You are on page 1of 51

Action Nominalizations and the Lexicalist Hypothesis

Author(s): Ilan Hazout


Source: Natural Language & Linguistic Theory , Aug., 1995, Vol. 13, No. 3, Special
Hebrew Issue (Aug., 1995), pp. 355-404
Published by: Springer

Stable URL: https://www.jstor.org/stable/4047840

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Springer is collaborating with JSTOR to digitize, preserve and extend access to Natural
Language & Linguistic Theory

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ILAN HAZOUT

ACTION NOMINALIZATIONS AND THE


LEXICALIST HYPOTHESIS*

This paper provides an account of the mixed verbo-nominal p


nominalization construction in Hebrew and Arabic. Such an a
postulating an underlying representation in which a VP is th
abstract nominal head, a nominalizer, and a derivation in whi
to the nominalizing morpheme to derive the action nominal tha
construction. A central assumption of this analysis concerns t
the argument structure, of the nominalizer and the way these i
thematic structure of action nominalizations. Finally, the articl
of lexicalism and develops arguments in favor of the non-lexica
nominalizations developed in this work.

1. INTRODUCTION

The Action Nominalization construction is one of a class of verbo-


constructions occurring in Modem Hebrew and Standard Arabic. The
verbo-nominal nature of such constructions consists of a mixture of proper-
ties that are typical of verbs and VPs as well as properties that are typical
of nouns and NPs. The following are representative examples (H and A
stand for Hebrew and Arabic, respectively, throughout this paper).

(1)a. (H) axilat Dan et ha-tapuax


eating Dan ACC the apple

Dan's eating the apple

b. (A) taSriib al-?awlaad al-Haliib


making-drink the kids the milk

making the kids drink the milk

In order to understand why we view these constructions as verbo-


nominal, let us first note that both examples in (1) are NPs as far as their
sentence-internal distribution is concerned. That is, they relate to their
syntactic environments like any other NP. However in their internal pro-
perties, these constructions are typically headed by a noun that is morpho-
logically related to a certain verb. Moreover, just like the verb, the head

* I am greatly indebted to Edwin Williams and Roger Higgins for discussions of and
comments on the material presented in this paper. The comments of an NLLT reviewer led
to improvements in the organization of this material. H. Borer's editorial comments were
helpful in preparing the final version.

Natural Language and Linguistic Theory 13: 355-404, 1995.


? 1995 Kluwer Academic Publishers. Printed in the Netherlands.
This content downloaded from
81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
356 ILAN HAZOUT

of an action nominalization may be accompanied by a direct object marked


with accusative Case, as in (la), as well as by other constituents that
would normally be taken to be the arguments of the corresponding verb.
We may say that the noun heading action nominalization constructions
inherit the argument structure of a verb, and they also allow for the
occurrence of adverbs, a typical verbal property.
As to the internal nominal properties of action nominalizations, it
suffices to note at this point that the noun heading this construction is
always accompanied by an NP, an argument of the related verb, in a
genitive type of relation. This clearly shows on the Case marking of the
genitive NP or its head. The head of an action nominalization may also
be modified by an adjective as is demonstrated in the discussion below.
This, obviously, is a typical nominal property.
This mixture of properties is problematic to the extent that the
verb/noun distinction is based on the traditional observation that a certain
cluster of properties is associated, in the general case, with a certain class
of elements to the exclusion of others. In view of the mixed properties of
action nominalizations (and other verbo-nominal constructions, cf. Hazout
1990, to appear), it is, at the least, unclear to what extent our traditional
view of the basic inventory of lexical elements in the languages under
consideration, as well as generally, can be maintained or is in need of
revision.
Thus, the main question facing us when dealing with the action nominal-
ization construction is how to explain, or derive, the demonstrated mixed
nature of this construction. Here two main approaches come to mind. The
first would be to admit that, besides verbs and nouns, the languages in
question have also a class of lexical elements that are of a mixed nature,
that is, of elements belonging to a category which shares properties with
both verbs and nouns. The constituents of which they may be said to be
the heads share some of the syntactic properties of both NPs and VPs. A
second approach would deny such a possibility and maintain that the basic
underlying inventory of lexical elements includes only 'strict' nouns and
verbs. Elements and constituents of mixed, or verbo-nominal, nature
would be derived in such an approach rather than basic. That is, they
would be only the output of certain processes to which verbs and nouns
may serve as input. Thus, in this second approach the properties of verbal
nouns are derived from the properties of verbs and nouns.
This paper presents an analysis of action nominalizations along the lines
of the second approach. As is shown in section 3, a satisfactory analysis
of this construction is possible if one assumes an underlying structure
involving a nominal head which selects a VP. The derivation of the surface

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 357

structure would involve, as one of its derivational steps, head movement


of the verb, adjoining it to the nominal head that governs it. The output
of this operation is the derived nominal, which is the head of the construc-
tion. In this, the approach advocated here may be considered to be
strongly non-lexicalist since, unlike the first approach, it allows for a
morphologically complex word to be the output of a syntactic operation.
A large portion of this paper will be devoted to an attempt to develop
arguments in favor of such a non-lexicalist approach.
The paper is organized as follows. Section 2 is devoted to a detailed
presentation and consideration of the data. In section 3 we present the
details of the proposed analysis of the action nominalization construction
in its different variants. Section 4 is devoted to a discussion of the issue
of lexicalism as it relates to this construction.

2. THE DATA

In this section we present the main facts about action nominalizations,


with which this paper is concerned. These include, mainly, the mixture of
verbal and nominal properties of this construction that we describe in
section 2.1. In their general outlines, these facts are strongly reminiscent
of facts which have been often observed in similar constructions in other
languages, such as, notably, the gerund in English. However, as will be
pointed out (See also section 3.4), there are certain particularities of the
constructions studied in this paper that render them particularly complex.
Some of these issues are related to the distribution of adverbs and adjec-
tives in these constructions and are described in section 2.2. Others are
related to the genitive construal of elements within action nominalizations.
As is demonstrated by the examples below, Hebrew and Arabic action
nominalizations come in two main variants, those which have an overt
logical subject and those which lack one. This alternation also raises
difficult problems for which this paper attempts to provide an account an
to which there is no parallel in the English gerund. Previously available
analyses which have been developed mainly in work on corresponding
English constructions are therefore insufficient and certain new ideas will
have to be developed.

2.1. Verbo-Nominal Properties of Action Nominalizations

Action nominalization constructions have the regular distribution of NPs.


One finds action nominalizations as subjects, objects and objects of prep-
ositions as shown in examples (2), (3) and (4), respectively.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
358 ILAN HAZOUT

(2) (H) harisat ha-cava et ha-ir hayta axzarit


destruction the army ACC the city was cruel

The army's destruction of the city was cruel.

(3) (H) ra?ayt-u iRtiyaal -a l-waziir -i


saw -I assassination-ACC the-minister-GEN

I saw the assassination of the minister.

(4) (A) fuji?t -u bi-Rtiyaal -i zaid-en


surprised-I with-assassination-Dat zaid-GEN the

al-waziir -a
the minister-ACC

I was surprised by Zaid's assassination of the minister.

As can be seen in the Arabic examples in (3) and (4), the head of an
action nominalization construction is a typical noun in that it shows Case
marking, depending on the position it occupies in the sentence. Thus, the
noun iRtiyaal 'assassination' is marked for Accusative and Dative in (3)
and (4), respectively.'
In their internal syntax, action nominalizations are nominal in that they
manifest the assignment of genitive Case to an NP (logical Subject or
Object). Examples (2) and (3) both have an NP standing in the bound
genitive (construct state) construction to the head.
In modern Hebrew the head and the subject may also appear in the
free and double genitive constructions as follows.

(5)a. (H) Free Genitive: ha- axila Sel Dan et ha-tapuax


the eating Sel Dan ACC the apple

Dan's eating the apple

b. (H) Double Genitive: axilat-o Sel ha-tapuax


eating-his Sel the apple

The eating of the apple

Free and double genitives allow for the occurrence of an adjective, modi-

1 The examples in (2) and (3) are from Standard Arabic. They are given here for the sak
of illustrating morphological Case marking, which is mostly absent in the spoken dialects.
Otherwise, our discussion is limited to the dialect known as Syrian Arabic, spoken in Syria,
Lebanon and Palestine.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 359

fying the head noun; The adjective intervenes between the head and its
subject.2

(6)a. (H) ha-harisa ha-maftia Sel ha-cava et ha-ir

the destruction the surprising Sel the army ACC the city

The surprising destruction of the city by the army

b. (H) harisat -a ha-maftia Sel ha-ir


destruction-her the surprising Sel the city

The surprising destruction of the city

Proceeding to the verbal properties of action nominalizations, most


important among these is the marking of accusative Case on a direct object
as shown in examples (2) and (4) above.3 In general, a nominalized head

2 The impossibility of adjectives in the bound genitive variant is due to a general prope
of bound genitives rather than to a property of action nominalizations. (i) is a simple bound
genitive construction.
(i) bayit (*gadol) ha-mora
house big the teacher

Likewise, the admissibility of intervening adjectives in free and double genitiv


property which has nothing to do with the particularities of action nominalizations. This is
shown in (ii) and (iii).

(ii) ha- bayit ha-gadol Sel ha-mora


the house the big Sel the teacher

the teacher's big house

(iii) beyt -a ha-gadol Sel ha-mora


house-her the big Sel the teacher

The teacher's big house

Obviously, a major task of this paper is to integrate a generally motivated analysis of Semi
genitive constructions within an account of action nominalizations. For work on the differen
variants of genitive constructionss, see Borer (1984), Fassi-Fehri (1988), Hazout (1990),
Ritter (1986, 1990), and Siloni (1990), among others.
3 There is some dispute in recent work concerning the accusative marker et and the nature
of accusative Case assignment in action nominalizations (cf. Borer 1984, Siloni 1990) Analy-
ses of this construction that, unlike the one defended in this paper, do not attribute its verbal
properties to the occurrence of a verb (and a VP) in its underlying structure (cf. Ritter 1986,
1990; Siloni 1990) must account for the assignment of accusative Case in this construction
in some other way. Here there are two main possibilities. One is to claim that the accusative
Case is assigned by the particle et itself. The other is to claim that the accusative Case assigned
in action nominalizations is of a different kind from that assigned in regular sentences. In
our view, both options are wrong. The first is wrong since, generally, NPs marked by et
have all the properties of direct objects and, therefore, must be viewed as governed by a
verb rather than by something else. Besides, the corresponding Arabic construction has
nothing which would be the equivalent of et. Rather, Arabic action nominalizations show

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
360 ILAN HAZOUT

manifests the same argument structure as its corresponding verb. Thus,


an arbitrary requirement by a verb for a certain type of semantically empty
preposition is observed also by the corresponding nominal form.

(7)a. (H) Dan histalek min ha-misra


Dan left from the position

Dan resigned the position.

b. (H) histalkut-o Sel Dan min ha-misra


leaving -his Sel Dan from the position

Dan's resignation from the position

Likewise, action nominalizations with nominal heads that are related to


double-object verbs take the same number of arguments and show the
same dative alternation characteristic of the corresponding verbs.

(8)a. (A) Zaid ?a9ta 1-walad attufaaHa


Zaid gave the boy the apple

Zaid gave the boy the apple.

b. (A) ?i9ta Zaid al-walad attufaaHa


giving Zaid the boy the apple

Zaid's giving the boy the apple

(9)a. (A) Zaid ?a9ta ttufaaHa li-l- walad


Zaid gave the apple to the boy

Zaid gave the apple to the boy.

b. (A) ?i9ta Zaid attufaaHa 11-1- walad


giving Zaid the apple to the boy

Zaid's giving of the apple to the boy

Finally, action nominalization constructions in both Hebrew and Arabic


allow for the occurrence of adverbs.

the same morphological accusative Case as in simple sentences.


The second option raises the general issue of the parallelism betwen action nominalizations
and corresponding sentential structures. The point is that if what appears to be accusative
Case is something different in the two construction types then the fact that only verbs which
assign ordinary accusative Case have nominal counterparts which assign special accusative
Case becomes entirely accidental. Indeed, it is the explanation of this very parallelism which
is the task of any analysis of these constructions.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 361

(10) (A) ?akl Zaid ttufaaHa bi-sur9a


eating Zaid the apple quickly

Zaid's eating the apple quickly

(11) (A) harisat ha-cava et ha-kfar be-axzariyut


destruction the army ACC the village cruelly

The army's destroying the village cruelly

It seems that the occurrence of adverbs in these constructions is not


licensed simply by the action or process interpretation associated with
these constructions, but rather is syntactically conditioned. This view is
supported by the fact that, although the nouns in question may appear all
by themselves, as in (12), the occurrence of an adverb is not allowed
unless it is accompanied by the rest of the complement system required
by the corresponding verb, as in (13).

(12)a. (H) ha- harisa nimSexa Saatayim


the destruction lasted two hours

The destruction took two hours.

b. (A) kaana 1- iRtiyaal waHSiyy-an


was the assassination barbaric-ACC

The assassination was barbaric.

(13)a. (H) *ha- harisa be-axzariyut


the destruction cruelly

b. (A) *al- iRtiyaal bi-sur9a


the assassination quickly

Besides the general fact that adverbs are an indication of the verbal
nature of the syntactic environment in which they occur, there are certain
facts about the distribution of adverbs (and adjectives) in action nominaliz-
ations, an account of which necessitates certain assumptions concerning
the internal structure of this construction. We consider these facts in the
next section.

2.2. The Distribution of Adverbs in Action Nominalizations

In this section we discuss two points concerning the admissibility and


distribution of adverbs. We consider first the kinds of adverbs which are

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
362 ILAN HAZOUT

admitted. Thus, while manner and time adverbials may freely occur in
these constructions, sentential adverbs are strictly excluded. (14) below
shows sentences with sentential adverbs at different positions in the sen-
tence. In (15) the action nominalizations corresponding to these sentences
are bad.

(14)a. (H) lelo safek Dan katav et ha-avoda


doubtlessly Dan wrote ACC the work

Dan doubtlessly wrote the work.

b. (H) Dan karov le-vadai axal et ha-tapuax


Dan probably ate ACC the apple

Dan probably ate the apple.

c. (H) ha- oyev behexlet haras et ha-ir


the enemy definitely destroyed ACC the city

The enemy definitely destroyed the city.

(15)a. (H) *ktivat Dan et ha-avoda lelo safek


writing Dan ACC the work doubtlessly

b. (H) *axilat Dan et ha-tapuax karov le-vadai


eating Dan ACC the apple probably

c. (H) *harisat ha-oyev et ha-ir behexlet


destruction the enemy ACC the city definitely

(16) and (17) below are the corresponding paradigms for manner and
time adverbials. These types of adverbs are perfectly admissible as demon-
strated in (17).

(16)a. (H) Dan katav et ha-avoda bi-mehirut


Dan wrote ACC the work quickly

b. (H) Dan axal et ha-uga be-nimus


Dan ate ACC the cake politely

c. (H) ha- oyev haras et ha-ir emeS


the enemy destroyed ACC the city last night

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 363

(17)a. (H) ktivat Dan et ha-avoda bi-mehirut


writing Dan ACC the work quickly

Dan's writing of the work quickly

b. (H) axilat Dan et ha-uga be-nimus


eating Dan ACC the cake politely

Dan's eating of the cake politely

c. (H) harisat ha-oyev et ha-ir emeS


destruction the enemy ACC the city last night

The enemy's destruction of the city last night

Given the general assumption that manner and time adverbials are at-
tached to the tree structure at the VP level (or are at least internal to
VP), whereas sentential adverbs are attached higher on the tree structure,
the ungrammaticality of the examples in (15) may be taken to indicate
that an underlying verbal constituent, to be postulated, goes up to the VP
level and no further. This implies that the required configuration is avail-
able at an underlying level of representation in order for such a distinction
as the one between the VP and the IP levels to be possible.
The claim that the distribution of adverbs in Hebrew action nominaliza-
tions is syntactically, rather than semantically, determined is further sup-
ported by the following facts. As is mentioned above, adjectives can follow
the head noun in free and double genitives.

(18)a. (H) ha- axila ha-menumeset Sel Dan et ha-uga


the eating the polite Sel Dan ACC the cake

Dan's polite eating of the cake

b. (H) axilat-o ha-menumeset Sel Dan et ha-uga


eating-his the polite Sel Dan ACC the cake

Dan's polite eating of the cake.

Note, however, that in contrast to adjectives, adverbs cannot follow the


head noun in free and double genitives. Compare the following paradigm
with (17) above.

(19)a. (H) *ha- ktiva bi-mehirut Sel Dan et ha-avoda


the writing quickly Sel Dan ACC the work

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
364 ILAN HAZOUT

(19)b. (H) *axilat -o be-nimus Sel Dan et ha-uga


eating-his politely Sel Dan ACC the cake

c. (H) *harisat -o emeS Sel ha-oyev et ha-ir


- destruction-his last night Sel the enemy ACC the city

To complete the picture, note that adjectives, as opposed to adverbs,


are bad in final position. Compare also the following paradigm with that
in (17).

(20)a. (H) *ktivat Dan et ha-avoda ha-mehira


writing Dan ACC the work the quick

b. (H) *axilat Dan et ha-uga ha-menumeset


eating Dan ACC the cake the polite

c. (H) *harisat ha-oyev et ha-ir ha-axzarit


destruction the enemy ACC the city the cruel

Thus, adverbs and adjectives are in complementary distribution with


respect to both final position and the position immediately following the
head. This is also demonstrated by examples in which both an adjective
and an adverb occur, as in the following. Switching the order between
adjective and adverb in (21b) below is totally impossible.

(21)a. (H) harisate -nu ha-axzarit et ha-ir im


destruction-our the cruel ACC the city with

alot ha-Saxar

break the dawn

our cruel destruction of the city with the break of dawn

b. (H) *harisate -nu im alot ha-Saxar et


destruction-our with break the dawn ACC

ha- ir ha-axzarit
the city the cruel

The facts just observed indicate that the configuration underlying these
constructions involves a distinction between a verbal domain, which allows
the occurrence of adverbs and excludes adjectives, and, conversely, a
nominal domain, which allows the occurrence of adjectives and excludes
adverbs.
Clearly, the admissibility of adverbs in action nominalizations is not a

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 365

fact that has simply to do with some kind of semantic compatibility.


Rather, it is a syntactic fact that has to do with the principles which
determine which elements can be attached at which positions in the tree
structure. This is not to be interpreted as a claim that the admissibility of
adverbs has nothing to do with the event interpretation associated with
action nominalizations. Indeed, some such correlation, or even depen-
dency, seems very likely. But it is clear that the interpretation of action
nominalizations as denoting an event is not sufficient as an explanation of
the severe restrictions on the distribution of adverbs within the construc-
tion.4
Thus, all of the facts considered above point clearly to the necessity of
assuming an underlying configuration of which a substructure is a verbal
projection of a certain level, namely, the VP level.

3. THE STRUCTURE OF ACTION NOMINALIZATIONS5

3.1. Bound Genitive Variants

In this section, we are concerned first with the core cases of action nomina-
lizations, that is, cases in which the head noun and the subject form a
bound genitive as in (22) below. These cases are common to both Hebrew
and Arabic. Free and double genitives which are special to Hebrew will
be examined later.

(22)a. (H) axilat ha-yeled et ha-tapuax bi-mehirut


eating the boy ACC the apple quickly

The boy's eating the apple quickly

4 What we have in mind here is mainly the recent influential work of Grimshaw (1990)
derived nominals in English. Grimshaw makes the distinction between two types of nominal
complex event nominals, on the one hand, and simple event and result nominals, on the
other. She claims that the first type has an argument structure which must be syntactically
realized while the second type does not. Hebrew and Arabic action nominalizations are,
however, very different from Grimshaw's complex event nominals and are much more similar
in their properties to English gerunds, in particular the variant known as POSS-ING. (Our
use of the term 'nominalization' (or 'nominal') for the Semitic construction is due partly to
the fact that in the already established terminology in work on Hebrew grammar, the term
gerund is used for other constructions (see Berman 1978; Hazout 1992).) Thus, unlike English
event nominals and like the English POSS-ING gerund, Semitic action nominalizations allow
for the assignment of accusative case to an object and for the occurrence of adverbs. It is
our view, therefore, that Grimshaw's important work on event and result nominals is of
little relevance for the constructions studied in this paper.
S The analysis proposed in this section makes use of a variant of theta theory due mainly
to the work of E. Williams. The reader is referred to Williams (1987a), (1987b) and (1989).

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
366 ILAN HAZOUT

(22)b. (A) akl attufaaHa bi-sur9a


eating the apple quickly

The eating of the apple quickly

We first present the structural derivation assumed here and then go into
further detail. Beginning with the cases involving an overt subject, the
underlying configuration assumed for (22a) is the following.

(23) DP

D NPi

I /
POSS NP2 N'

the boy N VP

NOM V NP3 Adv

eat the apple quickly

In (23), a nominal bound morpheme, NOM, figures as the head of an


action nominalization. NOM subcategorizes for a VP headed by a verb of
which any lexical requirements must be satisfied within the domain of its
government. The NP which is construed as the logical subject of the
underlying verb is base-generated in the specifier position of NP. The
actual deverbal form of the head noun is derived by head movement of
the underlying verb to NOM. Since NOM is a bound morpheme, this
operation must apply. The result is the configuration in (24).

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 367

(24) DP

D NPi

POSS NP2 N'

No VP

V N V NP3 Adv

NOM e

No in the above configuration is the derived head o


The trace left by the verb inherits the properties of the verb, in particular,
its Case-assigning features. The next step in the derivation is the move-
ment operation which is part of the general process by which bound
genitives are formed. The result of that process is shown in (25) below.7
The source of genitive Case assigned to the subject in (25) is the genitive
morpheme POSS, the head of DP.

6 The transformational operation known as 'head movement' is by now a generally ass


mechanism within the government and binding framework. The development of this mech
ism is due mainly to the work of Travis (1984) and, most prominently, Baker (1985, 1988
7 The analysis of bound genitives assumed here is based mainly on the work of Ritter (19
which makes use of the DP hypothesis of NP structure. Variants of Ritter's analysis are
developed in Fassi-Fehri (1988), Siloni (1990) and Hazout (1990).

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
368 ILAN HAZOUT

(25) DP

D NPi

N POSS NP2 N'

eat NOM the boy N VP

e V NP3Adv

the apple

The two main verbal characteristics of action nominalizations are di-


rectly accounted for by this derivation. First, accusative Case assignment
is a manifestation of the government of the direct object by the verb (or
by its trace). Second, adverbs are licensed in the configurations (23)-(25)
as subconstituents of VP. Given the above configurations, it is only VP
adverbs, and not sentential adverbs, that may occur.
A note on the morphology is in order at this point. As is well known,
verbs in both Hebrew and Arabic appear in a certain number of different
morphological patterns known as Binyanim. Generally, each pattern has
a unique nominalized form. Otherwise, if more than one nominal form is
associated with a verbal pattern, each single verb in that particular class
is associated with a unique nominal form. A potential difficulty for the
present analysis is that it raises the possibility that one may have to
postulate a different NOM morpheme depending on the verb form ocur-
ring as head of VP. This raises another difficulty, namely, of accounting
for the dependency between the form of the verb and the particular
morphological form of NOM, given that both are inserted independently.
However, no such difficulty arises if one assumes NOM to be an abstract
form rather than an item having an actual morphological shape. Being an
abstract form, NOM may be thought of as a function which maps a verb
onto the unique nominal form which is lexically listed for that particular
verb. In (26a) and (26b) are examples of some Binyan patterns in Hebrew
and Arabic with the corresponding verbal and nominal forms.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 369

(26)a. Hebrew Binyan pattern verb form noun form


i. CaCaC SaVaR SViRa
break breaking
ii. CiCeC DiBeR DiBuR
speak speaking

b. Arabic
i. (Form I) CaCaCa NaQaLa NaQL
transport transporting
ii. (Form II) CaCCaCa SaRRaBa taSRiiB
make drink making drink

The derivation of the subjectless variant in (22b) includes one additional


step by which the object NP moves to the specifier position. The underly-
ing structure of (22b) would, thus, be the following.

(27) DP

D NPi

N'

N VP

NOM V NP2 Adv

eat the apple quickly

For reasons discussed below, NP2 in the configuration above cannot be


assigned accusative Case by the governing verb and, therefore, must move
to Spec-of-NP as follows in (28).

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
370 ILAN HAZOUT

(28) DP

NPi

NP2 N'

N VP

NOM V e Adv

The rest of the derivation of the subjectless variant involves head move-
ment of V to N and N to D and is identical to that of the active variant
as shown in (23)-(25) above. There is, therefore, no need to repeat it
here.
The role played by NOM is central and many of the properties of the
construction are explained by making certain assumptions about its nature.
As a bound morpheme taking a VP complement, NOM is a morphological
nominalizer in that it combines with a verb to produce a nominal head.
This aspect of our analysis is familiar from studies of corresponding ele-
ments (e.g., the English -ING) which occur in similar constructions in
other languages. Such assumptions have been made in much of the rel-
evant literature (See section 3.4). Here we would like to suggest that
NOM is a nominalizer in an additional sense, namely, in its thematic
function. Making use of a particular variant of theta theory developed by
Williams, we can give this intuition a concrete expression by thinking of
NOM as an element with an argument structure. First, as a noun, NOM
has an external R(eferential) theta role. The external theta role of NOM
is assigned up ("vertically") to become the external theta role of the entire
NP. For reasons that will become clear immediately, we assume that NOM
has an additional R-role which is necessarily internal. Thus, the lexical
representation of NOM contains the following information.8

(29)a. Argument structure of NOM: (Ri, Rj)


b. Subcategorization frame of NOM: [ VP]

8 Nominalizing morphemes of this general type can vary according to the numb
ments they may have and the indices associated with dtwese arguments. Hazout (1990, to
appear) develops a typology of nominalizers along these lines and studies the way in which
various properties of nominalization constructions can be explained on the basis of the
thematic properties of their head.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 371

The Theta Structure of an action nominalization construction looks,


partially, as follows.

(30) ,

Ni AVP

NOM V NP,
(R,R j) (LQj, BO)

In this structure the external theta role of the verb is first assigned vert-
ically to VP to become its external theta role. It is then assigned to NOM
or, more precisely, to the internal R-role of NOM. As the sister of the
VP subcategorized by it, NOM is always in a position in which it may be
assigned the external theta role of VP. Moreover, just as it is the case
with any instance of theta role assignment, here too the relation is not
symmetric. Rather, it is an asymmetric relation of linking in the sense of
Williams (1989). It is the external theta role of VP that is dependant on
Rj for its referential content. Therefore the linking of both theta roles is as
illustrated above. As far as the formal operation of theta role assignment is
concerned, the external theta role of VP is, at this stage, satisfied. More-
over, the operation in (30) must apply, since the external theta-role of
VP must be assigned. Most importantly, this role couldn't be vertically
assigned to N' and become its external theta role, since N' already has
an external theta role, namely, the external theta role of NOM. Therefore,
the assignment of theta-roles as indicated in (30) is the only available
option. Since the external theta role of the verb in (30) is satisfied, the
application of any further theta-theoretic operations is not necessary.9
We can now provide the following account for the lack of accusative
Case in the subjectless variant. Our explanation of this fact makes use of
the idea that accusative Case assignment may only be licensed in the
environment of certain functional elements (e.g., INFL, NOM, etc.). Fur-
thermore, we make use of the characterization of such elements as [+No-
minall depending on the way they interact thematically with the rest of
the construction. For example, if an element such as NOM absorbs the

9 In the variant of theta theory assumed here (cf. Williams (1987a, 1989) the assignment
an internal theta role is not obligatory. Therefore the internal R-role of NOM may remain
unexpressed.

This content downloaded from


fff:ffff:ffff:ffff:ffff:ffff:ffff on Thu, 01 Jan 1976 12:34:56 UTC
All use subject to https://about.jstor.org/terms
372 ILAN HAZOUT

theta role of a predicate (VP in this case), we may say that it functions
as an argument, that is, it serves as the substitute for an overt subject. In
such a state of affairs, it qualifies as [+Nominal]. [+Nominal] elements
do not license accusative Case assignment by the verb in their domain. In
the construction under consideration here, such a state of affairs arises in
the subjectless variant. In the active variant (See below), the subject theta
role is transmitted further to the subject NP and, therefore, NOM is not
nominal. The same generalization holds for the construction known as
Agent Nominalization studied in Hazout (1990, to appear).
We see then that the function of the internal R-role of NOM is crucial
to the syntax of this construction. If things stay as they are in (30), the
result is an action nominalization lacking an overt subject, with the internal
R-role of NOM receiving an arbitrary or generic interpretation. Crucially,
the analysis in (30) incorporates the claim that the subjectless variant of
action nominalizations is not an instance of passivization. This view is
argued in detail in section 3.3.
We consider next the variant exemplified by (22a) in which an overt
subject occurs. Its thematic structure is illustrated below.

(31) NPii

NP2j N

Ni VP.

NOM V NP3k
(Ri, R j) (Aj, BO)

Here the thematic integration of NP2 is achieved through a generally


motivated thematic relation holding between a genitive NP and a head
noun. As it happens in any instance of genitive construction, a genitive
NP or, more precisely, its external R-role is coindexed with the internal
thematic index of the head of the NP in which it is embedded. In this
particular case, the genitive NP is coindexed with the internal argument
of NOM. Now, the internal argument of NOM is coindexed with (assigned)
the external theta role of VP. It follows that NP2, or the external R-role
of NP2 and VP share the same index. That is, the reference of NP2 and
the reference of the external theta role of VP is identical. Thus, within
the assumed theta-theoretic approach, theta role assignment amounts to

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 373

the coindexation of thematic indices which takes place under certain con-
figurational conditions.10
We can say, therefore, that the subject NP is assigned a theta-role
through the mediation of NOM. That is, the subject theta role which was
assigned first to NOM may be said to have been passed on (although not
externally or vertically) rather than to have been absorbed by NOM. Note
that in a case where an NP is present in specifier position the operation

10 In the variant of theta theory adopted in this paper, nouns have an external theta
designated as R (for 'reference'). In Hazout (1990), we make the additional assumption that
nouns in general also have an internal thematic index that is unspecified for thematic content.
The argument structure of a regular noun, thus, looks as follows.

(i) (Ri, j)
The thematic structure of a simple genitive, such as that in (iia), would be as in (iib).

(ii)a. (H) mazkir ha-miflaga


secretary the party

the party's secretary

b. NP.

the parly NI

secretary

(Ri,j)

The genitive NP 'the party' is coindexed with the internal argument of the head 'secretary'.
The relation between the two entities referred to by the two members of the construction
is understood to be thematically free, thus, the absence of a thematic lable for the internal
argument of the head noun. A failure of this thematic coindexation to apply will result in a
construal of 'the party' as a modifier (in the manner of an adjective), that is, as a non-
referential NP. The expression as a whole would then denote a secretary of a certain type.
In this way we account for the ambiguity of interpretation generally associated with simple
genitives. An indexation of the genitive NP with the external R-role of the head is impossible,
since it would result in a coindexation of this NP with the higher NP, creating a situation in
which an NP is construed as coreferential with the NP in which it is contained (cf. the
NPi/NP, constraint of Williams 1982). We may say, therefore, that the coindexation of the
genitive NP with the internal thematic index of the head noun applies by default rather than
by virtue of a specific rule. The same theta-theoretic mechanism applies if the genitive NP
is in a Sel-phrase. Thus, (iiia), the free genitive counterpart of (iia), has the thematic structure
in (iiib).

(iii)a. (H) ha- mazkir Sel ha-miflaga


the secretary Sel the party
the party's secretary

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
374 ILAN HAZOUT

(coindexation) illustrated in (31) must take place, since this is the only
way for this NP to be thematically licensed.
Binding relationships between NP2 and NP3, coarguments of the under-
lying verb in the configuration in (31), are directly accounted for. NP2,
being assigned the external theta role of VP, stands to NP3, an internal
argument, in the same asymmetric relationship that would hold between
a subject and an object in a sentence."1 The contrast in (32) is, thus,
explained.

(32) a. (H) reiyat Dan et acmo


seeing Dan ACC himself

Dan's seeing himself

b. (H) *reiyat acmo et Dan

seeing himself ACC Dan

The present analysis allows also for a simple account of subjectless


action nominalizations involving an anaphor such as the following.

(33) (H) reiyat acma ba -mar?a


seeing herself in the mirror

seeing herself in the mirror

Given our binding theoretic assumptions (see footnote 11), there is no

b. DP

D NP

the N'

N' SellP

N Sel the party

secretary

1 We are assuming here the particular theta-theoretic variant of the binding theory
veloped in Williams (1987a) and (1989). In that approach, binding relations are cons
as relations between theta roles rather than as relations between NPs occupying syn
positions. In this approach a theta role acquires a binding-theoretic status depending o
NP to which it is assigned. For example, a theta role is a theta-anaphor if it is assign
an anaphor and so forth.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 375

need for an NP occupying a syntactic position to serve as an antecedent


to the anaphor in (33). The configuration corresponding to (33) would be
the one in (30). The external theta role of the verb is the antecedent of
the internal theta role, which, being assigned to an anaphor in object
position, is a 'theta-anaphor'. Note that, since the anaphor is in the bound
genitive construction with the head and must, therefore, be assumed to
move to the specifier position of NP, there is no place for a PRO or any
other antecedent in that position.
It is also clear that the antecedent could not be an NP external to the
construction as the following example shows.

(34) (H) reiyat acma ba- mar?a garma le-kax Se Dina


seeing herelf in the mirror caused to it that Dina

hexlita la-lexet la- marpea


decided to go to the clinic

Seeing herself in the mirror caused Dina to decide to go


to the clinic.

In the above example the anaphor acma 'herself' is construed as coreferen-


tial with Dina. However, it cannot be assumed that it is Dina which
satisfies the binding-theoretic requirement (Principle A) holding with re-
spect to acma. This is because Dina is embedded in a lower clause which
is itself embedded in an NP which is, again, embedded in a PP complement
of the main verb. Principle A, therefore, must be assumed to be satisfied
within the subjectless action nominalization construction which figures as
subject of the sentence in (34).

3.2. Free and Double Genitive Variants

In this section we show how our analysis accounts for action nominaliz-
ation constructions which involve the free and the double genitive constru-
als. These cases are particular to Hebrew. Our main underlying assump-
tion concerns the nature of Sel-phrases. We take Sel-phrases to be adjuncts
and adopt Lebeaux's (1988) approach to adjuncts. In this approach, ad-
juncts may be freely adjoined to the tree structure at a later stage in the
derivation.12 In particular, a stage in the derivation later than NP-Struc-

12 We are assuming here the model of grammar developed by van Riemsdijk and William
(1981), who argue for the existence of an intermediate level of representation between D
and S-structure, a level to which they refer as NP-Structure. A'-movement, of which head
movement is a particular instance, maps NP-Structure onto S-Structure.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
376 ILAN HAZOUT

ture, which is the level at which thematic relations are established.'3 We


start with the free genitive variant as in the following example.

(35) (H) ha- axila Sel Dan et ha-tapuax


the eating Sel Dan ACC the apple

Dan's eating the apple

The two relevant derivational stages of (35) are as follows.

(36)a. NP-structure

NPk

det Nt

the Nk VP.

NOM V NP
(9Rk') NlO

eat

(4i, Bj)

13 This approach to Sel-phrases is discussed and argued for in detail in Hazout (1990).

This content downloaded from


81.194.27.167 on ffff:ffff on Thu, 01 Jan 1976 12:34:56 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 377

b. S-structure

DP

D NP

the N'

N' VP.

N Sel Dani V NP
A~~~~( (i) ll

V N e the apple

eat NOM
(Rgk,Ri)

In (36a) thematic relations are established. The thematic roles available


in the thematic structure of the verb are assigned and satisfied. In parti-
cular, its internal theta role(s) are satisfied within VP and its external
theta role is satisfied by NOM. (36b) is the configuration resulting from
the application of two operations: first, the usual head movement of V to
NOM and, second, the adjunction of a Sel-phrase. Here the Sel-phrase is
construed with the head in the way that this happens in any free genitive
construction (see fn. 10), it is coindexed with the internal theta role of
the head, NOM in this case, which is at this point assigned the external
theta role of VP. In this way, the NP in the Sel-phrase shares the same
index with the internal R-role of NOM and with VP, that is, it is coreferen-
tial with the external theta role of VP and is, therefore, its logical subject.
Here, again, we see how independently motivated mechanisms of genitive
formation interact with the particular syntactic structure of an action
nominal to produce the desired effect. The crucial aspect of this analysis
is that the external theta role of VP is satisfied at the level at which
thematic relations are established, namely, NP-structure. This is made
possible by the thematic properties of NOM.
The double genitive variant is exemplified by the following subjectless
action nominalization.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
378 ILAN HAZOUT

(37) (H) axilat-o Sel ha-tapuax

eating-his Sel the apple

the eating of the apple

The relevant representations of (37) are the following.

(38)a. NP-structure

DP

D NP

POSS N'

N VP_

NOM V NP
QRk, R ) (6j, Bj) l

he

b. S-structure
DP

rF D NP

P0SS NP N'

he N' VPj

N Sel V NP
the apple

V N e e

eat NOM

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 379

In (38a) the pronoun 'he' ('it') is assigned the internal th


verb. Due to the lack of accusative Case, as in any subjectless nominaliz-
ation, it moves to the Spec-of-NP, a position in which it can be assigned
genitive Case in the manner usual to the bound genitive construal. A
pronoun in the bound genitive is realized as a clitic pronoun. The surface
word order of (37) is derived by a movement of N to D as shown by the
arrow in (38b). The Sel-phrase is adjoined to this structure as shown in
(38b). However, the NP in the Sel-phrase is not coindexed with the
internal R-role of NOM. If it were coindexed with the internal theta role
of NOM (and, therefore, the external theta role of VP), the structure
would not be a subjectless action nominalization, but rather an active
action nominalization with 'the apple' construed as a subject. The result
would be the following grammatical but nonsensical NP in which the
object pronoun is assigned accusative rather than genitive Case.

(39) (H) ha- axila Sel ha-tapuax ot-o


the eating Sel the apple ACC-his

the apple's eating of him

Thus, in (37) the Sel-phrase is not integrated into the structure by the
theta-theoretic mechanisms of coindexation that are generally assumed in
this work. Rather, it is licensed by the mechanism involved in the phenom-
enon known as clitic doubling, the principles of which are independent of
the syntax of action nominals. The Sel-phrase in this structure is, thus,
licenced only by the relation in which it stands to the pronoun.
Our views with respect to clitic doubling in Hebrew genitive construc-
tions are discussed in detail elsewhere.14 The main idea of our account is
that the relation between the Sel-phrase and the clitic pronoun in a double
genitive is of the same logical type as that between an operator and a
bound variable and is regulated by the same principles which regulate
constructions with resumptive pronouns.15 At the relevant level of repre-
sentation (LF) the clitic pronoun is construed as a variable in an open
expression or a function which takes the genitive Sel-NP as its logical
argument, as follows.

(40) Ax[x's eating] (the apple)

By the semantic operation of lambda-conversion, 'the apple' replaces


variable in the bracketed part of (40) and its interpretation as the o
of 'eating' is, thus, made possible.

14 See Hazout (1990), Chapter 2.


15 We follow the spirit of Chomsky's (1982) approach to resumptive pronouns.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
380 ILAN HAZOUT

3.3. Passivization in Action Nominalizations

The analysis of action nominalizations presented above reflects the view


that the assignment of the external theta role of V/VP operates in exactly
the same way in both the subjectless and the active variants. In particular,
the derivation of the subjectless variant, according to this view, involves
no process of passivization. Under some views of passivization, this phe-
nomenon involves a lexical process which affects the external argument
of a verb by way of internalizing it. An internalized argument becomes
syntactically inactive in that its overt syntactic realization is not obligatory.
In the following we present four arguments, three of which based on
data from Hebrew and one on data from Arabic, showing that the sub-
jectless variant of action nominalizations does not involve such a process.
These arguments provide additional support for the theta-theoretic ap-
proach advocated in this paper. This is for the following reason. Given
that, as we show, no passivization is involved in subjectless action nominal-
izations and that their underlying structure involves a VP, it becomes
necessary to explain how the external theta role of VP is satisfied. The
internal R-role of NOM, thus, becomes crucial for this explanation.

3.3.1. Verbs Which Don't Passivize

There are verbs in Hebrew which may figure as transitive verbs in active
sentences but cannot be passivized. The two following pairs of examples
show the verbs calax 'cross' and hitxil 'begin, start', respectively. (41a)
and (42a) are the active sentences and their (b) counterparts are the
ungrammatical passives.

(41)a. (H) ha-cava calax et ha-teala


the army crossed ACC the canal

the army crossed the canal

b. (H) *ha- teala niclexa al-yede ha-cava


the canal was crossed by the army

(42)a. (H) ha- saxkan hitxil et ha-hacaga


the actor started ACC the play
the actor started the play

b. (H) *ha- hacaga hutxela al yede ha-saxkan


the play was started by the actor

There is apparently no morphological, phonological or semantic reason for

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 381

this deviance. Rather, it seems that the inability to undergo passivization is


an arbitrary lexical property encoded in the lexical entries of verbs like
calax 'crossed', hitxil 'started' and others.
Within the proposed analysis of subjectless action nominalizations, it is
these lexical items that would be inserted in the corresponding underlying
structures. It follows that a passivization analysis predicts that subjectless
variants of action nominalizations with these verbs should be impossible.
The following pair of examples shows this prediction to be wrong.

(43)a. (H) cixat ha-teala bi-zehirut


crossing the canal carefully

crossing the canal carefully

b. (H) hatxalat ha-hacaga al-yede ha-lahaka


starting the play by the group

the starting of the play by the group

Note that the inability of these verbs to undergo passivization is not


limited to their morphology. If this were so, it would have been possible
for these verbs to undergo all the operations involved in passivization
except for the morphological one. This would lead to the prediction that
the following active sentences, involving the verbs calax 'cross' and hitxil
'begin, start' should be possible with a passive interpretation. This is
clearly wrong: (44b) is good, but not with a passive interpretation.

(44)a. (H) *ha- teala calxa


the canal crossed

b. (H) *ha- hacaga hitxila


the play started

3.3.2. Verbs Which Do Passivize

A converse argument can be made as follows. One finds in Hebrew verbs


such as hiSpia 'influence' which take a prepositional complement.

(45) (H) Dan hiSpia al Dina


Dan influenced on Dina
Dan influenced Dina.

Interestingly, the verb in (45) can be passivized. In this process, however,

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
382 ILAN HAZOUT

the preposition al 'on' is dropped and its object becomes the derived
subject.

(46) (H) Dina huSpea al-yede Dan


Dina was influenced by Dan

Dina was influenced by Dan.

If a passive verb like the one in (46) were the input to the nominalization
process proposed for this construction, one would wrongly predict the NP
in (47) to be grammatical.

(47) (H) *haSpaat Dina al-yede Dan


influence Dina by Dan

The verb hiSpia 'influence' is far from being an isolated case in this
respect. In (48) we list verbs which manifest the same phenomenon. It is
likely that the possibility for a preposition to be dropped is related to its
semantic contribution. That is, a preposition may be dropped only if it
has no semantic import and its presence is arbitrarily required by a certain
verb.

(48)a. Verbs taking the preposition be 'in':


tipel be 'take care of, deal with', bagad be 'betray', baat be
'kick', tamax be 'support'

b. Verbs taking the preposition al 'on':


diveax al 'report', hexlit al 'decide', hixriz al 'declare', himlic
al 'recommend'

c. Verbs taking the preposition le 'to':


hirSa le 'allow, permit', azar le 'help, support', kara le 'call'

3.3.3. Anaphors

A passivization analysis of subjectless action nominalizations faces a seri-


ous problem with structures involving anaphors. Such an analysis leads
one to expect example (49) below to be bad like its sentential counterpart
in (50). This prediction is clearly wrong.

(49) (H) re'iyat acmo be-ofen barur


seeing himself clearly

seeing himself clearly

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 383

(50) (H) *acmo nir'a


himself was seen

Interestingly, under the analysis developed in this pap


ational operations applying in both cases are parallel: I
moves to Spec of NP in order to be assigned genitive Case, whereas in
(50) it moves to Spec of IP in order to be assigned nominative Case. If
we assume that it is also the same lexical operation which applies in both
cases (under the approach to passivization relevant to this discussion), we
would wrongly predict both (49) and (50) to be bad.
In this paper we adopt Williams' (1987a, 1989a) approach to the binding
theory. Under this approach, anaphoric relations hold not between NPs
occupying syntactic positions, but rather between theta roles in the argu-
ment structure of a given lexical element. The status of a theta role
with respect to the binding theory (as anaphoric, pronominal, etc.) is
determined in this approach by the constituent to which it is assigned.
Under our analysis of subjectless action nominalizations, no manipulation
of the argument structure of the verb took place in (49). We, therefore,
expect no reversal of asymmetries. The antecedent of the internal theta
role of ra'a 'see' (a theta-anaphor in Williams' terms) is in this case the
external theta role.

3.3.4. Control Structures in Arabic

The phenomenon of control of subjectless action nominalizations in Arabic


raises another senrous problem for a passivization analysis. The argument
here is strikingly simple. Consider the following example.

(51) (A) yuriid zaid [?akl attufaaHa]


wants Zaid eating the apple

Zaid wants to eat the apple.

The material in brackets in the above example is a subjectless action


nominalization with the object 'the apple' being assigned genitive rather
than accusative Case. However, (51) can only mean: 'Zaid wants to eat
the apple'. It is hard to imagine what would be the paraphrase of a passive
construal of this sentence. It certainly does not have the interpretation:
'Zaid wants the apple to be eaten' (in which, in any case, 'Zaid' is not a
controller of anything). In this, (51) is identical to its English counterpart:
John wants to eat the apple, in which an indisputably active form of the
verb eat occurs. It is, therefore, hard to see what it would mean to say

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
384 ILAN HAZOUT

that the embedded subjectless action nominalization in (51) is the output


of a passivization process.

3.4. A Note on Previous Work

The present analysis of action nominalizations, in particular the assump-


tion of an underlying verb, bears certain similarities to previous proposals
in the literature, the best known of which are concerned with the syntax
of the English gerund. The variant of English gerunds which is most
similar to Hebrew and Arabic action nominalizations is the construction
known as Poss-ING gerund, to be distinguished from Acc-ING and ING-
of. (52) is a Poss-ING gerund.

(52) John's eating the apple

Although similar in many respects, English Poss-ING gerunds and


Semitic action nominalizations show certain important differences. Gen-
erally, while both construction types have a mixture of verbal and nominal
properties, one might say that the semitic construction is more nominal.
First, adjectives can occur in the semitic construction but not in the En-
glish. Second, there is nothing in the English construction which corre-
sponds to the variety of genitive construals that are possible in the Hebrew
construction. Moreover, the free genitive variant of the Hebrew action
nominal (e.g., example (35) above) can have a determiner, while this is
impossible in the English construction. Finally, the subjectless variant of
the semitic construction has no counterpart in English. The closest thing
in English to the Semitic subjectless action nominalization is the NP in
(53).

(53) eating the apple

In (53) the object is assigned accusative Case, an impossibility in Hebrew


and Arabic. On the other hand, (54), which corresponds more closely to
the semitic construction, is bad.

(54) *the apple's eating

These are, in our view, substantial differences. Therefore, an analysis of


the Semitic construction cannot be based entirely on the results of previous
research on the English gerund. Those results are important and substan-
tial, but clearly new elements must be added. The following is a brief
review of previous work.
To the best of our knowledge, all major analyses of the Poss-ING gerund
within the generative tradition share the assumption of an underlying verb

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 385

and a VP. Thus, Chomsky (1981) (and the slightly different proposal of
Schachter 1976) suggests the structure in (55). (55) is assumed to be a
configuration in which the subject NP is assigned genitive Case.

(55) NP

NP VP

However, (55) is unsatisfactory to the extent that it does not conform to


standard X'-theoretic assumptions.
A proposal more similar to the analysis developed in this paper is the
one of Horn (1975).

(56) NP

NP N'

ING VP

Here the nominal affix -ING is the head. Thus, genitive Case assignment
to the subject receives a more natural explanation and the surface ordering
of elements is accounted for by assuming the application of Affix-Hopping
to -ING.
Somewhat intermediate between Chomsky's and Horn's positions is that
of Jackendoff (1977), which assumes the following PS-rule for gerundive
nominals.

(57) N" -- ING - V'

In Jackendoff's analysis also, -ING is attached to the verb by Affix-


Hopping.
More recently, new proposals have been put forth by Abney using the
DP hypothesis of NP structure. Abney (1986, 1987) suggests the following
structure for POSS-ING gerunds.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
386 ILAN HAZOUT

(58) DP

POSS D'

D VP

ING

While the verbal properties of


by the assumed VP substructure, the unavailability of determiners and
adjectives follows from the fact that, generally, they are not licensed by
D but by N. Following earlier work of Jackendoff, Abney (1987) suggests
also a slightly different variant of the above proposal which looks as
follows.

(59) DP

John's D'

D NP

ING VP

V DP

sing the Marseillaise

Finally, following the spirit of Abney (1986, 1987), but adopting a more
elaborate view of the structure of Noun Phrases, Ouhalla (1988) captures
the same intuitions by proposing the following structure.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 387

(60) A rP

Spec A'

Agr NOMP

NOM' XP P

ING Sc XI/V

John X/V Obj acc)

t the ball

In (60) the subject moves to the Spec of AgrP in order to receive genitive
Case by coindexation with Agr. The appropriate morphological output is
derived by the successive applications of head movement as illustrated by
the arrows. An interesting feature of Ouhalla's analysis is the idea that
the head, hit, is unspecified for grammatical category. It is recategorized
as V by percolation of features following head movement to the verbal
morpheme -ING. This categorial specification affects the entire constituent
up to its maximal node (deviating slightly from Ouhalla's 1988 pre-
sentation, we express this idea here by the notation XP/VP).
We see, then, that the verbal aspect of the syntax of POSS-ING gerunds
is captured in more or less the same way throughout the developement
of generative grammar, since at least the early seventies. The main idea
of these analyses, that of an underlying verb, is assumed in the present
work as well. As already stated, however, none of these proposals is
sufficient in our view to deal with the additional complexities of the Semitic
construction which differs from the English gerund in the ways observed
above.

4. THE ISSUE OF LEXICALISM

The account outlined in the previous section may be characterized as non-


Lexicalist in that it assumes a syntactic process by which the noun that is

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
388 ILAN HAZOUT

the head of an action nominalization construction is derived, the head


being a single word as far as morphological and phonological criteria are
concerned. The analysis developed here is, therefore, sharply incompat-
ible with any theoretical conception which wishes to exclude the possibility
of any word formation processes in the syntax. However, by taking this
line, we are able to explain many of the properties of the construction,
some of which are configurationally sensitive (e.g., binding facts and the
distribution of adverbs) and some of which have to do with the properties
that may be attributed to lexical items by virtue their of being members
of a certain word class (e.g., Accusative Case assignment by a verb). The
burden of providing an explanation in this approach is put on the proper-
ties of the configuration and the syntactic operations that it makes possible,
as well as on properties of certain lexical elements (e.g., NOM) whose
existence may be only postulated within a non-lexicalist approach.
The issue of lexicalism is still a matter of dispute in linguistic literature,
and alternative lexicalist approaches to the action nominalization construc-
tion are not too hard to imagine.16 In this section we present two argu-
ments to support the non-lexicalist (or syntactic) analysis presented in
section 3. The first is based on certain facts concerning the interpretation of
Hebrew action nominalizations and the second on the analysis of causative
constructions in Arabic.
Most lexicalist approaches to action nominalizations take the head of
the construction to be a noun. Analyses along these lines are proposed in
Borer (1984), Ritter (1986), Doron (1989) and Siloni (1990). In our view,
all of these accounts are unsatisfactory in explaining the verbal properties
of action nominalizations. With respect Case properties, lexicalist analyses
of this type must assume that the source of accusative Case in this construc-
tion is the head noun (cf. Ritter 1986, Doron 1989 and Siloni 1990). Under
such an assumption, one is forced to the conclusion that action nominals
form a special class of nouns capable of assigning accusative Case. This,
as noted in fn. 3, is undesirable at least to the extent that the relation
between these special nouns and their verbal counterparts becomes unpre-
dictable. Otherwise, it may be assumed that the Case assigner is the
particle et, rather than the noun itself (cf. Borer 1984). This option is
criticized in fn. 3. Other data, which we present in section 2.2 as well as

16 For a recent defense of the lexicalist hypothesis, see DiSciullo and Williams (1987).

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 389

in this section, in favor of a non-lexicalist approach is unaccounted for,


as far as we can see, by the lexicalist analyses mentioned above.17

4.1. The Interpretation of Action Nominalizations

The argument in this section is based on the observation of a certain


correlation between Case marking and interpretation in action nominaliza-
tions. Our claim is that an explanation of this correlation is best achieved
in terms of a non-lexicalist analysis of action nominalizations. The relevant
contrasts are shown by the following pairs.

(61) a. (H) reiyat Dan et Dina


seeing Dan ACC Dina

Dan's seeing Dina

17 An interesting option in keeping with the lexicalist hypothesis is proposed in Fas


(1988) who follows ideas of Abney's (1986) analysis of the English gerund. Fassi-Fehr
takes the action nominal head to be a verb as in (i).
(i) DP

D VP

NPj V'

V NP2

Under this proposal, V moves to D and NP1 is assigned genitive Case by D, a nominal
category.
Fassi-Fehri's proposal is, however, extremely sketchy and, therefore, hard to evaluate.
We can only point out certain potential difficulties. First, the head of an action nominal is
morphologically a noun in that it carries, in the standard Arabic dialect studied by Fassi-
Fehri, morphological Case marking. It is not clear how this fact can be accommodated into
an analysis which views an action nominal head as a verb. Secondly, the analysis in (i) has
nothing to offer with respect to the particularities of the Hebrew construction. These include
the free and the double variants as well as the occurrence of adjectives. The analysis in (i)
adopts an idea of Abney's which was designed for the treatment of English gerunds. How-
ever, as observed in fn. 6, there are crucial differences between the English and the Semitic
constructions which may be summarized by characterizing the latter as more nominal. Thus,
our criticism of Abney's (1986) analysis as inadequate for the analysis of the Hebrew
construction carries over to Fassi-Fehri's proposal. Finally, a potential problem for this
proposal is the analysis of the subjectless variant which is not discussed by Fassi-Fehri (1988).
It seems that the only available view of this variant under Fassi-Fehri's proposal is as an
instance of passivization. An explanation by which the same effect is achieved through the
interaction between a verb and a nominalizer, as this is done in our analysis, is not available
under this proposal. However, it seems to us that the evidence against passivization presented
in section 3.3 is overwhelming and, therefore, problematic for Fassi-Fehri's (1988) analysis.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
390 ILAN HAZOUT

(61)b. (H) reiyat Dina Sel Dan


seeing Dina Sel Dan

(62)a. (H) harisat picasso et yeruSalayim


destruction Picasso ACC Jerusalem

Picasso's destroying Jerusalem

b. (H) harisat yeruSalayim Sel picasso


destruction Jerusalem Sel Picasso

Picasso's destruction of Jerusalem

While in the (a) examples the object argument is in the accusative Case,
the arguments in the (b) variants are in the genitive Case, one in the
bound and the other in the free genitive. The most likely interpretation
of (61a) is one in which it refers to the event of Dan's seeing Dina. (61b),
on the other hand, although somewhat vague, is more likely to be assigned
a generic interpretation, in which reference is made to Dan's way of
viewing Dina in general.
The contrast in (62) is much sharper. (62a) can only have the event
interpretation by which it refers to the event of the destruction of Jerusa-
lem by Picasso (say, a Roman emperor). For (62b), on the other hand,
the most likely interpretation is radically different. In a pragmatic context
in which Picasso is widely known as a famous painter, it is most likely to
be construed as referring to a picture depicting the destruction of Jerusa-
lem which was painted by Picasso.
In fact, an NP of the form of (62b) may also have an event interpreta-
tion, in which the Sel-NP is construed as an agent. This may possibly be
brought out more clearly by (63), in which the Sel-NP is the name of the
real destroyer of Jerusalem.

(63) (H) harisat yeruSalayim Sel Titus


destruction Jerusalem Sel Titus

Titus' destruction of Jerusalem

This freedom of thematic interpretation is totally absent in (62a). In this


example 'Picasso' is strictly interpreted as the agent of 'destruction'.
The freedom of thematic interpretation in simple noun-to-noun con-
structions is a well-known fact. Thus, in a simple genitive construction
such as John's book, 'John' may be understood as standing in any of a
wide range of possible relationships with respect to the book, such as
ownership, authorship and so forth. The same has been observed by

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 391

Williams (1987a) for English derived nominals. Thus, according to Willi-


ams, your destruction of the city may have, among others, the interpreta-
tion 'your account of the destruction of the city'. Under a lexicalist ap-
proach to action nominalizations, both variants in (61) and (62), although
involving a different ordering of elements, are viewed as instances of a
noun-to-noun relationship. The difference of interpretation between them
would, therefore, be unexpected. It is hard to see how this difference
between the two variants could be attributed to differences in word order
and Case marking without invoking different syntactic structures.
Within the analysis proposed in this paper, these facts receive a natural
explanation. Considering the pair in (62), (62b) is a typical instance of a
genitive construction involving two lexically given nouns ('destruction' and
'Picasso'). The interpretation of such constructions is typically open to
pragmatic influences. (62a), on the other hand, is the output of a syntactic
process of nominalization and has the following underlying structure.

(64) DP

D NP

POSS NP N'

Picasso N VP

NOM V NP

destroy Jerusalem

The obligatory agent interpretation of Picasso in this case is an outcome


of the obligatoriness of the assignment of the external theta role of VP
first to NOM and then from NOM to the NP in SPEC. Thus, the interpreta-
tion of the subject in this case is strictly determined by syntactic factors
to the exclusion of any pragmatic ones. The object NP in (64), Jerusalem,
may only be assigned accusative Case. In this way, the correlation between

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
392 ILAN HAZOUT

the interpretation of the subject and the Case marking of the object
receives a simple account.'8

4.2. Arabic Causatives and the Syntax of Action Nominalizations

The argument in this section is based on the analysis of causative construc-


tions in Arabic. The structure of the argument is as follows. We observe,
first, an alternation between two types of causative constructions to which
we refer as the double accusative and the dative causative (or simply
dative). We argue that certain facts about these constructions necessitate
an analysis of the double accusative variant which takes the causative verb
to be the output of syntactic head movement, that is, a non-lexicalist
analysis. On the other hand, we show that in the second type, the causative
verb is base generated in its complex mophological form.'9 We then show
that Arabic action nominalizations manifest the same alternation between
the double accusative and the dative variant, and we conclude that the
option of a non-lexicalist derivation must, for this reason, be open for
action nominalizations.
The following pairs are typical examples of causative constructions.

(65)a. (A) zaid 9allam al-qasm haad'a l-kitaab


Zaid teach the class this the book

Zaid taught the class this book.

18 The difference between the (a) and the (b) examples in (61)-(62) is reflected in other
syntactic properties. The most striking difference has to do with the phenomenon of clitic
doubling. In a structure of the type in the (b) exampes, if the object NP is a pronoun, then
it must be coreferential with the NP in the Sel-phrase.

(i)a. (H) *harisat -a Sel ha-romaim


destruction-her Sel the romans

b. (H) harisat -am Sel ha-romaim


destruction-their Sel the romans
the destruction of the romans

On the other hand, clitic doubling is impossible for the variant in the (a) examples.

(ii) (H) *harisat -ai et ha-irs


destruction-her, ACC the cityi

A detailed discussion of this contrast is undertaken in Hazout (1990).


'9 Similar alternations have been previously observed in the literature.
reviews some of the literature, accounts for such alternations in terms of
developed in that work by distinguishing between merger at s-structure

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 393

b. (A) zaid 9allam haad'a l-kitaab li-l- qasm


Zaid teach this the book to the class

Zaid taught this book to the class.

(66)a. (A) zaid Sarrab al-Pawlaad al-Haliib


Zaid made-drink the kids the milk

Zaid made the children drink the milk.

b. (A) zaid Sarrab al-Haliib li- l-?awlaad


Zaid made-drink the milk to the kids

Zaid made the children drink the milk.

In the (a) examples above, the verb is followed by two accusative marked
NPs, whereas in the (b) examples, the order of the two NPs is reversed.
The first is marked for accusative and the second is governed by the
dative preposition li. The two variants may be schematically represented
as follows.

(67)a. Double Accusative: NP1 V NP2 NP3


b. Dative Causative: NP1 V NP3 P NP2

The first set of facts in our argument is based on anaphoric relation-


ships.20 Using the representations in (67), we observe that in (67a), NP1
can be the antecedent of NP2, as shown in (68a), but not of NP3, as shown
in (68b).

(68)a. (A) samma9- na ba9da-na l-Ranaani


made hear-we selves- our the songs

We made ourselves hear the songs.

b. (A) *samma9- na l-awlaad ba9da-na


make hear-we the children selves our

Equally, NP2 may serve as an antecedent to NP3, but the reverse does
not hold.

(69)a. (A) Hanna 9addad aHmad Haalu


John made count ahmed himself

John made Ahmed count himself.

20 The data in this section is borrowed mainly from Hoyt (1989), as well as from Mushaweh
(1986).

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
394 ILAN HAZOUT

(69)b. (A) *Hanna 9addad Haalu Ahmad

John made count himself Ahmed

In the dative construction, on the other hand, NP3 may be anaphorically


related to the subject.

(70) (A) Hanna 9addad Haalu la-aHmad


John made count himself to Ahmed

Johni made Ahmed count himself .

A second set of facts concerns the possibilities of citicization in the


causative variants. Using the patterns in (67), in a double accusative, if
NP2 is a pronoun, then it can (in fact, must) be cliticized onto the verb,
as in (71b), however, this is impossible for NP3, as shown in (71c). In
(71a) NP2 and NP3 are two non-pronominal (feminine and masculine,
respectively) NPs.

(71)a. (A) fahham aHmad hind addars


made understand Ahmed Hind the lesson

Ahmed made Hind understand the lesson.

b. (A) fahham -a aHmad addars


made understand-her Ahmed the lesson

Ahmed made her understand the lesson.

c. (A) *fahham -o aHmad hind


made understand-it Ahmed hind

In the corresponding dative, on the other hand, it is perfectly possible for


NP3 to be cliticized on the verb.

(72) (A) fahham -o aHmad la-hind


made understand-it Ahmed to Hind

Ahmed explained it to Hind.

A third set of facts is related to issues of interpretation. Hoyt (1989)


observes the following facts. Double accusative sentences with a modifying
adverb are ambiguous between two possible construals of the adverb, as
indicated in the following examples.

(73) (A) Hanna massak xaalid I-Sunta bi-beit l-'awlaad


John made hold Khalid the suitcase in house the kids

John made Khalid hold the suitcase in the kids house.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 395

(i) John caused the following: Khalid held the suitcase in the
kids house.

(ii) It was in the kids house that John made Khalid hold the
suitcase.

(74) (A) Hanna ?akkal xaalid I-sanduiS bi-tariqa biSa9


John made eat Khalid the sandwich in manner rude

John made Khalid eat the sandwich in a rude manner.

(i) Khalid's actions were rude.

(ii) John's actions were rude.

Hoyt (1989) also notes an interesting contrast between double accusative


sentences and semantically equivalent sentences of which the main verb
is not a morphological causative. In the following example with the causa-
tive verb 'cause-to-die', the same ambiguity is observed with respect to
an accompanying adverb as in (73)-(74).

(75) (A) Hanna mawwat xaalid bi- bila9


John caused to die Khalid with swallowing

1-mikrofilm
the microfilm

John made Khalid die by swallowing the microfilm.

(i) Khalid swallowed the microfilm.

(ii) John swallowed the microfilm.

However, in the following example, with the non-causative verb 'kill',


only one construal is possible for the adverb, namely, the one in which it
is related to the subject.

(76) (A) Hanna qatal xaalid bi-bila9 l-mikrofilm


John killed Khalid with swallowing the microf
John killed Khalid by swallowing the microfilm.

(i) John swallowed the microfilm.

We assume (77) to be the underlying structure of the double accusative.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
396 ILAN HAZOUT

(77) IP

NP1 VP

V NP2 VP

\ I
CAUSE V NP3

(Ai, Bj Y , DI

In this structure the abstract verbal element CAUSE subcategorizes for a


VP and an NP object to which it assigns a theta role. The surface structure
corresponding to (77) is derived by a single application of head movement
adjoining the lower verb to the morpheme CAUSE.

(78) IP

NPi VP

V NP2 VP

Vi CAUSE V NP3

In (78) both NP2 and NP3 are govemed, either by a lexical verb or by
the trace of a verb, and are in this way assigned accusative Case. Within
the variant of binding theory assumed here (based on Williams 1987a,
1989), the binding facts with respect to (77) are directly accounted for.
Given the principle that two elements X and Y may be anaphorically
related if they are coarguments (or coarguments of coarguments), NP1
and NP2 in (77) may be anaphorically related, since they are coarguments
of the verb CAUSE. The same holds with respect to NP2 and NP3 as
indicated by the arrows in (77). Since NP1 and NP3 are not coarguments,
they may not be anaphorically related. Note that the two verbs in (77)
constitute two separate thematic (argument) complexes and, therefore,

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 397

no ungrammaticality results from the fact that NP2 is


by both verbs at the same time.
As to the cliticization facts, our explanation is based on the general
observation that as far as cliticization phenomena in Hebrew and Arabic
are concerned a pronoun may only be cliticized onto a lexical head (V,
P, N) directly governing it. In the configuration in question, (77) and (78),
only NP2 is governed by the verb CAUSE-V. NP3 has no governing lexical
verb to be attached to. The contrast between (71b) and (71c) is, thus,
explained.
In the proposed structure for the double accusative, an adverb may be
attached at each one of the VP nodes, and each construal corresponds to
one of the two possible interpretations. The facts in (73)-(76) are thus
explained.

(79) IP

NPi VP

V NP2 VP Adv.

I
CAUSE V NP3 Adv.

The proposed analysis also explains more elementary facts of word


order. As (80) shows, a reversal of order of NP2 and NP3 is impossible.

(80)a. (A) zaid Sarrab 1-9awiaad l-Haliib


Zaid made-drink the kids the-milk

Zaid made the kids drink the milk.

b. (A) *zaid Sarrab l-Haliib l-?awlaad


Zaid made-drink the milk the kids

This fact follows from the hierarchical structure assigned to this construc-
tion by our analysis. In this structure each of the two NPs is governed by
a different verbal head, the two heads being strictly linearly ordered with
respect to each other.
Summing up our analysis of the double accusative variant, this analysis
relies crucially on the assumption of a double-VP configuration, the two
VPs being headed by the morpheme CAUSE and a verb, respectively.
The surface causative verb is the output of a syntactic operation.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
398 ILAN HAZOUT

Turning now to the dative variant, we note first the following obser-
vation made by Cowell (1964).

In some cases ... the first object of the causative may be replaced by a la-phrase and put
after the remaining object: Sawwaf jeruHak le-ddaktoor 'show your wounds to the doctor',
samme9 darsak la- ?abuuk 'recite your lesson to your father'. The use of a prepositional
complement with the causative in lieu of a first object generally implies a certain idiomatic
specialization with respect to the underlying simple verb: samma9 meaning 'to recite', kauab
meaning 'to dictate', etc. (p. 241)

Similar facts are observed by Mushaweh (1986) for Syrian Arabic, as well
as for the standard dialect. Thus, the interpretation associated with the
double accusative is strictly compositional and can be paraphrased as
'make, or force, someone to do something'. In contrast, a dative construc-
tion may receive idiomatic and derived meanings.
One might say that the ability to undergo semantic drift is the property
par excellence of lexical items. If this is so, then one is led to the conclusion
that the causative verb which occurs in dative constructions is a single
lexical item inserted in its complex morphological form at the level of NP-
structure.
The effect of the lexical process by which a causative verb is formed is
twofold: First is the addition of an external agent theta-role, second is the
internalization of the external theta role of the verbal form. The resulting
verb has only one accusative Case feature to assign and, therefore, a
special mechanism is invoked by which a dummy Case-marker, li, is
inserted. The structure of a dative causative construction is the following.

(81) IP

NPi VP

CAUSE-V NP3 li-NP2

The facts observed earlier with respect to the dative construction are
accounted for as follows. First, NP1 and NP3 can be anaphorically related
by virtue of being coarguments. As to cliticization, NP3 in (81) is governed
by the lexical head CAUSE-V and can be cliticized onto it (cf. (72)).
This analysis of the dative variant is further supported by facts about
adverbial modification. Given the analysis in (71), we should not expect
ambiguity in the interpretation of adverbs of the kind observed for the
double accusative variant. (82) confirms this prediction.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 399

(82) (A) Hanna ?akkal l-sandwiS la-xaalid bi -ssikiin


John made eat the sandwich to Khalid with-the knife

John made Khalid eat the sandwich with the knife.

As reported by Hoyt (1989), (82) allows only for an interpretation in


which Hanna uses the knife to make Khaalid eat the sandwich. An inter-
pretation in which Khaalid uses the knife is not available.
To sum up, our conclusion is that in the double accusative variant the
causative verb is the output of a syntactic process whereas in the dative
causative the main verb is the output of a lexical process. The point of
this argument is that the same double accusative/dative alternation is
attested in action nominalizations as demonstrated by the following pairs.

(83)a. (A) ta9liim zaid al-qasm haad'a l-kitaab


teaching Zaid the class this the book

Zaid's teaching of this book to the class.

b. (A) ta9him zaid haad'a 1-kittab li-l- qasm


teaching Zaid this the book to the class

Zaid's teaching of this book to the class.

(84)a. (A) taSriib hind al-?awlaad al-Haliib


making-drink Hind the kids the milk

Hind's making of the kids drink the milk.

b. (A) taSriib hind al-Haliib li- l-Pawlaad


making-drink Hind the milk to the kids

Hind's making of the kids drink the milk

It follows that at least the double accusative variants, the (a) examples in
(83)-(84), must involve a syntactic derivation of the type argued for
above. The head nouns in (83)-(84) may be viewed as a combination of
the three morphemes NOM, CAUSE, and V. However, at least with
respect to the double accusative, the combination CAUSE-V must be
viewed as syntactically derived, that is, it must have at least the substruc-
ture (85a) and, therefore, may only be assumed to have the underlying
structure (85b).

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
400 ILAN HAZOUT

(85)a. VP

V NP2 VP

CAUSE V NP3

b. NP

NPi N'

N VP

NOM V NP2 VP

CAUSE V NP3

We conclude that the option of a syntactic derivation for action nominaliz-


ations must be open in principle and is, therefore, given the empirical
advantages of such an assumption, the actual derivation of action nominali-
zations.
Our approach is confirmed by the fact that word order restrictions in
causative action nominalizations are identical to those attested in the
corresponding sentential structures and receive a natural explanation
within the proposed analysis. Thus, a reversal of order of NP2 and NP3 is
possible only if the dative option is taken (compare with (83) and (84)).

(86) a. (A) *ta9liim zaid haad'a l-kitaab al-qasm


teaching Zaid this the book the class

b. (A) *taSriib hind al-Haliib al-9awlaad


making-drink Hind the milk the kids

A lexicalist approach would assign these structures an analysis like the

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 401

following and would be unable to account for the respective order of NP2
and NP3.

(87) DP

D NP

NPi N'

N NP2 NP3

Another fact which is an outcome of the structural assymetry between


NP2 and NP3 in our analysis concerns the subjectless variants of (83a)
and (84a). Generally, in action nominals an object may not be assigned
accusative Case in the absence of a subject. The object must, in such a
situation, be construed in a bound genitive construction with the head
noun. In action nominalizations of the double accusative variant, only
NP2 may be affected by this process, that is, only NP2 may appear as
a genitive NP in the absence of a subject. NP3 must be marked for
accusative.

(88)a. (A) ta9liim al-qasm haad'a l-kitaab


teaching the class this the book

teaching the class this book

b. (A) *ta9liim haad'a 1- kitaab al-qasm


teaching this the book the class

(89) a. (A) taSriib al-Pawlaad al-Haliib


making-drink the children the milk

making the kids drink the milk

b. (A) *taSriib al-Haliib al-?awlaad


making-drink the milk the children

Regardless of the correct explanation for this general phenomenon,


our proposed non-lexicalist analysis of causative action nominalizations
provides a hierarchical, bi-clausal (or rather 'bi-VP') structure that makes
it possible for the relevant process to distinguish between the two NPs

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
402 ILAN HAZOUT

and to affect NP2 while ignorin


our non-lexicalist analysis. A lexicalist approach lacks the motivation for
introducing the necessary hierarchical distinction to the structure and,
therefore, an analysis along lexicalist lines, such as the one represented
in (87) above, is unable to account for this asymmetry.

5. CONCLUSION

The main task of this paper has been to provide an explanation for the
mixed verbo-nominal properties of the action nominalization construction
in Hebrew and Arabic. It was shown that a satisfactory explanation of
these properties is possible by postulating an underlying representation in
which a VP figures as the complement of the abstract nominal head
NOM. It was further shown that the postulation of such an underlying
representation not only makes possible an explanation of the verbal pro-
perties of this construction, but also of some configurationally conditioned
phenomena, such as the semantic construal of adverbs and the respective
distribution of adverbs and adjectives. The main element of the proposed
analysis is the abstract nominalizer NOM. The most important assumption
about NOM concerns its argument structure and the way it interacts with
the rest of the construction. This aspect of the analysis was executed
within the particular approach to theta theory proposed in Williams
(1987a, 1987b, 1989). NOM is a thematic nominalizer in that it provides
the external R-role which makes it possible for the construction as a whole
to be an argument. Given its internal R-role, NOM can also satisfy the
external theta role of VP and in this way function as a substitute for a
missing NP subject. It is only within a non-lexicalist approach that abstract
elements such as NOM may be postulated. Section 4 presented evidence
for a non-lexicalist approach to Hebrew and Arabic Action Nominaliza-
tions.

REFERENCES

Abney, Steven: 1986, 'Functional Elements and Licensin


bridge, Mass.
Abney, Steven: 1987, The English Noun-Phrase in its Sentential Aspect, Ph.D. dissertation,
MIT, Cambridge, Mass.
Baker, Mark: 1988, Incorporation: A Theory of Gr-anmatical Function Changing, The Uni
versity of Chicago Press.
Berman, Ruth: 1978, Modern Hebrew Structure, Tel-Aviv, University Publishing Projects.
Borer, Hagit: 1984, Parametric Syntax, Foris, Dordrecht.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
ACTION NOMINALIZATIONS AND THE LEXICALIST HYPOTHESIS 403

Borer, Hagit: 1988, 'On the Morphological Parallelism Between Compoun


in G. Booij and Van Marle (eds.), Yearbook of Morphology, Foris, Dordrecht.
Borer, Hagit: to appear, 'Derived Nominals and the Causative-Inchoative Alternation: Two
Case Studies in Parallel Morphology', unpublished ms., UC, Irvine.
Chomsky, Noam: 1981, Lectures on Government and Binding, Foris, Dordrecht.
Chomsky, Noam: 1982, Some Concepts and Consequences of the Theory of Government and
Binding, MIT Press, Cambridge.
Chomsky, Noam: 1986, Barriers, MIT Press, Cambridge.
Cowell, Mark: 1964, A Reference Grammar of Syrian Arabic, Georgetown University Press,
Washington.
Di-Sciullo, A. M. and Williams, Edwin: 1987, On the Definition of Word, MIT Press,
Cambridge, Massachusetts.
Doron, Edit: 1989, 'Derived Nominals in Hebrew', unpublished ms., Hebrew University of
Jerusalem.
Fassi-Fehri, A-K.; 1982, Linguistic Arabe: Forme et Interpretation, Publications de la faculte
des lettres et sciences humaines, Rabat.
Fassi-Fehri, A-K.: 1988, 'Generalized IP Structure, Case and VS Word Order', MIT Working
Papers in Linguistics 11.
Glinert, Luis: 1989, The Grammar of Modern Hebrew, Cambridge University Press.
Grimshaw, Jane: 1990, Argument Structure, MIT Press, Cambridge.
Hazout, Ilan: 1990, Verbal Nouns: Theta Theoretic Studies in Hebrew and Arabic, Ph.D.
Dissertation, University of Masschussetts, Amherst.
Hazout, Ilan: 1992, 'The Verbal Gerund in Modern Hebrew', NLLT 10, pp. 523-554.
Hazout, Ilan: to appear, 'Nominalizers in Theta Theory', The Linguistic Review.
Horn, George: 1975, 'On the Nonsentential Nature of the Poss-ING Construction', Linguistic
Analysis 1.
Hoyt, Katherine: 1989, 'Verb Raising in Lebanese Arabic', MIT Working Papers in Linguis-
tics 11.
Jackendoff, Ray: 1977, X'-Syntax: A Study of Phrase Structure, MIT Press, Cambridge.
Lebeaux, David: 1988, Language Acquisition and the Form of the Grammar, Ph.D. Disserta-
tion, University of Massachusetts, Amherst.
Milsark, G. L.: 1988, 'Single-ING', Linguistic Inquiry 19.
Mushaweh, Lubana: 1986, De la Syntaxe des Petites Propositions, These de Doctorat, Univ-
ersite de Paris VIII.

Ouhalla, Jamal: 1988, The Syntax of Head Movement: A Study of Berber, Ph.D. Dissertat
University College, London.
Ritter, Elisabeth: 1986, 'NSO Noun Phrases in Modern Hebrew', J. McDonough and B.
Plunkett (eds.), Proceedings of NELS 17.
Ritter, Elisabeth: 1990, 'Two Functional Categories in Noun Phrases: Evidence from Mode
Hebrew', unpublished ms., UQAM.
Riemsdijk, Henk van and Williams, Edwin: 1981, 'NP-Structure', The Linguistic Review 1.
Schachter, Paul: 1976, 'A Nontransformational Account of Gerundive Nominals in English',
Linguistic Inquiry 7.
Schachter, Paul: 1985, 'Parts of Speech', in T. Shopen (ed.), Language Typology and
Syntactic Description 1, Cambridge University Press.
Siloni, Tali: 1990, 'Hebrew Noun Phrases: Generalized Noun Raising', unpublished ms.,
Universite de Geneve.
Travis, Lisa: 1984, Parameters and Effects of Word-Order Variation, Ph.D. dissertation,
MIT, Cambridge, Massachusetts.
Williams, Edwin: 1981, 'Argument Structure and Morphology', The Linguistic Review 1.
Williams, Edwin: 1982, 'The NP Cycle', Linguistic Inquiry 13.
Williams, Edwin: 1987a, 'Implicit Argument, Binding Theory and Control', NLLT 5.
Williams, Edwin: 1987b, 'NP Trace in Theta Theory', Linguistics and Philosophy 10.

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms
404 ILAN HAZOUT

Williams, Edwin: 1989, 'The Anaphoric Nature of Theta Roles', Linguistic Inquiry 20.
Wright, W.: 1988, A Grammar of the Arabic Language, Cambridge University Press.

Received 30 May 1991


Revised 25 July 1992

9 Kalaniyot St.
Haifa, 34353
Israel

This content downloaded from


81.194.27.167 on Sun, 14 Nov 2021 19:54:36 UTC
All use subject to https://about.jstor.org/terms

You might also like