Upload
nathan-n
View
214
Download
0
Embed Size (px)
Citation preview
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
1/68
MARA LUISA RIVERO and MILENA MILOJEVIC SHEPPARD
INDEFINITE REFLEXIVE CLITICS IN SLAVIC: POLISH ANDSLOVENIAN
ABSTRACT. In this paper, we argue that certain Slavic reflexive clitics should be analyzedas indefinite defective pronouns in both syntax and semantics, and we go on to identify syn-tactic and semantic parametric variation among reflexive clitics in Slavic. Reflexive cliticsthat correspond to people represent nominative indefinite pronouns in Polish and Slove-nian, and accusative indefinite pronouns in all the Slavic languages, so there is syntacticvariation among such indefinites. In syntax, indefinite clitics stand for explicit argumentsthat are defective because they contain a human feature and no gender, number, or person,and move to repair deficiency. In semantics, they contain human variables and quantifiers
reminiscent of some, which can be deleted by existential disclosure. When quantifiersare deleted and adverbs bind indefinite clitics, such clitics may resemble everyone, manypeople, andfew people. In constructions with datives, Slavic clitics display a complex webof semantic and syntactic variation due to dative existential disclosure in logical form.Dative existential disclosure combines quantifier deletion with one of two operations bind-ing datives to disclosed variables. In Polish and Slovenian, constructions with indefiniteclitics and datives have the same syntax but differ in truth conditions because quantifiersare deleted, and datives bind disclosed variables in one way in Polish and another wayin Slovenian. In Czech and Bulgarian, dative existential disclosure affects reflexive cliticsstanding for implicit arguments with different syntactic properties. Thus, there is syntacticvariation, with Polish and Slovenian forming one syntactic group and Bulgarian and Czechanother. Variation extends to semantics, because the meaning of constructions with reflex-ive clitics and datives is the same in Bulgarian and Slovenian, while Czech belongs to thesame semantic group as Polish.
Research for this paper has had a long gestation, partially supported by Social Sci-ences and Humanities Research Council of Canada Research Grants 410-97-0242 and410-2000-0120 to the first author. Portions were presented by the first author at GoingRomance in Utrecht in December 1998, the Fundacion Ortega y Gasset in Madrid in June1999, the GLOW Summer School in Mytilene in July 1999, the Linguistic Symposiumon Romance Languages 30 in Gainesville in February 2000, the Universities of Brasiliaand Campinas in June 2000, McGill University in Montreal in October 2000, and the 11thColloquium of Generative Grammar in Zaragoza, Spain, in April 2001. Joint work wasread by the second author at the Workshop on Pronouns in Generative Grammar at theSocietas Linguistica Europeae in Ljubljana in July 1999, at Formal Description of SlavicLanguages 3 in Leipzig in December 1999, and at the Theoretical Problems of Pronouns
Seminar of the 5th European Society for the Study of English Congress in Helsinki inAugust 2000. We are grateful to the various audiences for useful comments.Special thanks go to Gennaro Chierchia, source of the ideas in section 4, who has
provided constant inspiration and help for the proposals in section 5, and to Janez Oresnikfor useful comments on many aspects of the paper and for help with Slovenian. We owe
Natural Language & Linguistic Theory 21: 89155, 2003. 2003Kluwer Academic Publishers. Printed in the Netherlands.
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
2/68
90 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
1. INTRODUCTION
In this paper, we argue that certain Slavic reflexive clitics should be ana-
lyzed as indefinite defective pronouns in both syntax and semantics andwe go on to identify parametric variation among these clitics.A first type of variation is syntactic. Such defective indefinite pronouns
correspond to syntactic subjects in Polish (Pol) and Slovenian (Slo), but notin Bulgarian (Bul) or Czech (Cze). However, they correspond to syntacticobjects in all the Slavic languages. A second type of variation unnoticedin the past distinguishes Pol from Slo. Namely, constructions combiningsubject indefinite pronouns and datives have the same syntax, but differin meaning and truth conditions. To account for this semantic variation,we propose two different semantic operations connecting indefinites todatives. A third type of variation involving syntax and semantics distin-guishes on the one hand Pol from Cze, and on the other hand Slo from other
South Slavic languages. Certain Pol and Cze constructions with reflexiveclitics and datives have different syntax and similar semantics. We arguethat in Pol they contain a subject indefinite pronoun and in Cze an impli-cit argument. Certain Slo and Bul constructions with reflexive clitics anddatives also differ in syntax from each other because in Slo they containan explicit pronoun and in Bul an implicit argument, but their meaningis similar. We account for this web of syntactic and semantic similaritiesand differences as follows. We first argue that indefinite pronouns containan existential quantifier in syntax and semantics, and implicit argumentscontain an existential quantifier in semantics, so the same type of semanticoperations can apply to both. We then distinguish between two such se-
mantic operations. One type connects datives to Pol indefinite pronounsand to Cze implicit arguments, so different syntactic constructions in Poland Cze share the same meaning. A second type of semantic operationconnects datives to Slo indefinite pronouns and to Bul implicit arguments.Thus Slo and Bul may share syntactic constructions with Pol and Cze, butthe shared constructions differ in meaning in the two groups of languages.
Our proposals should be of interest for linguistic theory for severalreasons. First, we present one of the few analyses of Slavic reflexive cliticsinvolving both syntax and semantics. Second, variation is assigned here tothe reflexive clitic, a functional category. Hence it follows that semanticvariation, as well as syntactic, resides in functional categories. Third, we
many thanks to Anna Boron, Robert B. Borsley, Magdalena Goledzinowska, Ewa Jawor-ska, Adam Przepiorkowski, and Ewa Willim for Polish, to Olga Arnaudova for Bulgarian,and to Denisa Lenertova for Czech.
We also thank four anonymous reviewers whose comments have contributed to the ideasin this version.
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
3/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 91
establish parallelisms between explicit indefinites and implicit arguments,arguing that both contain existential quantifiers, and go on to claim thatparametric variation in semantics is limited to UG operations involving
such quantifiers.The basic assumptions that form the background to this paper comefrom generative syntax and formal semantics. As to syntax, we adopt theMinimalist Program (Chomsky 1995, 1999, 2000), with feature checkingtriggering syntactic movement, and the version of the Binding Theoryin Reinhart and Reuland (1993). As to semantics, we follow DiscourseRepresentation Theory (DRT) (Heim 1982; Kamp 1981), in particular thedynamic version of DRT in Groenendjik and Stokhof (1991) and Chierchia(1995a).
The paper is organized as follows. Section 2 introduces in informalterms three types of Pol and Slo constructions with indefinite defectivepronouns in the form of reflexive clitics. In the first type, the clitic cor-
responds to a subject pronoun, in the second to an object pronoun, and inthe third to a subject pronoun resuming a dative. Section 3 discusses themorphology, syntax, and syntactic variation of the first two types, and sec-tion 4 is for their semantics. More precisely, section 3 argues that subjectand object reflexive clitics are Simplex Expression anaphors in the senseof Reinhart and Reuland (1993), because they lack gender, number, andperson, and move to repair this deficiency. This section identifies syntacticvariation in Slavic, showing that reflexive clitics can correspond to subjectpronouns in Pol and Slo but not in Bul or Cze, and to object pronounsin the four languages. Section 4 discusses the semantics of subject andobject reflexive clitics, arguing that they are indefinites with a human vari-
able and an existential quantifier that can be eliminated by an adverb bymeans of existential disclosure. Section 5 is for the morphology, syntax,and semantics of reflexive clitics as resumptive pronouns, and for theirvariation, which can be semantic or combine syntax and semantics. Theproposal is that Pol and Slo subject reflexive clitics that resume dativesbehave like defective indefinite pronouns. In syntax, such resumptives areSE-anaphors; in semantics, they are indefinites with a quantifier eliminatedby a dative. This section discusses parametric variation in Slavic reflexiveclitics with datives, which has gone unnoticed in the past. Pol and Slo differin semantics due to two different rules connecting datives and indefinitereflexive clitics, and other Slavic languages differ from each other in bothsyntax and semantics. Pol and Cze, for instance, share semantics but differ
in syntax because in the first language reflexive clitics stand for pronounsand in the second for implicit arguments, and they both differ from Slo andBul in semantics. Slo and Bul are similar in semantics, but differ in syntax
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
4/68
92 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
because in Slo clitics stand for pronouns and in Bul for implicit arguments.Section 6 concludes the paper.
2. CONSTRUCTIONS WITHI NDEFINITER EFLEXIVEC LITICS INPOLISH ANDSLOVENIAN
In this section, we introduce the data to be discussed, and state why itis of general interest. We also informally outline our analyses and theirconsequences for linguistic theory. We examine three different sets of con-structions with reflexive clitics that we claim stand for defective indefinitepronouns. Such clitics are shown in boldface from now on, and glossedRefl (Reflexive).
2.1. Nominative Indefinites
The first construction we study is traditionally called impersonal and isillustrated in (1):
(1)a. Te
this
ksia zke
bookACC
czytao
readNEU
sie
Refl
z
with
przyjemnoscia .
pleasure
(Pol)
People read this book with pleasure
b. Danes
today
dopoldne
morning
se
Refl
je
be3S
jedlo
eatNEU
jagode.
strawberriesACC
(Slo)
This morning people ate strawberries
c. Tutaj
Here
sie
Refl
pracuje
work3S
sporo.
much
(Pol)
Here people work a lot
d. Tukaj
Here
se
Refl
veliko
work3S
dela.
much
(Slo)
Here people work a lot
In (1), the sieand se clitics have a use with an indefinite reading that we
dub Subject/Nominative Indefinite Human Pronoun, or Nom Indefin-ite. A variety of syntactic and semantic characteristics make such NomIndefinites particularly interesting for linguistic theory. A first syntacticcharacteristic is that impersonal constructions with such a clitic cannot
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
5/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 93
contain an overt Nominative NP, so they seem nominativeless. However,we will see that in Pol and Slo they exhibit several nominative propertiesattributed to the reflexive clitic itself. A second syntactic characteristic is
that the verb in such constructions is invariable, that is, in a default formwithout agreement. In the Past, the verb displays a Neuter (NEU) suffix, asinczyta-oin (1a), and in the Present it is Third Singular (3S), as inpracujein (1c). By contrast, verbs must agree with Nom NPs. One example ofsuch agreement is in (2a) where Pastczyta-adisplays a Feminine Singularsuffix -a. Another example is the so-called middle construction in (2b),where Present prowadza is Plural, and agrees with the Plural NominativeNP:
(2)a. Mania
ManiaNOM
czytaa
readFEM
te
this
ksia zke.
bookACC
(Pol)
Mary read this book
b. Te
these
samochody
carsNOM
prowadza
drive3PL
sie
Refl
atwo.
easily
These cars drive easily
A third important characteristic of the Pol and Slo constructions with NomIndefinites is observed when verbs are transitive, as in (1ab). Such verbsappear with an overt NP in the Accusative (ACC), which indicates that thepattern is active and supports the view that the reflexive clitic is Nom. Insum, Nom Indefinites are interesting for syntactic theory because they areconstituents of active impersonal constructions that cannot manifest overt
or morphological Nom case or verb agreement, and may manifest overt ormorphological Acc case. For semantics, Nom Indefinites are interestingbecause they have an arbitrary (i.e., impersonal or indefinite) readingand must denote indeterminate humans. Semantic variability is anotherinteresting feature of Nom Indefinites. Some denote people in general, asin the translations in (1), and therefore can be close in meaning to universalexpressions likeeveryone. Others denote many people, as in (3):
(3) Jesli
If
sie
Refl
gra
play3S
zle,
badly,
zazwyczaj
usually
sie
Refl
przegrywa.
lose3S
(Pol)
if one plays poorly, one usually loses. Many people who play
poorly lose
In this paper, we account for the syntactic characteristics of Nom In-definites by proposing that they are defective pronouns that lack gender,
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
6/68
94 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
TABLE I
Nom indefinites in Slavic and Romance.
Yes No
SLA:Polish; Slovenian Bulgarian; Czech; Slovak; Serbo-Croatian (?)ROM:Italian; Portuguese; Spanish French; Rumanian
number, and person. Therefore, they cannot surface as ordinary Nom-inatives or trigger agreement on verbs. We account for their semanticcharacteristics by assuming that they contain a human variable and anexistential quantifier similar to a/some that can be deleted by a rule ofexistential disclosure. To illustrate, the indefinite reflexive clitic in (3)may mean many people because the quantifier it contains is removed, andthe adverbzazwyczajusually provides a meaning equivalent to many. In
this way, indefinite reflexive clitics always human in meaning may mani-fest different quantificational forces and thus resemble one, many, and aswe shall see,few.
Another aspect that makes Nom Indefinites interesting for linguistictheory is their parametric variation. Rivero (1999) argues that NomIndefinites separate Slavic and Romance into the groups of Table 1.
In section 3 we show that Nom Indefinites exist in Pol and Slo, wherethey have hardly been studied at all, and to a certain degree in varieties ofCroatian and Serbian (SC) to be mentioned only in passing here. However,they are absent in other languages, including Bul and Cze. The hypothesisthat most Slavic languages lack Nom Indefinites raises an important ques-tion in need of an answer. The problem is that many Slavic languagesdisplay a subset of the impersonal constructions with reflexive clitics foundin Pol and Slo. The question then is what is the appropriate analysis ofreflexive clitics in impersonal constructions in languages that do not haveNom Indefinites. Pol and Slo display two varieties of impersonal construc-tions, those with transitive and those with intransitive verbs, as illustratedin (1). Other Slavic languages display only impersonal constructions withintransitive verbs, as illustrated in (4):
(4)a. Tuk
Here
se
Refl
raboti
work3S
mnogo.
much
(Bul)
Here people work a lot
b. Tankovalo
DancedNEU
se
Refl
a do rana
into the morning
(Cze)
People danced into the morning (Ruicka 1992, (20))
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
7/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 95
If Bul and Cze lack a Nom Indefinite, what is the analysis of clitic seinthe above constructions? In this paper, we argue that such a clitic standsfor an implicit argument present in semantics, not for a Nom Indefinite
present in both syntax and semantics. Our proposal implies that impersonalconstructions with reflexive clitics and intransitive verbs that look identicalin several Slavic languages have two different syntactic analyses. In Poland Slo they contain an explicit Nom Indefinite, which is present in syntaxand semantics, and in other languages they contain an implicit argumentthat is present only in semantics.
In sum, the Nom Indefinite is found most clearly in Pol and Slo, whereits properties are not well known, and is absent in most Slavic languages,with the exception of some varieties of SC. Thus, parametric variation di-vides the family into two groups, with important morphological, syntactic,and semantic consequences explored in this paper. Variation hinges onwhether the reflexive clitic, a functional category, corresponds to a syn-
tactic nominative pronoun or not. In Pol and Slo, it corresponds to such apronoun, and in most of the other Slavic languages, it seems that it doesnot.
2.2. Accusative Indefinites
The second type of construction of interest to this paper is illustrated in(5):
(5)a. Marek
MarkNOM
sie
Refl
bije.
fight3S
(Pol)
Mark fights (other people)
b. Uciteljica,
Teacher,
Janezek
JanezekNOM
se
Refl
spet
again
poriva.
push3S
(Slo)
Teacher, Janezek is pushing (other people) again
In (5), the sieand se clitics have a use with an indefinite reading that wedub Object/Accusative Indefinite Human Pronoun, or Acc Indefinite,and gloss other people. In section 3, we argue that the Acc Indefinites of(5) are very similar to the Nom Indefinites of (1). In syntax, Acc Indefinitesare defective pronouns with a human feature, and no gender, number orperson. Such object pronouns stand for indeterminate humans, again with
arbitrary interpretations, so in section 4 we propose that they contain ahuman variable and an existential quantifier in semantics. The differencebetween Nom and Acc Indefinites is structural case. The sie andsecliticsin (5) stand for syntactic arguments that are Acc and bear Patient/Theme
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
8/68
96 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
roles, and co-occur with Nom NPs that bear Agent roles. Those in (1) areNom.
Acc Indefinites are interesting for linguistic theory for at least two reas-
ons. First, they make the syntactic distribution of Indefinite reflexives moregeneral than previously thought, which has an important theoretical con-sequence. A common view in the generative literature is that the indefiniteuse of reflexive clitics is restricted to subjects, hence to nominatives. Thisidea finds support in Romance languages like Spanish that have a NomIndefinite with the characteristics of (1), but no Acc Indefinite of type (5).This view also seems to find support in most uses of reflexive clitics, sincethey usually fall in the Acc class. For instance, the well-known passive useofsein (6) can be classified as Acc, because the construction has an overtNom NP with a Theme role, a verb that agrees with this NP, and no overtAcc NP:
(6) StariparentsNOM
seRefl
ubogajo.obey3P
(Slo)
Parents are obeyed
Most uses of reflexive clitics in Slavic and the indefinite use in someRomance languages seem to suggest that Indefinites can be defined ex-clusively in syntactic terms, that is, through their case. The idea is that allthe uses of reflexive clitics belong to the Acc class with the exception ofthe indefinite use. Our proposal that (5) contains an Acc Indefinite chal-lenges the received view, and implies that Indefinites cannot be identifiedin purely syntactic terms, namely their case. In this paper, we charac-terize indefinite/impersonal uses through a combination of syntactic andsemantic properties. The binding theory we adopt leads to the conclusionthat in syntax the clitics in (1) and (5) behave like defective pronouns, eventhough the first are Nom and the second are Acc. And we will see thatsemantically these pronouns are indefinite, even though they differ fromeach other in case.
The second feature that makes Acc Indefinites interesting for linguistictheory is their parametric variation. As indicated in Table 1, Nom Indefin-ites are limited to a few Slavic languages and are found in some Romancelanguages. By contrast, Acc Indefinites are found in all Slavic languages,but not in Romance. Variation revolves around whether the reflexive cliticcan be a syntactic Acc pronoun or not. It can be in Slavic, but not in
Romance.In sum, the constructions in (5) and their counterparts in other Slavic
languages display a human pronoun with an Acc reflexive clitic form,which shows that the Indefinite is not restricted to subjects, and suggests
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
9/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 97
that the syntactic property of case may be insufficient to characterizeimpersonal uses of clitics.
2.3. Indefinite Reflexive Clitics with DativesThe third and last pattern discussed in this paper is known as the invol-untary state construction in Polish grammar. We call it here the dativeexistential disclosure construction and illustrate it in (7)(8):
(7)a. Te
this
ksia zke
bookACC
czytao
readNEU
mi
IDAT
sie
Refl
z przyjemnoscia
with pleasure
(Pol)
I read this book with pleasure
b. Jankowi
JohnDAT
pracuje
work3S
sie
Refl
dobrze.
well
John works well
(8)a. Danes
today
dopoldne
morning
se
Refl
mi
IDAT
je
be3S
jedlo
eatNEU
jagode.
strawberriesACC
This morning I felt like eating strawberries
b. Janezu
JohnDAT
se
Refl
zelo
much
spi.
sleep3S
John feels very much like sleeping
The constructions of (7)(8) have the same morphology and syntax asthose of (1), but contain an additional dative (Dat). Like the constructionsof (1), those of (7)(8) do not and cannot contain overt Nom NPs andhave a verb with invariable morphology and no agreement. When the verbis transitive, these constructions also manifest overt Acc NPs, as in (7a)(8a), a fact that indicates that they are active and their clitic is nominative.However, they differ in interpretation from (1) because they lack indefinitereadings. In addition, the Pol and Slo sentences of (7ab) and (8ab) differin interpretation and truth conditions, a fact that has gone unnoticed in theliterature. (7a) (Pol) denotes a past eventuality with the Dat as (involuntary)Agent. (8a) (Slo) is a modalized statement, not a past eventuality, and
the Dat resembles an (involuntary) Experiencer. Thus, the speaker whotruthfully utters the Pol sentence must have read a book, while the speakerwho utters the Slo sentence need not have eaten strawberries. The secondtype of variation discussed in this paper, then, is that constructions such as
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
10/68
98 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
(7) and (8) that combine reflexive clitics and datives have the same syntaxand rather different semantics.
The constructions in (7) and (8) are particularly interesting for syntactic
theory because they contain the Nom version of the defective human pro-noun in (1) and (5), with similar consequences. The presence of such NomIndefinites explains why (7)(8) do not tolerate an overt Nom NP, show noverb agreement, and may contain an Acc NP. In addition, these patterns areparticularly interesting for semantic theory due to their dative, which weargue is the source of both the lack of indefinite reading and the differencein truth conditions between Pol and Slo. We propose that in (7)(8) thedative triggers two semantic operations on the Nom Indefinite. The firstoperation eliminates the quantifier contained in the pronoun and parallelsthe existential disclosure rule that deletes the quantifier contained in (3).The second operation establishes a connection between the dative and theindefinite pronoun and is the source of the semantic difference between Pol
(7) and Slo (8). In Pol, the connecting/binding procedure is reminiscent ofa Left Dislocation strategy, so (7a) compares to Mary1,she1read this bookwith pleasure. Sie and she are resumptive pronouns, and dative mi and
Maryinherit the Agent role of the pronoun. In Slo, the procedure connect-ing dative and Indefinite is akin to control, so (8a) resembles Mary1wishedPRO1to eat strawberries, whereMarycontrols PRO. These pronoun playsthe role of the obligatorily controlled PRO, and dative miplays the role ofcontroller and resemblesMary.
Recall that Nom Indefinites are not found in all Slavic languages, whichraises an apparent problem for constructions with reflexive clitics and dat-ives. The problem is that constructions similar to (7) and (8) are common
in all the Slavic languages. For instance, Cze Jankovi se pracovalo hezkyis similar in form to Pol (7b) and Slo (8b) and is similar in meaning tothe Pol construction: John worked with pleasure. If the Cze pattern doesnot contain a Nom Indefinite, what does it contain? Our answer in section5 is that the Cze se clitic stands for an implicit argument, not for a NomIndefinite. Since Cze constructions of this type have the same meaningas their Pol counterparts, it follows that the semantic rules that apply toNom Indefinites in Pol can also apply to implicit arguments in Cze. Thisis because in our analysis explicit indefinites and implicit arguments bothcontain an existential quantifier deleted by the same semantic rule.
In sum, the Pol and Slo constructions in (1), (5), and (7)(8) sharedefective Indefinite Human Pronouns in the form of reflexive clitics. The
goal of this paper is to explore the morphology, syntax, and semanticsof these pronouns. Another goal is to elucidate the differences betweensyntactic pronouns in Pol and Slo and implicit arguments in other Slavic
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
11/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 99
languages. The paper offers a novel syntactic and semantic analysis of sub-ject Indefinites in Pol and Slo, and of dative constructions based on suchIndefinites. It shows that Indefinites can also function as objects. It devel-
ops the first detailed account of Slo Indefinites, and identifies previouslyunnoticed similarities and differences between Pol and Slo. Set against thegeneral background of Slavic, it distinguishes between languages that havea subject Indefinite and those that do not. It also distinguishes Slavic dativeconstructions based on two new dimensions. One is their semantic type,with one type found in South Slavic and the other elsewhere. The other iswhether they contain explicit arguments, as in Pol, or implicit arguments,as in Cze.
3. INDEFINITES ASS IMPLEX E XPRESSION A NAPHORS
This section provides arguments for the view that the Nom and Acc In-definites in (1) and (5) are defective pronouns that should be analyzed insyntax as Simplex Expression anaphors (SE-anaphors) in the sense of Re-inhart and Reuland (1993). In section 3.1 we enumerate the syntactic andmorphological properties of Nom Indefinites that are important to supportthis hypothesis, in section 3.2 we examine Acc Indefinites, and in section3.3 we develop the technical analysis.
It seems useful to begin by listing in (9) the major constructions withreflexive clitics shared by all the Slavic languages:
(9)a. Intransitive impersonal
b. Reflexive/reciprocal
c. Passive
d. Middle
e. Anticausative
f. Inherent
As (9a) indicates, one general use of reflexive clitics in Slavic is withintransitive verbs in so-called impersonal constructions, which were il-lustrated in (4). The other general uses are as reflexives/reciprocals (10),passives (11) (obsolete in Pol as noted by Siewierska 1988 and others),middles (12), anticausatives (13), and inherents (14):
(10)a. Janek
John
ubiera
dresses
sie.
Refl
(Pol)
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
12/68
100 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
b. Janez
John
se
Refl
oblaci.
dresses
(Slo)
John gets dressed
(11)a. Dom
House
szybko
fast
sie
Refl
zbudowa.
built
(Pol)
b. Ta
This
hia
house
se
Refl
je
is
hitro
fast
zgradila.
built
(Slo)
The/this house was built fast
(12)a. Te
These
samochody
cars
prowadza
drive
sie
Refl
atwo.
easily
(Pol)
These cars drive easily
b. Ta
This
knjiga
book
se
Refl
lahko
easily
bere.
reads
(Slo)
This book reads easily
(13)a. Szklanka
glass
sie
Refl
rozbia
broke
(Pol)
The glass broke
b. Vejabranch
se
Refljeis
zlomila.broken
(Slo)broken
The branch broke
(14)a. Maria
Mary
boi
fears
sie
Refl
Janka.
John
(Pol)
b. Marija
Mary
se
Refl
boji
fears
Janeza.
John
(Slo)
With the exception of intransitives as in (9a), in the Government and Bind-ing framework the above uses of reflexive clitics have been consideredto be Accusative in one guise or another (see Burzio 1986 for an earlyimplementation of this general idea in Italian). This is because all such
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
13/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 101
constructions contain overt NPs that are Nom and verbs that agree withthese NPs, suggesting the accusativity of the reflexive clitic. A related ideain GB is that reflexive clitics as in (10)(14) indicate that the derivation
contains anaphors bound in their Governing Category under principle Aof the binding theory of Chomsky (1981). For the passive use in (11), forinstance, a GB analysis was for the Nom NP meaning house to raise fromobject to subject to check Nom Case, resulting in an anaphoric NP-tracechain with several links, with the clitic as one of them (see Dobrovie-Sorin1998 for a recent implementation of this general idea). From this point ofview, the Nom Indefinite of (1) differs from the uses of (10)(14) in atleast two ways. First, the construction with the Indefinite cannot manifestan overt Nom NP. Second, in the indefinite use the clitic is reminiscent ofa pronoun, not an anaphor. The Acc Indefinite in (5) has received limitedattention, but it also differs from the uses in (10)(14) in at least two ways.First, the construction with the Acc Indefinite can contain an overt Nom
NP, but this NP is an Agent. Second, the Acc Indefinite is also reminiscentof a pronoun, since it denotes a different individual than the Nom NP.The reflexive clitics of (1) and (5) lack antecedents, so they resemblepronouns like English he or one, not anaphors like English himself oroneself.
3.1. Morphological and Syntactic Characteristics of Nom Indefinites
Let us begin by looking at the Nom Indefinite, which has received con-siderable past attention from both syntactic and semantic perspectives (forSlavic, see Kanski 1986, 1992; Ruicka 1992; Franks 1995, among others;for Romance, see Chierchia 1995a; Cinque 1988; Dobrovie-Sorin 1998,among others). However, previous analyses fail to harmonize its syntacticbehavior with its semantic behavior. For instance, Kanski (1986, 1992) de-velops an interesting formal semantic analysis of the Pol clitic, but largelyneglects its syntax, while Ruicka (1992) offers a syntactic analysis in-spired by Cinque (1988), but is less interested in semantics. In this paperwe develop a novel syntactic analysis cast in minimalist terms of the NomIndefinite, combine it with a precise analysis borrowed from formal se-mantics, and identify parametric variation for indefinite uses both in syntaxand in semantics.
In Slavic, Nom Indefinites are found most clearly in Pol and Slo and to
a certain degree in varieties of SC to be mentioned only in passing here (forSC see Franks 1995; Rivero and Sheppard 2001; and footnotes 1, 2, and 3of this paper). However, they are absent in other languages, including Buland Cze.
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
14/68
102 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
In Pol and Slo, the Nom Indefinite displays four symptomatic morpho-syntactic properties. First, it co-occurs with a distinctive case and agree-ment pattern, an aspect that has attracted most of the attention in the past.
Second, it antecedes local, long distance, and reciprocal anaphors. Third,it is a syntactic controller. Fourth, it is semantically human, which hasmorphological consequences. In addition, in Pol the Nom Indefinite occurswith all the classes of predicates that allow overt NP subjects, with similarTheta roles. Let us examine these properties in turn.
3.1.1. Case and AgreementAs far as case is concerned, the Subject Indefinite requires Nom, that is,structural case. Four phenomena support this idea. First, if the verb istransitive, the construction displays an overt Acc NP in affirmative clauses(15a and c) and a Genitive NP (Gen) in negative clauses (15b and d). Thissituation indicates that the overt NP functions as syntactic object, and that
the pattern is active and not passive. If it is assumed that active sentencessuch as (15) must contain Nom because they contain Acc/Gen, then thereflexive clitic is Nom since overt Nom NPs are excluded:
(15)a. Te
this
ksia zke
bookACC
czyta/czytao
read3S/NEU
sie
Refl
z
with
przyjemnoscia
pleasure
(Pol)
People read this book with pleasure
b. Tej
this
ksia zki
bookGEN
nie
Neg
czyta/czytao
read3S/NEU
sie
Refl
z
with
przyjemnoscia .
pleasure
People do/did not read this book with pleasure
c. Stare
parentsACC
se
Refl
uboga/je
obey3S/be3S
ubogalo.
obeyedNEU
(Slo)
People obey/(have) obeyed parents
d. Starev
parentsGEN
se
Refl
ne
Neg
uboga.
obey3S
People do not obey parents
The second property indicating that the reflexive is Nom parallels the first.
The construction can display an overt Acc clitic pronoun together with thereflexive clitic, as in (16). This indicates that the Acc clitic stands for asyntactic object, and that the pattern is therefore active:
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
15/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 103
(16)a. Jesli
If
dziewczynka
little.girl
jest
is
niegrzeczna,
bad
kaze
punish3S
sie
Refl
ja.
her
(Pol)
If a little girl is bad, one punishes her
b. Ce
If
je
be3S
baterija
battery
izrabljena,
dead,
se
refl
jo
her
zamenja
change3S
(Slo)
If a battery is dead, one changes it
The third characteristic that shows that the reflexive clitic is Nom is thatPol Modals with Nom subjects appear with the Indefinite as in (17a and b),and those that disallow Nom subjects cannot as in (17c, d, and e) (Ka nski1986, p. 182). Slo modals are unproblematic but less symptomatic becausethey all take Nom subjects and Nom Indefinites, as in (18):
(17)a. Premier
prime.ministerNOM
powinien
should
byc
beINF
ysy.
baldNOM
(Pol)
The prime minister should be bald
b. Powinno
shouldNEU
sie
Refl
byc
beINF
ysym.
baldINST
People should be bald
c. Mania
MaryNOM
trzeba
must
pracowac
workINF
d. Trzebamust
sie
Reflpracowac.work
e. Trzeba
must
pracowac
workINF
People must work
(18)a. Ucenci
pupilsNOM
morajo
must3PL
biti
beINF
pozorni.
attentiveNOM
(Slo)
Pupils must be attentive
b. Mora
must3s
se
Refl
biti
beINF
pozoren.
attentiveNOM
People must be attentive
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
16/68
104 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
The fourth piece of evidence for Indefinites as Nom elements comes froma comparison ofTough-constructions in Slavic and Romance. Let us beginby looking at Romance. In Romance, Tough-constructions display Nom
subjects, as illustrated with Spanish in (19a). They also have well-formedversions with Indefinites as in (19b), which suggests that such Indefinitesare Nom:
(19)a. Ellos
TheyNOM.MASC
son
are
difciles
difficult
de
of
contentar.
pleaseINF
b. A
At
los
the
quince
fifteen
aos
years
se
Refl
es
is
difcil
difficult
de
of
contentar.
please
At fifteen, one is difficult to please
Now let us look at Pol and Slo Tough-constructions. Such constructionsdo not have overt Nom subjects, as (20ab) illustrate. In (20ab), a NP isscrambled from the embedded clause with the case required by the lowerV, which in Pol is Acc and in Slo is Dat. Since Nom is not available,Tough-constructions with the Indefinite are ill formed, as (20cd) show.The contrast between Slavic and Romance Tough-constructions, then, isanother piece of evidence that reflexive clitics as Indefinites are Nom:
(20)a. Jana
JohnACC
trudno
difficult
jest
is
zadowolic.
pleaseINF
(Pol)
b. Janezu
JohnDAT
je
is
teko
difficult
ustreci.
pleaseINF
(Slo)
John is difficult to please. It is difficult to please John
c. Trudno sie jest zadowolic. (Pol)
d. Teko
difficult
se
Refl
je
is
ustreci.
pleaseINF
(Slo)
One is difficult to please
Having examined case patterns with Indefinites, let us now turn to verbagreement. The agreement pattern of constructions with Nom Indefinites
is unusual. As noted in the introduction, when the clitic is indefinite, thetensed V does not agree with the overt NP, but must appear in an invari-able form that we consider the default setting. This form must be 3S inthe Present tense, and Neu in the Past tense, as illustrated in (15ab).
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
17/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 105
Periphrastic tenses as in Slo (15c) repeated now in (21a) and Pol (21b)are particularly interesting because they combine the two morphologicaloptions for default agreement. Namely, in constructions with periphrastic
tenses the auxiliary must be 3S, and the verb must be Neu:(21)a. Stare
parentsACC
se
Refl
je
be3S
ubogalo.
obeyedNEU
(Slo)
People (have) obeyed parents
b. Te
this
ksia zke
bookACC
by
would3S
sie
Refl
czytao
readNEU
z
with
przyjemnoscia .
pleasure
(Pol)
People would read this book with pleasure
In our view, Nom Indefinites are not found in most Slavic languages.
One piece of evidence for this proposal is that languages without NomIndefinites such as Bul and Cze do not have the above case and agreementpatterns, as shown by (22ab) and (23a and c).1 With reflexive clitics, thegrammatical option in Bul and Cze is an arguably Nom NP that agrees withV, as in (22c) and (23d). The clitic in this case is usually called passivese or middlese, but not indefinitese:
(22)a. V
in
tova
this
ucilite
school
ucenizite
students.the
se
Refl
nakazva.
punish3S
(Bul)
In this school, people punish the students
1 As far as SC is concerned, Rivero and Sheppard (2001) report the judgments in (i),based on several informants of Croatian and Serbian origin. Sentences with the human AccNP are possible for some, subject to dialectal variation (ia). However, Acc non-human NPsare impossible (ib):
(i)a. Roditelje
parentsACC
se
Refl
potuje.
obey3S
(SC)
People obey parents
b. Tu
this
knjigu
bookACC
se
Refl
cita
read3S
sa
with
zadovoljstvom.
pleasure
People read this book with pleasure
Tilburg (1986), however, who discusses SC varieties, lists precise locations in Croatiawhere patterns of type (ib) are found. A careful dialectal study is needed before solidconclusions about the status of each diagnostic of Nom Indefinites in SC can be reached.
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
18/68
106 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
b. Jade
eat3S
se
Refl
jabalkite.
apples.the
c. Jadateat3P
seRefl
jabalkite.apples.the
The apples are eaten/edible
(23)a. Rodice
parents
se
Refl.
posloucha.
obey3S
(Cze)
b. Rodice
parents
se
Refl.
poslouchaj.
obey3P
Parents are/should be obeyed
c. Jablka
apples
se
Refl
jedlo.
ateNEU
d. Jablka
apples
se
Refl
jedla.
ateFEM
The apples were eaten
3.1.2. Nom Indefinites as Binders and AntecedentsThe second set of diagnostics for Nom Indefinites comes from the fact that
they can function as binders and antecedents. The Nom Indefinite cliticcan bind local anaphors, that is, coargument reflexives such as sobie/sebeshown in boldface in (24):
(24)a. Terazsiemysli tylko osobie. (Pol)
b. Sedaj
now
se
Refl
misli
think3S
samo
only
na
on
sebe.
oneself
(Slo)
Now people think only of themselves
The clitic can also antecede long distance anaphors such as the clause-mate possessives in (25a) (Siewierska 1988) and (25b), and those in the
subordinate clause in (26), all of which are shown in boldface:
(25)a. Swoich
POSSGEN
przyjacio
friendsGEN
tak
so
sie
Refl
nie
Neg
traktuje.
treat3S
(Pol)
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
19/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 107
b. Svojih
POSSGEN
prijateljev
friendsGEN
se
Refl
tako
so
ne
Neg
tretira.
treat3S
(Slo)
One does not treat ones friends like that
(26)a. Mysli
think3S
sie,
Refl,
ze
that
swoje
POSSNOM
bedy
errors
(Pol)
sa
are
bardziej
more
usprawiedliwione
justified
niz
than
innych.
of.others
b. Verjame
believe3S
se,
Refl
da
that
so
are
svoje
POSSNOM
napake
mistakes
(Slo)
bolj
more
upravicene
justified
kot
than
napake
mistakes
drugih.
of.others
People think/believe that their own mistakes/errors are morejustified than the mistakes of others
The Slo Indefinite can antecede the reciprocal expression drug drugemueach other in (27) (Bolta 1988, p. 127). Pol lacks an equivalent reciprocalexpression, but sie can also antecede non-clitic reflexives in a reciprocalreading such assobain (28):
(27) Drug
Each
drugemu
otherDAT
se
Refl
prevec
too.much
gleda
look3S
v
in
lonec.
pot
(Slo)
lit., People look to each other in (the) pot too much.
People poke their noses into each others private affairs toooften
(28) Tutaj
Here
sie
Refl
ze
with
soba
ReflINST
rozmawia,
talk3S,
a
and
nie
not
koci.
argue3S
(Pol)
Here people talk, not argue, with each other
Under Reinhart and Reulands Reflexivity Theory (1993), the Indefinite inthe above constructions must indicate an argument position in the syntax.The sentences in (25)(26) do not describe reflexive actions, so thesie/seclitics cannot be markers of extrinsic reflexivity, in contrast to himselfin
John treats himself well. The non-local possessive anaphors are not argu-ments of traktuje/tretiraor mysli/verjame, so cannot reflexive-mark theseverbs. Thus, (25)(26) have two syntactic argument positions: (a) the In-definite indicates the antecedent position; and (b) the long distance anaphor
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
20/68
108 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
occupies the other position. The reciprocal in (27) shows that the clitic isneither an extrinsic nor an intrinsic marker of reflexivity. In this case, too,the clitic signals one argument and the reciprocal is its co-argument.
Current theories of focus (e.g., Rooth 1992 and others later) lead to theconclusion that in (24) the clitic corresponds to a semantic argument. Inthese sentences,sobieandsebereflexive-markmysliandmislirespectively,and as foci under the scope of the adverbs tylkoandsamomust correspondto a semantic argument variable. For Reinhart and Reuland (1993), if apredicate is reflexive-marked, it must also be semantically reflexive: i.e.,two of its coarguments must have identical semantic values. The two coar-guments ofmysliand misliare the clitic and the non-clitic anaphor, andthey must have identical values, so sie andsestand for argument variables.In sum, the sentences in (24) are similar to English Now people think onlyof themselves; the reflexive cliticssieand seare comparable to peopleandmust be used to compute the focus semantic value in such sentences.
We know that some Slavic languages lack Nom Indefinites and in suchlanguages the reflexive cannot function as an antecedent.2 To this effect,consider Cze (29ac) from (Ruicka 1992):
(29)a. Mluvilo
talkedNEU
se
Refl
tan
there
jen
only
o
of
sobe.
oneself
(Cze)
People talked there only of themselves
b. Mluvilo
talkedNEU
se
Refl
o
about
tom
that
navzajem.
with.each other
People talked about that with each other2 Rivero and Sheppard (2001) note that in SC, secannot bind a Possessive anaphor as
in (ia), but can bind a local reflexive as in (ib):
(i)a. Svoju
POSS
decu
childrenACC
se
Refl
uvek
always
slua
listen3S
sa
with
panjom.
attention
(SC)
One always listens to ones children with attention
b. Sada
now
se
Refl
misli
think3S
samo
only
na
of
sebe.
oneself
Now people think only of themselves
However, here too, grammaticality judgments seem to vary. For instance, Ruicka (1992)
mentions that for Ivic (ib) is ungrammatical. An anonymous reviewer reports that examplesof type (ib) are grammatical in some SC dialects that lack the Acc NP called attention toin footnote 1. Such a situation suggests that case and agreement may be less importantdiagnostics that proposed in the text. Variation in SC, then, could prove of theoreticalinterest, but for the moment its empirical basis remains unclear.
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
21/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 109
c. Mluvilo
talkedNEU
se
Refl
tam
there
jen
only
jeden
about
o
each
druhem.
other
People talked there only about each other
Sentence (29a) is similar to (24): a local reflexive anaphor is under focus,but cannot be bound by cliticse. (29b) shows that secannot co-occur withthe reciprocity adverb navzjem. (29c) is comparable to Slo (27) with areciprocal. The contrast with Pol/Slo suggests that Cze se does not standfor an argument variable in a syntactic position. In our terms, Cze is oneof the Slavic languages without Indefinitese.
3.1.3. Control by Nom IndefinitesThe third diagnostic in favor of the hypothesis that Nom Indefinites aresyntactic arguments is that they can function as controllers. Several lin-
guists have noted that Pol Indefinite sie
can control into adjuncts, as (30)illustrates (and see Dziwirek 1994 for references and discussion). We addthat Slo is similar to Pol in this respect:
(30) Te
this
ksia zke
bookACC
czytao
readNEUT
sie
Refl
siedzac
sitPART
przy
by
kominku.
fireplace
One read this book sitting by the fire (Dziwirek 1994)
However, it is important to distinguish between control into adjuncts asin (30), and control into complements as in (31). On the one hand, itis not clear if the controllers of adjunct clauses are implicit or explicit
arguments, which means that (30) does not provide strong evidence for theidea that Nom Indefinites are explicit/syntactic arguments. On the otherhand, Jaeggli (1986) proposes that controllers of passive complements aresyntactic arguments, which means that (31) provides the evidence neededto conclude that Nom Indefinites are explicit arguments:3
(31)a. Chce
want3S
sie
Refl
byc
be
lubianym.
liked
(Pol)
People want to be liked
3 Rivero and Sheppard (2001) report that control into passive complements gives rise to
conflicting judgments in Serbian, but seems always possible in Croatian. The anonymousreviewer mentioned in the first two footnotes reports that the SC dialect without the Accpattern and with binding of local anaphors also allows control into passive complements.Such a situation, if it occurs, could signify that case and agreement are not the most crucialdiagnostics of Nom Indefinites.
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
22/68
110 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
b. Vedno
always
se
Refl
eli
want3S
biti
be
obcudovan.
admired
(Slo)
People always want to be admired
3.1.4. The Human Feature in Nom IndefinitesA property of Indefinites supporting the idea that they are defective pro-nouns with restricted content has been noted by many: the Indefinitedenotes a human or personified being, like arbitrary PRO. Verbs whosemeanings do not imply human participation can serve to illustrate thisproperty, as in (32):
(32)a. Veter
Wind
je
be3S
pihal.
blownMASC
(Slo)
The wind was blowing
b. Pihalo
blownNEU
se
Refl
je
be3S
od
from
jeze.
rage
People were fuming/seething with rage
The verb in (32) may denote a meteorological event as in (32a), so is notintrinsically connected to human activities. However, if this verb combineswith Indefinite se as in (32b) (Bolta 1988, p. 123), the reference mustbe to people. The human feature is central to identifying the Indefinitewith intransitives because such verbs fail to display most of the otherdiagnostics. That is, intransitives disallow overt Acc/Gen NPs, binding of aco-argument anaphor, and control into a complement clause. In the absenceof these other clues, the human feature in intransitive constructions maysignal an Indefinite.
In sum, Pol and Slo share four diagnostics that we consider importantsymptoms of the Indefinite: case/agreement, binding/antecedence, control,and the human feature. Pol and Slo constructions with reflexive clitics dis-play the listed diagnostics because these languages have Nom Indefinites.By contrast Bul and Cze constructions with reflexive clitics cannot displaythe same diagnostics because these languages do not have Nom Indefinites.Thus, there are two syntactic and morphological types of Slavic languages:
those with Nom Indefinites and those without.As noted in the introduction, the conclusion that some Slavic languages
lack Nom Indefinites seems to raise a problem in need of an answer. Theproblem is that the four Slavic languages that we divide into two groups
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
23/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 111
share apparently identical impersonal constructions with reflexive cliticsand intransitive verbs, as (33ad) illustrate:
(33)a. TutajHere
sie
Reflpracujework3S
sporo.much
(Pol)
Here people work a lot.
b. Tukaj
Here
se
Refl
veliko
work3S
dela.
much
(Slo)
Here people work a lot.
c. Tuk
Here
se
Refl
raboti
work3S
mnogo.
much
(Bul)
Here people work a lot.
d. Tankovalo
DancedNEU
se
Refl
a
into
do
the
rana
morning
(Cze)
One danced into the morning. (Ruicka 1992, (20))
To account for impersonal constructions with intransitive verbs in (33ad), we borrow an idea of Dobrovie-Sorin (1998) for Romance: namely,such impersonal constructions have two different analyses. In our view,Pol (33a) and Slo (33b) contain a Nom Indefinite, which is a defectivepronoun present in both syntax and semantics. By contrast, Bul (33c) andCze (33d) do not contain a Nom Indefinite but an implicit argument thatis present in semantics and not syntax. In section 5, we motivate such ahypothesis.
3.1.5. Passive and Copular Constructions in PolishA last characteristic of Nom Indefinites is that they resemble ordinarypronouns in that they can combine with many predicate classes, and bearTheta roles for overt Nom subjects. While this characteristic is found bothin Pol and Slo, Slo Indefinites are more restricted in their distribution thanPol Indefinites. In Pol, Indefinites are found in passive and copular con-structions, as illustrated in (34ab). Slo native speakers differ in judgments,
but passive and copular constructions with Indefinites are often consideredungrammatical or marginal, as illustrated in (35ad). Another source ofdisagreement in Slo is the gender of adjectives in copular constructions.The Slo adjectives illustrated in (35c and d) are masculine like their Pol
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
24/68
112 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
counterparts in (34b), but some Slo speakers prefer in such constructionsneuter adjectives:
(34)a. Bywabe3S
sie
Reflkaranym
punishedMASC.INSprzezby
przyjacio.friends
(Pol)
From time to time people are punished by friends
b. Kiedy
when
sie
Refl
byo
wasNEU
modym,
youngMASC.INS
sie
Refl
byo
wasNEU
szcze sliwym.
happyMASC.INS
When one was young, one was happy
(35)a. Odfrom
casatime
doto
casatime
se
Refl
jebe3S
kaznovanopunishedNEU
odby
prijateljev.friends
(Slo)
From time to time people are punished by friends
b. ? Kadar
when
se
Refl
je
be3S
bilo
beNEU
sprejeto
received
pri
at
upanu,
mayor,
je
be3S
bilo
beNEU
treba
must
nositi
wear
bele
white
rokavice.
gloves
When people were received by the mayor, they had to wear
white gloves
c. Ko
when
se
Refl
je
be3S
bilo
beNEU
mlad,
youngMASC
se
Refl
je
be3S
bilo
beNEU
srecen.
happyMASC
When one was young, one was happy
d. ? Nikoli
Never
se
Refl
ni
not.be3S
srecen.
happyMASC
People are never happy
Above we concluded that in some Slavic languages there is no Nom Indef-inite. In such languages, reflexive clitics are clearly impossible in copularand passive constructions, and the gender of adjectives in copular con-structions does not influence grammaticality judgments. The clear contrast
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
25/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 113
between languages without Nom Indefinites and Pol can be illustrated byBul (36), which differs in grammaticality from the Pol passive in (34a):
(36)
Cestooftense
Reflebe3Spredadenbetrayed otby prijateli.friends (Bul)
People are often betrayed by friends
Pol passive and copular constructions are significant for our proposalsfor two reasons. First, participles and adjectives in such constructions es-tablish a morphological connection between the Indefinite and arbitraryPRO, which are both human. The Pol Indefinite usually co-occurs with amasculine singular adjective or participle in the Instrumental (INST), asillustrated in (34ab). Nom is excluded in this context. Likewise, Pol PROusually co-occurs with a masculine singular adjective or participle in Inst,
not Nom, as illustrated in (37):
4
(37) Byc
beINF
lubianym/
lovedINST/
lubianylovedNOM
jest
is
mio.
niceNEU
(Pol)
It is nice to be loved
Indefinite and PRO lack phi-features, so we must conclude that in Pol theconcord patterns in (34b) and (37) represent the default morphological set-ting for a human. Thus, default morphology always associates with NomIndefinites, which can nevertheless function as plurals in semantics, as (38)illustrates:
(38) Tuhere
sie
Refl
przychodzicome3S
weby
dwoje.twoNEU
(Pol)
Here people come two by two
Copular constructions are significant from a second perspective. Whenthey contain individual-level predicates such as intelligent, they bring anindividual variable into the semantic representation (and perhaps lack asituation variable) (see Carlson 1977; Kratzer 1989, among others). On thisview, the reflexive clitic that co-exists with an individual-level adjective
4 An anonymous reviewer suggests that predicative adjectives are Nom in Pol when
they can agree with a nominative item, lexical or not; the question then becomes whythe adjectives in (34) must be Inst when the reflexive clitic is Nom. With a small clauseanalysis, the adjectives in (34) could be in a nominativeless embedded clause (perhapswith subject PRO), and absence of phi-features would be the factor that accounts for theirconcord pattern, irrespective of the case on the reflexive clitic.
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
26/68
114 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
as in (34b) is a diagnostic for an individual variable. Combined with theother diagnostics, such a variable corresponds to a syntactically projectedNP that, as semantic argument of the predicate, can bind, antecede, and
control. Thus, the four diagnostics shared by Pol and Slo together withcopular constructions in Polish provide solid morphological, syntactic,and semantic grounds to postulate a syntactic or explicit Indefinite HumanPronoun in the form of a reflexive clitic.5
In sum, the most important characteristics of the Nom Indefinite are thefollowing:
(39)a. It must be represented by a reflexive clitic, which is a functionalcategory.
b. It combines with an Acc NP as in (15ad), or with an Acc clitic,as in (16ab). It also combines with a Modal that takes Nom
subjects, as in (17b). It is absent from Tough-constructions,which lack Nom, as in (20cd). Thus, it bears Nom, a structuralcase.
c. It combines with a tensed verb or auxiliary in a default form,that is without agreement, as in (15ad), etc.
d. It can bind or antecede many types of anaphors, thereby indic-ating that it is an argument in an A-position in the syntax, as in(24ab), among other sentences.
Long distance anaphors as in (26ab) and reciprocals as in(27) are particularly significant because they indicate the need
for two syntactically projected positions, one for the Indefinite,and one for the long distance anaphor or reciprocal. Focus as in(27) is also significant because it indicates that the syntactic slotof the Indefinite must contain a semantic argument variable.
e. It functions as a syntactic controller, as in (31ab).
f. It has a human feature and no phi-features, and imposes thesame default morphological concord pattern as arbitrary PROon adjectives and participles, as in (34ab).
5 Rivero (2000) suggests that Slo Indefinite se must be in construction with asituation/event variable and is marginal with passives and copular constructions withadjectives, which seem to lack this type of variable.
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
27/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 115
g. Like ordinary Nom NPs, including indefinites and Nom pro-nouns, it can be used with many types of predicates andassociate with a variety of external and internal Theta roles.
Constructions with individual-level adjectives as in (34b) showthat the Indefinite contains an individual variable in semantics(perhaps in the absence of a situation variable).
Slavic languages fall into two groups as to Nom Indefinites as in (39).Nom Indefinites are found in Pol and Slo, but not in Bul and Cze, so thereis parametric variation among such Indefinites.
3.2. Characteristics of Accusative Indefinites
In this section we examine Acc Indefinites:
(
40)
a. MarekMarkNOM
sie
Reflbije.fight3S (Pol)
Mark fights (other people)
b. Nie
neg
pchaj
pushIMP.2S
sie
Refl
pan!
man
(Pol)
Stop pushing (others), man! (Kanski 1986)
c. Uciteljica,
Teacher,
Janezek
JanezekNOM
se
Refl
spet
again
poriva.
push3S
(Slo)
Teacher, Janezek is pushing (other people) again
d. Pokai
show2S
kako
how
se
Refl
poljublja
kiss2S
(Slo)
Show me how you kiss (other people)
Acc Indefinites have also been labeled antipassives (Kubinski 1987),generic reflexives (Ruicka 1992), null object impersonals (Rivero1999), and right-oriented middles (Kanski 1986). However, they haveattracted by far less attention than Nom Indefinites and have not beenanalyzed in a way that harmonizes their syntactic and semantic properties.
We view Acc Indefinites as object counterparts of Nom Indefinites. UnlikeNom Indefinites, which as we saw are not found in some Slavic languages,Acc Indefinites are found in all of them limited to a few verbs. As far asregister is concerned, Acc Indefinites are often described as colloquial, or
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
28/68
116 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
an aspect of child language. This is true at any rate for Bul (Dimitrova-Vulchanova 1996, p. 143), for Pol (Kanski 1986; Kubinski 1987), and forCroatian and Serbian (Progovac 1998, p. 108). In Slo, however, both the
Nom and the Acc uses seem standard; indeed it is the Nom use that issometimes criticized by prescriptive grammarians.Acc Indefinites share four properties with Nom Indefinites. The first
is that they must also bear structural case, namely Accusative. The con-structions in (40) contain overt (or null) Nom NPs with an Agent role,and transitive verbs that agree with these Nom NPs, but cannot contain anovert Acc (or Gen) NP. In our view, overt Acc NPs must be absent in suchconstructions because the reflexive clitic is Acc and corresponds to theobject of the transitive verb or, more technically, to the syntactic internalargument of a two-argument predicate that bears the Theme role. On thisview, the constructions in (40) are active sentences with two syntacticarguments. One argument is the overt (or covert) NP that is Nom and is
the syntactic subject, and the other is the Acc object as reflexive clitic.The second property that Acc Indefinites share with Nom Indefinites is
that they resemble pronouns rather than anaphors. In the Binding Theory ofChomsky (1981), the reflexive clitics of (40ad) are expressions that fallunder principle B, that is, pronouns, because they are free in their Gov-erning Category and not A-bound by the Nom subject. For instance, thegloss indicates that NomMarekin (40a) does not bind an object, and is notcoindexed with clitic sie. Such a clitic, then, is comparable to a pronoun,not an anaphor. In the Reflexivity Theory of Reinhart and Reuland (1993),the sentences in (40ad) do not describe reflexive actions, but actions in-volving two different sets of participants. Thus, the verbs of such sentences
are not reflexive-marked and their clitics cannot be reflexivizers, and weargue later that they are defective pronouns or SE-anaphors. In additionto their binding characteristics, Acc Indefinites are objects of verbs thatdo not favor reflexive readings and facilitate pronominal readings becausethey often describe actions detrimental to the Agent. Pol verbs used withAcc Indefinites include kopackick,pchacpush,bicbeat, and drapacscratch, and Slo verbs include porivati push, tepsti beat, and griztibite. Such verbs favor an interpretation where the reflexive clitic meansother people, not himself. For instance, an anaphoric reading such asTeacher, Janezek is pushing himself again is possible in (40c), but it is aless natural interpretation than the indefinite reading other people.
The third property that Acc Indefinites share with Nom Indefinites is a
defective character. Like Nom Indefinites, Acc Indefinites denote a humanor sentient being and lack phi-features, and this is why they cannot surfaceas regular overt NPs with Acc case. We gloss Acc Indefinites by means
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
29/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 117
of other people, but they are neutral with respect to number and are alsocompatible with plurality. They can, for instance, refer to the speaker, asKanski (1986, p. 195) reports for (40b), or to an indeterminate group of
humans, as in (40cd).The fourth property that Acc Indefinites share with Nom Indefinitesis semantic variability, a sign of indefiniteness discussed in section 4. Asnoted in section 2, Nom Indefinites may be equivalent in meaning to Eng-lish expressions such as someoneor everyone. The Object Indefinite mayalso mean someone or everyone, depending on context. Kanski (1986,p. 195) considers that in (40b) the reflexive clitic stands for the speaker.Many have noted that Indefinite reflexive clitics may denote a speaker ora hearer, but the phenomenon is not well understood. However, it can besuggested that such readings derive from sie as an existential expressionmeaning someone: Do not push someone (who happens to be me)!Bycontrast with this existential reading, past sentences with imperfective
aspect favor for Acc Indefinites the universal reading everyone, as in(41ab):
(41)a. Kiedy
when
by
was
may
little,
Janek
Janek
sie
Refl
strasznie
horribly
drapa
scratchIMP
(Pol)
When he was little, John would scratch everyone horribly
b. Ko
when
je
is
bil
been
majhen,
little
se
Refl
je
is
Janez
Janez
grozno
terribly
grizel.
bittenIMP
(Slo)
When he was little, John would bite everyone horribly
Rivero (1999) suggests that imperfective aspect affects Acc Indefinitesin the same way quantificational adverbs affect Nom Indefinites. That is,Nom Indefinites display semantic variability because quantificational ad-verbs are operators that eliminate their existential quantifier and providethem with different forces, including universal force. Likewise, Acc Indef-inites display semantic variability because imperfective aspect in sentencesof the type of (41ab) is an operator that eliminates their existentialquantifier and provides them with the universal content of everyone.
In addition, Rivero (1999) suggests that Acc Indefinites are the overtcounterparts of object arbitrary little proin Italian (see Rizzi 1986). Kim(1991) proposes that such arbitrary little pro is a (Heimian) indefinite.From all of this, it can be concluded that indefinite versions of pronouns
exist for objects. In sum, Acc Indefinites resemble Nom Indefinites instructural case, in being defective pronouns with a human feature andno phi-features, and in displaying quantificational variability as a sign ofindefiniteness.
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
30/68
118 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
3.3. The Analysis of Indefinites as Simplex Expression Anaphors
In this section we propose a syntactic analysis cast in current minimalistterms for the Nom and Acc Indefinites in order to capture their morpho-
syntactic properties. For Rivero (1999), subject and object Indefinites area (new) kind of Simplex Expression anaphor (SE-anaphor), as in Reinhartand Reuland (1993). That is, in their syntactic representation, Nom andAcc Indefinite constructions contain an explicit null defective pronoun inthe VP with a human feature, but no gender, number, and person. This nullpronoun moves in the syntax to repair its deficiency and raises from theVP to the Specifier of a so-called base-generated reflexive clitic phrase,which results in a Simplex Expression anaphor chain. Here, we build onthis syntactic proposal.
3.3.1. Reflexivity TheoryWe first introduce basic concepts from Reinhart and Reuland (1993). Theydistinguish between the three types of expressions in (42) by means of thefeatures Refl(exivizing), and R(eferential independence):
(42)a. Pronouns [Refl; +R].
b. SELF-anaphors [+Refl; R].
c. SE-anaphors [Refl; R].
Pronouns are non-reflexivizers and contain phi-features, allowing themto be interpreted independently. SELF-anaphors are reflexivizers andlack phi-features. SE-anaphors are hybrid between pronouns and SELF-anaphors. The pronoun inJohn1 hates him2contains phi-features and doesnot reflexive-mark the predicate. By contrast, the SELF-anaphor in John1hates himself1reflexive-marks the predicate and does not contain a full setof phi-features, so cannot be interpreted independently.
The principles stated in (43) regulate coargument relations:
(43) Reflexivity Conditions:
a. Principle A: A reflexive marked predicate must be semanticallyreflexive.
b. Principle B: A semantically reflexive predicate must be reflex-
ive marked.
They ensure that the licit interpretation ofJohn1 hates himself1 is withthe two NPs coindexed. The SELF-anaphor reflexive-marks hates, which
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
31/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 119
thus must be semantically reflexive. That is, John and himself are itscoarguments and must have identical semantic values.
Let us now examine SE-anaphors, absent in English but present in
Dutch, as in (44ab):(44)a. Max
Max
legt
put
het
the
boek
book
achter
behind
zich.
SE-anaphor
(Dutch)
Max1puts the book behind him1
b. Max
Max
haat
hates
zich.
SE-anaphor
Like SELF-anaphors, SE-anaphors lack phi-features, cannot be interpretedindependently, and obtain content by chain formation under (45) (Reinhartand Reuland 1993, principle (80)):
(45) General Condition on A-chains:A maximal A-chain (a1, . . . an) contains exactly one link a1that is both +R and Case-marked.
In (44a), the SE-anaphor zichrepairs referential deficiency by adjoiningto Inflection in LF, which coindexes it with the subject, inheriting its phi-features. The resulting chain is well formed, since its head is Case-Markedand [+R], and its tail is [R]. But similarly to pronouns, SE-anaphors donot reflexive-mark a predicate. On this view, (44b) is deviant because itviolates principle B in (43). The predicate hatesis semantically reflexivesince its coarguments Maxand zichare coindexed, but is not inherentlyreflexive in the lexicon or reflexive-marked in syntax. In sum, a SE-
anaphor is a defective pronoun that acquires phi-features via the movementregulated by (45) that coindexes it with the subject.
In our view, Pol and Slo Indefinites are SE-anaphors for two reasonsstated now and discussed below. The first is based on chain formationand Condition (45): Indefinite reflexive clitics represent defective pronounsthat obtain the content for their interpretation via movement. The secondis Reflexivity Theory and the principles of (43): Indefinite reflexive cliticsare not reflexivizers. Certain Slavic reflexive clitics, then, may functionas SE-anaphors. However, we shall also see that the analysis of Indefin-ites as SE-anaphors is not identical to the one proposed in Dutch for theSE-anaphor zich.
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
32/68
120 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
3.3.2. Indefinite Reflexive Clitics and Chain FormationLet us examine how chain formation and Condition (45) apply to Indefinitereflexive clitics by first looking at the Nom Indefinite in Pol (15a) and Slo
(15c), partially repeated in (46ab):(46)a. Teksia zkeczytaosiez przyjemnoscia .
People read this book with pleasure
(Pol)
b. Stareseuboga.
People (should) obey parents
(Slo)
The basic skeleton of the clause that we assume for (46) is (47):
(47)
In (47), the lexical verb heads a VP with two arguments. The Indefinite isthe external argument NP1, which is a null defective pronoun with a humanfeature, but no phi-features. We assume that this NP has an uninterpretableformal feature which is (structural) Nom case (see Chomsky 1999, 2000).The internal argument NP2 is the overt Acc NP, which in (46a) isteksia zkeand in (46b) is stare. The Tense Phrase (TP) is headed by Tense (T), whichis defective because it lacks phi-features. TP takes VP as complement andV checks features against T. However, since T is defective, V must beeither 3S or NEU. The other functional projection Clitic Phrase (CLP)is headed by the reflexive. The core idea is that in (47), NP1, which is asyntactically projected argument of the predicate for the reasons given insection 3.1, must repair deficiency by raising to CLP, checking structuralcase. Thus, NP1 in the indefinite construction contrasts with littleproof the
ordinary null subject sentence, which is often assumed to check featureswith a non-defective T in TP.
The clitic in (47) can match and check the structural case of a phi-less Nom NP as subject Indefinite, since it has a structural case feature
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
33/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 121
unspecified for a value. That the clitic has such an unvalued feature canbe justified on the basis of morphology. Nom and Acc/Gen are the twostructural cases, and Sloseand Pol siedisplay the same form, whether they
play a Nom or an Acc/Gen role. Pol has no dative reflexive clitic, and theSlo dative reflexive is si, but lacks an indefinite use. Thus, if dative is nota structural case, then it can be assumed that the Slo se and Polsieformsare reserved for structural case. If it is further assumed that the reflexiveclitic is unspecified for case, then it can check the case of the two defectivepronouns that raise: the external one, which is Nom, and the internal one,which is Acc.
On this view, the clitic in the subject patterns in (46) attracts NP1 tocheck structural case. As stated, clitic and NP are each equipped with case,and match for checking. NP1 and T are defective: they lack phi-features.Following Chomsky (1999, 2000), their defective character prevents theestablishment of an appropriate checking relation between the two. If T
has an EPP feature and attracts NP1, T will not be able to remove the casefeature on the NP, so the appropriate checking relation is established whenNP1 moves up to the clitic. Thus, absence of phi-features, a prominentaspect of the Indefinite construction, makes the NP raise to a clitic with astructural case feature to satisfy formal needs: feature checking for case.The chain resulting from this movement is well formed according to thegeneral condition on A-chains in (45). On the one hand, the chain has caseon its upper link and its head counts as [+R] for reasons provided later insection 4. On the other hand, the tail of such a chain is [R], or a defectivepronoun without phi-features. Acc Indefinites as in Pol (40b) and Slo (40c),partially repeated in (48ab), receive a similar analysis:
(48)a. Nie pchajsie, pan!
Stop pushing (others), man!(Kanski 1986)
(Pol)
b. Uciteljica, Janezeksespet poriva.
Teacher, Janezek is pushing (other people) again!
(Slo)
In the object use, NP1 in (47) corresponds to little proin Pol (48a) and tothe Nom subjectJanezekin Slo (48b), and NP2 is the defective null objectpronoun that raises to the clitic. The VP in (47) is shown with a simplifiedstructure, but Chomsky (1995) and many others adopt an analysis of trans-
itive Vs with two layers of structure. The upper layer contains a light verbwith the subject, and the lower layer contains the lexical V with the object,as in [vPNP1 v [VPV NP2]]. With such a layered structure, an ordinary ob-ject NP that is Acc usually checks features against v. Under the assumption
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
34/68
122 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
that the constructions with object Indefinites in (48) also contain a layeredVP, then in their derivation NP2 in (47) must raise to the clitic, crossing vand T. We propose that NP2 raises directly to the clitic because it has no
phi-features, so it cannot establish the appropriate checking relation withthe light v. In other words, NP2 is defective, so it cannot be what Chomskycalls the active goal for little v: i.e., its features cannot be checked againstthose of v. NP2 is equipped with Acc case as uninterpretable feature andmoves up to the clitic to remove it. The reflexive clitic in (47) can matchand check the structural case of a phi-less Acc/Gen NP2 because it bearsan unspecified structural case feature. The movement of NP2 up to theclitic results in a chain that complies with the general condition on chainsin (45). On the one hand, the resulting chain has case in its first link thatis Acc. On the other hand, we postpone until section 4 the justification ofthe idea that the head of this chain is [+R], but it should be clear the tail is[R] since it contains a defective pronoun without phi-features.
In sum, clitics sie and seare entries of the lexicon with unspecifiedstructural case. They are merged in syntax in the CL position. NPs thatlack phi-features and can be external or internal arguments raise out ofthe VP to check case against this clitic, which results in a well-formedSE-anaphor chain.
On this analysis, Dutch and Slavic SE-anaphors exhibit two similaritiesand one difference. The first similarity is lack of phi-features, which makesboth types of expressions referentially defective. The second similarityis that raising results in a well-formed chain that provides the necessarycontent for interpretation, and hence repairs referential deficiency. Thedifference concerns how movement in each case repairs such deficiency.
On the one hand, Dutch SE-anaphors move and inherit phi-features fromthe subject, which repairs their referential deficiency by making them sim-ilar to definite pronouns like English him. On the other hand, SlavicSE-anaphors contain a human feature and move to a projection headedby a reflexive clitic without phi-features, so they do not repair referentialdeficiency by inheriting phi-features from another syntactic constituent. Insection 4, we will argue that Slavic SE-anaphors also repair their referentialdeficiency by movement because they form chains whose head is equiva-lent to an existential quantifier like some, and whose foot is equivalent toa variable restricted to humans. The intuition behind such a proposal isthat in (47), the syntactic position CL provides force, and NP1 for NomIndefinites or NP2 for Acc Indefinites provides the human feature, and the
chain is well formed because it can be interpreted like an indefinite likeEnglishsomeoneorsomebody.
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
35/68
INDEFINITE REFLEXIVE CLITICS IN SLAVIC 123
3.3.3. Indefinite Reflexive Clitics and ReflexivityThe principles of Reflexivity Theory in (43) also tell us that Nom and AccIndefinites are SE-anaphors. Recall that similar to pronouns, SE-anaphors
do not mark a predicate reflexive. On this view, then, the Nom Indefinite inSlo (46b) is comparable to EnglishHeinHe obeys (his) parents, because itdoes not mark the predicateubogaobeys reflexive. The Acc Indefinite inSlo (48b) is comparable to English themin John is pushing thembecauseit does mark the predicate spetpush reflexive.
In Dutch, SE-anaphors are pronouns coindexed with the subject be-cause they move to Inflection and obtain phi-features, forming a chain. InSlavic, SE-anaphors are also pronouns, but they move to a reflexive cliticforming a chain, are not coindexed with an overt NP, and do not obtainphi-features.
To conclude, the Indefinite is an entry of the lexicon that is a functionalcategory and has the form of a reflexive clitic. It has an unspecified struc-
tural case feature and is merged in CL in the structure in (47). A null NPwithout phi-features, a human feature, and an uninterpretable structuralcase feature is merged in VP as either NP1 or NP2, and checks case bymoving to this clitic. This results in a well-formed chain with a humanfeature.
4. THE SEMANTICS OFINDEFINITEREFLEXIVECLITICS
In this section, we examine the semantics of Indefinites and pair the syn-tactic analysis proposed in section 3 with a formal semantic one. One ideais that the analysis in which Indefinites are defective pronouns triggeringmovement in the syntax relates to the semantic proposal where they con-tain an existential quantifier and a human variable. These two analyses canbe unified under the assumption that the trace/copy of the human pronounthat raises in the syntax corresponds to the human individual variable inthe semantics, and the reflexive clitic in ClP in the syntactic structure cor-responds to the existential operator that in semantics binds such humanvariables. Thus we unify the idea that defective pronouns/SE-anaphorsmust trigger movement in the syntax and the view in dynamic semantics
that indefinites are inherently quantificational. Another important idea isthat the quantifier in Indefinites can be eliminated by a rule called exist-ential disclosure, which partially accounts for why they mean many andfew.
8/10/2019 Rivero Sheppard Indefinitve Reflexive Clitics in Polish and Slovenian
36/68
124 MARIA LUISA RIVERO AND MILENA MILOJEVIC SHEPPARD
4.1. The Logical Form of Indefinite Reflexive Clitics
We begin by looking at the logical form of Indefinites. Chierchia (1995a)proposes that in semantics, the Italian reflexive clitic siin constructions
comparable to (1ad) must be treated as a pronoun that is indefinite andhas human denotation. In what follows, we extend this proposal to Pol sieand Slose.
Let us first motivate the idea that these clitics share an important char-acteristic with indefinite NPs. A prominent diagnostic of indefinite NPsstudied extensively in DRT is quantificational variability (see Heim 1982and Kamp 1981, and other later work). To illustrate, in conditional sen-tences like If an Italian is tall, he is always/often blond, the force of theindefinitean Italianvaries and seems to be determined by adverbs of quan-tification (Q-adverbs) such as alwaysandoften. The sentence withalwaysis similar in truth conditions to Every Italian who is tall is blond, so the
indefinite has universal force. The sentence with oftenis similar to ManyItalians who are tall are blond, so the indefinite means many. Indefinitereflexive clitics share this characteristic. In conditional sentences, the forceof such Indefinites varies and seems to be determined by Q-adverbs, as theinterpretations of (49)(51) indicate:
(49)a. Jesli
If
sie
Refl
gra
plays
zle,
badly
zawsze
always
sie
Refl
przegrywa.
loses
(Pol)
b. Ce
If
se
Refl
igra
plays
slabo,
badly.
se
Refl
vedno
always
izgubi.
loses
(Slo)
If one plays poorly, one always loses. Everybody who playspoorly loses
(50)a. Jesli
If
sie
Refl
gra
plays
zle,
badly
zazwyczaj
usually
sie
Refl
przegrywa.
loses
(Pol)
b. Ce
If
se
Refl
igra
plays
slabo,
badly,
se
Refl
navadno
usually
izgubi.
loses
(Slo)
If one plays poorly, one usually loses. Many people who playpoorly loose
(51)a. Jesli
If
sie
Refl
gra
plays
zle,
badly,
rzadko
seldom
sie
Refl
przegrywa.
loses
(Pol)
8/10/2019 Rivero Sheppard Indefinitve Re