that reached by overt

contrary to the least effort


be the case if this analysis would not run

those involving fewer


principle, which favours shorter derivations-
at
those involving more steps. In the case
steps- over longer ones-
the
in the overt syntax and subsequent raising of
hand, lowering of T
involve two steps.
element at LF would obviously
complex verbal
be longer than the one which
derivation would
Consequently, this

This is so
unproblematic
in English.
X categories
4 below are
that sentences like (i) etc., are not
barely, hoardly,
OtIce
they do
like never, VP. Therefore,
negative adverbs adjoined to
duse categories
to be XP
ra ey are assumed

lot
not interfere with T lowering

(i) Bill never liked Mary.

Cf. Vil (33).


115
P r i n c i p l e s a n d P a r a m e t e r s

of T.
The derivation is doubly
former derivatin.
overt raising
involves only and ECP.
effort principle
the least
excluded by Tense ak in
is weak present-da
n
mentioned
above that
It has been
V-raising until
LF. Notice that..
fact relegates covert
English, which
English 1S reminiscent of
F
of V to T in present-day
movement
in, say Mandarin Chinese. (ce
movement of wh-elements to Spec-CP

and Audrey Li (1993: 191), (see below for more on this).. In


Aoun

Elizabethan English, however, Tense was strong, which fact triggers


movement ofthe verb from V to T in examples such as (21) below:

(21) a. Saw you my master? (Speed, Two Gentlemen of Verona, I.i)


b. Speakest thou in sober meaning? (Orlando, As you Like lt,

V.ii)
C. Know you not the cause?
(Tranio, Taming of the Shrew,
IV.ii) (Radford (2003: 117)
The derivation of (2la), for example, is thus shown in (22) below
where know moves in
a
dl
Successive-cyclic fashion, i.e. in two

movement operations, and where the verb-initial oru


results from
V-to-T-to-C movement:
(22) CP

Spec C

TP
Saw

DP T

you VP

Saw DP V'

you V DP

my master

(22) is identical in relevant aspects to the structure of French


examples
such as (15b) given in
(17) above, modulo, of course, the movement
of the
complex verbal element [r [V-T]] to C. Now, if checking is
ndeed contingent on raising processes, then we would have to

aSSume, as is natural, that in present-day English the verb obligatorily


unaergoes a raising operation at LF in order to check, and hence

eliminate, strong V-features associated with T. Otherwise, these


Catures would appear without interpretation at LF and eventually
u s e the structure to crash. There is a sense, then, in which we can

117
Principle
lack thereaf
T- o r the in the three
of V and
strength D.
that the
say

is responsible
for
movement
parametrization

Perhap We
languages that croso
push this
generalization
further and assume
cross-linguistic
e...
can
to the parametrization of
reduced- at
least in part- ural
variation is
we shall see below that sln
languages. an
properties in particular
results with regard to wh-questione in
yields the desired
assumption

Arabic.

5.2.2 Verb movement and Double Object constructions in

Arabic
which select two
Arabic has a class of di-transitive verbs
and (2) an outer or direct object.
objects: (1) an inner or indirect object
This is illustrated by the following examples:

(23) a. manaH-a l-qa:?id-u l-jundiyya-a wisa:m-an


the-soldier-Acc medal-Acc
gave-3SM the-leader-Nom

The leader gave the soldier a medal.

b. sallamt-u l-qa:Diyy-a l-eaqd-a


handed-1S the-judge-Acc the-contract-Acc
"I handed the
judge the contract."
C.
zawwaj-a l-sha:bb-a ?ibnat-a-hu
married-3SM the-young man-Acc daughter- ACC is
"He married the
young man his daughter.
This subsection is
basically a summary of Ennassiri (2004)
hove examples also have thematic "dative"
paraphrases where
he direct object iS adjacent to the verb and the indirect object is
contained inside a PP.

(24) a. manaH-a l-qa:?id-u wisa:m-an [pp li-l-jundiyy-il

gave-3SM the-leader-Nom medal-Acc to-the-soldier-Gen


The leader gave a medal to the soldier."

b. sallamt-u l-eaqd-a [Pp li-l-qa:D-i:]


handed-1S the-contract-Acc to-the-judge-Gen
"I handed the contract to the judge."

C. zawwaj-a ?ibnat-a-hu [Pp li-l-sha:bb-i]


married-3SM daughter-Acc-his to-the-young man-Gen
He married his daughter to the young man."
An important question arises here, viz. what is the constituent
structure of the VP in sentences such as those given in (23)? A priori,
Two possibilities present themselves, namely (25) and (26):

(25) VP

Spec

V NP1 NP2

119
VP
(26)

V
Spec

NP2

V NP1

represents the standard view according to which verbs such a


(25)
manaHagive', atTa: 'give', sallama "hand', zawwaja 'marry' are
three-place predicates. Two of their arguments are assigned internally,

whereas the third argument is assigned externally. An obvious


drawback of (25) is that it blatantly violates the Binary Branching
Hypothesis, (cf. Kayne (1984). As for (26), it presents problems of a
different nature. To see what is at stake, consider the following
example:
(27) arayt-u zaynab-a nafs-a-ha:
showed-1S Zaynab-Acc self-Acc-her
"I showed Zaynab herself."
If (26) were indeed the structure
constructions, it would
associated with
with double
double object
given that the wrongly predict (27) to be ug ingrammatical

indirect object would and


not be in
position
aposition to c-com
anaphor najasaha: "nerseIT. In fact, the
77
opposite picture would
arise.

Larson (1988) adopts a different analysis, according to which


the verb and the indirect object form a V at the level of
D-structure.
This V' is in turn predicted of the direct
object. Under this analysis,
the D-structure assOCiated With, say,
(24a) would be given as (28)
below, irrelevant details omitted:

(28) VP

Spec V

VP

NP

Wisa:m PP

manaHaa li-1-jundiyyi

that all internal


t 1s easy to observe here that the working hypothesis
to
constituents of the head
seems
cmatic roles are assigned to sister
Branching
dVe been abandoned. This may be due to the Binary

77
See Chapter VIll for more on this.

121
PTlCpio u Arabic

ead. itit has been


(1bid). Instead, been assumed
Condition, (cf. Kayne in the
literature that some
internal thematic
roles assigned to non-sisters
are assioned

& Lasnik (1991). With regard


regard to
to double object
(cf. Chomsky
(ibid) argues that they can be
constructions,
Larson
derived
transformationally from
dative constructions. In this respect, Larsor
respect

(1978), Dowty (1979), Green (1974)


departs from Allerton Hawkins
(1976), where it 1S assumed that the relatine
(1981) and Oehrle
between dative constructions and double object constructions is

transformational." Larson (1988) argues that the


lexical rather than
transformational approach to double object constructions is forced by

the Uniformity of Theta Assignment Hypothesis (UTAH), due to

Baker (1988):

(29) The Uniformity of Theta Assignment Hypothesis

Identical thematic relationships are represented by identical

structural relations between the items at the level of Ds.

analysis
"However, this correspondence is not sufficiently systematic to warrdn
for
in which one alternant is transformationally derived from the other. The reason for reason
this is that f ind
there are exceptions in both directions, i.e. there are
case o f

object sentences with no


prepositional counterparts, as in ()
prepositional sentences with no indirect object counterparts, DEt cee Colins
Collins
as in
(1995) for discussion):
) a.
Mary spared Bill the trouble.
b.
"Mary spared the trouble to/for Bill.
(ii) a.
Mary spoke a few words
b. to Bill.
*Mary spoke Bill a few words
hiously, the relationship between dative constructions
such as those
(24a-c) will be lost under the lexicalist approach.
However, Larson
ihid) makes a number of provisos, chief of which is
the idea that V.
AID NP has a
passive-like structure. Further, he assumes that the
sttbiect thematic role is not suppressed but assigned in an adjunct
configuration. This is expressed as (30) below:

(30) Argument Demotion

If X is a theta-role assigned by Y', then X may be assigned (up

to optionality) to an adjunct of Y'

According to this proposal, the double object counterpart of (24a) is


passivized, the
derived in the following way. With the verb now

is theta-bar
object cannot get Case, and the specifier position
a
indirect
that
This is reminiscent of Burzio's (1986) generalization
position.
Case from an object and
involve withdrawal of
structures
passive
Under the
the thematic role
from a subject position.
suppression of
realized as an

Demotion proposal,
the direct object is
Argument
theta-bar spec
m o v e s to the
indirect object
adjunct of V'. The
the direct object. Finally, the
c-commands
where it
position, from order to assign
Case
V-position in
1Ower V moves to the
higher empty in the
summarized
technicalities are
These
to the indirect obiect. of (24a):
representations
S-structure
D- and
OOWing respective
(31) VP

Spec V'

V VP

NP

V' NP

NP wisa:m

manaHa 1-jundiyya

Reanalysis applies to (31) and recategorizes V as V. The latter rule is


triggered whenever V has a thematic structure with an unsaturated
internal argument. The ultimate S-structure representation of (24a) 1s
then given as (32):
VP
(32)

Spec

V VP

manaH-a; NP V

1-jundiyyak V NP

WIsa:m-an

D0, 1t seems that Larson's (1988) analysis makes the right prediction
somewhat
with
wIth regard
regard to Arabic as well. Yet, this analysis is

(1991) & (1992)). Obviously,a


nbersome, or costly, (cf. Chomsky
most welcome. It
is to this approach
COsty approach would be
that we turn
below.
no real
argued that there is
n Ennassiri (2014b:53), it has been
hence
in Arabic, and
trig
TOr object m o v e m e n t (to ISpec, AGROP])
this
the assumed mechanism is unavailable in
Spec-head relation
then what
this is correct,
language, cast in the overt syntax. If
is how double obiect
can.

remains to be
established
tions are
derived in Arabic.
To do this, let us first
msky's
adopt Chomsky's (1995) VP
(10o

Chapter I):
shell given below, (cf.

vP
(33)

DP

VP

DP2 V

1-jundiyya V DP3

manaHa WIsa:man

ed by
Notice that the above VP shell is similar to the one propose
arson's

Larson (1988). The only difference lies in the fact that under La

(bid) analysis, the lower V substitutes for an empty posi verb

under Chomsky's analysis, V adjoins to v, the latter being


n g features. Th The
that is present in the numeration and, hence, has strong v e r d

Object, then, enters into a checking relation directly w1tn p o s a l ,

or the
complex [, V [vI1. This analysis supports Vergua t h a tt h e r e

captured in II (60), repeated below in (34) for


conve
principle, be only
should, in
one
argument for each verb, (cf.
Vergnaud's (1995) Fall class lectures, USC):

I verb I argument
(34)

Under this proposal a verb such as manaHa (give) would count as

three verbs or three atoms, to use Vergnaud's own words, represented

in clause structure in terms of the algorithm of chains. If this is


correct, notions such as ditransitive and multitransitive verbs would

merely be used as taxonomic or descriptive tems. Under this

conception, the derivation of double object constructions such as

(23a), for example, is outlined in (35) below (irrelevant details


omitted):

overt
in the
79
the schema in (34) applies
We have to is obvious, that where the light
assu assu
me, as
of the structure in (35),
one
nis would validate the adoption one and only
Var have each
verb, ,
dnd the lower (main/ lexical?)
arat
argument.
Prnciples

TP
(35)

vP

DP

1-qa:?id V VP

DP2

1-jundiyya V DP3

manaHa wisa:man

Case

In
(35),the indirect and direct objects are assigned structura
respectively. Under

by the light verb, v, and the lower verb, V, respoe assign


able to
minimality assumptions, the light verb, v, will not be
De a c t u a l

Case to the outer in


object as there is a potential, and a ther

Case is on
governor, namely V. If the here
analysis sketcncu o b j e c t

ot
track, then it is of
double

easy to see that the n


derivao than the one

Constructions sketched
above is more ecOno
by Larson (1988) in that it
propose

requires only the verb and the


matic
thematic subject to move higher in the clause structure for
nhological reasons. AbOve all, no
unnecessary movements or extra
nrOvisos are required here. This
pro
analysis also makes the
right
nrediction with regard to
examples such as (27), repeated below as
(36) for the indirect (inner) object will
always c-command the direct
(outer) object. Hence, no violation of either Binding Condition A or C
will arise.

(36) ?arayt-u zaynab-a nafs-a-ha:


showed-1S Zaynab-Acc self-Acc-her
"Ishowed Zaynab herself."

Notice also that the structure proposed for double object constructions

in Arabic assigns the indirect and direct objects a different status in

which fact predicts that they behave


terms of structural position,
differently with respect to certain syntactic phenomena. This
be
correct for whereas the indirect object may
prediction is, indeed,
may not, (cf. Ouhalla (1994).
passivized freely, the direct object
Consider the following examples:
TTa:lib-u l-kita:b-a
(37) a. ?ueTiyya
was-given the-student-Nom the-book-Acc

The student was given the book."


80
1-kita:b-u TTa:lib-a
b. PueTiyya
was-given the-book-Nom the-student-Acc

Lit. "The book was given the student.

Standard Arabic
status in
a marginal
80 has at best
that (37b)
ldlla (1994)
argues

hence the question mark.

129
where onl
conclusion
extends to (38a-b), the indirect
The same

idiomatically
as a clitic on the
the verb, (cf. Ouhalla
verh

object may appear

(ibid: 59):
1-kita:b-a
(38) a. ?asTayt-u-hu
gave-1S-him the-book-Acc

book."
"I gave him the

b. Pae Tayt-u-hu TTa:lib-a


gave-1S-it the-student-Acc

Lit. "I gave it the student."

If both the indirect and direct objects had the same structural position,
this distinction in grammaticality judgement would go unaccounted
for, as it would under an analysis like (39) below, where both objects
occupy the Spec position of an AGRO.

(39) AGRP

Spec AGR

NP AGR AGRPP

Spec AGR

NP2 AGR° VP

V
EXERCISES

1
Exercise :
5.1: Draw trees for the
following sentences. Indicate what
onsformations- if any- have derived these trees.

(a) We don't usually eat after 7 pm.

(b) We have always liked country-life.

(c) Jean n'aime pas la ville.

Jean NEG like NEG the city


"Jean does not like the city.
(d) lam ya-jaH zayd-un

not 3MS-succeed Zayd-Nom

Zayd has not succeeded."

(e) Veux-tu une pomme?

want-you one apple

"Do you want an apple?"


( Do you want a break?

Have you seen this film before?


VI. NP MOVEMENT

discuss the following topics:


This chapter will
Passive structures
Raising constructions

Passivization and raising interaction


Arabic
Raising constructions in

It has been seen above that Move a is a meta-rule which

subsumes all the movement operations in P&P theory. In addition. it is

an
an unconstrained rule which generates both grammatical and

ungrammatical sentences. This state of affairs is of course undesirable.

Hence, its output is subject to various constraints. With regard to head


movement, we saw that the output of this rule is constrained by the
head movement constraint (HMC), which bans movement of a head

Over another head.

6.1 Passive structures


In satisty
Chapter II, we have seen that syntactic strucurc
lexical this
properties of heads of which they are projectio11
namely NP-
chapter, weshall deal with another
instance of Move d
movement, which seems to begin
superficially disturb this patc"
with, consider the
following pair of exanmples:
(1) a. The

b.
policeman killed the wild bear.
The wild bear
was killed.
We have seen above that the
Projection Principle
entries to project onto all requires lexical
syntactic levels. So, the verb
kill has the
following lexical entry:

(2) kill: V, [--- NP] <Agent, Theme>


shows that kill c-selects an NP and s-selects two 0-roles: Agent
and Theme. (la) satisties the Projection Principle, but (1b) seems to
contravene it. However, (1b) is a grammatical sentence in
English. WWe
can now account for the apparent violation as follows. The DS of (1b)
is the one given in (3), where the verb kill is indeed followed by an

NP:

(3) TP

Spec T

VP

was Spec V

V DP

bear
killed the wild

133
C

to (3) and repositions the DP the wild ho.


Move a applies ear in Spec-
st be
TP. In P&P theory, all movement operations must be motis.
motivated,
there must a trigger for each instance of Move a. In the case
ie,

the trigger for movement relates to the Case Filter, which


nand,
ich ren.
requires that
overt NPs be Case-marked. The empty position e in (3) satisfies #h
e
EPP and provides a landing site for the object NP. The S-struchuwrs

resulting from movement is then given in (4) below, where the moved

element leaves a co-indexed trace in its extraction site:

4) TP

Spec

the wild bear; T VP

was
Spec N

V NP

killed

DP
Passivization is thus an instance that a

of A-movement i
nse

Therefore the
moves from an
A-position to another A-posiuo
resulting chain is called an A chain, i.e. (the wild bear, f).
However,
movemetnt is from a
theta-position to a theta-bar position. This
reauirement follows from the Theta
Criterion, (see previous section).
Consider now the following example in Arabic:

(5) Pukil-at t-tuffa:Hat-u

eaten-3FS the-apple-Nom
"The apple has been eaten."

Prima facie, (5) seems to indicate that the passive morphology does
not trigger movement in Arabic, and so the object DP tuffa:Hat-u (the

apple) may remain in situ. However, the object DP must move to a

position where it will be assigned Case. This is so because in Arabic,

feature of the verb, and


too, the passive morphology absorbs the Case
is
so the latter is unable to Case-mark the following DP. This position
moved DP will be assigned Nominative Case
Spec-VP, where the
from T under government. This is perhaps the reason why the subject

of passive structures is called na:?ib 1-fa:Eil (substitute for the subject)


although the S-structure subject of passive
in Arabic grammar,
agentive function. If these
constructions does not have any
outlined in
is derived in the
manner

assumptions are correct, then (5)


(6) below:

135
(6) TP

Spec

pro T VP

Pukil-at; Spec

ttuffa:Hat-u, V DP

********e*o****

So, it seems that the structure geometry of passive constructions in


Arabic is similar- in relevant respects- to that of active constructions,
in both
constructions, T assigns Nom Case to a following nomina
expression, (cf. Ennassiri (2014a)).
Or does it really? It may be argued here that Spec-VP is a

theta-position given that thematic subjects originate therein, pe aps


cross-linguistically, and so the DP ttuffa:Hat (the apple) cannot
ove

there. This prohibition would follow from the fact that this movement

would result in violation


of the Theta Criterion as the resulting
would have two , i.e.
-roles, one associated with the head of tne
the moved object
DP, and one with the trace left behind in

position. But recall that


effect of the verb
an
the passive morpnei
rpheme on

t the
Pukilat (was eaten) is to suppress its
external -role and
*
hility toassign AcCusalve a s e to the
following DP. These two twin
tions make
assumptions make it possible for the object DP to
move to the now
thematised Spec-VP posiion to get Case from T, exactly as
nOsed above. Perhaps it should be added at this
juncture that the
verb enters the derivation already inflected for the
passive
morphology, but it still needs to raise to T to
further inflect for the
tense features hooked therein8

English allows passives with no


NP-movement, such as (7)
below, where the subject position is filled with the pleonastic element

it. We argued in Chapter 4 that pleonastic elements are required by


EPP
(7) It was thought [cp that the policeman killed the wild bear]

In Arabic, too, the empty small pro may be phonologically realised if


it comes immediately after the complementizers ?inna or Panna (that):

(8) a. Pinna-hu ?ukil-at ttuffa:Ha-tu

that-it eaten-3FS the-apple-Nom

"The apple has indeed been eaten."

1-¬aduww-u
b. Hasibt-u ?anna-hu huzim-a
the-enemy-Nom
thought-I that-it defeated-3SM
had been defeated."
T thought that the enemy

lexical entry:
81 the following
It may be ned that ?ukil (was
eaten) has
assume
?ukil: V, - DP]
-0-subject>
137
have assumed
above the pleonasti
that the pleonastic pronomin
we
Recall that Case-feature
phonetic spell-out of the iated with
clitic -hu is
a

complementizers
Pinna and 82
Panna. Perhaps
Perha it is
the Arabic
more
consider it as a salvaging device that nmd
st apply to
convenient to

rescue
where
constructions
these complementizers are not
by a nominal expression.
immediately followed

6.2 Raising constructions

NP-movement is also involved in the following constructions,

which are referred to in the literature as Raising Constructions on


account of the fact that the subject of the embedded clause in raised to

the subject position of the main clause.

9) a. Paul seems to have a lot of syntax books.

b. Jillian seems to have read the novel.

The gist of raising constructions is that the raised italicised DP has a

grammatical function in the matrix clause and a thematic role in

embedded clause. This is clearly indicated in the following


resp
paraphrases of (9a) and (9b):
(10) a. It seems that Paul
has a lot of syntax boOK
b. It seems
that Jillian has read the novel.

inArabic, both ?inna and


can
introduce a root Panna are embedded clause
clause, (cf. II (71a-d). markel
Callowing generalisations will be made with
regard to (9) and
(10)

a. In English, seem does not assign structural Case to a


complement as a lexical property.

h. The subject of seem is not a


6-position. This is obvious in
(10a-b), where the subject positions are filled with the
pleonastic pronominal it.
C. seem is subcategorized as allowing insertion in a VP followed

by a clausal complement.
d. seem has one -role to assign (Proposition), which it assigns
to its CP complement.

Given property (a) above, the following example is ruled out:

(11) Paul seems that he has a lot of syntax books.

It follows then that the DS phrase marker associated with, say, (10a) is
where the subject position is empty and hence
the one given in (12),
can serve as a landing site for a moved element:

(12) e seems [TP Paul to like syntax books]


the
F a
well-formed structure in English,
) to surface as
in search
to the matrix subject position
nbedded subiect must move
architecture
SS
Therefore, the
A S e , Via NP-movement, of course.
associated with (9a) is given in (13):
(13) TP

DP T

VP
Pau

-Past] V TP

seems DP T

t, T VP

to like syntax books

and appear and


Raising predicates include verbs such as seem

adjectives such as be likely and be certain.

(14) a. John is likely to solve the problem.


b. John is certain to win.

c. John appears to like the place.

6.3 Passivization and raising interaction


the
Lonsider now the
following example. which illustra
interaction between
passivization and raising
(15) The cake seems to
have been burnt.
Che DS representation associated with (15) is given in (16) beloW:

e seems [TP2 e to have been burnt the cakel


(16) [TPI
Notice that the clausal complement of seems in (16) is a passive

structure. It has been argued above that the DP object in passive

structures must move to the subject position within the same clause

for it to be assigned Nominative case, (cf. (2). But in (16) even this
position is a non-Case assigning position given that TP is an
infinitival clause, where T is marked [-Tense]. Consequently, the DP
the cake needs to move to a higher position to satisfy the Case

requirements. The matrix subject position seems to be such a position.

It follows then that the SS associated with (15) is given in (17):

(17) ITP The cake; seems [TP2 ti to have been burnt t

(2) (1)

This results in the following chain, where the intermediate trace is


marked as .83

(18) (the cakei, ti , ti)

The initial trace, , is properly governed by the intermediate trace, t',


DP the
and the latter trace is in turn properly governed by the moved
Cake. In (17), then, NP-movement operates in two steps. The first step

33 an element takes up in
chain indicates the history of movement, i.e. the path
he
(landing site) is a chain link. The
r e e as a result of Move a. Each targeted point site- constitutes the foot/tail
dCe- which occupies the original position/extraction
the landing site) constitutes the
the moved element (sitting in
chain; and
E
head of the chain.

141
is motivated by EPP, and the second step 1s motivated by Case tha.
eory.
Because it operates in a ement has
stepwise fashion, NP-movement has been
.

described in the literature as a local movement. The localitycondit


was also observed by head-movement in Chapter V, whereexample
which involved too long a movement were deemed ungrammatical,

(cf. V (11)).

6.4 Raising constructions in Arabic

Consider now the following raising constructions in Arabic:

(19) a. yabdu: Panna zaynab-a waSal-at

seem-3MS that Zaynab- Acc arrived-3FS

It seems that Zaynab has arrived."

b. yabdu: Panna l-mu&allima:t-i waSal-na

seem-3MS that the-school mistresses-Acc arrived-3FP

It seems that the school mistresses have arrived."

c. yabdu: Panna l-junu:d-a ?intaSar-u:

seem-3MS that the-soldiers-Acc triumphed-3MP


It seems that the soldiers have triumphed."

d. Pinna-hu yabdu: Panna mu:sku: faqad-at ssayTarat-a tala:

that-it seem-3MS that Moscow lost-3FS the-control-Acc 0


-muwa:li:n-a la-ha: fi: ukra:nya:
the-allies-Gen to-her in Ukraine

It seems that Moscow has lost control over its allics in Ukraine."

The fact that the embedded complements in (19a-d) begin with the

complementizer ?anna (that) strengthens the case for taking yabdh

(seem) to c-sclect a CP in Arabic. Thus, the structurc associated with.

say (19a). is given in (20) below, which features the movement of the

matrix verb, but not that of the embedded subject:

(20) TP

DP T

pro T VP

yabdu:; V CP

C TP

L
Panna Zaynab-a waSal-at

AS mentioned above, the pleonastic empty pronomnal pro may have

if c-commanded by the complementizer Pinna


pnonological content
(that). This seems to also work for seem-type verbs in Arabie
(cf
(19d).4

(21) CP

TP

Pinna-hu DP

Upro T VP

yabdu: V CP

C TP

Panna mu:sku: faqad-at...

To be noticed in (19a-c) is the fact that the verb yabdu: (seem)


agrees with its third person singular empty pronominal subject in
person, number and gender. This mimics the agreement exhibited

between the pleonastic it and seems in English, (cf. (10a-b)). On the


other hand, the embedded complement clauses have their oWn

subject/topic; the latter agrees with the embedded verb in all


features.
nis does not ure
contradict the proposal that -hu is a spell-out of the Case Tea
associated with ?inna (that).
M . K . E n n a s s i r i

Rtt consider now the


roilowng more
interesting examples in
Arabic:

a. vabdu: 1-Eumma:l-u munhamik-i:na fi:


(22) Eamal-i-him
seem-3MS the-workers-Nom busy-Acc in work-Gen-their
The workers seem to be busy with their work."

b. yabdu: zayd-un fi: maktab-i-hi

seem-3MS Zayd-Nom in office-Gen-his

Zayd seems to be in his office."

On the face of it, (22a-b) do not seem to exhibit any movement

operation. But a closer consideration of the structure associated with


each of them reveals that there is indeed movement of 1-Eumma:l-u
(the workers) in (22a) and Zayd in (22b). We will follow the analysis

to which small clausal


proposed in Ennassiri (2014a), according
AGRPs whose head, AGR, encodes
complements such as (22a-b) are

Accordingly, then, the


structure
features.
number and gender
below, where the
associated with (22a), for instance, is given in (23)

indicesare identified, i.e. i =k:


PIIIl'

(23) TP

T VP

Spec V
[+Tense]

AGRP

yabdu: Spec AGR

1-Eumma:l-uk AGR AP

munhamiki:na, Spec A

A PP

t fi:Eamalihim

In (23), the overt agreement between the subject an0


1
adjective predicate is the reflex of
the relation holding between tne
specifier and head of AGRP.
The AP subject 1-&umma:-u the
Workers) raises at
S-structure, targetting the complement AUn hd

becoming its specifier. In this position, (the


the NP 1-Eumnm
Workers) should be
accessible to Case-marking from
an outside

governor, which in this case (seem). The


is the matrix verb yabau:
munhaniki:na (busy) likewise
predicate
moves to
ical reasons, Viz. AGR for
to
morphe support number and
gender features:
a
features would otherwise surface
without interpretation at
interface level of L . This is conceptually supported by the fact
aat head movement is
essentially a process of syntactic
affixation, (cf.
Chomsky (1989), Chomsky and Lasnik (1991) and Ouhalla (1991).
among others).

This is not the end of the story, though, for Spec-AGRP is now
a non-Case marking position because yabdu: in Arabic, like seem in
English, does not have the capacity to assign Case to a following

nominal expression. It is not a theta-position, either, given that yabdu

only has one propositional thematic role, which seems to have already
been assigned to the complement AGRP. If this is correct, then the

embedded small clause should to Spec-VP, where


subject of the move

it would be Case-marked by the matrix T. The verb yabdu: (seems)


this
must also move to T to check its tense feature therein. Obviously,
for passive structures in
analysis is pretty much like that proposed
associated with (22a)
Arabic in Section 6.1 above. The final structure
is therefore given in (24) below:
Hrabic
(24) TP

T VP

yabdu:; Spec V

1-Eumma:1-uk V AGRP

Spec AGR'

tk AGR AP

munhamiki:naj Spec

PP

fi:&amalihim

It has been argued above that in (19a-d), the verb yaD


abdu:

seems) patterns with its English a CP


counterpart in selecu
Complement. And as demonstrated in S
Chapter IV,
compleme
are Case assigners in Arabic. Obviously, then, the e
the embedded

Subject/topic is Case-governed by )in


the complementizer anu
19a-d). This being the case, we ement ofthe
would predict that mo
embedded This

subject/topic would vield mmatical structure.


an
ungrammatica
prediction is indeed born out, as the following grammaticality

judgements illustrate:

(25) a. zaynab-u; tabdu: ?anna t; waSal-at


Zaynab-Nom seem-3FS that arrived-3FS

Zaynab seems that has arrived."

b. -muEallima:t-u, tabdu:na ?anna t; waSal-na


the-school mistresses-Nom seem-3FP that arrived-3FP

The school mistresses seem that have arrived."

c. al-junu:d-ux yabdu:na ?anna t ?intaSar-u:

the-soldiers-Nom seem-3MP that triumphed-3MP

The soldiers seem that they have triumphed."

The ungrammaticality of the above structures may be accounted for in


various ways. For example, it may be advanced that the resulting

features, in contravention of the chain


chain has two different Case
must contain
Condition, which requires that every chain (a1,.., an)
B-marked position and exactly one Case-marked position.
exactly one
be argued that movement of the embedded
Alternatively, it may

principle, which states that


Subject/topic violates the Last Resort
derivation would otherwise
Syntactic operations may apply only if the
But here, movement of the
CSut in an ungrammatical representation.
subject/topic position is not
embedded subject to the inatrix

motivated; therefore, it cannot apply

149
following example
in ch is
English, which is similar
e.
Consider the

relevant aspects to (25a):

(26) John, seems that [Tp t; has left]

(26) is ruled out in English, being a Violation of the Last Reco


principle, too. Movement of John to the matrix Spec-TP position daas

not seem to be triggered by any grammatical principle. For examnle


at the time of raising, John has already been assigned Nominative
Case in its extraction site. It may be proposed that movement of John
in (26) is indeed motivated by EPP. But the
ungrammaticality of (26)
casts doubts on this proposal. In fact EPP may be
(must be) satisfied
here by inserting it in the matrix
Spec-TP, as the following example
illustrates:

(27) It seems that [Tp John has


left]
It seems,
then, that inserting it is more
economical in English than
moving John.
Consequently, the former
operation wins out.
regard to Arabic sentences w1u
like those in
embedded (25), movement
movement of of the
tne
subject/topic has no raison d'être,
device of either. The saivag
lexicalising the empty
Subject/topic position may not category in the the embedded
embeuu
This may be due apply here, as (28) below ates.
to the
fact that the illusua
from empty category inin (25a-c)
movement (25a-C) results
operation, and not from
Merger And in
operauo
general, traces of moved
elements may not be
cross-linguistically, 85
lexicalized, perhaps
(28) zaynab-u, tabdu: Panna-hu waSal-at
Zaynab-Nom seem-3FS that-3MS
arrived-3FS
Zaynab seems that has arrived."

6.5 Topic structures: A


cartographic approach
Another type of movement operation, called
Topicalisation,
has been discussed extensively in the
literature, (cf. (Andersson
(1975); Davison (1979); Emonds (1970,
2004); Green (1976):
Haegeman (1991, 2002a); Heycock (2002); Hooper &
Thompson
(1973); Maki etal (1999); and Rutherford (1970). This
operation
targets an element other than the subject and moves it to the left

peripheral position. This is illustrated by the following examples:


(29) a. This article, I will certainly read."

But
B6
see (30b).
I t has been argued in Arabic traditional grammar that in SV0 structures, the
Subject position is inherently topical. But we shall limit ourselves here to non
subject Topics.
87
sa) has the same argument structure as (i) below, in that both of them consist
E Event of reading, which involves two participants, viz. the article and .

() will certainly read this article.


word
semantically neutral
However, retlects the basic, i.e. pragmatically
and
marked; it is
order for English
"chglish sentences, whereas (29a) is more informationally
the m a n n e r
strings be linearized in
ed in a context requiring that the sentence

indicated in (29a).
151
knows who wrote it.
b. This manuscript, nobody

la-hu di:n-i
(30) a. (qul) ?alla:h-a ?a-¬budu (mukhliS-an
(say) Allah-Acc 1S-worship (taithful-Ac to-him religion.m
sincere devotion."
"Say: It is God I serve, with my
(The Quran, Zumar: 14)

b. (wa) 1-shuEara:?-u ya-tba&u-hum 1-Ra:wu:n


(and) the-poets-Nom 3MS-follow-them the-seducers
(And) the poets, it is those straying in evil who follow them."

(The Quran, Shuear?: 224)

c. (wa) 1-Dha:lim-i:na ?akadd-a la-hum &adha:b-an ali:m-an


(and) the-wrongdoers-Acc prepared-3MS to-them torment-Acc
extreme-Acc
"But the wrongdoers, for them has He prepared a dire torment."

(The Quran, Insän: 31)

In (29), the DP before the comma functions as the topic of the


sentence, i.e. the thing about which the sentence is predicated; the part
of the sentence following the comnma is the comment, i.e. what tne

Sentence says about the topic. In much the same way,


h)
Palla:h-a (Alauy
in (30a) and -shu&ara:?-u (poets) in (30b) also function as topics.
and

the sentences after the commas


are predicated of these topics.
There is, however, a slight (but very significant) differeince

between the (a) and (b) exanmples in (29) and (30), as we can scc from

the
fact that in the former examples, the DP
before the comma
ms

to Occur in two ment


different positions: in
once a post-veroa i

position, where it in åa
picks up the thematic role Theme,
e, and
and
once

Scope-discourse" position, where it serves as the topic


o f the
clause.

tOp
B y contrast
ot in the latter examples the DP set off by the
comma at the
beginning of the clause serves
only as the clause
which "stands 'proxy' for it in the topic and the
pronoun,
relevant position in the
functions as the direct
sentence"

object, (ef. Quirk et al. 1985).


(30b) have been referred to in the (29b)
literature as
left-dislocation
constructions. This squares with Lambrecht's (2001:1050) definition
af left-dislocation as "a sentence structure in which
referential
a

eonstituent which could function as an


argument or adjunct within a
Dredicate-argument structure occurs instead outside the boundaries of
the clause containing the predicate", Thus, the structure of left-
dislocation is schematically represented as (31):

(31) DP, ... (resumptive) pronoun, ..

It remains to be seen of course, how this differenceis

accounted for in terms of the Principles and Parameters model adopted


here. In Chomsky (1977) topies were introduced by the phrase
structure rule in (32), which yields the structure in (33), where S is the

88 89
Comment:""/

See also Bresnan (1978). the wh-phrase in Comp gets


39 wh-clause, where
nomsky (ibid) takes S' to be a also Rivero (1980)
for the
below for (29a), (see
as shown
Sequently deleted,
same analysis in Spanish):
[sI willcertainly
read twhichl
() ls lTop this article]ls [whieh]

153
(32) a. S " > T o p i c S

90
b. S Comp S

S
(33)

Topic S

Comp S

In the standard Minimalist Programme, however, topicalisation


involves adjunction to the category TP under CP. This yields the

following structure for topicalisation structures:


(34) CP

Spec

C TP

Topic TP
In
this subsection, topicalisation terms
of
tne will be nalysed in
cartographic approach to clause zi
structure, as
ou
90
Rule (32b) allows
recursion, as in the
s
(i) following exammple:
exampie
As for John, the
prizel think they will
give it to him.
(1997, 2001), Cinque (1999 and references
cited therein."
(1997) proposes an analysis to the structure of CP Rizzi
the formerly assumed a single
main- form
CP domain- according to which
discourse related functional
projection CP- is split into
seve

projections. This analysis- referred


i n the literature as tne
cartOgraphic approach- is motivated by the
at that several constituents may
appear in C'. Crucial to this
onroach isis the assumption that at least some
approach
functional notions, e.g.
topic, are now considered
"quasi primitives of language." Equally
important is the
working hypothesis that the hierarchical order of
fronted constituents 1s assumed to be fixed
crosslinguistically.
Accordingly, the split C displays the following hierarchy: 94
(35) Force> (Topic)> (Focus)>Fin9

This accounts for the appearance of elements such as

complementizers, topics, wh-phrases, focalised elements, etc.


Accordingly, topicalised DPs, such as those in (29a) and (30a) have
been assumed to be the result of a movement transformation which

91See, for example, Rizzi (1997) for a discussion of why an impoverished structure
cannot be the right structure for topicalisation constructions.
Ke(34)
92
& Parameters model
artography is a research programme within thea Principles
series of colloquia held in Italy
daopted here. It emerged and gained its name in the publication of Belletti
late 90's and became widely known through
ne
2004), Cinque (2002), and Rizzi (2004). notion of
structuralism, which takes the
93
DViOusly, this goes against American notions are
Therefore, functional
ory and constituency as fundamental.
and constituency.
in terms of relations of category
%ned
94 are activated only when
and Focus projections
9 7 ) posits that Topic
needed.
95. wh-elements cannot precede
This hierarchy will be adopted here for Arabic,
as
this approach to
Bakir (2011) for a critique of
this language.But see
Irani In
Iraqi Arabic.
155
Arabic
constituent to the leftmost position in the
relocates a

leaving a trace in the take-off site, (cf. Rivero (1980). Raa


phrase marker
and Haegemen (1991), among others). Within this cono
adford (1988)
structure associated with (29a) and (30a) is given in (36) heln
onception, the S.
(36) below
(36) TopP

Spec Top

This article; Top FocP


Pallah-a;
s Foc
aneensee..e..............as.......................... ...enanamuna****** TP

I will certainly read t


Pa-&budu t;

Support for the above structure for


Arabic comes from the foliowi
example, where [Spec,
FocPl also contains a
(37)
wh-phrase:
al-wajbat-u kayfa Padadt-a-ha:
The-meal-Nom how prepared-2MS-it
"The meal,
how did
you prepare it?
.We assume, for now, that
(37) has the
structure outlined in
below: (38)

(38) TopP

Spec Top

al-wajbat-uj Top FocP

Spec Foc

kayfa Foc TP

Pa&dadt-a-ha,:

to reconcile with (34), where the locus of wh-


OOViOusly, (38) is hard
structure than that of
Pirases S assumed to occur higher in clause
for the time
picalised DPs. We shall return to (30c) below, but

being, let us address the case of (29a&b) and (30a&b).


assumes that the Topicalised
ne derivation sketched in (36)
take-off
its
connected to
Din assumed to be
(29a) and (30a) is
movement,
"which sis
which

position by the
the algorithm of syntactic
the two
kinds
assigning
languages for
systematic USea
by natural
und Arabic

ofinterpretive properties to an element. The element is meroed :.

position in which it receives its argumental status, a


thematic rol.

then it is moved to another position dedicated to a


particular scon
cope-
discourse property." (cf. Rizzi
(2014:517-533). The trigger for thi
movement operation is the same as that
proposed for wh-movement
(see Chapter VIl). To anticipate somewhat, just as it was
assumed that
a
wh-phrase is attracted to [Spec, CP] by the [+wh] feature of C. so.
too, it will be assumed here that the
Topicalised DP is attracted to
[Spec, TopP] by the Top head. Evidence that (29a) and (30a) involve
movement comes from the following examples, which exhibit strong
island effects, perhaps reducible in the final analysis to an ECP
violation:

(39) a. This booki, I accept the argument that John should


read t,.
b. This book;, I wonder who read
t. (Chomsky (1977: 91))
(40) a. al-Su:rat-a; la: ?u-Saddiqu 1-?isha:Eat-a ?anna
zayd-an rasam-a t
the-painting-Acc not IMS-believe the-rumour-Acc that zayd-Ace
painted-3MS
"The painting, I don't believe the that
rumour Zayd painted."
*

b. al-ba:b-a; tasa:?al-tu man kassar-a ti


the-door-Acc wonered-1S who broke-3MS
The door, I wondered who
broke."
nd (40a)
(39a) and violate the
Complex NP
Condition, and (39b) and
siolate the
(40b) violate Wh-Island Condition." The
and
respective derivations of
(39a), (39b), (40a) (40b) illustrated below:
are

(41) a. TopP [DPi this book;] Irpi I accept [DPa the argument [cP that
TP2 I should read t;]]]

b. [TopP [DPi al-Su:rat-a;] [TPi la: ?u-Saddiqu [dPz 1-?isha:Eat-a


CP ?anna [TP2 zayd-an rasam-a t]]]]

(42) a. TTopP [DP1 this booki] [TPi I wonder [cp who, [T2 l, read t]]]
b. TopP [DP1 al-ba:b-a;] [TPi tasa:?al-tu [cP man; [P2 4 kassar-a

The structural representations schematised above all involve

displacement of a DP from its canonical (argumental) position to a

sentence-initial position. In (41a&b), DP1 crosses DP2 and TP1, both

being bounding nodes in English and Arabic. And in (42&b), DP1

moves across a wh-island.

have been argued to be


With regard to (29b) and (30b), they
constructions, where the DP in
instances of (Clitic) Left Dislocation

followed by an embedded
96 with a head noun
A COmplex noun phrase is one
two types of
adnominal
There are
head noun.
unominal clause that modifies the such as (i). Both
(i) and NP complements,
relative clauses, such
as
clau
are English:
assumed to be islands in the sense of Ross (196/

bought ]]]
(which/that/Ø) [Tp the actor
() lDe the car [ce
he bought a
Lamborghini
(i) lop the rumour [ce that [Tp
this position, and not preposed
ed fron
from within
ISpec. TopP] is merged in
the conmment" clause. A picce of evidence in favour of th..
the base-
generation approach comes from the fact that the relation betu. ation between this
DP and the argumental pronominal copy/clitic iS not subject to icl.
island
constraints. This is illustrated by the following examples:

(43) a. This book, I accept the argument that John should readit
b. This book, I wonder who read it.

(44) a. al-Su:rat-u, la: ?u-Saddiqu 1-?isha:Eat-a ?anna


rasam-a-ha:; zayd-an
the-painting-Nom not 1MS-believe the-rumour-Acc that
zavd.
Acc painted-3MS-it

The painting, I don't believe the rumour that Zayd painted it."
b. al-ba:b-u; tasa: '?al-tu man kassar-a-hu
the-door-Acc wonered-1MS who broke-3MS
"The door, I wondered who broke it."

The reason why the italicised DPs in


(43a&b) and (44a&b) seem to
cross over islands is that they do not cross any islands in the first place
as they are
base-generated, i.e.merged, in the left periphery of their
respective sentences. The left-dislocated DP and the
are
pronominal cop
coindexed via a rule of semantic
interpretation that establishes an

(obligatory) anaphoric relation between them akin to that discussed m

Chapter VII, (cf. Rivero


(1980). Put differently, the left-dislocateu
DP and the
pronominal copy refer to the same
individual/entity. Unuder
this
view, the pronominal copy is treated as a
bound variable
[Spec, TopP].
by the
DP in

et 1uS return to (30C), repeated for


convenience as (45) below:

(45)
(wa) 1-Dha:lim-i:na Paadd-a la-hum Eadha:b-an ali:m-an
(and) the-wrongdoers-Acc
prepared-3MS to-them torment-Acc
extreme-Acc

But the wrongdoers, for them has He


prepared a dire
torment."

example is hard to explain in terms of the


This
base-generation
approach to Left-dislocation structures adopted here, for the DP I
Dha:limi:na (wrongdoers) carries Acc Case, not the "default" Nom
Case normally assigned to topics in Arabic."7°
However, some Quran
interpreters, e.g. al-Qurtubi:, take the left-dislocated DP in (45) to be
Case-marked by a deleted verb akin to Eadhdhaba (torture) or the
preposition li (to). This would then strengthen the case for taking left-
dislocated DP to be base generated at the left periphery, as argued
above.
a0ove. We have seen above that topicalisation may 0Ccur in

Combination with wh-questions, (cf. (37). This is so because the


opicalised DP and the wh-phrase target different landing sites, viz.

97
This Case- Arab grammarians to be
Signed by exemplified in (44a&b)- is assumed by
nSsigned
In German,
an abstract notion of "topichood" (al-?ibtida:?).
the folowing
overt Accusative Case,
as
piece ofEdata left-dislocated DPs also carry

illustrates:
(i) Den Mann,
ich habe ihn gesehen.
Ihe man-Acc I have him-Acc seen
FTlcP
respectively. ly. However, structures
However

FocP],
and [Spec,
[Spec,
TopP]
bears a
Case
feature
other than
Iominative may
Nominati..

DP
99 This is illustr
illustrated by the following
fronted
where the This is
wh-questions.""

c o - o c c u r
with
not

example:
100
?a&dadt-a-ha:"
al-wajbat-a kayfa
(46)
The-meal-Acc h o w p r e p a r e d - 2 M S - i t

Here, we shall
for an explanation.
observation calls
This empirical
that the dislocated DP in (46)- and
to mean
take this impossibility
matter- and the wh-phrase compete for the
similar structures, for that
This is not a wild stipulation, given that
same Spec, FocPl
position."
functions as a contrastive
the DP normally
in this type of structure,
in (29a) and (30b)) which usually
focus, unlike topicalised DPs (like
to the parties involved ina
denote old information (known
is predicated. (30a). for
conversation) about which something
new

something like "It's Allah I


example, repeated as (47) below, means

worship, and not any of your deities", where ?alla:h-a (Allah) is used
contrastively.
(47) (qul) ?alla:h-a ?a-¬budu (mukhliS-an la-hu di:n-i)
"
Prof. Boudrae (pc).
(46) is ill-formed, irrespective of whether the DP al-wajbata (the meal) isfronte
to
or whether it is base-generated in its S-structure position and linked
coreferential pronominal clitic.
101
Rizzi (1997) for examples nd
See
Foci
bearing on the hypothesis that Wh-elements
compete for
the same structural
position.
This teams with Rizzi's sed
(1997:285) definition that "The topic is a prepo
element characteristically
and normally
set off from the rest
of the clause by "comma
ation"

in previous
ls
expressing old information, somehow available
discourse; the comment is a kind and salient sen sentence
predicated of the topic and of complex predicate, an op
introducing
new
information.
(eav Allah-Acc 1S-worship (faithful-Acc to-him
uSay: It is God I serve, with
my sincere religion-mv
m
devotion."
(The Quran, Zumar: 14)
e.her. Newson et al. (2006) argue that in structures such as (44a&b)
she topic forms an intonational unit by itself, with its own stress, and
the following sentence also has its own stress." However,
eanstructions such as (45) have "the fronted element within the same
izntonation unit as the rest of the clause and this element carries the

major stress of the sentence."

A nice consequence of the proposed analysis- if correct- is that


it provides a reasonable explanation of why Focus structures are

incompatible with wh-question operations. Thus, (47) should have the


alternative structure shown in (48), and not the one given in (36)

above:

(48) TopP

FocP
Top

Spec Foc

TP
Pallah-aj Foc

Pa-Ebudut;

163
Exercise 6.1: Discuss the movement operation(s) involved in each of

the following examples:

(a) John was believed to have been insulted.

(b) The senator wants to be re-elected.

(c) In the box, she put her late husband's belongings.

(d) Sushi, John said Mary ate for lunch.

(e) I read an article yesterday which contained examples from Turkish.

(f)A theory is developed which distinguishes lexical categories from


functional categories.

(g) Piyya:ka na-ebudu

to-you we-worship

You alone, we worship."


(h) Zayd-un man ta-Dhunnu ?anna-hu
sa-ya-ltaqi:
Zayd-Nom who 2MS-think that-it will-3MS-meet
Zayd, who do you think he will meet?"

Exercise 6.2: Draw the structures


associated with (a), (C) anu e).
VII. WH-MOVEMENT

TLie chapter will discuss the


following topics:
Root questions: non subject extraction
in English and
Arabic
> Root questions: subject extraction in
English and Arabic
Embedded questions: non subject extraction out of
complement clauses in English and Arabic
Embedded questions: subject extraction out of complement
clauses in English and Arabic

>That-trace Effect
Parametrising the differences between English and Arabic

So far, we have seen two types of movement transformations.


viz. Head movement and NP- movement. The former movement

moves a head to another head position, and the latter movement

moves an NP into a subject position. We can thus define NP-

movement as follows:

(1) NP-Movement3

Move NP to [Spec, TP]


NP-movement
that
We have indicated in the previous chapter
nominative Case
a
hcad occupies
Ouuces an A-chain, where the Therelore,
caseless posilion.
trace(s) occupy a
Ouonand the other
we shall discuss another
is Case-driven. In this chapter,
OVement in
constructions

103 and raising


well for passive
derinition works equally
l
English.
165
Arabic
moves a wh-expression- i.e. an
transformation that

containing an interrogative word beginning


expression
with wh-, e.g.
wh-, e.g
what/which/who(m)/ when/whV,
etc.- outside TP, This transfr
will be shown to exhibit properties which distinguish it from
NP
movement.

There two types of wh-questions: (1) direct


are
wh-questions.
ns,
where the entire sentence is interrogative, and (2) indirect wh
questions, where the interrogative clause is embedded in a declarative
e
sentence.

7.1 Root questions: Non-subject extraction in English and Arabic


Consider the following examples:
(2) a. LAP How expensive] will it be?
b. [ADVP How quickly] can he run?
c. [Pp To whom] are you talking?
d. [DP Which article] will you read?
It is
easy to from (2a-d) that different
see
types of constituents may o
targeted by
wh-movement, namely AP. ADVP, PP, and DP. NOUCE
also that in
each of the above
examples, is also
wh-movement
accompanied by another movement, viz.
head movement. As have
DOVe, the latter movement v
104
moves elements in T to The
Whether [+Q]
different triggers T-to-C
languages may employ movement is a parameter
in pai that reveals the options
question formation.
landing site of the wh-phrase tself is the specifier of CP. Thus, the S-
r e aSsociated with, say (2d) is given as (3) below:

(3) ce Ipe which article ji lc will, [rP you t read t:]11

Unlike French, English does not use the in-situ strategy in wh-
questions, as the following data from both languages show:

(4) a. It will be how expensive?

b. He can run how quickly?

c. You are talking to whom?

d. You will read which article?

(5) a. Tu as fais quoi?


you have done what

What have you done?"

b. Tu es arrivé ici comment?

you have arrived here how

"How did you arrive here?"

verb has to move to C in


ne data in (4a-d) begs the question why the
enterprise
sh? We will follow the current trend in the generative
abstract
Sme that direct questions are projections of an
which
the head C and
Q(uestion)-feature
uestion)-feature which is hooked on

contains a wil
exp
CSSes the interrogative force of the sentence. C also

167
Towards a
Tow
Contrastiv
SyntaxofoJ English
English and
P a r a m e t e r s
Theory:
Arab
Prnciples
and
in
verb raising in wh-questions we
overt vert
English
has
overt
will
feats
hat both feature are strong. A
feature.
Since
assuming
that
(1995)
in
Chomsky
follow
therefore
move in the o v e r t syntax to
check the wh-
must
wh-phrase the interrogative cla use
head C of under
with the
associated
foature

short, and a verbal al


Spec-Head
configuration,
SHAGR TOr
element
has to move
dummy do) un t o

(cither an auxiliary or a form o to

anchored therein. 1his explains the additionalIT.


check the Q-feature
movement in root questions in English.
to-C

Let us now address the following data from Arabic:

(6)
(6) a. [P li-man] ?a&ETayt-a 1-kita:b-a?
to-whom gave-2MS the-book-Acc

To whom have you given the book?*"

b.[DP ?ayy-a kita:b-in] nashart-a?


which-Acc book-Gen published-2MS
"Which book have you
published?"
c.
ADVP kayfa] faalt-a ha:dha:?
how did-2MS this

"How have
you done this?"
(6a-c) involve overt
Like movement of
of PP, DP and AdvP,
English, then, the respectivel

associated with an wh-phrase


empty category
app p p e a r s clause-initially
and is

variable
ry
marking thepo
the position of the
bound by this wh-phrase. 105 We can thus
Ise wh--in-situ
predict that Arabic does
strategy in the formation of not
regular questions 106
This
diction is indeed born out
given the wh-questions16
107
judgements: l07
following grammaticality

(7) a. a&Tayt-a l-kita:b-a li-man?


gave-2MS the-book-Acc to-whom
You have given the book to whom?"

b. nashart-a ?ayy-a kita:b-in?


published-2MS which-Acc book-Gen
You have published which book?"

c. faEalt-a ha:dha: kayfa?


did-2MS this how
You have done this how?"

105
Arabic has the following wh-words: man (who/whom), ma:dha: (what), ma:
(what), Payy (which), lima:dha: (why), Payna (where), mata: (when), kam (how
uch/ how many) and kayfa (how). These words may function as heads of phrases
or as specifiers:

Phrase head Phrase specifier


man (who) ?ayy (which)
ma:dha: (what) kam (how much/how many)
ma: (what)
lima:dha:(why)
?ayna (where)
mata: (when)
106 The
(7a-cl kayfa (how
in-situ strategy meanstthat a wh-phrase has not visibly moved to Spec-CP.
issue
107(7a-c) irrelevant to the
But this is
correct as echo-questions, of course.
under analysis.
69
derived in
n (8) beliOw,
for
instance,
ld be
would be as
d e r i v e d as
low, where
(6b), ich book)
book) to
Example
Pavy-a
kita:b-in
(which
mov
moves
Spec-CP, thus
wh-phrase
the counterpart
in (2d):
its English
mimicking

le- [Te nashart-a ]l1


Icr lor2ay-akita:b-in:};
(8)

intriguing phenomenon
to be observed in (6a-c). and
all
at
An
for that matter, is that whener
instances of wh-questions in Arabic, a

to its scope position in the overt syntax, the srfans


wh-element moves

word order is always Operator-finite verb. We shall see below that the

adjacency relationship between the operator and the inflected verbal

element is also required in embedded questions. Apparently, then,

Arabie resembles residual verb second languages, like English, for


instance, which restricts the occurrence of V2 (read verb second)
phenomenon to interrogative sentences such as (2a-d), where the
auxiliary verb is the second
constituent, and constructions where an
adverbial has been extracted to the
sentence-initial position such as (9)
below. 108

(9) Never again will I raise the issue with him. (cf. Ennassir
(1998:51)
m
English, the adjacency of
results from wh-operators and verbal elen
T-to-C movement.
element lands in T In Arabic, however, the verbal
and does not however,
not move
S
to by
C. This is
uf
up orroborated

For more
on this, see
Ennassiri (2014b).
following xample, where the
the
the complementizer ?in (if) and the
auxiliary ka:na (was)
between
occurs
subject.
(10) caPal-a-ni: 1-mudi:r. ?in ka:na
ttala:mi:d-u yafham-u.na
asked-3MS-me he-principal-Nom if was
asked-3MMS
understand- 3MP the-students-Nom
The principal asked me if the
students understood."
10) patterns with embedded
wh-questions in English, where the wh-
nhrase moves overtly to Spec-CP, but movement into
C is
disallowed,
(cf, gloss in(10). The fact that Arabic
allows the
G to be directly followed by the verb, as in
complementizer ?in
(10), also argues against
the possibility that T has moved to C. But
this leaves open the
question of how the Q-feature of C is checked in Arabic
wh-questions
Such as (6a-c).109

It has been argued in the literature that all syntactic variation


across languages is restricted to non-substantive parts of the lexicon.
(ef. Ouhalla (1990), Chomsky (1995), among others). It is therefore
plausible to assume that cross-linguistic variation like that observed
between English and Arabic with regard verb movement to C may be

handled exclusively in terms of feature strength, and that there is no

Recall that we have


Padneterization of the computational system.
and
above that the parametric differences between English
ascribed to the
cn with regard to V-movement have been
French
More specifically, in
differences
C s In the feature strength of T.

109
erhaps this is done via T-to-C movement at LF.
and
V-feature associated with
with T are
hoth the N-feature
DP m o v e m e n t to Spec.T strong.
French exhibits overt
Therefore,
T is
movement to T. In English,
the N-feature on
equally strong and
eau.

in Spec-VP) has to mo
so the thematic subject (base-generated
Spec-TP in the overt syntax
as well. However, the V-feature o4
of T has -feature
been assumed to be weak, thus delaying V-to-T movement as ate

LF.

Now, following universal requirement on wh-questions, let uc

assume that the head C of a wh-clause in Arabic is also endowed with

a wh-Operator feature and a Q-head feature. The wh-Operator feature

is equally strong, thus requiring a wh-phrase such as ?ayy-a kita:b-in


(which book) to move to Spec-CP. But the Q-feature is weak;

therefore, relegating movement of T to C to LF, in accordance with


Procrastinate, which minimizes the number of overt operations

necessary for convergence. The bottom line is that Cis not an attractor
for the verbal element in Arabic. The empirical intent of the

strong/weak distinction, then, is to distinguish overt from covern


movement. If this analysis is correct, then there is a sense in which we

can say that languages not only differ with regard to the richness of
their morphology, but also- and most importantly, perhaps- in ternissof

the strength of grammatical features, (cf. Chomsky (1995).

1aving said that, though, there is indication that there 1s ert

Ovement into C in Arabic yes-no qucstions such as ow:


(11a-c) De
(11) a.
?a-tajru?u Eala: mugqa:TaEat-i:?
O-particle-dare-2MS on interruption-me
Do you dare interrupt me?"

b.Pa-Panta faalt-a ha:dha: bi-?a:lihat-i-na:?


-part-you did-2MS this with-gods-Gen-our
"Are you the one who did this to our gods?"

c. Pa-turi:du ?an taku:n-a saEi:d-an?

Q-part-want-2MS to-be happy-Acc


"Do you want to be happy?"

Since the interrogative particle ?a is not a separate morpheme in


Arabic, it needs to attract a bearer to the C position. So, the questioned
word has to move overtly to support this affix; otherwise, it would
remain stranded, and the structure would eventually be ruled out. If

this analysis is correct, the sentence in (11a), for example, will be


110
derived in the manner outlined in (12) below:"

where the main verb


10 such as (i) below,
This is reminise
IScent of French examples
has moved to C via T:
(i) Parles-tu Français?
CP
(12)

Spec

TP
op Co

T VP
Pa-tajru?-u

V
Spec

pro V PP

Eala: muqa:TaEat-i:

one found
In (12), Spec-CP is occupied by a null Operator akin to the
involves
in yes-no questions in English. The derivation depicted above
V-to-T-to-C movement, thus accounting for the linearization exhibitea
in (1la), i.e. the fact that the verb is incorporated to the right or n

question particle.

But the examples given (1 la-c) have other variants where the

question word is hal (whether):


(13) a. hal tajru?u Eala: muqa:TaEat-i:?
whether darc-2MS on
interruption-me
Do you dare interrupt me?"

b. hal Panta faEalt-a ha:dha:?

whether you did-2MS this

Are you the one who has done this?

c. hal turi:du ?an taku:n-a saEi:d-an?

-part-want-2MS to-be happy-Acc


"Do you want to be happy?"

The structure associated with, say, (13a) is given as (14) below, where
hal (whether) is assumed to occupy the Spec-CP position:
(14) CP

Spec C'

hal
Cp-o TP

T VP

tajru-u Spèc

pro
PP

Eala: muqa:Ta&at-

175
Principles anai u
and A
e x t r a c t i o n
in English Arabic
Subject
questions:
7.2. Root
in (2a-d), it is fairl casy to
examples given
From the
English
indeed
taken place as elements c
wh-movement
has
see that positio
in their expected
do not appear
complements and adjuncts
In the case of subie
the left periphery/edge
of the clause.
ject
but at
not so clear cut. For the sake of
are
questions, however, things
the following examples:
exposition, consider

(15) a. Who will paint the wal1?

b. Who painted the wall?

In both examples, the subject does not seem to have moved anywhere.

But for the sake of generality, we will assume that who has indeed

moved to Spec-CP in a manner pretty much like that sketched in


(3) The derivation of (15a) and (15b) are thus given as (16) and
(17), respectively:

(16) cP [DP who, [c will, ITe4 paint the wall?]]]

(17) lcP [DP who [c- [r painted the


t4 wall?]]]
111
Butthere is independent evidence that
nis is
ilustrated by the subject wh-phrases do indeea
visibly vacated the following example, where the embedded subject has
embedded/complement clause:
i)
Who, do you think
[ cP2 t, will
repair the car?
In both
structures, the
wh-phrase has moved to
left a
trace in Spec-TP, even
though Spec-CP
the actual position and
a

word order
changed.
has not

Consider now the


following example of subject
movement in
Arabic:
(18) man katab-a Eala: 1-jida:r-i?

who wrote-3MS on the-wall-Gen

Who has written on the wall?"

It has been argued above that the underlying word order of Arabic is a

sVo. and that the superficial Vs0 order results from V-to-T

movement, (cf. 5.2). Accordingly, (18) has the derivation sketched in


(19) below, where the subject man (who) moves from Spec-VP to

Spec-CP:

177
(19) CP

Spec

1nank C-O TP

VP

katab-a DP

tEala:1-jida:r-i

In this view, subject extraction closely mimics object extraction, (cf.


(8)).

7.3. Embedded questions

In this subsection, we will be looking at the syntax of

embedded/indirect question CPs, i.e. questions that function as clausal


complements to verbs such as wonder, ask, doubt, etc. To begin With,
consider the following examples:
(20) a. I wonder [cp whether he will attend the meeting]**
b. He doesn't know [cP which article he
should read]

(20a) is an embedded
yes-no question.
They asked me lcP Who won the prizel

A Do you know lcP WHen tioward will


give his next lecture?

In
ad the indirect questions are
complements of the the

vOnder, remember, know, and asked,


verbs w o n d e r
respectively. The matrix
that embeds an indirect question may be
sentence

declarative, as in
203 -c), or interrogative, as in (20d). Note that, like direct questions.

indirect questions also require that the wh-phrase be moved to the

Specifier position of the embedded clause. But unlike in direct

questions, where an auxiliary or an agreeing form of "dummy do" is

hypothesized to occupy the Comp position, there is no T-to-C


movement in indirect questions. Evidence for this comes from the

following grammaticality judgements:'

(21) a. I wonder whether will he attendthe meeting.


read.
b. He doesn't know which article should he
his next lecture?]
d. Do you know [cp when will Howard give

seen
it will be
113 Filter. However,
Titer is dubbed the Doubly-filled
Comp contain overt
s
CP] and C
where both [Spec, and a
Arabic allows constructions wh-phrase
nat also allows
both a
(cf. Bavarian German
(28). Bayer (1984):
C S example from
illustrated by the following
entizer. This is

Woass ned wann dass da Xavea kummt.


comes
I know not when that the Xavea
Tdon't know when Xavea is coming wh-
with
associated

parameter
therefore a
Filter is
yfilled Comp
movement.
179
ofan
an indirect
indirect
Therefore, we shall assume that the structure question is
of course, for
direct question, except,
Cl.

the same as that of a ject


The respective
structures assOciated with
auxiliary inversion. say
thus given as (22) and (23), respectivelv:
(20a) and (20b) are

will attend the


(22) I wonder [cr whether [e- C [ip he meeting]|1

(23) He doesn't know [cp which article; le- C [Iphe should read t||

Consider now the following data from Arabic:

(24) a. 7a-tadhakkaru [cekayfa kunt-u ?a-sbaHu]

1S-remember how was-1S 1S-swim

I remember how I used to swim."

b.sa?al-a-ni:
[cpbi-man tazawwaj-at ibnat-u Eamm-i-hi]
asked-3MS-me with-whom married-3FS daughter-Nom uncle-
Gen-his
He asked me who his cousin had married."

c. la: ?aHada
ya-Erifu [cp lima: dha: nfaSal-a zzawja.n1
no one 3MS-know why separate-3MS the-couple
"No one knows why the couple
got separated.
Unlike in
English, the
adjacency relationship between the
se

and the verbal element is also


wn-p*
in
required in embedded quesuo
cf. Ennassiri (201 This is
Lrabic,

ammaticality judgements:
corroborated by the
following
25 a. a?al-a-ni: [cp bi-man ?ibnat-u
Camm-i-hi
asked-3MS-me with-whom daughter-Nomtazawwaj-at
married-3FS
uncle-Gen-his
He asked me who his cousin had married"

hla:?aHada ya-Erifu lcp lima:dha: zzawja:ni nfaSal-a:


no one 3MS-knoW why the-couple separate-3MDual

No one knows why the couple got separated."

It may be assumed that V-movement in such examples as (24a

c) above is driven by a principle on the scope of wh-operators, which

Rizzi (1991) calls the Wh-Criterion:

(26) The Wh-Criterion


with
A. A Wh-Operator must be in a spec-head configuration
an X+WH].

B. An Xw must be in spec-head contiguration witn a


Operator. head to
clausal CP
to the
According to this assumption, V must move

will have
(24b), for example,
bring the [+w feature to it.
Therefore.

The
partial derivation in (27):
(27)

V CP

sa?ala-n1 Spec C

bi-manji C TP

tazawwaj-at; Spec T

?ibnat-u &Eamm-i-hik T VP

....... ..
Spec

....
....
V

V PP

Underlying this assumption is the idea that the input structure ror
Move-a is one where the
subject occurs in the [Spec, TP] position.
Further investigation,
however, shows that this assumption docS
make the right
prediction, either conceptually or empirically. For one
thing,this analysis cannot account for sentences such as (20) ,
where a lexical
complementizer co-occurs with a verbal eleme
here is that there are three
roblem
pre-TP elements, but
namely the only two
structural positions, specifier of CP and its head:
(28)
eaala-ni: man lladhi: xalaq-a ssama:?-a
asked-3MS-me who that
created-3MS the-sky-Acc
He asked me who created the sky." (cf.
Ennassiri (2014b:38)
Eor another, it predicts that structures such as (29) are well-formed
contrary to fact:

(29) ma:dha: katab-u: 1-?awla:d-u?

what wrote-3mp the-children-nom

Now, if extraction is indeed from an underlying structure where the

subject is under [Spec, TP], there simply is no plausible way to

account for (29). Pre-theoretically, one might venture an analysis in


clausal CP head.
terms of which (29) would be ruled out because the
feature [-plural]. This
C, can only host a verbal element having the
must undergo a process of
Deing the case, then, the verbal element
to S-structure.
D-structure
Sngularization in the derivation from
it
wild nature of this analysis,
Iues the ad hoc, and obviously
mainstream
within
widespread assumption
wOuld run contrary to a features, 1or
other
phi-features, or any
Crative linguistics that no 114 This ban
derivation."

altered in the course of the


d t e r , can be Principle. Itl 1s
version of the Projection asS
C l Tollow from some
questions
such
embedded
that in
nerefore plausible to assume
But they
may
deleted.
hence
and
114 movement,

off by
Features may e
checked
derivation.

not be n of the
changed in the ourse

183
Principles a n d Pu

further
further
doesn't limb
climb than T, for
the
verbal
element
this
(24b). supertluous
at best.
would be
operation

extraction oul of complement ol.


s in
7.3.1 Non-subject
English and Arabic

7.1 and 7.2 above,


we dealt with the syntayX
of
In subsections
this subsection, we discuss wh-extraas
matrix/root wh-questions. In on

To begin with, consider the followit


out of complement clauses.

examples:

a. DP How many countries]i do you think (that) Phileas Fogg can


(30)
visit t, in 80 days?

b. [DP Which girl] did you say (that) Passepartout is likely to mary
t?

(31) a. DP Payya fusta:n-in]; ta-Dhunn-u ?anna zaynab-a sa-ta-shtar: t

which-Acc dress-Gen 2MS-think that Zaynab-Acc will-3FS-Duy

"Which dress do you think that


Zaynab will buy?
b.PP liman]; ta-Etaqidu ?anna-hum sa-vuET-u:na l-ja:?1zal-* "
tO-whom 2MS-believe that-they will-give-3MP the-pre Acc
"To whom do
you think they will give the prize?"
We have
alluded above to ne
the fact that both and NP
movement are local head-movement
head-mo
operations
operations in
in the
tne sense that they
operate
clically.Similarly, wh-movement also
despite surface structure operates SuccessIve-
cyclically

appearances. The
derivation of
(30a)
and a)proceeds in the standard fashion
shown in
b)below, with the moved
3
(32ad
wh-phrases targeting first [Spec,
ih
CP21- which functions as an
escape hatch for
wh-movement- before
eventually landing in [Spec, CP11.l5

(32) a. IcPI how many countries; lc do lTPI you think [cPz t' [c that
[TP2 Phileas Fogg can visit ti]]]

b. [CP1 ?ayya fusta:n-in; [TPi ta-Dhunn-u [cP2 tf [c ?anna [rTP2

zaynab-a sa-ta-shtari: t]]]]

thus legitimate since they pass the


The traces of the wh-phrases are

in (33) below, (cf. Rizzi


Empty Category Principle (ECP) defined
(1990)).116

must be:
ECP: A nonpronominal empty category
and
1) Properly head-governed
(Formal Licensing)

an escape
hatch is not availaole, W
when such
Seen in 7.4 below that
DE ungrammatical sentences.

movement cross a CP results in formulation


of the ECP
disjunctive

0)
Chomsky
A
7 6 b a ) proposes
the following
muSt
De
honpronominal empty category (emphasis addea
A antecedent-governed

eta-governed, or (ii)
as follows:
0:7) defines government
that:
X a-go Z such
Xa-governs Y if there is no

a-governor for Yand


potential
9is a typical

185
(ii) Antecedent-governed or
theta governed (Identification)ll8

7.3.2 Subject extraction out of embedded clauses in English

and the That-trace effect

It has been argued in the literature that constructions that

exhibit the That-trace Effect are those where the subject has been
extracted, (cf. Chomsky (1981), among others). So, the relevant sub-

structure of (34) is given as (35):

(34) Who do you think that loves Mary?

ii) Z c-commands Y and does not c-command X.

The following definition of antecedent-government should suffice here:

(in) X
antecedent-governs Y iff
(a) X and Y are coindexed.
(b) Xc-commands Y.
(c) There is no barrier between X and Y.
(35) CP

Spec C'

t C TP

that Spec T'

T
YP
-
loves Mary

Notice that no
comparable violation arises when objects or
adjuncts
are extracted, which fact may be taken to mean that the That-trace
Effect is uniquely related to antecedent-government of the subject
I19
trace, (cf. Chomsky (1981), (1986), Manzini (1992), among others)

(36) a. What do you think that I gave him for his birthday?

answered the
b. How does your teacher believe that you have
question?
only used as a
literature, that-trace effect is
in more recent
of certain
onic term to account for the ungrammaticality
some more generat
reduced to
Ostructions; it is very often than not
accounts for the
for example,
notions. Chomsky (1986b),

(1979).
ECP, (Chomsky
119Th dCe Effect was at the origin of
the
Pilp

as (34) above bv
(34) above by assuming
g that
that aa
examples
of such
of the comnl.
ungrammaticality

effect is induced ed by the presence


presence
complementizer
traca .
Minimality
a n t e c e d e n t - g o v e r n m e n t
of the root in [Spec,
blocks
that, which He (ibid: 42) dee.
intermediate
trace in Spec, CPJ. fines
IPI by the
Minimality as follows:

d ...y..0... B...
(37) In thestructure:
...

barrier for ß ify is a projection, the immediate projection


a is a

distinet from B.
of o, a zero-level category

(38) below are ruled in in English because the


Examples such as

does not constitute a Minimality


complementizer is abstract; hence, it
barrier:

(38) Who do you think is the best candidate?

Rizzi (1990a) feels skeptical about Chomsky's analysis


sketched above. In particular, he is not convinced that an overt

Complementizer blocks antecedent-government whereas a null one

does not. So, for reasons to do mainly with the inadequacy of the

disjunctive formulation of the ECP à la Chomsky (1981), (1986a),


(1986b) and Stowell (1981), Rizzi (ibid:29/71) proposes
proposes the
conjunctive definition of the ECP given in (33) above. This alternauive
definition accounts of
quite straightforwardly for the ill-formeanc
54). The contention here would be that the this
subject trace
example fails to fulfill the of
the
principle of formal licensing by ** e

fact that is not canonically


it , T.
governed by the functioa
The
principle
of identification is not
fulfilled either, as T does
qualify fo theta-government, ana the
potential
not

who is
too far removed from the antecedent-governor
subject trace under
Relativized
Minimality.

Erther, Rizzi (ibid) proposes to account for the ban on


subject
fiotn in English in terms of the
agreement-in-Comp strategy.
More specifically, he argues that in an example like (39) below, where
the head of the lower CP 1s null, there is agreement between the wh-
phrase, or its trace, in [Spec, CP] and the null head:

(39)
(39) Who do you think loves Mary?

This process has the effect of turning an inadequate head, i.e. one

which is inert for proper head-government, into an element capable of


the 120
fulfilling the requirementof head-government on traces. Only in
as it would be
latter case is a subject trace licensed in [Spec, IP],
that
the head C. The gist of this analysis is
properly head-governed by
be
the fact that it can
much the fact that C is null
as
t 1s not so
an ECP
saves structures
like (39) from
written as an AGR that
allowed by
AGR is one of the options
CCt. S0, transforming C into

where tne
below,
120 demonstrated
in (ib) » as in
Que->Qui Rule- que
«

s e s the complementizer

talicci ed used instead


the usual

e qui (=queavoid ECP violation, (cf. Rizzi (1990)).


+ Agr) is
ia)- oEssentially
(ia)-to

) d. Qui penses-tu que est un peintre

D. Qui penses-tu qui est un peintre?


Relativized Minimality (RM) theory to properly head.
Rizzi's (1990)
traces. Accordingly, the
and hencelicense non-pronominal
govern,
as (400):
structure associated with (39) would be represented

t loves Mary]]
(40) Who do you think [c»t' ØAgr lip

that cannot be turned into an Agr element may be


The reason why
inert for proper head-
accounted for by the fact that it is inherently

government for traces.

clauses in Arabic
7.3.3 Subject extraction out of embedded

in Arabic, where the


Consider now the following example
subject/topic has been extracted:

(41) man ta-tadhakkaru ?anna ?iltaq-a: 1-wazi:r-a?


met-3MS the-minister-Ace
who 2MS-remember that

Who do you remember that met the minister?"

The SS representation of (41) is given as (42) below:

1-wazi:r-a]]ll
(42) man, ta-tadhakkaru [cp r' [c ?anna [Tp t; ?iltaq-a:

apparently banned
Movement of man (who) across ?anna (that) is
ntizer

This ban may be accounted for by the fact that the compie
that) is not an
appropriate head-governor for
the trace
ace in the
specifier position of TP, just like its
counterpart in English. But could
sepostulate that subject extraction proceeds
from aa postverbal
aasition in (41)? lo answer this
postverbal
question, consider first the
proposed
alternative structure for this sentence, with
pro now
filling the
specitier position of TP instead:

(43) ICPI man ta-tadhakkaru lcr f'i le Panna [Tp pro ?iltaq-a: t; 1
wazi:r-a]]]

If (43) is indeed the S-structure associated with (41), then there would
be a way of accounting for its ungrammaticality, namely in terms of
Relativized Minimality. The argument would go as follows: the A'-
specifier of TP, which is now filled by pro, would block antecedent
government of t by t'; -the latter occupying the specifier position of
the intermediate CP- and hence would induce an RM effect. But this

extraction
wOuld leave unexplained why (44), which features subject
across an overt clitic, i.e. -hu
4)
(44) man ta-tadhakkar-u ?anna-hu shtar-a: sayya:rat-an
a-car-Ace
who remeber-2MS that-him bought-3MS

Who do you remember bought a car?

Casc-asigning
demonstrated above that the
nas been clearly wilh a
Occurrence
immediate

complementiz
Uzer ranna (that) resists

expression
in Arabic. This
allows

phonologically empty (pro)nominal


in two
ditferent,
but
like (44)
s, in principle, O account for examples

"
related ways. One explanation is that subject extraction do.
does not
proceed from a postverbal position, but from the specifier (t.

position of TP. In such cases, however, -huu (him) is suffixed as


an
enclitic onto the complementizer ?anna (that). In this case, (44) would

pattern with (45) below, where the clitic -ha (her) similarly
incorporates into the verbal element, (Ouhalla & Shlonsky (2002):

(45) min ?illi 1-7asad ?akal-ha mba:riH?

who that the-lion ate-her yesterday

Who did the lion eat yesterday?"


Another explanation is that the extraction of man (who) in (44) does
indeed proceed from a postverbal position in the manner outlined in
(43). Within this conception, the pro empty category in ISpec, TPJ
turns into a clitic and subsequently incorporates onto the

complementizer ?anna (that). In the second case, it would be more


plausible to take -hu to be a spell out of the phi-features of pro,
specifically the Case feature. We shall assume this to be the case here
noting at the same time that the issue is far from settled.

Let us further call this strong- syntax-internal- Case feature a

functional feature to distinguish it from a lexical Case feature, i.e. one

that is associated with a lexical head. Now, pushing this assumption to

ts extreme, in
we may conclude that in Arabic, and perhaps one
1anguages, too, a functional category that has a strong Case feature

associated with it must


assign this feature to an adjacent overt in
CIement. When no lexical nominal element is available, Arabic u
language-spe salvaging device, viz.,
Case feature.
lexicalizing this
functiona
Tuever, the above
analysis does not explain
Cubiect wh-movement why both subject
and across ?inna
results in
ontences, or at best
quesuOndoic Ones,
ungrammatical
as the
(47) illustrate: 21 examples in (46) and

(46) a. ?ayya kita:b-in; qul-ta Pinna zaynab-a shtar-at ?


which-Acc book-Gen said-2MS that
zaynab-Acc bought-3FS
Which book did you say that
Zaynab had bought?"
b. Payya kita:b-in; qul-ta 2inna zaynab-a shtar-at-hu?
which-Acc book-Gen said-2MS that
zaynab-Acc bought-3FS-it
Which book did you say that Zaynab had bought it?"
47) a. man; qult-a ?inna ti za:r-a-kum?

who said-2MS that visited-3MS-you


"Who did you say that had paid you a visit?"

b.man, qult-a ?inna-hu; za:r-a-kum?


who said-2MS that visited-3MS-you

121
The are provided by
aticality judgements in (46a-b), (47a-b), and (48a-b)
Col grammar in
the
lleague, Abderrahmane Boudraa, a professor of Arabic theoretical
Department O Arabic Studies, Faculty of
ented herein
Letters, Tetouan.
The
claims

share in this regard


are entirely based these
on judegements. We absolute
Tremblay's (2006) idea "theories of
that grammar
should not
assume

for (.)
idiolectal

ogeneity y among
ogeneity speakers in order to be able
to account

differences
193
visit?
Who did you that had paid you a

Payyu kita:b-in; qul-ta


?inna zaynab-a qara?-at-hu:?
(48) a.

Which-Nom book-Gen said-2MS that zaynab-Acc read-3FS-it

Which book did you say that Zaynab had read it?

b. Payya-1-u sayya:rat-in; qult-a Pinna zayd-an sa:far-a fi:ha:

which-F-Nom car-Gen said-2MS that zayd-Acc travelled-3MS


in-it

Which car did you say that Zayd travelled in it?"

Let us start with (46a), whose derivation is given in (49):


(49)
CP

C'
Spec

layvakita:bin C" TP1

Spec

pro VP

qult-a Spec

pro ICP2

Spec

C TP2

Pinna zaynab-a qara?-at 2

the fact that


the first trace of the
notwithstanding
41S ruled out,
The latter verb
by the verb qara?at (read).
2 , 1IS head-governed
the Empty Category
therefore, the trace passes
Cla-governs it; intermediate
trace,
Furthermore,
the
above.
re defined in (33) that the strategy
Notice also
t, seems to res both clauses of ECP. not yield
a
respect
does
of lexicalizing object position
the trace in the of
dZing
illustrates.
The impossibility
(46b)
grammatical sent
d s e n t e n c e
either, as

195
environment is puzzling, to saythe least
extraction from this Perhaps
of the wh-phrase across
Dss a barrier.
related to the movement
a

the issue is
be CP2 by dint ofthe fact that it : sL-
But this proposed barrier cannot
which assigns it the theta.nel
marked by the verb qult-a (said),
Proposition.

My colleague, Abderrahmane Boudraa, (p.c) has pointed out to


me that the complementizer ?inna (that) in Arabic has an inherent

property that it does not allow wh-phrases to move across it. On the

assumption that this is indeed the case, ?inna (that) somehow tums C
into a blocking category in (46a-b). / * This should be intuitively
true, at least on empirical grounds. Thus, CP2 would inherit
barrierhood from C'. If this analysis is on the track track, the first
empty category, t2, will not be antecedent-governed, in contravention
of ECP. This seems to team with Fukui's (2006) classification of
barriers into strong barriers and weak barriers. According to this

classification, a maximal projection constitutes a barrier by default.


But it ceases to be one if L-marked; otherwise, it remains a weak

4 Mark Baker (personal communication) has pointed out to me ther


d
languages also bar extraction from CP, e.g. Slavic languages and some 9
man

dialects.
This presupposes à la
a modification of the definition of "blocking categorY
Chomsky (1986).
t is argued in Wexler and Manzini (1987) that two lexical items in one and the
same
language may select different values for a given parameter
posal

functional categories. In particular, extraction hilitieshus


may be extended to Ssibilitie
in
Arabic may be
argued to depend largely on the type of poo
capturing the noted differences between compleme Panna,

and believe-type verbs, whicn SE


say-type verbs, which select Pinna.
tor
barrier. How a
strong barrier
remains a
barrier
Apparently, this is what happens with even when L
clausal
the verb qa:la (say) in
Arabic, for otherwisecomplements of
of (46a- the noted
ungrammaticali
would go
unaccounted for.

4 Standard Arabic vs. Moroccan Arabic: A


approach Microparametric

Moroccan Arabic, however, both


In
subject and non-subject wh-
phrases are extractable across balli (that), as the
125
following examples
illustrate:
(50) a. shku:n; qulti [cp balli ti ja]
who say-2S that arrived-3MS
Who did you say had arrived?"
a. LF: for which person x, you said x had arrived

b. shman ktab; qulti [ce balli qraw tlamad ri]


which book say-2S that read-3P the-pupils
"Which book did you say that the pupils had read?"
6. LF: which x, x abook, you said the pupils had read x

C. shman ktab; qulti [cp bolli qraw-ah; drari


which book say-2S that read-3MP-it the-pupils
had read?
Which book did you say that the pupils
read x
C.L: which x, x abook, you said the pupils had

be
This variat and Moroccan Arabic may
auon between Standard Arabic
Baker (ibid)
back to whwhat Baker (2008) terms
microparameters.

(that) in SA.
The compleme balli in MA corresponds
to Pinna
entizer

197
are microparameters
account for cross-dialectal
that account cros-
argues that there
between two related dialects. If w e e.
differences, i.e. differences further
assume that these microparameters are associated with lexical

clear why (S0a-c) are ruled in in Morocos


categories, it will become
Arabic, while their counterparts
are ruled out in Standard Arabic, (ef

(46a&b) & (47a)). More specifically,


the clausal complement of the

verb qa:1 (say) will not constitute a strong barrier in the fomer

126
language.

Let us now consider (48a&b), which seem to diametrically

reject the analysis proposed above, for the wh-phrases in both of them
appear in the specitier position of the matrix clauses, whereas the
direct object and the genitive expression are substituted by a

pronominal clitic. But if we assume that in (48a&b) wh-movement


targets the topic of the matrix clauses, and not an element from within
the embedded clauses, if becomes lucid clear why (48a&b) are ruled
in, while (46a&b) are not. This is further corroborated by the fact that

the moved topics carry (default) Nominative Case. In view of what

has been said, the derivation of (48a), for example, would then

proceed as in (51), where the trace of the moved wh-phrase sits in the

specifier position of TP1:27

126
Macroparameters, on the other hand, are associated with general princP
n-
Thatexplain natural language, e.g. the pro-drop parameter, the level at wu
movement operates (English vs.
Japanese), etc.
we have alluded in footnote 86 to the fact g r a m m a r i a n s

take the DP in
that traditional Arad B
SVO structures to be
the inherent topic of the sentenc ved

nat subject topics in Arabic do not need to be generated in [Spec,


therein for them to be IOpr
marked as topics.
u kita:
IcPI Payyu kita:bin [c
(51) lrei t lr lr
a qara-at-hu]]]ll] qult-a [vetlcz ?inna
zaynynab-
With regard to
(47ab), their status is a
(47a)
(46a&b).( resembles
bit different from
of (41), repeated for that
convenience as (52):
(52) man; ta-tadhakkar-u [cp ?anna t; ?iltaq-a:
who 2MS-remember
1-wazi:r-a?]
that met-3MS the-minister-Acc
Who do you remember that had
met the minister?"

The ungrammaticality of
(52) has been ascribed to the
that-trace
effect. the latter reducible to ECP, as we have
seen above. The same

analysis applies to (47a). But (47b) poses a


problem of a different
type; it is neither completely
grammatical nor completely
ungrammatical, hence the question mark. One way to account for it is
in terms of Pesetsky's (1987) proposal that there are two types of wh-
movement, namely D-linked- e.g. ?ayyu kita:bin (which book) in

(48a) and non-D-linked wh-movement- e.g. man (who) in (476). The

p e presupposes the existence of a specitic set of entities

Ontained in discourse- e.g, books in (48a). Therefore. (48a) is asking

about dcating/particularizing which item of the set of entities you

the other hand. does


Said that aynab
Zayna had read. The second type,
on

The diference
not presuppo similar set of entities.
OC Uhe existence of a

but it has
repercussions in
etween the two ypes
O is semantic in nature,
on
sentence

well as
syntax- pecially at the level of
LF-as
QDIc

processing. On this analysis, the questionable status of (47b) will he

set back to the fact that the pronominal clitic on the complementizer
er
Pinna (that) refers back to wh-phrase that is
a
non-D-linked
30
(nonspecific), namely man (who).

7.5. Argument-adjunct asymmetry

Argument-adjunct asymmetry has for many years been a


topical issue in the Principles and Parameters theory. To illustrate with
concrete examples, consider the following pieces of data:

(53) a. what does Paul wonder whether John will


give to Mary?
b. 'ma:dha: ta-tasa:?alu hal shtar-a:
zayd-un?
what 2MS-wonder whether
bought-3MS zayd-Nom
"What do you wonder whether
Zayd bought?"

(54) a. how does Mary wonder whether John


will fix the fan?
b. kayfa ta-tasa:?alu hal arsal-a zayd-un l-milaff-a?

128
A similar
proposal is advanced by Enç (1991) for nominals in
Enç (ibid) points out that specific terms of specificity.
set of
arguments bear an index on D which theidentifies
individuals of which the
have no such feature on D. If argument
is
member. But non-specific
a
we extend Enç's theory of arguments
D-linked wh-phrases such as nominals to wh-phrases,
?ayyu kita:bin (which book) and
(which car) would be specific, hence the ?ayyatu sayya:ratin
**"
grammaticality
of (48 a&b).
Notice that the
pronominal clitic on Pinna (that) in
phrase that is D-linked. (48a&b) refers back to a w
*This analysis may not
hold for (46a&b) because wh-movement
the embedded clauses.
proceeds trom
2MS-wonder whether sent-3MS
how
zayd-Nom the-file-Acc
How do you wonder whether Zayd sent the
file?"
To gain some understanding of what is
going on here, we have
orovided the derivations of (53a&b) in (55a&b), and those of (54a&tb)
in (56a&b):

(55) a. CPI what; does [TP1 Paul wonder jcP2 whether


[rP2 John wil
give t to Mary?]]

b.CPI ma:dha:; [TP1 ta-tasa:?alu [cez hal [TP2 shtara: zayd-un t


amsi?]]]
(56) a. lCP how; does [TPI Mary wonder [cp2 whether [TP2 John will fix
the fan ?]]]]

b. TCPi kayfa; [TPi ta-tasa:?alu [cez hal [rP2 arsal-a zayd-un l-


milaff-a t?]]]

Obviously, (55a&b) and (56a&cb) all involve the movement of a wh-


element from an embedded interrogative clause across a filled comp
position, thus failing the locality condition of Shortest Move (the

bounding conditions on overt movement). Yet, argument extraction in


(55a&b) is degraded, but it is not strictly ungrammatitrecal. On the
other hand, adjunct extraction in (56a&b) is strictly impossible.

for examples such as (55b) and (566) in


Chomsky (1995: 90) accounts

terms of the following economy condition:

(57) Minimize chain links

201
Principles and Parameters Theory:
Towards a Contrastive Syntax
ax of English
English and
and

Arabic

He (ibid) argues that when a chain link 1s created by movement, the


e
chain induces a subjacency violation if (57) 1s violated in its creation

If this violation survives at LF, the chain has an additional effect of an

ECP violation. In (56a&b), the operators how and kayfa cannot govern

their respective traces across whether and hal, which are themselves
operators, and hence potential governors for the traces left in the base-

position. Therefore, (56a&b) exhibit a subjacency effect as well as an


additional ECP violation, and are thus strictly impossible in English
and Arabic. (55a&b), on the other hand, the argument traces are
licensed by being properly head-government by the verbs give and
shtara: (buy), respectively. Therefore, (55a&b) exihibit only a

subjacency violation and are thus less deviant than (56a&b).


ENERCISES

1: Draw a tre diagram for each of the following


Evercise
examples.
indicatung what transformation(s)- if any- took place

(a)(pp The man (who/that/ you talked about] is my boss.

(b)[Dpal-walad-u lladhi: takallamt-a maEa-hu] dhakiyy-un.

What did you wonder lcP whether John gave to Mary?

d) Who seems [rp to have been injured?|

Exercise 7.2: Explain the ungramnmaticality of the following


examples.

(a) Dp al-walad-u lladhi: takallamt-a ma&a]


dhakiyy-un.
(6) What, did you suspect [Dp the claim
[cp that John bought ti?]]
C) CP What; did you wonder [CP2 who; John gave t; to ty?]|
a) What, did you forget [crz how, I fixed t,
t'
Towards a Contrastive Syntax of Englishh and Arabic
Principles and Parameters Theory:
.

VII. BINDING THEORY

This chapter will discuss the following topics:


The interpretation of overt anaphors and pronominals in

English
Principles A and B of Binding Theory
The notion of accessible SUBJECT
Overt r-expressions in English and principle C of Binding

Theory
Binding Theory and empty categories
The distribution of overt DPs in Arabic

distribution of NPs as determined by


Chapter III dealt with the
their Case properties in combination with the principles of Case

theory. We saw that all overt NPs occur in a position where they are

Case-marked. In this chapter we will deal with the distribution of NPs

as determined by their referential properties in combination with the

principles of Binding theory.5


semantic theory that is concerned with the
Binding Theory is a

two DPs at the level


possible referential dependency relation between
structural
of LF, i.e. with how to semantically interpret certain
conditions that
relationships between different types of DPs. The

phrases that have to


noun
Binding Theory "is concerned with connections among including tne
do with such semantic properties as dependence of reference,
52).
connection between a pronoun and its antecedent. (Chomsky (1988:
"
M K .E n n a s S s i r i

gOveI the istribution of these DPs are thus purely syntactic in

nature

1 The interpretation of overt nominal expressions in English


8.11
11Anaphors, pronomimals and BindingConditions
Binding is a semantic relation because it involves reference, the

dT being a relationship between an expression and the outside

rld. As far as their referential properties are conccrned, overt NPs

all into three categories: anaphors, pronominals, and r(eferential)-

etoressions. The first type includes reflexive anaphors, such as

mself himself, themselves and reciprocal anaphors, such as one

This class is characterized by the fact that its


another and each other.
elements do not have
intrinsic reference; rather, they depend for their

interpretation on an antecedent. The second type comprises pronouns

like , you, he, us, them, etc. Unlike anaphors, pronominals may have
class consists of names such as Zayd.
independent reference. The third
NPs such as the book and the students of
John, Mary and referring

two coreferential elements.


.Binding is thus a structural relationship between
and
133 two values of the features anaphor]
Chomsky (1982:78-83) uses the
manner:
the following
pronominal] to characterize overt DPs in

A. Reciprocals and reflexives


+Anaphor, -Pronominal
B. Pronouns
FAnaphor, +Pronominal]
C. R-expressions
-Anaphor, -Pronominal]
Class Four. This category of NPs get their meaning through dite.
ect
(deictic) reference to a specific entity in the outside (extra-linguistic
stic)
world.
The question Binding Theory strives to answer is: When do we

use anaphors. pronouns, and r-expressions? The core facts relevant to

this issue are given in (1) below


(1) a. John, likes himselfj. (himself must refer to John)

b. John, likes him. (him cannot refer to John)

c. John, believes that Mary likes him,j. (him may refer to John
or to someone else)

d. John; believes that Paul; likes himselfij. (himselfcannot


refer to John)

The above examples allow us to make the following statements

concerning the referential properties of anaphors and pronominals:

a. A reflexive must be coreferential with (refer back to) an


(2)
element within its clause.

b. A pronoun cannot be coreferential (refer back to) an element

within its clause, but may or may not refer back to an element

in the sentence.35

a reflexIVE
Chomsky (1973) points out that in an environment which allows
pronoun, e.g. (1a), a non-reflexive pronoun, or a full NP cannot
refer back to tne
antecedent, as in (1b).

We shall see below that some syntactic constraints apply if there is coreterene
M.n

thereflexive anaphor, himse picks out its reference from the


In
(1a)
NP,,John,
which serves an antecedent for it." Coreference is
subject

by co-indexation, and different reference, i.e.


contra-
indh
catedhere
is indicate by different indices. In (la), then,
John binds
indexation,

first principle of the Binding Theory can thus be defined


vlf
himse
The
below:
as in
(3)

3)Principle A of Binding Theory (preliminary)


(3)
An anaphor must be bound.

The term bind is itself defined as in (4):

4)
(4) Binding: a binds ß iff (read if and only i)
1.
ac-commands B and

a and B are co-indexed.

binding, consider the


For further illustration of the notion of
structure of (la) given in (5):

136
NP from which an
anaphor or pronoun draws its reference is called the
E
anfecedent. The way we normally indicate that two NPs are co-referential is by
mea
n index, usually a subscripted letter or number. Two NPs that share the
dexes/subscripts- i.e. coindexed- also share the same referent.
TP
5)

NP

T VP
John,

Spec

V NP

likes himselt,

the bindee-
NP John-the binder/antecedent- and himself-
are
In (5) the
clause. Clause (4i) ensures that the antecedent
contained in the same

the sentence would be ruled out,


must precede the anaphor; otherwise,
c f(6).137

(6) Himself likes John;,

(6) has the following S-structure:

137 A pronoun may precede the element to which it refers. This is a case or

cataphora (forward reference), as in (i) below:

(i) Before he, died, our neighbour had donated his entire fortuneE

charity.
MK. Bnmlus.)

TP

NP T

himself T
P
V NP

likes John

There is nothing in the above structure which c-commands and is

coindexed with himself. The potential antecedent of the reflexive

anaphor does not c-command it. This contravenes (41), and, therefore,

(6) is ruled out. In addition, the reflexive and its antecedent, on which
it is referentially dependent, must share phi-features of person,

number, and gender. Failure to do so will result in ungrammaticality,

too:

(8) John; hurts herself

In this respect, Haegeman (1991: 1930 points out:

The requirement that a reflexive and its antecedent agree with respect to their

nominal features follows from the fact that the reflexive depends for its

share
nterpretation on the antecedent, i.e. the reflexive and its antecedent
their referent. It would be rather odd to find that a reflexive has the property

referent to a male
malej, for instance, thus constraining the selection of the
y , and is co-indexed with an antecedent which
itself has the property I-
Principles and Parameters Theory: Towards a Contrastive Syntax of English and Are

malel. There would be a contradiction in the specification of the relevant

referent.
properties for the selcetion of the

Consider now the following examples:

(9) a. [TPi the coach; suspects Icp that [Tr2 Ihe players, blame each

other]l1
b. [TP the players, suspect [cp that [Tr2 lhe coach; blames each

other:]]]

ruled in that the reflexive anaphor each other is indeed


(9a) is given
bound in TP2, in that it has an antecedent within TP2, with which it is

co-indexed. Notice that the coach and the players are contra-indexed,
i.e. they carry different indices. This should be taken to mean that the

coach and the players don't have the same referent. But (96) is ruled

and bears the


out, for although the NP the players c-commands
same

index as the anaphor, it cannot bind it because the anaphor each other

is too deeply embedded in the syntactic structure. In other words,

condition (4i)- i.e. the fact that an anaphor must be c-commanded by

its antecedent- is not enough to determine the distribution of anaphors


in English. Therefore, we need a more precise definition of Binding
Condition A, one that should precisely indicate the domain in which

an anaphor must be bound. The grammaticality judgements in (9)

suggest that an anaphor must be bound within a "local domain". We

can thus restate BCA as (10) below: 8


M.K. E n n a s s i r i

Drinciple A of Binding Theory (revised)


(10)
An anaphor must be bound within its local domain'.

The
mmatical status of (9a) suggests that a local domain for
grammati

aphor binding in English ought to be the smallest clause/TP

sontaining the anaphor. However, the example given in (11) below

h s that an anaphor may also be bound within a domain that is

smaller than a clause, i.e. a DP


stories about each
(11) IP the coach; heard [DP [the players
otherijll
in (10), let deal
Before further tightening of the definition
us

The data in (la-d) show that


with the distribution of pronominals.
occur in environments where anaphors are
pronouns felic1tously
the case, principle B of Binding Theory may
disallowed. This being
approximately be defined
as (12) below:

(preliminary)
(12) Principle B of Binding Theory
free.
A pronoun must be

examples:
But consider now the following
blames
that [Tr2 the coach;
suspect [cP
(13) a. [TPI the players;
themAl] blamne
that [TP2 the players;
suspects [cP
6. TPI the coach;

them-a]J
c. [TP the coach, heard [Dr [the players']j stories about

themll
(13a-b) are identically constituted as (9a-b), except that they include

pronouns in the slot occupied by the anaphors in (9a-b). This indicates

that there is a complementary distribution between anaphors and

pronouns, i.e. where anaphors may occur, pronouns may not, and vice
versa. (13c) further illustrates that a pronoun may not be bound within

a DP that includes it. However, the data in (13a-b) show that a

pronoun may indeed be bound to an antecedent occurring outside its


minimal domain.Similarly, we can now restate principle B of
Binding Theory as follows:

(14) Principle B of Binding Theory (revised)


A pronoun must be free in its 'local domain'.

The next question we should address is what is an appropriate


definition of the notion local domain'? Basing ourselves on data from
English, (cf. (9a-b) and (11)), we have concluded above that the local
domain for anaphors ought to be TP and DP. However, the following
data casts doubt on this:

(15) a. John, considers [Tp himself to be the best


candidate]
139
(13b), the pronominal them must be free in its local
it to be
contraindexed with the players. domain, and this requires
outside its containing However, it may be bound to an element
clause, as in (13a), or it may
discourse antecedent, (cf. index k acquire its reference from Some
in (13a-c)).
Compare with (i), which is ruled out:
(i) John; considers [Tp him, to be the best
candidate]
M.A. DlU

b, Maryi drew [Dp a picture of herself1

ECM construction, where the anaphor himself is bound to


is an
(15a)
lalhn outside its local domain, 1.. TP. Similarly, hersel/ is bound to

1ar outside its local domain, i.e. DP. This suggests that principle A
Rinding Theory should be sharpened to accommodate the apparent

ontradiction in (15). In the standard' Principles and Parameters

theory, this sharpening is done by introducing the notion of governing

category, as in (16) below:

(16) Principle A of Binding Theory (final)


An anaphor must be bound in its governing category.

The notion governing category is itself defined as (17):

minimal category
(17) Bis the governing category for a iff Bis the
governor of a, and a SUBJECT accessible to
containing a ,
a

a141 (cf. Chomsky (1981: 211))

(18) a. l,... 8...], where y and o bear the sameindex.

b. Bis accessible to a iff a is in the c-command of Band


assignment to a of the index of ß would not violate (18a).

141 sitting NP in [Spec, TP] and [Spec,


The
P. term
This
subject is understood here to be any
solves the puzzle between the grammaticality status of (ii) and (ii) below:

is awful.
(i) John, thinks that [De Paul's picture of himself]
is awful.
() John, thinks that [pp this picture of himself]
of a subject inside
alference between (i) and (iü) relates mainly to the presence
domain where
extends the local
in (16)
an
t seems, then, that the definition
anaphor must be bound.
(cf. Chomsky (1981: 212)

calcgory, consider tho


To illustrate the notion of governing e

structure of (15a) given in (19) below:


following

(19) TP

NP

Joh VP

TP

considers NP

- --*
himselfi to be the best candidate

In (19), the governor of himself is the verb consider. The minimal

domain containing himself, its governor and an accessible subject is

the matrix clause, TP'. Therefore, himself must be bound in this


domain, which is indeed the case here. Following the working
hypothesis in the P&P framework, we shall assume that the notion of
governing category also applies to pronouns. Thus, principle B of

Binding Theory is redefined as (20):

(20) Principle B of Binding Theory (final)


A pronoun must be free in its governing category.
M . K . E n n a S S i r i

To see
how (20) accommodates examples such (13a), repeated below
as (21). consider the structure of this example given in (22):

TPI the players; suspect lcr that [r2 the coach, blames
(21)
them]]
TP
(22)

NP T

the players; T VP

CP

suspect Spec C

C TP

the coach; blames them;


that

blames. The minimal domain


n (22), the governor of them is the verb
governor and
an accessible subject is
COntaining the pronoun them, its
and this is indeed
Therefore, them must be free in this domain,
for
any other pronominal element,
uc case here. The fact that hem, or

clement outside its


governing
tha
natter, may be bound
to an

an antecedent
the sentence from
ate r y or select a
O reference outside

215
that has already been introduced in the discourse is irrelevant
to

Binding Theory as understood here.

8.1.2 R-expressions and Binding Theory

Up to now, we have dealt with the distribution and


interpretation of anaphors and pronominals in English from a binding-
theoretic perspective. But Binding Theory also characterizes the co-
indexing possibilities of R-expressions/names. Consider the following
examples: 42

(23) a. John, likes John.

b. He; likes Johni.


c. John; thinks that Mary likes Johni.
d. John; hates John;'s teacher.
e. He; hates John,'s teacher.

These facts suggest the following generalization:


(24) An R-expression must be free, (where free means "not
bound").
Consider the following example:

(23b) and (23e) may be filtered out


i) by the following condition:
Disjoint Reference Condition (DRC)
A pronoun X may not refer to the
as) a nonpronominal NP Y if
same thing as (have the same inde
X c-commands Y.
Aronoff & Rees-Miller (Baker (2003), ed
(2003:276))
Te theyi say lcp that [Tp2 the coach blames the players,11
(25)

shows that the notion of governing category defined above does


(25)
nlay a role in the distribution of r-expressions. Therefore, principle

C of the Binding Theory, which regulates the distribution of r-


eNDressions, 1s not expected to make reference to this notion.
ex
Accordingly. principle C should be strengthened to (26) below:

Principle C of Binding Theory (final)


(26)
An r-expression is free everywhere.

In Chomsky (1986a), the binding conditions discussed above


are reformulated as licensing conditions on a governed element a in

the expression E with indexing I in the manner outlined in (27) below:

(27) The licensing conditions on govemed elements


For some b such that (1) or (ii), I is BT-compatible with (a. b):

i) a is an r-expression and
is variable
(a) b is the domain of the operator if a a or

(b) b= E otherwise,

(i) a Is an anaphor or pronominal and b is the least


Complete Functional Complex (CFC) containing r for
which there is an indexingj BT-compatible with (a, b).
below:
Bl-compatibility is in turn defined as (28)

(28) Tis BT-compatible with (a, b) if:

(a) a is an anaphor and is bound in b under I.

17
in b under I.
(b) a is a pronominal and is free

and is free in b underI.


(c)a is an r-expression
As for the term Complete Functional Complex (CFC), it is the
minimal XP where "all grammatical Junctions compatible with ite
head are realized in it- the complement necessarily by the Projection

Principle, and the subject, which is optional unless required to license

a predicate, by definition." (cf. Chomsky (1986a).


Interstingly, Binding Theory is also relevant for the theory of

movement, where the moved constituent, i.e. the antecedent, binds its

trace. To illustrate, consider the following examples:

(29) a. [TP John; was attacked t]

b. [cP What; [c will [Tp you do t:?]]]

In (29a),.John A-binds its trace by virtue of the fact that John


occupies an argument position, namely [Spec, TP]. Therefore, John
and the trace are said to form an A-chain. From a binding-theoretic
perspective, the trace of A-movement- e.g. (29a)- is a non-overt
anaphor, and hence subject to principle A of Binding Theory. This is
indeed the case in (29a), where ; is bound
by John; in TP, the latter
being the governing category for the non-overt
anaphor. In (29b), on
the other hand, the wh-trace
occupies [Spec, CP], which is an A
position as no
grammatical function is assigned to this position, (CI.
Chapter IV: 4.3). Accordingly, the
wh-phrase is said to A-bar bind ls
trace (alias
variable). As with A-movement, the and he
variable form an
wh-phrase u
A-bar chain. From a
binding-theoretic perspecue
M.A

the
trace of A-bar movement- e.g. (296)-
too,
is
as a viewed
non-overt
a therefore. prineiple C of
r-eNpresIon;
Binding Thcory requires that it
hinding from an A-positic This is indecd
be free o f b i n d
the casc in
a h the trace is bound to
(295),
what, the latter sits in an A-bar
Osition. It may then be concluded that for Binding
P o s

Theory the term


dine is used to reter to
A-binding, not A-bar binding, i.e. to the
association of a DP with an A-antecedent.

8.2 The distribution of overt DPs in Arabic

8.2.1 The interpretation ofanaphors andpronominals

Consider now the following examples in Arabic:

a. zayd-un; jaraH-a nafs-a-hu;/k. (nafsahu must refer to Zayd)


(30)
Zayd-Nom injured-3MS self-Acc-him

"Zayd injured himself."

nafs-a-
b. ya-shtabihu zayd-un; fi: ?anna Eamr-ank ya-krahu
hu-k (nafsahu cannot refer to Zayd)
self-Acc-him
that Amr-Acc 3MS-hate
3MS-suspect Zayd-Nom in
hates himself."
Zayd suspects that Amr
clitic - u cannot
(the pronominal
C. zayd-un; yu-Hibbu-huri/k.
refer to Zayd)
Zayd-Nom 3MS-likc-him

Zayd likes him.


d. qa:1-a 1-mutafarriju:na, ?inna lla:Eibi:nak ?ashEaru:-hum
bi-1-malal-i. (the pronominal clitic hum may refer to -
mutafariju:na or to other discourse or world entities, but not

to lla:&ibi: na)
said-3MS the-spectators that the-players bored-them

"The spectators said that the players bored them."

(30a) exhibits the linear SVO order. In Ennassiri (2014b), it is argued


that this order results from the operation Merge. i.e. the topic Zaydun
merges with T' to form a full TP. So to determine the distribution of
anaphors in Arabic, consider the structure associated with (30a) given
below as (31):

(31) TP

NP

Zaydun T VP

NP

pro; NP

jaraH-a nafsahu;
In the above
structure, Zaydun c-commands nafsahu
(himself) in that
every XP that dominates
Zaydun- i.e.the matrix TP- also dominates
nafsahu. Therefore, Zaydun must function
as the antecedent of tne
anaphor. In (305), the anaphor
nafsa-hu (hinmself) occurs in
n
clause, and so the correct facts
embedded
follow naturally from
A
principle A
arinciple of the Binding Theory. 43
To illustrate further, consider
the structure associated with (30b) given in (32):

(32) TP

T VP

ya-shtabihu zayd-un; V

PP
binding
domain for
P CP the anaphor

fi
C Tp2

anna Eamr-an, ya-krahu nafs-a-hu-k

In the above structure, the governing category for the anaphor nafsahu

(himself) is TP, and nafsahu is indeed bound to Eamr-an (Amr)


Within this governing category. More crucially for our analysis here is

ne Tact that coreference between the anaphor nafsahu (himself) and

yaun (Zayd) in the matrix clause is barred for the same reason asS

udt advanced for (96) in English, i.e. in terms of locality violation,

143
The that the
in (30b) exhibits VSO order. Ennassiri (2014b) argues
matrix clause in
matrix of T is not strong
d t y in Arabic is a function of the fact that the D-feature
in the overt syntax.
5 attract the subject DP to a preverbal position
A violation. So given the
to Binding Principle
itself reducible
in (30a&b), it may be concluded tha
grammaticality judgements
applies in exactly the same way to both
Theory
principle A of Binding
English and Arabic.

Consider in this regard the


Let us turn now to pronominals.
,144
below:**
structure associated with (30c) given in (33)

(33) TP

NP

Zaydun; T VP

NP

proi NP

juHibu -hu*i/k

I n Ennassiri (2014b:97), it is argued that the topic in [Spec, TP] is nticall


Iinked to the null thematic subject, pro,both depict the same individual.
as
resembles (i) in English, where the reflexive anaphor, himself, is bound to PRO
the subject position of the infinitival complement clause, and PRO refers back
John:
() John, likes [Te PRO, to talk about himself]
In clitic -hu (him)
(33), the pronominal cannot refer to
C e in Arabie, too, Zayd. This is
SO
pronouns and their
ontained in the same domain as
antecedents cannot be
anaphors, (cf. (30a&b). This is
Grther illustrated in (Jua), wnere the
lates the clause-mate condition
antecedent-pronoun relation
posited for anaphors. In this
narticular case, there is a finite clause boundary- TP- between the
litic pronoun -hum (them) and its
antecedent, -mutafarriju:na (the
spectators):

(34) TP

VP

qa:la NP binding
domain for
-hum (them)
-mutafarriju:naj CP

C TP2

Pinna
lla:Eibi:nak ?ashEaru:-hum"w bi-l-malal-i

entertain a wider range of


4) shows that in Arabic, too, pronouns
than anaphors. The same
Structural relation with their antecedents
is given in (36):
Observation holds for (35). whose structure

223
IC

(35) zawj-u, hind-in yu-dallilu |-hak/w


nafs-a-ha:h/ nafs-a-hu
husband-Nom Hind-Gen 3MS-pamper her/ self-Acc-her/hiim

"Hind's husband pampers her/herself/himself."

(36) TP

(DP:) T

Dgen NP VP

zawju DP N' yu-dallilu DP

hind-ing N pro DP

t
nafsa-ha:k
nafsa-h
-ha:kw
In (36), the reflexive
anaphor nafsaha: (herself) may not be bound to
the encircled DP, although DP' does indeed
c-command the anaphor.
This is so because the
gender feature of the potential antecedent,
zawju hindin (Hind's husband) clashes with the
gender feature of the
anaphor. But recall that anaphors need a clause
mate antecedent wiu
Which it must share
all o-features. The
Oes not
potential antecedent, hinain,
C-command nafsaha:, either. The
first branching noade
M . K . E .

hindin, i.e. the encircled NP, does not dominate the


dominating

anaphor. This
This means that nafsaha: is free in TP, in contravention

A o fBinding Theory.
principle

The sentence is ok with the clitie pronoun, -ha: (her), referring

SSessor inside the subject DP', i.e. the NP hindin, because the
-ha:.
NPdoes not c-command Principle B of Binding Theory is
latte.

therefore respected

8.2.2 The interpretation of r-expressions

It has been seen above, subsection 8.1.2, that the interpretation

of r-expressions
in English is regulated by principle C of Binding
be free everywhere. Let us address
Theory, which states that they now

the case of r-expressions in Arabic. For ease of illustration, consider

the examples given in (37) below:

(37) a. zayd-un; ya-Htarimu zayd-an;


Zayd-Nom 3MS-respect Zayd-Acc

Zayd, respects Zayd;."


b. pro/hiyya, tu-Hibbu zawj-a hind-inj

she 3FS-likes husband-Acc Hind-Gen

"She; likes Hind's husband."

1-¬uzlat-a
C.qa:1-a pro;/zayd-un; ?inna 1-?aHmaq-a; yu-Hibbu
the-solitude-Acc
Said-3MS Zayd-Nom that the-fool-Acc 3MS-like

Zayd said that the fool; likes solitude."


d. huwwa,/pro, qa:1-a ?inna zayd-an, yu-Hibbu 1-¬uzlat-a

3MS-like the-solitude-Acc
he said-3MS that Zayd-Acc

"He, said that Zayd, likes solitude."

(37a-c) indicate that Arabic proper


The grammaticality judgements in
not tolerate a referentially
names and referring expressions do
for instance, the structure
dependent interpretation. Consider,
associated with (37a) given in (38) below:

(38) TP minimal
governing
categoryfor
DP zayd-an

Zayd-un; VP

ya-Htarimuk DP

pro; V DP

tk zayd-an;

zayd-an, which
in [Spec, TP] binds the proper name,
In (38), the topic
Condition C effects. The same thing holds for
fact induces Binding
r-expression must be pronoun free. ln
(37b), which indicates that an

containing the r-
(37c-d), the antecedent is outside the minimal clause
sentences instantiate
expressions, -?aHnaq-a and zayd-an. Yet, both
names ana
a
Binding Condition C violation. So the restriction that
expressio may not be bound by names
eferring
and
Arabic,
1.e. r-expressions have to be
pronouns holds
in
too. More generally. still, the
necessarily free in this
language,
principles of Binding Theory
above are represented in all human
languages and are thus
covered

of UniversalGrammar.
part
EXERCISES
Exercise 8.1: Explain why (a-f) are ruled out in English
(a) John; believes that himself, is the best candidate.
(b) Jane; resents Bill's criticism of herself.
(c) Mary; admires himselfi.
(d) He, thinks [that the fans admire Bill,]
(e) T expect [themselves, to help the students]
(f) She; expects [Mary, to do better]

Exercise 8.2: Analyse the grammatical status of the


following pairs of
sentences in Arabic from a
binding-theoretic perspective:
i. a.'qa:1-a l-mutafarrij-u:na ?inna lla:8ib-i:na; a:dha-w ba£D-a-
hum; baED-an.
said-3MSthe-spectators-Nom that the-players-Acc hurt-3MP
Some-Acc-them-Acc
The spectators; said that the
players; hurt each other."
b. qa:1-a l-mutafarriju:na; Pinna lla:Eibi:na; a:dha-w ba£D-a-hum;
baED-an.
said-3MS the-spectators-Nom that the-players-Acc hurt-3MP
some-Acc-them-Acc
The spectators, said that the players; hurt each otheri."

ii. a. zayd-un; ya-Etaqidu nafs-a-hu; dhakiyy-an


Zayd-Nom 3MS-believe self-Acc-him clever-Ace
"Zayd belicves himself to be clever."

b. zayd-un; ya-Etaqidu ?anna nafs-a-hu, dhakiyy-un


Zayd-Nom 3MS-believe that self-Acc-him clever-Nom
Zayd believes that himself is clever."
IX. CONTROL THEORY

discuss the following topics:


Thischapter will
The distribution of PRO in English
The nature of PRO

Control structures in Arabic: PRO vs. pro

9.1 The distribution of PRO in English

In chapters VI and VII, two types of empty categories were


discussed, namely NP-trace and wh-trace. The present chapter deals
with the distribution and interpretation of a more intriguing empty
category, i.e. PRO. The latter category may only appear in the subject

position of infinitival cBauses in English, i.e. clauses where T is


characetrised as (-finite]. The data relevant for the distribution of PRO

are given below:

(1) a. PRO likes syntax.


b. John likes PRO.
PRO.
C. We have already talked about
d. We believe PRO to be the best candidate.

in
is the fact that PRO may not appear
o noticed in (la-d) above
Cas ldrked positions.4 In (la). for instance, PR0 sits in the subject
5 put that PRO is assigned
null
But see Chomsk for a proposal
omsky and Lasnik (1993) "Efinitel*
Lase, which only PRO feature is licensed Dy
Cn can bear. This Null case
where finite T assigns Nominative Caco
position of a finite clause, to
of m-command, (ct. Chapter III). In (2b). PRO
Spec, TP] in terms

appears in the object position, where Acc Case is normally assigned


inside a prepositional phrase.
by the transitive verb. In (2c), it appears
where the prepositional object normally gets Oblique/Accusative Case
from the preposition. And in (2d), it appears in an ECM construction.

where the embedded clausal complement is transparent for Case

matrix transitive verb. From a Case-theoretic point of


marking by the
view, then, overt DPs and PRO seem to be in complementary
146
distribution, i.e. where the former may appear the latter may not."

9.2 The nature of PRO

Consider now the following examples:

(2) a. Mary wants [PRO to work hard]


b. John promised Mary [PRO to be on time]
c. John convinced Mary [PRO to leave the room]
d. It is not easy [PR0 to adapt to the weather in Alaskaj

On the face of it, (2a) resembles (3) below, where the embedded

clause is also infinitival in nature:


3) Mary seems [to work hard]

(1a-d) would all be ok with an overt DP


replacing PRO.
Dut the syntactic structures or (2a) and (3) are
quite different. In
particular, (3) is an instance of NP
movement, and so it has the
derivation given in (4), (cf. Chapter VI:

(4) Mary; seems [Tp t; to work hard]

In the earlier days of generative


grammar, constructions such as (3)
were analysed in terms of
Subject-to-Subject Raising, whereas those
like (2a) were treated in terms of Equi-NP Deletion, (cf. Rosenbaum
(1967). In P&P theory, constructions such as (2a) are treated as the
result of (obligatory) construal processes that relate PRO to an
antecedent in the sentence. Within this theory, (2a) would have the S-
structure given in (5) below, where the matrix subject, Mary, controls

PRO, i.e. Mary and PRO refer to the same individual in the outside
147
world. This is indicated via coindexing.
(5) Mary; wants [PRO; to work hard]

Chomsky (1982) argues that there is only one type of empty

determined.
category, but that its specific instantiation is functionally
mplicit in this statement is the idca that empty categories have no

have
nerent properties; rather, they derive whatever properties they

147 Hornstein and Polinsky (2010)


But
see, for example, Hornstein (1999, 2003);
treated as movement.
et al. (2010), where control structures are also
OOeCkx have the derivation
below is assumed to
(i)
c View, an example such as the
subsequently m o v e s to
in the lower VP and
(), where john merges
matrixi[Spec,
n TP1], targeting both [Spec, TP2] and [Spec, VP1].
John tried to embarrass mary. embarrass Maryl]l
Jehn to
(ii) lTP1 John lvP1 teha tried [rpz sehn to lvp2

231
from the environment in which they occur. Following this proposal

empty categories are assigned a value for each of the binary features
[+/-anaphor] and [+/-pronominal] by the mechanism of functional

determination stated below:

(6) The Functional Determination of Empty Categories

a. An empty category is a variable if it is in an A-position and


locally A"-bound.
b. An empty category in an A-position that is not a variable is
an anaphor.

c. An empty category that is not a variable is a pronominal if it


is free or locally A-bound by an antecedent with an
independent thematic role.
As stated in the previous chapter, this yields four different types of

empty categories with respect to the binding properties. Each of these


empty categories, except PRO, has a lexical counterpart:s

(7) a. [+anaphor, -pronominal] = lexical anaphors/ NP-trace

b. [-anaphor, +pronominal] =
pronouns/ (small) pro
c. +anaphor, +pronominal] = (big) PRO

d. [-anaphor, -pronominal] = R-expressions (names/ wh-trace)

It is easy to see from (7c) that PRO is a hybrid category, in that it has

the features of both an anaphor and a pronominal. Like pronominals,


PRO can either be free bound,
or as (8a) and (8b) show, respectively

However, Chomsky (1995: 41) takes English French on and German d to


one,
be partial overt counterparts to
arbitrary PRO and its restriction
PRO in that they share "the
modal interpretoO of
to subject position."
a. John is too stubborn [PRO to talk tol
(8)
b. John decided [PRO to vote]

the
Further, th antecedent of PRO, IT
any, has an
independent theta-role
aproperty which pronominals, but not
anaphors, have. But like
harS PRO does not have an
anapho
independent reference; rather, it is
referentially dependent upon another argument.
So, by being an
ananhor. PRO must be bound in its
governing category; and by being
a Dronominal, it must be free in its governing category. From
this, it
follows that the distribution of PRO is regulated
by principles A and B
of the Binding Theory. Consequently, and as a result of the
fact that
no element can simultaneously
satisfy these two conflicting
requirements, PRO must be assumed to have no
governing category;
thus, the PRO Theorem is derived:

9) PRO Theorem

PRO must be ungoverned.


A number of properties of PRO follow from the PRO theorem

subject position of
stated in (9). First, PRO may not appear in the
inite clauses, as it would otherwise be governed in that position by
In
ne verbal inflection or under spec-head agreement, (ct. (la).
of infinitives and
Onrast, it may only appear in the subject position
cannot be assigned.
So, the fact that
Case
s a position where

at
PROs are governed
149 controlled
that all
Kayne (1991:679) formulates a proposal
other things,
that PRO does

som
ome level Or
of representation. This
repres means, among
of (9).
deedVe a governing category, in complete violation
lexical counterpart is
accounted for by the fact that thio
PRO has no

be assigned Case.
category could
never

PRO Theoremn implies that the infinitival


Second, the
status of a CP. Perhaps the latter
complement has the categorical
barrierhood from TP and thus prevents PRO from
category inherits

being governed by the matrix verb. With regard to the verb want in
allow CP-deletion- as a lexical
English, it might be argued that it must
property- since an overt NP may appear in place of PRO. This is
illustrated by the following example:

(10) Mary wants [Bill to work hard]

The CP-deletion mechanism in (10) would allow the embedded

subject, Bill, to receive Case from the matrix verb, exactly as in ECM

constructions. But if that were the case, Chomsky (ibid) argues, the
following example would also be grammatical, contrary to fact:

(11) John; was wanted [; to succeed]

The ungrammaticality of (11) may be taken to mean that want-type


verbs do not permit CP-deletion as a marked option. Consequently.
Chomsky (ibid) argues that the lexical NP in examples such as (10)
receives its Case not so much from the matrix verb, since this option is

now barred, as from the prepositional complementizer for. The latter

complementizer undergoes a rule of for-deletion, in the PF level, "


instances where it in immediate postverbal
occurs positions, (compare
With Mary wants very much for Bill to work hard).
If this
is correct, why,
correct, wh then, is a similar derivation not possible
for
:deed for is selected
(11)?Ifindeed in (11), the trace will be
governed and
theECP wll eventually be satisfied. But it might be argued that for is
araner governor, as a result, (T1) would be ruled out as an
no
ECP
lation.1s0
violation. In fact, Chomsky (1981:452)
argues that the class of
ner governors
proper
should be restricted to lexical elements, i.e. to
having the features +N]
categories or
[+V]. thus excluding
arenositions from the
prep
class of proper
gOvernors. This, however, poses
problems for prepositions stranding in English, as the trace left behind
would not be properly governed, in violation of the ECP. We shall not
pursue this issue any further, but see Chomsky (1981: 492-493) for a

proposal.

The presence of PRO is required by the Projection Principle,


which demands that the lexical and theta-marking properties of heads
be represented categorically at each syntactic level. In (8b), for

example, the verb vote has an external theta-role to assign, and

Unerefore, there must be an argument available to receive it. From this


perspective, then, the Projection Principle should be part of the
aetinition of null pronouns, (cf. Gilligan (1987). The presence of
RO is also motivated by the fact that it is syntactically active. This is

PRO acts as a binder for the


Urared in (12), for instance, where
reflexive anaphor herself:

50 Th such as (i)
analysis may be extended to
instances of Comp-trace
below:
D.

succeed?
i) much for t, to
Who, do you want very

235
(12) Mary has decided [PRO to rely on herself]

The PRO Theorem has been met with scepticism, however, as


as

a number of linguists have argued that although the empty categorv

PRO has properties characteristic of both anaphors and pronominals, it


never shows up in a given construction having both properties
simultaneously, (cf. Koster (1984) and Manzini (1983), among
others). If this is correct, then PRO must be governed, and hence has a
governing category. As an anaphor, it will be A-bound in this
governing category; and as a pronominal, it will be A-free in it.

A more serious conceptual problem related to the PRO

Theorem is the assumption that in infinitival clauses in English, T


does not govern PRO. Within the GB framework, government was
determined in terms of m-command, so that a head can govern its
specifier. The stipulation that T is not a governor when it is
characeterised as [-finite] makes it more difficult to attempt to reduce
govermment to a configurational notion- i.e. m-command in this case-

ratherthan to a contentful notion.Further, there are empirical data


from several languages that show that PRO is licensed via
government. A case in point is Spanish, where a lexical

complementizer governs PRO in the subject position of an infinitival


clause, as we see from (13):
(13) Maria no sabe [c si [ip PRO comerlo]]
51.
nis "would be
equivalent, for example, to postulating that a constituent A T y
C-command a constituent al.
Y only if Xhas a given lexical feature." (Cf. Hornste
(2005: 129).
Maria not knows if to-eat-it

Maria does not know whether to eat it."

the PRO Theorem assumes control


A lso
to be a
property of
e
one single category, viz PRO. The prediction would be that a language
h as Arabic lackS COntrol structures as it does not have PRO among
nentory of its empty categories. This
prediction is not born out,
h a1oh. In the following subsection, it will be
seen that the null
suthiect of control complements in Arabic
belongs to the type pro and
is not a pronominal anaphor. So, to account for the
referential
Droperties of PRO, a new module has been introduced in the
grammar,
namely Control Theory, which specifies the structural and lexical
conditions under which the obligatory control
interpretation
occurs.

9.3 Control structures in Arabic

9.3.1 PR0 vs. pro


Let us now investigate, albeit briefly, the syntax of control

verbs in Arabic. To this end, consider the following examples:

(14) 3. yu-ri:du zayd-un [?an ta-Htarim-a zaynab-u nafs-a-ha:]|


MS-want zayd-Nom that 3FS-respect-subjunctive Zaynab-
Nom self-Acc-her

Zayd wants Zaynab to respect herself"

yuri:d-u zayd-un [?an ta-Htarim-a zaynab-u nats-a-huj


want-3MS M-Nom that 3FS-respect-subjunctive Zaynab-Nom
self-Acc-him

Zayd wants Zaynab to respect himself."


The above data suggest that the verb ?ara:da (want) in Arabic
subcategorizes for a CP complement given the presence of the
complementizer ?an (that).It is thus reasonable to take the
embedded TP to be the governing category for the reflexive anaphors
nafsaha: (herself) and nafsahu (himself). The grammaticality of (14a)
will be accounted for by the fact that nafsaha: (herself) will be A-
bound by Zaynbu within this governing category, thus
satisfying
Binding Condition A. In the same vein, the ill-formedness of (14b)
will be accounted for by the fact the reflexive
nafsahu (himself) will
lack an A-binder within the same embedded TP, in violation of
Binding Condition A.

Consider now the


following examples, where the
embedded subject is an empty pronominal:

(15) a.
yu-ri:du zayd-un [?an [TP2 ya-shtar-ia bayt-an li-nafs-i-hi]]
3MS-want Zayd-Nom thatsubi
3MS-buy-Subj to-self-Gen-him
Zayd wants to buy a house for himself."
b. Ha:wal-a zayd-un [?an [Tr2 ya-rta:H-a qali:lan
tried-3MS Zayd-Nom thatsubi 3MS-relax-Subj a bit

I t has been argued above


that the complementizer ?an must be to the
verbal element for
subjunctive mood assignment. This may now be adjace eted
as meaning that the
TP reinterp
complement of ?an does not project a specifier ion.
poi
relax bit."
"Zayd tried to
a

'vu-ri:du
C. zayd-un |Pan lTpz ya-rta:H-a nafs-u-hu qali:lan|1
3MS-want Zayd-Nom thatsubj 3MS-relax-Subj self-Nom-him a bit
Zayd wants himself to relax a bit."

Given that reflexive anaphors require a local antecedent, the reflexive


nakihi (hinmself) in (15a) must be bound within the same clause, i.e.
TP. Its binder is assumed to be "little" pro, which agrees with nafsihi

(himself) in all o-features, (cf. Ennassiri (2014a). This pronominal

element is in turn controlled by the matrix subject, Zayd, as (15a) is an


construction.
instance of obligatory subject control

With respect to obligatory subject control constructions-e.g.

observed that although the subject of the finite


(15a&b)- it has been
clausal complement is a pure pronominal, i.e. pro, it is obligatorily

'big' PRO
controlled by the matrix subject. Here, little pro resembles

independent theta-role. Second, it


In two respects. First, it has an

an overt DP. Yet, a crucial difference between


cannot alternate with
of conindexation between
pro and PRO lies in the fact that the nature

antecedents is
and their respective
SC tw0 empty pronominals
or
conindexation is required for identification
Crent. For PRO, this
unless PRO is coindexed with a
Tential purposes. In other words,

be identified, in
violation of the principle
Aargument, it cannot
identified. In English, and a
which uires all empty categories to be
coindexation is a
obligatory
other languages for that matter,
O

of
itself a lexical property
conseque
uence of obligatory coreference,
control verbs. In Arabic, however, pro is already fully (p-) identified

by the AGR features anchored onto the verb, (cf. Ennassiri (2014b)

Therefore, its coindexation is not so much forced by the Binding


Theory as by an independent theory of interpretation/construal.

Rosenbaum (1967) proposed a principle he called the Minimal


Distance Principle (MDP) for the controller of PRO in
to account

English. MDP states that the controller of PRO is the DP closest to


it. MDP works well for (15a&b) above and for
object-control
constructions such as (16) below. However, this
principle predicts that
pro in (17) should be controlled by 3amr-an, contrary to fact:

(16) Paqnaë-a zayd-un; Eamr-an; [cP Pan [Tp ya-qlaë-a pr0 ij an al


tadkhi:n-i]l]
persuaded-3MS Zayd-Nom Amr-Acc that
3MS-give-up-Subj
from the-smoking-Gen

Zayd persuaded Amr to give up smoking]


(17) waEad-a zayd-un, Eamr-an; [cp ?an
[TP ya-qla&-a proj/j Ean al-
tadkhi:n-i]
promised-3MS Zayd-Nom Amr-Acc that
from the-smoking-Gen 3MS-give-up-Subj
Zayd promised Amr to give up
smoking
We will return to the
case of
now it is worthwhile to
promise-type verbs below, but for
point out that subjunctive
complements to the
In this context, the Minimal
understood as the subject of Distance Principle would mean that
closest to this clause. the yarta:H-a the element
clause in (15b) is the
element that is
counterpart of lara:da (want) in Romance
Phenomenon, which is languages exhibit the
Obviative

and the
characterized by the fact that
embeddedan
A
matrix subjects must the
reference, (cf. Johnson (1983), Picallo
obligatorily be disjoint in
(1985), Kempchinsky
Dachette (1988), among others). This (1986),
phenomenon is illustrated
hu the following respective
examples from French and
Spanish:
(18) a.Je voudrais que je finisse cet article.
I would like that I
finish-subjunctive this article
"Iwould like to finish this article"

b. Je voudrais qu'il finisse son article.


I would like that he
finish-subjunctive his article
"I would like him to finish his article"

C. Il voudrais qu'il»ik finisse son article.

he would like that he finish-subjunctive his article

"He would like him to finish his article" (cf. Rochette (1988)

a . El maestro quiere que Arturo secalle/ calla.

ne teacher wants that Arturo be quiet-subjunctive/ indicative

"The teacher wants Arturo to be quiet

b. El maestro, quiere que pro«ia se calle.

the teacher wants that he be quiet

The teacher wants him to be quiet


Towards a Contrastive syntax of English and rabic
Principles and Parameters Theory:

in (18) and (19) contrast with the


The French and Spanish examples
and (15a&b), for in the latter
Arabic examples given in (14a)
Obviative Phenomenon seems to be optional. This
examples the
terms of what may be
interesting contrast may be accounted for in

dubbed a tense dependency relation between the matrix and the

subjunctive clauses in Romance languages. In these languages,

subjunctive clauses do not have an independent tense; rather, their


tense is dependent on- i.e. anaphoric to- the matrix clause tense. This
is based on the observation that the tense morphology of the matrix
indicative verb determines the tense morphology of the embedded

subjunctive verb, (cf. Johnson (1984) and Picallo (1984), (1985).


Under this view, subjunctive complement clauses have their temporal
reference established by the tense of the matrix verb. If this is correct,
the ungrammaticality of (18a), (18c) and (19b), where the embedded

subject corefers with the matrix subject, is straightforwardly


accounted for. Being anaphoric in nature, the embedded tense in these
examples must be bound by the matrix tense, where binding, again,
means coindexing with a
c-commanding antecedent. As a result of this
anaphoric/dependency relationship, the binding domain for the
embedded subject extends to the
matrix clause.* This means that the
matrix clause is the local domain
where the embedded
pronominal
subject must be free, but it is not, in
contravention of Binding
Condition B. In Arabic, on the other hand, it is the embedded clause,
154
A way to
achieve
this is to
the embedded TENSE, hence also assume, for example, that the
the embedded
matrix verb govern
(1988) for a similar subject position, (see Rochette
proposal).
M.K. E n u

nd not the matrix one. hich counts as alocal domain for


the
embedd
anominal subject/topic. The latter is
therefore frec in this
domain. and nd the examples are ruled in, as
required. We shall not
mpt at this stage an in-dep analysis of this phenomenon,

ontenting
ourselves wit merely describing it.
Summarising the above empirical observations, there is a sense
.hich it can be said that control structures in Arabic concern

siniations where a DP in the matrix clause participates in the event


matrix verb as well as that denoted by the embedded
denoted by the
verb, by proxy via a phonologically-null pronominal, (cf. Boeckx

2006)). However, unlike PRO in English, which has its own

distr1butional requirements- see Chapter 8- this null


idiosyncratic
control literature the controllee- exhibits
pronominal- designated in
as

characteristics as the phonologically empty


pretty much the same

of null subject sentences in Arabic. It has therefore been


subject
element is an instance of little
concluded that this empty pronominal
instances of pro seem to differ is in terms of pro
pro. Where the two
identification in each instance.

constructions
9.3.2 The structure of object control
verbs which
the syntax of
In this section we shall investigate
consider the
illustration,
control. For the sake of
8 0bligatory
Arabic:
from English and
W I n g respective data

herself
(20) John persuaded Mary to rely on

243
(21) a. Paqnae-a zayd-un fa:Temat-a ?an tu-sallim-a nafs-a-ha:/'nafs-a
hu

persuaded-3MS Zayd-Nom Fatima-Acc thatsubij 3MF-surender


Subj self-Acc-her/ self-Acc-his

Zayd persuaded Fatima to surrender herself/himself.

b. naSah-tu l-saji:n-a ?an ya-nshur-a mudhakkira:t-i-hi

advised-1S the-prisoner-Acc thatsubi 3MS-publish-Subj diaries-


Acc-his

I advised the prisoner to publish his diaries."


The first thing to notice about the matrix verbs in (20) and (21a&b) is
that they select two internal arguments, namely a DP and a CP. In the

LGB account, the first theta-role associated with the


theta-grid of
these verbs goes to the object, whereas the second theta-role goes to

the clausal complement. An important question arises at this stage,


viz. what is the syntactic structure of obligatory object control

constructions in (20) and (2la&b)? A priori, two possibilities present


themselves, (22) and (23), (cf. Ennassiri (2014a)):

(22) V

V DP CP
M.K. EnnaSsiri

V
(23)

V CP

V DP

A renresents the standard View


according to which verbs such as
per
in English and ?qnaea in Arabic are
three-place predicates.
Twa of their arguments are
assigned internally, whereas the third one
is assigned externally. An obvious drawback of (22) is that it violates
the Binary Branching Hypothesis discussed in 2.2 above. Further, the
standard view is incompatible with the more recent assumption that all1

the arguments of a predicate- including the subject- are asigned


internally. On the other hand, (23) is consistent with the Binary
Branching Hypothesis, but it presents problems of a difierent nature.
To see what it is at stake, let us consider first the case of persuade in

English. If (23) is indeed the structure associated with object control


be in a position to c-
Complements in English, the object NP would not
definition of c-
in the infinitival CP, assuming
a
command PRO
controlled PRO is a
Command à la Reinhart (1976). So, if obligatory
to Binding
Condition A,
Pure anaphor in English and is thus subject
Chomsky (1981, 1986a), Manzini (1983)and
NOster (1978, 1984),
if we further adopt
others), and
many
lams (1980), among
in (24),
Distance Principle repeated
Cmbaum's (1970) Minimal
for PRO.
controller
the DP under would not count as a
V

245
(24) Minimal Distance Principle (MDP)
An infinitival complement of a predicate P selects as its
controller the minimal c-commanding noun in the functional
complex of P.

Worse. perhaps, is the fact that under (24), the minimal c-


commanding noun for PRO would be the subject NP. All things being

equal, then, (23) would predict that persuade is a subject control verb,

contrary to fact.

With regard to Arabic examples, it has already been claimed


that the empty category in the canonical subject position of the clausal

complement is a pure pronominal, i.e. pro. We have also adduced

empirical data from Arabic to the effect that this null


pronominal
element does not require a c-commanding antecedent for
identification. Could (23), then, represent the structure of obligatory
object control verbs such as ?aqnaea (persuade) or naSaHa (advise) in
Arabic? Now, if (24) is a
property of UG, the answer will have to be
negative, for although pro does not require a
c-commanding
antecedent for identificational
purposes, the c-command relation is
still needed for construal
purposes. In other words, if there were no
constraints on what must count as a
controller for pro in Arabic,
DP in the matrix clause in any
(21a&b) would qualify as a controller for
the embedded pro, contrary to fact.

Recall that in section 2.5


above, double object constructions
were
analysed in terms of the
VP-shell hypothesis. This
be extended below to
analysis will
object control constructions as as well. In
nples such as
articular, examples (20) ana
as (20) d (2ia)
(21a) above will be
have the llowing (partial underlying form: argued to
(25) VP

v
DPsubject

John V VP

Zayd-un
DP V

Mary CP

fa:Temat-a
persuade PRO to rely..
Paqnae-a Pan tusallim-a pro...
The S-structure associated with (20) and (21a) is (26), which results
rom NP-movement in English and v-movement in Arabic:
(26) TP

Spec

John VP

Paqnae-a; DPsubject V

VP
Wohn
Zayd-un
DP V

Mary V CP

ifa:Temat-a
persuaded PRO to rely..
Pan tu-sallim-a pro...

The above structure


satisfies the MDP, thus making the right
prediction with regard to the controller of PRO and
pro.
However, this analysis comes short of
control verbs such
accounting for subjeC
as
(27a&b). To see why, consider the assumea
structure for these
sentences:
He promised his family to send enough money every month
7 )
a. th.
b. wa3ad-a zayd-un zaynab-a ?an ya-HDur-a l-Hafl-a

Z-Nom thatsubj 3MS-attend-


the-party-Acc
p r o m i s e d - 3 M S

7avd promised Zaynab to attend the party."

TP
(28)

Spec T

vP
He

waeada; DPsubject V

VP
tHe
zayd-un

DP2 V

his family V CP

zaynab-a
PRO to send
promised
Panya-HDur-a pro..

249
In (28), the potential (and in fact, the actual) controllers of PROb.and

pro, do not minimally c-command


them. In
controller, he,
English, the controller
originates in [Spec, vP]. but moves to the spccifier of TP far Case-

theoretic and EPP reasons. But even prior to movement, he does not

minimally c-command PRO, as there is a closer, more local (potentiolh


controller for PRO sitting in [Spec, VP]J. So here, too, MDP would

predict object control akin to (20) above, contrary to fact. The sam
thing may be observed for Arabic, for the actual controller, i.e
zavdun, fails to minimally c-command pro. Therefore, object control
may be erroneously predicted.

To render the MDP capable of accommodating subject control


constructions, Hornstein (1999:34) advances an analysis in terms
of
which promise takes an indirect
object introduced by the preposition
to. However, this
preposition "becomes null (perhaps by
into the verb) in the course of incorporating
the derivation." n
fact, promise-type
verbs do indeed with
occur an overt
the following examples:
preposition. This is illustrated by

(29) a. He promised to her that he would never leave


her.
b. She vowed to
him that she would protect his children.
c. She vowed to him
to protect his children.
With this
proviso, the alternative structure
thus be (30) below: associated with (27a) wil
M.K. EnnasSSir

(30)
TP

Spec

vP
He

DPsubjectv

tHe VP

promised PP V

P DP V CP

his hisfamily t PROto send..

The new thing about (30) is that the blocking DP, i.e. his family, no

Onger c-commands PRO. The first branching node dominating this


DP is PP, and, obviously, PP does not dominate PRO. Consequently,
C DP can now make the right prediction with regard to subject

should be pointed
control structures like (27a) in English. Perhaps it

251
out at this stage that the Case feature assigned by the verb promis
to
the DP his family is mediated by the null P. This is not a d

stipulation given that some e-roles have been assumed to be assigned

indirectly to arguments. A case in point is (31) below, where the


e
assignment of the second internal O-role has been assumed to he

mediated by the preposition on:

(31) He put [er the book] [e2on the shelf]

A number of other languages amenable to this


are
analysis. In
French, for example, the verb promettre subcategorises for a PP and
CP complements:

(32) Il a promis [pp à sa femme] de ne plus jamais revenir à une


chose pareille.

In Arabic, too, the verb waeada (promise) has another synonym which
takes two complements, a PP and a CP:

(33) taeahhad-a zayd-un li-zaynab-a ?an ya-HDur-a l-Hafl-a


vowed-3MS Z-Nom to-Zaynab
thatsun 3MS-come-Subj the-
party-Acc

"Zayd vowed to Zaynab to come to the


party."
For the sake of
generality, it is plausible to argue that waeada n
Arabic patterns with promise in
English. More specifically, the S-
structure associated
with, say (27b), is identical to that of
in (30), modulo, of course, (27a) given
subject movement to [Spec, TPJ:
(34)
( 3 4 )
TP

VP

wa3ad-a DP subject

Zaud-un VP

t'i PP V'

P DP V CP

zaynab-a ti Pan ya-HDur-a proj .

Obviously, the above discussion supports the wide-spread idea in

generative linguistics that natural languages are the same


underlyingly. Needless to say, of course, that the analysis presented
above is by no means definitive, as "the theory of control involves a

mumber of different factors: structural configurations, intrinsic

253
Principles and Parameters Theory: Towards a Contrastive Syntax ofEnglish and Arabic

properties of verbs, other semantic and pragmatic considerations,"

(Chomsky, 1981, pp. 78-9).


E X E R C I S E S

Exercise 9.1: Explain the contrast in (a-c) below:

seems
[John likes linguistics]
(a) It
to like linguistics]
b) It seems [PRO
(c) John wants [Mary to take a course in Linguistics]

(d)John; wants [PRO, to take a course in linguistics

(6Tahn: pleaded with Mary; [PRO%; to take a course in linguistics

Exercise 9.2: Identify the controller of pro in the embedded clauses in


(a) and (b) below. What does that tell you about the syntax of 3araD-a
in Arabic?

(a) 3araD-a zayd-un 3ala: zaynab-a [cr 2an yu-Sa:Hib-a-ha: ?ila:


ba:b-i
proposed-3MS Zayd-Nom on Zaynab-Gen thatsubj 3MS-accompany-
Subj-her to the-gate-Gen
Zayd proposed to Zaynab to accompany her to the gate.

tu-sa:fir-a ma3a-hu]
(6) 3araD-a zayd-un 3ala: zaynab-a [cp ?an

proposed-3MS Zayd-Nom on Zaynab-Gen thatsubi 3FS-travel-Subj


with-him
him."
Layd proposed to Zaynab to travel with

5
BIBLIOGRAPHY

Ackerman, F. (1987) *Pronominal incorporation, Approaches to


Hungarian, 2: 213-261.

Ackerman, F. and Ph. Lesourd (1997) *The lexical representation of


phrasal predicates, in A. Alsina, J. Bresnan and P. Sells (eds
Complex Predicates. Stanford: CSLI Publications, pp. 67-106.

Ackerman, F. And G. Webelhuth (1998) A Theory of Phrasal


Predicates. Stanford: CSLI Publications.

Adger, D. and G. Ramchand (2003) Predication and equation.


Linguistic Inquiry, 34: 325-359.

Akmajian, A., R. A. Demers, A. K. Farmer and R. M. Harnish (2001)


Linguistics: An Introduction to Language and Communication.
Cambridge MA: MIT Press.

Alexiadou, A. and E.
Anagnostopoulou (1998) Parametrizing AGR:
word order, V-movement and
EPP-checking?, Natural
Language and Linguistic Inquiry, 16: 491-539.
AI Sharifi, B. and L. Sadler (2009) 'The adjectival construct in
in Miriam Butt & Arabic
Tracy Holloway King (eds.), Proceedings of
the LFG09
conference, University of Cambridge, pp. 26-43.
Al-Shorafat, M. O. (2013) 'A Phase -Based Account of
in Standard Arabic', Wh-Questions
Linguistics and Literature Studies 1(4):
179-190.
Aoun, J. (1979) On Government,
Placement', ms., MIT. Case-marking and Clitic

Aoun, J. and D.
Sportiche (1983) On the formal theory of
government', The Linguistic Review, 2: 211-236.
Aoun, J. and Y.
Audrey Li
LF?', Linguistic Inquiry,(1993) 'Wh-Elements
24: 199-238.
in Situ: Syntax or
(1998)
PF and
E.
and E.
J . Movement',Lins
d O u n , .
Benmamoun
B enr
'Minimality,
Reconstruction and
ment', Linguistic Inquiry, 29, 569-591.
little y'
*On little " , MIT
(1999) Working Papers in
Irad.
M.
3 3 : 1 - 2 5 Linguistics,
o f . M. Word Formation
0TA
(197 in
Generative Grammar
MA:MIT Press.
'ambridge

Aronof, M. & J. Rees-Miller (eds.) (2003) The Handbook of


Linguistics. Oxford: Blackwell.

Authier, J . M (1992) 'Iterated Cps


. and Embedded Topicalization',
Linguistic lnquiry, 23: 329-336.
Baker, M. (1988) Incorporation: A Theory of Grammatical Function
Changing. Chicago: Chicago University Press.
Baker, M. (1997) "Thematic roles and syntactic structure', in L.
Haegeman (ed.), Elemnents of Grammar. Dordrecht: Kluwer, pp.
73-137.

Baker. M. (2008) The Macroparameter in a Microparametrie World',


in Biberauer, Theresa (ed.), The Limits of Syntactic Variation,
pp. 351-373.

Baker, M. (2015) Case: Its Principles and Its Parameters. Cambridge


University Press.

'Passive Arguments
Baker, M., K. Johnson and I. Roberts (1989)
Raised', Linguistic Inquiry, 20: 219-251.
Relativized minimality and
Baker, M. and H. Kenneth (1990) 12:289-297.
pronoun incorporation', Linguistic Inquiry,
Evidence from
DKIT, M. J. (2011) Against the Split-CP
hypothesis:
E. and H. Ouali (eds.), Perspectives
rdqi Arabic', in Broselow, Benjamin Publishing
On a b i c XXII-XXII, John
Linguistics
Company Handbook of
The Handbook of
(2003)
Baltin, M and C. Collins (eds.) Blackwell:
Oxford.

Contempoi Syntactic Theory.


257
Principles and Parameters Theory: Towards a Contrastive Syntax of Enalich
lish and Arabic

Barie, M. (2007) Control and WH-infinitivals, in Dau. &


in Davies
Dubinsky (eds.), pp. 263-279. Linguistic Theory n 14, pp. 269-
304.

Bars, A. and H. Lasnik (1986) *A Note on Anaphora and n


Objects', Linguistic Inquiry, 7: 347-354.
and Double

Belletti, A. (2004) (ed.) Structures and Beyond. The Cartogranl.


Syntactic Structures, Volume 3,. New York: Oxford Univereof
Press versity
Belletti, B. and L. Rizzi (eds.) (1996) Parameters and
Heads. Oxford: Oxford University Press. Functional
nal

Benmamoun, E.
(1992) Functional and Inflectional
Problems of Projection, Morpholoo.
Representation and
diss., University of Southern California, LA. Derivation, PhD
Benmamoun, E. (2000) The feature structure
of functional categories:
a
comparative study of Arabic dialects. Oxford: Oxford
University Press.
Bittner, M. and K. Hale
(1996) 'The Structural Determination of
and
Agreement', Linguistic Inquiry, 27: 1-68. Case
Bobaljik, J.D. (1995) Morphosyntax: the
PhD diss., MIT. syntax of verbal inflection,
Boeckx, C. (2006) Linguistic Minimalism:
Methods, and Aims. Oxford: Ox Origins, Concepts,
ford University Press.
Boeckx, C., N. Hornstein and J. Nunes
Cambridge: (2010) Control as Movement.
Cambridge University Press.
Borer, H.(1984) Parametric Syntar: Case
Romance Studies in Semitic ana
Languages. Dordrecht: Foris Publications.
Boakovi, Z. (1996), *Selection
and the
complements, Natural Languagecategorical status of infintival
269-304. and Linguistic Theory
78) 'A Realistic
Bresnan, J.
Halle. Joan
Transformational Grammar', in Morris
Rresnan. and George A.
Hal d Miller (cds.),
Theorr and.
Psvchological Reality, 1 59. Linguistic
Cambridge, MA: MIT
Press.
Bresnan.. 1982) Control and
(1982)

Inquirn: 13: 343-434. Complementation'. Linguistic


986) Italian syntax: A
Burzio.
L.
Dordrecht: Reidel. Government-Binding Approach.
Cardinaletti., (1997)
A. Subjects and clause
structure',
L in
Haegeman (ed.), The New Comparative Syntax. London:
Longman, pp. 33-64.

Cardinaletti, A. and M. T. Guasti (1993) 'Negation in epistemic small


clauses, Probus, 5: 39-61.

Chekili, F. (2004) The position of the Postverbal Subject and


Agreement Asymmetries in Arabie', PhiN 27:35-46.

Chomsky, N. (1957) Syntactic Structures. The Hague: Mouton.

Chomsky, N. (1977) On Wh-Movement', in P. Culicover, T. Wasow,


and A. Akmajian (eds.), Formal Syntax. New York: Academic
Press.

N. (1981) Lectures on Government and Binding.


Chomsky,
Dordrecht: Foris Publications.

Theory
Chomsky, N. (1982) Some Concepts and Consequences of the
Press.
Government and Binding. Cambridge MA: MIT
of
MIT Press.
N. (1986a) Barriers. Cambridge, MA:
Chomsky,
lis Nature, Orign,
Cnomsky, N. (1986b) Knowledge of Language:
and Use. New York: Praeger.
Knowledge: The
and Problems of
N. (1988) Language Press.
OSky, Cambridge, MA: MIT
Managua Lectures.
Principles and Parameters Theory: Towards a Contrastive syntax of English
ish and
and Arabic

Chomsky, N. (1991) 'Some notes on Economy of Derivation


and
Representations', in R. Freidin (ed.), Principles and Parametore
in Comparative Grammar. Cambridge, MA: MIT Press, 417.
454
Chomsky, N. (1994) 'Bare phrase structure', MlT occasional papers
in linguistics 5. Department of Linguistics and Philosophy, MT
Chomsky, N. (1995) The Minimalist Program. Cambridge, MA: MIT
Press.

Chomsky, N.
(1998a) 'Minimalist Inquiries: The Framework"',
Occasional Papers in MIT
MA.
Linguistics, 15, MITWPL, Cambridge.

Chomsky, N. (1998b) *Some observations on


grammar, in P. Barbosa, D. Fox, P. economyM.in generative
and D. Pesetsky
(eds.),
Hagstrom,
Is the Best Good McGinnis,
and Competition in
Syntax.
Enough? Optimality
115-27. Cambridge, MA: MIT Press, pp.

Chomsky, N. (1999) Derivation by Phase', in M.


Kenneth Hale: A life in Kenstowicz (ed.),
pp. 1-52. language. Cambridge, MA: MIT Press,

Chomsky, N. (2000) 'Minimalist


Martin, D. Michaels and inquiries:
J.
the
framework', in R.
essays minimalist syntax Uriagereka
on
in honour
(eds.), Step by step:
Cambridge, MA: MIT Press, pp. 89-155. of Howard Lasnik.
Chomsky, N. and H. Lasnik
Inquiry, 8: 425-504. (1977) 'Filters and Control',
Linguistic
Chomsky, N., and H. Lasnik
Theory', in J. Jacobs, A. (1991) Principles and Parameters
(eds.), Syntax: An von
Stechow, and T. Vennemann
Research. Berlin, International Handbook of
Walter de
1995. Gruyter (reprinted inContemporary
N. Chomsky
M.R. B n n a s s u I

Cinque, G. (2002) (ed.) Functional Structure


Cartography of Syntactic Structures, Volume
in DP
and IP. The
Oxford University Press. 1. New
York:
Comrie, B. (1981) Language Universals and
Syntax and morphology. linguistic typology:
Chicago Press. Chicago linois:
University of
licover, P. W.
Culicover, (199)
to Svntactic Theory, Principles and
Parameters.
New York: Oxford An Introduction
University Press.
ulicover, P. and R. Jackendoff (2005)
Oxford University Press. Simpler Syntax. Oxford:
navies, W. D. and S.
Dubinsky (2004) 7he
Grammar of Raising and
Control: A Course in
Syntactic Argumentation. Blackwell.
Davis, L. (1986) *"Remarks on the 8-Criterion and
Inquiry, 17: 564-568. Case', Linguistic
Davis, L. (1987) Remarks on Government and
Linguistic Inquiry, 18: 311-321. Proper Government',

De Vincenzi, M. (1991) Syntactic


parsing strategies in Italian.
Dordrecht: Kluwer Academic Publishers.

Enç, M. (1991) 'The Semantics of specificity', Natural Language and


Linguistic Theory 22: 1-25.
Ennaji, M.. (1995) Pronouns vs. clitics', Revue de la Faculté des
Letres et des Sciences Humaines, Fes.

Ennassiri, M. K. (2004) 'Differences between acquiring an LI and


leaming an L2: Some implications for TEF/SL', Ofshoot V, 2:
3-38.

Cnnassiri, M. K. (2014a) The Syntax of Complement Clauses in


Arabic. Tetouan: Al-Khalij Al-Arabi.
E
nassiri, M. K. (2014b) Issues in Arabic Syntax. Tetouan: Al-Khali

Al-Arabi.
Principles and Parameters Theory: Towards a Contrastive Syntax Of English and Arabie
abic

Fassi Fehri, A. (1993) Issues in the Structure of Arabic Clauses and


Words. Dordrecht: Kluwer.

Fassi Fehri, A. (1999) 'Arabic Modifying Adjectives and DP


Structures', Studia Linguistica, 53(2): 105-154.
Freidin, R. (1983) X-bar Theory and the analysis of English
infinitivals', Linguistic Inquiry, 14: 713-722.

Freidin, R. (ed.) (1991) Principles and Parameters in Comparative


Grammar. Cambridge, MA: MIT Press.

Freidin, R. (1992) Foundations of Generative Syntax. Cambridge,


MA: MIT Press.

Fukui, N. (2006) Theoretical Comparative Syntax: Studies in


in
macroparameters. London/New York: Routledge.

Greenbaum, S. & R. Quirk (1990) A Student's Grammar of the


English Language. Longman.
Greenberg, J. H. (1966) Universals of Language. Cambridge, MA:
MIT Press.

Haegeman, L.
(1982) INFL, COMP and nominative case assignment
in Flemish infinitivals', in P. Muysken and H.
Riemsdijk (eds.),
Feature projections. Dordrecht: Foris Publications, pp. 123-136.
Haegeman, L. (1991) Introduction to government and binding theory.
Oxford: Basil Blackwell.

Haegeman, L. (ed.) (1997) The New Comparative Syntax. London:


Longman.
Hendrick, R. (ed.) (2003) Minimalist Syntax. Oxford:
Blackwel.
Higginbotham, J. (1980a)
Anaphora and GB:
remarks, in J. T. Jensen
some preliminary
(ed.), Cahiers Linquistiques d'Otawa,
9:223-236.
M K .E n a S S I N 7

'S and X
N
Hornstein, N (1977) convention', Linguisticc
3:137-176. Analysis,
N. 1995) Logical Form. From GB to
ornstein,
Blackwell.
Minimalism. Oxford:
NN.
Hornstein, . (1999) 'Movement and Control', Linguistic Inquiry.
30:69-96.

nstein. N. (2001) Move! A Minimalist Approah to Construal,


Oxford: Blakwell.

Uarmstein. N. (2003) *On Control', in R. Hendrix (ed.), Minimalist


Syntax. Oxford: Blackwell, 6-81.

Hornstein, N. and D. Lightfoot (1991) On the nature of lexical


govemment, in Freidin (ed.), pp.365-391.

Hornstein, N., J. Nunes & K.K. Grohmann (2005) Understanding


Minimalism. Cambridge: Cambridge University Press.

Homstein, N. and M. Polinsky (2010) (eds.) Movement Theory of


Control. John Benjamins' Publishing Company.

Huang, C.-T. J. (1983) 'A note on the binding theory', Linguistic


Inquiry, 14: 554-561

Huang, C.-T. J. (1984) On the distribution and reference of empty


pronouns', Linguistic Inquiry, 15:531-574.

Jackendoff, R. (1997) The Architecture of the Language Faculty.


Cambridge, MA: MIT Press.
in
acobs, A. K. and P.S. Rosenbaum (1970) (eds.) Readings
Ginn.
Transformational Grammar. Waltham:
parameter.
acggli, O. and K. Safir (eds.) (1989) The mull subject
Dordrecht: Kluwer.
and the roles or
Onas, D. and J. Bobaliik (1996) 'Subject positions
TP, Linguistic Inquiry, 27: 195-236.
Contrastive Syntax ofEnglish and
Principles and Parameters Theory:
Towards a Arabic

'ECP extensions', Linguistic Inquiry, 12:93-133.


Kayne, R. (1981)

Kayne, R. (1984) Connectedness and Binary branching. Dordrecht


Foris Publications.

Kayne, R. (1994)The antisymmetry of syntax. Cambridge, MA: MIT


Press.

Koopman, H. (1983) The Syntax of Verb. Dordrecht: Foris


Publications.

Koopman, H. and D. Sportiche (1991) *The Position of Subjects',


Lingua, 85: 211-258.

Kremers, J. (2005) Adjectival Constructs in Arabic', Linguistische


berichte, 203:33 1-348.

Kurniawan, E. (2013) Sundanese complementation. PhD diss.,


diss.,
University of lowa.
Larson, R. (1988) On Double Object Construction', Linguistic
Inquiry, 19: 335-391.

Lasnik, H. (2001) 'A note on the EPP*, Linguistic Inquiry, 32: 356-
362.

Lasnik, H. (2003) 'On the Extended Projection Principle', Studies in


Modern Grammar, 31:1-23.

Lasnik, H. and M. Saito (1984) On the Nature of


Government', Linguistic Inquiry, 15, 235-289 Proper
Lasnik 1990). (reprinted in
Lasnik, H. and S. Saito
(1992) Move Alpha: Conditions on its
Application and Output.
Cambridge, MA: MIT Press.
Lasnik, H. and J.
Uriagareka (1988) A Course in GB
Cambridge, MA: MIT Press. Syntax.
Lightfoot, D. (1991) How to Set
Press. Parameters. Cambridge, MA: MII
M . . ENIM

rdi, G. (1994) Reference and


obard.
movement iin Syntax and
L o n g o

Proper Names: A
LinguisticTheory
Logical Form', of N-
609-665.

Inquiry, 25
(1996) A minimalist theory of PRO
R.
Martin,

University of Connecticut, Storrs.


and control.
PhD diss.,
1985) Logical Form.
May,
R.
Cambridge, MA: MIT Press.
(2009) Introducing English
C.
Meyer,
Linguistics. Cambridge: CUP.
Mohammad, A. M. (1989) The
Sentential Structure
diss., University Southern
of of Arabic, PhD
California, LA.
Mohammad, A. M.
(2000). Word Order,
Pronominalization in Standard and Agreement, and
and
Benjamins, Philadelphia. Palestinian Arabic. John

Newson, M. (2006) Basic English Syntax with Exercises.


HEFOP. Budapest:
Ouhalla, (1988) The Syntax of Head Movement: A Study
J.
PhD diss., University College, London. of Berber,

Ouhalla, J. (1991) Functional Categories and Parametric Variation.


London: Routledge.

Ouhalla, J. (1994) Verb Movement and Word Order in Arabic', in


Hornstein, N. and D. Lightfoot (eds.), Verb Movement.
Cambridge University Press.
Ouhalla, J. (1996) The Construct State in Berber'. Studies in
Afroasiatic Grammar, ed. By Jacqueline Lecarme, Jean
Lowenstamm and Uri Shlonsky, 278-301. The Hague: Holland
Academic Graphics.
J. (1999) Transformational Grammar: From Principles and
Ouhalla,
Parameters to Minimalism. London: Arnold.
and Hebrew
nalla, J. and U. Shlonsky (2002) Themes in Arabic
yntax. Kluwer Academic Publishers.
Pesetsky, D. (1987) Wh-in-Situ: Movement and Unselective
Binding'. in Eric J. Reuland & Alice G. B. ter Meulen (Eds.),
The Representation of (Indefiniteness, Cambridge, MA: MIT
Press.

Pesestky, D. (2000) Phrasal Movement and its Kin. Cambridge, MA:


MIT Press.

Pesetsky, D. and E. Torrego (2004) 'Tense, Case, and the Nature of


Syntactie Categories' in Gueron, J. and J. Lecarme (eds.) The
Svntax of Time. Cambridge, MA: MIT Press.
Pierce, A. (1992) Language Acquisition and syntactic theory: A
comparative analysis of French and English child language.
Dordrecht: Kluwer.

Pinker, S. (1994) The Language Instinct. New York, NY:


Perennial Modern Classics. Harper

Pollock, J. Y. (1989) 'Verb


movement, Universal Grammar, and the
structure of IP', Linguistic 20: 365-424.
Inquiry,
Quirk, R., S. Greenbaum, G.
Leech, and J. Svartvik (1985) A
Comprehensive Grammar of the English
Language. London:
Longman.
Radford, A. (1981) Transformational Symtax.
University Press. Cambridge: Cambridge
Radford, A. (1988) Transformational Grammar. A First Course.
Cambridge: Cambridge University Press.
adford, A. (2004) English Syntax: An Introduction. Cambridge:
Cambridge University Press.
adó, J.
(1998) Processing Hungarian: The Role
in Sentence of Topic and Focus
Processing, PhD diss., University of Massachusetts,
Amherst.

emsdijk, H. C. van and E. S. Williams (1986) Introduction


theory of grammar. to the
Cambridge, MA: MIT Press.
'NS orders in modern Hebrew', in
(1987)
Plunkett, B. (eds.), s). Proceedings of the North
East
McDonough, J..
i r .

vol. 17. pp. 521 5337.


Linguistic
Socien,

.1991) wo functional
categories in noun phrases', in S.
(ed.). Perspectives on Phrase Structure: Heads
Rothsto and
Licensing. Academic Press, California, pp. 7-62.
Rnen. M.L. 980) On
(1980) ft-Dislocation and
Topicalization in
Linguistic Inquiry 11. 2: 363-393.
Spani.

(1990) Relativized Minimality. Cambridge, MA: MIT


Rizzi. L. (1990)

Press.

Rizzi. L (1997) The fine structure of the left periphery',in


L. (1997)
Haegeman, L. (ed.), Elements of Grammar. Dordrecht: Kluwer,
pp. 281-337.

Rzi. L. (2004) (ed.) The Structure of IP and CP. The Cartography of


Syntactic Structures. Volume 2. Oxford: Oxford University
Press.

Rizi, L. (2014) 'Syntactic Cartography and the Syntacticisation


of
Scope-Discourse Semantics', in Reboul, A. (ed.) (2014) Mind,
Values, and Metaphysics, Switzerland: Springer International
Publishing, pp. 517-533.

Edward Arnold.
Roberts, I. (1986) Comparative Syntax. London:

Rosenbaum, P. (1967) The grammar of English predicate complement


constructions. Cambridge, MA: MIT Press.

deletion in English
Kosenbaum, P. (1970) 'A principle governing K. and P. S.
A.
sentential complementation', in Jacobs,
Grammar, pp.
Rosenbaum (eds.) Readings in Transformational
20-29
Doctoral
OSS,J. R. (1967) Constraints on variables in syntax.
dissertation, MIT, Cambridge.
New York:
Sa Chains. Cambridge and
( 1 9 8 5 ) Syntactic
Cambridge University Press.

267
C o n t r a s i u v

Towards a
Parameters Theory:
Principles and

Order in Hebrew and


and Word
Syntax. Oxford
Structure
Clause
Shlonsky, U. (1997)
Semitic
in Comparative
Arabic: An Essay
New York.
University Press,
and inalienable constructions
constructs
Siloni, T. (2002)
*Adjectival Themes in Arabic and
Ouhalla,J. and U. Shlonsky (eds.),
in Holland: Kluwer.
Hebrew Syntar.
Dordrecht,

Theory: An Introduction to Word


Morphological
Spencer, A. (1991) Granmar. Oxford: Basil
Blackwell.
Structure in Generative

'Topicalisation in English and the Trochaic


Speyer, (2005)
A.
in Linguistics, 10.2:
Requirement"', U. PennWorking Papers
243-256.

Sproat, (1984) INFL and the Configurationality of VSO


R.
Annual meeting of the
Languages', Proceedings of the 14"
North East Linguistics Society, pp. 418-431.

Stowell, T. (1982) "The tense of infinitives', Linguistic Inquiry,


13:561-569.

Stowell, T. (1983) 'Subjects across categories', The Linguistic Review,


2:285-312.

Tremblay, A. (2006) Theoretical and methodological perspectives on


the use of grammaticality judgement tasks in linguistic theory,
Second Language Studies 24 (1) 129-167.
Tsimpli, I.M. and J. Ouhalla (1990) 'Functional categories, UG and
Modularity'. Ms., University College and Queen Mary &
Westfield College, London.
Van Valin, R. (2001) An Introduction to Syntax. Cambridge:
Cambridge University Press.
Wahba, W. A. -B. (1991) LF Movement in Iraqi Arabic', in C.-T.J.
Huang and R. May (eds.),
Structure. Dordrecht: Logical Structure and Linguistic
Kluwer, pp. 253-276.
A

G.(995a)
(ed.) Government and
Goy
Binding Theory and the
lhath. Program. ford: Blackwell.
finimalis

K-bar theory and Case theory', in G.


(19956)
G.

i Webelhuth (ed.) Government and Binding Theory and the


lhuth.

Minimalis. Program. Oxford: Blackwell, pp. 15-95.

Hener.K. and
d M.R. Manzini (1987) 'Parameters and Learnability in
Binding', in Roeper, T. and E. Williams (eds.) Parameter
Setting. Dordrecht: Reidel.

Hoolford E. (1991) VP-Internal Subjects in VSO and


Nonconfigurational Languages, Linguistic Inquiry, 22: 503-
540.
Woolford, E. (2006) Lexical case, Inherent case, and Argument
Structure', Linguistic lnquiry 37: 111-130.

You might also like