Вы находитесь на странице: 1из 18

Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.

Pro-drop and Theories of pro in the Minimalist Program

Part 2: Pronoun Deletion Analyses of Null Subjects and
Partial, Discourse and Semi pro-drop
Pilar Barbosa*
Universidade do Minho


This article reviews the recent theories of pro-drop that explore the hypothesis that pro is a pronoun
that is deleted in Phonetic Form (Holmberg 2005; Roberts 2010b). Since most of the empirical
arguments given in favor of this hypothesis come from the partial null-subject languages, we
discuss the distinctive properties of this set of languages as opposed to the (rich agreement)
consistent null-subject languages. The pattern of pro-drop found in the partial null-subject
languages has features in common with discourse pro-drop, which is found in languages that lack
agreement (Huang 1984). Among the analyses that have been proposed in the literature on
discourse pro-drop is the hypothesis that it reduces to null-NP anaphora (Tomioka 2003). This
hypothesis relates discourse pro-drop with the availability of bare NPs in argument position. Here,
the possibility of extending the null-NP anaphora analysis to the partial null-subject languages is
considered. The article ends with a discussion of yet a fourth type of pro-drop language, the semi
pro-drop languages, in the light of recent developments concerning the Extended Projection
Principle (Biberauer 2010; Wurmbrand 2006).

1. Introduction
In Part 1 of this article, we reviewed the classic Government and Binding theory of pro,
according to which pro is an inherently unspecified nominal whose features are inherited
from Infl. We mentioned Holmbergs (2005) observation that such a theory is incompatible with the Minimalist Program as outlined in Chomsky (1995) and subsequent works,
where the u-features in T (=Infl) are assumed to be uninterpretable and thus not specified for a particular value. Hence, in this framework, it is not possible for an inherently
unspecified nominal to inherit its features from Agr.
Holmberg observes that there are two possible alternative hypotheses regarding a
theory of pro within the MP.

Hypothesis A: In null-subject languages, the u-features of T are interpretable:

Agr is a referential, definite pronoun phonologically expressed as an affix.
Hypothesis B: The null subject (henceforth NS) is specified for interpretable
u-features, values the uninterpretable features in Agr, and moves to Spec, TP,
just like any other subject. That pro is silent is thus a PF matter.

In Part 1, we discussed the proposals that pursue Hypothesis A as applied to the consistent (rich agreement) null-subject languages (NSLs). In this article, we will review the
analyses that explore Hypothesis B. Since most of the empirical arguments given in favor
of this hypothesis come from the partial NSLs, we discuss the distinctive properties of
2011 The Author
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

572 Pilar Barbosa

pro-drop in this set of languages (Section 2). This kind of pro-drop has features in common with yet a third kind, labeled discourse pro-drop, which is found in languages that lack
agreement inflection. These are discussed in Section 3. Among the analyses that have
been proposed in the literature on discourse pro-drop is the hypothesis that it reduces to
null-NP anaphora (Tomioka 2003). This hypothesis relates discourse pro-drop with the
availability of bare NPs in argument position. In Section 4 the possibility of extending
the null-NP anaphora analysis to the partial NSLs is considered. Section 5 discusses the
languages commonly referred to in the literature as semi pro-drop languages in the context
of recent developments concerning the nature of the Extended Projection Principle
(EPP), i.e., the requirement that Spec-IP be filled by a subject (Chomsky 1981).
2. The Partial Null-Subject Languages and the Deletion Analysis of Null Subjects


In his assessment of the different predictions made by the two types of approach to NSs
sketched in the introduction, Holmberg (2005) observes that the two hypotheses make
different predictions in the case of a language that has overt expletives and referential
NSs. Under Hypothesis B, pro occupies Spec-TP; therefore, this hypothesis predicts that
no expletive pronoun, overt or null should occur with a NS. Hypothesis A, on the other
hand, makes no such prediction given that it claims that pro doesnt raise to Spec TP due
to the pronominal properties of the head bearing the agreement features. Thus, if we find
a language that has overt expletives and referential NSs, and if the expletive cannot
co-occur with a referential NS, Hypothesis B is favored.
Holmberg argues that Finnish is such a language and concludes that Hypothesis B is
right for Finnish. Given that the null pronominal element behaves like an overt pronoun
for the purposes of the EPP, Holmberg (2005:538) suggests that the Finnish first and
second person NS is a pronoun that is not pronounced, i.e., a deleted pronoun.1
Even though first and second person subjects can be dropped in any environment in
Finnish, this language has a restricted pattern of third person subject drop. Other languages that have been described as having a similar behavior are Brazilian Portuguese
(BP), Marathi, Russian, and Hebrew. In the next section, we give an overview of the
most salient set of properties of these languages.


Some languages, such as Finnish, BP, Marathi as well as Hebrew, have systematic NSs,
but their pattern of distribution differs from that of the consistent NSLs in two ways: (i)
the NS is optional in some contexts in which it is mandatory in a consistent NSL; (ii)
the NS is excluded in many contexts in which it is possible in a consistent NSL. These
two facts can be illustrated by comparing the European and Brazilian varieties of Portuguese. Consider the following Portuguese examples:
(2) a. O Joao
the Joao
John said
b. [O Joao]
the Joao

disse que ele comprou um computador.

said that he bought
that he bought a computer
disse que [_] comprou um computador.
said that

2011 The Author

Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

Analyses of Null Subjects and Partial, Discourse and Semi pro-drop 573

In the European variety of Portuguese (EP), the NS option (2b) must be used when the
embedded subject takes the matrix subject as an antecedent. Unless it is strongly focalized,
an embedded overt pronoun in examples such as (2a) in EP is preferably interpreted as
non-coreferential with the matrix subject and signals topic switch. This phenomenon,
which is a characteristic feature of the consistent NSLs, came to be known in the literature as the Avoid Pronoun Principle.
In BP, by contrast, the overt pronoun in (2a) may be co-referent with the matrix subject; in fact, both options (2a,b) are available in this language whenever the embedded
subject is co-referent with the matrix subject. Thus, BP lacks the Avoid Pronoun Principle. The same observation holds for Finnish, Marathi (Holmberg et al. 2009), Hebrew
(Borer 1989), and Russian (Franks 1995).
Now consider a situation in which the embedded subject refers to an entity other than
John, who has been introduced in discourse. In EP (2b) is fine in such a context. In colloquial BP it is arguably not (cf. Ferreira 2000; Figueiredo-Silva 2000; Rodrigues 2004)
and (2a) is preferably used.2 Similar facts hold in Finnish, Marathi, and Hebrew (Holmberg 2005). According to the sources cited, a third person NS with definite reference
must have an antecedent in a higher clause.
Ferreira (2000) and Rodrigues (2004) claim that the relation between the antecedent
and the embedded NS in BP is one of obligatory control. Modesto (2008), however,
argues against this view. Holmberg et al. (2009) show that there is variation among Finnish, Marathi and BP regarding the structural conditions governing the relation between
the antecedent and the third person NS, but conclude that in all three languages the
relation is neither Obligatory Control nor Non-obligatory Control, but a third type of
control relation, whose precise nature is not well understood.3
Many of these languages show some kind of asymmetry between the third person and
the other persons.4 In Finnish, Hebrew, and Marathi a third person NS is generally not
allowed in a matrix clause in contrast to first or second persons.5
In all of the partial NSLs mentioned third person NSs can also be found in non-argumental subject constructions and when the subject is interpreted as a generic pronoun,
corresponding to English one, as in (3) below:

faz o doce
This is how one makes the cake

In (3) the generic 3SG NS in the embedded clause denotes people in general, including the speaker and the addressee. This reading of a third person NS is unavailable in a
consistent NSL. As already noted by Perlmutter (1971), a consistent NSL cannot use a
plain NS to convey the meaning of a generic (inclusive) subject and must resort to
some overt strategy. This contrast can be seen clearly when we compare BP with EP.
(3) is a well formed sentence in EP, but it has a different meaning, glossed as This is
the way he she makes the cake. The generic subject reading requires the presence of
the clitic se:
E assim que se faz o doce6

2011 The Author

Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

574 Pilar Barbosa



2.3.1. Holmberg (2005)

In order to capture the differences between the consistent NSLs and the partial NSLs,
Holmberg (2005) proposes that one of the parameters involved in regulating the pronunciation of subject pronouns is whether finite T hosts a D-feature encoding definiteness.
In consistent NSLs T hosts a D-feature, in partial NSLs it does not. In addition, he
proposes a typology of null pronouns: pronouns that are DPs and weak or deficient
pronouns, labeled uPs after Dechane and Wiltschko (2002). These are specified for
u-features but lack D; therefore, they are incapable of co(referring) to an individual or a
group. All NSs in the consistent NSLs are uPs and so are third person NSs in the partial
NSLs. In a language with a D feature in I, a null uP that enters into an Agree relation
with T is interpreted as definite. This is why the consistent NSLs must resort to overt
strategies to express the meaning of a generic subject pronoun. Absence of D in I, on the
other hand, means that a null uP subject is either bound by a QP or logophorically
linked to a DP in a higher clause; as a last resort, it may be interpreted as generic.
Holmberg (2005) discusses data from Finnish and BP that indicate that the definite null
third person subject raises to a high position in the clause (Spec-TP, in his terms) whereas
the generic NS must stay inside the vP. He concludes that the null uP in Finnish and BP
is accessible for binding by a higher DP if and only if it moves out of vP. If it stays in situ
it is inaccessible and the generic reading is the only option. Concerning first and second
person NSs, they are fully specified DP pronouns that are deleted in the phonology, by
the same process that applies to other kinds of ellipsis. Thus, there are two kinds of NSs:
one is an inherently deficient pronoun that needs to enter an Agree relation with T containing D to be interpreted as definite. In a language that lacks D in T, it can be interpreted as a bound or logophoric pronoun; in the absence of a binder it is interpreted as a
generic pronoun. The other is a fully specified DP that is deleted in PF. Regarding the
question why the non-NSLs do not allow almost any subjects to be null, Holmberg suggests that these languages have a stricter, phonological EPP-condition which not only
requires a filled Spec-IP, but a pronounced Spec-IP. In sum, Holmberg concludes that,
as far as core syntax is concerned, NSs in languages with overt agreement are like regular
pronouns; the fact that they are null is a PF matter: they are either deleted pronouns or
feature matrices that fail to have a PF realization.
Roberts (2010b) proposes to reduce the latter option to the former, by resorting to
Chain-Reduction, i.e., the deletion of all identical copies in a dependency except the
highest one (see Nunes 2004). We will review Robertss analysis in the next section.
2.3.2. Copy Deletion Analyses (Holmberg et al. 2009; Roberts 2010b)
Roberts (2010b) argues that NSs in the consistent NSLs are DP pronouns deleted in PF
and attempts to derive the conditions for pronoun deletion from an extension of Roberts
(2010a) theory of incorporation designed to account for Romance clitics and cliticization.
Roberts (2010a) proposes that Romance object clitics are uPs rather than DPs. He follows Chomskys (2000, 2001) Agree based system according to which the label of (active,
transitive) v contains a set of unvalued u-features. These constitute the trigger for Agree
with the matching object. Roberts observes that, since vs u-features are unvalued versions of the very u-features that make up the clitic, the clitics label is not distinct from
vs. More precisely, the clitics features form a proper subset of vs features. Roberts proposes that, given that copying the features of the goal exhausts the content of the goal,
the operation is not distinguishable from the copying involved in movement. Therefore,
2011 The Author
Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

Analyses of Null Subjects and Partial, Discourse and Semi pro-drop 575

the probe and the goal form a chain, which is subject to Chain Reduction, i.e., the deletion of all identical copies in a dependency except the highest one (Nunes 2004). As a
result of Chain Reduction, we see the PF effect of movement, with the u-features realized on the probe and the copy deleted.
Roberts (2010b) extends this idea to NSs in the consistent NSLs. He adopts Holmbergs suggestion that T has a D feature in these languages. Contrary to Holmberg, however, he proposes that the NS is a DP pronoun, more precisely a weak pronoun in the
sense of Cardinaletti and Starke (1999). Following these authors, he assumes that weak
pronouns must occupy a designated Specifier position, namely Spec-TP. Since T contains
a D-feature in addition to u-features that match those of the subject, the latter counts as
a defective goal in the sense that its features are exhausted by those of the probe. Being a
defective goal, it deletes in PF by Chain Reduction.
Roberts claims that the null DP subject has a D-feature valued as definite and values
Ts D-feature in this way. He adopts the following postulate relating definite D and
u-feature specification:

If a category a has D[def], then all as u-features are specified.

Holmberg et al. (2009) incorporate some elements of Robertss copy-deletion analysis of

NSs. The authors maintain that NSs in the consistent NSLs are uPs. For them, the value of
D is a referential index. Since a u-pronoun on its own doesnt carry a D-feature, it cannot
bear a referential index; therefore, it cannot be interpreted as definite. In a language with
D in T, however, a third person u-pronoun may be interpreted as definite if it merges as a
subject in the domain of a T whose D-feature is valued by a null topic and then incorporates with it in the same way as suggested by Roberts for object clitic incorporation. In this
case, the EPP is checked as a result of D valuation by the null topic. Holmberg et al.
assume Frascarellis (2007) proposal that third person NSs in Italian refer to an entity introduced as a topic of discourse, the Aboutness-shift Topic (A-topic), which is always syntactically represented in a designated A-topic position either overtly or covertly.7
Regarding the partial NSLs, when the subject is a u-pronoun, the derivation is essentially the same as that of consistent NSLs with the u-pronoun incorporated in T. Since T
doesnt have a D-feature valued by a null A-topic, the subject lacks a definite interpretation. Definite NSs, by contrast, raise to Spec-TP and check the EPP. Holmberg et al.
suggest that the reason why the definite NS doesnt incorporate is that it is a D-pronoun.
However, it is a deficient D pronoun in the sense that it has an unvalued D-feature.
Therefore, it must enter a control relation with a valued DP antecedent. In other words,
there are two kinds of D-pronouns in the partial NSLs: those that come with an inherently valued D feature and those that come with an unvalued D-feature. The former end
up spelled out in Spec-TP; the latter enter a control relation with a valued DP and are
deleted in PF as a result of an extended version of chain reduction.


Holmbergs work on the partial NSLs languages constitutes a major step in the understanding of the key properties of this type of language. Two strong empirical generalizations emerge: (i) there is a correlation between partial pro-drop and the existence of a
plain third person NS to convey the meaning of a generic (inclusive) subject; (ii) definite
NSs in the partial NSLs raise to a high position whereas the generic NS occupies a low
2011 The Author
Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

576 Pilar Barbosa

We start by noting that the fact that (ii) holds in the partial NSLs doesnt necessarily
mean that the same is true in the consistent NSLs. In fact, in the framework of Holmberg
et al. (2009), the NS in the consistent NSLs doesnt raise to Spec-TP and is rather an
incorporated pronoun, thus contrasting with the definite NS in the partial NSLs, which
raises and checks the EPP.
In Robertss analysis, on the other hand, the NS raises in the consistent NSLs. The
arguments given in favor of this assumption rely on Cardinaletti (1997) and particularly
on Cardinaletti and Starkes (1999) proposal that weak subject pronouns, like the Italian
egli series, are required to occupy a designated specifier position, namely Spec-TP. As
mentioned in Part 1, however, Cardinaletti (2004) provides evidence that, in fact, the
pronouns of the egli series are not in Spec-TP, but rather in a higher position, associated
with the notion subject of predication. The only overt pronoun that is required to fill
Spec-TP is tu you in subjunctive contexts. Cardinaletti shows that this requirement
doesnt hold of tu in indicative contexts and that the reason why it holds in subjunctive
contexts is that the second person subjunctive is the only instance in which Italian is not
pro-drop. Thus, the requirement to fill Spec-TP is not an inherent property of the pronoun itself; instead, it is due to the non-pro-drop nature of the syntactic environment.
This considerably weakens Robertss argument in favor of deducing the properties of the
NS from inherent properties of weak pronouns. Concerning the empirical evidence given
in Cardinaletti (1997) in support of pro raising to Spec-TP, it is argued in Barbosa (2009)
that similar facts hold of cliticization in general, so the Italian data are compatible either
with Holmbergs incorporation analysis or with the different versions of Hypothesis A
discussed in Part 1.
Setting aside the differences between Holmberg et al. (2009) and Roberts (2010b)
regarding the location of the NS in the consistent NSLs, their approach shares the idea
that the core property of this kind of language is that T has a D-feature encoding definiteness. Positing this feature has consequences for the licensing of pronouns that are
deleted in PF, but has no further implications for the syntax of overt subjects: these raise
to Spec-TP in order to check the EPP as happens in a non-NSL. However, as described
in Part 1, the consistent NSLs have a well-defined cluster of properties which includes:
(i) the availability of SV VS alternations (free-inversion); (ii) the fact that subject extraction proceeds from the post-verbal position (see Part 1 for details). Even though the
SV VS alternations can in principle be deduced from the presence of a null expletive in
Spec-TP, the question arises of how (ii) is obtained under this approach. This problem
doesnt arise under Hypothesis A, where both (i) and (ii) follow for free (see Part 1 for
discussion). Besides this, Barbosa (1995, 2009), Alexiadou and Anagnostopoulou (1998),
Ordonez and Trevino (1999), among others, discuss a number of differences between the
NSLs and the non-NSLs regarding scope interactions between overt pre-verbal subjects
and quantifiers elsewhere in the clause, asymmetries between referential and non-referential quantified subjects, and restrictions on the interpretation of pronouns as bound variables. Since the presence of the D-feature in T has no impact on the distribution of overt
subjects, these facts are left unaccounted for in this approach.
Focusing on the differences between the consistent NSLs and the partial NSLs regarding overt subjects, recall that one of the aspects that distinguish these two types of language is that there are environments where an overt pronoun is not allowed in a
consistent NSL, in accordance with the Avoid Pronoun Principle; in a partial NSL its
presence is optional (cf. (2)). Barbosa et al. (2005) argue that this difference follows under
the assumption that the pre-verbal overt subject in EP is a (clitic) left-dislocated topic
whereas in BP it raises (or may raise) to Spec-TP. Viewed in this light, the Avoid
2011 The Author
Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

Analyses of Null Subjects and Partial, Discourse and Semi pro-drop 577

Pronoun Principle simply reduces to preference for not merging a pronoun as a topic
unless it is required to signal topic switch or for emphasis empathy. Barbosa et al. (2005)
examine BP against the same set of phenomena where asymmetries in the behavior of
overt subjects can be detected between the consistent NSLs and the non-NSLs and
observe that BP patterns with the non-NSLs rather than with EP, thus concluding that
subjects in BP raise to Spec-TP.8 The data discussed there indicates that the status of
pre-verbal subjects in a consistent NSL differs from that of a partial NSL.
Under the deletion analyses described above it is not clear how these facts follow. In
Robertss analysis, the Avoid Pronoun Principle reduces to Chain Reduction. What distinguishes overt pronouns from their null counterparts is the presence of a Case feature:
overt pronouns have a Case feature that is not present in the probe; therefore, they cannot be deleted. However, the Case feature on the pronoun doesnt really explain its
inability to take a matrix subject as an antecedent. For Holmberg et al. (2009), overt pronouns are DPs with a valued D feature in the consistent NSLs as well as in the partial
NSLs, so they should behave similarly in both sets of languages, contrary to fact. In sum,
the endeavor to reduce NSs in both languages to regular pronouns that delete in PF falls
short of explaining the divergent behavior of overt pronouns in these two types of NSL.9
An assumption behind the deletion analyses under discussion is that the D feature in T
encodes definiteness and is interpretable. This assumption is designed to capture the
observation that the consistent NSLs cannot assign a generic interpretation to the third
person singular pronoun. However, in many consistent NSLs, the third person plural NS
can have an indefinite reading as shown below for EP.

a bater
a` porta.
at knock
at-the door
They are knocking at the door There is someone knocking at the door

(6) is ambiguous. It may mean that some contextually given set of people is knocking at
the door or it may mean that someone is knocking at the door. The latter is the arbitrary
reading of the 3PL NS (cf. Cinque 1988 and Jaeggli 1986, among others). This fact is
problematic for Roberts given that he claims that the NS is inherently definite and values
D as definite, but it is even more of a problem for Holmberg et al. (2009) (as well as Frascarelli 2007). In (6) the NS lacks an antecedent in discourse, so it is not possible to claim
that D in T is valued by a topic. Since, for Holmberg et al., the EPP is checked under
valuation by a null topic whenever the subject is null, the prediction is that the EPP
should fail to be checked in (6) and that the sentence should be bad for failure of EPP
checking, contrary to fact.
Thus, the idea that D in T is always interpreted as definite whenever the subject is null
is too strong. In sum, the deletion analyses discussed have shortcomings: on the one hand,
they lack the predictive power required to account for the cluster of properties that
differentiate the two sets of NSLs; on the other hand, they are too restrictive.
In spite of this, it seems clear that there is a correlation between lack of full u-feature
specifications and the availability of a generic NS. In effect the languages that lack agreement morphology and yet license NSs, such as Chinese, Japanese, and Korean, all have
plain generic NSs (see Holmberg et al. 2009). Thus, the availability of a generic (inclusive) reading of a plain third person NS could be taken as an indication that the mechanism of licensing the NS is not directly related to agreement. In this case, the partial
NSLs would be grouped with the discourse pro-drop languages, a hypothesis developed
by Modesto (2008) for BP and Finnish on the basis of the idea these languages are topic
2011 The Author
Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

578 Pilar Barbosa

prominent. In the next section we will discuss the discourse pro-drop languages and show
how they relate to the partial NSLs.
3. Discourse pro-drop Languages
Many East Asian languages including Chinese, Japanese and Korean, have no agreement
at all even though they allow pro-drop. In fact, in these languages argument drop is even
more widespread than in languages like Italian since any argument, a subject or an object,
can be dropped. These languages do not have overt expletives. A plain null argument can
be interpreted as generic (cf. Holmberg et al. 2009).
Typologically, these languages exhibit the cluster of properties characteristic of discourse-orientation (cf. Huang 1984). In particular, they are topic prominent in the sense
of Li and Thompson (1976). Huang (1984) proposes a theory of empty arguments based
on two parameters: one allowing zero topics; the other allowing a silent pronominal
argument. Zero topics bind a variable in argument position and are subject to the usual
constraints on movement. The silent pronominal argument is a minimally specified nominal (pro) and is subject to conditions on recoverability: it must be locally bound by either
a c-commanding DP argument or by rich agreement. Since these discourse-oriented languages have no agreement, pro can only appear as the subject of an embedded clause
where it is bound by the closest c-commanding argument. All other types of null argument in Chinese are bound by a zero topic.
Huangs proposal faces the problem that object drop in Chinese is not exactly subject
to the same constraints as movement (Huang 2000; Li 2007). In particular, a null object
may appear inside a relative clause whereas extraction out of relative clauses in Chinese is
ruled out.10
Speas (1994, 2006) offers an account of pro-drop that relies on the idea that it is the AgrP
projection rather than pro that must be licensed (see Part 1). In her system, AgrP can only be
projected if its head or specifier contains overt material. In languages with rich agreement
morphology the agreement affix is an independent lexical item, which can be inserted
directly in the head of AgrP, thereby licensing this projection. Consequently, pro can be
inserted in Spec-VP. In languages with poor agreement, AgrP is licensed only if it contains
an overt subject in its Spec. Thus, pro-drop is ruled out. In languages without any agreement, no AgrP needs to be projected, so pro can be inserted. This analysis, however, fails
to capture the differences between partial and consistent pro-drop and is problematic for
languages that have poor subject agreement and yet allow NSs, such as BP.
In a recent paper, Neeleman and Szendr}
oi (2007) propose to approach the issue of prodrop in agreementless languages by focusing on the pronominal paradigm. These authors
follow the line of inquiry that takes null arguments to be pronouns which are deleted in
PF. Richly inflected languages allow null arguments because the content of the deleted
pronoun may be locally recovered. In non-richly inflected languages the process of pronoun deletion is constrained by morphology: the only pronouns that can fail to have a
PF realization are those that have agglutinating morphology. Hence, the morphological
characteristics of the pronominal paradigm determine whether a language allows widespread pro drop or not. The authors argue that all of the discourse pro-drop languages
have agglutinating morphology on pronouns.
As mentioned above, a serious shortcoming of the pronoun deletion approach to
pro-drop is that it falls short of explaining the key properties that are associated with each
different kind of NSL. Thus, the fact that widespread pro-drop is observed in topic prominent languages is treated here as a mere coincidence. Likewise, the systematic differences
2011 The Author
Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

Analyses of Null Subjects and Partial, Discourse and Semi pro-drop 579

between partial and consistent pro-drop that are described above are left unexplained.
Besides, on Neeleman and Szendr}
ois account, pronouns in BP have non-agglutinating
morphology; since subject agreement morphology in BP is no longer distinctive for the
person feature, BP is predicted to be non-pro-drop, contrary to fact.
Tomioka (2003) offers a different perspective on discourse pro-drop. He observes that
all of the languages that allow discourse pro-drop allow (robust) bare NP arguments. He
examines the different interpretations of zero pronouns in Japanese and argues that they
are related to the inherent semantic flexibility of full-fledged NPs. Tomioka proposes that
the different uses of full-fledged NPs are derived from one basic meaning, property
anaphora (type <e,t>) and their differences are the result of two independently needed
semantic operations, namely Existential Closure (yielding indefinite interpretation) and
Type Shifting to an individual (yielding definite interpretation). Then he shows that the
same semantic tools that are used to interprete full NPs can be used to interprete pro.
Tomioka proposes that what underlies discourse pro-drop is the fact that languages
(almost) universally allow phonologically null NP anaphora (also known as N or NP
ellipsis).11 In a language that lacks determiners, this operation will give rise to phonologically unrealized arguments. In languages in which DPs are necessarily projected, a remnant D will always show up and so this process will never give rise to a silent argument.
This proposal captures the fact that the discourse pro-drop languages allow virtually any
argument to be dropped and is rather appealing, not only for its elegance and simplicity,
but also because it can be easily integrated in an independently motivated parameter: the
Nominal Mapping Parameter of Chierchia (1998), which determines the availability of a
bare NP in argument position in a language. Furthermore, it has the potential to relate
discourse pro-drop and topic prominence. As mentioned by Tomioka, all discourse
pro-drop languages seem to allow bare NP arguments, but not all bare NP argument
languages allow pro-drop. This observation could be captured if topic prominence is
somehow a condition for recoverability of reference of the null NP, particularly when it
is interpreted as a definite.
Independent evidence in support in favor of Tomiokas hypothesis comes from Spanish, Greek, and EP. Spanish and Greek have indefinite arbitrary null objects only in the
environments in which they allow bare plurals (cf. Campos (l986), Raposo (1998) for
Spanish and Giannakidou and Merchant (1997) for a discussion of Greek null indefinite
objects). EP has definite null objects. Raposo (1998) argued that this option is connected
to the unique distribution of bare plurals in this language as opposed to the other
Romance languages. He claimed that EP has a null definite D and that the null object is
a null D with a null NP complement. Thus, this is a case of null NP-anaphora that yields
a silent argument.
As mentioned by Tomioka, this approach raises the obvious question why English (or
Germanic in general) is not pro-drop, considering that it has bare plurals. In our view,
however, this doesnt mean that the hypothesis should be discarded altogether given that
there are intriguing crosslinguistic correlations between the availability of bare NP arguments and pro-drop. In the next sections we briefly discuss these.
4. The Null NP Anaphora Hypothesis and Partial pro-drop
The discourse pro-drop languages share two properties with the partial pro-drop languages
that set them apart from the consistent NSLs. The first one is the lack of Avoid Pronoun
effects of the type discussed in Section 2: an embedded non-emphatic overt pronoun in
Chinese may take a matrix subject as an antecedent.12 The second is that a plain NS can
2011 The Author
Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

580 Pilar Barbosa

have a generic (inclusive) interpretation. In Section 2, it was observed that in Finnish as

well as BP the generic third person NS stays in situ whereas the definite interpretation is
available just in case the NS raises to a high position. The relevant data are the following
(Holmberg et al. 2009):
(7) a.

Jari sanoo etta tassa istuu mukavasti.

Jari says
that here sits
Jari says that one can sit comfortably here.
Jari says that he sits comfortably here
b. Jari sanoo etta (han) istuu mukavasti
Jari says
that he
comfortably here
Jari says that he sits comfortably here.
Jari says that one can sit comfortably here

In Finnish, the EPP can be satisfied by other categories besides subjects. In (7a), the locative adverbial checks the EPP. In this case, the only reading available for the null subject
is the impersonal, generic interpretation. In (7b) the EPP is checked by the NS. Here,
the generic reading is not a possibility and the subject must be interpreted as a definite
pronoun controlled by the higher subject.
Holmberg (2005) assumes that the Finnish EPP position is Spec-TP. However, Holmberg and Nikanne (2002) show that this position is associated with topics and argue that
Finnish is a topic prominent language.13 On the basis of these observations and on the
fact that BP passes all of Li and Thompsons (1976) diagnostics for being classified a topic
prominent language (Pontes 1987), Modesto (2008), in a comparative study of BP and
Finnish, argues that the definite anaphoric NS in Finnish and BP is itself in topic position
i.e., is a null topic in the spirit of Huang (1984) thus collapsing partial pro-drop with
discourse pro-drop. One strong argument in favor of this approach is that both languages
have null objects interpreted with recourse to a discourse antecedent.
In this context, the null NP anaphora hypothesis would predict that both languages
should allow bare NPs in argument positions, and this prediction is confirmed: Finnish
doesnt have determiners; BP has determiners, but, unlike EP, it has bare singular and
plural nouns in subject or object position (cf. Muller 2001, Schmitt and Munn 2003).
Under the null NP anaphora hypothesis, the correlation between the two different positions (the Topic position or the VP internal position) and the available readings in (7)
would follow from the different configurations that serve as input to semantics: when the
null NP stays inside the VP it is interpreted by Existential Closure under the scope of a generic operator; when it raises to a topic position, the individual (definite anaphoric) reading
becomes available (see Kuno (1973) for arguments that topichood signals definiteness).
Other languages that have been shown to be partial pro-drop are Marathi, Hebrew and
Russian. Interestingly, all of these languages have null objects and bare nouns in argument
position. Marathi and Russian lack determiners, like Finnish. Hebrew has definite articles but
lacks an indefinite article, and has singular as well as plural bare nouns, with a range of interpretations that is similar to that of BP bare nouns (cf. Doron 2003). Thus, there appears to be
a correlation between partial pro-drop and the availability of bare NP arguments.
Even though the correlation holds, the picture is more complex than this. In Hebrew,
the availability of referential NSs is correlated with the presence of person agreement.
Present tense verbs in Hebrew are participles bearing number and gender agreement only.
In this tense, impersonal and non-argumental NSs are allowed; definite NSs are not
(Ritter 1995). In this respect, Hebrew differs from Russian. Past tense verbs in Russian
2011 The Author
Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

Analyses of Null Subjects and Partial, Discourse and Semi pro-drop 581

are also participles that are only marked for number and gender. Yet, definite subject
drop is possible in the past tense in Russian. This fact indicates that more than one
parameter of variation is involved. In Hebrew though not in Russian the definite (individual) interpretation of the NS appears to be contingent upon the presence of person
The pattern of NSs found in Hebrew present tense is not unique. It is found in some
creole languages, such as Cape Verdean Creole, as illustrated below:
(8) a. *(El) ta trabadja duro.
he asp works
b. Sta faze
making cold
c. Na
korda sedu.
in-the summer Asp wake early
In the Summer one wakes up early

Baptista (1995)

Cape Verdean only has non-argumental (cf. (8b)) and impersonal NSs (cf. (8c)). Similar
facts hold in Papiamentu (Muysken and Law 2001). Interestingly, both creoles have bare
nominals in argument position. Since they also lack agreement inflection, their behavior
is parallel to that of Hebrew past tense.15
The occurrence of impersonal NSs in correlation with the availability of bare NP arguments in these creoles as well as in Hebrew present tense indicates that Tomiokas
hypothesis is on the right track even though it requires further elaboration. More crosslinguistic studies need to be undertaken in order to determine whether these correlations
hold, but the facts described show that there is split among the languages discussed and
that the dividing line lies in the presence vs. absence of the resources required to assign
definite interpretation to the NS (in Tomiokas terms, for type shifting to an individual
to apply). Above it was suggested that topic prominence is one such resource, but the
Hebrew facts suggest that person agreement also plays a role.16
One issue raised by the null NP anaphora hypothesis is that it doesnt offer an immediate account of null expletives (it makes little sense to posit a non-theta bearing null NP).
This brings us to the issue of the status of null expletives, which will be briefly discussed
in the next section in connection with yet a fourth type of NSL.
5. Semi pro-drop Languages
In this section, we briefly discuss the phenomenon of expletive subject omission by
focusing on a set of languages that have been described in the literature as having only
expletive NSs. These are often referred to as semi null-subject or semi pro-drop languages
(Biberauer 2010; Falk 1993; Gilligan 1987; Grewendorf 1989; Hermon and Yoon 1989;
Koster 1987; Platzack 1987). It is standardly assumed that these fall into two subtypes:
(9) a.

Those which only permit non-argumental expletive omission, i.e., omission

of true expletives, which do not bear a theta-role e.g., Dutch, German.
b. Those which permit both non-argumental expletive omission and omission
of so-called quasi-arguments e.g., Icelandic, Yiddish.

All of the languages mentioned in (9) are V2 languages, i.e., they have V-to-C movement in root clauses with the additional requirement that Spec-CP be filled. In neutral
2011 The Author
Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

582 Pilar Barbosa

declaratives in which no XP has been focused or topicalized, expletives fill the preverbal
slot (Spec.CP). This is illustrated in (10a) for the Icelandic impersonal passive and in
(10b) for a German presentational construction:
(10) a. a var stundum hlegi a raerranum [Icelandic: Wurmbrand 2006]
was sometimes laughed at the minister
The minister was laughed at
b. Es ist heute ja doch ein Brief gekommen
[German: Biberauer 2010]
it is today MOD.PART.8 a letter come
There did after all come a letter today
Neither Icelandic nor German permit an overtly realized expletive to the right of V
raised to C in these contexts:
(11) a.

b. Heute

var (*a) hlegi

was it
laughed at
ja doch

the minister
ein Brief
a letter

Assuming that the position to the right of V in (11a,b) is Spec-TP and also the validity of
the EPP as formulated in Chomsky (1981), it was standardly assumed that German and Icelandic license proEXPL. Referential subjects located in this position cannot, however, be null.
German differs from Icelandic in requiring an overtly realized expletive in Spe-TP in
some contexts including weather-it (cf. (12)) and certain motion and experiencer constructions such as (13); in Icelandic, the presence of the expletive is precluded (examples
from Wurmbrand 2006):
(12) a. Heute schneit *(es)
today snows it
Today it is snowing
b. I dag hefur (*a) rignt
today has
(*it) rained
Today, it rained.


(13) a. weil *(es) ihm die Haare zerzaust hat

since *(it) him the hairs tangled has
since his hair got tangled; something made his hair tangled
[lit. It has tangled him the hair.]
b. Af husinu
bles (*a) strompinn
[Icelandic: Haider 2001]
off the. house blew (*it)
the chimney.Acc
The chimney blew off the house
Expletives of this type have been argued to be quasi-argumental (i.e., non-referential,
but yet theta-role bearing; cf. Chomsky (1981), Bennis (1986), Cardinaletti (1990), Vikner (1995)). The first argument for assigning (quasi)-argumental status to these expletives
is their ability to control PRO. The second argument comes from the distribution of
accusative case: accusative case on a DP is only possible in German when there is also an
(underlying) external nominative argument (Haider 1985; Wurmbrand 2006). Hence, if
an expletive is an argument that bears nominative Case, other DPs in the same clause
should surface as accusative. Haider (2001) shows that this predication is borne out in the
2011 The Author
Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

Analyses of Null Subjects and Partial, Discourse and Semi pro-drop 583

case of (13a), where the expletive is in Spec-TP. When the expletive is a Spec-CP filler,
as in (10b) above, the non-expletive DP surfaces as nominative. Thus, overt TP expletives in German are syntactically active contrary to Spec-CP expletives.
Turning to Icelandic, we find that the constructions corresponding to the German syntactic expletives have the same Case properties (cf. Haider 2001); however, no overt
expletives are present (cf. (12b), (13b)). Hence, Icelandics counterpart to the German TP
expletive is a covert expletive. To conclude, German and Icelandic exhibit a similar
behavior with respect to non-argumental proEXPL-A, whereas quasi-argumental proEXPL+A
is licensed in Icelandic only.
These facts have traditionally been related to the inflectional systems of each language,
Icelandic being inflectionally richer than German. According to Rizzi (1986), quasi-argumental proEXPL+A requires number specification on Infl whereas non-argumental
proEXPL-A doesnt. This explanation, however, is problematic in view of the fact that
proEXPL+A is attested in creoles that lack number agreement inflection (cf. (8b)).
In recent years, the idea that proEXPL-A exists has been challenged (Biberauer 2010;
Wurmbrand 2006). In fact, the sole motivation for positing such an entity is theory internal: assuming that the EPP is universal, it follows that Spec-TP must be filled by a covert
nominal in examples such as (11). However, Wurmbrand (2006) as well as Biberauer
argue against the idea that the standard EPP holds in these languages. Among the arguments given is the lack of syntactic evidence in favor of the existence of a covert expletive in (11), since it does not trigger accusative Case on the VP-internal DP and does not
seem to be present in the syntax for binding purposes.
Biberauer (2010) proposes an analysis of the distribution of overt expletives in Dutch
and Afrikans that does without proEXPL-A. If she is right, there is no need to posit the
existence of NSLs of Subtype 1, in which case we are left with only one kind of semi
pro-drop language, the Icelandic type. This conclusion answers the question posed at the
end of last section in connection with the null NP anaphora hypothesis: only theta-bearing nominals can be null.
Sigursson and Egerland (2009) observes that, in addition to the quasi-argument NSs
mentioned, Icelandic has impersonal null subjects that are syntactically active, as in the
Impersonal Modal construction illustrated below:

ma [ ]
fara a dansa.
go to dance
One may begin to dance now

Even though the availability of an impersonal NS in Icelandic is confined to specific constructions, its very existence indicates that Icelandic patterns with the creole languages
discussed in the previous section. Curiously, Icelandic has no indefinite article and displays bare singular nouns with indefinite interpretation. This suggests that an account
along the lines developed in the last section for Cape Verdean Creole and Papiamentu
might be suitably extended to Icelandic.17
6. Concluding Remarks and Directions for Future Research
In this article we have discussed the pronoun deletion analyses of pro-drop and we concluded that they have shortcomings. We examined three different types of NSLs: partial,
discourse and semi pro-drop languages. We showed that there are intriguing correlations

2011 The Author

Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

584 Pilar Barbosa

between these different types of NSL and the availability of bare NP arguments, as predicted by Tomiokas null NP anaphora hypothesis.
This hypothesis doesnt mean that if a language has robust bare NP arguments it will
necessarily display the range of properties associated with discourse or partial pro-drop.
Polish and Czech lack articles and, unlike Russian, they display the properties associated
with consistent pro-drop (Lindseth 1998). These languages differ from Russian in that
verbal inflection is marked for person agreement in all tenses. Holmberg (2005) and Roberts (2010a,b) associate consistent subject drop with the presence of D in T and attempt
to relate this property to rich agreement morphology. The versions of Hypothesis A presented in Part 1 of this article go a step further in claiming that the functional head bearing subject agreement has valued interpretable u-features in the consistent NSLs thus
behaving like a pronominal. If so, then the behavior of Czech and Polish follows from
the properties of Agr in these languages.
A number of questions arise. The Germanic languages have bare plurals but are nonpro-drop; Maori is a pro-drop language, yet it has no verbal agreement and apparently no
bare NP arguments either (Neeleman and Szendr}
oi 2007). These are some of the issues
that need to be addressed if this line of inquiry is pursued further, an effort that we
believe to be worthwhile.
Short Biography
Pilar Barbosa received a BA in Modern Languages and Literatures from the University of
Porto, Portugal, an MA in Linguistics from the University of Massachusetts at Amherst,
and a PhD in Linguistics from the Massachusetts Institute of Technology. She has been
on the faculty of the Instituto de Letras e Ciencias Humanas, of the University of Minho,
Portugal, since l997. Her primary research interests are theoretical and comparative
syntax. She has worked on issues such as pro-drop, cliticization, control and other aspects
of the syntax of the Romance languages.
* Correspondence address: Pilar Barbosa, Instituto de Letras e Ciencias Humanas, Universidade do Minho, Campus de Gualtar, 4715 Braga, Portugal. E-mail: pbarbosa@ilch.uminho.pt
As Roberts (2010b) points out, this constitutes a partial return to one of the main ideas in Perlmutters (1971)
analysis of NSs in that the NS arises through deletion of a subject pronoun.
It should be pointed out that examples of this kind are attested in spoken corpora, particularly when the matrix
subject is first or second person and the antecedent is highly accessible.
Gutman (2004) examines a variety of evidence from Finnish and Hebrew that indicates that a purely syntactic
analysis of the phenomenon cannot explain the whole range of data. He argues that only a theory of discourse
anaphora can account for the distribution of NSs in these languages.
These languages vary with respect to their verbal agreement paradigms. Compared to EP, BP has a reduced paradigm (Duarte 1995), but Finnish, Marathi, Russian, and Hebrew (past and future tenses) have relatively rich verbal
agreement morphology (see Holmberg et al. for discussion).
In Finnish, BP and Marathi, a third person NS occurs in yes-no questions (see Holmberg et al.). In BP it is possible to find a third person NS in a matrix clause as long as the antecedent is highly accessible. McShane (2009)
reports similar facts in Russian. According to Holmberg et al., Marathi only allows second person NSs in matrix
In Hebrew and Russian, the generic NS is marked as plural. The consistent NSLs also have an arbitrary third
person plural NS (Jaeggli 1986), but the range of interpretations available for this NS crucially differs from that of
the Hebrew and Russian third person plural NS. The latter has an inclusive reading that is never available with the
former (see Holmberg 2005).

2011 The Author

Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

Analyses of Null Subjects and Partial, Discourse and Semi pro-drop 585

Concerning first and second person null subjects, Holmberg et al. (2009) adopt Frascarellis proposal that every
clause has features representing the speaker and the addressee in the C-domain (cf. Sigursson 2004). Thus, the
speaker and the addressee are always available as antecedents.
This doesnt mean that a pre-verbal subject cannot also be a Topic in BP (see Pires (2007) on the different
structural positions occupied by pre-verbal subjects in BP).
The copy-deletion analysis faces yet another problem. Regular pronouns in the languages under consideration
have gender, verbal agreement has not. Thus, in order for the NS to count as a deficient goal, it must be inherently
unspecified for gender. Therefore, even within the framework of assumptions adopted by the deletion approach,
one is inevitably drawn to the conclusion that a NS has at least some inherently unspecified features, which are
otherwise present in overt pronouns.
Null subjects may also appear inside a relative clause, but they are subject to locality: no potential antecedent
can intervene between the null subject and its antecedent (see Huang (1984) and Li (2007) for discussion).
For arguments that null arguments in Japanese and Korean arise from elision of full fledged structures, see Kim
(1999), Oku (1998), Saito (2007) and Takahashi (2008). For an interesting discussion of the differences between
Japanese and Chinese regarding null subjects, see Takahashi (2007).
But note that an overt pronoun can not be bound to a quantifier in Chinese (see Huang 2000).
In fact, Holmberg and Nikanne argue that the EPP related position is higher than TP.
Since Russian lacks definite determiners, bare NPs can have a deictic or anaphoric interpretation. In Hebrew
this is not the case. It could be that the difference between Hebrew and Russian noted in the text is related to
this parameter.
On the distribution of null expletives in the creole languages see Nicolis (2008).
Ritter (1995) claims that verbal agreement in Past and Futures tenses in Hebrew has a D feature; third person
agreement is defective when compared to the other persons of the paradigm. Shlonsky (2009) argues that first and
second person agreement morphemes are incorporated subject clitics; third person agreement has an unspecified person slot. Both authors converge on the idea that third person agreement in Hebrew marks definiteness.
If so, then the term semi pro-drop language should more appropriately be used to apply to the pattern of pro-drop
that is restricted to the occurrence of quasi-argumental and impersonal NSs. It remains to be determined where
Yiddish stands given that its pattern of expletive NSs is not quite identical to that of Icelandic. In the Yiddish
counterpart to (11b) the presence of a null expletive is optional, not obligatory.

Works Cited
Alexiadou, Artemis, and Elena Anagnostopoulou. 1998. Parameterizing AGR: Word Order, V-Movement and
EPP-Checking. Natural Language and Linguistic Theory 16. 491539.
Baptista, Marlyse. 1995. On the pro-drop parameter in Cape Verdean Creole. Harvard working papers in linguistics,
Vol. 5, ed. by H. Thrainsson, S. D. Epstein and S. Peter, 317. Cambridge, MA: Harvard University.
Barbosa, Pilar. 1995. Null subjects. PhD dissertation, Cambridge, MA: MIT.
. 2009. Two kinds of subject pro. Studia Linguistica 63(1). 258.
, Eugenia Duarte, and Mary Kato. 2005. Null subjects in European and Brazilian Portuguese. Journal of Portuguese Linguistics 4. 1152.
Bennis, Hans. 1986. Gaps and dummies. Dordrecht: Foris.
Biberauer, Theresa. 2010. Semi null-subject languages, expletives and expletive pro reconsidered. Parametric variation: null subjects in minimalist theory, ed. by T. Biberauer, A. Holmberg, I. Roberts and M. Sheehan, 36475.
Cambridge: Cambridge University Press.
Borer, Hagit. 1989. Anaphoric AGR. The null subject parameter, ed. by O. Jaeggli and K. J. Safir, 69109. Dordrecht: Kluwer Academic Publishers.
Campos, Hector. 1986. Indefinite object drop. Linguistic Inquiry 17. 3549.
Cardinaletti, Anna. 1990. Impersonal constructions and sentential arguments in German. Padua: Unipress.
. 1997. Subjects and clause structure. The new comparative syntax, ed. by L. Haegeman, 3363. London:
. 2004. Towards a cartography of subject positions. The structure of IP and CP: the cartography of syntactic
structures, Vol. 2, ed. by L. Rizzi, 11565. Oxford: Oxford University Press.
, and Michal Starke. 1999. The typology of structural deficiency. Clitics and other functional categories in
European languages, ed. by H. van Riemsdijk, 145233. Berlin: Mouton de Gruyter.
Chierchia, Gennaro. 1998. Reference to kinds across languages. Natural Language Semantics 6. 339405.
Chomsky, Noam. 1981. Lectures on government and binding. Dordrecht: Foris.
. 1995. The minimalist program. Cambridge, MA: MIT Press.
. 2000. Minimalist inquiries: the framework. Step by step: essays on minimalist syntax in honor of Howard
Lasnik, ed. by R. Martin, D. Michaels and J. Uriagereka, 89156. Cambridge, MA.: MIT Press.
. 2001. Derivation by phase. Ken Hale: a life in language, ed. by M. Kenstowicz, 152. Cambridge, MA:
MIT Press.
2011 The Author
Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

586 Pilar Barbosa

Cinque, G. 1988. On si constructions and the theory of arb. Linguistic Inquiry 19. 52182.
Dechane, Rose-Marie, and Martina Wiltschko. 2002. Decomposing pronouns. Linguistic Inquiry 33. 40942.
Doron, Edit. 2003. Bare singular reference to kinds. Proceedings of Semantics and Linguistic Theory 13. 7390.
Duarte, Maria Eugenia. 1995. A Perda do Princpio Evite pronome no Portugues Brasileiro. Campinas:
UNICAMP PhD dissertation.
Falk, Cecilia. 1993. Non-referential subjects in the history of Swedish. PhD dissertation, University of Lund.
Ferreira, Marcelo. 2000. Argumentos nulos em Portugues Brasileiro. MA dissertation, Campinas: UNICAMP.
Figueiredo-Silva, C. 2000. Main and embedded null subjects in Brazilian Portuguese. The null subject parameter in
Brazilian Portuguese, ed. by M. Kato and E. Negrao, 12746. Frankfurt: Vervuert-IberoAmericana.
Franks, Steven. 1995. Parameters of slavic morphosyntax. New York: Oxford University Press.
Frascarelli, Mara. 2007. Subjects, topics and the interpretation of referential pro. Natural Language and Linguistic
Theory 25(4). 691734.
Giannakidou, Anastasia, and Jason Merchant. 1997. On the interpretation of null indefinite objects in Greek. Studies in Greek Linguistics 17. 14155.
Gilligan, Gary. 1987. A cross linguistic approach to the pro-drop parameter. PhD dissertation, University of Southern California.
Grewendorf, Gunther. 1989. Small pro in German. Scrambling and barriers, ed. by G. Grewendorf and W. Wolfgang Sternefeld, 294315. Amsterdam: Benjamins.
Gutman, Eynat. 2004. Third person null subjects in Hebrew, Finnish and Rumanian: an accessibility-theoretic
account. Journal of Linguistics 40. 46390.
Haider, Hubert. 1985. The case of German. Studies in German grammar, ed. by J. Toman, 65101. Dordrecht:
Foris Publications.
. 2001. How to stay accusative in insular Germanic. Working Papers in Scandinavian Syntax 68. 114.
Hermon, Gabriella, and James Yoon. 1989. The licensing and identification of pro and the typology of AGR.
Chicago Linguistic Society 25(1). 17492.
Holmberg, Anders. 2005. Is there a little pro? Evidence from Finnish. Linguistic Inquiry 36. 53364.
, Aarti Nayudu, and Michelle Sheehan. 2009. Three partial null-subject languages: a comparison of Brazilian
Portuguese, Finnish and Marathi. Studia Linguistica 63(1). 5997.
, and Urpo Nikanne. 2002. Expletives, subjects, and topics in Finnish. Subjects, expletives, and the EPP, ed.
by P. Svenonius, 71106. New York: Oxford University Press.
Huang, C.-T. J. 1984. On the distribution and reference of empty pronouns. Linguistic Inquiry 15. 53174.
Huang, Yan. 2000. Anaphora. A cross-linguistic study. Oxford: Oxford University Press.
Jaeggli, Osvaldo. 1986. Arbitrary plural pronominals. Natural Language and Linguistic Theory 4. 4376.
Kim, Soowon. 1999. Sloppy strict identity, empty objects, and NP ellipsis. Journal of East Asian Linguistics 8.
Koster, Jan. 1987. Domains and dynasties: the radical autonomy of syntax. Dordrecht: Foris.
Kuno, S. 1973. The structure of the Japanese language. Cambridge, MA: The MIT Press.
Li, Charles N., and Sandra A. Thompson. 1976. Subject and topic: a new typology of language. Subject and topic,
ed. by C. Li, 45790. New York: Academic Press.
Li, Y.-H. A. 2007. Beyond empty categories. Bulletin of the Chinese Linguistic Society of Japan 254. 74106.
Lindseth, M. 1998. Null-Subject properties of Slavic languages. With special reference to Russian, Czech and
Sorbian. Munchen: Sagner.
McShane, Marjorie. 2009. Subject ellipsis in Russian and Polish. Studia Linguistica 63(1). 98132.
Modesto, Marcelo. 2008. Null subjects in Brazilian Portuguese and Finnish: they are not derived by movement.
New horizons in the analysis of control and raising, ed. by W. D. Davies and S. Dubinsky, 23148. Netherlands:
Muller, Ana. 2001. Genericity and the denotation of common nouns in Brazilian Portuguese. UMOP 25: Proceedings of SULA (Semantics of Under-Represented Languages in the Americas), ed. by A. Weerle and J.-Y. Kim,
7280. UMass: GLSA.
Muysken, Pieter, and Paul Law. 2001. Creole studies. A theoretical linguists field guide. Glot International 5(2).
Neeleman, A., and K. Szendr}
oi. 2007. Pro-drop and the morphology of pronouns. Linguistic Inquiry 38(4). 671
Nicolis, Marco. 2008. The null subject parameter and correlating properties: the case of creole languages. The limits
of syntactic variation, ed. by T. Biberauer, 27194. Amsterdam: Benjamins.
Nunes, Jairo. 2004. Linearization of chains and sideward movement. Cambridge, MA: MIT Press.
Oku, Satoshi. 1998. A theory of selection and reconstruction in the minimalist perspective. University of Connecticut:
Doctoral dissertation.
Ordonez, Francisco, and Esthela Trevino. 1999. Left dislocated subjects and the pro-drop parameter: a case study of
Spanish. Lingua 107. 3968.
Perlmutter, David. 1971. Deep and surface constraints in syntax. New York: Holt, Rinehart & Winston.

2011 The Author

Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

Analyses of Null Subjects and Partial, Discourse and Semi pro-drop 587
Pires, Acrisio. 2007. The subject, it is here! The varying structural positions of preverbal subjects. DELTA 23. 113
Platzack, Christer. 1987. The Scandinavian languages and the null-subject parameter. Natural Language and
Linguistic Theory 5. 377401.
Pontes, Eunice. 1987. O topico no portugues do Brasil. Campinas, S. P: Pontes Editores.
Raposo, Eduardo. 1998. Definite Zero alternations in Portuguese: towards a unification of topic constructions.
Romance linguistics: theoretical perspectives, ed. by A. Schwegler, B. Tranel and M. Uribe-Etxebarria, 197212.
Amsterdam: John Benjamins.
Ritter, Elizabeth. 1995. On the syntactic category of pronouns and agreement. Natural Language and Linguistic
Theory 13(3). 40543.
Rizzi, Luigi. 1986. Null objects in Italian and the theory of pro. Linguistic Inquiry 17. 50157.
Roberts, Ian. 2010a. Agreement and head Movement: clitics, incorporation and defective goals. Cambridge, MA:
MIT Press.
. 2010b. A deletion analysis of null subjects. Parametric variation: null subjects in minimalist theory, ed. by
T. Biberauer, A. Holmberg, I. Roberts and M. Sheehan, 5887. Cambridge: Cambridge University Press.
Rodrigues, Cilene. 2004. Impoverished morphology and A-movement out of case-domains. PhD dissertation,
University of Maryland.
Saito, Mamoru. 2007. Notes on East Asian argument ellipsis. Language Research 43. 20327.
Schmitt, C., and A. Munn. 2003. The syntax and semantics of bare arguments in Brazilian Portuguese. Linguistic
Variation Yearbook 2. 185216.
Shlonsky, Ur. 2009. Hebrew as a partial null-subject language. Studia Linguistica 63(1). 13357.
Sigursson, Halldor. 2004. Meaningful silence, meaningless sounds. Linguistic Variation Yearbook 2004, Vol. 4, ed.
by P. Pica, 23559. Amsterdam: John Benjamins.
, and Verner Egerland. 2009. Impersonal null subjects in Icelandic and elsewhere. Studia Linguistica 63(1).
Speas, Margaret. 1994. Null arguments in a theory of economy of projection. UMass Occasional Papers in Linguistics 17. 179208.
. 2006. Economy, agreement and the representation of null arguments. Arguments and agreement, ed. by
P. Ackema, P. Brandt, M. Schoorlemmer and F. Weerman, 3575. Oxford: Oxford University Press.
Takahashi, Daiko. 2007. Argument ellipsis from a cross-linguistic perspective. Paper read at GLOW in Asia VI, the
Chinese University of Hong Kong.
. 2008. Noun phrase ellipsis. The Oxford handbook of Japanese linguistics, ed. by S. Miyagawa and M. Saito,
394422. Oxford: Oxford University Press.
Tomioka, Satoshi. 2003. The semantics of Japanese null pronouns and its cross-linguistic implications. The interfaces: deriving and interpreting omitted structures, ed. by K. Schwabe and S. Winkler, 32140. Amsterdam: John
Vikner, Sten. 1995. Verb movement and expletive subjects in the Germanic languages. Oxford: Oxford University
Wurmbrand, Susanne. 2006. Licensing case. Journal of Germanic Linguistics 18(3). 175236.

2011 The Author

Language and Linguistics Compass 5/8 (2011): 571587, 10.1111/j.1749-818x.2011.00292.x
Language and Linguistics Compass 2011 Blackwell Publishing Ltd

Copyright of Linguistics & Language Compass is the property of Wiley-Blackwell and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.