PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/ClRC/DaIeDuo.p65-p.15

BARE NOMINALS, FOCUS STRUCTURE, AND REFERENCE
IN GERMANIC, ROMANCE AND SEMITIC
By

Murad Salem

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Linguistics and Germanic, Slavic,
Asian and African Languages

2003

ABSTRACT
BARE NOMINALS, FOCUS STRUCTURE, AND REFERENCE
IN GERMANIC, ROMANCE AND SEMITIC
By
Murad Salem

This thesis is concerned with providing a uniﬁed approach to bare nominals
crosslinguistically by drawing on the properties of focus structure and word order facts.
It essentially seeks to assimilate the seemingly disparate behavior of bare nominals in
Palestinian Arabic (PA) and Spanish, on the one hand, and English, on the other, by
conceiving of such behavior as stemming from deeper distinctions between these two
language groups in the properties of focus and the differences in word order. The picture
that emerges is highly restrictive and thus desirably minimizes crosslinguistic variation
among these languages.

The proposed analysis argues in the ﬁrst place that there is no asymmetry in the
distribution of bare nominals in PA or Spanish in terms of structural positions. Relying
on the behavior of determinerless nominals in the Semitic Construct State, I advance an
analysis of bare nominals in the general case that views these nominals as being
focalized, or non-topical. Since bare nominals in both PA and Spanish are subject only
to an existential interpretation, it seems natural to predict that this analysis would carry
over to existential bare nominals in English. This should in fact be the null hypothesis.
The present dissertation argues for the accuracy of this prediction.

Once I have established that existential bare nominals are always focused, I set out

to explain the differences observed to hold between PA and Spanish, on the one hand,

and English, on the other, in the distribution of bare nominals as emanating from deeper
distinctions between these two language groups in the properties of focus and word
order facts. Word order in English is generally rigid, which rules out the possibility of
(de)focalizing constituents through movement. This language, therefore, has recourse to
marked focus, i.e. non-contrastively focusing a sentence-internal constituent. By
contrast, PA and Spanish enjoy a ﬂexible word order system which makes prosodic
movement an option at their disposal; marked focus is accordingly precluded. A bare
nominal in PA and Spanish cannot be non-contrastively focused in situ, but would have
to be placed in the lowest position in the syntactic tree in terms of c-command, where
Nuclear Stress is assigned in these two languages. These basic differences between
English, on the one hand, and PA and Spanish, on the other, translate into differences in
the distribution of bare nominals.

The proposed analysis also seeks to explain why English expresses genericity via
bare nominals whereas PA and Spanish lack that option. I argue that this difference can
be pursued along two tacks. First, I make the assumption that generic operators in PA
and Spanish cannot bind nominals in a DP with a null or empty head, whereas English
generic operators can. Second, due to their focal status, bare nominals in PA and
Spanish, just like their English counterparts, cannot be mapped into a restrictive clause
of a generic operator. These nominals are always mapped into the nuclear scope. It is
then predicted that existential bare nominals in PA and Spanish are never bound by
generic operators and these languages make use of the deﬁnite article to express

genericity.

Cepyright by
Murad Salem
2003

ACKNOWLEDGMENT

My deep appreciation goes to my supervisor Professor Alan Munn for being a
source of intellectual inspiration for me throughout my stay at Michigan State
University. I have worked very closely with Professor Munn, both as a student and a
teaching assistant, and I have always been inspired by our discussions and the
intellectual exchanges we have had throughout the years. I wish to thank him for his
continued support and encouragement during my stay at Michigan State University. I am
both proud and honored to have worked with him.

I would also like to thank Professor Cristina Schmitt for her input and comments on
issues related to my work. I wish also to thank her for the wonderful work she has done
as a member of my supervising committee. Her encouragement and positive attitude
have had a great inﬂuence on me.

My thanks also go to Professors Yin-Hwei Lin and Deogratias Ngonyani for their
help and support and their hard work on my supervising committee. I would also like to
thank Professor Patricia Lunn for her work as an external examiner.

Last, but not least, I would like to express my utmost gratitude and appreciation for
the loving support my wife Linda has showed me throughout the years, especially
during my work on this dissertation. I do not know if I would have been able to

accomplish any of this if it had not been for her encouragement and understanding.

Table of Contents

Chapter 1
Introduction
1.0 Overview .............................................................................................. 1
1.1 Bare Nominals: The Basic Problem ..................................................... 2
1.2 The Proposal ......................................................................................... 5
1.3 Organization ....................................................................................... 14
Chapter 2
The Internal Structure of the DP in Palestinian Arabic
2.0 Introduction .......................................................................................... 19
2.1 The Construct State: The Basic Properties ........................................... 20
2.2 Plurals in Construct State Constructions .............................................. 24
2.3 Strict Adjacency in the Construct State ................................................ 26
2.4 The Construct State Analyzed .............................................................. 31
2.4.1 Fassi Fehri (1999) ........................................................................ 31
2.4.2 Longobardi (1996) ....................................................................... 39
2.5 Deriving the Semitic Construct State .................................................... 41
2.5.1 The Status of D in Semitic ........................................................... 44
2.5.2 A Working Analysis of the Construct State in Arabic ................. 48
2.5.3 Bare Nominals in Non-CS Constructions .................................... 54
2.6 Bare Nominals in PA and English ........................................................ 63
2.7 English Bare Nominals and (In)deﬁniteness Feature
Checking .............................................................................................. 64
2.8 Conclusion ............................................................................................ 71
Chapter 3
Bare Nominals in PA: Distribution and Interpretation
3.0 Introduction .......................................................................................... 72
3.1 The Basic Facts .................................................................................... 76
3.2 BNs and Government ........................................................................... 78
3.2.1 The Lexical Government Approach ............................................ 80
3.2.2 Problems with the Lexical Government Approach .................... 83
3.2.2.1 The Nature of the Lexical Government .................................... 83
3.2.2.2 Modiﬁcation and Coordination of BNs .................................... 85
3.2.2.3 Focused Preverbal BNs ............................................................ 88
3.2.2.4 The Empty D as a Default Existential Operator ....................... 89
3.3 BNs and the Mapping Hypothesis ................................................. 91
vi

3.3.1 NPs or DPs .................................................................................. 91

3.3.2 NP-Movement Vs. DP-Movement .............................................. 94
3.3.3 The Mapping Hypothesis: Assumptions and Problems ............... 95
3.3.3.1 Heim's (1982) Conception of Indeﬁnites ................................. 95
3.3.3.2 Problems with the Mapping Hypothesis. ............................... 101
3.3.3.2.1 Reinhart (1995) ................................................................... 101
3.3.3.2.2 Pinto (1997) ........................................................................ 103
3.3.4 An Evaluation of Casielles-Suarez's (1997) Analysis .............. 107
3.4 A Uniform Account of Bare Nominals ........................................ 109
3.4.1 Bare Nominals and Informational Focus .................................. 111
3.4.1.1 Bare Nominals and Presentationalism ................................... 111
3.4.1.2 Nuclear Stress and Focus ....................................................... 114
3.4.1.4 Bare Nominals are Always Focal ......................................... 117
3.4.1.5 Singular Bare nominals ......................................................... 127
3.5 Modiﬁcation and Coordination of Bare Nominals ..................... 129
3.5.1 Adjectival Modiﬁcation ............................................................ 132
3 .5.1 .1 Intersective/Non-Intersective Adjectives ............................... 1 32
3.5.1.2 Ethnic Adjectives ................................................................... 135
3.5.1.3 Cardinal Adjectives ............................................................... 136
3.5.1.4 Relativization and Bare Nominals ......................................... 138
3.5.2 Coordination and the Licensing of Bare Nominas ................... 140
3.6 The Interpretation of Bare Nominals .......................................... 147
3.6.1 Stage-Level Vs. Individual-Level & Generic
Vs. Existential ........................................................................... 147
3.6.2 Do PA BNs Have Generic Reading? ......................................... 150
3.6.3 Why the Existential Reading for PA BNs? .............................. 156
3.6.3.1 BNs as Variables .................................................................... 157
3.6.3.2 Cohen and Erteschik's (2002) .............................................. 160
3.6.3.3 Categorical Vs. Thetic Judgments ......................................... 165
3.6.4 Conclusion ................................................................................ 169
Chapter 4
Focus and English Bare Nominals
4.0 Introduction .......................................................................................... 172
4.1 Diesing's Mapping Hypothesis Revisited ............................................. 175
4.2 Existential/ Generic & Focal/Non-Focal .............................................. 182
4.3 Object Bare Nominals and Focus Structure:
More Evidence ...................................................................................... 208
4.4 Affective Verbs and the Generic/Existential
Interpretation ........................................................................................ 214
4.5 Conclusion ............................................................................................ 219

vii

Chapter 5
Bare Nominals, Word Order and Reference

5.0 Introduction ...................................................................................................... 220
5.1 Sentential Stress Assignment, Focus, and Word Order ................................... 224
5.1.1 Zubizarreta (1998) ................................................................................... 224
5.1.1.1 The Nuclear Stress Rule (N SR) ........................................................... 224
5.1.1.2 The C-NSR and Prosodically-Motivated Movement ........................... 227
5.1.2 Cinque (1993) ......................................................................................... 233
5.2 Focus, Sentential Stress and Prosodic Movement in PA ................................. 238
5.3 The Distribution of Bare Nominals and Focus in English
and PA/Spanish ............................................................................................... 248
5.4 Focus Phrase (PP) and Bare Nominals ............................................................ 252
5.5 Bare Nominals and Reference in English and PA/Spanish ............................. 260
5.6 Conclusion ....................................................................................................... 265
Chapter 6
Conclusions
Conclusions ............................................................................................................ 267
Bibliography ......................................................................................................... 273

viii

Chapter 1

Introduction

1.0 Overview

In recent years there has been increased interest in exploring to what extent the
different branches of linguistics interface. Parallel to that there has also emerged a great
body of work on the similarities and differences across languages, mainly spurred and
invigorated by Chomsky's Principles and Parameters Theory (PPT) (Chomsky 1981,
1986, 1995). PPT has maintained that languages are not all that different, but that they
vary across a limited number of parameters: different sounds, different words or word
orders.

In the present dissertation I seek to combine both parallel lines of inquiry by
undertaking a cross-linguistic study that essentially attempts to investigate an aspect of
the interface between two facets of linguistic theory, namely, Syntax and Semantics.
This dissertation attempts to ﬂesh out what similarities (as well as differences) there are
between apparently different languages such as Palestinian Arabic, Spanish, Italian and,
of course, English, in terms of the structure of noun phrases (or, rather, determiner
phrases) generally, and bare nominals (i.e. bare plurals and singular mass nouns)
speciﬁcally. I attempt to ground the similarities between languages such as Palestinian
Arabic, Spanish and Italian, on the one hand, in their treatment of bare nominals, and
explain the differences these languages have with English, in terms of the larger
framework of similarities, or differences, that hold between these languages, or

language groups, in terms of focus and word order possibilities.

The focus of my attention in the current work will then be on a type of nominal that
is present in many languages with distinctly varied syntax and semantics, namely,
determinerless nominals or bare nominals. Upon closer scrutiny, the distinctive syntactic
and semantic properties of bare nominals across the aforementioned languages, it turns
out, are only surface manifestations of deeper differences between these languages in
focus and word order. In other words, the variations bare nominals reveal cross-
linguistically are predicted, or predictable, from the differences that set these languages
apart from one another in terms of the properties of the system of focus and word order

facts.

1.1 Bare Nominals: The Basic Problem

Determinerless nominals, or bare nominals, have been noted in the literature to
exhibit a curious asymmetry in distribution in many languages including Italian
(Longobardi 1994, 1996, 2000, 2002) and Spanish (Suﬁer 1982; Contreras 1986;
Casielles-Suarez 1997). On the one hand, bare nominals cannot occur preverbally under
normal conditions of stress; these nominals can show up in preverbal positions if, for
example, they are focused. This restriction on bare nominals, it has been argued, does
not hold when these nominals occur postverbally: focusing a bare nominal does not
seem to be crucial, or even required, if it is not preverbal in the sentence. The supposed
asymmetry just noted can be exempliﬁed by the following sentences from Palestinian

Arabic.l

 

‘ The purpose of the examples in (l) and (2) in the text is strictly expository, i.e. to illustrate the alleged

2

(1) a. 1i? bu wlaad ﬁ-l-hadiiqa
played-3MP boys in-the-garden
'Boys played in the garden'

b. * wlaad liS’bu ﬁ-l-hadiiqa
boys played-3MP in-the-garden

c., nizlat mayyeh min l-d3abal

came down-3FS water from the-mountain
'Water came down from the mountain.’

d. "‘ mayyeh nizlat min l-d3abal
water came down-3FS from the-mountain

The descriptive generalization the examples in (1) point to is that a bare nominal has to
occur postverbally, and not preverbally, in order for the outcome to be grammatical.
Additionally, as the standard argument in the literature would have it, this prohibition
against preverbal bare nominals is suspended or cancelled when the bare nominal
occurring preverbally is focused (2a, 2b) , modiﬁed (2c), or conjoined (2d) (focus is

indicated by capitalization):

(2) a. WLAAD linu ﬁ-l-hadiiqa
BOYS played-3MP in-the-garden
'BOYS played in the garden.’

 

asymmetry in distribution as represented in the literature. I do not commit myself at this point to the
correctness of the judgments concerning the grammaticality of these examples. As a matter of fact, I will
be arguing in the following chapters against some of the assumptions of the very view this set of examples
is said to represent.

 

b. MAYYEH nizlat min l-d3abal

WATER came down-3F S from the-mountain
'WATER came down from the mountain.’

c. wlaad min l-haara 1-mud3aawreh li‘ibu ﬁ-l-hadiiqa

boys from the-neighborhood the-next played-3MP in—the-garden
'Boys from the adjacent neighborhood played in the garden.’

d. wlaad w banaat li9bu ﬁ-l-hadiiqa

boys and girls played-3P in-the-garden
'Boys and girls played in the garden.’

The examples in (1) and (2) represent a predicament to any approach seeking to explain
or account for the distribution of bare nominals in languages such as Palestinian Arabic.
How could such an approach reconcile such apparently conﬂicting facts? On the one
hand, bare nominals are ungrammatical preverbally, grammatical postverbally, while, on
the other hand, they are seemingly acceptable in either position (i.e. pre- or post-
verbally) when focused, modiﬁed or conjoined.

These problematic issues have in fact remained unresolved within the many
approaches offered to explain the distribution of bare nominals. For instance,
Longobardi (1994) proposes to account for the asymmetry in the distribution of bare
nominals (in Italian) according to their pre— and post-verbal position in terms of a
lexical government requirement (see Chapter 3). Bare nominals, according to
Longobardi (1994), are lexically governed in their postverbal position while they fail to
fulﬁll this structural requirement preverbally. Descriptively, Longobardi's lexical
government condition provides an explanation to the paradigm in (1). However, the

paradigm of examples in (2) seems to argue against the correctness of this view that

endorses such a structural requirement.

As mentioned immediately above, the seemingly idiosyncratic behavior of bare
nominals seen in (2) has in fact remained mostly an intractable problem area for (purely)
syntactic approaches attempting to explain the distribution of bare nominals (Contreras
1986; Longobardi 1994; Casielles-Suarez 1997). Moreover, in attempting to provide a
motivated explanation for the behavior of bare nominals generally, these approaches
have inescapably posited a disjunction in their treatment of bare nominals. A theory that
does not posit such a disjunction is a priori more preferable to one that does.

The proposal I put forward in this dissertation attempts to surmount such
shortcomings while at the same time provide a motivated account of the distributional
facts of bare nominals in Palestinian Arabic and Spanish, on the one hand, and English,

on the other. In the next section I preview the proposal I advance in this dissertation.

1.2 The Proposal

The central claim of the current work is that there is no asymmetry in the
distribution of bare nominals in Palestinian Arabic or Spanish depending on their
structural position (i.e. preverbally or postverbally). These nominals are not subject to a
structural requirement such as lexical government (as in Longobardi (1994)), nor should
these nominals be claimed to be NPs unable to move out of the verb phrase (as in
Casielles-Suarez 1997) (see Chapter 3). To the extent that such a uniﬁed account of bare
nominals proves viable an important generalization will be captured in their
distribution. Such a generalization should be preferable on theoretical as well as

empirical grounds.

Now I would like to make my proposal a little more precise. The structural position
of a bare nominal plays the crucial role in licensing the bare nominal insofar as the latter
is placed in a position where nuclear stress would be assigned to it (hence, focus) by the
Nuclear Stress Rule. Nothing actually hinges on whether the nominal is lexically
governed or not as far as its licensing is concerned. To be licensed in Palestinian Arabic
or Spanish, the bare nominal has to be focalized by being assigned the nuclear stress by
the Nuclear Stress Rule (or else focalized contrastively).

The conception of the Nuclear Stress Rule I make use of in this work is that
articulated and elaborated on by Zubizarreta (1998), or Cinque (1993). Zubizarreta
(1998) conceives of the Nuclear Stress Rule as being non-monolithic, i.e. modularized.
Romance differs from Germanic, according to her analysis, in that nuclear stress in
assigned to a constituent in the former family of languages only if it is lowest in the
syntactic tree in terms of c-command in the sentence. This stress assignment mechanism
is equally operative in Germanic; however, in Germanic languages nuclear stress can
also be assigned in terms of the selectional ordering of constituents in the sentence. In
fact, Zubizarreta (1998) maintains that the second mechanism, although unordered with
respect to the ﬁrst in English, takes precedence in German.

Cinque's (1993) approach to stress assignment basically relies on the difference
between neutral and marked focus. Stress assignment, by the Nuclear Stress Rule, is
parasitic on the direction of syntactic recursion (i.e. the direction of the branching) in the
language. Nuclear stress is assigned to the constituent that is most deeply embedded,
once the direction of recursion is taken into consideration. Marked stress can be

assigned to a constituent that is not in the most deeply embedded position by a 'marked

 

focus rule' that shifts the stress away onto the focused constituent. More about both
Zubizarreta's and Cinque's analyses will be said in Chapter 5.

To reiterate, capitalizing on the role played by the Nuclear Stress Rule, as conceived
of either by Zubizarreta (1998) or Cinque (1993), I argue that bare nominals in
Palestinian Arabic and Spanish are only possible when assigned nuclear stress by the
Nuclear Stress Rule, in a position that is the lowest in the c-command ordering (or,
alternatively, most deeply embedded), which renders the bare nominal focal in the
sentence. Otherwise, if the bare nominal is not in a position to be assigned stress by the
Nuclear Stress Rule, the bare nominal will be contrastively focused as it occurs internal
to the sentence. Therefore, I argue, bare nominals in Palestinian Arabic and Spanish,
which are always subject to an existential interpretation can never be topical. This
focalized status of bare nominals is crucial even if the nominal is not assigned stress by
the Nuclear Stress Rule (i.e. it occurs sentence-internal). In this case the bare nominal
retains its focal status but it comes out contrastively focused, rather than informationally
focused. Therefore, according to my assumptions here, the examples in (1) can be recast
as in (3) (nuCIear stress assignment and (neutral) focalization indicated by underlining,

contrastive focus by capitalization):

(3) a. 1i? bu ﬁ-l-hadiiqa wlaad
played-3MP in-the-garden boys
'Boys played in the garden'

b. WLAAD li‘i‘bu (WLAAD) ﬁ-l-hadiiqa (mis BANAAT)
BOYS played-3MP (BOYS) in-the-garden (not GIRLS)

 

 

c. nizlat min l-d3abal mayyeh

came down-3F S from the-mountain water
'Water came down from the mountain.‘

d. MAYYEH nizlat (MAYYEH) min l-d3abal (mis BETROOL)
WATER came down-3F S (WATER) from the-mountain (not PETROL)

To expand the empirical coverage of my main argument I show that the same
analysis afforded to bare nominals in Palestinian Arabic and Spanish can be carried over
to existentially interpreted bare nominals in English. English bare nominals clearly
differ from their Spanish and Palestinian Arabic counterparts in that the former
distribute more freely in the sentence. However, bare nominals in English that are
subject to an existential interpretation, like all Palestinian Arabic/Spanish bare
nominals, can also be shown to be always focal. That existentially read English bare
nominals should behave like their Palestinian Arabic/ Spanish counterparts seems to be
the null hypothesis if cross-linguistic variation is to be kept at a minimum.

In order to illustrate that in fact the null hypothesis does hold, I examine more
closely the internal structure of the Palestinian Arabic DP in order to show how exactly
bare nominals in this language are deﬁcient in the general case so as to be subject to this
seemingly arbitrary restriction that they be focal in their linguistic environment. This in
turn will enable us to see how an understanding of such nominals in Palestinian Arabic
can enrich our understanding of bare nominals in English, too. With this goal in mind, I
isolate and investigate a construction in Palestinian Arabic that is both widespread and

well-studied in which bare nominals occur quite productively, namely, the so-called

'Construct State' (CS).2 In CS constructions, interestingly enough, bare nominals occur
freely, both pre- and post-verbally, and do not appear to be subject to the constraint that
they be focalized. Upon closer examination, I demonstrate, along the lines of what has
been standardly argued in the literature for the head N of the CS construction moving
overtly into a head D position, that bare nominals occurring in this construction are not
in the same structural position as bare nominals in non-CS structures occur in. I argue
that nominals are unmarked or unspeciﬁed for the +/- Deﬁniteness features, an
assumption that is the more plausible when looked at from a crosslinguistic perspective.
I assume further that the head D position possesses strong +/- Deﬁniteness features that
need to be checked overtly (i.e. in the syntax or before spell-out). Accordingly, bare
nominals in CS constructions substitute in the head D position, thereby checking in the
process the strong +/- Defmiteness features of D. Bare nominals in non-CS
constructions, by contrast, are unable to check those features since, by assumption, these
nominals are unmarked for +/- Deﬁniteness features (since these features are a property
of the D position to begin with) and are therefore unable to move into D. Bare nominals
in CS constructions, although they are equally unspeciﬁed for the +/- Deﬁniteness
features, inherit these features in CS constructions from the possessor DP obligatorily
present in this structure.

It seems to follow quite expectedly, then, that the same should hold of bare nominals

in English, since, by assumption, these nominals are unmarked for +/- Deﬁniteness

 

2 Even though the nominals that occur in Construct State constructions are morphologically identical to
the bare nominals encountered generally in non-CS constructions, it will be shown that the two groups of
nominals differ in the structural position they occupy in the syntactic tree. Therefore, using the term 'bare
nominal' to refer to those nominals occurring determinerless in CS constructions is for ease of exposition
only.

features and should be unable to substitute in the D position. English bare nominals
should also be subject to the requirement that they be focalized. This, I argue, is indeed
the case for existentially interpreted bare nominals in English. However, this state of
affairs does not hold in the case of generically interpreted bare nominals in this
language. As a possible solution to this apparent paradox I suggest that generically
interpreted bare nominals can be said to move into the speciﬁer position of a Topic
Phrase since, I argue, these nominals always carry a topical feature and are never
focalized. The DP in which these nominals are contained would then enter into a
checking relation with a head Top(ic) ﬁ'om which they would inherit their topical status.
These nominals, then, would be licensed in that speciﬁer position and interpreted
accordingly, as topics, hence deﬁnite. Another solution is to suggest that generically
interpreted bare nominals in English may be able to substitute in D at the level of
Logical Form, and are thus interpreted referentially, similar to the interpretation their
DP (with the deﬁnite article) counterparts in PA and Spanish have. The claim, therefore,
is that existentially-interpreted English bare nominals and generically-interpreted ones
are different entities. This conclusion is not unfounded especially that other languages,
such as Palestinian Arabic and Spanish, use an entirely different mechanism to express
the generic use (by using the deﬁnite article) than that used to express the existential
usage (via bare nominals).

The current proposal, then, fundamentally relies on the DP status of bare nominals
and the way or ways in which the strong features of the D position are checked and the
bare nominal is licensed. On the one hand, in CS constructions the features of D are

checked by a bare nominal moving into it overtly. The bare nominal can only do that

10

due to the obligatory presence of a possessor DP that provides the nominal with the
necessary +/- Deﬁniteness features to land in D. The (in)deﬁniteness of the DP in whose
head position the bare nominal substitutes would then be determined by which features
the bare nominal is supplied with from the possessor DP. But, what about bare nominals
in non-CS constructions? To say that these nominals are only acceptable when focused
is to state that focus somehow licenses these nominals, in a fashion that could be
parallel, though not identical, to what we see in CS constructions. If the bare nominal is
licensed in CS constructions via a possessive DP that supplies the former with the
necessary +/- Deﬁniteness features, we should then look for a licenser that performs a
similar function with respect to bare nominals in non-CS constructions. I would like to
suggest that focus does indeed license bare nominals in non-CS constructions and that
that takes place in the following way. To focus a bare nominal is to place it in a speciﬁer
position of a Focus Projection (FP), in an obvious analogy with marking the topical
status of a generically interpreted bare nominal by moving it into the speciﬁer position
of a Topic Phrase (Cf Rizzi (1997)). To say that bare nominals are DPs, rather than
NPs, means that the DP in which the bare nominal occurs would enter into a checking
relation with an F head in a Speciﬁer-Head conﬁguration within the FP projection (as, I
assume, the generically interpreted DP enters into a checking relation in a Speciﬁer-
Head conﬁguration with a Top head). Since focus is inherently a means of encoding
new information, it is therefore plausible to assume that focus is always indeﬁnite.3

This indeﬁniteness is passed on from the F head onto its speciﬁer via the very local

 

3 This does not apply to contrastive focus on deﬁnites since I argue in Chapter 4 that contrastively focused
deﬁnites are more appropriately dealt with as contrastive topics, rather than contrastive foci.

ll

relationship that holds between the two (Spec-Head conﬁguration). The DP that
contains the bare nominal is accordingly interpreted as indeﬁnite, which is exactly the
result we are looking for since bare nominals in Palestinian Arabic and Spanish are
always interpreted as indeﬁnite. Existentially-interpreted bare nominals in English
would be amiable to the same analysis.

Two questions become relevant at this point. First, if existential bare nominals in
English, and all bare nominals in Palestinian Arabic and Spanish, are focalized, hence
subject to a uniform treatment, why should English be different in terms of the
distribution of bare nominals from either Palestinian Arabic or Spanish? Second, why
should English be able to express genericity by using bare nominals, while Spanish and
Palestinian Arabic resort to the use of the deﬁnite article to express this usage?

As an answer to the ﬁrst question, notice that what we see in the examples in (3)
above is clearly different from what can be seen in their English counterparts. More
precisely, stressing an element or a constituent internal to the sentence in English does
not necessarily have to be interpreted as being contrastively focused, unlike what
happens in both Palestinian Arabic and Spanish. English makes ample use of what
Cinque (1993) calls 'marked focus' (see Chapter 5), an option unavailable for both
Palestinian Arabic or Spanish. The latter two languages exploit a different strategy,
namely, what Zubizarreta (1998) terms 'prosodically-motivated' movement, which
moves focalized constituents to be placed inside the scope of the operation of the
Nuclear Stress Rule, while at the same time displacing defocalized constituents so as not
to be assigned nuclear stress by this rule. English does not have access to this strategy of

prosodic movement since this language can instead utilize marked focus.

l2

Looking at word order in English, on the one side, and Palestinian Arabic and
Spanish, on the other side, it becomes quite expected and understandable that prosodic
movement should be available in the latter languages but not the former. English word
order is rigid, SVO for the most part. Palestinian Arabic and Spanish allow different
variations or permutations, SVO, V80, V08, to name some. Now, since a bare nominal
in both Palestinian Arabic or Spanish has to be assigned nuclear stress by the Nuclear
Stress Rule in the lowest position in the syntactic tree in terms of c-command, or most
embedded, it is predicted that a bare nominal in these two languages can be allowed to
occur in that position, and that position only. For that nominal to occur in a different
position in the sentence, i.e. sentence-internal, it can only be licensed by being
contrastively focused, an interpretively distinct option . In English, on the other hand,
since this language is far more constrained in terms of its word order possibilities and
the consequential absence of prosodic movement, a bare nominal can be focused in situ
without resorting to movement. This would in turn explain why bare nominals in this
language are afforded more freedom in the sentence, relative to what we see in both
Palestinian Arabic and Spanish .

As far as the second question is concerned, I make the assumption that generic
operators in these two language groups, English, on the one hand, and PA and Spanish,
on the other, differ in their binding possibilities. Generic operators in English can bind
nominals contained in a DP whose head D is empty or null. In PA and Spanish, by
contrast, this option is not available since generic operators are unable bind nominals in
a DP with a null head. Therefore, these two languages have to use the deﬁnite article to

express genericity. A second possible argument, which is not unrelated to my ﬁrst

l0

fi

('10

('2

2’1

assumption, is, as I argue in Chapter 4, that generically interpreted BNs in English, and
generic DPS with the deﬁnite article in PA, are never foci, but always topical. I will also
make the argument in Chapter 4 that when focused, these DPs turn out to be contrastive
topics, rather than foci. Notice that the difference between contrastive foci and
contrastive topics is a difference in the information status of the bare nominal.
Contrastive foci are essentially new information, unlike contrastive topics that do not
contribute ne_w information but rather depend on their prominence in the discourse. The
latter would plausibly include deﬁnite DPs, which I argue to be contrastive topics when
focused, because they depend for their interpretation on their prominence and the
addressee's ability to recognize them (either since they are prominent in the immediate
discourse or part of the 'shared knowledge/assumptions' or 'common ground').We can
hypothesize, then, that to be interpreted generically in a sentence a BN or a deﬁnite DP
have to be topical or defocalized. One way to cash out this solution is to say that when
topical or non-focal a BN or a deﬁnite DP is mapped into the restrictive clause of a
generic operator to be bound by that operator, a possibility which is available for
English BNs. Being focal, BNs in PA and Spanish do not ﬁt the bill, since BNs in these
two languages, due to their focal status, are arguably always caught in the nuclear scope
and never occur in the restrictive clause of a generic Operator. Therefore, these BNs

cannot be bound by generic operators.

1.3 Organization
The rest of this dissertation is structured as follows. Chapter 2 investigates the

structure of the DP in Palestinian Arabic and draws the necessary conclusions that are

also going to pertinent to bare nominals in English. The basic working hypothesis in this
chapter will be the following: by investigating the structure of the DP in Palestinian
Arabic through isolating a structure in which bare nominals in this language are
productively used, namely, the Construct State (CS), we will be in a better position to
understand the behavior of bare nominals as they occur generally in the language. By
comparing and contrasting the way bare nominals are used in the construct state to the
way they are used in non-CS contexts, we are able to pinpoint more precisely the
properties or features that non-CS bare nominals possess, or rather do not possess, to
cause them to be 'deﬁcient' somehow. As it turns out, non-CS bare nominals, being
deprived or unmarked for (in)deﬁniteness features, fail to be licensed in their
environment since they are unable to check the strong matching Def features of the D
position of the DP shell in which they occur. This very hypothesis enables us to
generalize the analysis further by predicting that bare nominals in English should be
’deﬁcient' in that respect too, a prediction that is borne out, at least with respect to
existentially read bare nominals. The cross-linguistic generality of the analysis should
be viewed as ultimately advantageous since it enables us to minimize cross-linguistic
variation as mush as possible.

In Chapter 3 I review the relevant analyses that have been proposed in the literature
as attempts to account for the seemingly idiosyncratic behavior of bare nominals. I
demonstrate that these analyses fall short of providing a satisfactory and adequate
account of the behavior of bare nominals and that there are important issues that remain
unresolved which these analyses tend to sidestep. Among the issues that stand as

problem areas for such analyses is the apparent effect focusing, modifying or

coordinating a bare nominal is said to have on the ability of the bare nominal to occur in
preverbal positions. In an analysis such as Longobardi's (1994), which basically argues
that a bare nominal is subject to a lexical government requirement, satisﬁed postverbally
but not preverbally, these issues plague the analysis to such an extent that they remain as
violations or exceptions to its generality.

In this Chapter I also make the relatively novel proposal that bare nominals in
Palestinian Arabic and Spanish, and existential bare nominals in English, are subject to
the requirement that they are focused or non-topical, a proposal that by and large
achieves a higher degree of uniformity in dealing with such nominals. The problematic
issues of modiﬁcation and coordination that are left over from traditional analyses are
not exceptions or violations according to my proposal and should not be dealt with as
such.

Since I seek to expand the empirical and conceptual coverage of my claims to
English, Chapter 4 deals primarily with illustrating that existential bare nominals in
this language behave in a fashion similar to their Palestinian Arabic and Spanish
counterparts. The central claim in this chapter is that generic bare nominals in English
are always topical, and existential ones are never so. Whether in subject or object
positions, existentially interpreted bare nominals in English can be shown to be
informationally (or contrastively) focalized, unlike the generic ones which always
surface as contrastive topics when focused. The notion of contrast is signiﬁcant here
since it will be pivotal to explaining away apparent counterexamples to this hypothesis.
The so-called aﬂective verbs, such as love, like, hate, loathe, ...etc can take object bare

nominals that can only be generically or universally interpreted but appear to be

16

informationally focused. I argue that the notion of contrast is part and parcel of the
meaning of such predicates and the object nominals these predicates take emerge as
contrastive topics, as I assume generally for generic bare nominals.

Chapter 5 attempts to derive the surface differences seen between English, on the
one hand, and Palestinian Arabic and Spanish, on the other hand, in the distribution of
bare nominals from deeper differences in the properties of the focus system and word
order facts. If existential bare nominals in English, and all bare nominals in Palestinian
Arabic and Spanish are always focal, then the differences in the distribution of these
nominals between these two groups of languages must be sought in the very properties
of focus, and the word order facts, in these languages. I show that since English is
characterized by a rigid word order, which generally rules out word order permutations
and hence signals the lack of 'prosodically-motivated' movement (Zubizarreta (1998)),
focusing in situ, or 'marked focus' in Cinque's (1993) terms, in this language is possible.
A bare nominal can be focused in English in a sentence internal position without
coming out contrastive. This, however, is not the case in Palestinian Arabic and
Spanish. A bare nominal can be informationally focused if assigned stress by the
Nuclear Stress Rule in the lowest position in the syntactic tree in terms of c-command or
in the most deeply embedded position. To focus a constituent that is internal to the
sentence would only result in contrastive focus. Therefore, or perhaps as a result,
prosodically-motivated movement is possible in these languages, a possibility made
available by the rich and ﬂexible word order possibilities that characterize both
Palestinian Arabic and Spanish. It is therefore predicted, the chapter concludes, that bare

nominals in these two languages would exhibit a limited freedom of distribution, unlike

 

what we see in English where bare nominals are more freely distributed.

In this chapter I also take up the issue of how exactly focus can license bare
nominals in Palestinian Arabic and Spanish, and existentially read bare nominals in
English. I argue that these nominals move into the speciﬁer position of a Focus Phrase
to enter into a checking relationship with the head F of that projection, thereby
inheriting the features necessary to be licensed and interpreted.

Another ﬁnding of this Chapter concerns reference in the two language groups under
consideration. Bare nominals in English can be used in generic sentences; however,
both Palestinian Arabic and Spanish use the deﬁnite article to express this usage. I
suggest that this difference has to do with the binding properties of generic operators in
these two language groups and also from the focal status of bare nominals in Palestinian
Arabic and Spanish.

Finally, in Chapter 6 I conclude this dissertation by delineating the assumptions and
the ﬁndings of the analysis argued for in this work. Moreover, areas that could be
candidates for further future research that could enrich and expand on the ﬁndings of the
current work, are alluded to in this chapter.

Before I move on to considering the structure of the DP in Palestinian Arabic in
Chapter 2, I would like to point out that I assume in this work the Principles and
Parameters theory, as in Chomsky (1986), and as amended and expanded on in

Chomsky's Minimalist Program (1995).

18

Chapter 2
The Internal Structure of the DP in Palestinian Arabic

2.0 Introduction

Understanding the behavior (or misbehavior, as some researchers prefer to dub it) of
bare nominals in Palestinian Arabic necessitates the understanding of the internal
workings and structure of the Determiner Phrase (DP) in this language. Ideally, we
should be able to ﬁnd a structure that involves the use of bare nominals, but in which
these nominals exhibit a distinct behavior and play a role markedly different from that
they would otherwise play in the language in the general case. Palestinian Arabic
provides us with precisely that opportunity for there exists in this language, as is the
case with other varieties of Arabic including the Standard variety, a construction whose
properties could potentially provide us with a deeper understanding of the characteristic
behavior of bare nominals generally. This construction is the Semitic Construct State. In
this chapter I take a closer look at the construct state in Palestinian Arabic with the
ultimate goal of shedding light on why bare nominals in this language behave the way
they do.

The rest of this chapter is organized as follows. In § 1.1 I will set the stage by giving
some descriptive preliminaries about the Semitic Construct State. § 1.2 is mainly
concerned with illustrating that bare nominals participate freely in the Semitic construct
state. In § 1.3 I argue that the strict adjacency presumed to hold between the head noun
of a construct state and its genitive argument/possessor does not always hold. In the

same section I also argue that there is some evidence to show that the deﬁnite article in

19

Arabic, which is presumably an affix, can be separated from the head it is afﬁxed to.

§ 2, with its different subsections are dedicated to a preview and a brief critique of
primarily two major approaches proposed in the literature to account for the peculiarities
of the construct state. In § 2.2.3 I present the analysis of the construct state I will be
adopting in this work. § 2.2.4 deals with the behavior of bare nominals outside of the
construct state environment. I also show in this section how an understanding of the
behavior of bare nominals in the construct state can enable us gain a better
understanding of bare nominals in the language generally. I propose that (bare) nominals
in Palestinian Arabic are unspeciﬁed for (in)deﬁniteness features and that what the
construct state environment does is help endow nominals occurring within such a
context with the necessary feature value. § 3 and its subsections investigate the
possibility of extending this analysis to bare nominals in English. I demonstrate in this
section that the analysis argued for in the present work helps us achieve a restrictive
account of the surface differences exhibited by apparently different languages, namely,

(Palestinian) Arabic and English. § 4 is the conclusion.

2.1 The Construct State: The Basic Properties

As noted in the introduction, to attempt an investigation of the behavior of bare
nominals in Palestinian Arabic (henceforth, PA), and their distribution in particular,
understanding the structure of the Arabic DP is a prerequisite. To gain a good
understanding of the Arabic DP one would perhaps be well-advised to examine a certain
structure that has aroused much interest and has been widely investigated by many

researchers, namely, the synthetic genitive possessive or, as is commonly referred to in

20

the Afro—Asiatic literature, the Construct State (CS) (see, among others, Ritter (1988,
1991), Fassi Fehri (1999), Siloni (1994, 1996, 1997), Longobardi (1996), Borer (1999),
Dobrovie-Sorin (2000), Danon (2001), and references cited there).

The construct state is a highly productive construction in Semitic languages such as
Arabic and Hebrew (see Longobardi (1996) for an attempt to argue that traces of such a
construction are also present in Romance as well as Germanic). A cursory look at the
works mentioned above would sufﬁce to show that the CS has a number of well-
documented, well-deﬁned characteristics and the attempt in these works has been to try
to account adequately for these characteristics. These characteristics include the

following:

(1) a. The head N occurs ﬁrst in the construction which has been argued to be a DP.

b. The head N carries the Case assigned to the whole DP.

c. An obligatory genitive argument/DP (or possessor) that carries genitive Case
follows N.

d. The head N lacks the (deﬁnite) article (whether the head N is interpreted as
deﬁnite or indeﬁnite).

e. The head N is interpreted as deﬁnite or indeﬁnite depending on the
(in)deﬁniteness of the genitive argument or possessor ((in)deﬁniteness
inheritance).

f. The head N must be adjacent to the genitive argument.

1 illustrate these properties with the following examples from PA:

21

(2) a. beet *(z-zalame)
house the-man
'the man's house'

a'. beet *(zalame)
house man
'a man's house'

b. (*l)-beet z-zalame
(the)-house the-man

b'. (‘l)-beet zalame
(the)-house man

0. beet-M (*l-kbiir—MS) l-mara l-kbiir-MS
house (the-big) the-woman the-big
'the woman's big house'

c'. beet-M (*kbiir) mara kbiir-MS

house (big) woman big
'a woman's big house'

d. d3aa2izit l-walad ?illi ribih-a ﬁ s-sibaag
prize-PS the-boy that won-3MS-3F S in the-race
'the boy's prize that he won in the race'

d'. d3aa2izit walad ribih-a ﬁ s-sibaag

prize-FS boy won-3MS-3FS in the-race
'a boy's prize (that) he won in the race'

(2a) above is an example of a deﬁnite construct state construction whereas (2a') is an
instance of an indeﬁnite CS. These two examples illustrate properties (1a) and (1c) as
can be seen from the initial placement of the head N beet 'house' (property (1a)) and the

resulting ungrarnmaticality if the genitive argument is omitted (property (lc)).l

 

' The relevant examples in the text would be acceptable with the deletion of the genitive argument, but
they would no longer be instances of CS. Once the genitive argument is deleted the examples would
merely be bare singular nouns with no understood possessive reading.

22

 

Examples (2b) and (2b') illustrate property (1d), i.e. the ungrarnmaticality of a deﬁnite
article with the head N, whether the latter is deﬁnite or indeﬁnite. As for (2c) and (2c'),
these examples illustrate (in)deﬁniteness inheritance (property (le)) where we see in
(2c) and (2c') that a modifying adjective (l-kbiir 'the-big' or kbiir 'big') of the head noun
beet 'house' agrees with the head it modiﬁes in deﬁniteness or indefmiteness,
respectively.2 In other words, the (in)deﬁniteness of the head N is parasitic on the
(in)deﬁniteness of the genitive argument. This is further illustrated in (2d) and (2d') with
a modifying relative clause. In (2d) a deﬁnite relative clause modiﬁes the head N that is

followed by a deﬁnite genitive argument. The complementizer PiIIi 'that' only appears

when the modiﬁed head N carries a deﬁniteness feature; its absence in (2d') is then

expected since the head N d3aa Pizit 'prize-FS' is indeﬁnite (by virtue of having an

indeﬁnite genitive argument walad 'boy'). Finally, (2c) and (2c') illustrate property (1 f)
above (i.e. strict adjacency): a modifying adjective of the head noun obligatorily shows
up at the right edge of the CS construction after the genitive argument and not after the
head noun. Notice that it is certainly possible to modify both the head noun and the
genitive argument, resulting in a nesting structure that could potentially give rise to

ambiguity if the agreement features were to overlap as in (3):

(3) d3aa?izit-FS l-binit l-kbiire-FS l-hilwe-FS
prize the-girl the-big the-beautiful

 

2 Note that adjectives in PA agree with the head they modify in gender, number, Case (see the proviso in

the text concerning Case in PA) and (in)deﬁniteness. 1n the case of (in)deﬁniteness the agreement will be
marked syntactically by the presence (deﬁnite) or the absence (indeﬁnite) of the article I- 'the.‘ Note also

that PA does not have an indeﬁnite article.

23

 

'the prize of the beautiful, big girl'
'the big girl's beautiful prize'
'the girl’s beautiful, big prize'3

It can be noted from the preceding discussion that the property (1 b) is not illustrated by
any of the examples in (2) since morphological Case is not realized on nominals in PA,
unlike standard Arabic that exhibits a rich pattern of Case and Case agreement on

nominals and their modifying adjectives. I assume that Case is abstract in PA.

2.2 Plurals in Construct State Constructions

A property of the construct state that does not normally get mentioned, and which
will prove of great interest to my main concerns in this work, is that the head noun in
CS constructions is not restricted to the class of singular nominals in PA. Plural
nominals can occur freely as head nouns of the CS and these plurals adhere quite closely
to the set of rules, or properties, characterizing CS constructions with singular head
nouns mentioned in (1) above. Most importantly, these plural nominals can, and, in fact,

have to, appear bare in the head position of the construct state as a consequence of (1d)

 

3 I ﬁnd the fourth logical possibility in (i) extremely marginal at best, downright ungrarnmatical at worst:
(i) 'the big prize of the beautiﬁrl girl'
This possibility is not realized if we follow Fassi Fehri (1999) (who follows Cinque 1996) in assuming
that adjectives are generated in the speciﬁer position of the head they modify. This being so, I-kbiire 'the-
big' in (i) has to be generated in Spec position of the head N d3aa Pizit 'prize.’ The possessor I-bim't 'the-
girl' would be generated in a position higher than both of the head N d3aa f’izit and its modifying
adjective, and I-hilwe ‘the-beautiful' in a position higher yet than the possessor I—bint. Assuming then
movement of N to a higher functional head (D, see below in the text), and the possessor to a position
higher than both adjectives but lower than the head N, this order for the modifying adjectives with the
reading in (i) would not be possible (see Fassi F ehri (1999) for details and execution of movement
operations). Note also that in the second possibility in the text (namely, the big girl '5' beautiﬁd prize) the
adjective beautiﬁd would be read predicatively, rather than attributively (meaning, the big girl 's prize
which is beauttﬁtl).

24

above.4 This latter fact concerning the head of construct state will turn out to be a

crucial element in our attempt to understand the overall behavior of bare nominals in

general.

For concreteness, consider the following examples:

(4) a. (*l-) byuut (*l-gadiime) l-zalame (l-gadiirne) inharagu—MP
(the-) houses (the-old) the-man (the-old) burned
'The man's old houses bumed'

b. (*l-) byuut (*gadiime) zalame (gadiime) inharagu-MP
(the-) houses (old) man (old) burned
'Old houses of a man bumed'

(5) a. hakeet ma? (*l-) Tullab (*z-zvaar) l-madrase (z—z‘taar)
spoke-l S with (the) students (the-young) the-school (the-young)
'I spoke to the young students of the school'

b. hakeet ma? (*l-) Tullab (*ztraar) madrase (zxaar)

spoke-1 S with (the-) students (young) school (young)
'I spoke to young students of a school'

(4) and (5) with plural heads illustrate the properties of the construct state mentioned in
(1) above. Above all, these examples serve to show that the singular-plural distinction is
of no consequence as far as the construct state structure is concerned. These examples
also illustrate that there are no structural restrictions on the placement of the construct

state. In (4) the CS occurs preverbally while it is a prepositional object occurring

‘

‘ However, as will be seen later on in the text, bare plurals/singulars in the CS are not in the same

Structural position they are normally in in non-CS constructions. Therefore the notion of bareness here has
‘0 be understood in this context and in these terms. Perhaps determinerless would be a more appropriate
way of referring to nominals in CS constructions. I will still keep using bare nominals for ease of
reference_

25

postverbally in (5).

2.3 Strict Adjacency in the Construct State

Since the goal of this chapter is to explore the internal structure of the DP, in this
section I would like to make a brief digression to discuss a fact concerning one of the
properties of the CS constructions that, to my knowledge, has not received enough
attention in the literature. It has been noted earlier that one of the characteristics of CS
constructions is the strict adjacency between the head noun occurring initially in the
structure and the genitive argument (or possessor) following it. The question that can be
raised at this point is the following: does this supposed adjacency hold all the time or is
it only a condition regulating the positioning of modifying adjectives alone? As it turns
out, this supposed condition of strict adjacency on CS constructions does not seem to

hold all the time. Consider the following examples from PA:

(6) hakeet maS‘ Saahib—M Galae bnuuk-MP (maShuuriin—MP)

spoke-1 S with owner three banks (famous)
'I spoke to an owner of three famous banks'

(7) hakeet ma? Saahib-MS (*muhtaram) 0ala9 bnuuk-MP (muhtaram-MS)

spoke-l S with owner (respected) three banks (respected)
'I spoke to an owner of three banks'

The state of affairs (6) seems to point to is the following: strict adjacency of the head

noun and the genitive DP may be broken by a numeral modifying the genitive DP.5

#

5 A question here arises regarding the difference between attributive modifying adjectives and numerals in

26

Note also that the fact that the numeral occurs between the head N and the genitive DP
does not interfere with latter's ability to be modiﬁed by an adjective exhibiting the usual

pattern of (in)deﬁniteness agreement. ma fhuuriin-MP 'famous' in (6) modiﬁes the

genitive DP bnuuk 'banks' and it agrees with it in indefmiteness (and gender and
number). (7) shows that the head N can also be modiﬁed by an adjective but the latter
would more plausibly be read as predicative, rather than attributive.

A similar, though not identical, state of affairs can be observed with examples
involving deﬁnite CS constructions as in (10). First, however, note that recursion is
possible in construct state structures as in (8). It can be further noticed that the deﬁnite

article attaches only to the last genitive phrase in the string:

(8) saTih-M (*l-)beet (*l-)Saahib l-bank (l-gadiim-M)
roof (the-)house (the-) owner the-bank (the-old)
'the (old) roof of the bank owner's house'

(9) hakeet ma? Saahib l-bnuuk-MP (l-maShuuriin-MP)

spoke-1S with owner the-banks (the—famous)
'I spoke to the owner of the famous banks'

turns of their ability to interrupt the strict adjacency between the head N and the genitive phrase.
Numerals can occur medially between the head N and the genitive DP. Modifying adjectives, on the other
hand, can never interrupt the adjacency seen between the head N and the genitive DP. The question takes
on more weight if we regard numerals as (attributive) modifying adjectives. Note, ﬁnally, that modifying
adj CCtives cannot generally occur prenominally, i.e. before the head N (cf Fassi F ehri 1999). For a
s“ggestion, see below in the text.

27

(10) hakeet maS‘ Saahib l-BalaB bnuuk-MP (l-maShuuriin—MP)

spoke-1 S with owner the-three banks (the-famous)
'I spoke to the owner of the three famous banks'

(11) * hakeet ma? Saahib 0a1a0 l-bnuuk-MP (l-majhuuriin-MP)
spoke-IS with owner three the-banks (the-famous)

(12) hakeet ma? Saahib-M (*l-muhtaram) l-GalaG bnuuk (l-muhtaram-M)

spoke-IS with owner (the-respected) the-three banks (the-respected)
'I spoke to a (respected) owner of three banks'

If recursion is possible in CS deﬁnite constructions with the whole construction taking
on a deﬁnite reading, provided that the deﬁnite article obligatorily attaches to the last
genitive DP in the string, then what should we make of the status of example (10) and
its ungrarnmatical counterpart (11)? Descriptively, what we have in (10) is a head N

Saahib 'owner' occurring initially as is customary in CS constructions, followed by a
numeral Halad 'three' carrying the deﬁnite article I- 'the' and a genitive DP bunuuk

'banks' occurring ﬁnally in the string. Mysteriously, the genitive phrase fails to carry the
deﬁnite article, as is expected with a recursive string, and (l 1) is consequently
ungrarnmatical; however, (10) is perfect and is still interpreted as deﬁnite as witnessed

by the fact that the agreeing adjective l-mafhuuriin 'the-famous' carries the deﬁnite
article obligatorily. In fact l-maﬂruuriin 'the-farnous' agrees with the genitive phrase in

deﬁniteness, gender, and number. In other words, the whole construct state is interpreted
as deﬁnite since the genitive phrase is deﬁnite. This is further illustrated in (12) where

the adjective l-muhtaram 'the-respected,' which is interpreted as predicated of Saahib

28

'owner,‘ carries the deﬁnite article.

The discussion immediately above shows that a numeral (or, NumP) can occur
between the head N and the genitive phrase. In fact, the sentences in (8)-(12) seem to
point to a stronger observation: If (9), repeated here as (13), is a typical example of a
deﬁnite CS construction, in which the article attaches to the genitive phrase, then (10),
repeated as (14), illustrates a case where adjacency is not even observed between the
deﬁnite article I- 'the' and the nominal it attaches to, bnuuk 'banks.’6 In other words, the
deﬁnite article in PA requires the same treatment afforded to the deﬁnite article in

English:

(13) hakeet ma? Saahib l-bnuuk-MP (l-maShuuriin-MP)
spoke-1 S with owner the-banks (the-famous)
'I spoke to the owner of the famous banks'

(14) hakeet ma? Saahib l-9ala9 bnuuk-MP (l-majhuuriin-MP)

spoke-1 S with owner the-three banks (the-famous)
'I spoke to the owner of the three famous banks'

 

6 Aside from their use in CS constructions, numerals such as HaIaB 'three' , tis i’a ‘nine' ...etc cannot
ordinarily combine with a deﬁnite nominal as in (i), but with indeﬁnites as in (ii). This could be due to the
fact that a numeral combining with a deﬁnite nominal in PA indicates a partitive reading, a reading only
possible to express with deﬁnites using a preposition as in (iii). The numeral can also modify a deﬁnite
nominal without triggering a partitive reading but in this case the numeral occurs postrrominally as in (iv).
However, I think that the point in the text concerning the lack of adjacency still stands.
(i) *Galae l-wlaad
three the-boys
(ii) GalaG wlaad
three boys
(iii) BalaB min l-wlaad
three of the-boys
(iv) l-wlaad l-Balae
the-boys the-three

29

Before bringing the present discussion to a close a word on the difference between
numerals and attributive adjectives in CS constructions, as noted in ﬁr. 5, is in order.
That strict adjacency between the head N of the construct state and its genitive phrase
can be interrupted by a numeral, but not an attributive adjective, could be a consequence
of where in the base structure numerals and attributive adjectives are generated.
Following Fassi Fehri (1999), who follows Cinque (1996), I assume that adjectives are
generated in the speciﬁer position of the head noun they modify before the latter moves
to an initial position.7 The genitive phrase/possessor would raise, as will be explained
shortly, to a position higher than all adjectives but lower than the head N. The numeral,
on the other hand, would be generated in NumP, a projection the discussion above has
showed to project optionally between the head N and the genitive phrase.

To sum up, the discussion in the this section has basically helped illustrate two
points: ﬁrst, strict adjacency between the head N of the construct state and the genitive

phrase is not always required, or necessarily observed. At least one projection has been

 

7 Fassi Fehri (1999) makes the claim that Arabic is, like English, and A-N (Adj-N) language where
adjectives occur prenominally in the base position and are placed postnominally as a consequence of the
noun moving leftward. He ﬁrst draws attention to the fact that adjectives in (Standard) Arabic are
generally postnominal, rather than prenominal. He then points out (1999: 108) that postnominal
adjectives in Arabic observe the mirror image ordering (M10) of prenominal adjectives in an A-N
language (like English) (for example, object adjectives have the ordering quality> size> shape> color>
provenance in Germanic, but the M10 in Arabic). This could be indicative of the prenominal origin of
Arabic adjectives and the movement of the head N. Second, purely attributive adjectives in English (as in
the alleged murderer), which show up prenominally, can be shown to fail the predicativity test in Arabic
as in (i):
(i) ” l-qaatil-u maz'uum-un
the-killer-nom alleged-nom
'the killer is alleged‘
This shows that although the adjective occurs postnominally in Arabic it is still attributive. So,
Postnominal positioning of the adjective is not indicative of predicativity. Third, according to Fassi Fehri
(P- l 12) prenominal adjectives do exist in Arabic as in (ii):
(ii) ?akal-tu ladiida t-ta'aam-i
ate-I delicious-ace the-food-gen
'1 ate the delicious (of the) food.

30

shown to occur between the position in which N sits and the projection containing the
genitive phrase, namely, a NumP. The second point our discussion has attempted to
bring out is that adjacency can be broken between the deﬁnite article in PA and the

nominal it attaches to.

2.4 The Construct State Analyzed

Accounting for the properties of the construct state listed in (1) has been a matter of
considerable debate among researchers. One common thread between the numerous
approaches to the construct state is that the surface position of the head N is a derived
one and the whole structure is a DP. Another point of agreement has been that the head
N raises to D. However, disagreement is the common theme among researchers when it
comes to trying to account for the other properties. In what follows I will attempt a
brief, yet somewhat critical, overview of two approaches to construct state in F assi Fehri
(1999) and Longobardi (1996). Whenever relevant, mention will also be made to other

analyses.

2.4.1 Fassi Fehri (1999)

F assi Fehri (1999) adopts a 'split' or 'ﬁssioned‘ DP structure in order to account for
the behavior of the construct state in Standard Arabic. In such a structure Case and
Def(initeness) are assigned by separate D heads in the DP. It is assumed in this system

that movement occurs for both the head N in the construct state as well as the

31

Ia
Pﬁip
lit a
Ola }
‘5 Lie

““4.
hng:

possessor.8 However, the target and the motivation for movement in both cases is
different. Therefore, the example in (15) from Standard Arabic would be represented by
the syntactic tree in (16) ((15) is his example (53a) and the tree representation is his
(56)) (I have taken the liberty to add onto the original tree from Fassi F ehri since some
projections or their heads are not very clear in his original structure. The lower section
of the tree, however, that includes the up projections, is not very clear to me from his

formulation):9

(15) htaraqa-t daar-u r-rajul-i l-waas'a—t-u
burned-fem house-nom the-man-gen the-large-fem—nom
'The man's large house burned.’

 

8 Fassi Fehri ( 1999: 121) claims that adjectives move in Arabic due to the richness of their inﬂectional
properties and that these features on the adjectives are checked against matching features in a DP domain.
He argues that adjectives "target DP, to check their agreeing Case, article, and phi-features against those
of a higher functional head, which could arguably be (a segment of) D” (ibid). He then "designate[s] this
D with a small d, for ease of reference, giving this notation no theoretical content" (ibid). The claim, then,
is that adjectives move to a dp projection containing Agr(eement) features. For Fassi Fehri, one piece of
evidence for movement of APs in Arabic is the placement of adverbial modiﬁers of adjectives as in (i):
(i) l-axbar-u l-mudaa'u mu?axxar-an

the-new the-broadcast late-acc

'the lately broadcast news'
He points out (p. 122) that assuming a modifying adverb is generated in the speciﬁer position of the AP it

mOdiﬁes then the positioning of the AP before the adverbial results from the presumed movement of the
341’ aeross the former.

It is not very clear in F assi Fehri's analysis what the status of npl (or np2) is .

32

DP2

02'
(16)
Spec DP1
DZ .
daar-ui D1
house
Spec dp
r-rajuli-k
the man D1 np2
I-waasi'a j np1
the-large
ek
e e

The basic intuition behind Fassi F ehri's analysis is that movement, or raising, occurs for
both the head N and the possessor DP, but that movement is triggered in each case for
different reasons. Since the head N inherits its deﬁniteness or indeﬁniteness from the
possessor, then it is reasonable, according to Fassi Fehri (1999: 127), to say that the
possessor raises to check the strong Def features of a head D (D, in his analysis), by
moving in the speciﬁer position of DP1. However, since it has been argued in the
literature that the head N moves or raises to a head D position, then the position to
which N moves has to be another D position, hence the ﬁssioned or split structure Fassi
Fehri argues for. But, Fassi Fehri contends, if the possessor moves to [spec, DP1] to
check the strong Def features of D], then it follows that the position to which N raises,
namely, D2, could not contain Def features to be checked, because the latter features are
a Preperty of D1.

The trigger for movement of N to D2, F assi Fehri points out, would then be the

33

strong Case features of the D2 position. The head N would then raise to check the strong
Case features of that position against its own Case features. The head N would then
inherit the Def features from the possessor by adjoining to D on its way to D2. Recall
that the possessor in Fassi Fehri's analysis sits in a Spec-Head conﬁguration with D1
(i.e. sits in the speciﬁer position of DP.) and the deﬁniteness value of the possessor
would be passed on to D1 to be picked up by the head N on its way to D2.This would
explain the dependence of the head N in its deﬁniteness speciﬁcation on the possessor.
Fassi F ehri further argues that the correctness of this ﬁssioned or split structure is
reinforced by the observation that in some cases the possessor does not move and
therefore defmiteness inheritance does not take place. The following examples are used

to illustrate his point (his 58a and b):

(17) haaaaa ?ax-ii wa haa5aa (?ayDa-an) flax-ii
this brother-mine and this (also) brother-mine
'This is my brother and this is (also) my brother.’

( 18) haa5aa bayt—u r-raj uli wa-haa5aa (?yD-an) bayt-u r—raj ul-i

this house—nom the-man-gen and-this (also) house-mom the—man-gen
'This is the man's house and this is (also) the man's house.’

If definiteness is understood in terms of either familiarity or individual or unique

l.efel‘ence then the example in (l 8) does not fulﬁll either since it is used predicatively
and non-referentially. This means, Fehri suggests, that construct state constructions do
not necessarily have to realize individual reference, hence deﬁniteness. He takes this to
mean that in indeﬁnite construct state constructions the possessor does not move.

ITIOVVever, he reasons, the head N has to move independently whether the possessor

34

moves or not since the former has to check the strong Case features of D2. This provides
the evidence needed to support such split structure.

I believe that Fassi Fehri's analysis unnecessarily differentiates between the
derivation of deﬁnite and indeﬁnite construct state constructions. Furthermore, it seems
to me that his argument concerning the examples in (17) and (18) above does not hold.
What seems to be relevant in the examples in (17)-(l 8) for Fehri is in fact semantic
deﬁniteness, and not syntactic deﬁniteness (see Danon 2001 for arguments and
distinction between syntactic and semantic deﬁniteness in Hebrew (construct state, in
particular». It can be demonstrated that smtactic deﬁniteness does hold in the examples
under discussion if we assume (as Danon (2001) does) that adjectival modiﬁcation
constitutes a good test for checking for such deﬁniteness. Consider, in fact, how

modifying adjectives have to obligatorily carry the deﬁnite article in the relevant

examples above. 1 demonstrate this observation with examples from PA where the same

facts holdz'0

( l 9) haa5a flax-uy z-zviir w haa5a ?ax-iuy l-ikbiir
this brother-IS the-young and this brother-1 S the-old
'This is my young brother and this is my old brother.‘

(20) haa6a beet z-zalarne l-id3diid wa-haa5a beet z-zalarne l-gadiim
this house the-man the-new and-this house the-man the-old
'This is the man's new house and this is the man's old house.’

\

10

“Fassi F€=hri does mention that the lack of individual reference, and therefore deﬁniteness, holds when no

de‘ngal modiﬁcation is involved. The fact still remains that adjectival modiﬁcation has to carry the

the Cite article which can be taken as a sign of syntactic deﬁniteness via agreement with the head noun in
° S)Intactic defmiteness is the relevant notion here, not semantic deﬁniteness, as Fehri implicitly

 

assumes

35

Therefore, in indeﬁnite CS constructions, the absence of the deﬁnite article on a
modifying adjective should be taken as agreement in indeﬁniteness as in the following

example from PA:

(20') haa5a beet zalame d3diid/*l-id3diid
this house man -new/ thenew
'This is a man's new house'

As seen in (19) and (20) syntactic deﬁniteness i_s realized on the head N and the
agreeing adjective. The deﬁniteness feature in Fehri's system is a formal feature (in the
sense of Chomsky 1995) that is not interpretable at LF and therefore has to be checked
before spell-out. I believe that it is the correct intuition to look at deﬁniteness as a
formal feature, as Fehri does, that is visible in the syntax. This means that the only type
of deﬁniteness that is, and in fact should be, relevant in construct state constructions is
syntactic deﬁniteness that is discharged via feature checking, but not semantic
deﬁniteness.

What has been said so far renders suspect a system that posits movement for deﬁnite
CS constructions while denying such movement for indeﬁnites. It appears more
theoretically appealing in light of the empirical evidence to treat construct state
constructions, whether deﬁnite or indeﬁnite, as a uniﬁed phenomenon that calls for a
uniﬁed approach.

As noted above in § 1.2, (bare) nominals can occur as heads in CS constructions. In
terms of the present system this means that these nominals raise to D2 overtly for Case

purposes. That the possessor, as argued by Fassi Fehri, does not raise/move in indeﬁnite

36

construct state constructions means that movement of N to D2 is taken to be independent
of raising the possessor. A question here arises concerning bare plurals in non-CS
constructions: if the possessor plays no role in the (overt) raising of N to D, thus
exhibiting the well-known feature (1d) above (no deﬁnite article), what prevents bare
nominals in general in non-CS constructions from raising overtly to D if Case is the
driving force for such movement? That bare nominals in non-CS constructions do not
exhibit property (1d) of the construct state, and therefore cannot be said to raise

(overtly) to D, is illustrated in (21a) contrasted with the CS-sentence in (22):

(21) a 2ad3-u (l-) wlaad
came-3MP (the) boys
' (The) Boys came.
(21) b. ?ad3-u (*l-) wlaad l-d3iraan z-z‘raar

came-3MP (the) boys the-neighbors the-young
'The neighbors' young boys came.’

The argument concerning (21a) and (21b) has two parts. First, the bare plural wlaad
'boys' in the non-CS sentence in (21a) should be read as deﬁnite if the bare plural in this
case truly moves into the D position. However, wlaad 'boys' in (21a) can only have an
indeﬁnite reading (if the deﬁnite article is not realized on N). On the other hand, wlaad
'boys' in the CS-sentence in (21b) is in fact read as deﬁnite as witnessed by the

modiﬁcation by the adjective z-z mar 'the-young' which carries the deﬁnite article.

Secondly, the deﬁnite article can be lexically realized with the bare plural in (21a), but,

expectedly, it cannot show up with the nominal in the CS-sentence in (21b). This ﬁrrther

37

indicates that whereas wlaad 'boys' in (21 b) sits in the D position, wlaad in (21a) does
not.

If the possessor plays absolutely no part in facilitating N-movement to take place
then we cannot account for this striking difference in the behavior of essentially
identical nominals. What would prevent the bare nominal in non-CS-constructions from
raising to D if checking the strong Case features of that position is the trigger of such
movement? (Actually, the nominal would have to raise obligatorily in this scenario to
check Case, if Case was truly the relevant feature). It is perhaps the wrong assumption
to presmne movement of the head N for Case checking since bare plurals and singulars
in non-CS constructions do not exhibit the same movement operation as (21a) clearly
illustrates. I therefore conclude that it may be wrong to assume that raising of the head
nominal to D can be independent of the raising of the possessor to some higher
projection below D. The movement of head N in the construct state seems to be strongly
parasitic on (the movement of) the possessor, and (in)deﬁniteness inheritance and the
lack of the deﬁnite article on the head N in D are symptomatic of this strong
relationship between the pair.

Summing up, Fassi F ehri's account misses the mark by assuming that indeﬁnite
construct state constructions do not involve raising of the possessor. It does not seem
correct either to assume that raising of N is independent of, and separate from, raising of
the possessor. By making this latter assumption we lose a uniform treatment of (bare)
plural nominals (and singular nominals) that would otherwise be preferred. Finally,
Case cannot be the motivation for N-raising since this assumption would not help us

understand why bare nominals generally in non-CS constructions fail to raise to D.

38

I will next look at an approach argued for in Longobardi (1996) that makes a

different set of assumptions than those made in Fassi Fehri (1999).

2.4.2 Longobardi (1996)

Longobardi (1996) is an attempt to parameterize the existence of the construct state
cross-linguistically. He argues that Romance and Germanic exhibit N to D raising (in a
class of lexically restricted words in Romance, unrestricted in Germanic), overtly in the
former covertly in the latter, that results in a non-prepositional genitive similar to the
construct state in Semitic. Longobardi attempts to offer a parametric approach for

Romance, Germanic, and Semitic that would deduce as a property of Universal

Grammar the following descriptive generalization (1996: 6):

(22) Movement of a common noun to (an empty) D is licensed only if an overt or
understood genitive argument is realized (in other words, if a corresponding

argument role is somehow discharged)

Longobardi shows that Romance exhibits overtly a variant of Semitic construct state
with a very limited stock of lexical items such as casa 'home.’ However, the Romance
construct state differs from that of Semitic in disallowing plural head nouns, disallowing
modiﬁcation by a restrictive relative clause, and by rigidity of designation and
transparency as in the case of proper nouns.

Longobardi (1994) has also argued for such movement of N to D, overt in Romance

and covert in Germanic. In the case of Romance, this movement has been restricted to

39

proper nouns. A fundamental claim he makes in this work is that raising to D only
affects proper nouns at S-Structure in Romance while raising of common nouns is not
possible at this level. This is due to the need to check a +/- R(eferential) feature of D (to
achieve direct reference via substitution into D as in the case of proper nouns).
However, languages such as Romanian exhibit a productive system of raising common
nouns to D at S-Structure which is unexpected in Longobardi's system. In order to
rescue his assumption that no S-Structure movement of common nouns to D is possible
Longobardi is therefore forced to assume two, and only two, types of movement into D:
aay'unction and substitution. Substitution into an empty determiner, he maintains, takes
place in Romance while adjunction crucially attaches the common noun to an overt
operator-the deﬁnite article- in Romanian.

In view of this state of affairs, the Semitic construct state provides primafacie
counter-evidence to Longobardi's (1994) account. Semitic construct state displays a
productive pattern of common noun overt raising to an empty D, like that of Romance
proper nouns, that is not lexically restricted to any class of words. The Semitic construct
state clearly does not ﬁt the Romance pattern of substitution into D due to the
perspicuous differences between the two in terms of productivity of the construction, the
class of nouns it involves, plurality and singularity of the head noun, and the rigidity and
transparency of designation or lack thereof. However, the Semitic construct state does
not seem to ﬁt the Romanian pattern either since in the former the head N appears to
substitute in D therefore precluding the appearance of the deﬁnite article (I-) unlike
what happens in Romanian where the noun left-adjoins to a lexical definite article in D.

To maintain that there are only two types of movement to D, Longobardi (1996) is

40

therefore forced to assimilate the raising of the head N in the Semitic construct state to
the Romanian pattern. He implements this in the following manner: suppose that raising
the head N in Semitic does not take place as substitution into an empty D as in
Romance, but rather as adjunction to an operator in D as in Romanian. However, since
no deﬁnite article is allowed in D in the presence of the head N in Semitic construct
state then the adjunction should be assumed to take place to a null operator in D (a null
article). ”

Before assessing Longobardi's arguments concerning the difference between
Romance on the one hand and Semitic on the other in terms of the properties of D, I
want to brieﬂy review Longobardi's assumptions and execution of the derivation of

construct state in the next section.

2.5 Deriving the Semitic Construct State

We have seen above that Fassi F ehri (1999) treats raising of the head N
independently of raising the possessor in CS constructions. I have also remarked above
that if the genitive phrase fails to raise in indeﬁnite CS constructions, but that the head
N raises anyway, this would leave the behavior of bare nominals in the general case in
PA unexplained. Speciﬁcally, we would not be able to say why, if Case is the
motivation for N movement in the construct state, bare nominals (both plural and
singular) fail to raise to D in non-CS contexts. Longobardi (1996), unlike Fassi Fehri,

does in fact assume that the raising of the head N is dependent on the raising of the

 

” Note that Longobardi's account makes raising to D of common nouns such as casa 'home' in Romance
parasitic on that of raising of proper nouns. In other words, raised common nouns in Romance are object-
referring just like proper nouns.

41

genitive argument/possessor.

Longobardi (1996) adopts the proposed analysis to Semitic construct state advanced
in Siloni (1994). According to that proposal, Longobardi (1996:25) points out, genitive
Case is a structural Case assigned or checked in a Spec-Head conﬁguration. The abstract
Case head in this case is Agrgen (i.e. Agrgenmve). Agrgen does not occur in D but in a
distinct head position directly below it in its own AngP (Agr Genitive Phrase)

according to the schema in (23) (before N raising has taken place):

(23)[DP [D e HAngP [Agr e lIXP-m]

Longobardi assumes that genitive Case in Semitic has to be both licensed and identiﬁed.
One of the two ways that Case can be licensed on a category or by a head 7 is if "or is a
member of the internal domain of a Chain headed by y" (1996:28). Identiﬁcation can
also be achieved in one of two ways. Relevant to Semitic is that if "or is in the Spec of a
designated category B" (ibid). Once N raises to D in the Semitic construct state Case-
checking of the genitive DP is licensed in D's internal domain and the genitive Case on
that DP is identiﬁed in the speciﬁer position of AngP. This so-called Case Checking
Principle of Longobardi is intended to capture nonstipulatively the following descriptive

generalizations he makes (p.24):

(24) a. If a common noun raises to D a prepositionless Genitive occurs.

b. If a prepositionless Genitive occurs a common noun raises to D.

42

Here Longobardi clearly establishes the link that was missing in F assi F ehri's system
considered above, that between the raising of N to D and the presence of the genitive
DP. However, he rejects the idea that N raises to D in order to check (Case) features of
the genitive DP or check features of its own. Rather, he argues that N raises to D in
order to check features of D itself, which to him argues against a version of Last Resort
in terms of Greed.

Longobardi further argues that the trigger for N raising in the Semitic construct state
is to check +/- deﬁnite features of the DP. He assumes that D in Semitic has an abstract
grammatical label [+article] that is part of the "feature content" of an empty operator
licensed in D but in need of identiﬁcation (pp. 31-32). This [+article] is -Interpretable

and strong in Semitic. It is checked according to the following condition (p.32):

(25) Defmiteness Condition:

A +article feature can be checked by any lexical element in D (i.e. attached to
it by Move) displaying a speciﬁed + or - deﬁnite value.

To quote, Longobardi executes the movement operations and feature checking in
Semitic construct state as follows (p.32):

Suppose[...] that Ang may inherit the deﬁniteness value of the genitive
argument in its Spec and that the raising of a noun to D takes place in the
following way: N ﬁrst adjoins to Ang and then the new complex so formed
adjoins to (or substitutes for) D . Now the condition on the identiﬁcation of D
will be satisﬁed: the raising of the lexical noun (along with Ang) to D in
Semitic will be functionally motivated and licensed by the need to check the
+article feature and the presence of a prepositionless genitive is explained

by the fact that this operation can be performed only by dragging to D an
Ang being in a Spec-Head relation with a genitive argument.

43

If I understand Longobardi correctly, it seems to me that although he makes the link,
missing from F assi F ehri (1999), between raising of N and the presence of the genitive
phrase, he still does not articulate precisely why it is necessary for a genitive phrase to
be present. To put it differently, if it is not +/— Def features of D per se that the head N is
checking, but a [+ article] feature of the latter position, then it is not very clear why a

genitive phrase has to be present in a Spec-Head conﬁguration with an Agr head.

2.5.1 The Status of D in Semitic

As has been noted above, Longobardi (1996) tries to assimilate the Semitic pattern
of N raising to D to that of Romanian. In the former, N will adjoin to a null operator,
differing in this respect from Romance, while in Romanian N will adjoin to an overt
operator in D, namely, the deﬁnite article. Longobardi's assumption of the existence of a
null operator to which the head N adjoins in Semitic is based on the observation that the
distribution of empty deterrniners in Arabic and Hebrew is free(r) than it is in Romance
or Germanic. Bare singular nouns, Longobardi argues, can occur freely with no
restriction on the empty D to a mass/plural reading as is the case in Romance (and most
Germanic). Hebrew (and Arabic) are therefore article drop languages which explains
the lexically generalized occurrence of common nouns in the construct state in Semitic.

Longobardi's analysis seems to beg the following question: if Semitic, speciﬁcally
Arabic and Hebrew, has a [+article] feature that is strong and has to be checked prior to
spell-out, and if in the construct state movement of N to D checks that feature (such

movement, of course, being parasitic on the generation of a possessive phrase to provide

N with the necessary + or - Def feature value),12 then should not the prediction be that
the [+article] feature is also present and in need of checking before spell-out (being
strong) whenever article drop takes place? In other words, bare singular count nouns in
non-CS constructions should not be possible or grammatical in Arabic or Hebrew (since
these nouns surface articleless) due to the strength of the [+article] features. However,
Longobardi claims that these nouns distribute freely in Hebrew articleless and are
possible in argument position anywhere. As such, the strong [+article] feature in these
cases would remain unchecked and the derivation should therefore crash.l3

Another point concerns bare plurals in Semitic, more speciﬁcally Palestinian Arabic.
Bare plurals in PA, as will be explained in more detail in subsequent chapters, have the
same distribution and interpretation as bare plurals in Spanish (and Italian, if we abstract
away from the observations made in Longobardi (2002) concerning the supposed varied
interpretations of bare nominals under modiﬁcation). Since bare plurals in PA and
Spanish, a Romance language, and, to a great extent Italian, behave in virtually identical
ways, these nominals seem to call for the same treatment.M In Longobardi (1994) an
analysis was proposed to account for the distribution and interpretation of bare plurals in
Romance in terms of a structural requirement on the empty determiner introducing these
nominals. Speciﬁcally, it was argued in this work that bare plurals in Romance need to

meet a lexical government requirement. Such a requirement prohibits their occurrence

 

'2 Recall, however, what has just been said in the previous section: it is not clear in Longobardi's system if
+/- Def features and [+article] features are one and the same.

'3 This is in fact the conclusion 1 will argue for in the present work: that bare nominals (singular and
plural) in a language such as Arabic are ungramrnatical unless certain conditions are met. However,
checking of [+article] strong features is not one of these conditions.

" Such a uniform treatment will be argued for in this work in later chapters.

45

in preverbal positions, positions that are not lexically governed. However, in postverbal
positions this requirement is met. I will not go through a detailed exposition of the
issues involved here since this will be dealt with in Chapter 3. For our purposes here,
sufﬁce it to say that bare plurals in PA appear to require the same treatment extended to
bare plurals in Romance in Longobardi (1994).15 In other words, PA bare plurals are
supposedly ungrarnmatical in preverbal positions and grammatical in postverbal
positions. This would presumably follow from the lack of lexical government in
preverbal positions. Since bare plurals are ungrarnmatical preverbally in the framework
of Longobardi (1994), this fact can be recast for PA bare plurals in the current terms of
Longobardi (1996) as follows: bare plurals surface articleless in PA; in other words,
they constitute an instance of article drop. The argument then would proceed in the
following manner: since the empty determiner in PA has a strong [+article] feature that
needs to be checked by moving a noun into D, this would explain the ungrarnmaticality
of bare plurals preverbally; the derivation would not converge since the [+article]
feature has not been checked before spell-out and is not interpretable at LF. Thus
Longobardi's (1996) analysis provides a nice explanation for this observed behavior of
bare plurals in PA (note that the same result would also be achieved through applying
Longobardi's (1994) requirement of lexical government). However, the shortcomings of
Longobardi's (1996) account become evident when we consider the behavior of PA bare
plurals mstverbally. If, within the framework of Longobardi (1994), PA bare plurals are

acceptable postverbally since the supposed lexical government requirement is met, then

 

‘5 However, the present dissertation actually argues against the approach proposed in Longobardi (1994).
So this last statement made in the text is for the sake of argumentation only.

46

we would have to assume that the strong [+article] feature is not 'operational'
postverbally (however that can be implemented), or else we would have to assume that
this strong feature is somehow checked (again, how or by what means is not very clear).
An additional shortcoming of the [+article] proposal is of a conceptual nature. As
observed earlier, bare plurals in PA are similar in interpretation and distribution with
Romance bare plurals. However, Longobardi's account would have to posit two
different ways for accounting for the behavior of bare plurals cross-linguistically:
Romance bare plurals are explained via the restrictions imposed on them by the lexical
government requirement; PA bare plurals, on the other hand, would have to be
explained in terms of the strength of the feature [+article] of the empty determiner.l6

To sum up, if bare nominals in PA have basically the same distribution and
interpretation as the Romance ones it does not seem plausible to bestow on the Arabic
(and Hebrew) empty determiner qualities, such as a null operator, different from those
Romance determiners have. In this light it is not plausible either to claim that N raising
in Romance is an instance of substitution into an empty determiner while in the case of
Semitic it is adjunction to an empty operator (on the analogy with N adjunction to an
overt operator in Romanian). I conclude that this proposal in terms of the strength of the
[+article] feature is untenable.

Notwithstanding this problematic aspect of Longobardi's analysis I still think that
the connection he makes between N raising and the presence of a genitive DP in

(Semitic) construct state is on the right track. I believe that the presence of the genitive

 

'6 Even this option does not work for PA as I have argued above in the text, since postverbally bare
nominals are arguably grammatical in terms of Longobardi (1996).

47

DP in the construct state plays an important role in licensing, in a pre-theoretical sense,
the raising of the head N to D. This fact comes out more strikingly when compared with
the behavior of bare nominals in PA in the general case (in non-CS constructions).
However, before investigating this issue further, I want to spell out more clearly, yet
brieﬂy, the analysis I will be adopting in this work for deriving the Semitic construct

state. This issue is the focus of the next section.

2.5.2 A Working Analysis of the Construct State in Arabic

An analysis that makes N raising parasitic on the presence of the genitive DP
should be theoretically more superior in that it could account for the behavior of bare
nominals in Arabic vis-a-vis their behavior in construct state constructions. In this light
an analysis such as Longobardi (1996), which is a modiﬁed version of Siloni (1994),
and abstracting away from certain problematic details in the Longobardi analysis, seems
to me to have the potential to offer us more explanatory adequacy than that upheld by
Fassi Fehri (1999).

Accordingly, I make the following assumptions-some of which are by no means
novel. 1 will assume, as is standardly the case, that the head noun of the construct state
in PA is generated in a position lower than that of the genitive DP in the base structure
and it moves to a head D position. I will also assume that N raising to D is functionally
motivated, not for Case-checking purposes, but rather to check another feature of the
head D to be determined shortly. The genitive DP will raise to a ﬁrnctional projection
higher than attributive adjectives (and lower than NumP, if the latter projects in the

structure, see § 2.3) and lower than the D projection to which the head N is to raise to.

48

Raising of the genitive DP takes place uniformly in deﬁnite and indeﬁnite construct
state sentences alike. To assume otherwise is not only a loss of a generalization but is
not conﬁrmed by the empirical facts either. The projection to which the genitive phrase
raises is argued to be DP 1 in Fassi Fahri (1999). In Longobardi (1996) and Siloni (1994,
1997) such a projection is assumed to be Ang(enitive)P. For my purposes in this work
I think it is inconsequential what the category label is of the projection which the
genitive DP moves into and would not actually affect the general argument pursued
here. Therefore, for my purposes I will call this functional projection F P (stands for
Functional Projection, which is different and distinct from Focus Phrase/Projection), to
abstract away from any commitment to the type of ﬁrnctional projection it rrright be.
However, I assume, with Longobardi and Siloni, that the genitive DP moves into FP to
check the Case features of the relevant functional head, F, with which it will be in a
Spec-Head structural conﬁguration. I similarly take the Case features of the head F to be
strong in the sense they trigger overt movement of the genitive DP. The resulting

structure for the example in (26) and (27) should be as in (28):

(26) byuut l-walad
houses the-boy
'the boy's houses'

(27) byuut walad

houses boy
'lit. a boy's houses'

49

 

byuut

(28)

I have assumed that the genitive DP raises into the Spec position of PP for Case reasons.
What I have not spelled out yet is what the functional feature motivating the movement
of the head N is. As argued above, this feature cannot be the strong [+article] feature
presumed in Longobardi's analysis. It could not be Case as assumed by Fassi Fehri
either. Intuitively, since I have maintained all along that raising of N has to be thought
of as parasitic on the genitive DP, the feature that 'attracts' the head N has to ﬁnd a
matching feature on the head N bestowed upon the latter by no other than the genitive
DP itself. If we grant that it is the genitive DP that endows the head N with the matching
feature we would be in a better position to explain why in non-CS constructions bare
nominals, not occurring with a genitive DP, cannot raise to D to check the relevant,
presumably strong, features of this head.

Now suppose that the head noun in its base position, i.e. before moving or raising to

50

D, is not speciﬁed for (in)deﬁniteness.l7 In other words, the head noun would be base
generated without + or - feature value for deﬁniteness. This does not seem unreasonable
especially if looked at from a cross-linguistic perspective. Consider, in fact, that in
languages such as Japanese or Russian bare nominals are indeterminate or ambiguous
between a deﬁnite or indeﬁnite interpretation (see, for example, Chierchia (1998)). The
context, in the case of Russian or Japanese, would play the disarnbiguating role.
Chierchia (1998: 361) observes that bare arguments in Russian have a free distribution
and could be read as deﬁnite or indeﬁnite depending on the context. Consider the

following from Russian Chierchia cites:

(29) V komnate byli malcik i devocka.
in (the) room were (a) boy and (a) girl.
Ja obratilsja k malciky.
I turned to (the) boy

(29) illustrates that the deﬁnite/indeﬁnite distinction is neutralized in the absence of the
article and that the context helps disambiguate which reading is intended. According to
Chierchia, the word malcik 'boy' and devocka 'girl' are read as indeﬁnite in such a
presentational sentence. However, the second instance of malcik is read anaphorically to

the ﬁrst and is therefore interpreted as deﬁnite.

 

'7 A similar claim has been made by Borer (1999) in the context of arguing for an incorporation account
of the construct state in Hebrew. Borer argues that the head N of the construct state is not speciﬁed for +/-
Deﬁniteness feature, and since D in Semitic is inherently unspeciﬁed for +/- deﬁniteness, the head N
cannot move to D. Incorporation of the head noun and the genitive phrase becomes necessary since the
latter would supply the head N with the relevant feature value. Among other things, the incorporation
analysis is undermined by the facts argued for in §l.3 in the text: strict adjacency between the head noun
and the genitive DP is not necessary, and sometimes impossible if a NumP projects. See Siloni (1997) and
F assi Fehri (1999) for arguments against the incorporation account of the construct state in Semitic.

51

If we assume that bare nominals are not speciﬁed for the feature value +/-
Defrniteness, it is quite plausible to say, contra Borer (1999), that the determiner
position is where the deﬁniteness/indeﬁniteness speciﬁcations lie. English is a good
example of how this might work since this language possesses two articles, a deﬁnite
and indeﬁnite, and nouns, which occur bare following the determiner, are interpreted
either as deﬁnite or indeﬁnite depending on the occurrence of the relevant article. '8 In
other words, deﬁniteness is a property of the DP, not nouns. I therefore assume that the
D position is endowed with strong +/- Def features that have to be checked overtly
through realizing a lexical determiner (for example, an article) or through movement of
a head N as is the case in the Semitic construct state. My assumption is not identical to
that of Longobardi's (1996), however. Recall that in Longobardi's system it is the strong
[+article] feature that needs to be checked in Semitic.

Keeping what has been said immediately above in mind, the raising analysis can be
implemented in the construct state as follows: supposing that bare nominals in PA
generally lack a +/- Def speciﬁcation, and assuming that D (in Semitic) has a strong +/-
Def feature, this means that the head noun would not have the relevant matching
features to check the corresponding features of D. However, Semitic construct state
constructions are generated with an obligatog genitive DP, and this genitive DP needs
to check its structural genitive Case in a Spec-Head conﬁguration with a Case-assigning
head, F in this case. Consequently, the genitive DP would raise into F P and land in its

speciﬁer position. The genitive DP is of course speciﬁed for (in)deﬁniteness in that it

 

'3 1 set aside for the moment bare nominals in English.

52

either carries the deﬁnite article or is articleless (i.e. indeﬁnite).19 The head F, by virtue
of being in a Spec-Head structural conﬁguration with the genitive DP sitting in [Spec,
FP], would acquire the relevant Def features of the genitive DP. Now, since D has
strong Def features to be checked overtly one option is to realize the deﬁnite article to
check those strong features. However, in this case raising of the head N would not be
functionally motivated or even required/permitted; it is precisely such a circumstance
that gives rise to the observed complementarity between the definite article and the
occurrence of the head N in D.20 The head N could not in this case raise since the
morphological trigger that would set the N raising in motion, namely, checking the
strong Def features of D, would be satisﬁed by the lexical realization of the deﬁnite
article in that position. However, if the lexical article is not realized in D, the strong Def
features of the latter would still remain unchecked which would trigger raising of the
head N to substitute in that position. But recall that nouns in Semitic are not speciﬁed
for Def features in the base position. At this point the genitive DP proves crucial. If the
head N has to adjoin to F on its way to D, as a consequence of the Head Movement
Constraint, then N would inherit the Def feature speciﬁcation of the genitive DP
transmitted to F by the former in a Spec-Head conﬁguration. This is precisely where

(in)deﬁniteness inheritance in the construct state occurs. N would acquire the necessary

 

'9 Recall that there is no indeﬁnite article in PA.

2° Of course, there has to be a way of ruling out a derivation where the lexical deﬁnite article would be
generated in D to check the latter's strong Def feature, thus eliminating the trigger necessary to prompt or
initiate N raising to that position, resulting in the head N staying in situ. In other words, in such a
derivation we would have an article in D and a raised possessor but a head N which is in situ. This
derivation would be ruled out on the principled ground that the deﬁnite article, due to its afﬁxal nature,
would need a host to attach to. Therefore, the derivation article-possessor-N is not a convergent
derivation. The article in D could not attach to the possessor since the latter is sitting in the speciﬁer
position of an FP projection, which is arguably in a complement position to D.

53

features to match the strong features of D.

I believe this analysis, which is minimally different ﬁ'om that of Longobardi (1996)
(and Siloni (1994), as previewed in Longobardi (1996)), successfully accounts for the
observed properties of the construct state in Semitic. The head N occurs initially in the
construction through movement to D. The obligatoriness of the genitive phrase is
expected if raising of the N is parasitic and dependent on it. (In)deﬁniteness inheritance
is a consequence of N's adjunction to the head of the projection in whose speciﬁer the
genitive DP sits. The impossibility of realizing the deﬁnite article on the head N follows
naturally from such an analysis by assuming that the strong +/- Def features of D can be
checked M in one of two ways, but not the two together: either raising of N to D or
lexically realizing the article. In a construct state construction, the head N has to raise
since an article in that position would need to be supported by a host due to its afﬁxal
nature, as explained in fn. 20. However, generating a deﬁnite article in D, and at the
same time raising N to that position, is not possible since the movement of N to D
cannot be initiated in the ﬁrst place due to the absence of the triggering feature in D
(i.e. the strong Def feature of D which would in this scenario be checked by realizing a

deﬁnite article).

2.5.3 Bare Nominals in Non-CS Constructions

One immediate advantage to the analysis expounded above is that it enables us to
account for the question raised earlier in § 2.2.2 above: if bare/determinerless nominals
Can raise to D in the construct state, how can we explain their inability to raise in non-

CS constructions?

54

As noted for example (21a) and (21b) above, repeated here as (30a) and (30b), bare
nominals in non-CS structures exhibit a possibility that has been argued to be
unavailable for CS head nouns, namely, carrying the deﬁnite article. This latter property,
i.e. the impossibility of the deﬁnite article on the head N in D, as noted earlier, is a
consequence of raising N to D in CS constructions. It follows that bare nominals in PA

cannot raise (at least overtly) to D in non-CS structures. (31) also shows that the same

facts hold for singular nominals.

(30) a 2d3-u (1-) wlaad
came-3MP (the) boys
' (The) Boys came.

b. ?d3-u (*l-) wlaad l—dziraan

came-3MP (the) boys the-neighbours
' The neighbors' boys came.’

(31) ?d3-a (l-) walad

came-3M8 (the) boy
' (The) Boy came.

Since I have argued that D has strong +/- Def features that need to be checked in the
syntax, and since bare nominals in PA in the base position lack the speciﬁcation for the
Def feature value, these nouns would not be able to raise since they do not have the
corresponding Def features. If these features are bestowed upon the head N by a genitive
DP in CS constructions, then the head nouns in non—CS constructions have no means of
acquiring the necessary Def features to be able to raise to D due to the absence of the

genitive DP. This analysis makes a strong theoretical prediction: if the head noun in

55

non-CS constructions can somehow acquire the necessary +/- Def features to be able to
check the matching strong features of D, then we predict that the head noun would
actually raise to D even in the absence of a genitive DP. I believe this prediction is

empirically borne out in PA as attested in the following sentences:

(32) beet-o inharaq

home-3M8 bumed-3MS
'His house burned.‘

(33) byuut-hum inharaq-u

houses-3MP bumed-3MP
'Their houses burned.’

First of all, recall that the genitive DP is obligatory in the Semitic construct state. In
addition, it has been argued by Longobardi (1996) that whereas it is possible for the
genitive argument of Romance construct state to be understood, i.e. not realized overtly,
this is not an option for the Semitic construct state.21 Under these assumptions the
sentences in (32)-(33) are instances of non-CS constructions. However, it has been
argued in the literature (e.g. Longobardi (1996)) that the clitics realized on the head
noun in (32) and (33), -o and -hum, respectively, are actually pronominal possessors. In
other words, these examples in (32) and (33) are essentially instances of CS structures,
although the possessor is pronominalized.

To begin to address this point, recall ﬁrst that Arabic (and Hebrew) have a

construction traditionally called clitic-doubling. Notice that (32) and (33) can be clitic-

 

2‘ Longobardi argues that the genitive argument in Romance can be a non-overt Pro, to differentiate it
from either PRO or pro.

56

doubled as is seen in (34) and (35) respectively:

(34) beet-o la l-walad inharaq

home-3M8 to the—boy burned-3 MS
'The boy's house burned.’

(35) byuut-hum la l-wlaad inharaq-u

houses-3MP to the-boys burned-3MP
'The boys' houses bumed'

The clitic-doubled argument in (34) and (35) are l-walad 'the-boy' and l-wlaad 'the-boys'
respectively. The clitic-doubled argument in these two examples are read as genitive
arguments or possessors (comparable to the Free Genitive of Hebrew). Now, if the
pronorrrinalized clitic on the head noun, -0 and -hum, are also possessors, then the
sentences above would conceivably have tlvg possessors realized in the same sentence.
However, no such situation can be found in traditional construct state constructions as

the following examples in (36a) and (36b) attest:

(36) a. * beet l-walad la l-walad/huwwe inharaq
house-3 MS the-man to the-boy/ he bumed-3MS

b. * byuut l-wlaad la l-wlaad/ huwwe inharaq-u
houses-3MP the-boys to the-boys / he bumed-3 MP

As seen in these examples the outcome is ungrarnmatical. Construct state constructions
cannot be clitic-doubled, a possibility still open for sentences with pronominlaized clitic

possessors. This seems to argue against viewing the pronominal clitic on the head N as

57

a possessor.
Another possibility is to regard the clitic as a determiner, as in the English
expression their home. However, the clitic -hum in byuut-hum homes-3MP,’ for
example, could not be analyzed as a determiner or a possessive adjective as in English
their since that would mean that byuut 'homes' has actually raised and occurs in [Spec,
DP] and, moreover, that the raising would be an instance of NP, rather than N raising.

This would incorrectly predict that NP raising could pied-pipe postrrominal adjectives:

(37) a byuut-hum l-wsiiﬁ’a inharqau
homes-3MP the-spacious bumed-3MP

b. *byuut-l-wsii‘r‘a-hum inharqau
homes-spacious-3MP burned-3MP

c. *l-wsiiYa- byuut-hum inharqau
the-spacious-3MP homes I bumed-3MP

It then seems that the clitic on the head could not be a determiner as is the case in

English. However, I believe that an account of such clitics on the head N is available.
Roberts and Shlonsky (1996) and Shlonsky (1997) propose an attractive analysis of

clitics in Hebrew and Palestinian Arabic. They argue that clitics in these two languages

have, among others, the following three characteristics (1996:183):

(38) a. Clitics occur on the right of their host, never on the left.
b. Clitics are always attached to the closest c-commanding head.

0. Clitics in Semitic appear on all lexical categories and on certain functional

58

categories.

These descriptive statements pretty much apply to the clitics under discussion in (3 2)
and (33). The authors above also propose that Semitic clitics are Agreement elements
whose function is to identify a null pronominal pro in argument position as in the

following schema:

(39) X+Clitic pro

(32) and (33) would then have the following structure:

(40) beet-o pro inharaq
home-3M8 burned-3M8
'His house burned.’

(41) byuut-hum pro inharaq-u

houses-3MP burned-3 MP
'Their houses burned.’

It is reasonably the case therefore that the clitic on the head N is an Agr marker whose
purpose is to identify the null pronominal pro. Now note that the head N in these
sentences exhibits the one of the same properties the head N of a traditional construct

state displays, namely, that it cannot carry the deﬁnite article:

(42) *l-beet-o pro inharaq
the-home-3 MS bumed-3MS

59

Also notice that the head noun carrying the Agr clitic in (40) and (41) is interpreted as
deﬁnite as attested by deﬁnite agreement on the modifying adjective in (43) and the

deﬁnite relative clause in (44):

(43) beet-o pro l-gadiim inharaq
home-3M8 the-old bumed-3MS
'His old house burned.‘
(44) byuut-hum pro ?illi dahanuhum mbarih inharaq-u

houses-3MP that painted-3 MP-3MP yesterday burned-3MP
'Their houses which they painted yesterday burned.‘

These observations can be taken to mean that the head N in (40) and (41) has raised
mix, as the case is in traditional construct state, to D. However, this possessive
construction, despite being very similar to the traditional construct state, should not be
treated as an identical phenomenon since an pier; genitive DP is obligatory in the latter.
This state of affairs argues against the assumptions contained in Longobardi (1994).
In this work, it is argued that overt movement to D is only possible for proper nouns at
S-Structure. Common nouns, on the other hand, are forbidden from raising and
substituting into D overtly, although they can adjoin to an overt operator in D as is the
case in Romanian. Raising of proper nouns at S-Structure in Italian/Romance is
assimilated with the raising of generic/kind-denoting common nouns at LF in English. If
an empty determiner is interpreted as a default existential operator, then generic/kind-
denoting common nouns have to raise, at LP, to substitute for the empty determiner at
that level and thereby preventing the noun from being interpreted existentially. In other

words, what happens in Italian at S-Structure in the case of proper nouns happens in

60

English at LP in the case of generically interpreted common nouns.

Longobardi's assumptions noted immediately above predict that if raising of a
common noun takes place overtly at S-Structure in any language, thus substituting into
an empty D, then that would preclude the existential reading for the common noun and
would mean the o_n_ly reading available for such a noun is a kind reading. However, this
is not what happens in PA. The common nouns in (40) and (41) are not interpreted
generically. In fact, they can only be interpreted existentially.

Raising of possessively interpreted common nouns to D at S-Structure is very
productive in PA and is quite lexically generalized. Raising of proper nouns to D is also

attested (in fact, obligatory, in the absence of a deﬁnite article as is the case in Italian):

(45) l-urdun l-dsamiil
the-Jordan the-beautiful
'the beautiful J ordan'

(46) faransa l-hadiiGa
France the-modern

(47) *l-faransa l-hadiiGa
the-France the-modem

As can be seen in (45)-(47) whether the article is attached to the proper noun or not, i.e.
whether the proper noun has raised into D or not, the proper noun is understood as being
deﬁnite as attested by the deﬁnite article marked on the modifying adjective. As a
matter of fact, even proper nouns such as l-urdun '(the-) Jordan,‘ which normally carries

the deﬁnite article, have to give up their deﬁnite article if they occur in possessive

61

sentences with Agr clitics:

(48) urdun-na l-d3amiil
Jordan-1P the-beautiful
'our beautiful J ordan'

(49) *l-urdun-na l-dgamiil
the-jordan-3P the-beautiful

However, we are still left with this nagging question: what enables the bare nominal to
raise to D in sentences that lack an overt genitive DP? As an answer to this question

consider ﬁrst the structure of (41), repeated here, in (41a):

(41) byuut-hum pro inharaq-u
houses-3MP burned-3MP
'Their houses bumed'

(4la)

 

 

 

 

62

Assuming that the Agr clitic -hum is generated in head position in the F P projection
(perhaps AgrP, as in Siloni (1994, 1997), Longobardi (1996), also Shlonsky (1997)),
pro would conceivably be generated in [Spec, FP] since the Agr clitic is there in the
ﬁrst place to identify the null pronominal pro. N would left-adjoin to F pied-piping -
hum with it and inheriting the deﬁniteness features from the latter. I think the preceding
discussion does not commit us to any conception of the null pronominal pro as a null
possessor (on analogy with Pro supposedly attested for Romance as discussed by
(Longobardi ( l 996)).

In the next section I will turn to the issue of applying such notions concerning the
strength of the Def features of the D position to English. I will argue that this notion
may in fact be carried over to English and thus afford us a more restrictive picture of the

surface differences between languages as different as PA and English.

2.6 Bare Nominals in PA and English

I have argued above that PA nominals are not speciﬁed for +/- Deﬁniteness feature
value. I have also maintained that D in PA has strong +/- Def features that need to be
checked in the syntax. In the general case, expectedly, bare nominals can not raise to D
overtly since they do not have the necessary corresponding +/- Def feature speciﬁcation
to check the relevant strong features of D.22 In the construct state, on the other hand,
bare nominals, unspeciﬁed for Def features at the base, can acquire the necessary

speciﬁcation for the Def feature value, needed to check D's strong features, by adjoining

 

22 'Bare nominals' are to be understood here and elsewhere to mean DPs without an overt or lexical D.

63

to a functional head, labeled F above, that is in a Spec-Head conﬁguration with a
deﬁnite/indeﬁnite genitive DP.

One consequence of this analysis is the following: bare nominals in PA non-CS
constructions, whether singular or plural, are predicted to be ungrarnmatical in argument
positions anywhere in the sentence. This is so because the strong features of the D head
of the DP which contains the bare nominal would go unchecked resulting in the
derivation non-converging.23 My claim is exactly that. This, obviously, goes against
arguments leveled for Romance in Longobardi (1994, 1996) where it is claimed that
bare nominals are ungrarnmatical preverbally, but not postverbally. I will also make the
claim that PA bare nominals are ungrarnmatical under normal conditions of stress. In
other words, they are always focused. These issues will be taken up again in greater

details in Chapter 3. For now, however, I will assume that my claims hold true.

2.7 English Bare Nominals and (In)definiteness Feature
Checking

PA nominals are ungrarnmatical in non-CS constructions since they are unspeciﬁed
for Def features and therefore unable to check the strong D Def features. Could this
situation be detected in a language such as English?

The answer to this question is, I think, positive. As a matter of fact, this seems to me
to be the null hypothesis. Note ﬁrst of all that English generally lacks bare singular

count nouns in argument position. Examples such as the following are ungrarnmatical:

 

23 Crucially, 1 am assuming here that only DPs can be arguments as argued by Longobardi (1994) and that
bare nominals in PA are contained in a DP shell.

(50) * Boy came

(51) "' I saw boy.

These examples contrast with the fully grammatical English sentences in (52) and (53):

(52) A/The boy came.

(53) I saw a/the boy.

The pattern in (50) and (51) is presumably attested in PA since articleless singular
nominals are possible in this language. However, recall that my claim here is that the
English pattern seen in (50) and (51) is also ungrarnmatical in PA under normal
conditions of stress. As is witnessed in (52) and (53), the English sentences in (50) and
(51) become grammatical once the D position is ﬁlled by a deﬁnite/indeﬁnite article.
Within the system of assumptions argued for here, and taking all the observations above
concerning examples (50)-(53) in tandem, this could mean that English D also has
strong (in)defmiteness features that need to be checked in the syntax.24

Longobardi (1996) argues that Romance and Germanic exhibit traces of the Semitic
construct state, overtly in the former and at LF in the latter. The head noun in the
English Saxon genitive in (54), Longobardi argues, raises at LP to D and the genitive
DP would occur in the Spec position of AngP (F P, in our terms). So, (54) would have

the structure in (55) which is essentially the PF structure of the Semitic construct state

 

2' This could be recast in the terms of Longobardi's (1996) analysis in terms of the strength of the
[+article] features. However, this approach has been showed to be inadequate for PA.

65

(his (81b) and (81a) respectively):

(54) The teacher's new book .....

(55) Book the teacher's new t about syntax

If Longobardi's analysis is on the right track then the unglammaticality of (56) is

predicted:

(56) *The teacher's a/the (new) book

The ungrarnmaticality of (56) is therefore parallel to the ungrammaticality of the
Semitic equivalent head of the construct state with a deﬁnite article with the obvious
difference that in Semitic raising of the head N occurs overtly. If bare singular count
nominals in English are not speciﬁed for +/- Def features either then one way of
implementing the correctness of (55) is to say that the head noun, unspeciﬁed for +/-
Def features as it is, acquires the deﬁniteness/indeﬁniteness features of the genitive DP
the teacher (which would be in Spec AngenP at LF, according to Longobardi's
analysis). This is further attested in the grammaticality of (57)-(58) and the
ungrarnmaticality of (59) (notice that, interestingly, book in (57) and (58) is interpreted
as indeﬁnite and deﬁnite, respectively, depending on the (in)deﬁniteness of the genitive

DP):

66

(57) a. a student's book about syntax
b. book a student's t about syntax

(5 8) a. the student's book about syntax
b. book the student's t about syntax

(59) a. *student's book about syntax

b. #book student's t about syntax

The grammaticality of (5 7) would follow directly from the fact that a student is
speciﬁed for the indeﬁniteness features in this case; therefore, book, moving covertly to
the head position D, would acquire the necessary indeﬁniteness features necessary to
check the Def features of D. Therefore, the sentence in (57) is predicted to be
grammatical. The same analysis can be carried over to (58). In this case, however, it is
the deﬁniteness features that get passed on to the bare nominal book. Once the head N is
speciﬁed for the relevant Def features it can covertly raise to D to check the
corresponding feature of the head D. (59) therefore is expectedly ungrarnmatical since
the genitive DP student, being deprived or unspeciﬁed for Def features, cannot pass
those on to book to enable the latter to raise and check the Def features of the head D. It
follows that the LF derivation in (59b) is unacceptable and the S-Structure (59a)
crashes.

Now compare this pattern witnessed in (57)-(59) to the following examples with

existentially interpreted bare plurals in English:

(60) *students' book(s)

67

The ungrarnmaticality of (60) is in line with the general argument put forth so far. If
English nominals, just like their PA counterparts, are not speciﬁed for the +/- Def
features, then the ungrarnmaticality of (60) is in fact two-fold: ﬁrst, LF-raising of
book(s) to D is not licensed since this nominal is not speciﬁed for the Def features in the
base position and it cannot acquire the necessary features from the genitive DP which is
itself unspeciﬁed for Def features. This brings us to the second source of
ungrarnmaticality of (60): being a bare nominal itself students cannot raise o_v_e_rt_ly to
check the strong Def features of the D head in the DP shell in which it is contained.
Either way, the derivation cannot converge. However, if the strong features of D of the
DP containing students were to be checked by another means, e. g. realization of the
lexical deﬁnite article the, or a determiner such as those, the derivation would be

expected to converge which is the case in (61 ):

(61) these/the students' book(s)

In (61) the +/- Def features of the head of the DP containing students are checked by
either these or the. The other source of potential ungrarnmaticality is eliminated since
now book(s) can raise to D at LF because it can acquire the necessary features from the
(deﬁnite) genitive DP these/the students sitting in the speciﬁer of the AngP projection.

The discussion so far makes this prediction: if the English D has strong +/- Def
features then bare plurals should not be acceptable in English, under normal conditions
of stress, as in the case of PA. I believe this prediction is also conﬁrmed. As will be

argued in the following chapters, bare plurals (that are existentially interpreted) are not

68

possible in English anywhere in the sentence under normal conditions of stress.25 In
other words the following examples containing existentially read bare plurals are

unacceptable if not focused, a position I argue for in this dissertation:26

(62) * Dogs are in my garden.

(63) * I can see spots on the ﬂoor.

A question immediately arises: if D in English has strong +/- Def features that need to
be checked why are generically interpreted bare plurals in English acceptable under

normal conditions of stress? Consider the following examples:

(64) Dogs are common in North America.
(65) Men smoke more heavily than women do.

(66) Women's place in society

As is clear from (64)-(66) generically interpreted bare plurals in English are acceptable.
To phrase the problem in more technical terms, generically interpreted bare plurals in
English can somehow satisfy the requirement that the strong +/— Def features of D be
checked. But how can these plural nominals achieve this?

Two potential solutions suggest themselves at this point. As will be argued in

 

2’ Diesing (1992), Kratzer (1995), and Longobardi (1996) assume that existentially quantiﬁed bare
nominals in English are impossible only outside of the VP.

2" Of course it remains to be addressed precisely how focusing a bare plural can exempt bare plurals ﬁ'om
violating the requirement that the D in their DP shell needs to have their features checked. I will return to
this issue in Chapter 5.

69

 

Chapter 4, generically interpreted bare plurals in English are distinct from existentially
quantiﬁed bare nominals in being non-focal. It is then reasonable to assume that their
behavior would diverge from that of existentially read bare nominals. Suppose, in
accordance with what has been suggested by Rizzi (1997) (see Chapter 5), that
generically read bare nominals carry a topical feature that needs to be checked in the
proper conﬁguration. Suppose, then, that generically interpreted bare nominals are
structurally located in the speciﬁer position of a TopicPhrase and the DPs in which
these nominals are contained enter into a checking relation with the head Top of that
projection. This way, these nominals would be able to be licensed and also interpreted
(as topics, presumably, deﬁnite). In Chapter 5, I argue that the same analysis can be
assumed for focused existential BNs, with the obvious difference that the latter enter
into a checking relation, in Speciﬁer-Head conﬁguration, with a head F of a Focus
Phrase.

Another solution could be the following: Longobardi (1994) argues that generically
interpreted bare nominals in English raise at LP to ﬁll D, a position I have argued is
problematic for languages such as PA where overt raising to D is attested. These
problems aside, suppose, then, that raising to D of a common noun is a necessary, but
crucially, not sufficient condition for the generic/kind interpretation. Now, recall that
generically interpreted bare plurals in English are read as (names) of kinds (as argued in
Carlson 1977, and Longobardi (1994)) which assimilates them with proper nouns.
Assume farther that generically interpreted bare plurals in English can raise to D at LF,
thus satisfying the feature checking requirement of the D position. Existentially read

bare plurals, on the other hand, have to satisfy this requirement in the syntax since they

70

do not have the kind reading necessary to raise to D.

2.8 Conclusion

I have argued in this chapter that we can gain a good understanding of the behavior of
bare nominals in PA by investigating the properties of the Semitic construct state. The
analysis I have adopted of the construct state is basically a modiﬁed version of the
analysis adopted in Longobardi (1996).

I have attempted to demonstrate that if we assume that bare nominals in PA are not
speciﬁed for the +/- Def feature, and if the D position in this language is equipped with
such a feature that is strong and in need of overt checking, we can gain a working
understanding of the reason why bare plurals in non-CS constructions cannot raise to D
overtly. I have also argued that this restriction on bare nominals in PA can also be
detected in English. If an analysis such as Longobardi's (1996) is correct, parameterizing
and extending the Semitic genitive structure to English, the system of assumptions
argued for here could reasonably be applied to English. Such an analysis, it seems to
me, may be able to reduce the surface differences noted between PA and English bare
plurals and achieve a higher degree of cross-linguistic uniformity.

In the next chapter I offer a more uniform view of bare nominals in PA and Spanish
where I argue that these nominals are always focused. I also extend the reach of this

analysis to English existential nominals in Chapter 4.

71

Chapter 3

Bare Nominals in PA: Distribution and Interpretation

3.0 Introduction

I have argued in Chapter 2 that to understand the overall behavior of bare nominals
in Palestinian Arabic it may be necessary to understand the structure of the Determiner
Phrase (DP) in the language. To achieve this goal, I have singled out a construction that
is both prevalent and widely investigated in the Afro-Asiatic literature, namely, the
Construct State (CS). A closer examination of the Palestinian Arabic construct state has
revealed an idiosyncratic behavior of bare (or, rather, determinerless) nominals, which
can freely and productively occur in this construction, both in the singular and plural
forms. Such behavior of bare/determinerless nominals as seen in CS-constructions is
perspicuously different from that generally observed of bare nominals in this language.

Most intriguing about the behavior of bare/determinerless nominals in the construct
state is their ability to substitute into the head D position of the possessive structure. In
this respect, bare nominals in non-CS constructions differ distinctly in that they are
unable to move into the D position. This distinction has proved both crucial and
illuminating; if we assume that bare nominals in CS constructions are somehow
endowed with certain features that make it possible for them to substitute into D, and
further assume that these features are 'inherited' or passed on from the possessive DP or
possessor, then we can explain the failure of non-CS bare nominals to substitute into the
determiner head D. These features, I have claimed, are (in)deﬁniteness features. Bare

nominals are devoid of the (in)deﬁniteness feature speciﬁcations in the general case.

72

The D head has, in Minimalist terms (a la Chomsky 1995), strong +/- Deﬁniteness
features that need to be checked overtly by N, that carries the matching feature
speciﬁcation- a possibility unavailable for bare nominals in non-CS constructions as
they arguably lack such feature speciﬁcation.

If bare nominals in Palestinian Arabic (PA) are unmarked for (in)deﬁniteness
features, it may be surprising, both on conceptual and empirical grounds, that the same
should not hold for other languages, English in particular. In Chapter 2 I have claimed
that such a state of affairs can be argued to hold true of English bare nominals as well.
To the extent that this claim is correct this would provide us with a truly minimalist, in
the pre-theoretical sense of the word, and a desirably restrictive account of the apparent
differences that exist between the two languages, Palestinian Arabic and English (or,
perhaps, language families). However, the parallels between Palestinian Arabic and
English can only be pushed so far since there are indeed conspicuous differences in the
distribution and interpretation of bare nominals in these two languages. These
differences, to be discussed in more detail in chapter 5, I wish to claim to be explicable
as originating from deeper and more principled differences, in part, in the 'focus system,’
as it were, and word order facts, of these two languages (or, again, language families).
This issue is the topic of Chapter 5.

The present system of assumptions naturally leads to the following prediction: being
unmarked for the relevant (in)deﬁniteness features non-CS bare nominals in PA should
be unacceptable, irrespective of their structural position in the sentence. A priori, the

same should also be said about bare nominals in English (excepting, of course,

generically interpreted bare plurals; cf. Chapter 2). This is in fact the position I would

73

like to argue for in the present work. This line of argtunent, however, ﬂies in the face of
various analyses advanced in the literature to explain the apparently idiosyncratic
distributional facts of bare nominals (for example, Longobardi (1994, 1996, 2000, 2002)
for Italian, Contreras (1986) and Casielles-Suarez (1997) for Spanish).

In this chapter I propose to analyze the distribution of bare nominals in PA (and
Spanish) in terms of the information structure of the sentence. This analysis is not,
however, entirely unconventional since a somewhat similar line was proposed for
Spanish by Sufier (1982). I wish to argue that a (purely) syntactic analysis leaves many
important questions unanswered concerning the distribution of bare nominals. In
addition, these analyses undesirably posit a disjunction in accounting for the distribution
of these nominals. As a common theme, some of these syntactic analyses make the
arguably unmotivated assumption that there exists a structural asymmetry in the
distribution of bare nominals: these nominals, the proponents of these analyses argue,
can occur postverbally in the sentence while their preverbal occurrence is far more
restricted so as to be impossible under normal conditions of stress (i.e. unfocused). The
proposal 1 advance here diverges in very important ways from such analyses by
completely dispensing with this alleged asymmetry in favor of a more uniform approach
to explaining the distribution of bare nominals. To be speciﬁc, I argue that bare
nominals are only possible if focused regardless of their structural positioning in the
sentence. More precisely, bare nominals in postverbal positions display the same
ungrarnmaticality under normal conditions of stress (i.e. unfocused) as do their
preverbal counterparts.

I begin in § 3.1 by introducing the basic facts regarding the distribution of bare

74

nominals cross-linguistically. § 3.2.1 reviews a popular proposal advanced by
Longobardi (1994) concerning a structural requirement bare nominals (in Romance and
Germanic) are allegedly subject to, namely, lexical government. I demonstrate in § 3.2.2
that Longobardi's proposal is not without problems and that it leaves much to be
desired. Speciﬁcally, Longobardi (1994) is unable to explain how focused preverbal
bare nominals can circumvent the lexical government requirement. Moreover, he is
forced, within the framework of assumptions he makes, to regard the apparent effects of
coordination and modiﬁcation claimed to hold in the literature of the distribution of bare
nominals as exceptions or violations of the generality of his analysis. This latter issue is
taken up in § 3.2.2.2. and the former in § 3.2.2.3. In § 3.3 I consider another proposal
made by Casielles-Suarez (1997) as an attempt to explain the behavior of Spanish bare
nominals. Since Casielles-Suarez 's approach depends primarily on the conceptual and
empirical validity of Diesing's (1992b) Mapping Hypothesis, I review the implications,
and the correctness, of the Mapping Hypothesis on the interpretative distribution of bare
nominals. In § 3.3.2.1 and § 3.3.2.2 two main sets of objections to the mapping
hypothesis are considered: those contained in Reinhart (1995) and Pinto (1997),
respectively. The conclusions arrived at from examining these two works point to the
inadequacy of the mapping hypothesis, theoretically and empirically; §3.3.3 is devoted
to such an endeavor. In § 3.4 a uniform account of bare nominals in PA is proposed. As
has just been noted, the major highlight of this account is that bare nominals in PA,
which can only be read existentially, are always focal. Such an account can also be
argued to hold for singular bare nominals in PA, which is the topic of § 3.4.1.5. Within

such a framework as the one I propose here coordination and modiﬁcation effects have

75

no role to play and as such do not constitute 'violations' or 'exceptions' as in Longobardi
(1994). Finally, § 3.6 and its subsections are concerned with the interpretation of bare
nominals in PA. I consider three proposals, which are not necessarily exclusive of one
another, that have been put forth in the literature to explain the interpretation of

indeﬁnites. § 3.6.4 is the conclusion.

3.1 The Basic Facts

The consensus in the literature seems to have taken for granted the following
descriptive generalization: the distribution of bare nominals (henceforth, BN(s)) (in
Romance)1 is far from uniform in that these nominals can occur only postverbally,
while their preverbal occurrence is ruled out under normal conditions of stress. Many
attempts have been made to explain this supposed asymmetry, most notably Longobardi
(1994, 1996, 2000, 2002) for Italian, Suﬁer (1982), Contreras (1986), and Casielles-
Suarez (1997) for Spanish. The following examples, taken from Longobardi (1994)
illustrate this alleged asymmetry in Italian (These are Longobardi's examples14a, 14b,

14c):

(1)*Acqua viene gil‘l dalle colline (Italian)
water comes down from the hills

(2) Viene gil‘l acqua dalle colline
comes down water form the hills

(3) Ho preso acqua dalla sorgente
I took water from the spring

 

' To my knowledge bare nominals in (Palestinian) Arabic have not been dealt with previously.

76

According to Longobardi (1994), and as seen in the contrast between the minimally
different (1) and (2), the BN acqua 'water' occurs preverbally in (1) which results in an
ungrarnmatical outcome. In (2), on the other hand, the BN is postverbal, hence the
grammaticality. The acceptability of (3) is parallel to the acceptability of (2) in that the
BN occurs within the VP. Descriptively, it appears that the BN in a preverbal position
fails to satisfy a requirement, structural or otherwise, that the postverbal BN is able to
satisfy. Notice that this requirement cannot be equated with a subject/object asymmetry
that certain syntactic structures are sensitive to since BNs can apparently occur
postverbally both as subjects and objects.2 Whatever is responsible for the difference in
grammaticality between (1 ), on the one hand, and (2)-(3), on the other, seems to be tied
closely to the structural position of the BN in the sentence (i.e. pre- vs. post-verbal).
This last observation has tempted researchers, Longobardi included, to attempt an
explanation of these facts on syntactic grounds. Such an attempt does not seem
implausible a priori for parallel asymmetries have been observed in the syntactic
literature. For example, Standard Arabic displays a curious pre-lpost-verbal asymmetry
in agreement in terms of b—features where the verb agrees fully with the preverbal
subject in gender and number. When the subject is in a postverbal position the verb
exhibits an impoverished agreement pattern with the subject and the verb can only carry
gender agreement (see for example, Aoun et al. (1994, 1999), Munn (1999)). Insofar as

these agreement facts can be explained structurally, proponents of a syntactic approach

 

2 Actually, Longobardi's claim would have more merit if the alleged asymmetry in the distribution of BNs
could be reduced to a subject/object asymmetry. This would be so since, as I point out in § 3.2.2.1, the
notion of government Longobardi assumes seems to be so tight as to demand a head-complement
structural conﬁguration between the governing head and the BN. Such a conﬁguration is evidently absent
in the case of subject BN5. See § 3.2.2.1 for details.

77

to the distribution of BNs in Romance can indeed be tempted to pursue a similar line of
inquiry. However tempting or appealing such a strictly syntactic approach to this
problem may be, I believe it is ultimately incapable of explaining certain observed facts
in the behavior of BNs about which I will have more to say below. In addition, such an
approach is indeed oblivious to the signiﬁcant contribution the information structure of
the sentence has in the distribution and interpretation of BNs so as to ultimately affect
the (un)acceptability of the relevant sentence. The signiﬁcance of information structure
is most evident in the widely attested, though ill-explained, observation that BNs
become legitimate preverbally in the sentence if they are focused. It seems to me that
any adequate theory of the distribution and interpretation of BNs has to have the
explanatory power to note and provide an account for such an observation. I wish to
show that the available theories fall short of this goal.

For concreteness, I consider in what follows two attempts that have been made in
this regard: Longobardi's (1994) work on Italian BNs and Casielles-Suarez (1997) on
Spanish. Both these researchers propose analyses that are more or less syntactic in
nature, more so in the case of Longobardi than in the case of Casielles—Suarez. I show
that such analyses fall short of the desired level of explanatory adequacy, however

appealing in their simplicity these analyses may be.

3.2 BNs and Government
As noted immediately above, to explain the alleged asymmetry in the distribution of
BNs in structural terms has a certain degree of appeal. Contreras (1986) was perhaps

one of the ﬁrst to take notice of the importance of syntactic factors and to claim that

78

they are instrumental in explaining the distribution of BNs in Spanish. I will only very
brieﬂy touch on Contreras' analysis since it does not differ in any substantial ways from
the more articulated approach of Longobardi (1994).

What Contreras (1986) offers is basically an analysis of the distribution of BNs
based on the Empty Category Principle (ECP). He assumes that Spanish BNs have the
structure [Np [QP e] N'] whose distribution is sensitive to proper government of the empty
quantiﬁer phrase (1986: 26). However, since the preverbal subject position is governed
by Inﬂ in Spanish, Contreras suggests, preverbal subjects in Spanish have to be
adjunctions rather than occurring in speciﬁer positions. Contreras’ claim is that if the
subject position was governed in Spanish examples such as (4) (his 6) would be

predicted to be grammatical, contrary to fact:

(4) * [s [NP [or e] café] [NFL [VP me gustall
‘1 like coffee.’

Inﬂ, Contreras argues, governs the NP and K-governs3 the QP which would incorrectly
predict the sentence to be grammatical (1986:38). However, if, Contreras argues,
subjects in Spanish are adjoined then the ungrammaticality of (4) is given a natural
explanation in terms of the ECP.

It is fairly clear from this brief overview of Contreras' analysis that government of

the preverbal subject position by Inﬂ, be it lexical government (in the sense of Rizzi

 

3 Contreras gives the deﬁnition in (i) of K(ayne)-govemment:
(i) A lexical category governs its sisters and the categories immediately dominated by its sisters.

79

1990) if Inﬂ is assumed in Spanish to be a lexical governor, or not, is absent in the
preverbal domain. This assumption on Contreras' part causes him to assume further that
preverbal subjects in Spanish are cases of adj unctions, rather than speciﬁer positions.

In a similar vein Longobardi (1994) argues for the existence of a structural
requirement to explain the distribution of BNs in Italian. Longobardi suggests that the
operative notion of government relevant to BNs is strictly government by a lexical head.
Therefore, he does not have to claim, as Contreras does, that (preverbal) subjects are

instances of adjunction. In the next section I take a closer look at Longobardi's analysis.

3.2.1 The Lexical Government Approach

Longobardi (1994) is an attempt to demonstrate that there exists an N to D
movement in Romance. Insofar as this can be shown to hold the DP hypothesis of
Abney (1987) would turn out to be the winning hypothesis and the NP would be shown
to be in complement position to a head D in a DP shell.

In order to achieve this end, Longobardi argues that BNs in Italian are DPs, rather
than NPs. To begin with, _b_ar_e (i.e. determinerless) singular nouns are impossible in
Italian in argument position, either preverbally or postverbally, as seen in (5) and (6)
below. However, in non-argument positions, these nominals turn out to be acceptable as

(7) illustrates (The examples are Longobardi's (6a, 6b, and 7a, respectively)).

(5) *(Un/Il) grande amico di Maria mi ha telefonato
(a/the) great ﬁiend of Maria called me up

(6) Ho incontrato *(un/il) grande amico di Maria ieri.
I met (a/the) great ﬁiend of Maria yesterday

80

(7) Caro amico, vieni a trovarmi
dear ﬁiend come to visit me

The same cannot be said of all bare nominals in Italian, however. Bare plurals and
singular mass nouns can appear articleless/determinerless in argument position as in (8)

and (9) (Longobardi's 123 and 12b):

(8) Bevo sempre vino.
I always drink wine

(9) Mangio patate.
Ieat/am eating potatoes

However, the occurrence of bare plurals and mass nouns in Italian, Longobardi points
out, is subject to a structural requirement in that their (un)grammaticality is determined
by whether they occupy a pre- or post- verbal position in the sentence. This is attested in

the contrast seen in examples (1)-(3) above, repeated here as (10)-(12):4

(10)*Acqua viene gil‘l dalle colline (Italian)
water comes down from the hills

(11) Viene gil‘l acqua dalle colline.
comes down water from the hills

(12) Ho preso acqua dalla sorgente.
I took water from the spring

 

4 We will see in § 3.3.1 the kind of evidence Longobardi provides for the DP status of BNs in Italian.

81

This contrast in grammaticality according to whether the BN is in pre- or post-verbal
position is reminiscent for Longobardi of a structural requirement empty categories in
the general case, and empty complementizer heads in English in particular, are subject
to, namely, lexical government (1994: 616). Longobardi takes this empty category,
which is presumably in need of lexical government, to be an empty D and claims further
that it is to be understood as an existential operator imposing a plurality (count or mass)
constraint on the head noun, hence the ungrarnmaticality of bare singular count nouns in
Italian. 5 Longobardi (1994:617) sums up this state of affairs in the following

generalizations concerning BNs in Italian:

(13) Empty determiners may occur at S-Structure in Italian only under the following
conditions:
a. They are restricted to plurals or mass head nouns like several other
determiners.
b. They are subject to a lexical government requirement like other empty heads.
c. They receive an indeﬁnite interpretation corresponding to an existential
quantiﬁer unspeciﬁed for number and taking the narrowest possible sc0pe

(default existential).

If BNs in Italian are headed by an empty D that needs to be lexically governed, this begs

 

5 Recall from Chapter 2 that this situation does not hold for PA since singular count nominals occur bare.
If we assume that the singular bare nominal in PA is also introduced by an empty D, an assumption both
reasonable and desirable at least conceptually, then the empty D in PA does not force such plurality
requirement on the N it quantiﬁes over.

82

the question why proper nouns in Italian are not subject to such a requirement especially
since they can occur determinerless at S-Structure. The situation at hand naturally leads
Longobardi to the conclusion that determinerless proper nouns have to substitute into D
at S-Structure to avoid violating the lexical government constraint. The details of the
analysis are not relevant at this point. What is relevant in this context is to what extent
the lexical government analysis holds and what problems it seems to give rise to or
leave unresolved. My contention is that this analysis leaves much to be desired to which

I turn directly.

3.2.2 Problems with the Lexical Government Approach
3.2.2.1 The Nature of the Lexical Government

As Longobardi himself notices (1994: 117-fn. 11), the lexical government
requirement null determiners are allegedly subject to is such that a loose deﬁnition of
government is ruled out. To be more precise, the government relation needs to be tight
enough so as not to allow governing the empty head across a small clause boundary as

in (14) which parallels the impossibility of governing an empty C in English as in (15):

(14) *Consideravo studenti intelligenti
Iconsidered students intelligent

(15) I expected *(that) John is intelligent (to be) an obvious truth.

Longobardi concludes that this lexical government has to be satisﬁed by a head-

complement relation holding between the lexical head and the BN, respectively. While

83

this may be the correct generalization for examples such as (8), repeated here as (16),

(16) Bevo sempre vino.
I always drink wine

where the BN is in complement position to the lexical head, or even in examples with
unaccusative verbs where the subject is arguably generated in object position, lexical
government seems to be far less straightforward when the BN is in the subject position
of a sentence with a transitive verb with a VOS order. Consider the VOS order example

in (17) from PA:

(17) banu 1-?ahraamat Sfabiid

built-3P the-pyramids slaves
'slaves built the pyramids'

It may very well be safe to assume that the counterpart of (17) is also acceptable in
Romance, at least in Spanish where the VOS order is attested (see for example,
Zubizarreta (1998), Ordor‘iez (1998)). Whether we assume that the VOS order is
generated via right-adjunction of S to the VP (as in Sufier (1994) and Torrego (1984)
(cited by Ordoﬁez (1998)), or by moving the object to a speciﬁer position higher than
the subject but lower than the verb as Ordoﬁez (1998) proposes, or even as in

Zubizarreta (1998) who argues that VOS order is derived from VSO by moving VP; that

84

contains the object to left adjoin to VP, stranding the subject,6 it is by no means clear
how lexical government obtains between the verb and the subject, let alone in a head-
complement relation.

To sum up, the lexical government requirement may well hold between a lexical
head and its BN object in complement position. However, this is by no means an
unproblematic assumption in the case of BN subjects, especially in the VOS order. To
implement such an assumption would require Longobardi (1994) to qualify his analysis

in ways that might make such an analysis lose its explanatory power or generality.

3.2.2.2 Modiﬁcation and Coordination of EMS
It has been claimed that BNs may show up preverbally in Spanish (and also Italian)
if modiﬁed or conjoined as in the following examples from Suﬁer (1982:226-228) (her

55a, 60a):7

(l8) Hombres 131 no deberian existir
‘Men like that ought not to exist.’

(19) Cubiertos, platos y vasos aparecieron como por milagro
“Flatware, plates, and glasses appeared as if by miracle.’

 

6 Zubizarreta argues that in Spanish VOS order is derived from VSO by moving the VP2 that contains the
object to left adjoin to VP. as in (i):
(0 [TP me regalo [ml m la botella dc vinoi lei [V2 eillli [vm Maria [Vi [esllll]

to-me gave the bottle of wine Maria

7 Sufler points out (1982:227) that the modiﬁed BNs in preverbal position are somehow contrastive. So, in
(18) in the text she explains that the sentence does not refer to men in the general sense, but to a subset of

men who happen to be 'like that.‘ This seems to me to be in line with the general analysis I argue for in this
Work.

85

Longobardi (1994: 618, fn. 12) makes the same observation. He acknowledges that the
generalizations in (l 3) above posited for BNs do in fact allow room for exceptions. To
start with, he notices that BNs can occur preverbally if modiﬁed by an AP, PP, or a

relative clause (however, he notes that this is acceptable at an elevated stylistic level) as

in (20)-(22):

(20) Meravigliose foreste/Foreste meravigliose si aprivano davanti ai nostri occhi.
beautiful forests opened in ﬁ'ont of our eyes

(21) Ragazze delle pil‘l varie origini affollavano i marciapiedi
girls of most varied origins crowded the sidewalk
tra Rue St. Denis e Boulevard Sébastopole
between Rue St. Denis and Boulevard Sébastopole

(22) Ragazze che Gianni non aveva mai visto affollavano i maciapiedi tra...
girls that Gianni had never seen crowded the sidewalk between...

Longobardi also says that BNs may be acceptable in post-copular argument position if

modiﬁed:

(23) La causa delle rivolte sono spesso marocchini che non voglioni tonare a casa
the cause of the riots are often Moroccans who don't want to go back home

Coordinated BNs also violate Longobardi's generalization in (13b):

(24) Cane e gatto erano gia addormentati
dog and cat had already fallen asleep

86

It is important to note that these examples stand as violations to the generality of
Longobardi's claims for his analysis. He does acknowledge the status of these examples
as exceptions in the language, however marked they may be. In the footnote he suggests
that perhaps modiﬁcation of a BN endows the D head that is empty with certain features
(which he does not specify) that allow it to circumvent or satisfy the lexical government
requirement. As for coordination Longobardi admits that this remains a mystery under
his account.

Note ﬁrst that to acknowledge that these examples are exceptions to the generality
of his analysis, Longobardi has unavoidably weakened his position somewhat. It would
certainly be more advantageous, on theoretical and empirical grounds, to propose an
analysis that does not view the examples in (18)-(24) as 'violations' or 'exceptions' to the
generality of the analysis. Second, while Longobardi claims that these exceptions hold
at a restricted stylistic level, the same cannot be said of BNs in other languages such as
PA. In the latter language, where BNs behave in a parallel fashion to Romance BNs,
such 'violations' or 'exceptions' are quite ubiquitous and not only limited to a stylistic or
elevated level of discourse. Therefore, Longobardi's analysis would not offer us any new
insights with respect to PA BNs. In the analysis argued for in this work, on the other
hand, modiﬁcation and coordination do not stand as exceptions or indeed violations to
the generality and explanatory power of the analysis proposed. As a matter of fact, 1
demonstrate later in this chapter that modiﬁcation and coordination play absolutely no
role in licensing BNs.

In more general terms, Longobardi's analysis is conceptually unappealing in that a

disjunction is assumed in the behavior of an essentially uniform phenomenon. BNs are

87

subject to precisely the same interpretation in both pre- and post-verbal positions; that
is, they are always interpreted existentially. This seems to call for a uniﬁed analysis.

Such an analysis will be forthcoming in § 3.4.

3.2.2.3 Focused Preverbal BNs

Perhaps more troubling about Longobardi's analysis is its failure to take the
interaction between BNs and the information structure of the sentence into account. It is
indeed the case that he observes that Ming a preverbal BN makes it more acceptable
(1994:616, ﬁ1.10) and suggests that the lexical government requirement on the head D
may be satisﬁed by reconstructing the preverbal BN to a postverbal position.8

A reconstruction analysis of a focused preverbal BN cannot be correct, however. It
contradicts Longobardi's general assumptions concerning the level at which the lexical
government requirement is satisﬁed in Italian. Longobardi's contention is that such a
structural requirement is fulﬁlled for Italian My at S-Structure (i.e. in the syntax),
unlike English in which this requirement is satisﬁed covertly at LF. A reconstruction
analysis must crucially assume that this process, reconstructing the BN to a postverbal
position, has to take place gym, at LP, in Italian. However, this is not the relevant
level for Italian since in this language, according to Longobardi's own formulation, the
lexical government requirement has to be fulﬁlled at S-Structure, otherwise Italian BNs
would be able to distribute across the board both pre- and post-verbally, exactly as in

English, unrestricted, which is obviously not the case.

 

8 Longobardi assumes here that the left-peripheral placement of the focused BN is a result of a
topicalization process.

88

Longobardi's second suggestion, based on personal communication with Luigi Rizzi,
is that it can be assumed that only the NP moves to a preverbal position, leaving the D
head behind to satisfy the lexical government locally. This suggestion, I think, is a
stipulation at best. One can hypothesize further that postverbal nominals that are
modiﬁed by overt determiner (e.g. an indeﬁnite article an in Italian or its plural
counterpart) would actually have this option of moving to a preverbal position leaving
the overt D head behind. This option is obviously impossible and empirically unattested.

I conclude, therefore, that Longobardi's attempts to explain away the acceptability of
focused preverbal BNs in terms of reconstruction, or particularly in terms of NP

movement that strands a determiner head, are unmotivated and conceptually inadequate.

3.2.2.4 The Empty D as a Default Existential Operator

Longobardi (1994:617) argues that the empty D head acts as an existential operator
that imposes a certain requirement of plurality (count or mass) on the head BN. This, he
points out, explains why BNs in Italian are restricted to plural count and singular mass
nominals. Cross-linguistically, however, this assumption is problematic.

PA does not possess an indeﬁnite article. Singular count nominals, therefore, occur
bare and distribute, and are interpreted, in identical ways to bare plurals in this
language. Furthermore, I have noted in Chapter 2 that plural BNs in PA also behave in
an identical fashion to Romance BNs. It seems to be highly stipulative to posit a
different internal structure for singular BNs and plural BNs in PA on the one hand, or
plurals BNs in PA and plurals BNs in Romance, on the other. In other words, it would

likely be an incorrect assumption to say that singular BNs in PA are NPs while plural

89

BNs in this language are DPs, with an empty D quantifying over the BN. This claim is
empirically problematic and leads to the wrong predictions. Recall from Chapter 2 that
singular and plural BNs participate equally and in identical ways in syntactic operations
argued to take place in the Construct State in PA, in particular substituting into a D
position. Such a syntactic operation, i.e. substituting a singular BN into a head D, would
not be possible if these nominals are assumed to be NPs, rather than NPs contained in a
DP shell. Finally, as will be shown in more detail in § 3.4.1.5, both singular and plural
BNs interact in identical ways with the information structure of the sentence in PA
which calls for a uniform analysis. It is in fact the null hypothesis that these nominals
should be subject to such a uniform treatment. The onus of proof, it seems to me, would
be on an analysis that argues for the opposite state of affairs.

These observations are not meant to argue against the existence of an empty D head.
In fact, my proposal in this work assumes the correctness of the DP hypothesis and that
BNs are DPs, and not NPs. What the preceding remarks attempt to contradict is positing
interpretive or structural constraints such as lexical government on this empty head.

Summing up, Longobardi's analysis leaves a number of issues unresolved. The nature
of lexical government would have to be tight enough to allow lexically governing a
subject, in a VOS sentence in particular. The implementation of this is far from trivial
and would require extra assumptions and machinery that may render the analysis
cumbersome. In addition, modiﬁcation and coordination remain weak links in such an
analysis in that they would constitute exceptions that do not lend themselves to a
straightforward explanation within the set of assumptions available in Longobardi's

approach. Longobardi's system also makes contradictory assumptions concerning the

90

level at which the lexical government requirement is presumed to hold in Italian. To
account for the acceptability of focused preverbal BNs in Italian Longobardi would have
to assume either reconstructing the BN at LF to a postverbal position or NP-movement
out of a DP shell into a preverbal position. Both assumptions are decidedly problematic.
An analysis in which these problematic issues do not arise would be more desirable.
Such an analysis seems to me to be tenable. I will have more to say about this in the
following sections. In what follows directly I turn to an examination of another

approach argued for by Casielles—Suarez (1997).

3.3 EMS and the Mapping Hypothesis

We have seen above that a lexical government account is unable to satisfactorily and
adequately account for the distribution of BNs. Casielles-Suarez (1997) seeks to account
for the distribution of BNs in Spanish without resorting to a government requirement,
contra Contreras (1986) and Longobardi (1994). I brieﬂy review the analysis in
Casielles-Suarez (1997) (henceforth, C-S) noting in conclusion that it too falls short of

adequately accounting for the distribution of BNs.

3.3.1 NPs or DPs

Before tackling the technical details of C-S's analysis, I would like to note that ﬁrst
and foremost C-S assumes, without argument, that BNs in Spanish are NPs, and not
DPS. I think there is enough evidence, especially of a cross-linguistic nature, to argue
against this assumption. The ﬁrst piece of evidence is in fact provided by Longobardi

(1994).

91

Perhaps Longobardi's major piece of evidence to the DP-status of BNs in Italian
stems from his contention that BNs are unable to fulﬁll a lexical government restriction
preverbally, unlike their occurrence in postverbal positions. We cannot, however, accept
such an argument due to the simple fact that I have immediately above argued against
the correctness of the applicability of this notion of government to BNs. The evidence
for the DP status of BNs has to be sought elsewhere. Longobardi does in fact provide
other empirical evidence which bears on this issue.

According to Longobardi (1994:618), determinerless predicative expressions with
mass or plural heads can be shown to contain an empty D head. Consider the following

(Longobardi's l9a-19f) :

(25) a. Gianni e medico.
Gianni is doctor

b. Gianni e un medico.
Gianni is a doctor

c. *Gianni e medico che si cura davvero dei suoi pazienti.
Gianni is doctor who really cares for his patients

(I. Gianni a: un medico che si cura davvero dei suoi pazienti.
Gianni is a doctor who really cares for his patients

e. Noi siamo medici che ci curiamo davvero dei nostri pazienti.
We are doctors who really care for our patients

f. Noi siamo dei medici che ci curiamo davvero dei nostri pazienti.
We are PARTIT ART doctors who really care for our patients

The minimal pair in (25a)and (25b) argues for the optionality of the indeﬁnite article in

these sentences. Relativization in (25c) is ungrarnmatical. The source of the

92

ungrarnmaticality is arguably the absence of the article. Evidence for this comes from
contrasting (250) with (25d) which minimally differs from the former in that the article
is present and relativization works here. In other words, the relevant factor to make
relativization work seems to be the presence of a D head.9 Now, Longobardi explains,
if relativization is parasitic on the presence of a D head, either an overt head in the case
of (25d) and (25f) with the partitive article, or empty in the case of (25e), taken in
tandem with the optionality of the article in (25a—b), this suggests that in the case of
plural BNs there always has to be a (n empty) D head.IO ”

Another cross-linguistic observation may provide further evidence in this regard.
Recall from Chapter 2, § 2.2.4, that BNS (or determinerless nominals) in PA can
substitute into a D position when they are marked with the necessary (in)defmiteness
features, as argued also for bare nominals in the PA Construct State. If such a head
position did not exist then no N movement would be possible out of the NP in
complement position to the empty D. These cross-linguistic observations waken C-S's

assumption that BNS in Spanish are NPS, rather than DPS.‘2

 

9 This line of argument, of course, takes for granted that the indeﬁnite article is in D, an assumption that is
not uncontroversial. Mum and Schmitt (1999) argue that the indeﬁnite article is the spell out of singular
number, which would explain the impossibility of bare singulars in predicative positions in English, unlike
what happens in Romance. This, Munn and Schmitt (199927) argue, would follow from the difference in
the structure of the predicate nominal in Romance and English. In the former, Agr and Num would be
split, unlike English in which Agr and Num are fused. So, in Romance the predicate nominal would be an
AgrP whereas in English it would be an Agr/NumP.
'0 Similar examples with mass nominals are also provided by Longobardi.
" An additional argument provided by Longobardi (p.620) is exempliﬁed by the following sentences:
(i) Ritengo Mario I"(un) bravo medico.

I believe Mario (a) good doctor
(ii) Ritengo Gianni e Mario (dei) bravi medici.

I believe Gianni and Mario (PARTIT ART) good doctors

The indeﬁnite singular article unis obligatory in (i). However, in the almost identical (ii), except for
plurality, the article is optional. (ii) would be good with an overt partitive article or a non-overt D head.
'2 Schmitt and Munn (2000) argue, however, that bare nominals in Romance are NumPhrases while bare

93

3.3.2 RIP-Movement Vs. DP-Movement

Acting on the assumption that BNS in Spanish are NPS, and not DPs- an assumption
noted immediately above to be problematic from a cross-linguistic perspective-
Casielles-Suarez (1997) contends that the asymmetry apparent in the behavior of BNS,
for her attested in the contrast between (26) and (27), follows from the fact that BNS,

being NPS, cannot move out of the VP (These are her examples (3) and (4), p. 95):

(26) jugaban nifios en la calle
played children in the street

(27) * Niﬁos jugaban en la calle
children played-3p] in the street

In other words, BNS in Spanish are always interpreted within the VP, C-S argues. This
would help explain why BNS in Spanish can only be interpreted existentially and the
BNS in examples (26) and (27) above can never be interpreted generically. This is also
seen in the ungrarnmaticality of (28) and (29) (her 17 and 19, respectively) with a

generic reading:

(28) * Nifios (generalmente) jugaban en la calle (generic)
children (generally) played-3pl in the street

(29) * jugaban (generalmente) niﬁos en la calle *generic
played (generally) children in the street

 

nominals in Brazilian Portuguese and English are DPS.

94

 

dil
is 1
bar
til
(19
me
anal
llie
SEW]
fold
of Di
(1997

lanai

To account for the acceptability of focused preverbal BNS, C-S unavoidably assumes a
different kind of movement for BNS out of the VP, namely, focus-preposing. The claim
is that while BNS can focus-prepose at S-Structure these nominals are reconstructed
back into the VP at LF. That also, 08 argues, explains why they are interpreted
existentially. This last observation follows from the Mapping Hypothesis of Diesing
(1992b), which C-S assumes to be correct and applicable to Spanish BNS. This brings
me to the ﬁrst major problem that can be noted with C-S'S analysis, namely, that the
analysis in essence relies on the correctness of Diesing's Mapping Hypothesis (MH).
The MH, however, has been Shown to be inadequate and empirically problematic by
several researchers, among whom are Reinhart (1995) and Pinto (1997). In the
following section I brieﬂy review the essential assumptions of the Mapping Hypothesis
of Diesing (1992b) and the criticism leveled against it by Reinhart (1995) and Pinto
(1997). However, I ﬁrst begin by looking at Heim's (1982) analysis of indeﬁnites as

variables on which Diesing (1992b) bases her own proposal.

3.3.3 The Mapping Hypothesis: Assumptions and Problems
3.3.3.1 Heim's (1982) Conception of Indefinites

Diesing (1992b) assumes a view of indeﬁnites as variables unselectively bound
based on the analysis presented in Heim (1982).l3 Heim (1982) argues that indeﬁnites
do not have any quantiﬁcational force of their own. The quantiﬁcational force

indeﬁnites seem to have is brought about by other expressions in the linguistic

 

'3 See Carlson (1977a, b) for arguments against viewing bare nominals as indefmites.

95

lan'a
more
their lg
i"deﬁn
COlilaini
‘ldt‘linil.

cliili’mla

environment of the indeﬁnite. So, adverbs of quantiﬁcation (such as always) act as

operators that bind two arguments according to the following schema:

(30) Q-Adv((D,‘1’)

The Q-Adv in this case unselectively binds the variables in the following manner

exempliﬁed for always:

(31) "always (¢,‘I’)" is true iff every assignment to the free variables

in (I) which makes (1) true also makes ‘1’ true.

The ﬁrst argument (D of the quantifying adverb acts as a restriction on the domain of
quantiﬁcation of the adverb, hence the term 'restrictive clause.‘ The second term in the
quantiﬁcation corresponds to what Heim labels 'nuclear scope.’ So Q—Adv ((D, if) is true
iff the variable assignment to '2” (the nuclear scope) belongs to the restricted set of
variable assignment satisfying (D(the restrictive clause). Indeﬁnites, Heim argues, act
more like variables than quantiﬁers. Indeﬁnites have this variable reading in virtue of
their lexical meaning and this variable reading is the only reading contributed by the
indeﬁnite to whatever sentence it occurs in. Thus, the quantiﬁcation present in sentences
containing indeﬁnites is triggered by something else in the sentence and not by the
indeﬁnite: either by an adverb of quantiﬁcation, overt or invisible, i.e. implicit, or by

existential closure. The mapping ﬁ'om S-Structure to Logical Form is achieved via a

96

number of rules Heim calls 'rules of construal.’l4 The end result of these rules of
construal is cutting up the logical form into a tripartite structure as demonstrated for the

example "Every man arrived" below:

(32) Every man arrived
every NP S
/\ A
man e arrived

Also consider the following example (32a) and its representation in (32b) (her examples
(l6) and (l7c')) ﬁom Heim (1982: 191) that shows how this work with a quantifying

adverb (such as always):

(32a) If a cat has been exposed to 2,4-D it always goes blind.

 

” According to Heim (1982), the ﬁrst rule of construal is called NP-Indexing: Assign every NP a
referential index. Another rule of construal is NP-Preftxing: Adjoin every non-pronominal NP to S (i. e.
1P), leaving behind a co—indexed empty category. A third rule of construal is Quantiﬁer Construal: Attach
every quantifier as left-most immediate constituent of S (1P). This rule of construal helps create the
tripartite structures characteristic of quantiﬁcational sentences. Finally, Existential Closure is a rule of
construal that subdivides into 2 parts: Adjoin a quantiﬁer 5’ to the nuclear scope of every quantiﬁer, and
the other division of Existential Closure takes care of indeﬁnites in unembedded sentences: Adjoin the
quantiﬁer 3to T(ext) (as in the sentence He went to a restaurant. It was expensive). Here it is assumed
that there is an operation or process of text formation where a number of sentences are conjoined under a
T node.

97

(32b)

 

 

S
always 1 NPl S
/\ 3/\
NP] 8' S UP]
a cat e1 that has been e1 goes blind

exposed to 2, 4-D

Depending on where the indeﬁnite is placed the interpretation will vary accordingly. If it
is caught in the restrictive clause of an implicit or explicit quantiﬁer/operator (as in

(3 2b)) it gets the generic interpretation. However, if it occurs in the nuclear scope of the
operator it gets bound by existential closure (the default existential operator (3 in
(32b))).

Diesing adopts Heim's analysis with very minimal changes. Among the
modiﬁcations that Diesing makes to the original Heim analysis is Diesing's assumption
that existential closure only closes off the nuclear scope and there is no existential
closure applying to the whole text (see fit. 14). The application of existential closure to

texts would give the wrong reading for a sentence such as the following:

(33) Oscar owns Sheep. Otto vaccinates them.

Applying text-level existential closure to this sentence would mean that there are some

98

Sheep that Oscar owns that Otto vaccinates. However, the meaning of the sentence is
that Otto vaccinates all of the sheep owned by Oscar.

Diesing further assumes that the mapping from the syntactic structure to the logical
form is straightforward. She Splits the (two-subject's) syntactic tree into material that
maps into the nuclear scope which is isomorphic with the VP (containing the VP-
intemal subject) and material that gets mapped into the restrictive clause which includes
the higher subtree dominating the VP. From there on it is a very Short step to produce
the relevant interpretations for indeﬁnites, Diesing argues. An indeﬁnite that is mapped
into the nuclear scope (i.e. within the VP) is interpreted existentially, where existential
closure is operational, whereas an IP-level indeﬁnite gets interpreted generically. This

Diesing forrnalizes as the Mapping Hypothesis (1992b:10) in (34):

(34) The Mapping Hypothesis

Material from VP is mapped into the nuclear scope.
Material from IP is mapped into a restrictive clause.

Diesing brings a good amount of empirical data from both German and English to bear
on the correctness of the mapping from the syntactic structure to the semantic
representation the Mapping Hypothesis is supposed to encode. German seems to
provide primafacie conﬁrmation of the MH Since in this language subjects can occur in

either of the two positions argued for the subject at S-Structure: in the VP-intemal

 

'5 Diesing assumes the VP-lntemal Subject Hypothesis which maintains that subjects are base-generated
in a speciﬁer position internal to the verb phrase (see, among others, Koopman and Sportiche 1991).

99

To
Lei
Iher
inter

r eC0:

position or in VP-external position. The placement of the subject in either of these
positions coincides with a difference in interpretation: in the VP-internal position the
indeﬁnite subject is read existentially whereas in the VP-extemal position it can only be
generic. '6 Since subjects in English only occur VP externally Diesing is forced to
assume LF lowering for existentially read subjects. Therefore, the indeﬁnite subject
ﬁremen in (3 5), which Diesing takes to be ambiguous between a generic and an
existential reading, would have to lower, or reconstruct, back into the VP at LF on the
existential reading. This option is unavailable when the indeﬁnite subject firemen is
generically interpreted in (35) and in (36) (or for German indeﬁnite subjects for that

matter Since the mapping in German takes place overtly).

(35) Firemen are available.

(36) Firemen are altruistic.

To execute this idea Diesing assumes, with Kratzer (1995), that subjects of Individual-
Level predicates (Carlson 1977a, b) are base generated outside the VP, and cannot
therefore reconstruct. Subjects of Stage-Level predicates are base generated VP-
intemally and their S-Structure position is derived. These subjects can accordingly
reconstruct back into the VP at LF.

This brief overview of Diesing's (1992b) analysis would be sufficient for my

 

'6 Diesing's judgments concerning the positioning of subjects at S-Structure, either pre- or post-verbally,
follow from her claim that certain adverbials such as ja doch 'indeed' mark off the edge of the VP. An
indeﬁnite subject occurring to left of this particle would be outside the VP whereas an indeﬁnite occurring
to the right would be inside.

100

purposes here. For the exact technical details of Diesing's analysis the reader is referred
to that work. I turn in the next section to an evaluation of the theoretical and empirical
soundness of the MH by brieﬂy considering the work of Reinhart (1995) and Pinto

(1997).

3.3.3.2 Problems with the Mapping Hypothesis
3.3.3.2.1 Reinhart (1995)

Reinhart (1995:63) points out that Diesing posits a parametric variation
(e. g. between English and German) at the LF level. Recall that the MH assumes a
lowering rule for the indeﬁnite subject in English, when existentially interpreted, to
reconstruct back into the VP at LF. This possibility, on the other hand, cannot be
entertained for German since indeﬁnite subjects in this language need not reconstruct
due to the availability of two S-Structure positions for subjects.17 To assume a
parametric variation between languages at the LF level, Reinhart explains, runs counter
to fairly standard assumptions to the effect that languages do not differ in their LF
representations.

Another problem Reinhart points to is the following. There-Insertion sentences

perhaps provide the strongest evidence for Diesing's analysis.18 However, Diesing fails

 

‘7 If a lowering rule is assumed for German, indeﬁnite subjects in German would be expected to be
ambiguous between a generic and an existential reading in preverbal position, as in English. Furthermore,
VP-intemal indeﬁnite subjects in German could also be interpreted generically since a 'raising' analysis
could in principle be argued for on a par with a lowering analysis. Neither possibility, lowering or raising,
is allowed or attested according to Diesing (1992b).

'8 This is so since it is widely known that there-insertion sentences give rise to the so-called "deﬁniteness
effects" (DE). DE are exempliﬁed by the fact deﬁnites are not allowed in there-insertion sentences as in
‘There is the boy in the garden. Bare plurals occurring in such sentences are interpreted existentially, not

101

to explain how exactly the facts follow in these sentences. If the mapping from the
syntactic structure to the logical representation takes place in English at LF, what,
Reinhart asks (p.63, fn), prevents the existential subject from moving at LF to an IP
position, as is commonly assumed for the associate in there-insertion sentences (e. g.
Chomsky (1995)), and in so doing be interpreted outside the VP (i.e. generically).

A claim of Diesing's (1992b) that also turns out to be problematic for Reinhart
(1995:71) is the fact that weak indeﬁnites‘g as in (37) are argued to be ambiguous

between a strong/presuppositional reading and a weak reading:

(37) Two/Some ﬁremen are available.

Diesing assumes such a distinction to be there, with no further argument, Reinhart
points out, nor does Diesing provide a sufficient test to detect this supposed ambiguity,
other than an imagined context which to Reinhart's mind leaves many questions
unanswered (p.72). As for Diesing's claim that a stressed SOME provides the
disambiguation necessary in that it indicates presuppositionality in examples such as
(3 8), Reinhart argues that the stressed SOME can actually occur in there-insertion

sentences, traditionally a testing ground for weak indeﬁnites where the strong reading is

 

generically. These BNS in there-insertion structure are clearly in the VP at S-Structure. So, according to
Diesing's MH they are predicted to have an existential reading. However, note Reinhart's criticism in the
text.

'9 Roughly, indeﬁnites that can occur in there-insertion sentences (see iii. 18). Milsark (1974), as cited in
Ladusaw (1994), makes a distinction in the class of determined NPS between those that have a cardinality
reading (sm men) (i.e. weak indeﬁnites), and those that are quantiﬁcational (the men) (i.e. strong
indeﬁnites). Diesing also claims that weak indeﬁnites (such as three men) are ambiguous between a strong
(presuppositional or partitive) reading and a cardinal reading.

102

excluded, as in (39) (1995:73):

(38) SOME ghosts are in the pantry, the others are in the attic.

(39) There are SOME ghosts in the attic.

If a stressed SOME can be used to diagnose the strong or presuppositional reading then
we lose our account of the ungrarnmaticality of sentences such as "*There is every ghost
in the garden," Reinhart explains. These problems are non-trivial in that they demand

from Diesing extra assumptions that could ﬁrrther complicate her analysis.

3.3.3.2.2 Pinto (1997)

Other problems with the MH have been pointed out by Pinto (1997) who
demonstrates that empirical evidence from Italian is at odds with the basic tenets of
Diesing's MH.

Pinto's (1997) main concern is to argue that (subject—verb) inversion in Italian is not
free but is syntactically determined. To begin with, she shows that inversion cannot be
connected to or reﬂective of unaccusativity Since some unaccusative verbs cannot invert
with a wide focus reading and some unergatives can invert with such a reading.20 The
availability of inversion with a wide focus reading for Pinto follows from the presence

of a locative argument LOC that is syntactically projected but could be lexically covert.

_k

20 A sentence with a wide focus reading is basically an answer to the question "what happened?" as
Opposed to a sentence with a narrow focus reading whose question could be, for example, "who v-ed
GDP)?" The importance of wide focus reading here is that the answer to a question with wide focus
reading in languages such as Italian, which has two surface positions for the subject, would usually be
aHSWered by an inverted structure.

103

 

“ "m...- \’- L). ‘ -.

Evidence for this comes from the difference in meaning between inverted (VS) and non-
inverted sentences (SV). In the former the meaning seems to involve a locative
argument or location that is deictic (i.e. Speaker oriented). Such an argument is absent
from non-inverted sentences. The syntactic implementation of Pinto's ideas is achieved
through adopting a Minimalist (in the sense of Chomsky (1995)) approach to the data.
Since the details or the execution of these ideas are not relevant here I will only give
Pinto's ﬁndings and generalizations that are in conﬂict with or critical of Diesing's MH.
Pinto (1997:199) observes that the mapping to logical forms in Italian must occur
overtly in the syntax (like German) Since with inversion verbs the indeﬁnite subject in
preverbal position has only a partitive reading ( i.e. strong, therefore mapped into the
restrictive clause at LF). Postverbally, the subject can only have a weak (i.e. existential)

reading. This pattern, which supports Diesing's MH, is exempliﬁed as in (40a—b):

(40) a. Due linguisti sono arrivati. partitive/ #existential
two (of the) linguists arrived

b. Sono arrivati due linguisti. existential/ #partitive
arrived two linguists

c. Due dei linguisti sono arrivati. partitive/ #existential
two of the linguists arrived

d. Sono arrivati due dei linguisti. partitive/ #existential
arrived two of the linguists

With overt partitives, however, the readings are not predicted by the MH.21 In both

_

2' However, Pinto realizes that it could be argued that overt partitives are not indeﬁnites as such. Still, She

104

main.

> a
‘1

“Edi:

 

(40c) and (40d) the overt partitive does not have an existential reading.22 Assuming the
MH, the overt partitive in (40d) occurring postverbally has to be mapped into the
nuclear scope (Since it is VP-intemal) and Should get an existential reading, a reading
which is in reality unavailable. To rescue example (40d) it could be argued that the
mapping takes place at LF where the postverbal partitive scrambles out of the VP to be
mapped into the restrictive clause. But then (41a) would be expected to be ambiguous
between a strong and a weak reading (depending on whether the indeﬁnite reconstructs
or not, like English BNS with stage-level predicates), which is again contrary to fact.

AS for indeﬁnite subjects in sentences with non-inversion verbs, Pinto argues that
the following generalization holds:

(42) Interpretation of indeﬁnites in non-inversion verb contexts:

preverbal subjects —-) partitive or existential
postverbal subjects —) not available with wide focus

Pinto points out that (42) is not necessarily problematic to the MH; however, to account
for the ambiguity in the interpretation of the preverbal subject the mapping would have
to be assumed to occur c_ov_elly; but this again conﬂicts with the mapping supposed to
take place o_vert_ly for examples (40a-b) above.

The following examples are also noted to pose a problem to the MH:

‘

maintains that the fact remains that Diesing' 5 analysis does not account for or consider overt partitives.
22A partitive reading for Pinto coincides with a strong (as opposed to weak) reading of the indeﬁnite. This
reading in Diesing' S system is possible in the 1P domain, not within the VP.

105

(43) a. Che cosa é successo?
what happened

b. Ha pianto un bambino.
cried a child

c. E’ svenuto un uomo.
fainted a man

These examples are acceptable postverbally with wide focus reading with non-inversion

verbs, Pinto argues (p. 204). So, She prOposes to modify (42) as follows:

(44) Indeﬁnite subjects with non-inversion verbs:
strong and weak indeﬁnites must occur in preverbal subject position where
they are ambiguous. When the determiner is the indeﬁnite article, the subject
can Show up both in preverbal and postverbal position, correlating with a
strong and a weak interpretation respectively.

In other words, it would have to be assumed that the mapping is overt for the indeﬁnite
article, while all other weak determiners, which can only show up preverbally, would
require covert mapping. This is surely a problematic assumption.
Apart from all these empirical problems with the MH Pinto (1997: 205) also points
out a conceptual problem. The MH comes at the cost of a redundancy at the lexical level
by positing an ambiguity approach to indeﬁnites. Indeﬁnites would be ambiguous
between weak and strong or existential and presuppositional.
To sum up, we have seen in the brief overview of Reinhart's and Pinto's analyses
t112311: the assumptions contained in Diesing's (1992b) proposal are problematic at best.

Reinhart demonstrates that the MH leaves some questions unanswered. The MH also

106

mil
pare
0111
the '
is el

data

3.3

COIT
he. 1

03

PM
insic
this
OI q
Pitt,

BN3

makes assumptions that are theoretically unorthodox and problematic such as stating
parametric variation between languages at level of Logical F orrn. Pinto in turn points
out some conceptual inadequacies of the MH, such as the redundancy at the lexical level
the MH assumes for indeﬁnites. More importantly, Pinto's analysis shows that the MH
is empirically inadequate in that it makes the wrong empirical predictions for Italian

data.

3.3.4 An Evaluation of Casielles—Suarez's (1997) Analysis
The discussion in the previous two sections has been intended to illustrate, and it
seems to me has also established, that for an analysis assuming the correctness of the
MH some conceptual and empirical problems emerge that would need to be overcome.
Casielles-Suarez's (1997) (C-S, henceforth) analysis essentially presupposes the
correctness of the MH. However, apart form this Signiﬁcant weakness in C-S's analysis,
i.e. relying entirely on the correctness of the MH, there remain some other problems that
C-S's proposal faces, to which I turn directly.

I{ecall from § 3.3.2 C-S'S basic proposal: BNS in Spanish are NPS, not DPS, which
canDOt move out of the VP. This explains, in C-S's view, their unacceptability in
preverbal positions as well as their existential reading Since these nominals are 'caught'
inside the VP, thus subject to the effects of existential closure. An immediate problem
this analysis faces is how to explain the alleged grammaticality of BNS when modiﬁed
or coordinated. If modiﬁcation and coordination could in principle rescue BNS in
p reverbal positions then it would have to be said that, when modiﬁed or coordinated,

B
NS a) are no longer NPS, or b) become DPS, or c) are NPS that can, somehow, DP-

107

move out of the VP. It is not clear to me how any of these assumptions can be
implemented or executed; or, at least, their implementation is far from trivial. As for the
acceptability of focused preverbal BNS C-S is forced to assume a movement account for
these nominals from within the VP. This movement is licit, She suggests, Since it is an
instance of focus-preposing.

To assume focus-preposing (i.e. movement) for focused BNS causes C-S's analysis
to lose the uniformity that would have been otherwise a welcome result of her approach.
Here are the cold facts of the analysis: unfocused BNS are NPS that cannot move out of
the VP at S—Structure; focused BNS can focus-move or prepose out of the VP at this
level. This bifurcation unnecessarily complicates the analysis of BNS and a Simpler
analysis would obviously be more desirable.

More damaging to C-S'S analysis is her argument concerning focus-preposing while,
at the same time, assuming Diesing's MH. It is abundantly clear from her analysis that
the mapping to logical representations She assumes for Spanish, in keeping with the

MH, takes place overtly in the syntax (i.e. at S-Structure). If focus-preposing occurs
overtly, then the focused bare nominal would have to reconstruct into the VP to be
Callght by the default existential operator closing off the nuclear scope. The mapping for
fC’cus‘Preposed BNS, therefore, Should inevitably be assumed to occur covertly, at LF,
not 0vertly. But, of course, this is not, nor Should it be in terms of C-S's overall system

0f assumptions, the relevant level for Spanish. This situation seriously undermines C-S's

analysis, I think.23

\

23
Th. . . . . . I o O
18 rs also remmlscent of the Situation we have encountered wrth Longobardi's (1994) solution to

 

108

C-S assumes that DP movement in Spanish is topic-driven and that BNS do not have
the necessary features and therefore cannot be topics. Abstracting away from whether
DP-movement is topic-driven movement or not, I believe C-S's assumption that BNS
cannot be topics is essentially correct. This is in essence the line of argument I propose
in the present work. However, the (syntactic) execution of the idea that C-S adopts
seems to me to beg numerous questions some of which have been raised in the
preceding discussion. In addition, C-S fails to recognize the close interaction between
BNS in English and information structure. As a matter of fact She assumes that English
BNS stand at odds with Spanish BNS with each at a different end of the Spectrum. I
believe this assumption is the wrong assumption. As has been alluded to in Chapter 2,
and as will be seen in more details in the chapters that follow (Chapter 4, in particular),
existentially interpreted BNS in English also resist a topic reading in that they are always
focal- I show in § 3.5.2 below that (existentially interpreted) English BNS behave much

more like Spanish BNS, and also PA BNS, than C-S seems to suggest in her analysis. I

conclude, therefore, that C-S's proposal is inadequate in some important respects and a

more uniform analysis is called for.

3-4 A Uniform Account of Bare Nominals

The attempts to explain the behavior of BNS based on purely syntactic grounds have

tumed out to be untenable. Analyses based on the idea of lexical government leave

 

\

fOc
used BNS preverbally.
l 09

many important issues unresolved, such as the effect modiﬁcation, coordination or focus
may presumably have on the acceptability of BNS. These issues are not satisfactorily
addressed in C-S'S (1997) account either which takes BNS to be NPS whose movement
out of the VP is ruled out, unless, of course, it is focus-preposing. Aside from the lack
of uniformity characteristic of C-S's approach, her account is also undermined primarily
by the unmotivated reliance on Diesing's Mapping Hypothesis.

Both of the analyses previewed in the previous sections, namely Longobardi (1994)
and CasielleS-Suarez (1997), have managed somehow to overlook the importance of the
information structure of the sentence in which the BN occurs. On the one hand,
Longobardi (1994) ignores the glaring relevance of the notion of focus in rehabilitating
BNS in preverbal position and unjustiﬁably assumes it to be an idiosyncratic behavior
that can ﬁnd a motivated explanation in terms of the syntax of BNS. CasielleS-Suarez

(1997), on the other hand, does recognize the signiﬁcance of information structure and
i ts relevance to explaining the behavior of BNS when she argues that DP movement in
Spanish is topic-driven and that BNS in this language do not have the necessary features
to Participate in such movement. In other words, BNS cannot be topics. However, C-S
fails to see that this restriction on BNS is much stronger and tighter than She assumes as
I Will eXplain directly below. Moreover, C-S'S analysis mistakenly posits a Split between
BNS in Spanish and English by failing to recognize that BNS in English actually exhibit
a noticeably strong interaction with the information structure of the sentence. This
panic‘llar issue is taken up in greater detail in the Chapter 4.

In the next section I propose a more restrictive account of BNS in PA and Spanish,

and English, as it turns out. In the analysis to be elaborated shortly no disjunction in the

110

behavior of PA or Spanish bare nominals is presumed. Bare nominals in these languages
are unacceptable in preverbal or postverbal position unless they are focal. No structural
requirement such as government is needed nor is it necessary to assume that BNS are
NPS that have always to occur within the VP. Furthermore, my analysis does not rely on,
nor does it assume, Diesing's Mapping Hypothesis, which, as I have attempted to
illustrate, is conceptually and empirically inadequate. I believe the analysis I argue for
here Should prove more superior in the way of explanatory adequacy and conceptually
more appealing. Moreover, it will enable us to present a truly restrictive view of the
differences in the behavior of BNS between English, on the one hand, and PA and
Spanish, on the other. Such a restrictive view would ultimately allow us greater

understanding of why these languages use DPS to refer the way they do. This will be the

topic of Chapter 5.

3.4. 1 Bare Nominals and Informational Focus

3-4. 1.1 Bare Nominals and Presentationalism

I begin by discussing the relevant notion of presentationalism. Presentational
Sentences have the order [VS]. Suffer (1982:125) points out that the communicative

funcﬁon of presentational sentences is asserting the existence of the NP (i.e. DP)

referent as exempliﬁed in the following:

(45) A parecio un homre
appeared a man

111

B.\
the!

PM;

(46) A somo el sol radiante
peeked (3 sg) the sun radiant
'The radiant sun peeked out'

That the referent denoted by the NP (DP) is what is being asserted (or, more

speciﬁcally, its existence) can be demonstrated through negation (p.192.fn 2):

(47) No aparecio un hombre, en realided no aparecio absolutamento nada
no appeared (3 sg) a man, in reality no appeared absolutely nothing

What is being negated in (47) is the existence of the referent denoted by a man and its
participation in the act of appearing. In other words, presentational sentences are
existential. Both the verb and the NP (DP) in presentational sentences are in the scope

of assertion, so both can be thematic. In declarative sentences (with the order SV(O)) S
is thematic while the VP is rhematic.

Suﬁer (1982: 212) ﬁrrther explains that presentaionalism is exclusively
Characteristic of semantically intransitive verbs. If BNs in Spanish are unacceptable in
Preverbal positions (under normal conditions of stress), and since in presentational
sentences the subject is thematic (and the verb may be), the expectation would be that
BNS are perfectly acceptable in presentational (i.e. intransitive) sentences. What remains
to be decided, however, is the status of transitive sentences and their interaction with
BNS: can transitive sentences be presentational? And if so, can BNS freely occur in

t - ° °
hem? The answer to the ﬁrst questron Suﬁer argues to be negative. If the ultimate

p W038 of presentational sentences is to introduce the DP referent into the world of the

112

discourse, uniquely and exclusively, then the prediction would be that transitive
sentences, which characteristically have two DP referents, cannot be presentational.
Suﬁer also answers the second question in the negative: postposing a BN in a transitive
sentence (i.e. creating a presentational context) does not license the BN. This is

exempliﬁed by the following examples (Suﬁer 's (l6)-(19)):

(48) a. *Hombres cazaron un ciervo
men hunted a deer

b. * cazaron hombres un ciervo
c. *cazaron un ciervo hombres

(49) a. *Rumores llenaban el cuarto
Rumors ﬁlled the room

b. *Llenaban rumores e1 cuarto
c. *Llenaban el curato rumores

(50) a. *Alumnos dieron regalos a la maestra
students gave presents to the teacher

b. *Dieron alurnnos regalos a la maestra

c. *Dieron regalos alurnnos a la maestra
d. *Dieron regalos a la maestra alurnnos

Suﬁel‘ speculates that since transitives have two arguments (V NP NP) ambiguity may
resuk under normal conditions of stress and intonation (1982: 213). Therefore the use of
unmodified BNS postverbally in transitive sentences can be allowed if such ambiguity

can SOrnehow be barred from arising as in the following examples:

ll3

(51) La cosecha la destuyeron langostas
the harvest (f.sg.) it (f.sg.) destroyed (3 pl.) locusts
'The harvest, locusts destroyed it'
(52) El cuarto lo llenaban rumores

the room (m.sg) ﬁlled rumors
'The room, rumors ﬁlled it'

Notice that (51) and (52) are acceptable since it has been made abundantly clear that the

BNS, langostas 'locusts' and rumores 'rumors,’ respectively, are the only rhematic or
f_oc_§ elements in the sentences. Leﬁ-dislocating the DPs La cosecha 'the harvest' and El
cuarto 'the room' in (51)-(52) serves to signal their 'topical' status and also highlight the
focal status of the BNS in the sentences. I believe Suﬁer's overall argument is sound and
can be shown to be on the right track. I also believe Suﬂer’s descriptive observations can
be formalized and carried over successfully to PA.
I have just noted concerning Suﬁer's examples (51)-(52) that what seems to be

crucial for BNs is that they be focal. Focus, then, plays the decisive role in licensing
BNS- In the next section I examine brieﬂy Zubizarreta's (1998) assumptions regarding

the interaction of focus and nuclear stress (in Spanish) and the notion of informational

focus that I adopt in this study.

3-4- 1 .2 Nuclear Stress and Focus
It has been pointed out by Reinhart (1995) that there are two schools of thought

when it comes to the relation between focus and phrasal stress. On the one hand, there
has been the argument that the stress of a sentence is phonologically determined, apart

from any discoursal consideration. This approach has been revived by Cinque (1993)-

114

On the other hand, the counterargument has been that focus is not phonologically
determined and it is primarily determined by discourse determinations. Such an
approach has been argued for, for example, in Zubizarreta (1998). For my present
purposes, it sufﬁces to say that both these approaches can be shown to be compatible
with the PA facts. For simplicity of exposition, and also due to its relevance to the

argument to be made in this chapter, I very brieﬂy introduce the main ideas of

Zubizarreta (1998) at this stage. A more detailed discussion of Zubizarreta's analysis
will have to wait until Chapter 5.
Zubizarreta (1998:37) points out that natural languages make use of one or more of
the following ways to mark the focus in the sentence: prosody, morphology or a
syntactically speciﬁed position. She puts forward an analysis of how prosody identiﬁes
the focus of the sentence. Phrasal prominence, as determined by the Nuclear Stress
Rule, mediates the relationship between prosody and focus, according to Zubizarreta.
T0 aChieve more empirical coverage she argues against a monolithic view of the
Nuclear Stress Rule (N SR) and proposes a modular view of this rule according to which
asymn‘letric c-command and selectional ordering play a pivotal role. Glossing over a lot
0f tech-Hi cal details to be reviewed in the Chapter 5, Zubizarreta argues that in Spanish
the NUClear Stress Rule assigns prominence to the constituent lowest in the syntactic
tree in terms of c-command. With this in mind, she motivates what she calls
p rosod i Cally-motivated movement which dislocates constituents that are not focal so that
focused constituents end up lowest in terms of c-command. The same facts carry over to

P . . .
A Sent€311ces; for now I assume the correctness of thrs hypothesrs until a thorough

exam ~ . .
1nEltlon in Chapter 5.

115

Taking Suﬁer's (1982) observations into account, and working within the framework
of assumptions in Zubizarreta (1998), I will demonstrate that BNs in PA are only
acceptable if focused. However, prior to doing that, I want to brieﬂy explain the notion
of focus I assume in this work.

The notion of focus I am assuming in this dissertation will be that adopted in
Zubizarreta (1998) (based on Chomsky (1971, 1976) and Jackendoff (1972)). Focus is
understood in terms of the discourse notion of presupposition as the non-presupposed
part of the sentence (Zubizarreta 1998: l). Zubizarreta uses the question/answer pattern
to determine the information structure of the sentence in terms of focus and
Presupposition (i.e. shared assumptions). She assumes that the focus structure of the
sentence should be represented in connection to and in tandem with the context question
Of that sentence as in the example in (53) below (the focus is indicated with the diacritic

F)~ The focused constituent in the answer will correspond to the wh-element in the

Context question:

( 5 3 ) a. [What happened?]
[F John [ate [the pie]]].

b. [What did John do?]
[John [1: ate [the pie]]]

c. [What did John eat?]
[John [ate [1: the pie]]].

(1. [Who ate the pie?]
[[F John] [ate [the pie]]].

e. [What happened to the pie?]
[[r John ] [[r ate] [the Pie]]]-

ll6

f. [What did John do with the pie?]
[John [[F ate] the pie]].

I think this simpliﬁed notion of focus will be sufﬁcient for my purposes in this work.
My next task will be utilizing this notion of focus in ﬂeshing out the details of my main

argument, namely, that BNS in PA are always focal. To this I turn directly.

3.4.1.4 Bare Nominals are Always Focal

Let me begin by considering the following example:

(54) *S’abiid banu Lahw
slaves built the-pyramids

AS is expected the sentence is ungrarnmatical since the BN fabiid 'slaves' occurs

pr eVﬁl‘bally under normal conditions of stress (i.e. unfocused). In accordance with
Zubizarreta's system of assumptions noted in the preceding section, the DP I-ahraam
'the‘PyI‘amids' in (54) would be assigned nuclear stress (NS) (indicated by underlining)
by the Nuclear Stress Rule (N SR). This means l-ahraam 'the-pyramids' has to be the
focused, i.e. the non-presupposed, constituent. However, the sentence would be
gramrrjlzltical if the BN was focused as in (55) (1 indicate contrastive focalization by

capital 3) :24

\

24
N
uu ote that the NS cannot be assigned to I-ahraam 'the-pyramids' in (55) as in (i) since the focus in the
Ce is YABIID 'slaves' in which case the latter is contrastively focused, thereby attracting all and the

 

ll7

ﬂ

0n

(55) YABIID banu l-ahraam
SLAVES built the-pyramids

This much is uncontroversial. The claim I wish to argue for, however, is stronger: BNs
are not possible postverballyﬁeither, under normal conditions of stress. Thus, the
postverbal BN in (56) is ungrarnmatical, contrary to common assumptions (for example,

as in Longobardi (1994) for Italian (Romance), C-S (1997) for Spanish):

(56) * banu cr‘abiid l-ahraam
built slaves the-pyramids

Let us for the moment abstract away from the ungrarnmaticality of (56) and attempt to
formulate a context question that would be appropriate for such a sentence. Recall from
the brief discussion above that the NSR in PA and Spanish, as argued by Zubizarreta,
“Signs stress to the lowest element in the syntactic tree in terms of c-command.
Therefore, stress (or nuclear stress (N 8)) would be assigned to l-ahraam 'the-pyramids'
in (5 6) since it is lowest in terms of the c-command ordering. According to (the

ungrilnumatical) (56), that the NS is assigned to l-ahraam 'the-pyramids' means that

I‘ahraam should be, or is, the non-presupposed part of the structure (i.e. the asserted or

 

\

Onl
(if; 8“"383 in the sentence:
5‘ ABIID banu l-ahraam.

ll8

C0nl<
Zubi;

quest

1m]! 1
that 3i
come

We

belc

rhematic part of the sentence). In line with the version of focus assumed in this work,
I-ahraam 'the-pyramids' should replace, and be represented by, a wh—phrase in the
context question as in (57a) or a variable in the logical representation as in (57b) (see

Zubizarreta 1998: 5). The prediction is that (57a) should therefore be the context

question for (56). However, (57a) is ungrarnmatical:

(5 7) a. *suu banu ci’abiid?
what built slaves

b. there is an x, such that slaves built x
the x, such that slaves built x = the pyramids

I will return in a moment to the source of the ungrarnmaticality of (573). For now, notice
that Since the context question in (57a) is ungrarnmatical, we need to seek an alternative

Context question for (56). Let me assume, for the sake of argument, that the correct or
felicitous context question for (56) is (58), an assumption I will revise immediately

below;

(5 8) miin banu l-ahraam?
who built the-pyramids

However, the context question for (56) could DOt POSSibly be (53)- To understand why

h‘s ‘3 the case, notrce the contradrctron or conﬂrct 1n the information status of l-ahraam

the‘pyrarnids' in the context question and its status in the answer: l-ahraam 'the-

pyl'al'nids' would be theme (i.e. presupposed) in the context question (i.e. (58)) and

r .
% (Le, focus) in the answer (i.e. (56)). Recall that in (56) l-ahraam would be

119

assigned the NS, in virtue of being most deeply embedded or lowest in the chain of c-

command, which would mean that it is the focus or the rheme.
Support for the infelicity of (58) as a possible context question to (56) comes from

the fact that the only possible answer (with a full sentence) to (5 8) must in fact be (593)

(placing the so-called narrow focus on the subject):

(59) a. banu l-ahraam Cr’abiid
built the-pyramids slaves
'Slaves built the puramids'

b.1—ahraam, banu-bin cr’abiid

the-pyramids built-3P-3P slaves
'The pyramids, slaves built them'

Alternatively, left-dislocation can be employed to mark the thematic or topical status of
l-ahraam 'the-pyramids' as in (59b). Going back to (59a), the non-presupposed (focal)

Constituent (fabiid 'slaves' in this example, which could be replaced by a variable such

as x t0 indicate its focal status: 'the x, such as x built the pyramids') would be assigned
the NS- Accordingly, there would be no conﬂict between the focal status of l-ahraam
'the‘PyTamids' in (5 8) and (59a) since it would be thematic (i.e. presupposed) in the
former (i.e. the context question) and also in the latter (i.e. the answer). Nor would there

be 001111 ict in the information status of Fabiid 'slaves' because it would be asserted

(i.e. thematic or non-presupposed) in both the context question and the answer. It

th . . . .
erefore seems that an impasse has been reached; no felrcrtous context questron could

so far be found for the utterance in (56).

120

To begin to address our predicament let us keep abstracting away from the
ungrarnmaticality of (56), repeated her as (60). Now, to reiterate, (60) could not be an
answer to the context question (57a), repeated here as (61), since ((5 7a)/(6l) is
ungrammatical. Since we have exhausted all the possible context questions for (60), and

since Yabiid 'slaves' seems to only be subject to a focal reading as the ungrarnmaticality

of (61) illustrates (see below), it follows then that (60) has to be an answer to a context

question in which fabiid 'slaves' cannot be presupposed or thematic; it has to be the

focus or the rheme. Therefore, (58), repeated here as (62), seems to be the only possible

or compatible context question to (60) (however, see below for a modiﬁcation in the

discussion of examples (67)-(70))2

(60) * banu S‘abiid l-ahraam
built slaves the-pyramids

(6 l ) *suu banu Sabiid ?
what built slaves.

(62) miin banu l-ahraam?
who built the-pyramids

In (62) l-ahraam 'the-pyramids' is the theme or the presupposed constituent; it should
not be assigned the NS in the answer in (60). Not only is the ungrarnmaticality 0f (60),
With the NS falling on I-ahraam 'the-pyramids,’ explained, but it is also expected. There
would be two ways (60) could be handled so as l-ahraam 'the-pyramids' would not be

assigned the NS. First, I-ahraam could be right-dislocated (i.e. outside the clause), with

 

l2]

S’abiid 'slaves' being assigned stress by the NSR. It would be necessary in this case for

l—ahraam 'the-pyramids' to be (right)-dislocated so that no contradiction would ensue
between its status in the context question ((62)), as theme, and its status in the answer
((60)), as rheme. The second option for I-ahraam 'the-pyramids' is to be inside the

clause (i.e. not dislocated) and Yabiid 'slaves' would crucially have to be focused

(contrastively, see Chapter 5) as in (63);”

(63) banu YABIID l-ahraam
built SLAVES the-pyramids

In the ﬁrst option where l-ahraam 'the-pyramids' in (60) would be right-dislocated, and
under Kayne’s (1994) assumptions of phrase structure rules prohibiting right-adjunction,
l-ahraam would arguably left-adjoin to IP in TopicPhrase (l-ahraam being topical), with

the Subsequent movement of banu f’abiid 'built slaves' to Focus Phrase in order to get
the word order facts correctly.26 Or, rather, Pabiid 'slaves' only would move to F P while

banu 'built' would move into a TopicPhrase since it is thematic or topical. However, I
tend to think that I-ahraam 'the-pyramids' may actually be in-sz'tu, and not right-

diSIOCated, and that Yabiid 'slaves' is contrastively focused. The choice is ultimately

 

\

25 ,

”$2115 Obviously contradicts Suﬁer’s claim (§ 3.4.1.1) that transitive verbs cannot participate in

foe ernational sentences. I believe that transitives can be presentational as long as the BN is assigned
us by the NSR or otherwise contrastively focused (for the differences between informational and

c tra .
26° :1 the focus see Kiss (1998)).
am assuming here Rizzi's (1997) structure of the left periphery where a focus phrase/projection (F P) is

ﬂ
anked on either side by a topic phrase/projection (TP) (see Chapter 5).

I22

theory-intemal. More about this will be said in the Chapter 5.

Returning to (57a)/(61) above, I believe the source of the ungrarnmaticality of this
context question stems from the fact that it has two focuses: the wh-phrase and,

according to the analysis being presented here, the focused BN, Pabiid 'slaves.‘ Such a
context question cannot be salvaged by contrastiver focusing Pabiid 'slaves' since in

this case the context question would again have two focuses: the wh-pronoun and

YABIID 'slaves.’ This line of reasoning is also conﬁrmed by the possibility of

contrastively focusing a bare nominal in a question where there is no wh-pronoun as in

(64):

(64) Q. banu l-ahraam BANAAT?
built the-pyramids GIRLS

A. la?.banul-ahraarn cr’ABIID
no built the-pyramids SLAVES

The main point that emerges from the discussion so far is that the BN is always focused.

The Sarne analysis can be extended to sentences containing PPS as in (65):

(65) *Ya3o wlaad Ya-l-beet
came boys to-the-house

(65) is Ungrammatical since the bare plural wlaad 'boys' occurs in the sentence under

no . . . .
I‘Inal conditions of stress. Thrs sentence could not be an answer to a context questron

123

such as (66) since the question is ungrarnmatical:

(66) *ween ?a30 wlaad?
where came boys

Again the unacceptability of (66) can be explained in a fashion similar to that argued for
(61) above: the utterance in (66) is out since it has two focuses. The only felicitous

context question that is possible for (65) is (67):

(67) miin ?a30 cIa—l-beet?
who came to-the-house

Ya-l-beet 'to-the-house' is thematic (presupposed) in the context question (67). It may

therefore not be assigned stress by the NSR in (65). It follows, as in the analysis for

(60), that Fa-l—beet 'to-the-house' in (65) has to be either right-dislocated, in which case

wlaad 'boys' would be assigned stress by the NSR, or, alternatively, wlaad by itself is

focused contrastively as in (68):

(68) ?ad30 WLAAD Ya-l-beet
came BOYS to-the-house

However, adopting the latter option for WLAAD 'BOYS,’ where it would be

c0ntrastively focused as in (68), may not be the optimal choice. Notice that a more

124

felicitous answer for the context question in (67) would in fact be (69):

(69) Yadgo Ya-l-beet wlaad
came to-the-house boys

As will be explained in more detail in the Chapter 5, in VOS order sentences the subject
would be assigned narrow focus (i.e. as an answer to who V-ed?). To count as a
felicitous answer to (67), where the subject is assigned narrow focus, the preferable
constituent order for the answer would need to be VOS (as in (69)). However, if we
were to adopt the second option where WLAAD 'BOYS' as in (68) would be

contrastively focused, a good context question for (68) would then be (70) or (70'):

(70) ?a30 BANAAT Ya-l-beet?

came GIRLS to-the-house
'Did GIRLS come to the house?’

(70') ?a30 Ya-l-beet BANAAT ?

came to-the-house GIRLS
'Did GIRLS come to the house?’

At this point the following question can be posed: why is the context question in
(71) grammatical? This is so in spite of the fact that the context question apparently has
two focuses, the wh-pronoun and the bare nominal, and should therefore be predicted to

be ungrarnmatical:

125

(71) miin saaf WLAAD?
who saw BOYS

I think the answer to the question is pretty straightforward. In line with my assumptions
so far, (71) is acceptable with WLAAD 'BOYS' focused contrastively and would be
unacceptable if the bare noun is unfocused. However, the reading we get in the question
is M reading similar to the pair-list reading we get in questions such as who saw
what? in which there is only one focus consisting of pairs: John saw a cat, Bill saw a
lion, etc. (Zubizarreta 1998). So, the dictum that every sentence should have one, and
only one, focus, formally expressed in Rizzi’s (1997) assumption that a sentence can
have multiple TPs (Topic Phrases) but only one FP (Focus Phrase) (see Chapter 5), is
not violated by our example.

To sum up, I have argued for a uniform account of BNS in PA. The picture that
emerges is very restrictive indeed, particularly when we consider that this analysis, I
believe, can happily be applied to Spanish (and English, see Chapter 4). The
information structure of the sentence proves vital in deciding the acceptability of BNs
since the latter have to always be focal.

Recall from Chapter 2 that I have argued that plural BNs are unmarked for the
(in)defmiteness features necessary to check the strong +/- Deﬁniteness features of the D
position. This argument predicts that singular bare nominals in PA could be afforded the
same treatment since these nominals occur determinerless. This prediction is actually
borne out. In the next section I wish to argue that singular BNS in PA pattern like plural

BNs in this language in being focused regardless of their structural position.

l26

3.4.1.5 Singular Bare nominals in PA

PA does not have an indeﬁnite article. Singular indeﬁnites in Arabic are therefore
determinerless, unlike the case in English, for example. I have argued in Chapter 2 that
plural bare nominals in non-CS constructions are ungrarnmatical since they lack the
(in)defmiteness features necessary to saturate the D position. I have also illustrated there
how bare singular indeﬁnites participate freely in construct state constructions, on a par
with bare plurals. If the D position is where the (in)defmiteness features reside, the
expectation would then be that bare singular in PA would not be able to raise to D in
non-CS constructions to check the strong features of that position since these nominals
would lack (in)deﬁniteness features. Bare singulars would therefore have to be focal in
the same way bare plurals are.

This prediction is conﬁrmed by the empirical data. Abstracting away from the
singular-plural distinction, singular BNs in PA display a similar behavior like that of
plural ones: both have identical distribution and the same interpretation as well. It
follows that singular BNs too should be subject to being focused, irrespective of their

structural position. This is what we see in the following examples:
(72) *walad ribih l-d3aa?ize
boy won-3M8 the-prize

(73) *ribih walad l-d3aa52ize
won-3M8 boy the-prize

(74) ribih l-d3aa?ize walad

won-3M8 the-prize boy
'A boy won the prize.’

127

(75) WALAD ribih l-dgaa‘ﬁze
BOY won-3 MS the-prize

(76) ribih WALAD l-dgaa?ize
won-3M8 BOY the-prize

The BN walad 'boy' is unacceptable under normal conditions of stress as in (72) and
(73). However, it is acceptable with a narrow focus reading as seen in (74) and the BN
is assigned stress by the NSR (being most deeply embedded or lowest in terms of the c-
command ordering). If the BN is not lowest in terms of the c-command ordering,
whether occurring pre- or post-verbally, the reading the BN receives is contrastive
focus, as seen in (75)-(76). This overall behavior is essentially the same as that noted
with respect to plural BNs in the previous section and the same arguments can be
constructed for singular BNs to show their focal status on a par with plural ones.

The discussion so far predicts the correctness of the following conceptual statement:
since singular and plural BNs have to be focused, and since a sentence cannot have
more than one focus, contrastive or otherwise, then no two BNS, singular or plural or a
mix of the two, should be able to co—occur. The correctness of this statement is

empirically attested. Consider the following:

(77) a.*§aaf walad banaat
saw boy girls
b. *saaf banaat walad

saw-3M8 girls boy

c. *WALAD saaf banaat
BOY saw-3M8 girls

128

d. *waladsaaf BANAAT
boy saw-3M8 GIRLS

e. *saaf WALAD banaat
saw-3M8 BOY girls

f. *saaf walad BANAAT
saw-3M8 boy GIRLS

g. *saafu wlaad banaat
saw-3MP boys girls

h. *saaf banaat wlaad
saw-3S girls boys

Any combination of singular or plural BNs, as seen in (77), is unacceptable.27 Since a
sentence can have no more than one focus, two BNs of any kind are unacceptable in the
same sentence, even if one of them is contrastively focused as in (77c-f). Provided that
BNs in PA are always focal, a straightforward explanation for this could be that a
sentence can never have more than one focus that attracts all and the only stress in that
sentence. One way to cash this out would be to say that a focused BN would occupy a
Focus Projection and when two PA BNS co-occur they would be basically vying or

competing for the same position, presumably that of [Spec, F P] (cf Chapter 5).

3.5 Modiﬁcation and Coordination of Bare Nominals
In Longobardi (1994) it is assumed that BNS can be allowed to occur in preverbal

positions if modiﬁed, coordinated, or, of course, focused. No satisfactory explanation

 

27 (77a) could have an acceptable reading where the sentence would have a pair-list reading as in the
following English paraphrase: a boy saw girls, a(nother) boy saw old ladies, etc.

129

for why modiﬁcation or coordination allow BNS preverbally is available in that analysis;
quite the opposite, in fact, since modiﬁed or coordinated BNS stand as exceptions (or,
violations) to Longobardi's analysis.

I have argued in § 3.4.1.4 and § 3.4.1.5 that BNs in PA have to be focal, irrespective
of their structural position. Within the system of assumptions I have argued for so far, in
which a lexical government requirement has no place as a licenser, the fact that
modiﬁed or coordinated BNs can occur preverbally may be explained in one of two
ways: either BNs, when modiﬁed or coordinated, are no longer bare, therefore not
subject to being obligatorily focal; or that modiﬁed and coordinated BNs are in fact
subject to the same restriction as non-modiﬁed, non-coordinated, BNS, namely, being
focused. Evidence in either direction has to be sought on empirical grounds, since
conceptually either hypothesis seems to be plausible. However, the latter hypothesis,
namely, that BNs, even if modiﬁed or coordinated, have to be focal, seems to me to be
more appealing. First of all, it should be noted that if the second hypothesis is adopted it
will guarantee us a greater degree of uniformity in dealing with the distribution of BNs.
Focalizing a BN would in that case be an across-the-board kind of constraint BNs are
subject to. Second, the consensus in the literature (for example, Longobardi (1994),
Carlson (1977a, b)) seems to take bare nominals to be those that are determinerless (i.e.
not determined by, for example, an article or quantiﬁer). In other words, BNs with
adjectival modiﬁcation (i.e. pre-modiﬁcation (at least in English)) are still considered to
meet the 'bareness' criterion, so to speak. It would be kind of surprising for the same not
to carry over to post-modiﬁcation, i.e. modiﬁcation with a relative clause (see

§ 3.5.1.4). If what has been said so far turns out to be correct, then BNs in PA or

130

Spanish are uniformly and obligatorily subject to being focal, regardless of whether or
not they are modiﬁed (or coordinated). I believe the empirical evidence points to the
correctness of this hypothesis: BNs have always to be focused, regardless of
modiﬁcation or coordination, or else they would be unacceptable, again, irrespective of
their structural position.

One of the tests for 'topic-hood' in English has been argued to be 'leﬁ-dislocation'

(Reinhart (1981)). For example, Felix in (78) and Matilda in (79) would be topics:

(78) Felix, it's been ages since I've seen him.

(79) As for Matilda, she can't stand him.

Topicalization, which differs from left-dislocation in lacking a pronoun referring back
to the topicalized argument (i.e. anaphoric pronoun), can also mark a topic (Reinhart:
(1981)). In this case, however, there would not be an anaphoric pronoun, but rather a

gap, co-indexed with the topicalized argument:28
(80) Your second proposal, the board found t unfeasible.
Another test I have made crucial use of in § 3.4.1.4 has been the context-

question/answer pairing. The focused part of a sentence corresponds to, or substitutes

for, the wh-phrase in the context question. So, for example, in (81) l-madiine 'the-city'

 

2“ (78)-(80) are taken from Reinhart (1981).

131

would be part of the presupposition of the context question (and therefore of the answer)
while miin 'who' stands for the focus of the context-question and would be substituted

by the focused material in the answer, d3'unuud 'soldiers' in (82):

(81) miin darnmaru l-madiine?
who destroyed-3MP the-city
'Who destroyed the city?’

(82) darnamru l-madiine d3unuud

destroyed-3MP the-city soldiers
'Soldiers destroyed the city.‘

I have used this last test to demonstrate that BNs in PA are always focalized in that they
can never be part of the presupposition of the context-question or its answer. The same
can be shown to hold for modiﬁed or coordinated BNs. In other words, it can be shown
that modiﬁed or coordinated BNs can not occur in the presupposition of the context-
question or the answer. The application of the ﬁrst two tests, left-dislocation and
topicalization, produce a similar outcome. To an illustration of how these tests interact

with the modiﬁcation and coordination facts I now turn.

3.5.1 Adjectival Modiﬁcation
3.5. 1. 1 Intersective/ Non-Intersective Adjectives

First, let us consider what effect, if any, the intersective/non-intersective distinction
in adjectives has on BNS. Larson (1998) exempliﬁes the ambiguity between intersective

and non-intersective reading by the following sentence:

I32

(83) Olga is a beautiful dancer.

(83) is ambiguous between these two readings as paraphrased in (84) and (85) (Larson,

1995: 1):

(84) a. 'Olga is a dancer and Olga is beautiful'

b. 'Olga is beautiful as a dancer'/‘Olga dances beautifully'

The intersective reading is exempliﬁed by (84a) which means that Olga is both a dancer
and she is also beautiful. However, beautiﬁil only applies to her person, excluding her
dancing, which could be far from beautiful. In the second reading in (84b), beautiﬁd is
said to hold of Olga's dancing, not of Olga herself. This is the non-intersective reading.
Modifying PA BNs by intersective adjectives does not license them under normal

conditions of stress:

(85) *muhamiin mashuriin ribhu d3aa?izit nobil
lawyers famous won-3MP prize Nobel

Applying any of the aforementioned topic-hood tests to the BN muhamiin mashuriin

'falnous lawyers' would illustrate the latter's unacceptability as a topic. First, in a

question, this BN cannot be part of the presupposition as in (85a):

133

(85) a. *suu ribhu muhamiin mashuriin ?
what won-3MP lawyers famous

'what did famous lawyers win?’
Second, mubamiin mashuriin 'famous lawyers' cannot occur in left-dislocated positions
as in (85b) or (85c):

(85) b. *bilnisbe la muhamiin mashuriin, humme ribhu d3aa?izit nobil
as for lawyers famous they won-3MP prize Nobel

(85) c. *muhamiin mashuriin, (humme) ribhu d3aa?izit nobil
lawyers famous (they) won-3 MP prize Nobel

Topicalizing the BN is not possible either as (85d) attests:

(85) d. *muhamiin mashuriin, ma-suﬁ-is t ﬁ-l-mahkame
lawyers famous neg-saw-lMS-neg in-the-courthouse

Adjectival modiﬁcation of BNs with non-intersective adjectives cannot license the BN

preverbally unfocused either:

(86) a. *muhamiin saabqin ribhu d3aa1’izit nobil
lawyers former won-3MP prize Nobel

The unacceptability of unfocused BNs with non-intersective adjectives is clearly

illustrated by their failure to be topics:

134

(86) b. *suu ribhu muhamiin saabqin?
what won-3MP lawyers former

c. *bilnisbe la muhamiin saabqin, humme ribhu d3aafi’izit nobil
as for lawyers former they won-3MP prize Nobel

d. *muhamiin saabqin, (humme) ribhu d3aa?izit nobil
lawyers famous (they) won-3MP prize Nobel

Topicalizing the BN is not possible either as (86c) clearly illustrates:

(86) e. *muhamiin saabqin, ma-suﬁ-is t ﬁ-l-mahkame
lawyers famous neg-saw-lMS-neg in-the-courthouse

It is clear, then, that modiﬁcation, whether by an intersection or non-intersective
adjective, does not play any role in 'licensing' BNs preverbally. What is of consequence

in effecting such an outcome is focalization of the BN.

3.5.1.2 Ethnic Adjectives

Modiﬁcation with ethnic adjectives does not license BNs under normal conditions

of stress:

(87) a. *muhamiin ‘Burdiniyiin ribhu d3aa?izit nobil
lawyers Jordanian won-3MP prize Nobel

b. *suu ribhu muhamiin flurdiniyiin?
what won-3MP lawyers Jordanian

135

c. *bilnisbe la muhamiin ?urdiniyiin, humme ribhu d3aa?izit nobil
as for lawyers Jordanian they won-3MP prize Nobel

d. *muhamiin ?urdiniyiin, (humme) ribhu d3aa1’izit nobil
lawyers Jordanian (they) won-3MP prize Nobel

e. *muhamiin ?urdiniyiin, ma-suft-is t ﬁ-l-mahkame

lawyers Jordanian neg-saw- l MS-neg in-the-courthouse

3.5.1.3 Cardinal Adjectives

The same pattern can also be seen when BNs are modiﬁed by cardinal adjectives:

(88) a.*9ala9 muhamiin (?urdiniyiin) ribhu d3aa?izit nobil
three lawyers (Jordanian) won-3MP prize Nobel

b. *suu ribhu GalaG muhamiin (i’urdiniyiin)?29
what won-3MP three lawyers (Jordanian)

c. *bilnisbe la GalaG muhamiin (?urdiniyiin), humme ribhu d3aa?izit nobil
as for three lawyers Jordanian they won-3MP prize Nobel

d. *Galae muhamiin (i’urdiniyiin), (humme) ribhu d3aa1’izitnobil
three lawyers Jordanian (they) won-3MP prize Nobel

 

29 As noted in § 2.3 from Chapter 2, strict adjacency is not always observed between the deﬁnite article in
PA and the head noun. Also when a NumP intervenes between the article and the head N the head N is
read as deﬁnite, although it cannot bear the deﬁnite article. This gives rise to the following pattern of
(un)grammaticality with modiﬁed BNs as in (1)—(iv):

(i) *S‘uu ribhu l-GalaB muhamiin ?urdiniyiin?
what won the-three lawyers Jordanian

(ii) ‘suu ribhu l-GalaG l-muhamiin ?urdiniyiin?
what won the-three the-lawyers Jordanian

(iii) ‘suu ribhu l-Galae l-muhamiin l-?urdiniyiin?
what won the-three the-lawyers the-Jordanian

(iv) sun ribhu l-Galae muhamiin l-?urdiniyiin?
what won the-three lawyers the-Jordanian

136

e. *GalaG muhamiin Purdiniyiin, ma-suﬁ-is t ﬁ-l-mahkarne
three lawyers Jordanian neg-saw-l MS-neg in-the-courthouse

The cardinal can also appear postnominally. The same remarks can be said concerning
the postnominal modifying cardinal.30 3 ' The BNs in the following examples are

crucially read under normal conditions of stress (i.e. not focused), hence the

ungrarnmaticality:

(89) a. * muhamiin (flurdiniyiin) Galaee ribhu d3aa?izit nobil
lawyers (Jordanian) three won-3MP Prize Nobel

b. *suu ribhu muhamiin (?urdiniyiin) GalaGe?
what won-3MP lawyers (Jordanian) three

c. *bilnisbe la muhamiin (?urdiniyiin) GalaGe, humme ribhu d3aa?izit nobil
as for lawyers Jordanian three they won-3MP Prize Nobel

d. *muharniin (?urdiniyiin) Galaee, (humme) ribhu d3aa?izit nobil
lawyers Jordanian three (they) won-3MP Prize Nobel

 

3° It must be noted, however, that in the case of postnominal modifying cardinals the cardinal may more
likely be read predicatively, rather than attributively. Therefore, the grammatical counterpart of (89a)
where the BN would be focused, the meaning of the sentence would roughly be "Jordanian lawyers who
are three, won the Nobel Prize."
3 ' Dual NPs behave in identical fashion when bare as seen in (i)-(iv):
(i)"‘waladiin kasar-u l-gazaaz
boys (dual) broke-3MP the-glass
(ii)*§uu kasar-u waladiin ?
what broke-3 MP boys-dual
(iii)"‘shuu kasar-u waladiin ‘i’urdiniyiin Gniin?
what broke-3MP boys-dual Jordanian two
(iv)*bilnisbe la waladiin, humme kasaru l-gazaaz
as for boy-dual they broke-3 MP the-glass
This shows that the relevant licensing element in PA has to do with the presence or absence of the deﬁnite
article.

137

e. *muhamiin Purdiniyiin GalaGe, ma-suﬁ-is t ﬁ-l-mahkame
lawyers Jordanian three neg-saw-lMS-neg in—the-courthouse

3.5.1.4 Relativization and Bare Nominals

Modiﬁcation by gerundive reduced clauses does not license unfocused BNs:

(90) a. *wlaad gandiin ﬁ l-gahwe xisr-u kull maSarii-hum

boys sitting at the-coffee shop lost-3MP all money-3MP
'boys sitting in the coffee shop lost all their money'

Applying any of the usual topic-hood tests would expectedly yield ungrarnmatical

results:

(90) b.*suu xisr-u wlaad gai‘diin ﬁ l-gahwe?
what lost-3MP boys sitting at the-cafe

c. *bilnisbe la wlaad gaidiin ﬁ l-gahwe, Omar haka maYhum
as for boys sitting in the-coﬁ‘ee shop, 0. talked to them

(I. * wlaad ga‘ifdiin ﬁ l-gahwe, (humme) xisr-u kull maSarii-hum
boys sitting at the-coffee—shop (they) lost-3MP all money—3MP

e. * wlaad gaidiin ﬁ l-gahwe, Omar ma-laaha5-is t
boys sitting in the-coffee-shop O. neg-noticed-3MS-neg

If Kayne (1994) is correct in arguing that postnominal adjectives are reduced
relatives, and since modifying adjectives as has been remarked above cannot license

BNs unfocused, then the prediction is that relative clause modiﬁcation should not be

138

able to license BNs.32 This prediction is empirically conﬁrmed. Modiﬁcation by
reduced relatives or indeﬁnite relatives cannot license BNs under normal conditions of

stress:

(91) a. *Sabiid maYruﬁin b-guwwit—hum banu l-ahraam
slaves known for-strength-3P built-3P the-pyramids

b. *suu banu Qabiid maimﬁin b-guwwit-hum?”
what built-3P slaves known for-strength-3P

c. *bilnisbe la Sabiid maS‘ruﬁin b-guwwit-hum, (humme) banu l-ahraarn
as for slaves known for-strength-3P (they) built-3P the-pyramids

(92) a. *S’ulamaa? ribhu d3aa?izit nobil hiDru l-i’idgtimaaﬁf
scientists won3-MP Prize Nobel attended the-meeting

b. *suu hiDru S’ulamaa? ribhu d3aa?izit nobil?
what attended-3MP scientists won-3MP Prize Nobel

c. *bilnisbe la cr‘ulamaa? ribhu d3aa2izit nobil,
as for scientists won-3MP Prize Nobel
(humme) hiDru l-Yid3timaa9

(they) attended-3MP the-meeting

At this point it is perhaps revealing to note that in all the examples cited above
where BNs are unacceptable in topic positions grammaticality is restored once the BN is

modiﬁed by the deﬁnite article. Consider, for example, what happens when the BNs in

 

32 Recall from Chapter 2 that Fassi Fehri (1999) argues that Arabic is an A-N language, where adjectives
occur prenominally, rather than N-A language and that the observed surface order for Arabic (i.e. N-A) is
derived.

33 A marginal reading that is available here is that of a contrastive nature where a pair-list juxtaposition is
available. So the pair-list reading would be what slaves who are known for their strength built contrasted
with what slaves who are known for their weakness built, for example.

139

(92) are determined by the deﬁnite article:

(92)a. l-S‘ulamaa? 21111 ribhu dsaa‘ﬁzit nobil hiDru l-ridgtimaas“
the-scientists that won3-MP Prize Nobel attended the-meeting

b. suu hiDru l-S‘ulamaa? ?illi ribhu dsaa?izit nobil?
what attended-3MP the-scientists that won-3MP Prize Nobel

c. bilnisbela I-9ulamaa? ?illi ribhu d3aa?izit nobil,

as for the-scientists that won-3MP Prize Nobel
(humme) hiDru l-‘l’id3timaa‘r’
(they) attended-3MP the-meeting

3.5.2 Coordination and the Licensing of Bare Nominals

Modiﬁcation and coordination have remained something of an idiosyncrasy in the
behavior of BNS in the analyses reviewed earlier, that of Longobardi (1994) and
Casielles~Suarez (1997). These phenomena, i.e. modiﬁcation and coordination, stand
out as violations or exceptions to the generality of these analyses. As has been remarked
in the previous sections, modiﬁcation, which Longobardi's analysis does not address in
any satisfactory manner, is to be viewed in the approach I argue for in this work as non-
exceptional in any way. Whether bare nominals are modiﬁed or not is in fact to be seen
as inconsequential to how BNS behave generally (in particular, how they distribute in
sentences).

In a similar fashion, I wish to argue that coordinated BNS are quite unremarkable in

that they do not present us with any need to make new, additional assumptions. Taking

 

3" Note that with deﬁnite relative clauses the presence of Pilli 'that' is obligatory, unlike in the case of

indeﬁnite relative clauses where it is disallowed.

140

the tests proposed in Reinhart (1981) for topic-hood to be correct, as I have assumed so
far, coordinated BNS can be shown to be unavailable as candidates for topical readings.
To begin with, coordinated BNs cannot be part of the presupposition of the context

question as in (93b) or the answer as in (933):

(93) a. *zlaam w niswaan hiDru l-?id3timaa&’
men and women attended-3P the-meeting

b. *suu hiDru zlaam w niswaan?
what attended men and women

Coordinated BNS also fail the "As for.." test as the following example attests:

(93) c. *bilnisbe la zlaam w niswaan, (humme) hiDru l-‘Pidgtimaa?
as for men and women (they) attended-3P the-meeting

Nor do coordinated bare nominals pass the left-dislocation or the topicalization tests:

(93) d. *zlaam w niswaan, suit-hum ﬁ l-?id3timaa&‘
men and women saw-1M8 at-the-meeting

e. *zlaam w niswaan,suﬁ t ﬁ l-?id3timaa9
men and women saw-lMS at-the-meeting

Again, as a reminder, what should be kept in mind here and elsewhere above is that the
BNs in these sentences are crucially not intonationally marked as focused. This

reminder is important since an utterance such as (93c) would be grammatical if the

MI

coordinated BNs are contrastively focused as in (94):

(94) ZLAAM W NISWAAN suft t ﬁ l-?id3timaa§‘,

MEN AND WOMEN saw-MS at-the-meeting

mis WLAAD W BANAAT

not BOYS AND GIRLS

'MEN AND WOMEN I saw at the meeting, not BOYS AND GIRLS.’

Example (94) seems to me to be an instance of focusing, not left-dislocation, where the
contrastively-focused coordinated bare nominals are co-indexed with a gap in the site of
extraction.

It is perhaps worth noting here that Casielles-Suarez (1997: 105-107) considers non-
contrastive (-ly focused) BNs in left-dislocated positions grammatical in Spanish and
she assumes that they are topical (an assumption, C-S points out, she makes contrary to
Suﬁer (1982)). Therefore, the following example, which C-S takes from Sufier (1982), is

an instance of a left-dislocated topical BN for C-8:

(95) Petroleo, no surgio pero si agua
oil not bubbled-up-3 sg but yes water
'Oil did not bubble up but water did'

C-S claims that this example contains a topical BN in left-dislocated position. However,
she notes that left-dislocated BNS are characterized by the lack of a clitic, unlike
genuine or typical cases of left-dislocation which are co-indexed with an anaphoric
pronoun or a clitic (1997: 108). As a way out, C-S argues that the absence of the clitic in

this case is closely tied to the very nature of what is being left-dislocated; in this case, a

142

BN. To illustrate the point she cites Vallduvi (1988) who points out the impossibility of

a clitic when BNs are dropped, as seen in (96) (her example, (39)):

(96) &Trajiste tenedores? Si, si traje
brought-23g forks yes, yes brought-15g
'Did you bring forks?’ *yes, I brought

C—S's claim is basically that only DPs trigger the appearance of a clitic. BNS in Spanish,
she argues (p. 109), are NPs, rather than DPs. It then follows that they do not trigger the
appearance of a clitic. To illustrate this point she provides the following examples (her

40, 41, 42, and 43, respectively):

(97) a. *Los libros, no traje

b. Los libros, no 103 traje (original emphasis)
'The books, I didn't bring them'
(98) -&Trajiste Los libros? a.-*No, no traje
'Did you bring the books?’ b.-No, no 103 traje 'No, I didn't'

(99) Libros, no traje
(100) -&Trajiste Iibros? -No, no traje

I believe C—S's analysis is incorrect. First of all, I think her judgment, contrary to Suﬁer's
(1982) who considers them contrastive, that BNs can appear left-dislocated as topics is
not correct. I have shown in the previous sections that PA BNS cannot occur topical in
sentences. It seems therefore unlikely that Spanish BNs, whose behavior is almost
identical to PA BNs, should be drastically different in terms of this interpretive aspect.

Second, C-S's claim that the failure of BNs in Spanish to trigger the appearance of a

143

clitic is due to their NP-status is not justiﬁed and seems to be the incorrect assumption.
Before I discuss this point let me note that the same facts expressed in (97)-(100) also
hold in PA as the examples below, which are equivalent to the Spanish examples,

clearly attest:

(101) a. *l-kutub, ma d3ibtis
the-books neg brought-1 MS-neg
”'the books, I didn't bring'
b. l-kutub, ma d3ibt-hum-is

the-books neg brought-l MS-3P-neg
'the books, I didn't bring them'

(102) d3ibit l-kutub?

brought—3S the-books
'did you bring the books?’

a. *la?, ma dsibtis

no, neg brought-1 S
'*no, I didn't bring'

b. la?, ma d3ibt-hum-is

no, neg brought-l S-3P-neg
'no, I didn't bring them'

(103) kutub, ma d3ibti§
books, neg brought-IS-neg

(104) a. d3ibt kutub?
brought-28 books
'did you bring books?‘
b. la?, ma d3ibtis

no, neg brought-1 S-neg
'*no, I didn't bring'

The empirical facts from PA are identical to those in Spanish. A BN fails to trigger the

144

appearance of an agreement marker or clitic that is anaphorically related. A determined
nominal, on the other hand, does entail the appearance of such a clitic. However, I
believe that C-S's explanation of the facts is ﬂawed. Crucially, as I noted above in

§ 2.3.4, C-S fails to see the effect information structure has on English BNs. Note, in
fact, that the lack of a clitic in Spanish or an agreement marker in Arabic corresponds to

a lack of pronoun in English as (105) illustrates:

(105) a. Did you bring forks?
b. *?Yes, I brought them.

c. Yes, I brought some.

The fact that (105a) cannot be answered with a pronoun anaphoric to the BN forks, or at
best it is marginal, is essentially an identical phenomenon to that noted above for
Spanish and Arabic (i.e. the emergence of an anaphoric clitic or Agr marker). In other
words, the ungrarnmaticality of the English glosses to examples (96) (ﬁves, I brought)
and (104b) (*no, I didn’t bring) stems from the fact that the argument grid of the two-
place predicate is not saturated (i.e. the object 0-role has not been discharged). To argue
otherwise, C-S would have to assume that BNs in English are also NPs, rather than DPs,
since they fail to trigger the presence of an anaphoric pronoun. This is a very
problematic assumption, to say the least. Note also that this assumption is falsiﬁed in
PA by N to D movement in the construct state construction considered in Chapter 2.

On the other hand, if BNs are always focal, as I assume in this work, a natural

145

explanation for the observed facts can be found. Since BNS are always focused,35

i.e. non-topical, they resist being anaphorically related to a pronoun in English, or a
clitic in Spanish, or trigger an agreement marker in PA. When non-focal, BN in English
would in fact trigger the presence of a pronoun anaphorically linked to the BN. This is
attested in the case of generically read BNs in English which I argue in the next chapter
to be non-focused (i.e. topical), unlike existentially read BNs. Genetically interpreted

BNs in English, therefore, pattern with determined NPs:

(106) a. Did you bring the forks?
b. Yes, I brought them.
(107) a. Do you like dogs?

b. Yes, I like them.

Once the determined NP the forks is mentioned in the context question (106a), or the
generic BN dogs in (107a), this makes either a possible topic for the ensuing discourse.
forks in (105a), on the other hand, is mentioned in the context—question which should,
on a par with the forks in (106a) and dogs in (107a), make it a possible topic for the
following discourse. However, this is not the case. This immediately follows ﬁom the
inability of existential bare nominals in English to act as topics and the restriction they

are subject to to be focal.36 The difference between English on the one hand, and

 

3’ This is the position I argue for in this work: BNs are always focal even in English, unless generically
read. This issue will be taken up in more detail in the next chapter.

36 It may seem undesirable to assume a divide between generic and existential BNs in English since a
uniform treatment would a priori be more appealing. However, such a distinction seems necessary

146

Spanish and PA on the other, follows from the fact that the object argument has to be
overtly realized in English, unlike the case in PA or Spanish .

The remainder of this chapter will be dedicated to investigating the interpretive
properties of BNs. I illustrate in what follows that the existential interpretation BNS in
PA and Spanish (and (one interpretation of) English BNs) are subject to is quite

expected and follows directly from the fact that these nominals are focused.

3.6 The Interpretation of Bare Nominals
3.6.1 Stage-Level Vs. Individual-Level 81 Generic Vs.
Existential

Cross-linguistically, BNs have been noted to possess one, or both, of these two
readings: an existential reading and a generic reading (Carlson (1977a, b), Diesing
(1992a, b), Longobardi (1994, 2002), Kratzer (1995), Cohen and Erteschik-Shir (2002),
among others). In Carlson's work this distinction in interpretation follows from a
distinction between predicates in terms of their being either S(tage)-Level or
I(ndividual)-Level predicate. According to Carlson's original theoretical framework,
English BNs are interpreted existentially with S-level predicates, while the generic

reading is reserved for BNs with I-level predicates. This contrast can be seen in the

 

anyway since BNs in languages such as Romance and PA can only get the existential reading. The very
fact that English BNs can also have a generic reading is reﬂective of a deeper distinction with or
difference from Romance and PA. This distinction is taken up in more detail in the Chapter 5.

147

following two examples:

(108) Tickets are available.

(109) Tickets are expensive.

The BN tickets in (108), with the S-level predicate available, would be read
existentially, whereas in (109), with the I-level predicate expensive, it can only be read
generically. However, Carlson's neat dichotomy has been challenged by Diesing (1992a,
b) who argues that S-level predicates such as available are, at least, two-ways
ambiguous. (108) would be ambiguous between an existential reading for tickets,
meaning roughly 'at the moment of speaking there are tickets available for sale,‘ and a
generic reading, meaning that it is a characteristic of tickets to be available.

However, the S-level/I-level distinction has been argued not to be the relevant
distinction in triggering the existential/ generic readings of BNs. Higginbotharn and
Ramch (1996), for example, argue that the difference in the readings we get in
examples such as (109) and (108), existential vs. generic, respectively, cannot be due to
the S-Level/ I-level distinction since the norm for verbal predicates in the simple present
tense is the generic reading. Nor can the transient nature of the predicate be correlated
with the existential reading since subjects of sentences such as Children are
fat/skinny/joﬁizl/sad, with predicates that would be predicted to allow the existential
reading due to their transient character, can only get a generic reading here.
Higginbotharn and Rarnchand (1996) consider the existential reading to be correlated

with predicates that involve spatio-temporal proximity to the speaker. J ager (1999) also

148

rejects the idea that S-Level/I-Level distinction is what is relevant for the interpretation.
He argues that in place of the S-Level/I-Level distinctions there are three contrasts that
need to be distinguished: whether the predicate is permanent or temporary, if it admits a
weak (i.e. existential) reading of its indeﬁnite subject, and whether it can occur as the
complement of a perception verb (1999:74-75). J tiger then considers the intricacies of
one of these contrasts, namely, the weak (existential)/strong (generic) readings of
indeﬁnite subjects. He suggests that the relevant distinction here is the thetic/categorical
(cf Ladusaw (1994), see § 3.6.3.2; McNally (1998)). According to Jager, German
differentiates between thetic and categorical judgments by placing the subject in the
former in-situ (within the VP) while scrambling it in the latter outside of the VP. The
subject in categorical sentences is read as Topic and in thetic sentences as comment. In
thetic sentences where no topic is available the event argument is considered as topic.
Subject effects (weak/strong contrast) are a consequence of a principle requiring every
clause to have a topic. Events, but not states, can function as topics, so stative predicates
cannot accommodate weak subjects ([-WS] in Jager's terminology) if none of their
arguments (including implicit ones) can serve as topic. In other words, if there is an
argument of the stative predicate that can be topic the stative predicate can be [+WS],
otherwise it cannot.

Cohen and Erteschik—Shir (2002) make a similar proposal to that of Jager's.
However, for these authors the S-Level/I-Level distinction is still important, though not
in the traditional, clear-cut sense as expressed in Carlson (1977a, b) and Diesing (1992a,
b). Cohen and Erteschik-Shir argue that it can be demonstrated that existential subjects

can be disallowed with some Stage-Level predicates or that such subjects are allowed

149

with some I-Level predicates. Both results, of course, conﬂict with traditional
conceptions of the S-Level/ I-Level contrast as articulated in Carlson's and Diesing's
works. The subject effects, borrowing J ager's terminology, for Cohen and Erteschik-Shir
(2002), follow from the focus structure of the sentence. I will come back to a more
detailed exposition of these authors' analysis below.

Despite the criticism leveled against the relevance of the original S-Level/ I-Level
contrast I still think that it should not be dispensed with. The distribution of the
interpretational possibilities of nominals in PA turn out to correlate closely with whether
the predicate is Stage or Individual level. To demonstrate this, I would like to consider
Longobardi's (2002) analysis of Italian BNs and compare/contrast the facts observed for

Italian with those that hold in PA. I consider this topic in the next section.

3.6.2 Do PA BNs Have Generic Readings?

English BNs, as is well known, are subject to two kinds of interpretations:
existential and generic. Longobardi (2002) claims that Italian BNs exhibit this
bifurcation of interpretation, although the interpretations in this case distribute
differently than in the case of English BNs. In other words, Italian BNS are generically
read only in 'independently quantiﬁcational environments' (2002: 340), unlike English
where BNs can be generic in all environments. The question that arises here is whether
or not the interpretive facts argued to hold for Italian can also be carried over to PA. In
what follows I consider an answer to this question and I conclude that the answer is in
the negative.

Before I consider Longobardi's analysis, a brief typology of genericity is in order.

150

 

Krifka et a1. (1995: introduction) point out that there are two kinds of generics. The ﬁrst
is what can be referred to as reference to kind as in The potato was ﬁrst cultivated in
South America/Potatoes were introduced into Ireland by the end of the I 7th century.
(their examples (1a, 1b), p. 2). The second type of genericity has to do with
"propositions which do not express speciﬁc episodes or isolated facts, but instead report
a kind of general property, that is, report a regularity which summarizes groups of
particular episodes or facts" (Kriﬂ<a et a1. (1995:2)). The second kind of genericity is
exempliﬁed by sentences such as John smokes a cigar aﬁer dinner and A potato
contains vitamin C, amino acids, proteins and thiamine, which are called characterizing
sentences. The latter sentences make up most of the instances of generic sentences in
English.

Italian BNs, according to Longobardi (2002: 340), have only an existential reading
with episodic Stage-Level predicates.37 In this respect PA patterns exactly like Italian.
38 However, the similarity ceases in characterizing sentences with stage level
predicates. Longobardi (p. 340) claims that Italian BNs can be read generically in this
case while the existential reading is marginal. As (110) and (111) below show, PA BNs
in such sentences cannot be read generically:

(Gen=generic; EX=existential):

(l 10) * ?alaat hadiiGe yimkin tGiir kGiir min l-quuul Gen

tools modern may raise a lot of curiosity
'Modem equipment can raise a lot of curiosity'

 

37 1 model the examples that follow after those offered by Longobardi for Italian.
38 Of course, with the proviso that BNs are focused, even if modiﬁed.

151

(l 11) *sayyaraat beiDah yimkin tkallif ‘Bakear Gen

cars white may cost more
'White cars can cost more'

Nor can a BN in PA be read generically in episodic sentences with a generalizing adverb
with S-Level predicates which Longobardi argues to give rise to a generic reading in
Italian. In fact, not even an existential reading is possible for the PA BN here due to the

presence of 'always.‘

(112) *?alaat hadiiBe/ﬁyala beiDah dayman/Yaaliban 20aarat kBiir min l-quuul

tools modem/elephants white always/often raised a lot of curiosity
'Modem tools/White elephants always/often raised a lot of curiosity'

As far as I-Level predicates go, Longobardi (p. 341) claims that Italian BNs behave non-
uniformly in that they pattern in one of two ways: with 'more eventive' predicates, BNs
are possible with a generic reading while with 'more stative' predicates they are

degraded (his examples, 8, 9, and 10, respectively):

(113) a. Stati di grandi dimensioni sono pericolosi Italian
'States of large size are dangerous.’

b. ??Stati di grandi dimensioni sono prosperi
'States of large size are prosperous'

(l 14) a. Cani da guardia di grosse dimensioni sono pir‘r efﬁcient/aggressivi.
'Watchdogs of large size are more emcient/aggressivef

b.??Cani da guardia di grosse dimensioni sono pi1‘1 pelosi/neri.
'Watchdogs of large size are more hairy/black.’

152

(115) a. Uccelli di zone paludose sono ghiotti di insetti.
'Birds from marshy areas are greedy for insects.’

b.??Uccelli di zone paludose sono semi/intelligenti.
'Birds from marshy areas are dark/intelligent!

Descriptively, Longobardi sums up the readings BNS have in Italian as follows:

(116) S-level a. episodic—> Ex
b. characterizing (habitual aspect) —) Ex/Gen
I-level a. lexically characterizing A (habitual aspect) —> Gen39

b. lexically characterizing B (non-habitual aspect) Q

What is important to note here is that the PA counterparts of(113)-(115) are all
ungrarnmatical. PA BNS are unacceptable with I-level predicates of either group, the

eventive or the stative, as seen in the equivalent examples to (113a-b):

(117) a. *wilayaat had3im-hin kbiir xiTraat
states size-3F P big dangerous-3F P

b. *wilayaat hadsim-hin kbiir mintistaat
states size-3F P big prosperous

No generic reading is available for either (117a) or (1 17b). Not even an existential

 

39 A group corresponds to examples with more eventive l-level predicates that Longobardi characterizes
as being 'habitual.‘

153

reading

a gener

reading is possible either in these examples. The sentences can only be acceptable with

a generic reading with the deﬁnite plural:

(118) a. l-wilayaat ‘Billi hadsim-hin kbiir xiTraat
the-states that size-3FP big dangerous-3FP

b. l-wilayaat ?illi had3im-hin kbiir minti‘r’saat
the-states that size-3FP big prosperous.

In fact, even if the tense in such sentences is changed into a tense with an episodic

interpretation these sentences remain unacceptable as witnessed in (1 19b):

(119) a. * ﬁyala beiDah xitraat/ Baldwat
elephants white dangerous-3 P/ intelligent-3P

b. *ﬁyala beiDah kaan-in xitraat/ 5akiyyat
elephants white were-3P dangerous-3P/ intelligent-3P

PA BNs in object position are also unacceptable with a generic reading, unlike Italian
BNs which are claimed by Longobardi to be acceptable with such a reading in object

position (his 27, 28):

(120) Una buona legge esclude solo cittadini stranieri dal diritto divoto. Gen/Ex
'A fair law only excludes foreign citizens from the right to vote.‘

(121) Amo/Adoro/Mi piacciono arance di grandi dimensioni. Gen
'1 love/adore/like oranges of large size.‘

154

In fact the only way to express a generic reading in (120) and (121) in PA is by using the

deﬁnite article as seen in (122) and (123):

(122) l-qanuun l-‘i’aadel byista0ni bas l-muwaTiniin l-?d3aneb .....
the-law the-just excludes only the-citizens the-foreign...

(123) ?ana bahibb l-burtgaan l-kbiir
I like the-oranges the-big

Finally, PA BNs, just like their Italian counterparts, are impossible with Kind-denoting

predicates such as extinct, widespread:

(124) *ﬁyala beiDah Saru munqaridiin
elephants white became extinct

(125) *ﬁyala beiDa muntisriin
elephants white widespread

If the facts claimed by Longobardi (2002) to hold for Italian BNs do actually hold, it
becomes obvious from this discussion that PA BNs behave differently as far as the S-
level/I-level distinction is concerned.40 As seen above, PA BNS are only subject to an
existential reading in episodic sentences with Stage-Level predicates. The generic
reading is impossible for PA BNs with characterizing S-level predicates or episodic

sentences with generalizing adverbs. PA BNs are also unacceptable with I-level

 

4° Schmitt and Munn (2000) observe that the interpretive facts argued for by Longobardi (2002) do not
hold for Spanish. This would basically suggest that Longobardi's argument cannot even be carried over to
all Romance.

155

predicates or Kind predicates. The picture that emerges here is very streamlined: PA
BNs are only acceptable with an existential reading in sentences with S-level predicates
and are unacceptable with I-level predicates. PA BNS are never possible with a generic
reading, irrespective of modiﬁcation or coordination. Their occurrence is also conﬁned
to sentences with S-level predicates. To achieve the generic interpretation for PA
sentences the use of the deﬁnite article is necessary.

All that has been said, of course, suggests that the S-level/I-level contrast should not
be dispensed with as it proves relevant after all. As a matter of fact, even for some
authors who argue that this contrast may not be the crucial distinction triggering the
weak/ strong readings of indeﬁnite subjects, they still maintain that it becomes relevant
if understood correctly. Cohen and Erteschik-Shir (2002) provide such a perspective.
Their analysis can in fact be instrumental in shedding some light on, or at least
providing an alternative view for, why BNs in PA can only be read existentially, and not

generically. This issue will be taken up in the next section.

3.6.3 Why the Existential Reading for PA BNs?

I have argued above that PA BNs cannot be interpreted generically, pace
Longobardi. The existential interpretation is the only interpretation available for BNs in
PA. I have also maintained that the S-level/ I-level contrast should not be discarded
since BNs in PA can co-occur with the former, but not the latter, which argues for the

relevance of this distinction. This last fact seems to argue that the reason that only the
existential interpretation is available for BNs in PA should be rooted in, or attributed to,

SUch a distinction. In what follows I investigate this possibility. Before I do that,

156

however, I demonstrate that an analysis of PA BNs as variables, as in Heim (1982),
coupled with my assumption in this work that PA BNs are always focused, can still get
us the correct results. It can explain why BNS are without exception interpreted

existentially.

3.6.3.1 BNs as Variables

In § 3.3.3.1 I have illustrated Heim's (1982) treatment of indeﬁnites as variables
unselectively bound. I have also noted that Diesing (1992b) adopts, with slight
modiﬁcations, Heim's analysis of indeﬁnites and carries it over to BNs in German and
English. Abstracting away from Diesing's mapping hypothesis, which has been shown in
§ 3.3.3.2 to be conceptually and empirically problematic, Heim's analysis of indeﬁnites
adopted by Diesing can still give us the correct results in the case PA BNs.

Partee (1991) argues that there is a correlation between the tripartite structure
posited for the logical representation of sentences with indeﬁnite nominals by Heim
(1982) on the one hand, and the topic-focus structure or articulation of the sentence, on
the other hand. More speciﬁcally, Partee points out that the restrictive clause of the
tripartite structure can be correlated with the topic or non-focused part of the sentence,
whereas the nuclear scope can be shown to coincide with the focus of the sentence. For

concreteness, a sentence such as (127) would get the representation at LF as in (128):

(127) Mary always took [John]p to the movies.

157

Restrict r
operator 0 Nuclear Scope

always Ex (Mary took x to Mary took John to the movies at e
e .
(128) the movres at e)

The operator in (127), in this case the adverbial always, would bind an event variable.
John, being the focus of the sentence, would be mapped into the nuclear scope of the
logical representation of the sentence, while the sentence, with the focus replaced by a
variable existentially quantiﬁed, maps into the restrictive clause. Partee brings much
empirical evidence to bear on the main argument, evidence ranging from sentences with
focus sensitive elements such as only and even, counterfactuals, modals, frequency
adverbs, generics, to sentences with emotive factives and attitude verbs.

Now recall that in the original Heim (1982) account existential closure takes place at
two levels: at the textual level and at the level of the nuclear scope. Diesing (1992b)
dispensed with Heim's assumption concerning text-level existential closure. Diesing,
along with Heim, assumes that existential closure (only, in Diesing's case) closes off the
nuclear scope. This assumption is sufﬁcient for my current purpose: nothing really
hinges on whether or not existential closure is operating at the level of the text. Nothing
hinges either on whether or not the domain of existential closure, and that of nuclear
scope, is concurrent with the VP, as in Diesing's MH. However, to make my point, I
make the following assumptions: I assume with Diesing that BNs are, like indeﬁnites,
variables operator-bound (for a different view, see Carlson (1977a, b); I also assume

with Partee that the focus coincides with the nuclear scope and with Heim and Diesing

158

that existential closure closes off the nuclear scope. Now recall that I have argued above
that BNs in PA are always focused. Taking these assumptions in tandem we arrive at the
conclusion that BNs in PA, being focused, would always be mapped into the nuclear
scope, rather than the restrictive clause. Therefore, they would always be within the
scope of existential closure. In other words, they would always be bound by an
existential quantiﬁer/operator having scope over the nuclear scope and would therefore
be interpreted existentially. Note, again, that none of what has been said assumes
Diesing's mapping hypothesis: no assumption is made here that the VP maps into the
nuclear scope at S-Structure or LF, and being inside the VP, BNs are interpreted
existentially.

Looking at matters in this light we are able to accomplish the results we have set out
to accomplish. No generic reading is possible, or even allowed, for PA BNs which
follows from their being always focused and, therefore, always mapped into the nuclear
scope at LF. This effectively rules out the possibility that PA BNs be mapped into the
restrictive clause (and be interpreted generically in the process).

This analysis, then, proves to be viable. In what follows, I would like to consider
another way of achieving exactly the same result. It is important to keep in mind here
the key assumption I have made so far, namely, that BNs are always focused,
irrespective of their structural position. Such an assumption essentially renders any of
these analyses, the one just reviewed and others to be looked at directly, compatible
with an adequate treatment of PA BNs. In the next section I will consider Cohen and
Erteschik-Shir's (2002) approach to BNs in English to see how well it can be rendered

to ﬁt my current goals.

159

3.6.3.2 Cohen and Erteschik-Shir's (2002)

The basic claim of Cohen and Erteschik-Shir (2002) (C&E-S henceforth) is that the
generic/existential readings of indeﬁnite subjects follow from the focus structure of the
sentence. Topics are interpreted generically while focused BNs are interpreted
existentially.

Although the Stage-level/Individual-level distinction is not considered adequate by
C&E-S they do not discard it. What the authors call the Topic Constraint, that every
sentence has to have a topic, follows from the assumption that sentences express
predications over topics. Topics can be subjects or objects, i.e. arguments, as in the

following examples (their examples (14) and (16)):

(129) Johnmp [is clever]poc

(I30) MARY loves John.

However, C&E-S maintain that it is hard to construe adjuncts as topics since topics are
not 'superﬂuous' elements in sentences, as adjuncts may be. This contrast between

argument and adjuncts is seen in (131) and (132):

(131) As for the problem, John will work on it.

(132) ? As for the office, John will work there.

They also point out that topics could be spatio-temporal locatives where the italicized

expressions are read as topics. These locatives can be overt or implicit:

I60

(133) a. There was a narnetag near every plate."1
b. A ﬂag was hanging in ﬁont of every window.
c. A student guide took visitors to two museums.
d. A spectator put contributions in the hat.
e. A ghost appeared at midnight.
f. A student arrived during class.

g. sTOPt [A policeman arrived]mc42

Topics are usually arguments which also holds true for the spatio-temporal locatives in
the sentences immediately above. To make their overall argument, C&E-S draw on the
analysis of the S-level/ I-level distinction in Kratzer (1995), which I brieﬂy review here.
Kratzer (1995) claims that S-level predicates differ from I-level predicates in their
argument structure. The former have an additional spatio-temporal argument that I-level
predicates lack. Crucial evidence for the difference in argument structure comes from

the following examples Kratzer (1995: 129) provides (her 15 a-d):

(134) a. *When Mary knows French, she knows it well.
b. When a Moroccan knows French, she knows it well.

c. When Mary knows a foreign language, she knows it well.

(1. When Mary speaks French, she speaks it well.

 

" C&E-S point out that since topics have to be speciﬁc, singular indeﬁnites in (133) cannot be topics
which makes the locatives topics.

'2 'sTOPICt’ stands for 'stage topic' when the topical element is non-overt. The index '5' indicates the
'spatial' location while the 't' index denotes the 'temporal' localization of the topic.

l6l

Kratzer observes that when an I-level predicate, such as knows in (134a), occurs in the
antecedent of a conditional, and supposing that the antecedent clause of a conditional
maps into the restrictive clause of an operator, along with the assumption that I-level
predicates lack a spatio—temporal argument, sentence (134a) is predictably bad. (134b),
which is minimally different from (134a), is grammatical since the indeﬁnite, a
Moroccan, provides the variable needed for the operator to bind in order to fulﬁll the
principle of Full Interpretation and the prohibition against Vacuuous Quantiﬁcation."3 44
Vacuuous Quantiﬁcation does not result in (1340) since the variable needed for the
operator to bind is provided by a foreign language even though the predicate in this case
is I-level (i.e. 'knows'). The crucial difference for Kratzer is provided by (134d). This
example does not contain an indeﬁnite in the antecedent of the conditional which can
provide the variable needed for the operator to bind. However, the predicate in (134d) is
S-level. If it is assumed that S-level predicates contain an extra spatio-temporal
argument that can provide the variable needed for the operator to bind, then the
grarmnaticality of the sentence is explained. Kratzer (pp.129-l 30) argues for this
position and provides the following logical representations for (134a) and (134d) (her

(15'a) and (15'd), respectively):

(134') a. *Always [knows(Mary, French)] [knows-well(Mary, French)]

(1. Alwaysl [speaks(Mary, French, 1)] [speaks-well(Mary, French, 1)]

 

'3 Recall Heim (1982) treatment of indefmites as bound variables.

‘4 The Prohibition against Vacuuous Quantiﬁcation can be deﬁned as follows (Kratzer, 1995: 131): For
every quantiﬁer Q, there must be a variable x such that Q binds an occurrence of x in both its restrictive
clause and its nuclear scope.

162

In the tripartite quantiﬁcational structures in (134'a) and (134'd), the difference in
argument structure, according to Kratzer, is what triggers the (un)grammaticality of the
relevant sentence. In (134'd) the prohibition against Vacuuous Quantiﬁcation is
respected since the operator, always in this case, binds the extra argument variable
provided by the S-level predicate. No such argument exists for knows and (l 34'a) is
accordingly an unacceptable logical representation.45 46
Cohen and Erteschik-Shir (2002) exploit Kratzer’s line of argument here concerning
the difference in argument structure between S—level and I—level predicates. C&E-S
argue that since an I-level predicate such as brave in (135) lacks such an extra spatio-

temporal argument, and by the Topic Constraint which says that all sentences have to

have topics, the BN boys would have to be read as the topic.

(135) Boys are brave.

(136) Boys are present.

In (136), on the other hand, since present is an S-level predicate the extra spatio-

 

45 See De Swart (1991) for criticism of Kratzer’s arguments here.

‘6 Kratzer (1995) utilizes this line of argumentation to propose a syntactic account of weak
(existential)/strong (generic) interpretations of BNs. She argues that the extra argument is always external;
therefore subjects of S-level predicates cannot be base-generated outside of the VP but inside. Since I-
level predicates lack such an argument the subject can be base-generated outside of the VP, in the IP
domain. Kratzer further draws on Diesing’s (1992b) Mapping Hypothesis to argue that the mapping to
logical representations is straightforward from that point on: the VP material is mapped into the nuclear
scope (and get existentially closed off) whereas the IP material gets mapped into the restrictive clause and
is therefore interpreted generically. In spite of its appeal, Kratzer’s and Diesing’s analyses are problematic
since, as Cohen and Erteschik-Shir's (2002) point out (p. 140), the evidence for the VP-intemal subject
hypothesis is equally applicable to both S-level and I-level predicates. Note also that I have shown in the
text above that the Mapping Hypothesis of Diesing (1992b) is conceptually and empirically problematic
(see § 3.3.3.2).

163

temporal argument can be the topic freeing the BN boys to be focal, which is the
preferred reading for the BN. This sentence, however, is ambiguous: boys can also be
read generically. In this case, the BN itself is read as topic, and not the extra argument.

The argument-hood of topics ﬁgures in in examples such as the following:

(137) Boys are hungry.

C&E—S argue that (137) are infelicitous. The existential reading, they point out (p.141),
is hard to obtain here since the stage topic (i.e. the extra argument) is an adjunct, rather
than an argument of the predicate. This can be seen in the contrast between (138) and

(139):

(138) In the dining room, John was hungry.

(139) ? In the dining room, John was present.

As far as how the existential and the generic readings of BNs are generated, C&E-S
assume the mapping between the topic and the restrictive clause and the focus and the
nuclear scope (cf. §3.6.3.1 above). Generic BNs with kind-denoting predicates denote or
refer to kinds. In characterizing sentences the BN gets mapped into the restrictive clause
to get bound by the implicit Gen operator (after type-shifting to resolve the type
mismatch; see C&E-S (2002) for details). As far as the existential reading is concerned,
C&E-S assume that BNs are not variables but that they get incorporated to the predicate.

The existential interpretation of the BN is provided by the lexical meaning of the verb. 1

164

will not engage in the technical details of the incorporation account of existential BNs
referring the reader to C&E-S (2002).

For my purposes here, in view of the above discussion, C&E-S's analysis can carry
over to BNs in PA. Recall that BNs in PA are always interpreted existentially and they
cannot co-occur with I-level predicates. If the latter, according to C&E-S ’5 account
(based on Kratzer, (1995)), lack an extra spatio-temporal argument that can act as a
topic, then the non-occurrence of PA BNs with I-level predicates is predicted. On the
other hand, if S-level predicates have such a spatio-temporal argument then this
argument can act as a topic thereby freeing the BN to be focal. PA BNs are always
focused, which is a natural outcome of assuming C&E-S's approach. Therefore, this
analysis extends quite naturally to the observed facts concerning BNs in PA. Then, these
BNs get assigned the existential reading supposedly through incorporation to the verb.

Despite the appeal of this analysis, I ﬁnd it a bit problematic. It does not consider the
S-level/I-level distinction paramount. This distinction has been demonstrated earlier to
be relevant for PA BNs. Furthermore, the assumption that S-level predicates have an
extra argument that I-level predicates lack has been challenged in the literature
(e.g. De Swart (1991)). In the next section I consider an analysis, which I adopt, put
forth by Ladusaw (1994) that was touched upon very brieﬂy in § 3.6.1 above in my brief

discussion of Jager (1999), concerning the categorical/thetic distinction.

3.6.3.3 Categorical Vs. Thetic Judgments
In the last two sections I have brieﬂy considered two analyses conceming the so-

called subject effects (mainly, the existential/ generic distinction). I have pointed out that

165

these analyses could adequately be utilized towards an explanation of the interpretive

distribution of BNs in PA. In this section I brieﬂy review, and adopt, a proposal

advanced by Ladusaw (1994) to explaining the interpretive facts of BNS (in English).
Ladusaw (1994) bases his proposal on the hypothesis presented by Yuki Kuroda

((1972) and (1992), as cited by Ladusaw)"7 Ladusaw quotes Kuroda as follows:

"This theory assumes, unlike either traditional or modern logic, that there
are two different fundamental types of judgments, the categorical and the
thetic. Of these, only the former conforms to the traditional paradigm of
subj ect-predicate, while the latter represents simply the recognition or
rejection of material of a judgment. Moreover, the categorical judgment is
assumed to consist of two separate acts, one the act of recognition of that
which is to be made the subject, and the other, the act of afﬁrming or
denying what is expressed by the predicate about the subject. With this
analysis in mind, the thetic and the categorical judgments are also called
the simple and the double judgments (Einfache Urteil and Doppelurteil)."

What is important, Ladusaw explains, is to note that the difference between the two
types of judgments is that in the case of the categorical judgment there has to be a
presupposed subject of which the predicate is predicated. In the case of a thetic
judgment, on the other hand, no presupposed subject needs to exist. For concreteness,

consider Ladusaw’s examples (his 1 and 2):

(140) Some unicoms entered the garden.

(141) Many people were at the party.

 

'7 I do not use page numbers in reference to Ladusaw (1994) due to their absence in the copy of his article
I am consulting.

166

Ladusaw observes that there is an ambiguity in each of these examples that reﬂects the
difference in the type of judgment involved. Examples such as (140) have been noted to
be ambiguous between an indeﬁnite reading (i.e. weak) and a partitive one (i.e. strong):
it could mean either that there is an indeﬁnite number of unicoms that entered the
garden (indefmite/weak reading) or that some of (contextually identiﬁed) unicoms
entered (while the others have not) (partitive/strong). In the latter case, where the
indeﬁnite subject is partitive/strong, the categorical judgment singles out a presupposed
subject (some unicoms) and then it predicates the predicate of it. In other words, we
single out a contextually deﬁned set of unicoms and then say of them that they entered
the garden. So, a prerequisite to saying 'entered the garden' is the existence of a deﬁned
set 'some unicoms' that is presupposed. On the other hand, the thetic judgment does not
presuppose the existence of a set of unicoms, but it merely reports the event of an
indeﬁnite number of unicoms entering the garden. It follows from this discussion,
Ladusaw reasons, that I-level predicates must be predicates in categorical judgments.
The thetic/categorical distinction is deﬁned over the difference between the bases
for judgments, Ladusaw argues. The basis for a judgment could be simple (hence
resulting in a thetic reading) or compound (resulting in a categorical reading). These
judgments come in two modes, that of afﬁrmation and denial. As far as thetic judgments
are concerned, these are presentations of an object (an entity or an eventuality). To
affnrn such a presentation is to commit oneself to the existence of an object or an
eventuality that satisﬁes such a presentation. To deny that presentation, on the other
hand, is to deny the existence of that presentation. Ladusaw equates a presentation with

a description. He further takes objects to range over individuals and eventualities. What

167

we end up with, Ladusaw argues, is a description of individuals or eventualities in the
case of thetic judgments.

As for the basis of a categorical judgment it is not simple, as the case is for thetic
judgments, but compound: an object satisfying the presentation is frrst posited or
presupposed and then a DIEM is afﬁrmed or denied of that object. Properties, of
course, differ from descriptions of individuals or eventualities. In short, a thetic

judgment is an existential commitment (a denial or afﬁrmation) while a categorical

 

judgment is a predication (again, a denial or aﬂirmation of the applicability of that
property to the object). Ladusaw gives the following in way of summary of his

assumptions in this regard:

(142) a. Properties cannot be the basis of a thetic judgment, only part of the basis of
a Categorical J udgrnent.
b. The subject of the basis of a categorical judgment must be an object, not a
description of an object.
c. Strong construals are either object denoting or quantiﬁcational.
(1. Therefore, the subject of an ILP must have a strong construal.
The set of assumptions made by Ladusaw just presented in the discussion above can be
immediately generalized to carry over to PA BNs. My argument so far has been that
BNs in PA are always non-presupposed, i.e. focal. Recall also that these BNs do not
combine with I-level predicates. If, according to Ladusaw, I-level predicates are
promrties, and not descriptions, they cannot combine with descriptions (of individuals
or eventualities). This directly accounts for the non-occurrence of PA BNs with I-level

predicates. Since PA BNs are descriptions of objects (and not objects themselves) they

cannot combine with I-level predicates, which are properties. Moreover, the existential

I68

reading such as bare nominals have is a direct result of existential closure due the thetic
judgment involved. No generic reading is possible since, by argument, BNs in PA are
focused.

I believe Ladusaw’s analysis provides us with an elegant explanation of the
observed facts. No commitment as to what the syntactic domain of the thetic judgment
is is necessary (for example, as in Diesing’s (1992b) Mapping Hypothesis where the
domain of existential closure is the VP). Therefore not only do we not have to assume
the domain of existential closure to be the VP, but this would in fact be also an entirely
redundant move.

To conclude, bare nominals in PA are always subject to the existential reading due
to their very nature as descriptions of objects, rather than objects themselves.
Categorical judgments are cases of predicating a property of a presupposed object, not a
description. Only I-level predicates can be predicates in categorical judgments. It closely
follows that BNs in PA may not co-occur with I-level predicates. The existential reading
of BNs is due to the existential closure characteristic of sentences of thetic judgment
(recall that the latter are denials or afﬁrrnations of an existential commitment per

Ladusaw (1994)).

3.6.4 Conclusion

In conclusion, the preceding discussion in this chapter has been a mix of literature
review, critique and an attempt at the introduction of some fairly novel ideas. In
reviewing some of the proposals offered concerning the distribution of bare nominals

cross-linguistically I have argued that such proposals suffer from the lack of an adequate

169

and uniform account of the behavior of BNs.

I have started out the discussion in this chapter by recalling the facts from chapter 2.
Bare nominals in PA non-construct-state constructions are ungrarnmatical under normal
conditions of stress and intonation. This is due to the strength of the +/- Deﬁniteness
features of the head D position and the inability of BNs to satisfy, or check, such
features due to the latter being unmarked or unspeciﬁed for the relevant (in)deﬁniteness
features. In the current chapter, I have argued against analyses that posit a structural
requirement, such as government, of BNs explaining that such an analysis leaves a
number of issues unresolved. More speciﬁcally, it is unable to account for why
coordinating, focusing or modifying a BN apparently supposedly rescue the nominal
from violating such a requirement. 1 have proposed, instead, not to view coordination or
modiﬁcations as exceptions, or idiosyncrasies, but rather to take such phenomena as
inconsequential in the analysis of BNs. Existentially interpreted BNs are always focal,
irrespective of coordination or modiﬁcation.

As for the existential reading itself I have reviewed a number of proposals that I
claim to be quite in line with the general thread of my proposal here. In all these
analyses, Partee (1991), C&E-S (2002) and Ladusaw (1994), and in keeping with the
spirit of the analysis I propose here to the effect that PA BNS are focal, the generic
reading for these nominals is not only unexpected, but it is in fact ruled out.

In the next chapter I propose to expand the reach of the analysis advanced here to
include existentially read BNs in English. I wish to show that English BNs that are
existentially interpreted are also focal, just like their PA counterparts. This outcome is

actually expected in view of my analysis in Chapter 2 where I have argued that BNs are

170

generally devoid of (in)deﬁniteness features. I also wish to demonstrate that what
C&E-S (2002) have argued for, that English topical BNs are generic, does in fact hold
in the opposite direction too. English generic BNs are always presupposed, and never

focal, in contrast to existentially interpreted BNs in both PA and English.

171

Chapter 4

Focus and English Bare Nominals

4.0 Introduction

The driving force behind the current analysis has been a desire to achieve a uniﬁed
account of the distribution and interpretation of bare nominals in Palestinian Arabic.
This goal, I have attempted to demonstrate, is indeed attainable. To the extent that the
claims regarding the distribution and the interpretation of BNS in PA (and Spanish) in
this work turn out to be on the right track, the following question assumes extra
signiﬁcance: could such an analysis be generalized to encompass languages as varied
and different from PA and Spanish as English? The most perfunctory examination of
the phenomenon across this spectrum of languages sufﬁces to illustrate the apparent
implausibility of such an endeavor, since bare nominals, though attested in all these
languages and arguably exhibit very similar semantics, seem to exhibit a remarkably
varied syntax. This leads to the inevitable conclusion that it appears, at least on the
surface, that such a phenomenon would not lend itself with any degree of ease to a
uniform analysis. The sole purpose of the present chapter is to underscore and establish
the inaccuracy of such a conclusion.

In the previous two chapters I have proposed to view the phenomenon under
discussion in a manner which promises more uniformity and dispenses with any
disjunctive accounts that lack such uniformity. Moreover, I would like to argue, such a
view as the one being proposed in this work may prove more fruitful in accounting for,

or rather eliminating, the differences seen cross-linguistically with respect to bare

172

nominals. I have essentially disposed of the need for any structural requirements or
constraints in regulating the distribution of bare nominals generally. The claim I have
made is that bare nominals in PA and Spanish are subject to a focal interpretation which
makes them unsuitable, or resistant, to being read as topics. That these nominals are
always read existentially follows from their status as foci. The generic reading, then,
turns out to be neither possible nor permissible for these nominals, irrespective of how
we choose to analyze them.

In my examination of the structure of the DP in PA in Chapter 2 I have observed a
telling difference between construct state (CS) constructions and non-CS constructions.
The former, but not the latter, allow bare nominals (BNs) to move overtly into a head D
position. This, I have suggested in that chapter, can be explained by the hypothesis that
BNs lack (in)defmiteness features in the general case. All things being equal, this will in
turn make these nominals unsuitable candidate checkers of the strong +/- Deﬁniteness
features of the D position. In CS structures, on the other hand, BNS are empowered to
perform such a function due to the crucial presence of a possessor that passes on the
necessary (in)deﬁniteness features to the BN. I have also suggested that the same can be
hypothesized for English at the level of Logical Form (based on remarks by Longobardi
(1996)). Such an extension of the facts to English should be expected, and perhaps,
welcomed, since the phenomenon of determinerless nominals is not a prerogative of or
exclusive to PA (and Spanish). In fact, since Carlson (1977a, 1977b) English has proven
to be a very productive and, indeed, fertile testing ground for the patterns of behavior
noted of BNS generally.

Recall further from the previous chapter my contention that PA and Spanish BNs are

173

foci, and therefore always subject only to an existential interpretation. Ideally, we
should expect to ﬁnd the same state of affairs to hold in the case of English BNs, thus
pushing the claims of uniformity even farther. However, a cursory look at the behavior
of the latter conﬁrms the apparent inaccuracy of this expectation: English BNs allow a
generic, as well as an existential reading. Looking at the issues in this light, it becomes
apparent that English BNS do not lend themselves to an analysis such as the one
presumed for PA and Spanish BNs. However, in order to minimize cross-linguistic
variation as much as possible, I wish to argue that this latter hypothesis is also
inaccurate, or at least, too strong. More precisely, I believe English BNS that are
existentially interpreted do meet the null hypothesis, namely, that they should behave
like their existential counterparts in PA and Spanish. Obviously, insofar as this could be
proven to be the case, we would reach a desirable outcome, that of reducing as much as
possible cross-linguistic variation.

Therefore, my intention in this chapter is to demonstrate that such a task is not
intractable. I show that English BNS that are subject to an existential interpretation are
always foci, just like their PA and Spanish counterparts. English generic BNS, on the
other hand, are topics and can not be focused.1 Focusing generic BNS in English, and
generic DPs in PA (i.e. with the deﬁnite article) can only attain us a contrastive topic
reading, unlike focusing an existential BN which can be non-contrastive foci.

If existential BNs in English are foci as in PA and Spanish, this begs the question of

why English BNs should be able to express genericity while their PA and Spanish

 

' In other words, I wish to argue that what Cohen and Erteschik Shir (2002) argue for, that topics are read
generically and foci existentially, holds in exactly the opposite direction: existential BNs have to be foci,
and generic BNs cannot be focused.

174

counterparts can not. I think there is a principled answer to this question. This, however,
will have to wait until Chapter 5.

The rest of this chapter is structured as follows. In § 4.1 I reconsider Diesing's
(1992b) Mapping Hypothesis to argue that what the MH accomplishes can be
accomplished by putting the focus structure of the utterance under scrutiny. In § 4.2 I
argue that there is a very tight correlation between generically interpreted BNS in
English and being presupposed or non-focal, and being existential and being focalized.
This correlation holds regardless of the structural position of the BN, be it in subject or
object position, and is reminiscent of the situation noted of existential BNS in PA. In
§ 4.3 more evidence is presented to the correctness of this correlation by considering
more closely English BNs in object position. A category of verbs that presents us with
primafacie evidence to the inaccuracy of such a correlation is taken up in § 4.4 where I
argue that gong-as; is an inherent component of the meaning of these verbs.
Consequently, the BN object, which is interpreted generically, is predictably contrastive.

§ 4.5 contains some concluding remarks.

4.1 Diesing's Mapping Hypothesis Revisited

One of the points that has emerged from the discussion in the previous chapter is
that the mapping algorithm of Diesing (1 992b) has turned out to be insufficient or, at
least, problematic. The Mapping Hypothesis (MH) gives the wrong results and,
furthermore, makes the wrong predictions, both on conceptual and empirical grounds.
Aside from all the problems noted in the discussion in Chapter 3 with respect to the

MH, Diesing herself observes, and dismisses, what appears to be a noticeable weakness

175

in the MH, namely, its redundancy.

Diesing (1992b: 49-53) observes that the transparency of the mapping ﬁ'om syntactic
structures to logical representations predicted by the MH is violated by focusing the bare
nominal. To begin with, consider the following German examples from Diesing (1992b)

which conform to her MH (her 78a-b and 79a-b, respectively):

(1) a. ...weil Professoren ja doch verﬁigbar sind
since professors 'indeed' available are
'since (in general) professors are avilable
b. ...weil ja doch Professoren verﬁigbar sind
since 'indeed' professors available are
'since there are professors avialable'
(2) a. ...weil Skorpione ja doch giftig sind
since scorpions 'indeed' poisonous are
'since (in general) scorpions are poisonous'

b. *weil ja doch Skorpione giftig sind
since 'indeed' scorpions poisonous are

The positioning of ja doch 'indeed' in (1) and (2) enables Diesing to determine the
placement of the BN in the sentence since ja doch is assumed by Diesing to mark off the
edge of the VP. Accordingly, in (1a) Professoren 'professors' is to the left of ja doch,
which means the BN is located VP-extemally, or in the 1P domain. Therefore, the BN is
subject to a generic interpretation as the gloss makes clear. Being a Stage-level
predicate, verﬁigbar 'available' also allows its subject to take on an existential
interpretation as is expected in Diesing’s analysis; (1 b) exempliﬁes this possibility

where the BN is read existentially. Notice also that the placement of the BN in (lb)

176

conforms to the MH by occurring VP-internally. giftig 'poisonous,’ on the other hand, is
an Individual-level predicate and the BN Skorpione 'scorpions' can only occur outside
the VP. Therefore, the BN is interpreted generically. As far as (2b) is concerned, this
sentence should be predicted to be ungrarnmatical within Diesing's framework of
assumptions. This prediction is borne out since (2b) is ungrarnmatical. The BN in this
sentence is placed within the VP and should be assigned an existential reading.
However, this is not possible with an I-level predicate such as g1ﬁig'poisonous.’

So far this paradigm of examples does indeed argue for the correctness of the MH.
However, Diesing observes that destressing the BN in (2b), while stressing or focusing

the predicate, restores grammaticality to the sentence as seen in (3) (her (80)):

(3) ...weil ja doch Skorpione GIFT'IG sind
since 'indeed' scorpions POISONOUS are
'since (in general) scorpions are poisonous

Unfortunate as it turns out for the MH, the example in (3) is acceptable. Notice that the
subject Skorpione 'scorpions' occurs igsiﬁ the VP, assuming of course, with Diesing,
that ja doch 'indeed' delimits the VP. Notwithstanding the position of the subject, and
the fact that GIFT IG 'POISONOUS' is and I-level predicate, the sentence is acceptable if
the subject is deaccented and the predicate is focused. The subject BN is interpreted
generically even though it occurs VP internally, a clear violation of the MH. A point that
stands out from the discussion of this example, which will turn out to be crucial in what
follows, is the fact that the generic BN is deaccented (non-focal) whereas the predicate

is focalized. I argue below that this is not an isolated example in German but is a

177

recurrent pattern also in English, in which generically interpreted BNs, I argue, are
always presupposed (i.e. non-focal).

The MH of Diesing (1992b) appears to have been formulated in order to prevent
subjects of Individual-level predicates such as intelligent in (4) from being interpreted

existentially:

(4) F iremen are intelligent.

To prevent ﬁremen from acquiring an existential interpretation in (4), the MH acts as a
constraint or ﬁlter to rule out the wrong reading. To recap the discussion ﬁ'om the
previous Chapter 3, this will be achieved by assuming that the IP-level material maps
into the restrictive clause at LF, and not into the nuclear scope. firemen, being outside
the VP at S-Structure, is prevented from mapping into the nuclear scope at LF

(i.e. lowering into the VP at the LF level and hence be mapped into the nuclear scope).2
However, if we consider the following examples from Partee (1991) it becomes evident
that what the MH is trying to accomplish can be achieved concisely and adequately by
considering the focus structure of the sentence containing the indeﬁnite bare nominal

(Focus is indicated by capitalization):

(5) a. TICKETS are available.

b. Tickets are AVAILABLE.

 

2 Subjects of l-level predicates, according to Diesing's (1992b) assumptions, which are in turn based on
Kratzer (1995), are base-generated outside the VP and cannot therefore reconstruct back into the VP at
LF, unlike subjects of S-level predicates. Since subjects of S-level predicates are base generated VP-
intemally they can move back into the VP at LF.

178

(6) a *TICKETS are expensive.

b. Tickets are EXPENSIVE.

First of all, the difference between (5) and (6) is essentially a difference in the type of
predicates the BN co-occurs with. (5) contains an S-level predicate, whereas the
predicate in (6) is I-level. What is seen in (5) is basically a more liberal pattern of focus
possibilities that in (6) 3 In (5), both the BN subject or the predicate can be focused. In
(6), however, only the predicate can be focused" In an out of the blue utterance of the
sentences above, the BN TICKETS in (5a) is read existentially. On the other hand,
Tickets in (5b), with the predicate focused, is read generically. The BN in example (6)
cannot, in an out of the blue reading, be focused (see fn. 4, and below). Only the
predicate in (6b) can be focused. These interpretive differences seen in (5) and (6) are
precisely the same possibilities predicted by Diesing’s MH.5 Since available is a stage-
level predicate, it follows, according to Diesing's analysis, that the sentence should be
ambiguous between a weak and a strong reading for the BN, i.e. the BN should be
subject to both a generic as well as an existential reading, depending on whether or not
it reconstructs back into the VP at LF. In (6a), where the indeﬁnite subject with an
Individual-level predicate is focused, the sentence is unacceptable (unless TICKETS is

read contrastively). (6b) shows that when the predicate is focused the sentence is

 

3 This of course begs the question why an S-level predicate can allow the BN to be focal or non-focal,
whereas I-level predicates only allow the BN to be non-focal. I leave this question open for the moment.
4 The BN Tickets in (6a) can be focused contrastively, however. More will be said about this possibility
below.

5 Recall that we have seen in § 3.6.3.2 that this was also a proposal made by Cohen and Erteschik-Shir
(2002), namely, the mapping of bare plurals to logical forms according to their focus structure.

179

acceptable and the indeﬁnite subject is read generically.6

The preceding discussion serves two purposes. First, it highlight the observation
made earlier that the MH is cumbersome in that it makes unnecessary assumptions to
explain phenomena that could otherwise quite easily be explained by a focus theory.
Second, and more importantly, the discussion brings out an important point: existential
BNs are always focused whereas generic BNs are non-focused, or presupposed.

Despite the differences, there is a common thread between (5) and (6): in both sets
of examples the generically interpreted BN resists a focal reading. When the generic BN
is focused, it is contrastively focused, as is attested in the following grammatical

counterpart of (6a):

(7) TICKETS are expensive, not BOOKS.

The existentially interpreted BN in (5) can be non-contrastively focused with a S-level
predicate such as available. When the BN is not focused with a S-level predicate it is
assigned a generic reading. As a matter of fact, (5) simultaneously illustrates two
possibilities for the BN, and it also displays the two options available for an S-level

predicate: either being focal or non-focal. Each possibility for the predicate in terms of

 

6 The same analysis argued for concerning examples (5) and (6) in the text automatically carries over to
the familiar examples in (i) and (ii) Diesing (1992) uses to argue for the MH:

(1) Firemen are available.

(ii) Firemen are altruistic.
In (1), where the BN is ambiguous between an existential and a generic reading, the ambiguity correlates
with different focus assignments. If the BN is focused, the existential reading follows; if it is non-focal,
the generic reading follows. In (ii), the predicate only can be focalized. The BN is topical and is read
generically. If focused, the BN gets a contrastive reading.

180

focus correlates with an opposite focal status for the BN subject; if the S-level predicate
is focal, as in (5b), the BN subject is non-focal, and therefore, generically interpreted. If,
on the other hand, the S-level predicate is non-focal, as in (5a), the BN subject is focal,
and, therefore, existentially interpreted. Only the former possibility is open for I-level
predicates, unlike S-level predicates. To put it differently, I-level predicates have to be
focal, since they combine with BNs that are non-focal (i.e. generic).

This intriguing descriptive pattern noted immediately above for English in examples
(5) and (6) ﬁnds further support in PA. Recall from the previous chapter that PA BNS
only co—occur with S-level predicates to the exclusion of I-level predicates. Furthermore,
PA BNS, I have argued, have to be foci and are only subject to an existential
interpretation. It can be argued, then, that I-level predicates must always be focused,
unlike S-level predicates, which can be focal, but do not have to be. We could speculate
that the impossibility for a PA BN to co-occur with an I-level predicate follows
naturally from this difference. PA BNS, being focal, cannot co-occur with I-level
predicates, the latter being always focal, assuming, crucially, that one focus is possible
per sentence.

However appealing this hypothesis appears to be, it seems to me that it may not be
accurate. The focal/non-focal status of the predicate, whether it is an S- or I-level, I
believe, follows ﬁom the focal/non-focal status of the BN subject in this case, and not
the other way around. Existential BNS, being focal, impose a non-focal status on their
predicate. Generic BNs, on the other hand, force a focal status upon their predicate
because these nominals are obligatorily non-focal, otherwise they would get the

existential interpretation. This observation explains the fact that when a generically

l8]

interpreted BN co—occurs with an S-level predicate the latter becomes focused, as seen
in (5b). The pattern seen in PA BNs is that of the English sentence in (5a): a focal,
existentially-interpreted BN imposing a non-focal status on its predicate. The English
pattern seen in (5b) is not attested in PA, however. This can be explained by making
two important assumptions, the ﬁrst of which I have argued for in the previous chapter
and the second will be taken up in the next chapter. First, BNS in PA, like their
existential English counterparts (see below), are always focused and therefore impose a
non-focal status on their predicate; second, as I claim in Chapter 5, generic operators in
PA cannot bind bare nominals contained in a DP shell with a null or empty determiner.

This line of argument is what I would like to develop in what follows. Therefore, I
set out to demonstrate that existential BNs in English are inherent foci. Conversely,

generically-interpreted ones are non-focal, or presupposed.

4.2 Existential] Generic & Focal/Non-Focal

Bare nominals in Palestinian Arabic have been argued in Chapter 3 to be always
foci. Furthermore, these nominals, it has been noted, are only subject to an existential
interpretation. It would certainly be ideal if the same focal status of PA BNs could be
shown to carry over to existentially interpreted bare nominals in English. This task, I
believe to be attainable. Once we are able to show that existentially read BNS in both
Palestinian Arabic and English can be handled equally on the same footing, what
remains is to show that the differences in the general behavior of BNS in these two
languages can be explained on principled grounds. This second task I put off until

Chapter 5. For now I turn to an exposition of the ﬁrst task.

182

As a test to differentiate between object-referring (i.e. existential) and kind-referring
(i.e. generic) DPs, Kriﬂta et al. (1995:13) note that in upward-entailing contexts (i.e.
non-negative sentences) object-referring DPs can be replaced by "less informative" DPs
without affecting the truth-conditions of the sentence. Kind-referring DPs, in contrast,

do exhibit such a difference in truth-condition:

(8) Berber lions escaped from the zoo. :> Lions escaped from the zoo.

(9) Berber lions are extinct. #:> Lions are extinct.

The fact that Berber lions escaped from the zoo clearly entails that lions escaped from
the zoo. However, that Berber lions are extinct by no means entails that lions are
extinct. The BN in (8) is therefore read existentially (i.e. object-referring) whereas that
in (9) gets the kind reading or a generic reading. Now, since the context question-answer
pairing has proven instrumental in my discussion of BNs in PA in Chapter 3, I believe
such a strategy will turn out to be just as useful in the case of English. Let us, then,
consider the relevant context questions of the examples just cited in (8)-(9).

Since Berber [ions in (9) is interpreted existentially a felicitous context question

(CQ) for (8) would be (10)

(10) What escaped from the zoo?

in which Berber lions would be focalized since it substitutes for the wh-pronoun.

However, (1 1) would be an infelicitous CQ for (8):

I83

(11) # Where did Berber lions escape from?

This follows from the fact that the BN in the CO in (11) is more felicitously read
generically (as kind-referring) rather than existentially (i.e. object-referring). Actually,
(11) may not be a felicitous or an acceptable question at all due to the clash between the
(kind-referring) generic subject Berber lions and the episodic S-level predicate escape,
unless a very speciﬁc context is conjured up for the question.

Consider next a possible CQ for (9) as in (12):

(12) # / * What/ Who is extinct?

The CQ in (12) is awkward at best, ungrarnmatical at worst. This seems to say that
Berber lions in (9) cannot be focalized. A more plausible CQ for (9), then, would be

(re-)phrased as in (13), or even (14):

(13) What kind(s) of animal(s) is/are extinct?/What animal is extinct?

(14) Tell me about Berber lions: What about them?/What do you know about them?

Both (13) and (14) are acceptable as CQs for (9) since they clearly signal the Elk—31
status of the kind-referring or generic DP Berber lions. However, these two CQs are not
actually equivalent. On the one hand, if (9) is a reply to (13) then Berber lions would be
read contrastively (i.e. juxtaposing extinct and non-extinct animals) contrasting Berber

lions to other animals which are not extinct. In this case (9) would be more plausibly

184

rendered as in (15):

(15) BERBER LIONS are extinct (as opposed to some unspecified animal(s)).

As such BERBER LIONS would be carrying contrastive stress and the kind-predicate,
extinct, would be destressed, quite an expected result since the predicate in (15), and its
CO in (13), is outside the scope of the assertion (i.e. presupposed). If, on the other hand,
(9) is an answer to (14), Berber lions would be topical, and not contrastive. In this case,
Berber lions would be deaccented, since it is presupposed or non-asserted, and the kind-

predicate would be focalized as the rendering in (15') illustrates:

(15') Berber lions ARE EXTINCT.

A similar argument can be formulated for the following example also cited in Kriﬂta

et al. (1995: 24):

(16) Typhoons arise in this part of the Paciﬁc.

This example is ambiguous between two reading paraphrased in (17) and (18) (Kriﬂta et

al. 1995: 24). T yphoons in (l 7) has a generic interpretation while it is interpreted

existentially in (18).

(17) Typhoons in general have a common origin in this part of the Paciﬁc.

185

(18) There arise typhoons in this part of the Paciﬁc.

The paraphrases in (17) and (18) correlate with distinct CQs. To start with, note that

(19) would be a perfectly plausible CQ for (16):

( 1 9) What arises in this part of the Paciﬁc?

This CQ would get us the second reading in (18) argued to be available for (16). Notice
that this part of the Pacific in the CQ in (19) is presupposed, i.e. thematic. It should
therefore have the same thematic status in (16). T yphoons, by contrast, is not
presupposed in (19) and should therefore be asserted in (16), i.e. rhematic or focused.
The reading available here, paraphrased in (18), presupposes that this part of the Paciﬁc
is what is under discussion, or part of the common ground or common assumptions of
the interlocutors. In other words, this reading does not commit to any existential
presupposition with respect to typhoons; typhoons is not part of the background
knowledge of the interlocutors. Accordingly, typhoons in (16) would be focalized.

The CO in (19), however, cannot get us the other reading available for (16), namely,
the one paraphrased in (1 7). This is so since typhoons would be focal in the CO in (19)
but thematic or presupposed in (16). A plausible CQ that would generate the reading in

(17) would rather be (20)

(20) Where do typhoons arise?

186

or even (21)

(21) What about typhoons: tell me about them?/What do you know about them?/

What do you know about typhoons?

and in both CQs typhoons would be presupposed, thus read generically. Under such a
reading, the focus of (16) would either be the locative PP in this part of the Paciﬁc, or
the VP arise in this part of the Pacific, depending of course on which CO (16) would be
an answer to (i.e. either (20) or (21), respectively). Possible yet, this alone could be
contrastively focused, but in this case the relevant CQ would be as in (22), which is

different from either (20) or (21):

(22) Do typhoons arise in THAT part of the Paciﬁc?

As an answer to the CO in (22), (16) would be rendered as in (23):

(23) Typhoons arise in THIS part of the Paciﬁc.

A quick comparison to PA is revealing here. Consider the literal equivalent to (16)

in PA as in (16'):7

 

7 A postverbal position for the BN is most felicitous to indicate focalization. A preverbal position would
be possible but the BN would come out contrastive (i.e. TYPHOONS, not TORNADOES, arise in this part
of the Pacific. More about this in the next chapter.

187

(16') b—tiTlaY min ha5al-dsuzu? min l-muhiiT l-haadi ‘Pa‘i‘aaSeer

arise ﬁ'om this the-part of the-ocean the-Paciﬁc typhoons
'Typhoons arise from this part of the Paciﬁc (Ocean).'

As is clear from my assumptions so far, the only reading available for the (16') in

Palestinian Arabic is where the BN, Pai’aaSeer 'typhoons' is both focused (assigned

stress by the NSR) and read existentially; the generic reading cannot be expressed by the
BN. So, (16') can only have the PA counterpart of (19) as a felicitous CQ, as seen in

(19'):

(19') sun b-iTla‘i‘ min haaa l-dsuzu? min l-muhiiT l-haadi ?

what emerge from this the-part of the-ocean the-Paciﬁc
'What arises in this part of the Paciﬁc (Ocean)?'

By contrast, the CQs in (20) and (21), exempliﬁed here in (20') and (21'), would not get
us the reading expressed in (16'). Notice also that a CO such as (21") is ungrarnmatical,

a result by now expected in view of my contention that BNS in PA are always focalized.
Thus, a question such as (21") would be out on the principled ground that the CQ would

have two focuses, an unacceptable outcome.

(20') (min) ween b-iTlaSTu l-?a9aaSeer?

(from) where arise the-typhoons
'Where do (the) typhoons arise?

(21') suu b-tinaf 9am l-?a9aaSeer?

what know-ZS about the-typhoons
'What do you know about typhoons?’

188

(21") *suu b-tiiiraf cIan Paﬁ‘aaSeer?

what know-ZS about typhoons
'What do you know about typhoons?’

(20') and (21') would be acceptable CQs to the PA example in (16"), where the nominal

is determined by the deﬁnite article and the reading is, of course, generic:

(16") I-?a9aaSeer b-iTla‘Zu min ha5a l—d3uzu? min l-muhiiT l-haadi

the-typhoons arise from this the-part of the-ocean the-Paciﬁc
'Typhoons arise from this part of the Paciﬁc (Ocean).'

Non-contrastively focalizing 1- Pa f’aaSeer 'the-typhoons' in (16") is unacceptable. First,

if, for the sake of argument, we assume I- Pa YaaSeer 'the-typhoons' in (16") to be

informationally (i.e. non- contrastively) focused, (19'), repeated below, would not be

felicitous as the relevant CO

(19') suu b-iTlaS? min haaa l-d3uzu? min l-muhiiT l-haadi ?

what emerge from this the-part of the-ocean the-Paciﬁc
'What arises in this part of the Paciﬁc (Ocean)?'

the reason being that the preverbal position is reserved for topical or presupposed (non-
asserted) elements. The subject, which would replace suu 'what' in (19'), would carry
narrow focus and should therefore be placed in a postverbal position as in (19") which

is, however, ungrarnmatical.

I89

(19") *b-tiTla‘i‘u min ha6a l-d3uzu? min l-muhiiT l-haadi l-PaYaaSeer
arise from this the-part of the-ocean the-Paciﬁc the-typhoons
'Typhoons arise from this part of the Paciﬁc (Ocean).'

Thus, 1- Pa PaaSeer 'the-typhoons' in (16") cannot be informationally focused. It can

either be topical, in which case another constituent or element in the sentence would be

focused, or else, if focused, it would be contrastively focused8 (i.e. notationally, it

would be capitalized) as in (l9'"). In this latter case, a felicitous CQ would be (19""):

(19'") L-i’ASTAASEER b-iTlaS‘U min ha5al-d3uzu? min l-muhiiT l-haadi

THE-TYPHOONS arise from this the-part of the-ocean the-Paciﬁc
'TYPHOONS arise from this part of the Paciﬁc (Ocean).'

(19"") suu Pilli b-iTlaS‘ min ha5a l-d3uzu1> min l-muhiiT l-haadi,
what that arise ﬁ'om this the-part of the-ocean the-Paciﬁc
l-‘i’aa‘aaSeer willa l-‘i’awasef?

the-typhoons or the-hurricanes
'Which (natural phenomenon) arises in this part of the Paciﬁc (Ocean),
typhoons or hurricanes?’

Note also the addition of Pilli 'that' which indicates the deﬁnite status of what is being

questioned is added in (19"").

The purpose of this little detour to PA is twofold. First, it serves to conﬁrm our

intuitions regarding the availability in English of the two interpretations to examples

such as (16). These interpretations correspond to two different utterances in PA, one

 

" See below for an important qualification of the 'focal' status of generic DPs.

I90

with a BN and the other utilizing the deﬁnite article to express the generic reading. My
second purpose behind this digression is to illustrate that the null hypothesis is in fact
true: if generically interpreted BNs in English are non-focal, then determined DPs that
are read generically in PA should be predicted to resist being focalized. Both generic
BNs in English and generic DPs with the deﬁnite determiner in PA are only open for a
contrastive topical interpretation. This result is not trivial; it will turn out to be
signiﬁcant in the next chapter.

The same observations made with respect to (16) can also be made concerning the
following two examples cited in Krifka et al. (1995: 24).9 These authors point out that
these examples too are ambiguous between a generic and an existential reading of the

BN, as paraphrased in a and b, respectively:

(24) Dogs must be carried. (Sign in front of the escalator).
a. A dog must be carried to use the escalator.

b. Any dogs on the escalator are to be carried.
(25) Shirts must be worn. (Sign at a restaurant entrance).
a. If you want to enter this restaurant, you must wear a shirt.

b. The only thing you can do with a shirt is to wear it.

The readings in (24a) and (25a), where the BN is subject to an existential reading,

 

9 The examples are due to Halliday (I970).

191

would have (26a) and (26b) as possible CQs, respectively:

(26) a. What must be carried (to use the escalator)?

b. What must be worn (to enter the restaurant)?

Both these CQs entail that the BNS in (24) and (25) are asserted and not presupposed,
hence focused or thematic. This would basically produce the existential reading for
these nominals, therefore entailing the unintended, ludicrous reading for (24a) where the
reader of the sign would take it to be mandatory to carry a dog in order to use the
escalator. '0 However, such readings as (24a) and (25a) could not possibly arise in

answer to the following CQs:

(27) a. What about dogs? / What do you know about dogs? Tell me (more) about
them?
b. What about shirts?/ What do you know about shirts? Tell me more about

them?

What is in common between (27a) and (27b) is that the BNS, dogs and shirts,
respectively, are presupposed, i.e. non-asserted, therefore subject to a generic reading.
The reading generated by these CQs would be (24b) and (25b), respectively, resulting in

the unintended, ludicrous reading for (25b) where the reader of such a sign at a

 

'0 The mechanism by which the existential reading is achieved for the bare nominal can be the same
mechanism argued for with respect to BNs in PA in the previous chapter (cf. Ladusaw (1994)).

I92

restaurant would take it as giving pointers on the numerous purposes or uses shirts could
be put to (or, in this case, the only purpose).

Link (1995) considers the following two examples:

(28) a. Computers assist EXPERTS.

b. COMPUTERS assist experts.

Link observes that by shifting the stress from experts in (28a) to computers in (28b) the
V/3 (i.e. universal-generic/existential) structure of the sentence is shifted or reversed
(1995: 374-375). In (28a) EXPERTS would be read existentially, while computers
generically. In (28b), on the other hand, COMPUTERS is subject to an existential
reading whereas experts is generically interpreted this time around. Consider how it can
be demonstrated that when the BN gets the generic reading it is non-focal. (2 8a) would

plausibly be an answer to the following CQ:

(29) a. What can/ do computers do (these days)?

b. Who do computers assist?

In both CQs, (29a) and (29b), EXPERTS would be non-presupposed. This means that it
is rhematic or focused in the answer in (28a). Notice that (29a) and (29b) entail slightly
different focus structures for (28a). If (29a) is the relevant CQ for (28a), then the

predicate assist as well as the BN EXPERTS would be part of the assertion. Both would

in that case be focused. Conversely, assist would be outside of the focus structure of

193

(28a) if the relevant CO is in fact (29b). What is fairly clear here is that neither of the
CQs in (29a) and (29b) would generate the reading, or rather the focus structure, in
(28b), where COMPUTERS is asserted and experts is presupposed. There would
obviously be a clash between the focal status of the BN computers in (29), where it
would be presupposed, and its informational status in (28b), where the bare nominal

COMPUTERS is focalized. A more likely CQ for (28b) would either be (30a) or (30b):

(30) a. What assists experts (these days)?

b. What about experts?

Notice, however, that (28a), repeated here as (28a') for convenience, may have a totally
different focus structure than has been argued for it. The BN in object position, experts,
can also be read generically. That this reading is also available is attested by the

possibility of replacing the BN object with a generic DP (i.e. with the deﬁnite article) in

PA as in (28a"):

(28a') Computers assist EXPERTS.

(2 8a") l-hawasib bi-tsaa‘i‘id l-xubaraa‘i>

the-computers assist the-experts
'computers assist experts'

In this case, as exempliﬁed in (28a'), if the BN experts is focused, the reading would be

contrastive and the sentence would be an answer to a CO such as (29c):

I94

(29c) What kind of people do computers assist (as opposed to others)?

A good paraphrase for the answer to (29c) would be that experts are the kind of people
assisted by computers, but that novices, for example, or high school students, would not
beneﬁt from such technology. It is therefore evident that to focus a generically
interpreted BN is to trigger the contrastive (topical, see below) reading, unlike the
situation found with existential BNs.

A similar situation holds for (31a) and (31b) also cited in Link (1995):

(31) a. Cats chase MICE.

b. CATS chase mice.

An identical argument utilizing CQs can be brought to bear on these sentences. For that

reason, I limit myself here to considering the following CO in (32):

(32) Which animals chase mice?

In (32) the BN mice is part of the presupposition of the sentence. The element in an
answer substituting for which animals should therefore be focalized. But which animal
is more likely here to be asking about the kind-referring, rather than about an obj ect-
referring DP. This would contradict the general argument that a generic DP cannot be

focalized, especially when it is noted that (3 3) is a felicitous answer to the CO in (32):

I95

(33) CATS chase mice.

As an answer to the CO in (32), CA TS in (33) would be focalized. However, as I have
just remarked regarding the CO in (29c) and its answer in (28a'), I believe that the focus

here is contrastive. In other words, the answer in (33) could be re-phrased as (34):

(34) CATS chase mice (not DOGS).

The example, then, does not constitute an exception to or violation of my argument.
More will be said about contrastive topics/foci below.

Finally consider (35)-(40), also taken from Link (1995) (capitalization mine):

(35) Cowboys carry GUNS.

(36) Indians make BASKETS.

(3 7) Soldiers wear BERETS.

(38) CIA agents perform COVERT ACTIONS.
(39) Lions have MANES.

(40) Unicorns have HORNS.

These examples can lend themselves easily to a similar treatment. Take (3 5), for
example. Cowboys in (3 5) is most appropriately interpreted generically, whereas GUNS
would get the existential reading. For concreteness, a possible CQ for (35) would be

(41), but not (42):

I96

(41) What do cowboys carry?/ What do cowboys do?

( 42) Who carries guns?

The CQ in (42) is not unacceptable, however. It is acceptable, not with the reading or
focus structure in (3 5), but with a reading where cowboys would be both generic and

contrastive as in (35'):

(35') COWBOYS carry guns.

It also seems to me that when COWBOYS is contrastively focused guns appears to be
more appropriately interpreted as generic. The existential interpretation is, however, still
available for guns, I think. Consider, for example, that in PA guns could still be
expressed with a BN (therefore, existential) even when cowboys is generic (therefore

determined by the deﬁnite article):

(43) R-RUS‘AA b-yihmilu musaddasaat

THE-COWBOYS carry-3MP guns
'cowboys carry guns.’

This means that guns should still be focalized, even though cowboys is contrastively
focused. This situation is potentially problematic for any account assuming one, and
only one, focus per sentence, as in Rizzi (1997) and Zubizarreta (1998). However, I
believe there is a way out of this problem. If contrastively focused, generically

interpreted BNs can co-occur with focalized existential BNS, this may mean that generic

I97

-3.-‘v.—.

BNs that are contrastively focused are actually contrastive topics, rather than contrastive
M- Most signiﬁcantly, this hypothesis is more kindly to, and accommodating of, my
main argument since I have maintained so far that generically interpreted BNS cannot be
foci, but are always presupposed (i.e. thematic).

Other evidence pointing in the same direction of my general argument comes from

Kratzer's (1995) following examples:

(44) PONDS belong to this property.

(45) COUNTEREXAMPLES to this claim are known to me.

(44) and (45) indicate that subjects of I-level unaccusative (or passive) predicates can be
interpreted existentially. In both these examples the subjects have to be focalized,
otherwise the sentences would be unacceptable. In fact, notice that if (44) is minimally

modiﬁed to include another existential BN the sentence would become unacceptable:

(46) *PONDS belong to BOYS from London.

Since BOYS in (46) is existentially read it has to be focalized, as we have seen in (44).
To focalize PONDS in this case, along with BOYS, would result in two focuses for the
sentence, which is unacceptable. In other words, the reason why (44) is grammatical is
due to the focal status of the existential BN which means that (belong to) this property
in (44) is thematic.

Consider next the following example from Carlson (1977a):

198

(47) Dinosaurs ate kelp.

Carlson points out that this example is ambiguous between a generic statement about
dinosaurs or as simply reporting an event. Note that the two readings come out under

different assignments of sentential stress:

(47) a. DINOSAURS ate kelp (while Grog watched). (EX)

b. Dinosaurs ATE KELP (before they became extinct; MS) (GEN)

As Carlson notes, this ambiguity has consequences on truth conditions. This observation
follows directly from the focus structure assigned in (47a) and (47b) and is reminiscent
of the difference in truth conditions, which results from different focus assignments,

noted by Rooth (1992) for the following two examples:ll

(48) a. [Mary]p always took John to the movies.

b. Mary always took [John]p to the movies.

What is important to note about (47) is that the existential/ generic readings of the BN

Dinosaurs correlate with being focused/presupposed, respectively. As soon as

Dinosaurs is understood in generic terms the focus of the sentence shiﬁs away from it to

 

” A difference in focus structure in (48a,b) results in a difference in the truth conditions of the sentences
as noted in the text. So, for (483) to be true, it has to be Ma_ry that always takes John to the movies, not,
for instance, Sara. On the other hand, for (48b) to be truth-conditionally sound it has to be John who Mary

always takes to the movies, but not lﬂ.

I99

the predicate ate kelp. The sentence would then be understood as saying that it was
characteristic of dinosaurs (in general) to eat kelp. When the BN dinosaurs is focused,
however, the whole sentence is understood as reporting a speciﬁc event of kelp-eating
by a group of dinosaurs. Only the existential reading is available in this case.

Another of Carlson's (1977a) examples can be accorded a similar treatment.

Consider (49):

(49) a. Trucks hauling dynamite ARE ILLEGAL IN NEVADA. (GEN)
b. TRUCKS HAULING DYNAMITE were spotted on Nevada highways. (EX)

0. * Trucks hauling dynamite WERE SPOTTED ON NEVADA HIGHWAYS.

As can be clearly seen in (49a), the generic reading of the BN Trucks hauling dynamite
coincides with the placement of sentential stress on the I-level predicate, ARE ILLEGAL
IN NEVADA. With a Stage-level predicate, such as were spotted in Nevada, the BN is
subject to an existential interpretation. Sentential stress is therefore placed on the BN,
not on the predicate, as seen in the contrast between (49b) and (49c). The topic-hood of
the BN in (49a), and its absence in the case of the BN in (49b), can be brought out by

using one of the tests utilized in Chapter 3 from Reinhart (1981):

(50) a. As for trucks hauling dynamite, they are illegal in Nevada.

b. *As for trucks hauling dynamite, they were spotted on Nevada highways.

This test brings out the topical status of the BN in (49a). However, the BN in (49b)

200

cannot be read generically, hence the ungrarnmaticality of (50b).
The same test for topichood just employed can be successfully utilized to

disambiguate the following ambiguous sentence:

(51) Girls are sick.

The adjective sick is ambiguous between physically sick or mentally sick. These two
readings correspond to distinct interpretations for the subject BN. If sick is taken to
mean physically sick (i.e. as an S-level predicate, temporary or transient), the subject
BN would be read existentially. By contrast, if sick is understood in the sense of being
mentally sick, which is a more permanent property than being physically sick (i.e. it
would be an I-level predicate), Girls would be read generically. Note, moreover, that
there is another level at which these two reading differ, namely, the level of information
or focus structure. The difference in terms of being an S-level or I-level predicate results
in a difference in how the BN is to be rendered in terms of its information structure. If
what is meant by sick here is physically sick (S-level predicate), and Girls would
therefore be read existentially, the focus of the sentence would be the BN, as in (51'a).
If, on the other hand, sick is construed as being mentally sick (an I-level predicate, since
potentially permanent), and Girls is therefore understood generically, the focus of the

sentence would be placed on the predicate itself, not the BN, as seen in (51' b).

(51') 3. GIRLS are sick.

b. Girls ARE SICK.

201

The meaning of (5 1' b) would be that girls in general are characteristically mentally ill
(or, perverted, bogus,...etc). Again this is a rather more permanent property than
physical illness.

The generic/existential distinction interacts in the same interesting ways seen so far

in the following examples:

(52) a. I know WOMEN.
b. I KNOW women.
(53) a. I know LAWYERS.

b. I KNOW lawyers.

Again, when understood existentially, the BN bears the sentential stress in (523) and
(53a) (or, conversely, focusing the BN makes the existential interpretation far more
appropriate than the generic one). To be interpreted generically, or if interpreted
generically, the stress has to be shifted away from the BN to be placed on the predicate,
as witnessed in (52b) and (53b). So, (52a) would mean that there are women such that I
know them. By contrast, (52b) means that I know what women are like (perhaps, they
are kind, mischievous, ﬁckle, ...etc, again, rather permanent properties). The same can be
said about (53a) and (53b). Note, moreover, that if the BN in either (52) or (53) is
focused, but, at the same time, understood generically, the reading that will come out is

contrastive (i.e. as contrastive topics), as in (52') and (53'):

202

(52') I know WOMEN (but not MEN). (Gen)

(53') I know LAWYERS (but not REAL STA TE AGENTS). (Gen)

Another piece of evidence in favor of my argument is provided by remarks made in
Carlson (1977b). Carlson notes that predicate nominals select a generic subject whereas
prepositional phrases select an existentially interpreted subject. This can be seen in the

contrast between (54a, b) and (55a, b):

(54) a. Dogs are sweet animals.
b. Dentists are book collectors.
(55) a. Dogs are in the next room.

b. Children were without parents.

Dogs in (54a) and Dentists in (54b) are more appropriately interpreted generically due to
the I-level predicates co-occurring with these BNs. So, what is being discussed is dogs
and dentists in general terms and we are making the observation that it is characteristic
of dogs to be sweet animals and of dentists to be book collectors. In (55a) and (55b), on
the other hand, it is not a general tendency of the referent of the BN that is under
discussion. So, Dogs and Children in (55) are more accurately interpreted existentially.
These pre-theoretic remarks, which seem intuitively plausible, are conﬁrmed by

applying the by- now familiar test for topichood to the ENS in (54)-(55):

203

(56) a. As for dogs, they are good animals.
b. As for dentists, they are book collectors.
(57) a. *As for dogs, they are in the next room.

b. *As for children, they were without parents.

Let us now see what status each one of these BNs would have in terms of the focus

structure of the sentence. First, take Dogs and Dentists in (54). A possible set of context

questions for these two sentences would be (5 8):

(5 8) What about dogs I Dentists?/What do you know about dogs | Dentists? Tell me

(more) about them?

Notice that ('59) would not be a felicitous CQs for (54)

(59) a. What animals are sweet?

b. (*?) Who are book collectors?/ What kind of people are book collectors?

unless Dogs and Dentists in (54a) and (54b), respectively, are read as contrastive topics.
In this case distinctions would be drawn between animals that are sweet and people who
are book collectors, on the one hand, and animals that are not sweet and people who are
not book collectors, on the other hand. If (58) is the relevant CQ, then it is fairly
obvious then that the BNS in (54) are non-focal or presupposed, unlike the situation in

(55)- The examples in (55) can be appropriate answers to the following CQs:

204

(60) a. What is in the next room?

b. Who was without parents?

Note also the unacceptability of the following CQ as a context question to (55):

(61) (*) Where are dogs? [To be answered: Dogs are in the next room.]

This will follow directly from the focal status of Dogs in (55a) which would clash with
its topical status in (61). The very awkwardness (or, perhaps, ungrammaticality) of the
question in (61) follows immediately from that the question has two focuses, dogs and
the wh-element, which is presumably always focal.

As has been noted earlier, the focus structure of the utterance can affect the truth
conditions of the sentence. For example, Rooth (1992) points out that an answer such as

(63) is unacceptable to the question in (62) with the focus structure shown:

(62) Who cut Bill down to size?

(63) * Mary cut [Bill]F down to size.

Also, a question such as (64) cannot be answered by (65):

(64) Who did Mary cut down to size?

(65) [Mary]p cut Bill down to size.

205

This is essentially due to the conﬂict between the question and the answer in terms of
focus structure. It follows, then, that the question in (62) can be answered by (65), and
the question in (64) can be answered by (63), since in this case there would arise no
clash in the focus structure of the CO and its answer. So an answer to question (62)
would be of the form "x cut Bill down to size" while an answer to (64) would take the
form "Mary cut y down to size."

Now consider the following set of data:

(66) a. *Who cut boys down to size?

b. Who cut the boys down to size?
(67) *Where did boys go?
(68) *Why did girls leave early? '2

(69) *When will boys leave?

Notice ﬁrst of all that all the predicates (66)«(69) are stage predicates and that the BNS
should thus be interpreted existentially. The question in (66a) is ungrarnmatical (unless,
of course, it is interpreted in a pair-list fashion). Abstracting away from the
ungrarnmaticality of the question, an appropriate answer to (66a) would take the form "x

cut boys down to size" where the BN boys would be presupposed or non-focal. That it is

 

'2 A question arises concerning the acceptability of (i), where the BN, which is existentially interpreted,
can co-occur in a CO with why:

(i) Why did people leave early?
It seems to me that the acceptability of (i) could be pragrnatically facilitated in that the existential reading
is alluded to, but somehow masked in the CO. In fact, if the BN people is further anchored, so as to
clearly mark its existential status, the sentence would no longer be acceptable as seen in (ii):

(ii) *Why did people in the room leave early?

206

the status of the BN boys that is problematic is evident when the grammatical (66b) is
considered where a determiner is used.

The interpretation the BNs have in (66)-(69) is existential due to the stage-level
predicates involved. With individual-level or habitual predicates where the BNS are
interpreted generically we do not encounter the same unacceptable outcomes. The
reason for the difference stems from the non-focal status of the BNS in (70)-(72). These
context questions would therefore be felicitous due to the fact that only one focus exists

per sentence, that of the wh-element.

(70) Why do boys hate school?
(71) What do girls like?

(72) What do boys smoke?

The major point the discussion so far has helped establish is that generic BNs can
never be focal and existential BNS can never be presupposed. When a generic BN is
focused, the focus turns out to be contrastive. In fact, we have seen some evidence
suggesting that contrastively focused generic BNS are not foci, but rather contrastive
topics. In the following section I consider further evidence to the correctness of my
claims that has been pointed out by Laca (1990) concerning BNS in object positions.
The evidence reviewed will further strengthen my argument. Furthermore, I propose in
what follows a solution to a potential problem for the current analysis posed by
predicates such as hate, love, ..etc, whose BN objects can only be read generically. More

about this will be said in the next section.

207

4.3 Object Bare Nominals and Focus Structure: More Evidence

My main concern so far has been subject BNs and their interaction with the focus
structure of the sentence, although object BNs have also been sporadically touched on.
A closer look at object BNS is, however, warranted and could provide further support
for my general analysis.

Laca's (1990) work is primarily an attempt to demonstrate that the lexical
classiﬁcation of verbs is not the only relevant factor in determining the interpretation a
BN in object position is assigned. The information structure of the sentence also plays a
role. BNs in object position that are read generically are presupposed, unlike existential
BNs which are asserted. However, she argues that the terms 'universal' and 'existential'
with respect to BNS could be misleading since these readings do not correspond exactly
to the quantiﬁed meanings ( 17’, :7, respectively) in allowing exceptions or triggering

inference. Consider (69)-(72):

(73) Bill hates dogs.
(74) Bill hates all dogs.
(75) *? Bill hates dogs except for his own terrier.

(76) Bill hates all dogs except for his own terrier.

As seen in the contrast between (75) and (76) quantiﬁed nominals allow for exceptions

unlike BNs which do not. Moreover, BNs and quantiﬁed nominals differ in terms of the

inferences they trigger as seen in (77)-(79):

208

(77) Bill mistrusts politicians.
(78) Bill mistrusts all politicians.

(79) Bill mistrusts George Bush.

The inference from (78) to (79) goes through; so, if Bill mistrusts all politicians it then
follows that he does not trust George Bush. On the other hand, the inference from (77)
to (79) is blocked; for Bill to mistrust politicians does not necessarily mean he mistrusts
George Bush. Laca, therefore, opts to use the terms inclusive (for universal) and non-
inclusive (existential). She further argues that information structure plays a role in
sorting out these two readings for object BNS.

To make her argument, Laca ﬁrst observes that inference is blocked when the bare

plural is generically interpreted. Consider (80):

(80) a. The Gwamba-Mamba eat salmon.

b. The Gwamba-Mamba eat ﬁsh.

The fact that the Gwamba-Mamba eat salmon does entail that the Gwamba-Mamba eat

ﬁsh and the inference from (80a) to (80b) goes through. Salmon is interpreted

existentially in this case. Now consider (8]):

(81) a. The Gwamba-Mamba hate salmon.

b. The Gwamba-Mamba hate ﬁsh.

209

For the Gwamba-Mamba to hate salmon does not necessarily mean that the Gwamba-
Mamba hate fish. The inference is therefore blocked from (81a) to (81b), where only the
generic reading is possible for the BN salmon. The same blocking-of-inference effect

can also be seen with predicative adjectives:

(82) a. I see big green spots when the light goes off.

b. I see green spots when the light goes off.

Again, (82a) entails (82b). If I see big green spots, which is an existentially read BN,
then that entails that I am able to see green spots when the light goes off. The same
holds for (83) where if Mary wears silk blouses it will also be true that she wears

blouses.

(83) a. Mary wears silk blouses.

b. Mary wears blouses.
However, when a predicative adjective is used, which causes the BN to be read
generically, the inference is blocked. So, in (84), the fact that I see big spots green does

not entail that I am also able to see spots green when the light is off.

(84) a. I see big spots green when the light goes off.

b. I see spots green when the light goes off.

210

Nor would it hold that if Mary wears silk blouses unironed that she wears (all her)
blouses unironed. The same line of reasoning can be carried over to (86) where eating

salmon, a generically interpreted BN, raw does not entail eating ﬁsh raw.

(85) a. Mary wears silk blouses unironed.
b. Mary wears blouses unironed.
(86) a. The Gwamba-Mamba eat salmon raw.

b. The Gwamba—Mamba eat ﬁsh raw.

Manner adverbials also behave similarly (Laca 1990: 34). The inference does not go

through between (87a) and (87b):

(87) a. The Gwamba-Mamba eat salmon with their ﬁngers.

b. The Gwamba-Mamba eat ﬁsh with their ﬁngers.

This follows from the fact that the BN salmon in (87a) is interpreted generically, hence
the blocking effect noted.

Laca (1990: 35) observes that predicative adjectives and manner adverbials are
interpreted as the focus of the sentence while the rest of the sentence is interpreted as
presupposed. This is further conﬁrmed by her observation that the relation between

(88a) and (88b) is that of presupposition since it is retained under negation as in (88c):

21]

(88) a. Los guamba—mamba comen el salmon crudo. (Spanish)
The Gwamba-Mamba eat (the) salmon raw

b. Los guamba-mamba comen salmon.
The Gwamba-Mamba eat salmon

c. Los guamba-mamba no comen el salmon crudo.
The Gwamba-Mamba don 't eat salmon raw

This clearly illustrates that the BN salmon is read generically and is also presupposed,

not asserted. The same point can also be made by considering (89a—d)

(89) a. Linguists read thrillers.
b. Linguists read books.
c. Linguists read thrillers at an incredible speed.

(I. Linguists read books at an incredible speed.

The inference between (89a) and (89b) goes through; if linguists read thrillers, then it
follows that they read books. However, if they read thrillers at an incredible speed that
does not entail that they read books at an incredible speed. Since at incredible speed
tends to stand out as being focal, the BN thrillers, which is read generically, comes out
as being presupposed in (89c).

Laca ﬁnds further conﬁrmation for her argument by observing that when there is a
close or intimate relation between the object and the verb, such a relation either being
based on a semantic relation or world knowledge, the non-inclusive (i.e. existential)

reading is available (1990: 36).

2l2

(90) Henry smokes cigars.

This follows from the fact that we cannot interpret the verb as being focused while at
the same time excluding the object. So, in (90), Laca points out, the sentence can be
taken to mean either that Henry does something, i.e. smoking cigars, or that he smokes
something, i.e. cigars. In the former case the object cigars would be part of the focus of
the sentence, in the latter case it would be the focus of the sentence. The interpretation
that Henry does something to cigars (i.e. smoking them), is, however, unavailable, since
that is what we normally do with cigars anyway.

To sum up, what emerges ﬁ‘om the discussion of Laca's (1990) arguments is the
following. The generic (i.e. inclusive, in Laca's terms) and existential (i.e. non-inclusive)
readings correlate with whether the object BN is interpreted as part of the focus or not.
If the object is indeed part of the focus of the sentence, or if it is the focus of the
sentence, then the preferred interpretation would be existential. If, on the other hand, the
object is outside the focus of the sentence then the generic interpretation is more
prominent.

A potentially problematic case, also noted by Laca (1990), is the case of the so-
called aﬁrective attitude verbs. These are verbs such as like, love, loathe, admire,
respect, ...etc, that exclude a non-inclusive or existential interpretation for their direct

object. I consider the problem in more detail in the next section.

213

4.4 Affective Verbs and the Generic] Existential Interpretation

There is a category of verbs that constitute primafacie counter evidence to the
analysis I propose in this work. The problematic nature and idiosyncratic behavior of
these verbs have also been noted by Laca (1990) who uses the term aﬂective verbs to
refer to this category of verbs. Examples of affective verbs include like, love, despise,
admire, loathe, fear,...etc.

Recall from my earlier discussion the following basic claims. First, existentially
interpreted BNS are always focalized, unlike generic BNS which can never be focalized
(non-contrastively). Second, generically interpreted BNS can be focused contrastively,
and in that case they are interpreted as contrastive topics rather than contrastive foci.
Looking at matters from this perspective, affective verbs present us with an apparent

problem. Consider the following:

(91) John likes novels.
(92) My boss respects women.
(93) Everybody admires athletes.

(94) Mary loves coffee.

The BN in object position is interpreted generically. This is conﬁrmed by the
impossibility of a BN in PA in the same position, since BNS in this language are only
subject to an existential reading as can be seen in the minimally different PA
counterparts of (95)-(98). The deﬁnite article is used to express the generic

interpretation of the DP:

214

(9S) omar b—ihibb l-riwayaat
Omar likes the-novels

(96) mudiiri b-ihtrim l-niswaan
my-boss respects the-women

(97) kull n-naas bi-qaddru I-riyaDiyiin
all the-people admire the-athletes

(98) salma b—ithibb l-gahwe
Salma likes the-coffee

So far all of this is unproblematic and is quite expected. The problem arises however
when we consider that context questions such as (99)-(102) are acceptable, contrary to

what my assumptions so far have lead us to expect:

(99) What does John like?
(100) Who does my boss respect?
(101) Who does everybody admire?

(102) What does Mary love?

Recall from earlier discussion that the CO in (99), for example, is of the form What is
the x, such that John likes x? and (100) would be Who is the x, such that my boss
respects x? Consequently, the answer to (99)-(102) should be of the following logical

forms, respectively (with some unnecessary details omitted):

(103) John likes x.

215

(104) My boss respects x.
(105) Everybody admires x.

(106) Mary likes x.

This being so, whatever substitutes for the variable x in both the context question and
the answer should be focalized. But remember now that the ENS in object position in
(91)-(94), and (95)-(98), are interpreted generically. Generic BNS, according to my
assumptions so far, are always presupposed or non-asserted, and should therefore not be
able to be focalized in the manner just illustrated in (99)-(102) and the informal logical
representations in (103)-(106).

Laca (I990) recognizes this problem and suggests the following. Affective verbs are
a different breed altogether since they are not linked to any spatio-temporal events.
Moreover, these verbs signal relations that hold between individuals rather than stages
of individuals, as in Carlson (1977b).13 In addition, the objects of affective verbs are
interpreted intensionally, rather than extensionally. Finally, the subjects of these verbs
are non-agentive but are interpreted as experiencers.

It seems to me that what Laca (1990) overlooks is another important property of
these verbs. Before I state that property, however, recall from our discussion so far that
generically interpreted BNS always come out contrastive if they are focalized, most
likely, contrastive topics. With this in mind, let me return to our examples with affective

verbs. Consider (91) again, repeated here as (107), and its CO in (99), repeated here as

 

'3 However, Laca casts doubt on the correctness of this notion of individuals vs. stages of individuals
arguing that it is hard to imagine that when, for example, John beats his wife it is a slice of John that is
involved but when John hates his wife it is the 'whole' individual that is involved.

2l6

(108):

(107) John likes novels.

(108) What does John like?

Although (107) seems to say that generically read BNS can be focused, it can be noticed
that the BN novels in (108) is contrastively focused, although the contrast is implicit,
rather than explicit. The contrastive reading is triggered by the affective verb like which
basically takes (the) things John likes and places them in juxtaposition against (the)
things he does not like. That this contrast is inherent in the lexical meaning of I-level
predicate like is evident when compared to another I-level predicate such as own in

(109) and its answer in (110):

(109) What does John own?

(1 10) John owns houses.

Notice that novels in (107) would be used with a deﬁnite article in PA since it is
generically interpreted, as we have seen in (95) above. Houses in (110), on the other
hand, would not be used with a deﬁnite article which means it is interpreted

existentially. This contrast is seen in (111)-(112):

(1 l 1) omar b—ihibb l-riwayaat
Omar likes the-novels

217

(112) omar b-yurnluk byuut
Omar owns houses

It is clear, then, that contrast is not an inherent property of any I-level predicate, but that
it is indeed an inherent property of affective verbs such as like or respect. That contrast

is not part and parcel of the meaning of own is exempliﬁed in the following questions:

(113) Who likes novels?
(114) Who owns houses?

(115) Who owns this hotel?

A good answer to (1 15), for example, is

(116) Criminals own this hotel.

where no contrast is necessarily part of the meaning of the answer. It is also important to
point out that it does not necessarily follow that with own in (110) and (116) a contrast
is drawn between the things John owns and the things John does not own, or the things
criminals own vis-a-vis the things criminals do not own. However, in (107) it is quite
plausible to say that a contrast is drawn between the things John likes and the things he
does not like. When you say that your boss respects women, you are generally taken to
mean that he respects women in comparison to someone or something else or another.

Summing up, the claim that generically interpreted BNS can never be non-

218

contrastively focalized can still be maintained. This applies to BNS in both subject as
well as object positions. The case of affective verbs does not constitute an exception to
this analysis since contrast is part of the meaning of affective verbs which results in the

objects of these verbs being read contrastively.

4.5 Conclusion

I have attempted to put forward and establish the following claims. Existential bare
nominals in English are focused, on a par with their PA counterparts. Generically
interpreted BNs in English are always presupposed or non-asserted (i.e. non-focal or
thematic). If these nominals occur in the scope of focus in the utterance or are the focus
of the utterance, they are interpreted as contrastive (topics).

This view of the issues seems to me to be appealing in its simplicity and explanatory
power. Languages that seem on the surface to be vastly different are given a uniform
and a uniﬁed analysis from which not only do the differences follow naturally, but are in
fact quite predicted. In the next chapter I wish to show exactly how such differences
between Palestinian Arabic (and, Spanish) and English, can be predicted ﬁ'om the little
we have uncovered in relation to the interaction of the focus structure of the utterance

and BNs in these languages.

219

Chapter 5

Bare Nominals, Word Order and Reference

5.0 Introduction

I have set out in Chapter 2 to explore the properties of bare nominals in Palestinian
Arabic. My working hypothesis has been that insofar as bare (i.e. determinerless)
nominals can be found and studied in other special environments, environments in
which their behavior is distinctly different ﬁ'om the general case, we may be able to
provide a more adequate description of the idiosyncratic behavior generally noted of
these nominals. This guiding assumption has turned out to be instrumental and, indeed,
illuminating in shedding some light on the properties of bare nominals, not only in
Palestinian Arabic, but also in English.

One such environment, as it turns out, is the so-called Construct State (CS) in
(Palestinian) Arabic. Bare nominals (BNS) in this construction do not exhibit a
sensitivity to the usual restrictions noted generally of BNs, such as their supposed
grammaticality in postverbal positions and their ungrarnmaticality preverbally. BNS in
CS constructions distribute freely, a fact I have attributed to the presence of a possessor
DP in this environment, which enables the BNS to substitutes into D. I have further
hypothesized that BNS in non-CS constructions lack certain features, (in)deﬁniteness
features to be exact, which makes these nominals unﬁt, and thus unable, to land in D. In
CS constructions these features are inherited from the possessor DP. If the D position
has strong +/- Defmiteness features to be checked overtly, and since BNS generally do

not have these features, then it is to be expected that BNs should be unacceptable (if

220

they occur under normal conditions of stress)] since the strong +/- Deﬁniteness features
of D go unchecked.2

I have reasoned that if PA BNS in non-SC constructions are not possible, unless
focused, it would be surprising if this state of affairs was an oddity peculiar to PA. The
null hypothesis would be that the same situation holds in other languages which may be
remarkably different from PA. Therefore, I have argued, the null hypothesis turns out to
be true in English as we have seen in Chapter 4, where I have demonstrated that
existentially interpreted BNs, like their PA and Spanish counterparts, are always
focalized.

So, if the facts are as such, then the following two questions become ever more
signiﬁcant. First, if existentially interpreted BNS in both PA and English lend
themselves to such a uniﬁed analysis, then why can BNS in English be used to express
genericity-both with kind-referring predicates or in characterizing sentences- while this
possibility is unavailable to PA BNS? In order to express genericity, PA (and Spanish)
has to resort to the use of the deﬁnite article, unlike what happens in English. Second,
why would BNs in English, on the one hand, and PA and Spanish, on the other hand,
differ in terms of their distribution? BNs in English distribute quite freely relative to the

somehow more constrained distribution of PA and Spanish BNs.

 

' The assumption here is that for (existentially interpreted ) BNS to be acceptable they would need to
occur in, presumably, speciﬁer position of a Focus Projection (FP). This would explain the
ungrarnmaticality of more than one BN co-occurring in the same utterance noted in Chapter 3 since they
would compete for the same position (Cf Rizzi (1997)). The issue of the nature and implementation of
the F P projection will be taken up in greater detail in § 5.4 below.

2 Of course, this begs the question why the derivation would not crash anyway when the BN occurs in the
speciﬁer of the Focus Projection and the strong features of D would still remain unchecked (recall that the
strong features of D would be checked by a determiner or by substituting a BN with the relevant features
into this position as in the Construct State construction). An answer to this question will be proposed in

§ 5.4.

221

 

An answer to the ﬁrst question may be pursued along two tacks. First, I would like
to make the assumption that generic operators in these two language groups, English, on
the one hand, and PA and Spanish, on the other, differ in their binding possibilities. 3
Generic operators in English can bind nominals contained in a DP whose head D is
empty or null. In PA and Spanish, by contrast, generic operators are unable to realize
this possibility, and these two languages have to use the deﬁnite article to express
genericity. A second possible argument, which is related to the suggestion concerning
generic operators just made, is to recall from Chapter 4 my contention that generically
interpreted BNS in English, and generic DPs with the deﬁnite article in PA, are never
foci, but always topical. I have also argued that when focused these DPs turn out to be
contrastive topics, rather than foci. It could then be hypothesized that to be interpreted
generically in a sentence a BN or a deﬁnite DP have to be topical or defocalized. One
way to cash out this solution is to say that when topical or non-focal a BN or a deﬁnite
DP is mapped into the restrictive clause to be bound by the generic operator. Being
focal, BNs in PA and Spanish do not ﬁt the bill, since they are arguably always caught
in the nuclear scope and cannot be bound by generic operators.

As far as the second question concerning the distributional facts is concerned, the
differences, I would like to argue, are merely surface manifestations of much deeper
differences in the properties of focus and word order between English, on one side, and
PA and Spanish, on the other side. The differences in the properties of focus are

reﬂective of, or reﬂected in, the differences that exist between these language groups in

 

3 I thank Cristina Schmitt for suggesting this line of argument.

222

word order. PA and Spanish assign ('neutral' or non-contrastive) focus to the constituent
lowest in terms of c-command (a la Zubizarreta (1998)), or, alternatively, the most
deeply embedded (per Cinque (1993)). English has the additional possibility of
assigning sentential stress, thus marking focus, more freely (but in a marked way, see
below) than PA or Spanish do. To destress a constituent, i.e. to mark it as non-focal, the
latter languages resort to movement (in the terms of Zubizarreta (1998), prosodically-
motivated movement). This is made possible, of course, by the relatively free word order
these language possess; no need arises in PA or Spanish to employ marked focus
(informally, focus that is assigned to a constituent not most deeply embedded). English,
on the other hand, restricted as it is by a rigid word order, has to resort to the marked
focus strategy since movement is generally not an option. It is therefore able to assign
focus non-contrastively to constituents that are not lowest in terms of c-command (or,
alternately, most deeply embedded). The freedom with which BNS distribute in English,
and the lack of it in the case of PA or Spanish BNs, thus follows from the properties of
focus and word order in these languages.

I would like to proceed as follows. In § 5.1 I consider the interaction of sentential
stress, focus and word order. In this regard, two analyses prove relevant, that of
Zubizarreta (1998) and Cinque (1993). I consider these two analyses in § 5.1.1 and
§ 5.1.2, respectively. Along the way the workings of the Nuclear Stress Rule and
prosodically—motivated movement are considered in § 5.1.1.1 and § 5.1.1.2. Prosodic
movement in PA and its interaction with focus and sentential stress are taken up in
§ 5.2. Before offering an argument to account for the differences between English and

PA/ Spanish in terms of expressing genericity in § 5.5, I consider ﬁrst the differences

223

that exist between these two language groups in the distribution of bare nominals in
§ 5.3 and the nature of the Focus Phrase and its interaction with BNs in § 5.4. § 5.6 is

the conclusion.

5.1 Sentential Stress Assignment, Focus, and Word Order

In order to ﬂesh out the details of my argument regarding the differences with
respect to BNs between PA and Spanish, on the one hand, and English, on the other
hand, it is imperative to look at the interaction of sentential stress assignment, focus and
word order variation in these two language groups. Two proposals thus become of
particular interest in this regard; a proposal advanced by Zubizarreta (1998) and another
by Cinque (1993). My goal will be to brieﬂy preview both analyses with the intention of
illustrating that adopting either would serve my purposes well. I begin by examining

Zubizarreta's (1998) work.

5.1.1 Zubizarreta (1998)
5.1.1.1 The Nuclear Stress Rule (NSR)

Zubizarreta (1998) proposes a modular version of the Nuclear Stress Rule (N SR)"
utilizing syntactic notions such as c-command and selectional ordering to account for

differences between Germanic and Romance in computing phrasal prominence.

 

4 Zubizarreta (1998238) explains that the NSR that was ﬁrst formulated by Chomsky and Halle (1968) was
essentially an algorithm that construes phrasal boundaries as cycles assigning stress to the rightmost
element or word in the cycle. Not only does this algorithm assign primary stress but it also assigns
secondary stress assuming that the NP as well as the IP constitute cycles for the operation of the NSR.
This is illustrated in the following example (Zubizarreta's (3), p. 38):

(i) ((The cat (in the blue MD) (has written (a book (about rats)))).

224

Zubizarreta (1998:3 7) points out that natural languages make use of one or more of
the following ways to mark the focus in the sentence: prosody, morphology or a
syntactically speciﬁed position. She puts forward an analysis of how prosody identiﬁes
the focus of the sentence. Phrasal prominence, as determined by the Nuclear Stress
Rule, mediates the relationship between prosody and focus according to Zubizarreta. To
achieve more empirical coverage she argues against a monolithic view of the Nuclear
Stress Rule (N SR) and proposes a modular version of this rule according to which
asymmetric c-command and selectional ordering play a pivotal role. Such a modular
view will be able to explain satisfactorily the positioning of nuclear stress (NS) in

paradigms such as the following ﬁom German (her 43, 45, 50a, 50b, respectively):

(1) Hans hat das/ein Birch gelesen
Hans has the/a book read

(2) Karl hat ein Buch ins Regal gestellt
Karl has a book on-the shelf put

(3) a. Peter hat an einem Papier gearbeitet
Peter has on a paper worked
'Peter worked on a paper.’
b. Peter hat an einem kleinen Tisch gearbeitet

Peter has on a small table worked
'Peter worked on a small table.’

In (1) the NS falls on the direct object in V-ﬁnal transitive structures whereas in (2) the
PP complement gets the NS since it is selected by the verb. (3) illustrates an
argument/adjunct asymmetry. If the PP is an argument of the verb as in (3a) it is

assigned the NS; however, it is not assigned stress if it is an adjunct as in (3b).These

225

examples argue for the importance of selectional properties (see fn.5) to the assignment
of the NS in German. However, in V-second sentences this asymmetry is not exhibited
by these structures and the NS is assigned uniformly on the last constituent as in (4) (her

54a, 54b, respectively):

(4) a. Peter arbeitet an einem Papier
Peter is-working on a paper

b. Peter arbeitet an einem kleinen Tisch
Peter is-working on a small table

The examples in (4) suggest that only 'constituent structure' is relevant in the assignment
of NS in V-second structures in contrast to the examples in (3) that show clearly that the
'selectional ordering' is what is relevant in assigning the NS. Zubizarreta calls the part of

the NSR that is sensitive to selectional ordering the S(electional)-NSR5 while that part

 

5 Zubizarreta (1998: 52) forrnalizes this ordering of selected heads as in (i):
(i) (C, T, V.,...V,~, PN,,, D”) with possibly m=l

(C, T,..., V,, D,), for i=1, 2,..., m-l (for the cases where m>l)

where D, i=1, 2,..., m-l is the nominal argument of V,- (for the cases

where m>l) and D,,, is the nominal argument of the lowest (possibly only)

verb or prepositional predicate (PN,,,) in the selectional ordering.
According to (i), Comp selects Tense, Tense selects a VP. V1,...V,-, PN,,., D", is an ordering of the lexical
verb into elementary verbs or prepositions while D,, is the nominal argument of PN, and D,- is the
nominal argument of V, when it exists. A category to the right is lower than the one to its left.
As an illustration of how the S-NSR works consider the following example from German (Zubizarreta
1998: 57):

(ii) [CP K3111 [hat [e [V] I ein BﬁChz [V2 gekauft [92 111111]
Karl has a book read

D, (=Karl) and [D2 V2] (=[ein Birch; [gekauftz [m e]]] are metrical sisters (see fn. 6) that are not
selectionally ordered since D1 (=Karl) has no metrical sister that is a head, hence a selector. Notice that
[ ein Buchz [V2 gekauft [02 e]]] is metrically non-distinct (in other words, equivalent) from [ein Bt’rchz [V2
gekault ]] since empty categories are metrically invisible (i.e. do not count toward the computation of the
NS). Since there is no selectional ordering, but there is an ordering in terms of asymmetric c-command,
the C-NSR applies assigning prominence to the rightmost sister, [ ein Birch: [V2 gekauft [e2 1]]. Then the
cycle is repeated for the metrical sisters D2 (=ein Buch) and V2 (=[gekault2 [Dz e]]]. These two sisters are
selectionally ordered since there is a selecting head, gekauﬁ. The S-NSR applies assigning prominence to

226

of the NSR that does not take selectional properties into account she calls the
C(onstituent)-NSR. She also argues that both the S-NSR and C-NSR are active in
German and English with only one difference: in German the S-NSR takes precedence
while the S-NSR and the C-NSR are unordered in English. She states the NSR as

follows (p71):

(5) NSR
Given two sister nodes C, and Cj, (a) if C, and C,- are selectionally
ordered, in the sense of (57) (=(58)) [(i) in fn. 5], the one lower

in the selectional ordering is more prominent, (b) otherwise, the one
lower in the asymmetric c-command is more prominent.6

5.1.1.2 The C-NSR and Prosodically-Motivated Movement

We have just noted that both modules of the NSR, according to Zubizarreta (1998),
namely, the S-NSR and C-NSR, are active in both German and English, ordered in the
former but unordered in the latter. In Romance languages such as Spanish, on the other

hand, the S-NSR is not active since these languages position the NS on the lowest

 

the lower constituent D2 (=ein Buch) (Notice that [ein Buch] is lower than the verb since it has moved
from complement position to the right of the verb). Buch is therefore assigned the NS.

6 Zubizarreta deﬁnes c—command as in (i) and argues that at least a weaker version of Kayne's (1994)
notion of asymmetric c-command and linear precedence holds as in (ii):

(i) or c-commands 13:,” a and B are visible to the syntactic computation (i.e. either heads or maximal
projections (excluding segments)) and (a) a and B are sisters or (b) there exists a 7 such that a and y are
sisters and y dominates [3.

(ii) Given two constituents A and B, if A asymmetrically c-commands B then every terminal that A
dominates precedes every constituent that B dominates.

Zubizarreta adopts this formulation of c-command in order to avoid the contradiction that would result in
requiring sisterhood and asymmetric c-command in formulating the C-NSR. She also adopts what she
calls "metrical invisibility" in order to expand on the notion of sisterhood. Some elements in the sentence
are metrically invisible such as anaphoric DPs in English and German as in "Mary walked in. John kissed
her." and defocalized elements that are outside the focal domain. Empty categories are also metrically
invisible. Therefore, no one-to-one correspondence necessarily exists between syntactic sisterhood and
metrical sisterhood since metrical sisters could be interrupted by defocalized elements.

227

constituent in the c-command domain; therefore, only the C-NSR is active in Spanish.
Since only the C-NSR is operational in Romance another type of movement can be
motivated, not driven by feature checking (as in Chomsky 1995), but motivated by
prosodic considerations. Such movement Zubizarreta labels prosodicalIy-motivated
movement (or p-movement).

The C-NSR in Spanish, for example, is responsible for the contrasts seen in (6)
between the Spanish examples and the English ones (note that stressed constituents are
underlined in Spanish, accented in English). (6a) is ungrarnmatical with prominence on
(EL) bebe’, unlike the English example (6c) where phrasal prominence can be placed on
(the) baby. As seen in (6b) and (6e), once the NS is located on Ilora and salié, being the
lowest elements in the c-command ordering, the examples are acceptable again. Similar

contrasts between Spanish and English can also be seen in (6g-l).

(6) a. *El bebe' llora. vs b. Elbebéllora
c. The baby’s crying.

(1. *El g1 salio. vs e. E1 sol salio
f. The sun came out

g. Maria me regalo la botella de m
h. Maria to-me gave the bottle of wine

i. El sindicato hablo contra el gobierno
j. the union talked against the government

k. Ana escondio debajo de la cama la mufleca
Ana hid under the bed the doll
1. Ana hid the (1611 under the bed'

It is clear, then, that only the C-NSR is operative in Spanish which Zubizarreta

228

h

(1998:124) formulates as follows:

(7) C-NSR in Romance

Given two sister nodes C, and C,, the one lower in the asymmetric
c-command ordering is more prominent.

Stressing a constituent other than the lowest in asymmetrical c-command in the Spanish

examples (6) would result in a contrastive reading as in (8):

(8) EL SINDICATO hablo contra el gobiemo (y no el partido)
THE UNION talked against the government (and not the party)

Alongside the NSR, there exists another rule that Zubizarreta (1998: 21) calls the
F (ocus) (P)rominence (R)ule. This rule is active since the focus structure of the sentence

is determined by the placement of the main phrasal prominence:

(9) FPR

Given two sister nodes C, (marked [+F]) and C,- (marked [-F]), C, is
more prominent than Cj.

A fundamental difference between Germanic and Romance is that in the former
languages defocalized constituents are reanalyzed as metrically invisible to the NSR. In
Spanish, on the other hand, all phonological material is metrically visible to the NSR. It

is thus inevitable that conﬂict would arise between these two rules, the NSR and the

229

F PR. Germanic, as has been just mentioned, resolves the conﬂict by marking the
constituents that are not focal as metrically invisible.7 This option, however, is
unavailable to Spanish and the conﬂict that may arise between the requirements made
by the NSR and the FPR would be resolved by moving or shifting defocalized
constituents around so as to ensure that the focalized constituent is assigned stress by the
C-NSR. Such is prosodicaIIy-motivated movement.

SVO order and VSO order in Spanish admit a focus-neutral interpretation where the

last constituent gets the NS (Zubizarreta 1998: 125):

(10) a. Maria me regalo la botella de vino.
Maria to-me gave the bottle of wine

b. me regalo Maria la botella de vino
to-me gave Maria the bottle of wine

If Maria gets assigned the NS instead of the lowest constituent in terms of c-command

then it becomes contrastive,

(l l) MARIA me regalé 1a botella de vino (no Juan)
(not Juan)

(12) me regalo MARIA la botella de vino (no Juan)

where the presupposition is denied (or reasserted). Sentences that are generated by the

 

7 So in (6c) above, since the baby is accented, the VP is crying would be rendered metrically invisible.
Zubizarreta signals metrical invisibility by italicizing the constituents affected as in (i):

(i) The baby is crying.
This possibility does not exist in Spanish, and PA.

230

E‘W-c-n-‘l

 

contrastive/emphatic rule such as in (11) and (12) cannot be an answer to a question

such as (I 3) which places narrow focus on the subject:

(13) Quién te regalo la botella de vino?
who to-you gave the bottle of wine?

Another order is also possible in Spanish, namely VOS. VOS order, unlike VSO and
SVO, is not compatible with a focus-neutral interpretation (where the subject (S) in the
VOS sentence is not right-dislocated). It only admits to a narrow focus on the subject

and the interpretation is not contrastive:

(14) me regalo la botella de vino {Maria}
to-me gave the bottle of wine {Maria}

(14) is the only answer to the question in (13) with a full sentence (where narrow focus
is placed on the subject). The main stress in (14) is generated by the NSR, not the
Emphatic/Contrastive Stress Rule (Zubizarreta 1998:126). The main stress is on the
subject, and it is impossible to focus the object in such sentences, unless, of course, the
subject is right-dislocated, in which case the subject would be set off from the rest with
an intonational break (notationally, a comma). However, this is not the case in either

(15) or(l6).

(15) *me regalo la BOTELLA de vino Maria
to-me gave the BOTTLE of wine Maria

231

(16) *me regalo la botella de VINO Maria
to-me gave the bottle of WINE Maria

Since the subject gets the NS in VOS sentences it can be argued that the VOS order is
derived from either VSO order or SVO order with the ultimate goal of stressing the
subject. Zubizarreta argues that in Spanish VOS order is derived from VSO by moving

VP; that contains the object to left adjoin to VP, as in (I 7):

(17) ['11) me regalo [vp1[ vpz la botella de vinok [ek [V2 ek]]], [vpr Maria [V,
[61111]]

to-me gave the bottle of wine Maria

The reason for movement is prosodic, not feature checking;8 hence the name
prosodically motivated movement (p-movement).

To summarize, Zubizarreta's (1998) analysis revolves around a modular view of the
NSR. The difference between Germanic and Romance in this respect is which module
of the NSR is operative in the language. In Spanish, only the C-NSR is active while
both modules, the S-NSR and the C-NSR, are operative in Germanic. Another rule is
active in these languages, namely, the FPR, resulting in potential conﬂict from the
overlap between the NSR and the FPR. Germanic and Romance possess different
strategies in resolving this conﬂict; in the former by marking invisible defocalized

constituents, in Romance by resorting to prosodically-motivated movement.

 

8 In other words, the movement is motivated by the need to stress the subject by placing it into a position
where the NSR is operational (in Spanish, lowest in the tree in terms of c-command).

232

“'“I

ll *3-

 

Prior to an illustration of how these notions of focus and nuclear stress, and
prosodically-motivated movement are applicable to PA, I would like to brieﬂy consider
Reinhart's exposition of Cinque's (1993) work on stress assignment. My modest goal
here is to demonstrate that Cinque's analysis is compatible with the empirical facts in

PA, as will also be demonstrated of Zubizarreta's (1998) analysis.

5.1.2 Cinque (1993)

In this section I consider brieﬂy Reinhart's (1995) to-the-point exposition of
Cinque's (1993) analysis of stress assignment. It has to be pointed out, however, that
Reinhart's take on and implementation of the analysis discussed in Cinque's work differ
in very minimal ways with no difference in the outcome, as she points out (1995: 27).

Cinque's (1993) proposal basically revives Chomsky's (1971) analysis of focus as
being determined with reference solely to the PF structure itself without reference to
discourse (Reinhart 1995: 20). The core of Cinque's analysis is that sentence stress
assignment, as deﬁned on PF structures, and as assigned by the Nuclear Stress Rule
(N SR), does not need to be parameterized in order to capture the varied stress
assignment patterns across languages. Rather, the NSR can be assumed to apply
cyclically to syntactic constituents assigning sentence stress to the 'most embedded'
element in the sentence. Of course, the notion of 'most embedded' becomes crucial here
especially when cross-linguistic differences are considered, such as the difference
between English and Dutch. In a sentence such as (18), as Reinhart (1995: 26) points
out, identifying the most embedded constituent does not seem to be straightforward

when English and Dutch are compared:

233

(18) a. I read the book.

b. (Dat) I bet boek las (Dutch).9

The crux of the problem, Reinhart points out, is that if it is straightforward to say that
(the) book in (18a) is most embedded, hence assigned sentential stress, complications
ensue if we say that stress is assigned in Dutch in a left-headed fashion (to get the stress
correctly in (1 8b)) since that would generate the correct stress for (18b) (since boek is
leftmost in the VP), but will not work in the case of intransitive sentences (since the
subject would be leftmost) (Reinhart, however, does not illustrate this possibility).
Cinque's solution to this problem is simple. We need to know the direction of the
branching or recursion in a language anyway. The most embedded constituent would
then be identiﬁed in tandem with the direction of the recursion in the language. English,
being right-branching as in (19a), would identify (the) book in (18a) as the most
embedded constituent/element. Dutch, on the other hand, being left-branching in the VP

as in (19b), would single out back as most embedded in (18b).

 

9 Reinhart does not provide an independent gloss to the example in (18b), which I take to mean that it is to
be translated in a similar fashion to the English example in (18a).

234

't‘-CLW
. . . .5.- I .n’

Sentential stress would thus fall on (the) book in (18a) and on boek on (18b), both being

the most embedded constituents/elements in their respective structures.

Cinque also argues that the neutral focus is assigned in the sentence according to the

sentential stress assignment on the PF structure in accordance with the following focus

rule (Reinhart 1995: 30):

(20) The Focus Rule

The focus of IP is a(ny) constituent containing the main stress of IP, as
determined by the stress-rule.

The Focus Rule in (20) would then predict that in a sentence such as (21) any of the

three constituents bearing sentential stress could serve as focus:

(21) [My neighbor [is building [a desk H]

a: at a:

a) NP cycle: [* l
b) VP cycle: [ * ]
c) IP cycle: [ * ]

On the word cycle every constituent is assigned an asterisk. On the NP cycle, the NSR

assigns the most embedded word an asterisk, in this case a desk (or the NP (i.e. DP) a

235

I‘mﬂ

 

 

desk)), on the ﬁrst line (a). On the VP cycle, the most embedded word in the VP, again a
desk, is assigned an asterisk, on the second line (b). Finally, on the IP cycle, a desk,
being the most embedded word in the IP, is assigned an asterisk, on line (c). Sentential
stress is then assigned by the NSR to a desk on all cycles. Now, any of these three
constituents, i.e. NP, VP, or IP, can serve as focus, in accordance with the Focus Rule
above, as Reinhart's following example illustrates (her example (26a-e), p. 30) (stress is

indicated by bolding, focus by underlining):

(22) a. What's this noise?
-Mv neighbor is building; degk.

b. What's your neighbor doing?
-My neighbor is building a desk.

c. What's your neighbor building?
-My neighbor is building a desk.

d.-Has your neighbor bought a desk already?
# My neighbor is building a desk.

e. -Who is building a desk?
#-My rﬁighbor is building a desk.

(22a) illustrates the instance where the IP constituent is focused with sentential stress
assigned to the most embedded word or element, namely, a desk. When the VP
constituent is focused we get the focus structure and stress assignment in (22b). Finally,
(22c) illustrates the case where the stress assignment and focus coincide to make the
most embedded word in any and all constituents most prominent, namely, a desk.

Notice that the subject my neighbor and the verb alone, building, are not possible

236

.I ’ m

"

 

neutral foci according to the Focus Rule. The subject and the verb can be assigned focus

by another rule, namely, the Marked Focus Rule:

(23) Marked Focus Rule:

Relocate the main stress on a constituent you want to focus.

The Marked Focus Rule, when applied, generates (24a) and (24b), for (22d) and (22e),

respectively:

(24) a. My neighbor is building a desk.

b. My neighbor is building a desk.

As Reinhart (1995: 32) points out, this is essentially the essence of Cinque's
analysis, namely, the distinction between neutral and marked stress and it is precisely
here where such analyses have been challenged (due to the idea that a distinction needs
to be systematically drawn at the sentence level between 'neutral' and 'marked' stress).
The opposing analyses look at the matter differently in that they consider focus as
determining sentential stress, rather than the other way around (Reinhart (1995: 32)).

I think this brief look at Cinque's (1993) analysis suffices to enable us to consider
whether or not his analysis would prove applicable to PA data. In the next section I will
show how both Zubizarreta's (1998) and Cinque's (1993) analyses are compatible with

the data from PA.

237

5.2 Focus, Sentential Stress and Prosodic Movement in PA

Upon reviewing Zubizarreta's (1998) analysis, two questions become important.
First, does PA follow the pattern of nuclear stress assignment presumably typical of
Germanic-where both the S-NSR and the C-NSR are active, or would it instead follow
the Romance pattern-where only the C-NSR is active? The second question would be
whether or not prosodically-motivated movement could be attested in PA, or English for
that matter? In what follows I take up these important questions in turn.

Concerning the ﬁrst question, whether PA follows the Germanic pattern or the
Romance one, let us start by considering the following sentences (I follow Zubizarreta

(1998) in underlining stressed constituents):

(25) a. *l-Tiﬁl bibki.
the-baby (is)-crying
b. l-Tiﬁl 121110;
c. The baby is crying.

(26) a. *l-sams Til‘i‘aT.

the-sun came-out
b. l-sams Til‘r’aT

c. The sr’m came out

(27) a. Slama ?a9Tatni gazazitl-S’aSiir

Salma gave-3FS-1MS bottle the-juice
'Salma gave me the juice bottle.’

b. l-‘Bittihaad Sawwat Ded l-hukuume

the-union voted against the-govemment
'The union voted against the government.‘

c. Salma xabbat taht s-sriir l-luSTbe

Salma hid-3FS under the-bed the-doll
'Salma hid the doll under the bed.’

238

d. Salma xabbat l-lu‘i‘be taht s-sriir

Sahna hid-3FS the-doll under the-bed
'Salma hid the doll under the bed.’

The contrast seen between (25a-b) and (26a-b), on the one hand, and (25c) and (26c), on
the other hand, shows that PA and English (can) differ in the placement of nuclear
stress. English can stress (the) baby and (the) sun in situ, i.e. where the stressed element
is not moved to a position lowest in c-command, or most deeply embedded, as seen in
(25c) and (26c). By contrast, the PA examples above illustrate that to stress a
constituent in this language it needs to be placed in a position lowest in the syntactic
tree in terms of c-command or most deeply embedded. So, in (25a), (non-contrastively)
stressing the preverbal DP, l-Tifil 'the-baby’ while destressing the VP bibki '(is)-crying'
is unacceptable. Similarly, assigning the nuclear stress to l-sams 'the-sun' while

destressing Til f’aT 'came-out' in (26a) results in an unacceptable outcome. These

observations are further supported by the examples in (27) where only the constituent
lowest in c-command is assigned the nuclear stress. The contrast seen in (27c) and (27d)

is especially instructive. In (27c), the focused constituent is I-lu Ybe 'the—doll' and the

sentence means that what Salma hid under the bed was the doll. In contrast to (27c),

(27d) says that where Salma hid the doll was under the bed, and the PP, taht l-sriir

'under the-bed' gets the NSR and is therefore focused.
It is important to note at this point that embedding a sentence does not change the
stress assignment facts just observed. This is attested in the following examples where

some of the examples in (25)-(27) are embedded within super-ordinate clauses:

239

(28) smii’it ?inno l-Tiﬁl bibki.

heard-1 S that the-baby (is)-crying
'I heard that the baby was crying'

(29) hakitil-hum finno Slama ?39Tatni gazazit l-SfaSiir

told-1 S-3MP that Salma gave-3FS-l MS bottle the-juice
'I told them that Salma gave me the bottle of juice'

(30) ma bagdar-is ‘Pinno fa?akkid ?inno l-Ittihaad Sawwat Ded l-hukuume

neg able-neg that conﬁrm that the-union voted against the-government
'I can't conﬁrm that the union voted against the govemment'

It is thus clear that embedding does not change the facts concerning nuclear stress
assignment. The stressed constituent continues to be the one lowest in terms of the c-
command ordering or the most deeply embedded.

The preceding discussion also allows us to answer the second question raised at the
outset: does PA utilize what Zuizarreta (1998) labels as prosodically-motivated
movement? I believe the answer to this question is positive. First of all, note that, since
only the C-NSR is active in PA, stressing a sentence-intemal constituent would give rise
to a contrastive reading to the sentence as (313) and (323) clearly illustrate for (25) and

(27b), respectively. Notice also the obligatory absence of nuclear stress on bibki '(is)-

crying' and l-l‘lukuume 'the-government', as the ungrarnmaticality of (3 lb) and (32b)

attests:

(31) a. L-TIFIL bibki (mis 2immo)
THE BABY (is)-crying (not his mother)

b. *L-TIFIL bibki (me time)

240

(32) a. L-ITIThAAD Sawwat Ded l-hukuume (mis l-hizib)
THE-UNION voted against the-government (not the-party)

b. *L-ITTIhAAD Sawwat Ded l—hukuume (mis l-hizib)

This would also predict that cases of conﬂict between the NSR and the FPR would be
resolved by moving the defocalized constituents so the focused constituent would be
assigned stress by the NSR, as has been observed for Spanish. Therefore, a felicitous
answer to (34), where the subject would be stressed (i.e. focused), with a full sentence,

is (35), rather than (36):

(34) miin 1’39Tatak l-xubiz?

who gave-3FS-IMS the-bread
'Who gave you the bread?’

(35) i’aYTatni l-xubiz Salma

gave-3FS-1MS the-bread Salma
'Salma gave me the bread'

(36) * Salma 239Tatni l-xubiz '0
Salma gave-3FS- 1 MS the-bread

In (36), a conﬂict ensues between the NSR, which would want to assign the nuclear

 

'0 (i) would not be a felicitous answer to (34) either since Salma in (i) is contrastively focused. A
felicitous context question for (i) would be (ii):
(i) SALMA ‘23? Tatni l-xubiz
SLAMA gave-3FS- 1 MS the-bread
'SALMA gave me the bread.’
(ii) fa‘r’T‘atak l-xubiz LAILA?
g3ve-3FS-2MS the-bread LAILA
'(who) gave you the bread Laila?‘

241

stress to l-xubiz 'the-bread' by virtue of being lowest in terms of c—command, and the
FPR, whose job would be to assign stress to the subject Salma (as an answer to the CQ
in (34)). This conﬂict is sorted out by moving the constituent that is focused to be
placed in the lowest position in terms of the c—command ordering. This means that
l-xubiz 'the-bread,‘ being defocalized, would be moved to a position where it would be
outside the scope of the NSR, as in (3 5), and the word order resulting would be VOS.
Further support for this analysis comes from the fact that l-xubiz ‘the-bread’ in (3 5)

cannot be (contrastively) stressed, unless Salma is right-dislocated:

(3 6) * ?a‘i‘Tatni L-XUBIZ Salma

gave-3FS-IMS THE-BREAD Salma
'Salma gave me the bread.’

The preceding discussion allows us to conclude that PA follows the Romance
pattern of nuclear stress assignment where only the C-NSR is active. The constituent to
which nuclear stress is assigned has to be placed lowest in the syntactic tree terms of
c-command. Furthermore, prosodically-motivated movement has been demonstrated to
take place in PA, as the case is in Spanish.

Cinque's (1993) assumptions regarding stress assignment can also be assimilated
quite nicely into the current framework with respect to PA. PA is similar to English in
terms of the direction of recursion in that it is also right-branching. Therefore, the most
embedded constituent bears the stress and the pattern seen for English in (223-c) above,

repeated here as (3 73-0), can be shown to hold for PA as well as seen in (3 8):

242

.2-1
in

’1‘... '

 

(3 7) 3. What's this noise?
-My neighbor is building a desk.

b. What's your neighbor doing?
-My neighbor is building; desk.

c. What's your neighbor building?
-My neighbor is building a desk.

(38) a. suu haaﬁi d-dad33e?
what this the-noise

- b-yibni dzaari b-Tawlg”

is-building neighbor-l S (3)-table
'My neighbor is building a table.‘

b. suu b-yi‘i'mal d3aarak?

what is—doing neighbor-2M8
'What is your neighbor doing?’

-d33ari b-yibni b—Tawla
neighbor-1M8 is-building (3)-table
'My neighbor is building a table.‘

c. suu b-yibni dgaarak?

what is-building neighbor—2M8
'What is your neighbor building?’

- dsaari b-yibni b-Tawla

neighbor-1 MS is-building (3)-table
'My neighbor is building a table.’

(383) represents the option where the whole sentence is focused. In this case the NS is
assigned to the most embedded constituent, T awla 'a table,‘ although the whole sentence

is in focus. In (38b), only the VP is focused and the nuclear stress is still assigned to the

 

” In a so—called out-of-the-blue answer, where the whole sentence is in focus, the more preferred word
order is VS(O), rather than SVO.

243

most embedded word in the sentence, namely, Tawla 'a table.’ Finally, when only the NP
(i.e. DP), Tawla 'a table' is focused, it also gets the NS, as in (38c). So far, English and
PA behave similarly. Now consider the so-called marked focus in English as

exempliﬁed by (22d-e) above, repeated here as (3 7d-e), and their PA equivalent (38a-b):

(3 7) d.-Has your neighbor bought a desk already?
# My neighbor is building a desk.

e. -Who is building a desk?
#-Mv neighbor is building a desk.

(3 8) a.?i§tara dsarak Tawla?

bought-3M8 neighbor-2M8 (3)-table
'Has your neighbor bought a table?’

-#d33rai b-yibni b-Tawla
neighbor-2M8 is-building-3MS (3)-table
'My neighbor is building a table.'

b. miin b-yibni b-Tawla?
who is-building (3)-table

-# d3arai b-yibni b—Tawla
neighbor-3M8 is-building-3MS (3)-table

Notwithstanding the similarity in the grammaticality judgments in (3 7d-e) and (383—b)
between English and PA, these two languages go about resolving the unacceptability of
these sentences in completely different ways. In other words, English and PA diverge in
their implementation of Cinque's marked focus. English, as has been noted in (243-b)
above, repeated here as (39a-b) for convenience, resorts to using marked focus by

destressing a desk, i.e. the constituent that would be assigned the NS by the NSR, and

244

stressing the focused element, building and my neighbor, respectively,

in situ:

(39) a. My neighbor is building a desk.

b. My neiﬂbor is building a desk.

PA, on the other hand, does not resolve the conﬂict between stress assignment by the
NSR and focus via destressing the most deeply embedded constituent, while leaving it
in situ, as the case is in English. Rather, as I have noted above, PA defocalizes the non-
focal constituent by p-moving it to a position outside of the scope of the NSR, and by
placing the focalized constituent in the most embedded position within the scope of the
NSR as witnessed in (40b) (the PA counterpart to the English (3%)). If this proves to be
unavailable, and p-movement does not turn out to be an option as in (403) (the PA
counterpart of the English (393)), the focused element in PA would be read contrastively

(since it is focused internal to the sentence), unlike what happens in English.‘2 '3

(40) a. dgarai B-YIBNI b-l-Tawla

neighbor-2M8 IS-BUILDING-3MS the-table
'My neighbor is building the table.’

 

'2 I have replaced the indeﬁnite '3 table' with the deﬁnite 'the table' since, as I have argued in Chapter 3,
the singular indeﬁnite is subject to the same conditions as plural BNs in PA, in that all these nominals
have to be assigned stress by the NSR, or contrastively focused. In (40b) in the text, if Tawla '(a)-table' is
defocalized and moved to the left in order to assign the NS to d3arai 'neighbor-l MS,’ this would generate
an unacceptable outcome since the singular BN would be defocalized and destressed.
'3 P-movement is unavailable in (403) since the outcome of moving the verb to the right periphery is very
marginal as in (i):
(i) ‘?d33rai b-l-Tawla b—yibni

neighbor-2M8 the-table is-building-3MS

245

b. b-yibni b-l-Tawla d33rai

is-building-3MS the-table neighbor-3M8
'My neighbor is building the table.’

The crucial examples here are (39b) and (40b). One immediate observation concerning
(39b) is the following. English, (39b) clearly says, does not have the option of
Zubizarreta's prosodic movement. This sets English apart from both PA and Spanish
since the latter two languages make use of this option, i.e. the option of focalizing or
stressing constituents by moving them into the scope of the NSR (or, alternatively,
defocalizing or destressing a constituent or element by moving it to a position outside
the scope of the NSR). PA and Spanish, then, do not need to (in fact, they cannot) resort
to the so-called marked focus that English utilizes since a 'less marked' focusing option
is available via prosodically-motivated movement.‘4

This ﬁmdamental difference between English on the one hand, and PA and Spanish,
on the other hand, is in fact quite expected. These two groups of languages differ in
another important respect, namely, that of the word order possibilities available to them
(Cf. Reinhart (1995)). Recall that English word order is generally more restricted and
less ﬂexible than that of either PA and Spanish. Word order in English is more or less a
ﬁxed SVO. PA and Spanish, by contrast, allow SVO, VSO, VOS, among other

permutations. It should come as no surprise then that English would allow in situ focus

 

'4 Reinhart (1995: 32) makes similar observations in her exposition of Cinque's (1993) analysis of the
differences between Italian and English. She notes that focusing in-situ, as is the case in English, is a more
marked choice (hence, marked focus) than moving elements around to express neutral focus as is the case
in Italian. The notion of markedness, and the related notion of economy, is a thorny issue since movement,
especially in terms of Chomsky's (1995) minimalist assumptions, is always the less economical, hence
more marked, option.

246

(i.e. marked focus) since this is the only option that is available for this language.
Prosodically-motivated movement, as has been just noted, is generally unavailable to
English. If English had a less rigid word order, like PA and Spanish, then the option of
marked focus would be quite unacceptable. This is actually the case for PA and Spanish
since word order is relatively free in these two languages. Therefore, using the more
marked option- i.e. marked focus- (ch fn. 14) is not acceptable non-contrastively.
Consider the contrast seen between the English examples in (41) and the Italian ones in
(42), which is essentially the same contrast noted between English and PA and Spanish
(the examples are cited in Reinhart (1995: 32) from Cinque (1993)) (bolding indicates

stress):

(41) a. Johnson died.
b. Johnson died.
(42) a. Johnson é morto.
b. E morto Johnson.

c. # Johnson é morto.

English can stress the subject or the verb in situ, as seen in (4la-b). The same cannot be
said of the Italian examples in (42). To stress the subject Italian resorts to (prosodically-
motivated) movement, thus moving the subject Johnson to the most deeply embedded
position to get assigned stress by the NSR. To use the more marked focus option in
(42c) would be unacceptable with the stress on the subject that is not read contrastively.

Of course, the same paradigm in (42) can be reproduced in PA, and Spanish, I am sure.

247

The only available reading for the equivalent of (42c) in PA would be where Johnson is
interpreted contrastively. Again, (42c) turns out to be unacceptable in an out-of-the-blue
utterance since a 'less marked' option is available, attainable via prosodic movement as
in (42b), and its corresponding equivalent in PA.

With the preceding discussion in mind, I believe we are now in a position to address
our main concern here, namely, the reason why bare nominals differ the way they do
between English, on the one hand, and PA and Spanish, on the other, in terms of

distribution. This will be the subject matter of my next section.

5.3 The Distribution of Bare Nominals and Focus in English and
PA / Spanish

Let me very brieﬂy recapitulate my assumptions so far. Stress assignment in PA and
Spanish is achieved via the C-NSR, where the lowest constituent in terms of c-
command, or most embedded, is assigned stress. The conﬂict between the FPR (in the
terms of Zubizarreta (1998)) and the NSR, or alternatively, between the focus structure,
expressed via the Focus Rule and PF stress assignment (in the terms of Cinque (1993)),
is resolved in PA and Spanish through prosodically-motivated movement (a la
Zubizarreta (1998)). In English, on the other hand, this conﬂict is resolved via
destressing non-focal constituents through using the option of marked focus. Prosodic
movement is generally absent from English. '5 I have also noted that these differences

between both language groups occur concurrently with the more fundamental

 

'5 Abstracting away ﬁ'om Heavy NP Shift, if it turns out that this option is in fact an instance of prosodic
movement.

248

differences in word order. Word order in English is more rigid, almost a ﬁxed SVO.
Unlike English, PA and Spanish are privileged with a ﬂexible system of word order
possibilities.

Let me further recap my assumptions regarding BNs in these two language groups.
BNs in PA and Spanish, I have argued, are always existentially interpreted and,
additionally, focalized. The generic reading for these nominals in not possible. English
BNS can be read both existentially and generically. In their existential usage English
BNs also turn out to be focalized. Genetically interpreted BNS in English, and their PA
and Spanish determined DPs (with a deﬁnite article) counterparts, cannot be focused,
unless they are read contrastively (and in that case they are read more plausibly as
contrastive t_opi£s, rather than fgcj), as has been explained in Chapter 4.

As we have seen in Chapter 3, the claim in the literature has been that BNs in
Romance, and by argument, also PA, are subject to a lexical government requirement. I
have argued against this idea claiming that a more uniﬁed approach is called for to
account for the distribution of these nominals. The distribution of BNs in PA/Spanish, I
have argued, is constrained rather by the requirement that they be non-topical, i.e. focal.
English differs in this respect in that its BNS can, but do not have to, be focal. The
difference between the distribution of BNs in the two language groups (families?), could
therefore be, as I have suggested at the beginning of the present chapter, closely
reﬂective of the difference in how Eggs is achieved in these languages.

For concreteness, consider the following examples from English and PA:16

 

'6 I use PA as also representative of the facts in Spanish.

249

(43) 3. Boys played in the park.
b. # Boys played in the park (not girls).

(44) 3. REM f-l-hadiiqa wlaad
played-3MP in-the-park boys
'Boys played in the park.’

b.* wlaad liSIbu f-l-hadiiqa
boys played-3MP in-the-park

c. WLAAD liS‘bu f-l-hadiiqa (mis BANAAT)
BOYS played-3MP in-the-park (not GIRLS)

d. lifbu WLAAD f—l-hadiiqa (mis BANAAT)
played-3MP BOYS in—the-park (not GIRLS)

Boys in (43), according to my assumptions so far, is focused since it is interpreted
existentially (1 indicate that through underlining). This is basically the only word order
option available for English, other permutations being unacceptable. Crucially, Boys in
(43) is n_ot contrastively focused, as in (43b) (or, at least, does not necessarily have to be
contrastively focused). Now, the PA word-for-word equivalent of (43), namely (44b), is
ungrarnmatical since the BN wlaad 'boys,‘ which still gets the existential interpretation
typical of all PA BNs, occurs under normal conditions of stress (i.e. unstressed, hence
non-focal). Since the C-NSR is the only viable option for PA, to be assigned stress the
BN wlaad 'boys' has to move to the lowest position in terms of c-command, or the most
embedded position in Cinque's terms, where it would be assigned the NS by the NSR.
This would get us (443), which is the acceptable equivalent of the English (43). wlaad
'boys' in (443) is assigned the NSR in its most deeply embedded position. Therefore, just

like its English counterpart in (43), it is non-contrastively focujsgd in th_at position. Now,

250

signiﬁcantly, to occur in any other position, i.e. outside the scope of the NSR, wlaad
'boys' would only get the contrastive reading as in (440-d).

Due to its rigid word order, English can make use of the so-called marked focus, as
in (433), where the BN is not most deeply embedded to be assigned stress by the NSR
(in terms of Cinque (1993)). Notwithstanding its sentence-intemal position in the
English sentence, the BN is focused non-contrastively. Conversely, due to its ﬂexible
word order, PA does not have the option of using marked focus available to it. This is so
since a 'less marked' focusing option is available, viz. that of prosodically-motivated
movement, thus displacing the defocalized element and removing it ﬁ'om within the
scope of the NSR. Using the marked focus option in PA, i.e. by stressing the BN in a
position that is sentence internal as in English, would in fact result in a contrastive
reading, as witnessed in (44c-d).

To sum up, these underlying differences between English and PA are, then,
responsible for the distinct and differing distribution of BNs in these two languages (or
language groups, counting Spanish in). On the one hand, English enjoys a ﬂexible focus
system, while it suffers from a rigid word order. PA and Spanish suffer from a rigid
focus system, while they are privileged with ﬂexible word order permutations. In fact,
one is tempted by this intriguing state of affairs to hypothesize that because English has
such an impoverished or restricted word order it compensates for that by a rich focus
system (or, perhaps, vice versa). We can additionally and equally hypothesize that
because PA and Spanish have an impoverished or restricted focus system, they make up
for that by permitting different word order possibilities (or, again, quite logically, vice

versa).

251

In the next section I would like to consider an issue that has not been adequately
explored. I have claimed at different points in this dissertation that one way to treat
focalization is to say that a focused BN moves into a speciﬁer position of 3 Focus
Phrase (FP). However, I have not spelled out how this would license the BN, how being
in an FP interacts with the NSR, and ﬁnally, at what level moving into an FP can be

assmned to take place. To these issues I now turn.

5.4 Focus Phrase (FP) and Bare Nominals

My basic assumption has been that to be informationally focused in PA and Spanish
the bare nominal has to be in the lowest position in terms of the c-command ordering
(or, the most deeply embedded position) so as to be assigned the NS by the C-NSR.
Otherwise, in all other positions, the bare nominal cannot be informationally focused,
but has to be contrastively focused. At this point, Rizzi's (1997) conception of clausal
structure becomes relevant. In the following remarks I present a truncated discussion of
his assumptions in this work.

Rizzi (1997: 285) points out that traditionally the left periphery of the clause has
been said to involve dichotomies such as topic and comment as the following

topicalization English example demonstrates (his (1)):

(45) Your book, you should give t to Paul (not to Bill).

In Italian, Rizzi points out, the topic-comment dichotomy is traditionally expressed by

the so-called Clitic Left Dislocation (CLLD) structures where a resumptive clitic is

252

involved in the comment coreferential with the topic as in the Italian example in (46)

(his example, (3)):

(46) II tuo libro, lo ho letto
"Your book I read (,not his)"

Another dichotomy that has been said to involve the left periphery of the clause is

the focus-presupposition articulation as seen in (47) (his (2)):

(47) YOUR BOOK you should give t to Paul (not mine).

Rizzi gives the following structures for the left periphery of the clause involved the

aforementioned dichotomies (his (5) and (6)):

(483) (48b)
TopicP FocusP
XP Topic' ZP oc'
Top YP Foc WP
XP= topic ZP= Focus
YP= comment WP= Presupposition

Rizzi (p. 287) makes the important assumption that a constituent possessing a topic or

focus feature would end up in a Spec-Head relationship with a Top head or 3 Foo head,

253

respectively. This way movement of a focused or topicalized constituent would be
subsumed under movement for feature checking.
It is also argued by Rizzi (1997: 290) that a clause can involve many topics but that

there is only one unique focus as the examples that follow illustrate (his (21) and (22)):

(49) H libro, a Gianni, domani, glielo daro senz'altro
"the book, to John, tomorrow, I'll give it to him for sure"

(50) *A GIANNI IL LIBRO daro (non a Piero, l'articolo)
"TO JOHN THE BOOK I'll give, not to Piero, the article"

Furthermore, Rizzi argues that a focus and one or more topics can co-occur as seen in

(51) (his (23)):

(51) A Gianni, QUESTO, domani, gli dovrete dire
"To Gianni, THIS, tomorrow, you should tell him"

If we adhere to this conception of clause structure articulated in Rizzi (1997), where
3 Focus Phrase is said to occur in the left periphery of the clause, ﬂanked on either side
by a (number of) Topic Phrases, then the PA SVO order, where S is, of course,
contrastively focused preverbally, is quite straightforwardly accounted for. S would be
in the speciﬁer position of the Focus Phrase, and would be licensed in that position. The
potentially problematic case would be as in (44d) above, repeated below for

convenience,

254

(44d) IiSfbu WLAAD f-l—hadiiqa (mis BANAAT)

played-3MP BOYS in-the-park (not GIRLS)
'Boys played in the park (not girls).'

where the bare nominal WLAAD 'BOYS' occurs postverbally. By assumption, the bare
nominal in this position is contrastively focused. This claim, however, does not seem to
be compatible with the view of the clause structure assumed here where the Focus
Phrase occurs in a left peripheral position. So, the question now becomes how and in
what position the postverbal contrastively-focused S, WLAAD 'boys' in (44d), is
licensed.

It may be argued that a possible answer to this question may lie in the analysis
offered by Belletti (2001). Belletti argues that the postverbal positioning of the subject
(i.e. subject inversion) in Italian should be looked upon as a way of focalizing the
subject. Overt DPs are either licensed by Case or by Focus. The postverbal subject is
therefore licensed in 3 Focus Phrase that is internal to the clause (in the lowest
functional projection, a higher functional position than the VP) and the subject enters
into a checking relation in the speciﬁer of that position (presumably, with the functional
head of the Focus Phrase). Looking at matters this way, the impossibility of the VSO
order in Italian, Belletti points out, and the marginality of the VOS order in that
language, would then be accounted for in terms of Relativized Minirnality (RM)
violations. In the VSO order, the subject, according to Belletti's analysis, would move to
Focus Phrase from its VP-internal position to be licensed. Since the speciﬁer of the
Focus Phrase is ﬁlled by the subject, the object would have to be licensed through Case,

rather than Focus. The object would then have to move into a speciﬁer position of a

255

Case-checking projection to be licensed. But that is not possible since the subject is
sitting in the lowest functional projection of the clause. Further movement of the object,
crossing the subject, would then be prohibited as 3 RM violation. The marginal VOS
order would also be ruled out in a similar fashion, but in this case it is the subject that
gets trapped in the VP, hence unlicensed. The obvious, and immediate, objection to
Belletti's analysis, which she notes herself (2001: 72-73), would be that in other
Romance languages such as Spanish, and also in Palestinian Arabic, both VSO and
VOS orders are acceptable and productive, which is not predicted by Belletti's analysis.
Belletti suggests that this tells us that the subject in Spanish is in a different position
than in Italian. At any rate, I think Belletti's assumptions seem to be inadequate as far as
both PA and Spanish are concerned.

It seems to me that the conception of the clause structure offered by Rizzi (1997)
could still be maintained. Take (44d). From what I have assumed so far, this example

cannot be an answer to (52), but rather (53):

(52) # miin li9bu f-l-hadiiqa?
who played-3MP in-the-park

(53) liS’bu BANAAT f-l-hadiiqa?
played-3MP GIRLS in-the-park

This so since the BN in (44d) is contrastively focused, whereas answering (52) with a

BN in subject position would necessarily entail narrow focus on the subject (whose

stress would be assigned by the NSR), not contrastive focus. So, if the context question

256

to (44d) is (53), not (52), it then follows that li S’bu 'played-BMP' in (44d) is part of the

presupposition, not the assertion (i.e. topical). It can then be assumed that li Ybu 'played-

3MP' would sit in a Topic Phrase projection ﬂanking the Focus Phrase on the left, since
it has a topical feature (as in Rizzi (1997)). WLAAD 'boys' would therefore occur, or

rather, would have moved into, the peripheral Focus Phrase projection. f-l-l‘ladiiqa 'in-

the—park' in (44d) could then be in Topic Phrase on the right edge of Focus Phrase,

again, 3 reasonable assumption in terms of Rizzi's analysis, since f-I-badiiqa carries a

topical feature. The same analysis can be applied to (54):

(54) banu YABIID l-ahraam
built-3P SLAVES the-pyramids

In this case banu 'built-3P' would be in Topic Phrase, while YABIID 'SLAVES' would

be in Focus Phrase. l-ahraam 'the-pyramids' could then be in another Topic Phrase on
the right edge of Focus Phrase, or else it could be said to be in its canonical object
position.

However, how could this discussion be applied in the important case of (44a) in the

text, repeated here as (55):

(55) mm f—l-hadiiqa wlaad

played-3MP in-the-park boys
'Boys played in the park.’

257

Since wlaad 'boys' is focalized, therefore assigned the NS by the NSR in its lowest
position in the syntactic tree in terms of c-command, then it does not seem to sufﬁce, or
work, to claim that wlaad 'boys,' being a focalized constituent, would have to occur,
overtly in the syntax (i.e. at S-Structure), in Focus Phrase in the left periphery of the

clause. However, recall that the context question for (5 5) would be (56):

(56) miin liS’bu f-l-hadiiqa?
who played-3MP in-the-park
'Who played in the park?’

Taking the CQ in (56) into consideration, it becomes evident that li fbuf—l-hadiiqa

'played-3 MP in-the-park' is presupposed, or topical in (55). It could then be argued that

Ii S’buf-l-hadiiqa 'played-3MP in-the-park,‘ since it has a topical feature, is in Topic

Phrase ﬂanking the left edge of Focus Phrase in which wlaad 'boys' would be sitting.
Viewing things in this manner enables us to maintain that a focused BN in PA and
Spanish moves into the speciﬁer position of an FP overtly in the syntax (at S-Structure).
However, another option is still logically possible, as has been suggested by Rizzi

(1997: 287) in regards to the following Italian example (his (7)),

(57) Ho letto IL TUO LIBRO (,non il suo)
"I read YOUR BOOK, not his"

which is to say that when focalized a BN in PA and Spanish moves into the speciﬁer

position of 3 Focus Phrase covertly at the level of Logical Form (LP) to enter into a

258

 

checking relation with 3 Focus head F in a Speciﬁer-Head conﬁguration. This option
would minimize cross-linguistic variation since the level of LF is the relevant level
anyway for English focused BNs, if we abstract away from English examples such as
(47) above, since overt movement to FP (Focus Phrase) is constrained by the rigid word
order in English.

The last issue I would like to consider in this section is how focusing a bare nominal
licenses it. Recall from Chapter 2 that I have claimed that BNs in non-Construct-State
constructions in PA are unmarked for +/- Deﬁniteness features, and are therefore unable
to substitute into D. I have also assumed that the same can be carried over to existential
BNS in English. In Construct-State constructions, on the other hand, a BN is supplied
with the features necessary to check the strong Def features of D, and in the process, the
BN receives the appropriate interpretation. The +/- Defmiteness features are passed on
to the BN by a possessive DP in the construct state (CS). Ideally, focusing a BN in non-
CS construction should perform a similar function to that performed by the possessor
DP, namely, endowing the BN with the necessary features to be interpreted.

I would like to argue that focus does just that. It enables the BN to check its features
against an F head in a Speciﬁer-Head conﬁguration to achieve an interpretation. Focus
is standardly taken to encode new information. It can thus be said to encode
indeﬁniteness features. Even contrastive focus can also be argued to encode new
information and thus signaling indeﬁniteness since the contrasted BN still supplies new

information as witnessed in the next example:

259

 

(58) A. hakeet ma‘t.‘ DAKATRAH?

talked-2M8 with DOCTORS
'(Did) you talked to DOCTORS?‘

B. 13?. hakeet ma‘t‘ MYALLMAAT

no. talked-MS with TEACHERS—3F P
'No. I talked to (FEMALE) TEACHERS.’

If the DP that contains the BN Sits in a speciﬁer position in FP in a checking
conﬁguration with a focus head F, and since focus is inherently indeﬁnite, the
indeﬁniteness features are passed on to the DP containing the BN in the speciﬁer
position. It is then predicted that the only features that can be inherited by a BN sitting
in an FP would be indeﬁniteness features. This prediction is exactly what we encounter
in PA. BNS in this language can m be interpreted as indeﬁnite. Furthermore, the
prediction is also conﬁrmed for English, since the sole interpretation an existentially
read BN can get is indeﬁnite.

In the next section, l tentatively explore another difference between English and PA
and Spanish concerning how generic or, rather, kind reference, is achieved in both

groups in languages.

5.5 Bare Nominals and Reference in English and PA/ Spanish
Chierchia (1998) proposes a neocarlsonian view of bare nominals and suggests that

these nominals, in English and Italian, denote kinds and that the generic and existential

uses of BNS in these languages are accomplished via type-shifting. Although interesting

and well worked-out I believe Chierchia's analysis has some problems. I will not

260

consider a critique of his analysis here. '7 I would just note that Chierchia's assumption
that BNS in Italian denote kinds is quite problematic, and cannot be carried over to PA
BNS (for example, see Zamparelli (2002), who rejects Chierchia's contention that Italian
BNS denote kinds).

I would rather like to suggest that when we deal with generically interpreted BNS in
English, on the one hand, and existentially interpreted BNS, on the other hand, we are in
fact dealing with two somewhat different entities. In fact, this should be the null
hypothesis, since other languages, languages such as PA and Spanish, express these two
usages by resorting to two quite different mechanisms. On the one hand, the existential
usage is expressed via BNS, and the generic usage is achieved by using a deﬁnite article
with the nominal. That generic BNS in English are different from existential ones in this
language has been suggested in Chapter 2. In that chapter, I have argued that generically
interpreted BNS in English are able somehow to satisfy the +/- Defmiteness features of
the D position, unlike existentially interpreted BNS. I have speculated then that two
solutions could be offered to this problem. In accordance with Rizzi's (I997)
assumptions, I have suggested that generically read bare nominals carry a topical feature
that needs to be checked in the proper conﬁguration. I have suggested that these
nominals are structurally located in the speciﬁer position of a TopicPhrase and the DP
in which the nominal is contained enters into a checking relation with the head Top of
that projection (Cf. (483) above). Licensing the bare nominal is then accomplished in

this conﬁguration. Moreover, interpreting the BN would also be accomplished this way

 

'7 Consider for example Mum and Schmitt (2000) and their objections to Chierchia's Nominal Mapping
Parameter and the problems it encounters when applied to Brazilian Portuguese.

261

since generically interpreted BNS are interpreted as topics, presumably, deﬁnite. The
other solution, I have suggested in Chapter 2, is that the generic BN is somehow able to
substitute in D. This position seems to get support from Zamparelli's (2002) contention
that when D is ﬁlled the expression is read referentially.‘8 This does not seem to be
implausible in the case of generic BNS, kind-referring BNS in particular, in English
since these nominals do give rise to a referential reading. Further support for this last
point comes from PA and Spanish where the generic reading, always referential, is
expressed by ﬁlling the D position with 3 deﬁnite article.

So, if generic BNS in English and determined nominals with the deﬁnite article in
PA and Spanish are alike, this begs the question why the former language can express
genericity using BNS while the latter two languages have to use the deﬁnite article for

this purpose.‘9

Let us ﬁrst consider the following diagram:

 

'8 Zamparelli (2002: 14) argues that referentiality could be linked to the presence of lexical material in the
D position (at LF). This, he speculates, is responsible for the narrow-scope reading of the indeﬁnite in (i)
and the wide scope reading in (ii), since in the latter there is no material to substitute in D:

(i) John is looking for a piece of paper.

(ii) John is looking for pieces of paper.
In (i) it is possible to interpret the sentence as meaning that there is a piece of paper such that John is
looking for it. In (ii) such an interpretation is unavailable. However, Zamperelli's analysis assumes that the
indeﬁnite article is in D, which is not uncontroversial.
'9 That generically interpreted BNS in English have more in common with determined nominals with the
deﬁnite article than is usually assumed is in fact the crux of Zamparelli's (2002) analysis.

262

(59)

a. PA BNS English BNS
EX EX Gen Kind
b: PA 'the'-Plurals English 'the'-Plurals
Ex Def Gen Kind Ex Def
c. PA 'the'-Singulars English 'the'-Singulars
EX Def Kind EX Def Kind

The most striking observation that can be made concerning (593-0) is the virtual
symmetry seen between English and PA. Interestingly, (59a) and (59b) exhibit 3 striking
pattern. PA BNS can be used only existentially, whereas BNS in English have a three-
fold usage where they could be used either existentially, generically, or in reference to a
kind. These latter two usages English BNS can be put to are realized in PA by
the-plurals. The differences, thus, should follow from principled reasons.

I believe the differences can be seen to stem from two related assumptions. First, I
would like to make the assumption that generic operators in these two language groups,
English, on the one hand, and PA and Spanish, on the other, differ in their binding
possibilities. In English, generic operators can bind nominals contained in a DP whose
head D is empty or null. In PA and Spanish, by contrast, generic operators are unable to
realize this possibility, and these two languages have to use the deﬁnite article to

express genericity. The second assumption I would like to make, which is not unrelated

263

to the ﬁrst assumption concerning generic operators, is to recall from Chapter 4 my
contention that generically interpreted BNS in English, and generic DPS with the deﬁnite
article in PA, are never foci, but always topical. When focused, these DPs turn out to be
contrastive topics, rather than foci. It could then be hypothesized that to be interpreted
generically in a sentence a BN or a deﬁnite DP have to be topical or defocalized. One
way to cash out this solution is to say that when topical or non-focal a BN or 3 deﬁnite
DP is mapped into the restrictive clause to be bound by the generic operator. Being
focal, BNS in PA and Spanish, and existential BNS in English, do not ﬁt the bill, since
they are arguably always caught in the nuclear scope and cannot be bound by generic
operators.

It should then be expected that English can express genericity using BNS since the
generic operators in this language can bind a nominal contained in a DP with a null head
D. Furthermore, English bare nominals can be defocalized and mapped into the
restrictive clause of a generic operator to be bound by the latter. Conversely, PA and
Spanish generic operators may be taken to be unable to bind nominals in DPS with
empty or null heads. A BN in PA and Spanish is also always focused, which rules out
the possibility of binding the nominal in the restrictive clause since these nominals are
always mapped into the nuclear scope. The latter two languages have to use the deﬁnite
article in order to express the generic use, unlike English, which can express it using a

bare nominal.

264

5.6 Conclusion

I have argued in this chapter that the differences seen between English, on the one
hand, and PA and Spanish, on the other hand, in terms of the distribution of BNS are
quite predictable and predicted from the following assumptions. Existential BNS in
English, and all BNS in PA and Spanish, are always focused and always bear sentential
stress as assigned by the NSR. Since these nominals behave similarly in all three
languages, or perhaps, in the two language groups, then the differences should be sought
in the very fact that all these nominals are focused.

The properties of focus in these two language groups, I have argued, are different.
Such differences in turn reﬂect on how these nominals distribute in the utterance. In
English, both the S-NSR and the C-NSR are active or, alternatively in terms of Cinque's
(I 993) analysis, marked focus is an option. The most embedded constituent does not
have to be focalized but can be destressed so as to focus, non-contrastively, a
constituent that is internal to the utterance. In Spanish and PA, on the other hand, only
the C-NSR is active and the marked focus option open to English is unavailable to the
former languages. To stress a constituent, PA and Spanish move this element and place
it where it can be assigned stress by the NSR (lowest in the syntactic tree in terms of c-
command or in the most deeply embedded position). Using the marked focus option in
PA and Spanish would only generate a contrastive interpretation.

Finally, I have argued that reference in these two language groups can perhaps be
explained through utilizing two assumptions. I have assumed that generic operators are
different in these two language groups. English generic operators can bind nominals in

DPS with null heads, whereas PA and Spanish generic operators cannot. Additionally, I

265

have suggested that if BNS in PA and Spanish are always focal these nominals could not
be caught in the restrictive clause of a generic operator to be bound by that operator.
Assuming that focused BNS are mapped into the nuclear scope, whether in English or
PA/Spanish, would rule out the possibility of being bound by generic operators. These

nominals would instead be bound by existential closure, and be interpreted existentially.

266

 

Chapter 6

Conclusions

I have started out the discussion of bare nominals in Palestinian Arabic by putting
forward this hypothesis: one way of understanding the workings of bare nominals in
Palestinian Arabic, or in any language for that matter, would be for there to be, in the
ideal case, a certain environment in which bare nominals occur, perhaps quite
productively, but in which they display a pattern of behavior manifestly and markedly
different from their behavior in the general case. By closely examining the behavior of
bare nominals in such an exceptional context, and then juxtaposing that behavior with
what is witnessed of bare nominals in the usual case, we should be in a better position to
understand the behavior of these nominals.

Palestinian Arabic (PA), as it turns out, provides us with precisely that option since
this language possesses a construction, namely, the Construct State (CS), in which bare
nominals occur productively and in which their behavior seems pretty unconstrained.
The head noun of the construct state has been standardly assumed, in the Afro-Asiatic
literature, to substitute into the D position. This, I have argued, provides the
underpinnings of a fundamental difference between bare nominals (BNS) in CS
constructions and nominals in non-CS constructions. In other words, BNS in CS
constructions are able to move into D, thereby checking the strong +/-Deﬁniteness
features of that position, once these nominals inherit the necessary matching features
from the possessive DP obligatorily present in the CS. Since, by assumption, BNS are

unmarked for such features, and are only able to acquire them through the presence of a

267

possessor DP, BNS in non-CS constructions fail to perform this function, and the strong
+/- Deﬁniteness features of the D position of the DP shell in which these nominals
occur would remain unchecked. This in turn would render ungrammatical any utterance
in which these nominals would occur and would also explain, for instance, why BNS
can occur preverbally only when they are contrastively focused, since they would not be
licensed otherwise.

I have then attempted to push this line of reasoning a little farther. First, I have
demonstrated, drawing on the placement of nuclear stress via the Nuclear Stress Rule
and the subsequent focal status of the constituent stressed, that BNS in PA and Spanish ,
which are only subject to an existential reading, are always focal (informationally,
otherwise, contrastively, if sentence-intemal). If the BN is assigned the NS, via the
C-NSR module operative in PA and Spanish (in Zubizarreta's (1998) terms), or via the
NSR to the most deeply embedded position (in Cinque's (1993) terms), that nominal
comes out informationally focused. If, on the other hand, the BN occurs internally in the
sentence it surfaces contrastively focused.

I have used the occurrence of bare (or, determinerless) nominals in CS
constructions, and the mechanism by which these nominals are licensed, to argue that
focus performs the ﬁmction of the licenser for BNS in non-CS constructions by analogy
with the function performed by the possessive DP in CS constructions. In CS
constructions the features of D are checked by a bare nominal moving into it overtly due
to the presence of a possessor DP from which the nominal acquires its (in)deﬁniteness
features. In non-CS constructions the BN is licensed by focus. To focus a bare nominal

is to place it in a speciﬁer position of 3 Focus Projection (F P). Since BNS are DPS it

268

follows that the DP in which the nominal occurs would enter into a checking relation
with an F head in a Speciﬁer-Head conﬁguration within the F P projection. Since focus
is inherently a means of encoding new information, I have argued that it is quite
reasonable to assume that focus is always indeﬁnite. This indeﬁniteness is passed on
from the F head onto its speciﬁer via the very local relationship that holds between the
two (spec-head conﬁguration). The DP that contains the bare nominal is accordingly
interpreted as indeﬁnite, which is exactly the result we are looking for since bare
nominals in Palestinian Arabic and Spanish are always interpreted as indeﬁnite.

To argue that BNS in the general case are devoid of (in)deﬁniteness features
necessary to check the strong features of the D position begs the obvious question
whether English BNS should be different in this respect from their PA and Spanish
counterparts. I have argued that in fact English BNS are not different but can be shown
to conform to the pattern seen of BNS in the other two languages. More speciﬁcally, I
have argued that existential BNS in English are also focused, and that these nominals are
different entities from BNS that are subject to a generic reading in this language. Such a
move should be welcomed on the principled ground that it minimizes cross-linguistic
variation. Furthermore, to say that existential BNS are different from generic ones in
English should not be as far-fetched as it may at ﬁrst seem since other languages, PA
and Spanish are only two examples, have two distinct mechanisms of expressing the
existential use (via BNS) and the generic one (by using the deﬁnite article). Another
difference that points to the accuracy of this assumption is that, as I have attempted to
demonstrate, unlike existential BNS in English, and all BNS in PA and Spanish, which

are focalized, generic BNS, and generic DPS with the deﬁnite article in PA and Spanish,

269

are never foci. If focused, generic BNS in English, and generic DPS with the deﬁnite
article in PA and Spanish, emerge as contrastively focused topics, not foci. This
reasonably illustrates that we are dealing with two different entities, not one, when
dealing with BNS in English. I have also argued that existential BNs in English would
be licensed by focus in a way identical to what I have argued for BNS in PA and Spanish
(although the level at which the movement into F P occurs could be different).

Once we look at the issues involved from this point of view, I have further argued,
the other differences between English, on the one side, and PA and Spanish, on the
other, in terms of the distribution of BNs fall out naturally from this analysis. I have
reasoned that since existential BNS in English and all BNS in both PA and Spanish are
focused, then the difference in their distribution in these two language groups should be
pursued by examining two other fundamental differences that exist between these
languages, namely, the differences in the system of focus and word order facts. English
is characterized by a restricted word order, but a ﬂexible focus system, therefore making
it possible to focus bare nominals, non-contrastively, that occur sentence-intemally, by
exploiting what Cinque (1993) terms marked focus. This also explains the absence of
Zubizarreta's (1998) prosodically-motivated movement from English. PA and Spanish,
by contrast, have rich word order possibilities and a more restrictive focus system than
that seen in English. Focusing a bare nominal internal to the sentence results in a
contrastive reading, unlike what happens in English. Marked focus, therefore, is
unavailable for either PA or Spanish. Prosodic movement, however, is an option
available for these languages, which should be expected Since only the C-NSR is

Operational in PA and Spanish (or, alternatively, stressing the most embedded element

270

in the sentence via the NSR as in Cinque (1993)). To be informationally focused, which
as I have argued is the case for all bare nominals in PA and Spanish, a bare nominal is
therefore conﬁned to such a position (i.e. lowest in c-command or the most deeply
embedded), otherwise it would only be contrastively focused. These deep contrasts
between English, on the one hand, and PA and Spanish, on the other hand, are
responsible for the difference in the distribution of BNS seen in these two language
groups.

Finally, I have argued that taking all the preceding assumptions in tandem we are
able to hypothesize as to why genericity is expressed differently in the two language
groups. I have argued that reference in these two language groups can perhaps be
explained by pursuing two related tacks. On the one hand, I have assumed that generic
operators differ in these two language groups in that English generic operators can bind
nominals in DPS with null heads, whereas PA and Spanish generic operators cannot. On
the other hand, I have suggested that if BNS in PA and Spanish are always focal these
nominals could not be mapped in the restrictive clause of a generic operator. Assuming
that focused BNS are mapped into the nuclear scope, whether in English or PA/Spanish,
would rule out the possibility of being mapped into the restrictive clause and thus be
bound by generic operators.

I believe that the picture that emerges from the analysis in this work is desirably a
restrictive one. Languages that seem to be noticeably different with respect to the syntax
and the semantics of bare nominals such as English and PA and Spanish turn out to have
more in common than meets the eye. The differences follow from deeper distinctions

that can be made at the level of stress assignment, focus and word order.

271

This in fact leads us to one of the interesting questions the study undertaken in this
work raises, which is obviously beyond the scope of this dissertation, is how such an
analysis as the one espoused here could be applied, if at all, to bare nominals in
languages such as Russian, and perhaps, Chinese. Russian allows bare nominals in
argument positions to occur ﬁeely. Obviously, the system of focus in this languages
needs to be examined before any conclusions can be reached. Also word order becomes
relevant as it has turned out to be with respect to English, PA and Spanish. Such an
inquiry into the properties of BNS in Russian, I believe, could ultimately enhance our

understanding of the issues discussed in the current work.

272

Bibliography

Abney, S. 1987. The English noun phrase in its sentential aspect. Doctoral dissertation,
MIT, Cambridge, Mass.

Aoun, J, Elabbas Benmamoun, and Dominique Sportiche, 1994. Agreement, Word
Order, and Conjunction in Some Varieties of Arabic. Linguistic Inquiry, Volume 25,
Number 2: 195-220.

Aoun, J, Elabbas Benmamoun, and Dominique Sportiche, 1999. Further Remarks on
First Conj unct Agreement. In Linguistic Inquiry, Volume 30, Number 4: 669-681.

Belletti, A. 2001. Inversion as Focalization. In Aaﬂ<e Hulk and Jean-Yves Pollock, eds.,
Subject inversion in Romance and the theory of universal grammar. Oxford; New
York: Oxford University Press.

Borer, H.1999. Deconstructing the construct. In K. Johnson and I. Roberts, Eds,
Beyond Principles and Parameters. Dordrecht: Kluwer publications.

Carlson, Greg. 1977a. A United Analysis of the English Bare Plural. Linguistics and
Philosophy 1: 413-457.

Carlson, G.: 1977b, Reference to Kinds in English, PhD dissertation, University of
Massachusetts, Amherst.

Carlson, G. and F. Pelletier (eds.). 1995, The Generic Book. The University of Chicago
Press, Chicago.

Casielles-Suarez, Eugenia. 1997. Topic, Focus, and Bare Nominals in Spanish.
Unpublished Ph.D. Dissertation. University of Massachusettes, Amherst.

Chierchia, G.l998. Reference to Kinds across Languages. Natural Language

273

Semantics 6, 339—405.

Chomsky, N. 1971. Deep structure, surface structure and semantic interpretation. In D.
Steinberg and L. Jakobovits, eds., Semantics: An interdisciplinary reader in
philosophy, linguistics and psychology. Cambridge: Cambridge University Press.

Chomsky, N. 1976. Conditions on rules of grammar. Linguistic Analysis 2: 303-351.

Chomsky, N. 1981. Lectures on Government and Binding. Dordrecht: F oris.

Chomsky, N. 1986. Knowledge of language. Its nature, origin, and use, New York,
Praeger.

Chomsky, Noam. 1995. The Minimalist Program. Cambridge, Mass: MIT Press.

Chomsky, N. and M. Halle.1968. The Sound Patterns of English. New York: Harper
and Row.

Cinque, Guglielmo.1993. A null theory of phrase and compound stress. Linguistic
Inquiry 24, 2: 239-298.

Cinque, G. 1996. The 'antisymmetric' programme: theoretical and typological
implications. Journal of Linguistics 32: 447—464.

Cohen, Ariel and Nomi Erteschik-Shir. 2002. Topic, Focus, and the Interpretation of
Bare Plurals. Natural Language Semantics 10: 125—165. Kluwer: the Netherlands.

Contreras, Heles. 1986. Spanish bare NPS and the ECP. In Ivonne Bordelois et al., eds.,
Generative Studies in Spanish Syntax 25-49. Dordrecht: Foris.

Cristina Schmitt and Alan Munn. 2000. Bare Nominals, Morphosyntax, and the
Nominal Mapping Parameter. To appear in Linguistic Variation Yearbook (20002).

274

Danon, Gabi. 2001. Syntactic deﬁniteness in the grammar of Modern Hebrew.
Linguistics 39, 6: 1071-1116.

De Swart, H. 1991. Adverbs of Quantiﬁcation: A Generalized Quantifiers Approach.
Ph.D. dissertation, Rijksuniversiteit GrOningen.

Diesing, M. 19923. Bare plural subjects and the derivation of logical representations.
Linguistic Inquiry 23, 353-397.

Diesing, M. 1992b. Indeﬁnites. Cambridge, Mass: MIT Press.

Dobrovie-Sorin, Carmen. 2000. (In)deﬁniteness spread: From Romanian genitives to
Hebrew construct state nominals. In Virginia Motapanyani, ed., Comparative Studies
in Romance Syntax. Elsevier, Oxford, pp. 177-226.

F assi F ehri, A. 1999. Arabic Modifying Adjectives and DP Structures. Studia
Linguistics 53, 2: 105-154.

Halliday, M.A.K. (1970). A Course in Spoken English: Intonation. Oxford University
Press, Oxford.

Heim, I. 1982, The Semantics of Deﬁnite and Indeﬁnite Noun Phrases, PhD
dissertation, University of Massachusetts, Amherst.

Higginbotharn, J. and G. Ramchand: 1997, ‘The Stage-Level/Individual-Level
Distinction and the Mapping Hypothesis’, in D. Willis, ed., Oxford University
Working Papers in Linguistics, Philology and Phonetics, vol. 2: 55—83.

Jackendoff, R. 1972. Semantic interpretation in generative grammar. Cambridge,
Mass.: MIT Press.

Jéiger, G.: 1999. Stage Levels, States, and the Semantics of the Copula. ZAS Papers in
Linguistics 14: 63-94.

275

Kayne, Richard 8.1994. The Antisymmetry of Syntax. The MIT Press, Cambridge,
Massachusetts.

Kriﬂ(a, M., F. Pelletier, G. Carlson, A. ter Meulen, G. Link, and G. Chierchia.l995.
Genericity: An Introduction. In G. Carlson and F. Pelletier, eds. The Generic Book".
1—124.

Kiss, Katalin E. 19983. ldentiﬁcational Focus Versus Information Focus. Language 74,
2: 245-273.

Koopman, H. and Dominique Sportiche. 1991. The position of subjects. Lingua 85:
211-258.

Kratzer, A.: 1995. Stage-Level and Individual-Level Predicates. In G. Carlson and F.
Pelletier (eds.), The Generic Book: 125—175.

Kuroda, Shige-Yuki. 1972. The categorical and the thetic judgment: Evidence from
Japanese. Foundations of Language 9:153-85.

Kuroda, Shige-Yuki. 1992. Judgment forms and sentence forms. In Japanese syntax and
semantics: Collected papers, 13-77. Dordrecht: Kluwer.

Laca, 8.1990. Generic Objects: Some More Pieces of the Puzzle. Lingua 81: 25—46.

Ladusaw, William. 1994. Thetic and categorical, stage, and individual, weak and strong.
In Proceedings of SALT I V, eds. Mandy Harvey and Lynn Santelmann: 220-229.
Cornell U. DMLL, Ithaca, NY.

Larson, R. 1998. Events and Modiﬁcation in Nominals. In D. Strolovitch and A.
Lawson, eds., Proceedings ﬁom Semantics and Linguistic Theory VIII. Ithaca: Cornell
University.

Link, G.l995. Generic Information and Dependent Generics. In G. Carlson and F.
Pelletier, eds., The Generic Book: 358—3 82

276

Longobardi, G. 1994. Reference and Proper Names: A Theory of N-Movement in
Syntax and Logical Form. Linguistic Inquiry 25. 4: 609-665.

Longobardi, G. 1996. The syntax of N-Raising: a minimalist theory. Ms. U. di Venzia.

Longobardi, G. 2000. Postverbal subjects and the mapping hypothesis. Linguistic
Inquiry 691-702.

Longobardi, G. 2002. How Comparative is Semantics? A Uniﬁed Parametric Theory of
Bare Nouns and Proper Names. Natural Language Semantics 9: 335-369. Kluwer: the
Netherlands.

McNally, L. 1998. Stativity and Theticity. In S. Rothstein (ed.), Events and Grammar,
293—307. Kluwer Academic Publishers, Dordrecht.

Milsark, G.L. 1974. Milsark, Gary. 1974. Existential sentences in English. PhD
dissertation, MIT.

Munn, Alan, 1999. First conjunct agreement: Against a clausal analysis. Linguistic
Inquiry 30:643-668.

Munn, Alan and Cristina Schmitt. 1999. Bare nouns and the morpho-syntax of number.
Proceedings of the Linguistic Symposium on Romance Languages 1 999.

OrdOi‘lez, Francisco. 1998. Post-Verbal Asymmetries in Spanish. Natural Language and
Linguistic Theory 16, 2:313-346. Kluwer: the Netherlands.

Partee, B. H. 1991. Topic, Focus and Quantiﬁcation. In Proceedings of SALT I : 257—
280.Comell University, Ithaca, NY.

Pinto, Manuela. 1997. Licensing and interpretation of inverted subjects in Italian .
Ph.D. dissertation, Universiteit Utrecht.

Reinhart, T. 1981. Pragmatics and Linguistics: An Analysis of Sentence Topics.

277

Philosophica 27, 53—94.

Reinhart, T. 1995. Interface Strategies. OTS Working Papers TL-95-002,
Utrecht University.

Ritter, Elizabeth. 1988. A head-movement approach to construct-state noun phrase.
Linguistics 26: 909-929.

Ritter, E. 1991. Two functional categories in noun phrases: evidence from modern
Hebrew. In S. Rothstein, ed., Syntax and Semantics 25: 37-62. New York: Academic
Press.

Rizzi, Ligui. 1990. Relativized Minimality. MIT Press, Cambridge, Mass.

Rizzi, Ligui. 1997. The Fine Structure of the Left Periphery. In L. Haegeman, ed.
Elements of Grammar: 281-337. Kluwer: the Netherlands.

Roberts, Ian and Ur Shlonsky.1996. Pronominal enclisis in VSO languages. In The
Syntax of the Celtic Languages, ed. Robert Borsley and Ian Roberts: 171-199.
Cambridge: Cambridge University Press.

Rooth, Mats. 1992. A Theory of Focus Interpretation. Natural Language Semantics 1:
75-1 16. Kluwer: the Netherlands.

Shlonsky, Ur.1997. Clause Structure and Word Order in Hebrew and Arabic. Oxford
University Press, New York and Oxford..

Siloni, T31. 1994. Noun phrases and nominalization. Ph.D. dissertation. U. de Genéve.

Siloni, T31. 1996. Hebrew Noun Phrases: Generalized Noun Raising. In Adriana Belletti
and L. Rizzi, eds., Parameters and Functional Heads. Oxford University press.

Siloni, T31. 1997. Noun Phrases and Nominalization: The Syntax of DPS. Kluwer:
Dordrecht.

278

Suffer, M. 1982. Syntax and Semantics of Spanish Presentational Sentence- Types.
Georgetown University Press, Washington, DC.

Suffer, M. 1 994. V-Movement and the Licensing of the Argumental Wh-phrases in
Spanish. Natural Language and Linguistic Theory 12: 335-372.

Torrego, Esther. 1984. On Inversion in Spanish and Some of its Effects. Linguistic
Inquiry 15: 103-129. MIT Press.

Vallduvi, E. 1988. Functional Load, prosody and syntax: Left-detachment in Catalan
and Spanish. CLS 24: 391-404

Zamparelli, Roberto. 2002. Deﬁnite and Bare Kind-Denoting Noun Phrases. To appear
in Proceedings of Going Romance 2000, Oxford University Press, 2002.

Zubizarreta, ML. 1998. Prosody, Focus, and Word Order. Cambridge, Mass: MIT
Press.

279

llllllllllllllll