ITIUIL
~ SJournal
Name
~
I 11 11 51
Manuscript No.
6
B
Dispatch: 5.1.09
Author Received:
Journal: STUL CE: Blackwell
No. ofpages: 25 PE: Indumathi
HEBREW AS A PARTIAL NULL-SUBJECT
LANGUAGE*
Ur Shlonsky
Abstract. To account for the tense-wise and person-wise uneven distribution of
null subjects in Hebrew, we argue first that Hebrew finite T can have either a full
phi set - with person and number, a partial one - with number only or no phi set
at all. Second, Hebrew pro is argued to lack a person feature. Feature matching
with T consequently fails when T has [person], but succeeds when T lacks
[person], albeit with the subject interpreted impersonally. Third, 'contr' f pro
involves the assignment of a person feature to pro, rendering it
e of
reference. Fourth, the fact that Hebrew pro can only be first or sec
n is
an illusion: The speech act participants are associated with a f
ead
SAPo, to which the first and second person pronouns c1iticize. Th
independently-referring pro at all in Hebrew
1. Introduction
31
Hebrew is a partial null subject language, mani(!,sM
ree intertwined
asymmetries in the distribution of covert subje
nouns.
The referentiality asymmetry:
Non-referential argumental null subject
possible in every tensed
environment but one.
The person asymmetry:
Referential null subjects are peJl1ll'[t~gi9
first and second person
inflection; third person covert hip"t<:'i"~,l'P only possible in contexts of
(non-standard) binding and/or'Ca'fitXQI
The tense asymmetry:
Referential null subjects ~r
possible in past and future tense
clauses. They are ruled-out
sent tense clauses.
Hebrew covert pronoun
the subject of numerous studies: E.g.,
Borer (1986, 1989), D
3, 1988), Gutman (1999, 2004), Landau
(2004), Vainikka &
9), Shlonsky (1987, 1990, 1997). The present
contribution reco
work and the data it has unearthed in light
engendered by the minimalist tum of Chomsky
of the conceptu
subsequent work.
(1995: Chapter
* I ,,,,",l¥"iefUl to the organizers of the Workshop on Partial Pro Drop at Cambridge
University;iti4une 2006 for soliciting this paper and encouraging this work, to IdanLandau
andlo tw6$tudia Linguistica reviewers for written comments. Parts of this paper have been
preserit,da't the University of the Aegean in Rhodes and at the University of Siena. I am
grateful to the audiences at these venues for their comments and suggestions.
Studia Linguistica 63(1) 2009, pp. 1~25. © The author 2009. Journal compilation
© The Editorial Board of Studia Linguistica 2009. Published by Blackwell Publishing Ltd.,
9600 Garsington Road, Oxford OX4 2DQ, UK, and 350 Main Street, Malden, MA 02148, USA
2 Dr Shlonsky
1.1. The referentiality asymmetry
The sentences in (I) illustrate three types of non-referential null subjects,
atmospheric, arbitrary quasi-existential and arbitrary generic/quasiuniversal. These sentences are grammatical independently of whether
they are affirmative or negative. The negating particle is 10.
(1) a. (10) qar.
neg cold
'It is (not) cold.'
b. (10)· dofqim
b-a delet.
(neg) knock + pres-mpl on-the door
'Someone is (not) knocking on the door'
lit.: 'Somebody (/Nobody) is knocking on the door.'
c. (10) ma'arixim
et ha truma
selneg appreciate + pres-mpl acc the contribution 0
'People (don't) appreciate her contribution.'
In present tense negative sentences, such as thos
, there is an
alternative to 10, namely,eyn. (Eyn cannot occur wi
r future tense
verbs.) This negative particle can bear a suffix,
rus', as in (2a), in
which case it resembles an inflected (negative) au
. y. Eyn can also occur
in a bare, non-agreeing form, with the clausal subject on its right. Tbis
option is illustrated in (215) and, apart fro
rief mention at the end of
section 4.1, will not be further discussed'
aper, (but see Shlonsky
1997,2000.)
(2) a. hu eyn-o
dofeq
he neg-3ms knock + pres'He is not knocking on e'doo.
b. eyn hu dofeq
elet.
neg he knock + pre~rfu
e door
'He is not knocking':.\:llt'the"'door.'
The sentences in (3) ar
sharp contrast to th
sentences in (1) an
matical. Their ungrammaticality stands in
mmaticality of the la-negated variants of the
0 the descriptive generalization stated in (4).
(3) a. *eyn-o
neg-3ni~
cold
'It isndfcbld'
b. *eyUSamg.ofqim
ba-delet.
rieg;3pl'!Cnock + pres-mrj\on-the-door
'Sofueone is not knocking on the door'
c. Teyncam ma'arixim
et ha truma
sel-a.
rieg~3pl appreciate + pres-mpl acc the contribution of-3fs
'People don't appreciate her contribution.'
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
Hebrew as a partial null-subject language 3
(4) Non-referential null subjects are illicit as subjects of agreeing eyn.
Section 3 is dedicated to an explanation of (4).
1.2. The person asymmetry
1
~,
The distribution of referential null subjects is split along the lines of
grammatical person: Third person null pronouns, as opposed to first and
second ones, are illicit in the past and future tenses. This split is
exemplified by the contrast between (5a) and (5b) for the past tense and
between (6a) and (6b) for the future tense. l
(5) a. *lamad
study+past-3ms
:
,;
*lamd-da
study+past-3fs
\C:
*lamd-du
j')
t
albanit.
+ Albanian
+
study+past-3pl
'He/she/they studied Albanian.'
b.
lamad-ti
T
study+past-l s
,.1.·
lamad-ta
+,.
study+past-2ms
J
lamad-t
J
/
study+past-2fs
_:' "
lamad-nu
\
Ii
study+pas -~1"
.~~
t
J
J
'I1you-m/you-f/we/you-pl studied Albanian.'
LThe glosses provided for the agreement affixes here and elsewhere in this paper are only
intuitiv.eajid do not do justice to their true morphological makeup. For an influential
Distributed Morphology account of Hebrew verbal inflection, see Halle (2000).
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
4
Dr Shlon sky
(6) a. *yi-lmad
)
-f.
{
r
3ms-study+fut
*ti-lmad
+·albanit.
3fs-study+fut
fAlba nian
(
I
..[
*yi-lmed-u
3m-study+fut-pl
t,
b.
H\
e-lmad
ls-study+fut
,
ti-lmad
,
I
j
'He/she/they will study Albanian.'
[2
I
j7
')
'j.g
:1
I
,
/
oil
i\
--
2ms-study+fut
j
f
ti-lmed-i
k·lubanit.
2s-study+fut-f
,/<Albanian
ni-lmad
+
+
1pl-study+fut
ti-lmed-u
"
.1:1
I
,
I
t
I
!
"
•
)
2m-study+fut-pl., J
/
.1'
'Ilyou-mlyou-f/we/you-pl will
Sectio n 4 discusses this asymm etry
1.3. The tense asymmetry
Null refere ntial prono uns ar
oss Ie in the presen t tense, as shown in (7):
(7) *lomed
-' -'
II"
study+pres-ms
*lomed-et
I
I
.~
*lomd-im
!
Albanian
j
!
l'
study+pres-mpl
*lomd-ot
sh.lgy+pres-fpl
Jl
'IIyOu-m,flhe/she/we-mpl,fpIlyou-mpl,fpIlthey-mpl,fpl study/s Albanian.'
© The author 2009. Journal compilation © The Editorial Board
of Studia Linguistica 2009.
l
t"
f
Hebrew as a partia l null-subject language
5
The sentences in (8) contrast with those in (7) and show that
in
eyn-negated environments, referential pro is acceptable, though
solely
when eyn is inflected for first or second person (compare (8a) and (8b).)
Thus, the first/second vs. third person split, manifested in the past/future
tense paradigm reappears in present tense sentences with inflected eyn.
(8) a.
eyn-eni
lomed
~l~
neg-Is
study+pres-ms
·r
eyn-xa
lomed
,.
neg-2ms study+pres-ms I
eyn-ex
lomed-et
)Alba nit
neg-3fs study+pres-fs
Albanian
,
eyn-enu
lomd-im
,I
neg-Ipl
study+pres-pl .1
eyn-xem lomd-im
1
neg-2pl
study+pres-pl
'I am Iyou-m are not studying Albanian'
b. *eyn-o
lomed
)
I
neg-3ms study+pres-ms
*eyn-a
lomed-et
neg-3fs
*eyn-arn
t
study+pres-fs
l
lomd-im
!
!
I
i
neg-3pl
J
The following figure s
referential null sub'
I
marlZes the distribution of referential and nonebrew.
2/@cenSinglP!Il.. i;
~
The researcp.pfogram initiated by Lectures of Government and Bindin
g,
(Chomskyi;] 98 I) and, in particular, the leading idea that differences
amon g gfl\lJliuafs are traceable to different settings of a shared pool
of
param ~ters ,spaw neda large number of propo sals for
the grammatical
licens ingb[ pro, (e.g., Adam s 1987, Alexiadou & Anag nosto poulo u 1998,
BOrer 1986, 1989, Gilligan 1987, Huan g 1984, Jaeggli & Safir 1989a
,
Jaegg li& Safir 1989b, Kenstowicz 1989, McCloskey and Hale 1983,
© The author 2009. Journal compilation © The Editorial Board
of Studia Linguistica 2009.
6 Dr Shlonsky
The distribution of Hebrew Null subjects
referential
Non-referential
~
past/future tense;l present tense
eyn-negated
eyn-negated
other contexts
~
1 & 2 person 3 person all persons
I
OK
*
*
*
Figure 1. The distribution of Hebrew Null subjects
Platzack 1987, Rizzi 1982, 1986, Roberge 1990, RQ
1978, Vainikka & Levy 1999, a.o.) These propOs~~
vided a formal
characterization of the observed (though treq~~ntly partia~ cross
linguistic correlation between the degree tQ',.Vhicli''person and number
features are discretely represented in a ve~barcQnjugation paradigm and
the occurrence and distribution of cove.r;Qf\g"lninal subjects?
The Minimalist Program introdu
£phceptual innovation with
regard to the syntactic role of ver
eatures and the process of
agreement. Chomsky has frequen!'PQIll d out that while the [number]
value of subject nominal phrase
d the [person] value of pronouns are
relevant for their inte~retatio
0tUpare: 'We invaded Iraq' with 'They
invaded Iraq',) these featur
ot serve any inte~retative role when
occurring as component
a1 inflectional morphology, i.e., as
components of T. FrQ
erspective of semantic inte~retation,
agreement in number a'
on of a subject and a verb is superfluous,
since it does not cont
an interpretation which cannot be directly
gleaned from the si:(bjecFHself. Clearly, this kind of redundancy cannot
be tolerated in f\f\',~Qptif\1al1y designed' system.
These considhatiops lead Chomsky to the idea - which underpins
much curref\tre~~<l,rch including the present paper - that agreement plays
a front stage role in the syntactic computation which renders features
2 The",,~xist."ilce of such a correlat.ion in inflect.ing languages is hardly in doubt.. What. is
more pi6bleiriatiC is it.s non-uniformit.y. See e.g., Miiller (2006) for a recent. discussion. A
differenfapproach a1t.ogether seems t.o be called for t.o handle null subject.s in languages
lacking verbal phi feat.ures, such as Chinese (see Huang 1984). Jaeggli & Safir (l989b)
fonIlulat.e,a paramet.er int.ended t.o handle both agreement. and agreemeut-Iess pro drop
syst.ems,as does Speas (2006). See also Holmberg (2005).
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
oj
Hebrew as a partial null-subject language 7
visible to the external interpretive modules and drives displacement.
Without going into the details, which I assume the reader to be familiar
with, take agreement in phi-features to be established through the process
of Agree, in which the phi features on the subject are probed by T and
come to license, or value the inherently uninterpretable phi features of T.
T then contains no unvalued features and meets the condition of Full
Interpretation. Agreement in phi features is, additionally, a necessary
condition for the valuation of the Case feature on the subject nominal.
The implication of taking the phi-features on T to be uninterpretable is, as Holmberg (2005) cogently points out, that these features
cannot in turn be used to license or identify the features oLa null
subject (e.g., in the sense of Rizzi 1986.) Indeed, the
'ninlalist
perspective on T's phi features has the consequence of re
the
licensing relationship: One should now ask how the featu
Tare
valued by a null subject and not how T licenses or assi
es to
the features of the (null) subject.
With this reasoning in the background and, assumin
the implementation of Agree as valuation, (Choms~y"~,<",? 004), let us
consider a formal implementation of the distriblitig,!i of Hebrew null
subjects, starting with those that occur in sente
with eyn.
3. Non-referential subjects
ith eyn, or T eym carries
As a first step, assume that the T head as~
a complete but unvalued set of phi ft;?illjlTes;"'signaled by the agreement
suffixes on eyn. The phi set of T eyn is cl1l<m~,tized in (9).3 T eyn probes for a
goal bearing a set of intejrprt:
e phi-features and valuation is
implemented by assigning a va
!ling-in the person and number
4
slots in T's phi-set matrix.
(9) Phi-set matrix of Teyn
[N(number)
Non-referentia!,pfoq9UIls are impersonal by definition. This means that
such pronouns jack '~' specification for [person], although they possess a
lexical numper feature - plural for arbitrary pro and singular for other
3 AlthqughJIlse the terms 'phi set' and 'phi matrix', I am not committed to the view that
phi feattifesCOristitute a 'bundle' in Chomsky's (2001, 2004) sense. For the purposes of this
paper,phifeatures should probably be considered as unbundled.
',It is plausible that T is also endowed with a gender feature, which marks second and
thirdpeJ;soh inflection in Hebrew. Since gender does not playa role in the present discus. sian, I ignore it.
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
8 Ur Shlonsky
occurrences. Consequently, probing by T eyn for a value for [person]
doesn't turn up anything. The sentences in (3) fail Full Interpretation,
since they contain an unvalued feature on T ey".
The failure to value [person] on T eyn in (3) should not be attributed to
the phonetically-null status of the non-referential subject but rather, to its
non-referential status. Hebrew has an impersonal 'it'-like pronoun, ze, the
subject in (lOa) (see Hazout 1994 and Sichel 2001.) Despite being overt,
this pronoun is incompatible with eyn, as (lOb) shows, since it lacks a
[person] specification.
(l0) a.
ze
It
'It
b. *ze
It
'It
kase
le-daber rusit.
(neg)
difficult to-speak Russian
is(n't) difficult to speak Russian.'
eyn-o
kase
le-daber rusit.
neg-3ms difficult to-speak Russian
isn't difficult to speak Russian.'
(10)
Covert non-referential subjects are fine with present t~ ~~v~r s without
eyn, as in (I). They are also perfect with past andutUrefense verbs, as
shown in (ll).
(ll) a. (10) haya
qar
(neg) be+past cold
'It was (not) cold'
b. (10) dafqu
b-a delet
(neg) knock + past-mPr\on-the
'Someone did (not) knock 0
c. (10) he'erixu
sel-a
(neg) appreciate + past-m
c,. the contribution of-3f,{
'People (didn't) appreciitteh~r cbntribution'
Ifpast/future Tin Hebrew ha,fl the§a,)i1e phi-feature matrix as T eyn (see (9)),
then sentences containing
nominal subject bereft of a person feature,
namely, a non-referenti
uld fail Full Interpretation since such
sentences would not co
egitimate goal for Agree in person by T.
To resolve this
, let us assume that the T of past and
future tense verbs
verbs are discussed separately in section
5 - contains a
set matrix, lacking the slot for [person].
(12) Phi-set maWix
i
r
[N(number)_J
Characterized in this fashion, past/future T is compatible with both
referential and non-referential subjects since it does not probe for [person].
© The author 2009. Journal compilation © The Editoriaf Board of Studia Linguistica 2009.
Hebrew as a partial null-subject language 9
This conclusion raises the immediate question of how to account for
the person split with past/future T and, from a morphosyntactic
angle, how to characterize the content of the verbal affixes which
code for person features. I approach this problem from the perspective of
eyn and then extend the discussion to the past and future tense forms.
4. Referential subjects - Person
.!
Null subjects ofagreeing eyn sentences can only have first and second person
reference, as illustrated by the contrast between the sentences in (8a) and (8b).
This pattern reproduces the person split which characteriz~s null
subjects with past and future tense verbs in (5) and (6). Contjriuwg to
nifests
assume that Teyn's phi-feature matrix is as in (9), that is, that i
a complete but unvalued set of phi features, the observatios t6 be
that among null pronouns, only first and second perses are
legitimate goals for phicprobing by T.
I suggest an interpretation of this observati
terms of
Kayne's (2000: §9.11) idea that covert pronouns can
'rd person
and its corrolary, that "an agreement suffix havi
roperties of a
pronoun can only be first or second person." (
00: 176.)
4.1. First and second person
Concretely, I take the first and seco
n inflectional endings
ination of prefixes and
(suffixes with eyn and past tense ver
suffixes with future tense verbs; see
(6)), to be the morphological
cts a Speech Act Participant
correlates of a syntactic head w .
06) terminology.
Phrase, or SapP, to employ Bianc
As Bianchi notes, first and
ond person pronouns are discoursec!etermined by the changing discourse
dependent, in that "their refer~
roles [... J Third person pr0!tcltj ,Instead, are not discourse-dependent in
this sense." (p. 2034) This~Fl1Q§J!1e familiar idea that only the participants
in the speech act - th~,'§p~ak~r and the addressee, represented by first
and second person -!1~v'l"~rue grammatical person (cf. Benveniste 1956).
I suggest a structltraliilJ,t¢rpretation of this idea: Sentences containing
first or second pers5q"§ubjects contain the category Sapp. 5
SapP can be"th8light'of as a particular species of Rizzi's (2005, 2006)
SubjectP, thaf'i§, a,¢riterial position encoding discourse-relevant infor5 Bianchi's'S"vP,a cartographically-situated adaptation of Harley and Ritter,,(2002)
schema of thefeatuf" hierarchy of pronouns, is associated with all persons, as is war;anted
by the HaJjllP,(lata which she studies. In Hebrew, SapP is restricted to first and second
person proriotins: See Tenny (2006) for a recent discussion in a different but related context.
It islikel),that in some languages, first and second person are associated with distinct
functional projections. French might be such a language; see De Crousaz & Shlonsky (2003;
426--428). Since there is no evidence for such a split in Hebrew, I will take SapP to be
associated with both speech act participants.
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
"I"
10 Ur Shlonsky
mation, in this case, discourse roles, (see Cardinaletti 2004 and Rizzi &
Shlonsky 2007.) Concretely, take SapP to be projected "l3,m"l below C
and above TP, as in (13).
(13) b ... C ... [SapP' .. Sap ., '[TP'" T ...J]]
·'"1
SapO is filled by a pronominal clitic, moved from the position of the
thematic subject. Cliticization is implemented by moving the pronominal
subject to Sapo, along the lines suggested by Belletti (1999) for object clitics
in Romance, and by subsequently moving the verbal complex in TP&F or
T evn and adjoining it to the cIitic head, (seee.g., Shlonsky 2004). Sap~can be
coilstrued as containing an acc (head) feature, which attracts 'l1~&F'and .
T eyn ' Schematically, the DP containing the clitic is first moveq~g;~Pffif/T,
then the clitic head is extracted and adjoined to Sapo. Finally, .0 is ad'oined
to Sapo (see a related discussion in Cardinaletti & Repetti2
The cIitic in Sapo can be doubled by an overt subject p
, III which
case both the clitic and its double have their source in
ex DP (as
in Belletti 2005, Uriagereka 1995), but given the
the subject
pronoun can be separated from the verbal comp
an adverb or a
parenthetical expression, as shown in (14), it is .
at the landing site
of the overt pronominal subject is actually highe
n Spec/Sapo. This is
not surprising, since there are likely to be several Ifferent positions for
preverbal subjects in the clause, (Cardinal
04 and Shlonsky.(2000n
(14) a. ani behexlet lamad-ti
1 certainly study + past-Is
'I certainly studied Albani.'f
b. ata be'emet lamad-ta
albanit
you really study + pa·st~2j11sAlbanian
'You really studied
.ari1
To summarize, Hebrew la
yen first and second person pronouns. 6
Rather, it possesses ave
f clitics, which may be doubled by full
pronouns. When the
s not doubled, there is an illusion that
there is a covert prgng?ssociated with first or second person inflection but in fact this in etion is an incorporated Sap head. 7
Typologically •. .\lrii"~l?ted languages provide evidence for a distinct
syntactic positiOn fq;rthe Sap persons. Poletto (2000), for example,
discusses m?)J.ife~t?ti6ns of the structural distinction between the Sap
persons andihirdperson in many Veneto dialects with respect to the
6 Motepte8isely, it lacks covert first and second pro subjects since, as Landau points out,
(pers6nalcommunication), controlled first and second person PRO is possible in Hebrew, as
in fOnglish,a non-null subject language.
7 FIebre¥ is different from Finnish, where, according to Holmberg (2005), 1 & 2 person
null subject sentences have a filled Spec/IP, satisfying the EPP.
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
"t
Hebrew as a partial null-subject language
11
positioning of subject ditics, as do Manzini & Savoia (2005) who
consider an even larger spectrum of North Italian dialects. Rice's (2000)
discussion of Athapaskan agreement prefixes goes in the same direction
_(see also Linn and 'f& a~ Rosen 2003 and references cited therein.).
The distribution of pronominal subjects in Hebrew eyn sentences
provides language-internal evidence for a configurational person split.
Shlonsky (2000) discusses the pattern illustrated in (15a,b), which points
t6 the cartographic generalization in (16).8
(15) . a.
??ani
rodef
I
.
seek-ms .1'
??ata
you-m
rodef
seek-ms
??at
rodefet
you-f
seek-fs
hu
rodef
eyn
he
seek-ms
neg
hi
rodefet
.1
she
seek-fs
J
??anaxnu rodfim
\
I
we
i
J
I
??atem
I
you-pI
t
I
I
salo
[
I
I
J
hem
they
'I1you-m/yo
'I1you-m/Y9
seeker/pee
~r
s e/we/you-pllthey do not seek peace'
bhe/we/you-pllthey am/are/is not a peace
ekers.'
8 Note the remarkable s
ilarity between (I5a,b) and the position of subject clities relative
to negation in the Veneto dialect of Polesano in (i) and (ii) (from Poletto 2000).
(i) A
no vegno
sci-Is neg come.
'I. atri.,':riof'c'6ining.'
(ii) No Ia
vien
neg scl-3fs comes.
~She i$ not coming.'
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
12 Dr Shlonsky
b. am
eyn-em
I
neg-Is
eyn-xa
you-m
seek-ms
rodef
neg-2ms seek-ms'
at
eyn-ex . rodefet
you-f
neg-2fs
seek-fs
hu
eyn-o
rodef
he
neg-3ms seek-ms
ata
j
,
..
'
rodef
1,
-f.
/(
I
/
'J
I
J
j
r
'r
hi
eyn-a
rodefet
she
neg-3fs
seek-fs
anaxnu
eyn-enu
rodfim
-I
I
we
neg-Ipl
seek-pI
'r
atem
eyn-xem rodfim
you-pI
neg-2pl
seek-pI
i'
-I
hem
eyn-am
rodfim
-j
they
neg-3pI
seek-pI
~
"
/
salom
peace
,
'I1you-m/you-f/he/she/we/you-pl/.tbeyg.o not seek peace'
'I/you-m/you-flhe/she/we/youey::am/are/is not a peace
seeker/peace seekers.'
J
[
t1£16) he lowest position that
erson pronouns is above T ey
below and above T eyn '
occupied by first and second
person pronouns can occur both
SapP is therefore higher th@e,nd the overt Sap persons must move at
least as high as SpecjSap:p,:01r\K~t third persons enjoy a greater freedom:
They may remain low~ thilu"l:p or may raise above it (e.g., to SpecjT or
higher.)/
4.2. Thirdpersg
Turning to c9,yerttpitd person pronouns, we must now explain why these
are incompa.tibleWithboth T eyn and Tp&F. Since only the first of these
two T'sga.s . il.nimvalued person feature, (compare (9) and (12)), we
cannot atffll:>iife this asymmetry to some malfunction in the Agree or
feature1r~lU,ition mechanism. We should seek an explanation in the
propeftiesof the covert third person pronoun itself.
Third person covert pronouns are indeed unacceptable in examples like
(Sa) and (6a). However, they are fine in contexts where an overt
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
./
Hebrew as a partial null-subject language
13
antecedent for the null subject is accessible. Several species of what we
might call referentially-dependent pro can be discerned in Hebrew.
(These were first discussed in Borer 1989).
Landau (2004) studies one such species, restricted to future tense
complements of matrix predicates which select subjunctive complements in
other languages. Landau argues that future tense morphology can
designate subjunctive mood in Hebrew and demonstrates that clauses of
this sort manifest the cluster of properties typically associated with
subjunctive mood. The crucial fact about null subjects in such environments is that they cannot have independent reference but must be
controlled.
Thus, (17a) (Landau's (3b)) shows that the covert subject of 'g9'in the
embedded clause must be controlled by the matrix subject, (inSQi:tttfl,~t to
the overt pronoun.) (17b) demonstrates that the embedded,,:;verl:J" annot
be inflected for past tense when the subject is covert and (17
s that
a factive matrix verb - which does not select for
lIve - IS
incompatible with a null subject in the embedded clau
J
/
h,11.
J~
d
I
(17) a. heml kivu
se heml/2/prol/*2 yelx
They hoped-3pl that they/pro
3-g
J.nmukdam
. '-early
'They hoped that they will go honW~fl,rly.'
b. hem kivu
se hem/*pro.nfl,lxu
habayta
They hoped-3pl that they/pr W'eu,t,,3pl home
.Jmukdam
'learly
",.
'They hoped that they wenfhql:pe early.'
c.;\.em
micta'arim se
*pro yelxu
habayta
They regret-3pl th
pro 3-go + fut-pl home
timukdam
.~early
'They regret t4
will go/.;early.'
,'f
\
Landau diagnoses th
bject in examples like (17) as PRO, not pro,
pointing out som
,~etive similarities between the null subject of
such finite clauses.a ,t4e subject of non-finite control complements. His
analysis is emQ~ddedin a theory of Control which is geared to explain,
inter alia, th~,.alt~l]~tion of PRO, overt pronouns and lexical subjects in
environments of Control into finite clauses. A detailed evaluation of
Landau'scalSll1us of Control lies beyond the purview of the present
investigatidnQut clearly, if his analysis is valid, then the pattern in (17) is
irrelevari(tOthis study.
Itis,rionetheless, tempting to relate (17a) to other cases of
referentially-dependent third person covert subjects, in particular III
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
14 Dr Shlonsky
contexts of control or binding into adjuncts which are problematic for
his system.
Gutman (1999, 2004) discusses numerous cases which do not fall
under Landau's generalizations. In (18), (Gutman 2004's (14a),) for
example, a covert subject, obligatorily dependent on a nominal in the
matrix clause (the subject, in this case), is grammatical in a past tense
adjunct clause.
J
~
(18) Dani j kibeI
mi-Dafna
ve~Rina
matana yafa
Dani received-3ms From-Dafna and-Rina present pretty-fs
Saxarei se Proj/*2 siyem
et ha doktorat.
I lafter
that
finished-3ms acc the doctorate
'Dani received a fine present from Dafna and Rina af
had
finished his doctorate.'
Such null subjects can also be referentially-depende
(indirect) object, as in (19) (Gutman's 2004 (14b).)
(19) Dafna ve-Rina natn-u le-Dani j matana
Dafna and-Rina gave-3pl to-Dani presel}f"'prefty-fs after
F
J
~~at prolj*2 '1~~~d-3ms :~c ~:e ~~~:~:~,
'Dafna and Rina gave Dani a fine p
doctorate.'
after he had finished his
It might appear that these tempCJ!'
clauses are also in the
subjunctive mood, as they woulgF'b~"
example, in Romance. Null
subjects are also possible in u!'P0se adjuncts, which also manifest
subjunctive in Romance.
,..
J
(20) Dafna ve-Rina
Dafna and-Rina
..rPro1j*2 y-saye
l
3ms-fi
'Dafna and Ri
Dani 1 kedey se
acc Dani in order that
ha doktorat.
acc the doctorate
ssed Dani so that he finish his doctorate.'
Consider, now,
6ntrast between a non-finite adjunct clause in
(2Ia) and afil}it~ one, in (21 b). Covert subjects appear in both
structures l,Juf'theiiinte!'Pretation is different. In the non-finite adjunct,
the subjecfis controlled by the matrix subject: (2Ia) cannot mean that
Dani willwirithe prize. In (2Ib), on the other hand, there is a clear
preferenc¢for' object "control" and it is quite awkward to construe the
senterice",as meaning that Dafna will win the prize.
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
Hebrew as a partial nuil-subject language
15
(21) a. Dafnal nisq-a
et Dani kedey PR01j*2 li-zkot be-pras
Dafna kissed-3fs acc Dani in order
to-win in-prize
'Dafna kissed Dani in order to win a prize.'
et Dani kedey se pr0??1j2 ;/'1'l"\
, b. Dafna 1 nisq-a
Dafna kissed-3fs acc Dani in order that
,fti-zke
y-zke
be-pras
F L3fs-win + fut 3ms-win + fut in-prize
'Dafna kissed Dani so that/she/he win a prize.'
(2la) instantiates run of the mill subject Control and the null suqject in
the embedded clauses is PRO. Object Control is blocked in thisijxawple,
presumably because the direct object does not c-command the 'unct
clause (but see note 10.).
Now consider (2lb) and the flip from subject to object conro. tis not
obvious why a finite tense as opposed to a non-finite one~h'OtI,lifengender
even an optional flip from subject to object control. O~e po§sibility, of
course, is that the null subject of finite clauses hase''Option of being
realized as pro, not PRO.
We thus have at least a plausibility arguI11ij111
surmising that,
Independently of the question of grammatica.tJilood of the adjunct
clause, (21 b) does not involve Control into a finite <;lause of the Landau
type.
As for the relevance of mood, (22) ,hC,W'
the mood specification of the adjunct clause is orthggul1"""""u the manifestation of
referentially-dependent covert
the Gutman variety. In
Romance, reason/cause adjuncts
the indicative and not in
the subjunctive mood.
(22) Dafna nisq-a
et
Dani 1 biglal se pro 1j*2 slyem
Dafna kissed-3fs acc Dani because that
finished-3ms
~r et
ha doktorat.
~acc the doctor
'Dafna kissed Da.n
The referential qijP
ijllcy of the covert third person pronoun poses two
distinct questiops, ria.mely, how the pronoun gets its interpretation and
why it must qe rijferentially-dependent.
ConsiderifirsLthe how question. Gutman's examples demonstrate that
c-commandis ,not a necessary condition for licensing referentially-
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
~I
,
16 Dr Shlonsky
dependent null subjects so that the covert subjects under discussion are
not bound in any standard syntactic sense. 9 ,10
Following Ariel (1990, 2001), Gutman views the accessibility of the
antecedent as a necessary condition for the grammaticality of sentences
with covert subjects. One of the factors that determine accessibility is the
discourse salience of the antecedent. Consider, in this context, the
paradigm in (23).
J
.
j~,
(23) a. sam'anu se lifney svu'ayim Dani] xakar
et
heard-lpl that before two-weeks Dani interrogated-3ms acc
(Dafna, ve-se
axarey xames dakot proj maca
.H I Dafna and-that after five minutes
'Iota asema.
lher guilty-fs
'We heard that two weeks ago, Dani interrogate
that after five minutes he found her guilty.'
9 Borer (1989) held that the binding of these covert PEPl)guns was anaphoric and provided
sentences purporting to show that they could not hgve split, antecedents. Ariel (1990) and.
Gutman (2004) provide several counterexamples whie
Borer's claim into question. The
~cited by Gutman.
following example is from Ariel (1990: chapter
et
Shimonj al
Noga; bikra
Noga criticized-3fs acc
Shimon
kset\ pr'1,
li-yrusalayim.
whel],"\',
to-Jerusalem
'Noga criticized Shimon on his
to Jerusalem."
10 It is actually not entirely cl
material, see e.g., Bianchi (19
manifestation of robust Co
adjunct clause (some spegl<'
'cle when they drove
junct clauses are not c-commanded by VP-internal
osition of temporal adjuncts. In (my) Hebrew, the
!fects suggesKthat a direct object c-commands into an
"pts (i), though).
(i) *Dani pagas otai':I!fiiey se Rina, nihyta mefursemet.
her'before that Rina became famous-fs
Dani met
Dani met her.before Rina became famous."
(ii) *Rina! harga':'::otol Qiiglal se Danl] nihy/a mefursam.
()
'~i:: kirigd'hi~~~~~:~:u~::ih::c::efa~:~,;,e famous-ms
Ou the oth"rlIand"adjunct clauses can harbor parasitic gaps, which are presumably not
c-commanded,pythe genuine wh gap.
(iii) eizesefer ',' tiyakta
()
axarey se
lwhiClIboOk filed-2ms after
(
qarata?
that read-2ms
r,'Which book did you file after you read?'
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
J/
,
Hebrew as a partial null-subject language
J
17
b. sam'anu se 1ifney svu'ayim Dafna]
heard-1p1 that before two-weeks Dafna
r~i:texkera
a1-yedey Dani, ve-se axarey
Dani and-that after
...\.!nterrogated + pass-3fs by
I xames dakot pro] nimc'ea
asema.
tfive
minutes
found + passive-3fs gui1ty-fs
'We heard that two weeks ago, Dafna was interrogated by
Dani, and that after five minutes she was found guilty.'
c. *sam'anu se lifney svu'ayim Dafnaj
heard-1p1 that before two-weeks Dafna
fnexkera
a1-yedey Dani,
Dani
linterrogated + pass~3fs by
. Ive-se axarey xames dakot pro] maca
'land-that after @ve minutes
found-3ms IT
Iota asema. .
\.b.er guilty-fs
'We heard that two weeks ago, Dafna was inte
and that after five he found her guilty.'
In both (23a) and (23b), the covert subject i~ Co ferentia1 with the
matrix subject. The ungrammaticality of (23c);might be attributed to
the fact that in a passive construction, the demoted agent is not
sufficiently salient and cannot serve as a ccessib1e antecedent for the
covert subject in the second conjunc not~, in passing, that the
antecedents in the acceptable sentences
, J do not c-command the
null subject.)!
Although it remains to be dete
antecedent accessibility is
grammatically encoded, we can
It more concrete proposal with
respect to the implementation
9P standard binding/control, henceforth NSB/C. The idea is tha
involves the assignment of a person
feature to the covert subje
er 1989.) A pronoun in possession of
a person feature can b
··reted as a referential pronoun (see
Longobardi 1994, a.o.) nce e phi set of the covert subject is modified
- in this case, by th. qItion of a feature - a parallel modification is
implemented on T
i::.#1atrix. Feature synchronization is a natural
extension of thea.nism of agreement: Once the probe-goal relation
establishes an a.gfi~e#1~nt link, any modification of one of the terms of the
agreement re1atipn ell'tails a parallel modification in the other terms. (The
relevance ofthisc:reC:fuirement will become apparent in section 5.) Thus,
NSB/C assigns ?person feature to pro and this feature is then copied
onto T. ll
LetU$hbWturn to the question of why covert third person pronouns
musfbereferentially"dependent. An overt third person pronoun can refer
make
11 NSB!C assigns afeature to pro, not a value for a feature. Feature synchronization, in
turn, involves copying the person feature to T.
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
18 Dr Shlonsky
freely (as well as be bound) whereas a covert pronoun is referentiallydependent. This difference between the two seems to preclude an analysis
in which null third person pronouns are overt pronouns minus phonetic
content. We can also reject the view that Hebrew simply lacks null third
person pronouns. If that were the case, it would be puzzling why they
suddenly come into being in the presence of an appropriate antecedent.
(There is also a more general question regarding the fate of the subject
theta role if there is no constituent to bear it.)
Suppose that all covert "third person" pronouns possess a number
feature but lack any specification for [person]. Such pronouns are
therefore different from overt pronouns, which are endowed witl}"both a
person and a number feature. Hebrew thus has a single lexicalprQ;:one
that lacks a person feature. There is no lexical difference be
nonreferential and referential covert subjects; both are 'reduced'1.l1.ls. 12
Such covert pronouns can be successfully probed by Tl'
this T
head only has a number feature in its phi set. Since th
un lacks a
specification for person it can only be interpreted impe
, I.e., nonreferentially. Lacking an intrinsic person feature,
ances of pro
cannot, however, refer. They rely on NSBjC for
ss!gnment of this
feature, which renders them capable of referenc;,¢:
To conclude this section, consider a case which,prings out more clearly
the inherently non-referential status of null third person pronouns. What
I have in mind is pronoun obviation in slJ:bjuDJFJive contexts.
In addition to the controlled subjunctjyes}~xe#iplified in (17), there is a
small class of verbs in Hebrew which (lmbre'Romance subjunctives in
imposing non-coreference or obviaf
Pl1iithe subject of their complement clauses. The examples in (24
Laiidau's (2004) (9a,b).
;t
~1
(24) a. Rina! ®acta se hi~l/~ ~j:z:ke
ba-pras.
c
Rina wanted that she
3L win + fut-s
in-the-prize
'Rina~wanted tha
e.j'~:'would win the prize.'
b. Gil! canx
yeda
carfatit.
Gil
needs
3m-know + fut-s French
'Gill! needs
know French.'
Note, now, that
covert, as showg
Landau's (59a).)
(25) a. *Rinaj
Rina
b. *GiI;
Gil
facta
wanted
canx
needs
of the embedded clauses in (24) cannot be
ungrammaticality of (25a,b), (adapted from
se pro tizke
that
3f-win + fut-s
se pro yeda
that
3m-know + fut-s
ba-pras.
in-the-prize
carfatit.
French
1~ Fordstructural implementation of this idea, see Ritter (1995), my own (1997) adaptation "fRitter (1995) and, more recently, Dechaine & Wiltschko (2002) and Holmberg
(2005).
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
Hebrew as a partial null-subject language
19
.Assume a fairly standard view of obviation, namely, that in 'true'
subjunctive clauses, the binding domain relevant for the embedded
subject is extended to the matrix clause (see Quer 2006 for a recent
survey.) Consequently, coreference between the overt subject pronouns in
the embedded clauses in (24a,b) and the respective matrix subjects is
excluded by Condition B. The pronouns must be obviative in order not to
violate the binding condition.
Now consider covert pronouns. If these are assigned features by the
matrix subject via Control or non-standard binding, then they violate
Condition B. If feature assignment is withheld, then they presumably lack
the minimal content necessary for independent (obviative) re(~rence.
Hence, covert pronouns must be barred from the subject po'" on of
embedded subjunctives of this variety.13
5. Referential subjects - Tense
Although compatible with null non-referential pr
(see (1»,
negated by
present tense sentences - with the exception 0
y of partial proeyn - eschew all covert referential subjects. The e
rent reasons for
drop developed in this paper leads us to exr:i.~ct
third person null
the absence of first and second person
subjects.
I assume that the absence of first anqsecq):ld person clitic subjects
with present tense verbs is due to the~b&YPecY'·of an appropriate host
for the clitic. Recall from section .:'1'tl.",that the cliticization process
involves two stages. First, the clitiqj§fj,~jgahed to Sapo and then,
is
adjoined to Sapo. It is the moverp6l±t o{''f to Sap that provides a host
for the clitic. Suppose that Sa,pg;:s 0@c head feature is restricted to
T eyn and Tp&F' TPres is not at cted'" Hence, it cannot host the clitic in
Sapo. .
..,m.
Non-clitic (or full) first ap"d . & gnd pronouns are attracted to Spec/Sap.
Their full acceptability iy!+ftense sentences shows that they are not
obligatorily associateq
llclitic: They appea.r with a clitic in the
.
company of TP&F hIJJ,p aPe sentences with TPres'
Let us now turn 0 the'xeferentially-dependent pro discussed in section
4.2. This kind oLp!'
unacceptable in present tense sentences, compare
e.g., (19) (repel\ted low) and (26).
r
13 The question remains. of course. how to express the differences between Romance-style
subjunctives, as in (24) and Balkan-type subjunctives, as in (17). The former, but not the
latter, give rise to obviation effects. In Landau's system, the difference boils down to the
presence"s. the absence of Agr in C.
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
20 Ur Shlonsky
]
(19) Dafna ve-Rina natn-u le-Dani l matana yafa
axarei
Dafna and-Rina gave-3pl to-Dani present pretty-fs after
Jse pro 1/*2 siyem
et ha doktorat.
I"[that
finished-3ms acc the doctorate
'Dafna and Rina gave Dani a fine present after he had finished his
doctorate.'
(26) *Dafna ve-Rina
notnot le-Dani matana yafa
kol
Dafna and-Rina give-3pl to-Dani present pretty-fs every
,.fpa'am se pro mefarsem ma'amar xadas
, ltime
that
publish-3m article
new
'Dafna and Rina give Dani a fine present each time he
a new article.'
A covert pronoun differs from an overt one in lacking [person
it can
s that
receive through NSB/C. The contrast between (19) anq,(26
NSB/C can apply into clauses with Tp&F but not with T Prd~i, Int4~ following
paragraphs; I argue that it is not NSB/C itselfwhichi,$'b,!,g'ckedwith Tpres ,
but that this operation entails a computational!'ir~gkdown when the
features of modified pro and copied onto the phi$e~"pft1),e agreeing head.
I suggest that we attribute the relevant diffet~nce between Tp&F and
TPres to properties of their phi sets and the means by which the agreement
mechanism functions in present tense cla1.!§es,
The crucial fact here is that presen
e verbs in Hebrew are
participles, not only morphologically
ntactically as well (see
Shlonsky 1997, Siloni 1995.) In (f1g)1 ~he participle appears in a
compound tense structure, under,g~;,auxlliary.14 Exactly the same form
appears in the present tense sent~pceiIJ,(27b). The simplest assumption
here is that (27b) contains a
~;tically unrealized TPres.
The agreement features bqrfhe participial verb in (27b) are the
participial agreement featl!res
er] and [number], not the ([number]
and [person]) agreementli'
s which appear on, say, Teyn- The
presence of [gender] aIJ,,<f1Q
ence of [person] features on the Hebrew
participle are not argitrfl,ty acts about this motphological form, but
derive from its beiIJ,g~::p,9jTIinal category, akin to nouns and adjectives.
The Hebrew participJ~ lacks a specification for [person] not because a
[person] slot happensfobe lexically absent from its phi set (as is arguably·
the case of Tp&F), !'itIt because there is a conflict between its nominal
nature and.,the pOssession of such a slot.
(27) a. Dani haya
loves
sveder.
Dani be + past-3ms wears + pres-ms sweater
'Dani used to wear a sweater.'
]4
inte,rpnetation of such auxiliary-participle structures is habitual, not perfective.
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
Hebrew as a partial null-subject language 21
b. Dani loves
sveder.
Dani wears + pres-ms sweater
'Dani is wearing a sweater.'
c. Danieyn-o
loves
sveder.
Dani neg-3ms wears + pres-ms sweater
'Dani is not wearing a sweater.'
While there are two phi sets in (27a) and in an eyn sentence such as (27c), a
participial one and a T-related on~ tllris res1"oot, T pres lacks a phi set
altogether; the phi features that appear in (27b) are contributed by the
participle.
One should now ask where the (nominative) Case feature is found in
present tense sentences like (27b). The choice is limited to
the
participial head. The latter is an unlikely source for a
re,
because if the participial head had a Case feature, then (27a ,
tains both a participle and a finite auxiliary, would have fwo Case
features (as would, by analogy, (27c),) an unwelcome s e of"affairs.
Suppose, then, that the Case feature in (27b) is cOnt' e(Fby TPres but
since this type ofT lacks a phi set, it cannot probelil'lq
e Case on the
subject. Following Borer 1995, Shlonsky 1997 apgu,~S'f4at the participial
verb systematically adjoins to T in Hebrew (mor~ . .ptecisely, it adjoins to a
null auxiliary in T, but this detail can be overlOoked). Using standard
diagnostics, ~. showl that the participial.pr~sent tense verb in (27b)
occupies the same position as the auxiliai'y in'(27a). Adjunction of the
participial verb to Tpres/makes unval
hi(frorn the participle) and
Case (from Tpres) available on a sing)
plex head and the subject can
be successfully targeted. Since th~ipcontributed by the participle
lacks a [person] feature, it has t4e douSequence of barring a referential
pro subject, although both qv~ft ubjects (which are independently
endowed with reference) an
non-referential subjects (which do
not need association with
feature) are both perfectly fine with
present tense verbs. In t
ect, TPres is functionally equivalent to
Tp&F: Both are compatior~,
non-referential null subjects. There is a
major difference bet~e~jj. the two T's however, since only Tp&F is
compatible with aJ:tefet~ll,tial pro under NSB/C.
Tp&F lacks a leXiea)"slof for [person]. This T head can, nonetheless, be
granted a persol1'slofn,li'ider the feature synchronization process discussed in
section 4.2: NSB/C a$signs a person feature to pro which is then copied onto
Tp&F'
TPres lacks a phi set and depends on the participial phi set in order to
enter anagr~ement relation with the subject. Although nothing prohibits
NSB/Cftomoperating into clauses containing a present tense verb and
assigning [person] to the covert subject in such sentences, a problem arises
when it domes to synchronizing the modified phi set of pro with that of
the participle: The phi set of the participle simply cannot carry a person
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
22
Dr Shlonsky
slot, since it is a nominal category. Put succinctly, a participle granted a
person feature would no longer be a participle and would fail to match
the selectional requirements of T. Thus, while NSB/C of covert subjects
of present tense verbs is, in principle, possible, it has computational
consequences leading to ungrammaticality.
Eyn sentences differ from eyn-Iess present tense ones in that T eym as
opposed to Tpresfontains a [person] slot. It is thus predicted that NSB/C
should be possible into eyn sentences. Indeed, (28) patterns with (19) and
not with (26).
.
10 notnot
le-Dani] matana
Dafna and-Rina neg give-3pl
to-Dani present
t-\kse pro eyn-ol/*2 mefarsem ma'amar xadas
rwhen
neg-3ms publish-3m article
new
'Dafna and Rina don't give Dani a present when
doesn't publishes a new article.'
(28) Dafna ve-Rina
6. Summary and conclusion
From the perspective of their phi sets, Hebrew q,,§
ee finite T heads:
One which contains a full set of phi features,:.Qn~"which only contains
number and one which lacks a phi set altogethef.:ii'[;qe first is associated
with a negative auxiliary that occurs in pre§lenttense sentences; we calI it
Teyn- The second, labeled TP&F, chara8~t:iz;;~i past and future tense
sentences and the third, TPres, occurs iq:'pJ;eslept tense sentences.
The covert nulI subject in Hebrew~~,(*is "n intrinsic person specification. To adopt the terms of Ritter
; it: 18 a Num(ber)P (equivalent to
Dechaine & Wiltschkq( 2002 ~
reement between pro and T
predictably fails when pro is tq~'sU1:ljeet of T eym since the latter's person
feature cannot be valued. Pro isi,'poss!~le with TP&F, since this species of T
only possesses a number
tnte:'lri its phi set. However, such a pro
cannot be interpreted refe
, because it lacks a person feature.
However, pro can end
a person feature when it is controlled or
bound. (Although the
erlature of this kind ofreferential dependency
remains elusive.) Ng ard binding/Control involves the assignment
of a person featu.rei."f6pro, rendering it capable of reference. Such
assignment mogifies:tliephi set of pro and requires a parallel modification
of the phi set off. Wh.ile TP&F can be modified by the addition or copying of
the person fleatu.tefrom pro, TPres cannot be modified because it lacks a phi
set. Subjeetagreeinent in present-tense sentences without eyn is participial
agreementwhichcannot host a copy of the [person] feature assigned to pro.
ConseqgentlY, present tense sentences featuring a non-standardly bound/
controIIedcovert subject are judged ungrammatical.
F'jnaI1y, Hebrew lacks covert subjects referring to the first or second
persOn. This is not an accidental gap, but folIows from the fact that the
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
Hebrew as a partial null-subject language
23
speech act participants are syntactically encoded by means of a functional
category. The existence of covert first and second person subjects is an
illusion; they are ciitics which appear on the verbal head.
These are the senses in which Hebrew is a partial pro drop language.
7. References
ADAMS, M. 1987. From Old French to the theory of pro-drop. Natural Language
& Linguistic Theory 5(1): 1-32.
ALEXIADOU, A. & ANAGNOSTOPOULOU, E. 1998. Parameterizing AGR: Word
Order, V-Movement and EPP-Checking. Natural Language & Linguistic Theory
16(3): 491-539.
ARIEL, MIRA 1990. Accessing noun-phrase antecedents. London: RoutIe<lge;
ARIEL, MIRA. 2001. Accessibility theory: An overview. Text repr~sefttation:
Linguistics and psychological aspects, eds. T. Sanders, J. S
eroord &
W. Spooren, 29-87. Amsterdam: John Benjamins.
BELLETTI, ADRIANA. 1999. Italian/Romance elitics: Structure
erivation.
Clitics in the languages of Europe: Empirical approaches to
ge typology,
ed. H. van Riemsdijk, 543-579. Berlin & New York: Mout
ruyter.
BELLETTI, ADRIANA. 2005. Extended doubling and the
ery. Probus
17(1): 1-35.
BENVENISTE, EMILE. 1956. La nature des pronoms. Fo
911. Jakobson: Essays
on the Occasion of his Sixtieth Birthday, eds.
Balle, H. G. Lunt,
H. MacLean & C. H. van Schoonveld, 34-37 (1971
e Hague: Mouton.
BIANCHI, VALENTINA. 1997. On the structural position ottime clauses. Quaderni
dei Laboratorio di Linguistica 11: 66-90.
BIANCHI, VALENTINA. 2006. On dIe syntax of
rguments. Lingua 116(12):
2023-2067.
BORER, HAGIT. 1986. I-subjects. Linguisti
7(3): 375-416.
'BORER, HAGIT. 1989. Anaphoric Agr. Tb
Parameter, ed. O. Jaeggli
& K. J. Safir, 69-110. Dordrecht: I)"U~er.
BORER, HAGIT. 1995. The ups and ,<lQwrrs'Qf Hebrew verb movement. Natural
Language & Linguistic Theory U(3}: 5;27,-606.
CARDINALETTI, ANNA. 2004. Towlj.rds ai,cartography of subject positions. The
structureofCP andIP, ed. L. . '1152£165. New York: Oxford University Press.
CARDINALETTI, ANNA & REI'E
LORI. 2007. Proclitic vs enclitic pronouns in
Northern Italian dialects a
-subject parameter. Stony Brook University
& University of Venice/'
CHOMSKY, N. 1981. Lecture~,"'on Government and Binding. Dordrecht: Foris.
CHOMSKY, N. 1995. r.bi}''Mi&'irnalist Program. Cambridge, MAf: The MIT Press.
CHOMSKY, N. 200L"jPel'i'vatlon by phase. Ken Hale: A life in language, ed.
M. Kenstowicz,h§p;,c;li.mbridge, MA: MIT Press.
CHOMSKY, N. 2004. Beyond explanatory adequacy. Structures and beyond: The
cartography ofjsyntq2tic structure, ed. A. Belletti, 104-131. Oxford: Oxford
UniversityPtess.
DE CRousAZ, I & SHLONSKY, U. 2003. The Distribution of a Subject Clitic
Pronoun in a franco-Provencal Dialect and the Licensing of Pro. Linguistic
Inquiry J4(:l): 413-442.
DEcHAlNE,R-M & WILTSCHKO, M. 2002. Decomposing Pronouns. Linguistic
Inquiry 33(3): 409-442.
DORON, E. 1983. Verbless predicates in Hebrew. Ph.D. dissertation. University of
Texas at Austin.
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
/1
24 Vr Shlonsky
DORON, E. 1988. On'the complementarity of subject and subject-verb agreement.
Agreement in natural language: Approaches, theories, descriptions, eds. M.
Barlow & C. A. Ferguson, 201-218. Stanford University, Center for the Study
of Language and Information.
GILLIGAN, G. 1987. A cross-linguistic approach to the pro-drop parameter. Ph.D,
dissertation. University of Southern California.
GUTMAN, E. 1999. Null subjects: A theory of syntactic and discourse-identification.
PhD. dissertation. University of Delaware.
GUTMAN, E. 2004. Third person null subjects in Hebrew, Finnish and
Rumanian: an accessibility-theoretic account. Journal of Linguistics 40(3):
463-490.
HALLE, M. 2000. Distributed Morphology: Impoverishment and fission. Research
in Afroasiatic grammar, eds. J. Lecarme, J. Lowenstamm & U.
125149. Amsterdam: John Benjamins.
HARLEY, H & RITTER, E. 2002. Person and number in pronouns:
..
geometric analysis. Language 78(3): 482-526.
HAZOUT, 1. 1994. The Hebrew Pronoun ze and the Syntax of Sent<!ntial Sj1bjects.
Lingua 93(4): 265-282.
HOLMBERG, A. 2005. Is there little pro? Evidence from Finnis
36(4): 533-564.
HUANG, C.-T. J. 1984. On the distribution and referenCe
Linguistic Inquiry 15(4): 531-574.
JAEGGLI, O. A. & SAFIR, K. 1989b. The null subject
r and parametric
theory. The Null Subject Parameter, eds. O. A
gl
K. Safir, 1-44.
Dordrecht: Kluwer.
KAYNE, R. S. 2000. A note on clitic doubling in French: Parameters and Universals. 9: 163-184.
.
KENSTOWICZ, M. 1989. The 11ull subject parameter iri):nodern Arabic dialects. The
null subject parameter, eds. O. A. Jaegg~ &".:K,iiSafir, 263-276. Dordrecht:
Kluwer.
LANDAU, 1. 2004. The scale of finiteness'c"a.ria !J1~ calculus of control. Natural
Language & Linguistic Theory 22(4)
1-87'f:
LINN, M. S. & FWI!L"~ ROSEN, S..
.~e functional projections of subject
splits. Texas Linguistic Forum. T..~eRo.le of Agreement in Natural Language 53,
ed. W. E. Griffin, 135-146.
. ,..
.
LONGOBARDI, G. 1994. RefereI\ge a'ri~~roper Names: A Theory of N-Movement
in Syntax and Logical Fofll1.' Mfigu!stic Inquiry 25(4): 609-665.
MANZINI, M. R. & SAVOIA,.[".M;',;2Q05. I dialetti italiani e romanci. M orfosintassi
generativa. Alessandria:, Edi2:igrii'dell'Orso.
MCCLOSKEY, J. & HALE,'Rf"o;198J' On the syntax of person-number inflection in
Modern Irish. Naturcill;cifiguage & Linguistic Theory 1(4): 487-533.
MULLER, G. 2006. P~<i:qtl5P'and impoverishment. Form, structure and grammar.
A festschrift pre§~1J!edi.t0 Gunther Grewendorf on occasion of his 60th birthday,
eds. P. Brandt&E:fuss, 93-115. Berlin: Akademie Verlag.
PLATZACK, C. 19~7. Th¢ Scandinavian languages and the null-subject parameter.
Natural LalJgucl'ge,&Tinguistic Thepry 5(3): 37'f-401.
POLETTO, C.;: 2000.;: The higher functional field: Evidence from Northern Italian
dialects. New york: Oxford University Press.
QUER, . J.2006. Subjunctives. The Blackwell companion to syntax, eds.
M. Evera<!tf,B. Hollebrandse, H. van Riemsdijk & R. Goedemans, 660-684.
Oxford: Blackwell.
RICE, K.2000. Morpheme order and semantic scope. Word formation in the Athapaskan verb. Cambridge: Cambridge University Press.
para.
3.
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.
Hebrew as a partial null-subject language
25
RITTER, E. 1995. On the syntactic category of pronouns and agreement. Natural
Language & Linguistic Theory 13(3): 405-443.
RIZZI, L. 1982. Issues in Italian Syntax. Dordrecht: Foris.
RIZZI, L. 1986. Null objects in Italian and the theory of pro. Linguistic Inquiry
17(3): 501-557.
RIZZI, L. 2005. On some properties of subjects and topics. Proceedings of the XXX
Incontro di Grammatica Generativa, ed. L. Bruge, G. Giusti, N. Munaro, W.
Schweikert & G. Turano, Venezia: Cafoscarina.
RIZZI, L. 2006. On the form of chains: Criterial positions and ECP effects.
Wh-movement: Moving On, eds. L. L. Cheng & N. Corver, 97-134. Cambridge,
MA: MIT Press.
RIZZI, L. & SHLONSKY, U. 2007. Strategies of subject extraction. Interfaces +
Recursion = Language? Chomsky's Minimalism and the View from . SyntaxSemantics, eds. H. M. Gartner & U. Sauerland, 115-160. Berlin: MOiltgn de
Gruyter.
ROBERGE, Y. 1990. The syntactic recoverability of null argume
Mohtreal:
McGill-Queen's University Press.
ROBERTS, 1. G. 1993. Verbs and diachronic syntax: A comparat
hi&iory of
English and French. Dordrecht: Kluwer.
SHLONSKY, U. 1987. Null and displaced subjects. Ph.d. disserta
ssachusetts
Institute of Technology.
SHLONSKY, U. 1990. Pro in Hebrew subject inversion. L'
275.
SHLONSKY, U. 1997. Clause structure and word orde'
and Arabic: An
essay in comparative Semitic syntax. New York:lKord University Press.
SHLONSKY, U. 2000. Subject positions and copu1arConstructions. Interface
strategies, eds. H. Bennis, M. Everaert & E"Rel.l1and, 325-347. Amsterdam:
Royal Netherlands Academy of Arts and Sc;iences,'
SHLONSKY, U. 2004. Enclisis and proelisis.,rhli"·'~!n4'iureof CP and IP, eds. L.
Rizzi, 329-353. New York: Oxford Uni¥er§~ty rress.
SICHEL, 1. 2001. Studies in the syntax prp~~nqJ/hs and features with particular
reference to Hebrew. Ph.d dissertatiqil);Cit)iUniversity of New York.
SILONI, T. 1995. On participial relatives"a!lg complementizer DO: A case study
in Hebrew and French. Natura GYjguage & Linguistic Theory 13(3): 445487.
SPEAS, M. 2006. Economy, agr~et,.ahd the representation of null arguments.
Arguments and agreement,
ckema, P. Brandt, M. Schoorlemmer &
F. Weerman, 35-75. OXD
rd University Press.
TARALDSEN, K. T. 1978.0
, vacuous application and the that-trace filter.
Bloomington: Indiana!)!versIty Linguistics Club.
TENNY, C. 2006. Ev~get;\!iality, experiencers, and the syntax of sentience in
Japanese. Journal.pf~GstAsian Linguistics 15(3): 245-288.
URIAGEREKA, J. 122?;:'~spects of the syntax of elitic placement in Western
Romance. Lin/fuisilc!r(quiry 26: 79-123.
VAlNIKKA, A. &&EVY, y.. 1999. Empty Subjects in Finnish and Hebrew. Natural
Language &rLiftguisHc Theory 17(3): 613-671.
Ur'Shlonsky
Departement de linguistique
Faculte des Lettres
Universite de Genevel
1211-CH Geneve 4
Ur.shlonsky@~!unige.ch
© The author 2009. Journal compilation © The Editorial Board of Studia Linguistica 2009.