DOI: 10.1515/qal-2016-0001
ISSN: 2299–8942
What does żeby introduce? Old and new research questions about
the Polish żeby complementizer
Marcin Orszulak
ABSTRACT The article explores the status of żeby-clauses in the Polish mood system. Clauses introduced by the
complex complementizer żeby defy a unified description and traditional Polish grammars classify them either as
indicative or conditional/subjunctive. The paper provides evidence that żeby-clauses can be classified as irrealis on
the basis of their temporal properties. Furthermore, żeby-clauses introduced by non-veridical verbs share affinities
with the subjunctive structures in other languages; specifically, they exhibit subjunctive-related phenomena, such as
obviation effects. The present discussion also gives more general insights into the traditional mood classification in
Polish and its problems.
Keywords: mood, irrealis, subjunctive, complementizer, żeby, subordinate clause
1
Introduction
Żeby-clauses constitute a theoretical problem for the systematic description of the Polish mood
system (see Bańko, 2012a, p. 162). Traditional Polish grammars distinguish between three moods
in Polish: indicative (tryb oznajmujący), conditional/subjunctive (tryb przypuszczający) and
imperative (tryb rozkazujący) (e.g., see Nagórko, 2007, p. 103). This division is mainly based on
functional properties: the indicative mood is used to express statements, the imperative mood is
meant to express orders/appeals and the conditional/subjunctive mood functions to express
possible but unreal events (Nagórko, 2007, pp. 102–104). Some linguists classify żeby-clauses as
a subtype of the conditional/subjunctive mood (e.g., Tokarski, 1973/2001; Laskowski, 1984),
whereas others claim that clauses introduced by żeby should be treated as indicative (e.g.,
Puzynina, 1971; Nagórko, 2007; see Section 2.3 for a short review of this debate).
The aim of the present paper is to discuss the status of clauses introduced by the Polish
complementizer żeby with respect to the Polish mood system and provide arguments for
classifying żeby-clauses as irrealis. Specifically, I will seek answers to the following research
questions:
– Should żeby-clauses be classified as realis or irrealis clauses?
– What are the semantic properties of verbs that select for żeby-clauses?
– What are the affinities between żeby-clauses and subjunctive structures in other languages?
At this point, a terminological excurse is necessary. The Polish term tryb przypuszczający in the
literature is translated either as ‘conditional’ or as ‘subjunctive.’ To avoid possible confusion,
QUESTIONS AND ANSWERS IN LINGUISTICS, VOLUME 3, NUMBER 1, 2016, PAGES: 27-47
© 2016 Center for General and Comparative Linguistics, University of Wrocław, Poland
Unauthenticated
Download Date | 6/19/17 6:19 AM
WHAT DOES ŻEBY INTRODUCE?
28
I need to stress that in the present discussion I will use the term ‘conditional/subjunctive mood’ to
refer to tryb przypuszczający understood as a category traditionally recognized by Polish
grammarians, which encompasses a variety of irrealis moods, such as conditional, hypothetical,
subjunctive and optative. The term ‘conditional/subjunctive mood’ should be distinguished from
‘subjunctive mood’ (discussed in Section 3), which I define as an irrealis dependent mood
usually found in embedded clauses (following Quer, 2006).
The paper is organized as follows. Section 2 provides a traditional grammar description of żebyclauses with a special focus on their distribution and debatable status in the Polish mood system.
Section 3 shows that – on the basis of their temporal properties – żeby-clauses can be classified as
irrealis clauses that share various semantic and syntactic properties with subjunctive structures in
other languages. Section 4 presents conclusions and questions for further research.
2
A descriptive look at żeby
The aim of this section is to discuss grammatical properties of żeby, such as its internal
composition, syntactic distribution and variants, following the Polish descriptive grammar
tradition. Observations about żeby will then serve as a basis for discussing controversies about its
status as part of the conditional/subjunctive mood in Polish.
2.1
Żeby as a complex complementizer
From the morphological perspective, żeby is a complex complementizer1 composed of the simple
complementizer że (subordinating conjunction equivalent to English that) and the particle by,
which is traditionally taken as the marker of the conditional/subjunctive mood in Polish
(Sadowska, 2012, p. 404). Żeby (analogously to że) also functions as a subordinating
complementizer but its distribution is restricted to specific contexts often determined by the
selectional properties of matrix predicates (see Section 2.2):2
(1)
Moja żona chce,
że-by
nasze dzieci
my
wife want.PRS.3SG that-COND/SBJV
our
children
skończyły
studia.
complete.PST.PTCP.PL.NONVIR
studies
‘My wife wants our children to complete their studies.’
Crucially, by in żeby-clauses is not moveable and thus cannot be separated and adjoined to
a subordinate clause verb. This is a crucial difference which distinguishes żeby-clauses from other
contexts in which conditional/subjunctive by appears (see Section 2.3); compare (1) and (2):
1
We will use the term complementizer instead of the traditional grammar term conjunction to make the discussion of
the traditional grammar sources compatible with the second part of the paper, which is based on the generative
grammar research.
2
The glosses in the present paper describe only those grammatical features that are relevant for specific points in the
discussion. In the glosses l-participle is marked as PST.PTCP (past participle), whereas -no/-to constructions are
marked as NO/TO. In terms of gender, plural l-participles can have either virile or nonvirile forms (marked in the
glosses as VIR and NONVIR, respectively). The remaining abbreviations are taken from The Leipzig Glossing Rules
(2015).
Unauthenticated
Download Date | 6/19/17 6:19 AM
29
(2)
MARCIN ORSZULAK
*Moja żona chce,
że nasze dzieci skończyły-by
studia.
my
wife want.PRS.3SG that our children complete.PST.PTCP.PL.-COND/SBJV studies
Intended meaning: ‘My wife wants our children to complete their studies.’
The picture of żeby distribution is additionally complicated by its different variants: aby, ażeby,
iżby, coby and by, which are not morphologically related with że (‘that’) (Tomaszewicz, 2009,
p. 222). According to Tomaszewicz (2009, p. 222, fn. 2), “aby, by are more formal, iżby, ażeby
sound dated, coby is very colloquial. Also, coby somehow fits well only in purpose clauses, but
not in complements to volitionals, whereas iżby seems fine only with volitionals.” It seems that
variants of żeby are generally governed by stylistic preferences and register. Such a view is
shared by Bańko (2012b), who treats żeby as a neutral form, by, aby, iżby and ażeby as formal
variants and coby as a colloquial form.
Finally, żeby carries person and number suffixes that mark 1st and 2nd person singular and plural;
3rd person singular and plural have zero affix (see (3) based on Sadowska, 2012, p. 455):
(3)
żeby-m
żeby-ś
żeby-Ø
żeby-śmy
żeby-ście
‘that I would’
‘that you would’
‘that he, that she, that it, that they would’
‘that we would’
‘that you would’
(1SG)
(2SG)
(3SG/PL)
(1PL)
(2PL)
Żeby as well as person and number suffixes form an agglutinative complex that cannot be
separated, i.e., these grammatical features cannot be alternatively expressed on a verb.
2.2
Distribution and composition of żeby-clauses
As Pisarkowa (1972, p. 185) notes, żeby can introduce two major types of sentential
complements classified according to the Polish descriptive grammar as complement clauses
(zdania dopełnieniowe), which serve as complements to verba sentiendi et dicendi, and subject
clauses (zdania podmiotowe), selected by verbs without prototypical subjects (see (4) and (5)
from Nagórko, 2007, pp. 306, 308):
(4)
Radził,
że-by-m
przyznał
się
advise.PST.PTCP.3SG.M
that-COND/SBJV-1SG plead.PST.PTCP.SG.M REFL
do
winy.
to
guilt
‘He advised me to plead guilty.’
(5)
Nie
wypada,
że-by-ś
palił
przy
be.good.manners.PRS.3SG
that-COND/SBJV-2SG smoke.PST.PTCP.SG.M next.to
rodzicach.
parents
‘It is not good manners for you to smoke when your parents are around.’
NEG
In terms of selectional properties of matrix verbs, complement clauses can be classified into three
groups. There is a group of verbs that can select for complements introduced only by żeby (e.g.,
compare (6) and (7)): błagać (‘to beg’), chcieć (‘to want’), kazać (‘to tell’), pragnąć (‘to desire’),
Unauthenticated
Download Date | 6/19/17 6:19 AM
WHAT DOES ŻEBY INTRODUCE?
30
rozkazać (‘to order’), radzić (‘to advise’), wzywać (‘to summon’), zabiegać (‘to strive’), zachęcać
(‘to encourage’), zezwalać (‘to permit’), zmuszać (‘to force’), zakazać (‘to prohibit’), żądać (‘to
demand’), żebrać (‘to plead’) (this group roughly corresponds to non-veridical verbs defined by
Gianakidou, 2009, discussed further in Section 3.4). Apart from this list, there are classes of
verbs which select for żeby-complement but can also select for sentential complements
introduced by other complementizers, e.g., uważać (‘to mind’), troszczyć się (‘take care’) (select
for żeby and czy (‘if’), consider (8) and (9)); mówić (‘to say’), pamiętać (‘to remember’) (select
for żeby, czy (‘if’) and że (‘that’); consider examples (10)–(12)) (cf. Pisarkowa, 1972, pp. 185–
1863):
(6)
Błagał,
że-by
żona
beg.PST.PTCP.3SG.M that-COND/SBJV
wife
wróciła.
come.back.PST.PTCP.SG.F
‘He begged his wife to come back to him.’
(7)
*Błagał,
że
żona do
niego wróciła.
beg.PST.PTCP.3SG.M that wife to
him come.back.PST.PTCP.3SG.F
Intended meaning: ‘He begged his wife to come back to him.’
(8)
Troszczyła
się,
że-by
wszyscy
take.care.PST.PTCP.3SG.F
REFL that-COND/SBJV
all
mieli
miejsca siedzące.
have.PST.PTCP.PL.VIR seats
‘She took care of the sufficient number of seats for all the guests.’
(9)
Troszczyła
się, czy nie
zabraknie
miejsc siedzących
take.care.PST.PTCP.3SG.F REFL if
NEG
lack.3SG.PRS seats
wszystkich
gości.
all
guests
‘She took care of the sufficient number of seats for all the guests.’
(10)
Pamiętaj,
że-by
dzieci
wzięły
jutro
remember.IMP that-COND/SBJV
children
take.PST.PTCP.PL.NONVIR tomorrow
książki.
books
‘Remember that children should take their books tomorrow.’
(11)
Nie
pamiętam,
czy
o
tym
remember.PRS.1SG
if
about it
‘I can’t remember if I mentioned that earlier.’
NEG
(12)
do
to
niego
him
goście
guests
dla
for
wcześniej wspomniałem.
earlier
mention.PST.PTCP.1SG.M
Pamiętam,
że kiedyś
najlepszy chleb sprzedawali
na rynku.
remember.PRS.1SG that in.the.past best
bread sell.PST.PTCP.3PL.VIR on market
‘I remember that in the past the best bread was sold on the market square.’
3
Still, one can find data that deviate from Pisarkowa’s classification. For example, some Polish speakers use
rozkazać (‘to order’) with że (not only żeby).
Unauthenticated
Download Date | 6/19/17 6:19 AM
31
MARCIN ORSZULAK
Discussing żeby-selectors in Polish, we need to note that some verbs that typically select for
a clause introduced by że (indicative complement) can take żeby-clauses as their complements in
the negative context (see (13)–(15)):
(13)
Wierzę,
że / *żeby
believe.PRS.1SG
that / that-COND/SBJV
konkurs.
competition
‘I believe that our team will win the competition.’
(14)
Nie
zespół wygra
team win.PRS.PFV.3SG
wierzę,
że
nasz zespół wygra
believe.PRS.1SG
that our
team win.PRS.PFV.3SG
‘I don’t believe that our team will win the competition.’
NEG
(15)
nasz
our
konkurs.
competition
wierzę,
że-by
nasz zespół wygrał
konkurs.
believe.PRS.1SG that-COND/SBJV our team win.PST.PTCP.SG.M competition
‘I don’t believe that our team could win the competition.’
Nie
NEG
As illustrated in (13), wierzyć (‘to believe’) selects for an indicative że-complement; however,
when negated, it can select for a że-clause (see (14)) and a żeby-clause (see (15)) with a subtle
difference in meaning (in the version with żeby the event of winning the competition is less
probable). The sentence in (15) is an example of polarity subjunctive, which is not a result of
lexical selection but of the presence of an additional element that functions as a subjunctive
licensor – in the discussed case this is negation (see Stowell, 1993, after Quer, 1998, pp. 32, 59).
Additionally, except for complement and subject clauses, żeby (together with other
complementizers, such as by and aby) can also introduce adverbial clauses that express purpose
(see (16)) (Nagórko, 2007, p. 312), e.g.:
(16)
Zapłaciłem
bratu
za
kurs niemieckiego, że-by
pay.PST.PTCP.1SG.M brother.DAT for
course German
that-COND/SBJV
znalazł
pracę w
Niemczech.
find.PST.PTCP.SG.M job
in
Germany
‘I paid for my brother’s German course so that he could find a job in Germany.’
However, this type of żeby use should be clearly distinguished from complement and subject
clauses as it is not restricted by selectional properties of matrix verbs. Finally, according to
Nagórko (2007, p. 307), żeby can also introduce a relative clause and thus function as a variant
relative pronoun (though this context seems marginal and archaic4); see (17):
(17)
Najchętniej kupił-by-m
mieszkanie, że-by
preferably
buy.PST.PTCP.M-COND/SBJV-1SG
flat
that-COND/SBJV
pomieściło
wszystkie
moje książki.
accommodate.PST.PTCP.SG.N
all
my
books
‘Preferably, I would buy a flat that would accommodate all my books.’
4
To illustrate this use of żeby, Nagórko (2007, p. 307) uses a fragment from Adam Mickiewicz’s “Pan Tadeusz” (an
epic poem from the 19th century).
Unauthenticated
Download Date | 6/19/17 6:19 AM
WHAT DOES ŻEBY INTRODUCE?
32
The relative clause in example (17) describes a wish that the apartment would be spacious
enough to accommodate all the books. However, it should be noted that such meaning in Polish is
usually conveyed by a clause that is introduced by a relative pronoun and that comprises a verb
with by (as a conditional/subjunctive marker); compare (17) and (18):
(18)
Najchętniej
kupił-by-m
mieszkanie, które
preferably
buy.PST.PTCP.M-COND/SBJV-1SG
flat
that
pomieściło-by
wszystkie
moje książki.
accommodate.PST.PTCP.SG.N-COND/SBJV
all
my
books
‘Preferably, I would buy a flat that would accommodate all my books.’
Apart from embedded contexts, in the literature there is also evidence that żeby can appear alone
in the matrix sentence to mark the conditional/subjunctive mood (see (19) based on GębkaWolak, 2010, p. 38):
(19)
Że-by
// o-by
// by
that-COND/SBJV // PARTICLE-COND/SBJV // COND/SBJV
się
chowały!
REFL grow.PST.PTCP.PL.NONVIR
‘May our children thrive!’
nasze dzieci
our
children
zdrowo
healthily
The sentence in (19) can by introduced by three elements: żeby, oby (combination of the particle
o and the conditional/subjunctive marker by) and by alone.5 Gębka-Wolak (2010, p. 38) treats
such examples as semantic equivalents in which by functions as a mood operator which marks
such sentences as conditional/subjunctive. However, Gębka-Wolak does not explain why there
are variants in which by needs to adjoin to complementizers and how żeby in such sentences is
different from żeby in subordinate clauses. Tomaszewicz (2009, p. 231), however, refers to such
structures as optatives (expressing desired states), which are related to the subjunctive mood, but
– in contrast to it – appear unembedded.
Also the internal structure of żeby-clauses is far from being unified. Subordinate clauses
introduced by żeby can comprise three forms of verbs: past participle (l-participle without past
reference), infinitival form or impersonal form (so-called ‘-no/-to constructions,’ discussed in
Section 3.2) (Pisarkowa, 1972, p. 186); consider examples (20)–(22):
(20)
Zachęcał,
że-by
wybrali
wycieczkę do Włoch.
encourage.PST.PTCP.3SG.M that-COND/SBJV choose.PST.PTCP.PL.VIR trip
to Italy
‘He encourage them to choose a trip to Italy.’
(21)
Zachęcał,
że-by
wybrać
wycieczkę
encourage.PST.PTCP.3SG.M that-COND/SBJV choose.INF trip
‘He encouraged choosing a trip to Italy.’
do
to
Włoch.
Italy
5
There is also another type of optatives that start with niechby (combination of the particle niech and the
conditional/subjunctive marker by). However, they can have a pejorative meaning of a threat/curse, e.g., Niechby go
szlag trafił! (‘Damn him,’ intended meaning: wishing someone to die).
Unauthenticated
Download Date | 6/19/17 6:19 AM
33
(22)
MARCIN ORSZULAK
Zachęcał,
że-by
wybrano
wycieczkę
encourage.PST.PTCP.3SG.M that-COND/SBJV choose.NO/TO trip
‘He encouraged choosing a trip to Italy.’
do
to
Włoch.
Italy
These three variants differ in referential properties. The variant of żeby-clause with l-participle
(example (20)) allows for an overt subject but excludes any coreference between the subject of
the main clause and the subject of subordinate clause; even if the grammatical values on verbs
(person and number) are the same (consider (23)):
(23)
Zachęcali,
że-by
wybrali
wycieczkę do Włoch.
encourage.PST.PTCP.3PL.VIR that-COND/SBJV choose.PST.PTCP.PL.VIR trip
to Italy
‘They encouraged them to choose a trip to Italy.’
In the examples with the infinitival and impersonal forms of verbs in the subordinate clause (21–
22), the subject of the embedded clause seems unspecified. In example (21) the subject of the
matrix clause could also participate in the event described by the żeby-clause, but an
interpretation in which there is no coreference between the matrix and embedded subject is also
possible. The variant with the impersonal form of the verb (22) is more difficult to assess, but
I would argue against the coreference with the matrix subject in this case.
To sum up the distribution of żeby, I can conclude that żeby appears in a variety of irrealis
contexts related to expressing wish, desire and purpose.6 Although traditionally in Polish only
three moods are distinguished (indicative, imperative and conditional/subjunctive; see Nagórko,
2007, pp. 103–104), żeby functions as a marker of different irrealis moods, such as conditional,
subjunctive and optative. Such multifunctionality is found also for grammatical markers of mood
and modality in other languages, which blurs the precise boundaries between specific mood
categories (Nordström, 2010, p. 16).
2.3
Żeby and the conditional/subjunctive mood in Polish
Żeby and other complementizers containing by have constituted a theoretical challenge in the
description of the conditional/subjunctive mood in Polish. As already mentioned, the particle by
is a conditional/subjunctive marker that is adjoined to a verb to create the conditional/subjunctive
form.7 However, there is also a possibility of detaching by and move it to the position after the
subject (see Sadowska, 2012, p. 405); compare (24) with (25):
(24)
Zjadła-by
eat.PST.PTCP.SG.F-COND/SBJV
‘She would eat something sweet.’
coś
something
(25)
Ona by
zjadła
she
COND/SBJV
eat.PST.PTCP.SG.F
‘She would eat something sweet.’
słodkiego.
sweet
coś
something
słodkiego.
sweet
6
In line with Palmer (2001, p. 1), I assume the binary distinction into realis and irrealis; however, in Section 3.1
I will adopt Mezhevich’s (2006) definition of the irrealis mood based on the temporal properties of a clause.
7
Migdalski (2006, p. 253) calls by “an invariant conditional auxiliary” that can encliticize to verbs and
complementizers. Also Borsley and Rivero (1994) analyze by as a conditional auxiliary.
Unauthenticated
Download Date | 6/19/17 6:19 AM
WHAT DOES ŻEBY INTRODUCE?
34
What is important, this moveability is blocked when by is part of a complementizer, e.g., in the
case of żeby (compare (26) and (27)):
(26)
Mama chce,
że-by-śmy
mum want.PRS.3SG that-COND/SBJV-1PL
‘Mum wants us to rest.’
(27)
*Mama
chce,
że
odpoczeli-by-śmy.
mum
want.PRS.3SG that rest.PST.PTCP.PL.VIR-COND/SBJV-1PL
Intended meaning: ‘Mum wants us to rest.’
odpoczeli.
rest.PST.PTCP.PL.VIR
As Puzynina (1971, pp. 131–132) stresses, in the case of żeby not only by stays immobile, but
also person and number inflectional suffix must be adjoined to the complementizer and it is not
possible to express it on the verb. These properties are shared by other complementizers with by,
such as gdyby and jakby (‘if’). However, the division into moveable and immoveable contexts is
additionally blurred by sentences in which the matrix sentence verb can select for both że and
żeby (see (28) and (29), examples after Puzynina, 1971, p. 134):
(28)
(29)
Powiedział,
że
on
say.PST.PTCP.3SG.M that he
‘He said that he would do it.’
by
COND/SBJV
Powiedział,
że-by
say.PST.PTCP.3SG.M that-COND/SBJV
‘He told him to do it.’
on
he
to
it
zrobił.
do.PST.PTCP.SG.M
to
it
zrobił.
do.PST.PTCP.SG.M
The described intricacies of by separation led the debate about its unified status as
a conditional/subjunctive marker. In this discussion two main stands evolved: (i) to treat all the
instances of the particle by as the conditional/subjunctive mood contexts; (ii) to distinguish the
contexts with the immovable by and treat them as a separate mood.
The first stand is expressed by Tokarski (1973/2001, p. 202), who treats żeby as a complex
complementizer in which the simple complementizer że “absorbs” the particle by to create
a construction with the past form of a verb (l-participle, past participle). For Tokarski such
a structure should be classified as the conditional/subjunctive mood because l-participle, although
morphologically marked as past, does not have a past reference. This view is shared by
Laskowski (1984, p. 135), who admits that constructions involving a complementizer with by and
a past participle constitute a problem for the description of the Polish conditional/subjunctive
mood category. However, he argues that such structures are contextually determined variants of
the conditional/subjunctive mood; the claim supported by the lack of the past reference of the
past participle as well as the non-factual and conditional meaning.
The second stance, in which moveable and immoveable by contexts are distinguished as separate
moods, was thoroughly presented by Puzynina (1971), who followed Gołąb’s view on two forms
of the conditional/subjunctive mood in Polish. Gołąb (1964; cited after Puzynina, 1971, pp. 133–
134) argued that the immoveable by is reserved to subordinate clauses of subjunctive and
purposive function, whereas the moveable by appears in matrix clauses. Nevertheless, this claim
was refuted by Puzynina (1971, p. 134) on the basis of the aforementioned optative sentences in
which żeby is placed unembedded at the beginning of a sentence (consider (30) and (31)):
Unauthenticated
Download Date | 6/19/17 6:19 AM
35
MARCIN ORSZULAK
(30)
Że-by-śmy
zawsze
byli
that-COND/SBJV-1PL always
be.PST.PTCP.PL.VIR
‘May we always be in good health.’
zdrowi.
healthy.PL.VIR
(31)
*Że zawsze
byli-by-śmy
that always
be.PST.PTCP.PL.VIR-COND/SBJV-1PL
Intended meaning: ‘May we always be in good health.’
zdrowi.
healthy.PL.VIR
As observed in examples (30) and (31), in such contexts we cannot separate że and by, which is
obviously at odds with Gołąb’s claim. Alternatively, Puzynina (1971, p. 136) proposed that the
instances of the immoveable by should be subsumed into the indicative mood irrespective of the
matrix clause/subordinate clause context: “these forms should be treated as a positional variant
[of the indicative mood – M.O.] characterized by past morphology on the verb and
person/number suffixes immobilized at the conjunction or particle” [translation – M.O].
According to her reasoning, sentences with żeby belong to the indicative mood in which the
category of mood expressed on the verb is neutralized by the modal meaning expressed by the
complementizer or the modal verb in the matrix clause (Puzynina, 1971, p. 136). From the
diachronic perspective, such forms result from the process in which complementizers and
particles take over modal functions from verbs (Puzynina, 1971, p. 139).
It must be noted that Puzynina’s view is followed in many more contemporary works (e.g., see
Nagórko, 2007; Gębka-Wolak, 2010). Nevertheless, as Gaszyńska-Magiera (1998, p. 52) notices,
the literature on the mood categories in Polish is rather scarce and there are theoretical problems
which have yet to find a universally accepted account.8 Among such issues, Gaszyńska-Magiera
(1998, p. 54) pinpoints the case of the immoveable by and the problem of mood selection in
subordinate clauses, in which we can find interesting mood oppositions, such as those illustrated
by the verb mówić (‘to say’), which selects for both że- and żeby-clauses (see examples (28)
and (29)).
In the next section, I will follow the conditional/subjunctive view on żeby-clauses and show that
they can be classified as irrealis on the basis of their temporal properties.
3
Żeby-clauses and the realis/irrealis distinction
In this section I follow the basic mood distinction into realis and irrealis and adopt Mezhevich’s
proposal of treating mood as a dyadic predicate that relates two times (Mezhevich, 2006) to show
that żeby-clauses actually belong to the irrealis side. Next, I have a closer look at various verb
forms that are licit in żeby-clauses and elaborate on their temporal interpretation, drawing an
interesting analogy between Polish and Greek data. Finally, I focus on the notion of subjunctive,
understood as a dependent mood found in embedded clauses (see Quer, 2006), and demonstrate
that żeby-clauses can be classified as subjunctive on the basis of selectional properties of matrix
verbs (the contrast between veridical and non-veridical verbs, proposed by Giannakidou, 2009)
8
Gaszyńska-Magiera (1998) analyzed descriptions of the Polish conditional/subjunctive mood in textbooks for
learning Polish as a foreign language. She showed that the lack of a unified theoretical account on the descriptive
side results in prescriptive inconsistencies.
Unauthenticated
Download Date | 6/19/17 6:19 AM
WHAT DOES ŻEBY INTRODUCE?
36
and various transparency effects that are found in embedded subjunctive clauses
crosslinguistically (as described by Quer, 2006).
3.1
Żeby-clauses as irrealis clauses
Mezhevich (2006, p. 119) proposes that mood should be analyzed as a dyadic predicate that
relates two times: the evaluation time (“time relative to which the situation described by the
utterance is evaluated”) and the utterance time. As a consequence, the relation between these two
times is a basis for the realis/irrealis distinction: “in realis mood, the utterance time is interpreted
as the evaluation time: the situation is evaluated relative to the utterance time. […] In irrealis
mood, the utterance time is interpreted as not being the evaluation time: the situation cannot be
evaluated relative to the utterance time” (Mezhevich, 2006, p. 124). For example, consider the
pair of sentences in (32) and (33) (from Mezhevich, 2006, p. 125):
(32)
I had a car.
(33)
I wish I had a car.
In the realis sentence I had a car the utterance time is actually the evaluation time, which means
that the situation time (the time of the propositional content) is interpreted with respect to the
utterance time: the event of having a car was prior to the utterance (past tense interpretation). On
the other hand, the interpretation of the irrealis sentence in (33) involves two different time lines:
the time line at which the utterance time is located and the time line at which the situation time is
placed (the event of having a car).
In her analysis, Mezhevich (2006) follows Iatridou (2000, p. 247), who analyzed mood as
a dyadic predicate that relates two types of worlds: topic worlds (“the worlds that we are talking
about”) and actual worlds (“the worlds that for all we know are the worlds of the speaker”).
However, Mezehevich’s (2006) modification to define mood not as the relation between worlds
but as the relation between times (or time lines) allows capturing a number of properties that
tense and mood share (literature review in Mezhevich, 2006, pp. 122-123):
– mood and tense systems are structured based on binary distinctions: realis/irrealis for mood,
past/non-past and future/non-future for tense;
– languages do not have special verbal morphology to express the indicative/realis mood and if
a clause has a temporal interpretation, then it is indicative;
– crosslinguistically, the irrealis mood is commonly expressed by past morphology without past
interpretation (“fake past” as Iatridou (2000, p. 244) labels it).
Similar connections between tense and mood are visible in the Polish data. Żeby-clauses
comprise l-participles (apart from infinitives and impersonal verbs as already mentioned), which
are morphologically past, but lack a past interpretation. However, when an l-participle is found in
an indicative clause, it receives the past interpretation, consider the following examples:
(34)
Mówi,
że
brat
kupił
nowy samochód
say.PRS.3SG that
brother
buy.PST.PTCP.3SG.M new car
w tamtym tygodniu /
*w następnym
tygodniu.
in last
week
/
in next
week
‘He says that his brother bought a car last week/*next week.’
Unauthenticated
Download Date | 6/19/17 6:19 AM
37
(35)
MARCIN ORSZULAK
Mówi,
że-by
brat
kupił
say.PRS.3SG that-COND/SBJV
brother
buy.PST.PTCP.SG.M
samochód
*w tamtym tygodniu /
w następnym tygodniu.
car
in last
week
/
in next
week
‘He tells his brother to buy a new car *last week/next week.’
nowy
new
The pair of sentences presented in (34) and (35) features the verb mówić (‘to say’), which
alternates between the indicative and subjunctive complements (i.e., it can select for both). The
indicative clause in (34) is only grammatical with the past tense adverbial w tamtym tygodniu
(‘last week’), but not with the future tense adverbial w następnym tygodniu (‘next week’). In
contrast, the subjunctive clause in (35), despite being morphologically past, is not compatible
with the past adverbial. The same contrast was observed in Russian by Mezhevich (2006, p. 148),
who points out that “embedded subjunctives typically denote a hypothetical situation in the future
relative to the matrix event.”
To recapitulate the main tenets of Mezhevich’s proposal, indicative clauses have an independent
tense interpretation and the time of the event that they describe is evaluated with respect to the
utterance time, which is characteristic of the realis mood. On the other hand, subjunctive clauses
have a temporal interpretation dependent on the context (for matrix subjunctives) or on the matrix
clause and the event described by a subjunctive clause is not interpreted relative to the utterance
time (Mezhevich, 2006, pp. 126, 151).
3.2
Żeby-clauses with -no/-to verb forms and infinitives
The discussion at this point must turn to other verb forms found in żeby-clauses: -no/-to
constructions and infinitives. I will show similarities that they share with l-participles in terms of
temporal interpretation.
The -no/-to constructions are impersonal verb forms ending with -no or -to suffix. Such forms
exhibit rare syntactic properties, such as no overt subject, PROarb interpretation and the ability to
assign the accusative case to their complements (cf. Lavine, 2005 and Ruda, 2014 for discussion).
However, what is important for our analysis is the fact that -no/-to forms can be used only with
past tense adverbials if they appear in a realis/indicative sentence (see Ruda, 2014, p. 210);
consider (36):
(36)
W dawnych czasach / *obecnie / *w przyszłości budowano
domy z drewna.
in old
times / at.present / in future
build.NO/TO houses of wood
‘In the past/*at present/*in the future houses were made of wood.’
However, if the -no/-to forms are used in żeby-clauses they lose their past reference and are
interpreted as hypothetical future relative to the matrix predicate event; see (37):
(37)
Mieszkańcy chcą,
że-by
wybudowano szkołę
residents
want.PRS.3PL that-COND/SBJV
build.NO/TO school
*w tamtym roku / w następnym roku.
in last
year / in next
year
‘The residents want the school to be built *last year/next year.’
Unauthenticated
Download Date | 6/19/17 6:19 AM
WHAT DOES ŻEBY INTRODUCE?
38
The examples in (36) and (37) show that the verbs with -no/-to suffixes have the past
interpretation (i.e., compatible with past tense adverbials) only in indicative sentences. This is an
exact analogy to the l-participle, which is interpreted as past in realis clauses, but loses this
interpretation in irrealis contexts. Again here we deal with “fake past,” which – as already
mentioned – is typical for irrealis contexts crosslinguistically.
According to Mezhevich (2006, p. 125), irrealis sentences do not have a temporal interpretation
because “[a] clause can have a temporal interpretation only if it describes a situation whose time
is located on the same time line as the utterance time.” Similar properties are attributed to
infinitives. Wurmbrand (2007) argues that infinitives are tenseless, i.e., they lack their own
temporal denotation, consider the following sentences (examples from Wurmbrand, 2007,
p. 409):
(38)
Leo decided a week ago that he will go to the party (*yesterday).
(39)
Leo decided a week ago to go to the party yesterday.
(38) presents the example of a finite future statement that has an absolute interpretation: the event
of going to the party needs to occur after the utterance time; hence the incompatibility with the
past tense adverbial. In contrast, the infinitival event in (39) must follow the matrix event; the
utterance time is irrelevant here. Such a contrast can be also observed in Polish: embedded realis
sentences with the compound future construction9 have an independent future interpretation as
opposed to their irrealis infinitival counterparts; compare (40) and (41):
(40)
Marek powiedział
tydzień temu, że
jutro
/ *wczoraj
Mark say.PST.PTCP.3SG.M week ago that
tomorrow / yesterday
będzie
kupował
nowy samochód.
be.PRS.PFV.3SG
buy.PST.PTCP.IPFV.SG.M
new car
‘Mark said a week ago that tomorrow/*yesterday he would buy a new car.’
(41)
Marek powiedział
Mark say.PST.PTCP.3SG.M
wczoraj
kupić
yesterday
buy.INF
tydzień temu, że-by
week ago that-COND/SBJV
nowy samochód.
new car
jutro
/
tomorrow /
The clause with finite future construction in (40) is acceptable only with the future tense
adverbial jutro (‘tomorrow’); however, its infinitival counterpart introduced by żeby is fine with
both past tense and future tense adverbial (see (41)). This is so because the event expressed by
the infinitive should occur after the matrix clause event (hypothetical future) and thus any
adverbial that describes an event happening later than a week ago fulfils this requirement.
In sum, the common denominator of all the forms that can be found in żeby-clauses –
l-participles, -no/-to constructions and infinitives – is the property of losing independent temporal
interpretation that is evaluated with respect to the utterance time. Such a conclusion is in line with
Mezhevich’s (2006) insights into the nature of the irrealis mood, in which the situation time and
utterance time are located on different time lines and thus the propositional content of a clause
9
Polish has two constructions to express future time: the so-called simple future with perfective verbs and the
compound/periphrastic future with imperfective verbs (see Sadowska, 2012, pp. 398ff). The incompatibility with
past tense adverbials holds for both types of future constructions in Polish.
Unauthenticated
Download Date | 6/19/17 6:19 AM
39
MARCIN ORSZULAK
cannot be evaluated with respect to the utterance time. Furthermore, as illustrated with the
examples in this section, verb forms that have an independent temporal reference are compatible
with the realis mood since only in such cases the situation time and the utterance time are located
on the same time line, which allows for an absolute temporal interpretation.
3.3
The puzzle of fake past
At this juncture, one needs to pose a question about l-participles and -no/-to constructions and
their ability to lose past interpretation. If we assume that these verb forms are past and in the
subjunctive/conditional context they lose this interpretation, then we posit that there are actually
two different l-participles and two different -no/-to constructions – one type with past
interpretation found in indicative clauses and the other type without past interpretation found in
the subjunctive/conditional clauses.
Another line of reasoning would be to assume that l-participles and -no/-to constructions are
temporally unspecified and that they acquire specific temporal properties depending on a context,
i.e., in connection with other elements such as subjunctive particles. A similar analysis was
proposed by Giannakidou (2009) with respect to Greek perfective nonpast forms (PNP).10
In Greek the PNP cannot appear on its own and it must be licensed by other particles: the
subjunctive na, the future tha, the conditional an and future-oriented connectives, e.g., prin
‘before’ (examples from Giannakidou, 2009, pp. 1885-1886):
(42)
Na/as to
pis.
SBJV it
say.PNP.2SG
‘You may say it.’
(43)
Tha
to
pis.
it
say.PNP.2SG
‘You will say it.’
FUT
(44)
Prin to
pis.
before it
say.PNP.2SG
‘Before you say it,…’
(45)
*To pis.
(PNP: *on its own)
Assuming the pronominal theory of tense, Giannakidou (2009, p. 1906) considers the PNP
“a referentially deficient dependent variable […] a variable that cannot be used deictically to
refer to a particular time specified by the context only.” Thus, the specific interpretation of PNP
forms is that of time intervals starting with t, a variable that cannot be interpreted deictically, and
moving forward to unspecified future (Giannakidou, 2009, p. 1898). The particles like na and tha
are necessary for the PNP to be grammatical as they provide a now variable n with which
t variable is identified. Crucially, Giannakidou (2009, p. 1906) proposes a clear interpretational
difference between embedded uses of the PNP with the subjunctive particle na and main uses
with the future particle tha. Embedded PNPs are interpreted with respect to the matrix event:
“The temporal n of the subjunctive in embedded clauses remains a relative now, an idea
consistent with the fact that embedded na is triggered by a higher predicate” (ibidem). Main
10
I would like to thank the anonymous reviewer for suggesting the analogy between Polish and Greek data.
Unauthenticated
Download Date | 6/19/17 6:19 AM
WHAT DOES ŻEBY INTRODUCE?
40
clause PNP forms, on the other hand, are interpreted with respect to the utterance time since the
now variable n introduced by tha is identified with the utterance time and, as a result, “futurate
reading” arises (ibidem).
Drawing an analogy between the PNP in Greek and l-participles and -no/-to constructions in
Polish, we can notice that żeby behaves in a similar way to Greek na as it introduces embedded
subjunctives whose temporal interpretation is analogous to their Greek counterparts. Specifically,
events expressed by żeby-clauses have a future interpretation with respect to the matrix event and
thus, in Giannakidou’s terms, żeby introduces a now variable that is identified with the attitude
verb. However, the discussed similarity between Greek and Polish is weakened by the fact żebyclauses can feature infinitives, whereas Greek has only finite complementation (see Giannakidou,
2009, p. 1885).
Another parallel between Polish and Greek concerns the future particle tha, which licenses main
clause PNP forms. Tha bears resemblance to the subjunctive/conditional by used in main clauses
in Polish. What is especially interesting is an observation that tha and by appear lower in the
structure than na and żeby. Still, tha has a future tense interpretation which is equivalent rather to
Polish simple future construction than to the hypothetical by (still the future interpretation may be
considered irrealis similarly to the hypothetical mood; see Dahl, 1985, p. 103).
To sum up this part of the discussion, the analogy between Greek and Polish provides grounds
for an alternative analysis of fake past forms in subjunctive clauses in Polish. Instead of
proposing a kind of homonymy, comprising “real” past and “fake” past, we may posit that
l-participles and -no/-to constructions are temporally deficient and their temporal interpretation
must be completed by another element (polarity behavior) that in subjunctive clauses would be
żeby, which would introduce the now variable n and thus provide a starting point for the time
interval. Nonetheless, such an analysis requires further research to account for the past
interpretation of these verb forms in indicative clauses.11
3.4
Żeby as a subjunctive complementizer
So far I have shown that żeby-clauses are irrealis clauses; however, the irrealis mood is an
umbrella term that encapsulates such moods as conditional, subjunctive, hypothetical or optative.
In the next section I will show that żeby-clauses introduced by a specific group of predicates can
be classified as subjunctives (see also Section 4).
Żeby-clauses in Polish have a close affinity with subjunctive structures in numerous European
languages. Here I understand the subjunctive mood as a mood found typically in embedded
clauses which are complements to verbs with particular semantic properties, such as volitionals
and directives (see Giannakidou, 2009; Quer, 2006). On the interpretation side, in contrast with
realis/indicative, subjunctive is an irrealis mood, which is often used to express wish, obligation
and request. From the typological perspective, the category of subjunctive may be realized in
various forms: separate verbal paradigm (Romance languages), inflectional particles (e.g., Greek
na) and complementizers (e.g., Russian čtoby) (Quer, 2006, p. 660). According to Giannakidou
(2009, p. 1884), subjunctive realization generally follows two patterns: verbal subjunctive
(“a piece of morphology on the verb specific to this category”) and subjunctive external to the
verb, i.e., expressed by uninflected particles, also combined with complementizers.
11
Dornisch (1997) proposed a cover past tense auxiliary that would be responsible for the past interpretation.
Unauthenticated
Download Date | 6/19/17 6:19 AM
41
MARCIN ORSZULAK
Against this background, Polish żeby-clauses pattern with non-verbal forms of the subjunctive in
which this mood is expressed by a special complementizer (this is also the case of other Slavic
languages, e.g., Russian čtoby and Czech/Slovak aby; see Tomaszewicz, 2009, p. 222). At this
point, it is crucial to discuss selectional properties of verbs which select for żeby-clauses in
Polish, because, as Giannakidou (2009, p. 1887) argues, crosslinguistically subjunctive clauses
are selected by specific classes of predicates.
Giannakidou (2009, p. 1885) elaborates on the distinction between veridical and non-veridical
verbs, which is based on the availability of truth inference. If a verb expresses the speakers’s or
subject’s commitment to the truth of the complement clause, then it is veridical and selects for the
indicative (ibidem). Veridical verbs include such categories as assertives (to say), fiction verbs (to
dream), epistemics (to believe), factive verbs (to know), semifactives (to remember) (ibidem).
Non-veridical verbs, on the other hand, do not ascertain truth, e.g., from the sentence John wants
to find a snake, one cannot know if John actually found a snake (example from Giannakidou,
2009, p. 1889). The class of non-veridical verbs includes volitionals (to want), directives (to
order), modals (must), permissives (to allow) and negative verbs (to refuse) (Giannakidou, 2009,
p. 1888).
Although Ginnakidou’s classification is based on Greek verbs, it seems more universal and
applicable to the data from Polish. It is important to stress that all the verbs that select for żebyclauses in Polish (see Section 2.2) fall into the non-veridical category associated with the
subjunctive (mainly volitionals, such as chcieć ‘to want’, pragnąć ‘to desire,’ and directives, e.g.,
kazać ‘to tell’ and rozkazać ‘to order’). Therefore, under the selectional criterion, żeby-clauses
selected by non-veridical verbs can be categorized as subjunctives. If this assumption is true, then
such complex sentences should exhibit specific syntactic transparency phenomena typical for
similar structures in other languages.
Quer (2006, pp. 661-669) discusses several syntactic effects that show transparency of
subjunctive clauses which allow for some syntactic and semantic relations which are not possible
for indicative clauses (“subjunctive-related phenomena”). To illustrate this property, he collated
linguistic data from various languages referring to obviation effects (in Spanish), long-distance
anaphoric binding (in Icelandic) and NPI-licensing (in French) as well as subject raising 12
(Romanian and Greek).
Żeby-clauses exhibit obviation effects, which are “a ban on coreference between the embedded
and the matrix subject” (Quer, 2006, p. 662); see (46) and (47):
(46)
Piotri chciał,
że-by
pro*i/j
Peter want.PST.PTCP.3SG.M that-COND/SBJV pro
z
pokoju.
from room
‘Peter wanted him to leave the room.’
(47)
Piotri powiedział,
że
Peter say.PST.PTCP.3SG.M that
‘Peter said that he left the room.’
proi/j
pro
wyszedł
leave.PST.PTCP.SG.M
wyszedł
leave.PST.PTCP.3SG.M
z
from
pokoju.
room
12
Raising verbs in Polish do not select for żeby-clauses and therefore this subjunctive diagnostic cannot be applied to
Polish data.
Unauthenticated
Download Date | 6/19/17 6:19 AM
WHAT DOES ŻEBY INTRODUCE?
42
As observed in (46) and (47), there is a difference between the subjunctive complement (46) and
the indicative complement (47) in terms of coreference and lack of coreference between
embedded and matrix subjects: żeby-clause disallows coreference. However, obviation effects
become less strong in żeby-clauses with infinitives (see (48)):
(48)
Piotri chciał,
że-by
PROi + j
Peter want.PST.PTCP.3SG.M that-COND/SBJV
PRO
z
pokoju.
from room
‘Peter wanted us (including himself) to leave the room.’
wyjść
leave.INF
As already mentioned in Section 2.2 (see example (21)), in such sentences as (48) the matrix
subject can take part in the embedded event but together with other unspecified participants, i.e.,
Peter wanted to leave together with other people. Still, the interpretation that Peter was the only
subject of wanting and leaving is blocked.
Obviation effects also influence long-distance anaphoric binding possibilities, i.e., if the subject
of the embedded subjunctive clause cannot be coindexed with the matrix subject as in (50), then
the reflexive pronoun sobie cannot refer to the subject in the main clause, in contrast with (49)
and (51), where the matrix subject can serve as an antecedent for the reflexive pronoun.
(49)
Piotri chciał
PROi ogolić
Peter want.PST.PTCP.3SG.M PRO shave.INF
‘Peter wanted to shave his head.’
sobiei głowę.
REFL head
(50)
Piotri chciał,
że-by
Peter want.PST.PTCP.3SG.M that-COND/SBJV
głowę.
head
‘Peter wanted him to shave his head.’
proj
pro
(51)
Piotri powiedział,
że
proi/j
Peter say.PST.PTCP.3SG.M that pro
‘Peter said that he shaved his head.’
ogolił
sobie*i/j
shave.PST.PTCP.SG.M REFL
ogolił
sobiei/j
shave.PST.PTCP.3SG.M REFL
głowę.
head
The data presented by Quer (2006, p. 664) actually show that in Icelandic the reflexive pronoun
sig located in the subjunctive clause can be coindexed with the matrix subject, which is not the
case in Polish. Still, Polish subjunctive and indicative complements differ with respect to
reference possibilities: the subjunctive context (50) excludes the possibility of the coreference
between the reflexive pronoun and the matrix subject, whereas in the indicative context (51) such
coreference is allowed.
As far as NPI-licensing is concerned, żeby-clauses pattern with indicative clauses and NPIs13
cannot be licensed over the clause boundary; see (52)-(53):14
13
Here we assume, after Giannakidou (2011, p. 1661), that the core property of negative polarity items is “exclusion
from positive assertions with simple past.” Mind that in (52)-(53) we only refer to n-words understood as a type of
NPIs.
Unauthenticated
Download Date | 6/19/17 6:19 AM
43
MARCIN ORSZULAK
(52)
*Maria nie
chciała,
że-by
kupiła
Mary NEG want.PST.PTCP.3SG.F that-COND/SBJV
buy.PST.PTCP.SG.F
Intended meaning: ‘Mary didn’t want her to buy anything.’
(53)
*Maria nie
powiedziała,
że
kupiła
Mary NEG say.PST.PTCP.3SG.F that
buy.PST.PTCP.3SG.F
Intended meaning: ‘Mary didn’t say that she bought anything.’
niczego.
nothing
niczego.
nothing
Indicative and żeby-clauses also behave alike with respect to Genitive of Negation (GoN). In
Polish transitive verbs assign the Accusative case to their objects, but when a verb is negated
objects are marked as Genitive (Witkoś, 1995, p. 246); compare (54) and (55):
(54)
Maria kupiła
Mary buy.PST.PTCP.3SG.F
‘Mary bought a new car.’
(55)
Maria nie
kupiła
Mary NEG buy.PST.PTCP.3SG.F
‘Mary didn’t buy a new car.’
nowy
new.ACC
samochód.
car.ACC
*nowy
new.ACC
samochód / nowego samochodu.
car.ACC
/ new.GEN car.GEN
However, for both indicative and żeby-complements, GoN cannot be assigned over the finite
clause boundary, i.e., when negation is placed in the matrix clause; compare (56) and (57):
(56)
Maria nie
chce,
że-by
Jan
Mary NEG want.PRS.3SG that-COND/SBJV
John
*nowego
samochodu / nowy
samochód.
new.GEN
car.GEN
/ new.ACC
car.ACC
‘Mary doesn’t want John to buy a new car.’
kupił
buy.PST.PTCP.SG.M
(57)
Maria nie
powiedziała,
że
Jan
kupił
Mary NEG say.PST.PTCP.3SG.F that
John buy.PST.PTCP.3SG.M
*nowego
samochodu / nowy
samochód.
new.GEN
car.GEN
/ new.ACC
car.ACC
‘Mary didn’t say that John bought a new car.’
These four contexts show that there is no systematic difference between embedded żeby-clauses
and indicative clauses in Polish: they differ in the case of coreference possibilities but not in the
case of NPI licensing and Genitive of Negation. The contrast between these two types of clauses
is blurred in the case of movement operations: long-distance wh-extraction and clitic climbing.
Żeby-clauses in Polish allow for long-distance wh-extraction but only in the case of verbs that
also select for infinitival complements (such as chcieć ‘to want’) (see Willim, 1989). In contrast,
the extraction out of the indicative complement is blocked; compare (58) and (59):15
14
However, in Italian the matrix negation can license NPIs in embedded subjunctive clauses (see Quer, 2005,
pp. 664–665).
15
See Willim (1989) and Witkoś (1995) for the details concerning extractions out of indicatives and subjunctives as
extraction possibilities differ with respect to varied factors, e.g., bridge verbs, subject-position extraction, multiple
wh-extraction, etc.
Unauthenticated
Download Date | 6/19/17 6:19 AM
WHAT DOES ŻEBY INTRODUCE?
44
(58)
Co
Maria chce,
że-by
what Mary want.PRS.3SG that-COND/SBJV
‘What does Mary want Peter to buy?’
Piotr kupił?
Peter buy.PST.PTCP.SG.M
(59)
*Co Maria sądziła,
że
Piotr wcześniej
studiował?
what Mary think.PST.PTCP .3SG.F that
Peter before
study.PST.PTCP.3SG.M
Intended meaning: ‘What did Mary think that Peter studied before?’
What is interesting is that the movement contrast disappears in the case of clitic climbing: the
object pronoun can move out neither from the żeby-complement nor from the indicative
complement (see Witkoś, 1995, p. 245); consider (60) and (61):
(60)
*Maria
go
chciała,
że-by
Jan uderzył.
Mary
him want.PST.PTCP.3SG.F that-COND/SBJV John hit.PST.PTCP.SG.M
Intended meaning: ‘Mary wanted John to hit him.’
(61)
*Maria
go
powiedziała,
że
Mary
him say.PST.PTCP.3SG.F that
Intended meaning: ‘Mary said that John hit him.’
Jan
John
uderzył.
hit.PST.PTCP.3SG.M
The contrast between wh-extraction and clitic climbing may result from different types of
movement (wh-movement and NP movement as suggested by Witkoś, 1995, p. 245) and thus
different constraints that determine such operations. Nonetheless, these two contexts also show
that in Polish it is difficult to classify clausal complements into transparent subjunctives and
opaque indicatives. Żeby-clauses are transparent only for certain relations/operations: obviation
effects and long-distance wh-extraction. This is actually predictable because as Quer (2006, p.
661) pointed out:
[…] subjunctive does not constitute a syntactically uniform object, either cross-linguistically or even
within the same language. For instance, some of the subjunctive-related phenomena described for
one language do not hold for other languages having subjunctive mood. Similarly, some allegedly
subjunctive-related phenomena show up in a subset of the subjunctive clauses in a language, but not
in all of them.
Quer’s observations are confirmed by the Polish data: in Polish we can find only some of the
subjunctive-related phenomena and, moreover, there are phenomena like long-distance whextraction which are restricted to “a subset of the subjunctive clauses” – in this case subjunctive
clauses that also select for infinitives.
To sum up, the data presented in this section support the claim that subordinate żeby-clauses can
be classified as subjunctive clauses. Based on the comparison with other languages that exhibit
various forms of the subjunctive mood, I showed that żeby-clauses are selected by non-veridical
verbs, which are typical subjunctive selectors, and are transparent for some syntactic and
semantic relations. At this juncture, I need to return to the theoretical discussion about the status
of żeby-clauses in the Polish mood system (summarized in Section 2.3). I cannot agree with
Puzynina (1971) and Nagórko (2007), who treat subordinate żeby-clauses as indicative clauses,
first because of the distinction into veridical and non-veridical selectors and second because of
the different syntactic and semantic properties that they have. Furthermore, classifying żebyclauses as indicative would obscure the distinction between realis and irrealis mood as well as be
Unauthenticated
Download Date | 6/19/17 6:19 AM
45
MARCIN ORSZULAK
at odds with the observation that predicates that select for żeby belong to the irrealis side as they
do not ascertain truth and express non-actualized events (see Nordström, 2010, p. 30).
4
Conclusions and further research problems
In the present article I delved into the problematic status of żeby-clauses in the Polish mood
system. I reviewed the Polish literature on the subject and showed that żeby-clauses lack
a unified account since they are classified either as conditional/subjunctive (Tokarski, 1973/2001;
Laskowski, 1984) or as indicative (Puzynina, 1971; Nagórko, 2007). In this debate about the
mood value of żeby-clauses I took a stance that they are irrealis clauses, which is supported by
their temporal properties. On the basis of Mezhevich’s (2006) definition of mood, I demonstrated
that żeby-clauses are irrealis since they are interpreted relative to the matrix event and not to the
utterance time as the situation time and the utterance time are placed on different time lines.
Furthermore, żeby-clauses can only comprise verb forms that have a dependent tense
interpretation in certain contexts, that is, l-participle and -no/-to constructions as well as
infinitives. I also recognized a problem with two different temporal make-ups of l-participles and
-no/-to constructions (absolute tense, deficient tense) depending on the mood of
a clause. As a preliminary solution, I suggested an alternative analysis of these forms based on
the Greek data.
In the further discussion, I focused on żeby-complements and showed that they are selected by
non-veridical verbs, which is the property of subjunctive complements (Giannakidou, 2009).
Embedded żeby-clauses that function as complements to non-veridical verbs bear similarities to
subjunctive structures in other languages. I followed Quer’s (2006) digest of subjunctive-related
phenomena and showed that żeby-clauses are transparent domains for some semantic and
syntactic operations; specifically: obviation effects and long-distance wh-extraction. However,
I found that in Polish there is no systematic difference between indicative and subjunctive żebycomplements: they differ in terms of reference possibilities and wh-extraction but pattern alike
with respect to clitic climbing, Genitive of Negation and NPI licensing.
The present account of żeby-clauses cannot be treated as exhaustive. I mainly elaborated on the
subset of those żeby-clauses that function as complements to non-veridical verbs. As discussed in
Section 2.2, the complementizer żeby can also introduce purpose clauses, unembedded optative
clauses, relative clauses and subject clauses. The first three contexts can be subsumed under the
irrealis mood based on their semantic properties, i.e., they express unreal/hypothetical events.16
Still, their status in the Polish mood system requires further research.
An interesting aspect, which actually escapes the present analysis, is the context of subject
clauses introduces by żeby. As already mentioned, such żeby-clauses function as subjects and are
selected by verbs without prototypical subjects (e.g., wypadać ‘to be good manners’). It would be
interesting to compare such verbs with non-veridical verbs to seek semantic properties which
trigger the emergence of żeby in both contexts.
It must be also stressed that the analysis of the mood system in Polish should be supplemented by
thorough corpus research. It would be crucial to verify the existing classifications of indicative
16
Giannakidou (2009, pp. 1888-1889) presents examples of subjunctive relative clauses in Greek and suggests that
they are licensed by non-veridicality.
Unauthenticated
Download Date | 6/19/17 6:19 AM
WHAT DOES ŻEBY INTRODUCE?
46
and subjunctive selectors found in descriptive grammars of Polish. Furthermore, corpora can
show whether że-selectors (indicative) and żeby-selectors (subjunctive) actually overlap with
Giannakidou’s veridical/non-veridical distinction. What would be also interesting is to see which
verbs that typically select for an indicative complement can be turned into subjunctive selectors
by the presence of matrix negation (polarity subjunctive).
Finally, the issues discussed in the present paper contribute to a wider discussion about the
category of subjunctive. As noted by Quer (2006), subjunctive structures do not exhibit uniform
properties both within a given language and crosslinguistically. Still, the mood value of
a subordinate clause (realis/irrealis, indicative/subjunctive) has an impact on various syntactic
and semantic relations between the main clause and the subordinate clause.
References
Bańko, M. (2012a). Wykłady z polskiej fleksji. Warszawa: PWN.
Bańko,
M.
(2012b).
Żeby
i
iż.
Retrieved
http://sjp.pwn.pl/poradnia/haslo/zeby-i-iz;13517.html.
December
29,
2015
from:
Borsley, R. D. and Rivero, L. M. (1994). Clitic auxiliaries and incorporation in Polish. Natural
Language and Linguistic Theory, 12, 373-422.
Dahl, Ö. (1985). Tense and Aspect Systems. Oxford: Basil Blackwell.
Dornisch, E. (1997). Auxiliaries and functional projections in Polish. In W. Browne, E. Dornisch,
N. Kondrashova, and D. Zec (Eds.), Formal Approaches to Slavic Linguistics (FASL) IV:
The Cornell Meeting 1995 (pp. 183-209). Ann Arbor: Michigan Slavic Publishers.
Gaszyńska-Magiera, M. (1998). Tryb przypuszczający w nauczaniu języka polskiego jako
obcego. Acta Universitatis Lodziensis, 10, 51-60.
Gębka-Wolak, M. (2010). Ile form bezokolicznikowych jest w paradygmacie czasownika?
Problem trybu przypuszczającego bezokolicznika. Poznańskie Studia Polonistyczne. Seria
językoznawcza, 17(37), 25-39.
Giannakidou, A. (2009). The dependency of the subjunctive revisited: Temporal semantics and
polarity. Lingua, 199, 1883-1908.
Giannakidou, A. (2011). Positive polarity items and negative polarity items: Variation, licensing,
and compositionality. In C. Maienborn, K. von Heusinger, and P. Portner (Eds.), Semantics:
An International Handbook of Natural Language Meaning (pp. 1661-1712). Berlin: Mouton
de Gruyter.
Gołąb, Z. (1964). Conditionalis typu bałkańskiego w językach południowosłowiańskich.
International Journal of Slavic Languages and Poetics, 8, 1-36.
Iatridou, S. (2000). The grammatical ingredients of counterfactuality. Linguistic Inquiry, 31(2),
231-270.
Laskowski, R. (1984). Kategorie morfologiczne języka polskiego – charakterystyka
funkcjonalna. In R. Grzegorczykowa, R. Laskowski, and H. Wróbel (Eds.), Gramatyka
współczesnego języka polskiego. Morfologia (pp. 121-169). Warszawa: PWN.
Unauthenticated
Download Date | 6/19/17 6:19 AM
47
MARCIN ORSZULAK
Lavine, J. L. (2005). The morphosyntax of Polish and Ukrainian -no/-to. Journal of Slavic
Linguistics, 13(1), 75-117.
Mezhevich, I. (2006). Featuring Russian Tense: A Feature-Theoretic Account of the Russian
Tense System. (Doctoral dissertation). University of Calgary.
Migdalski, K. (2006). The Syntax of Compound Tenses in Slavic. Utrecht: LOT.
Nagórko, A. (2007). Zarys gramatyki polskiej. Warszawa: PWN.
Nordström, J. (2010). Modality and Subordinators. Amsterdam/Philadelphia: John Benjamins.
Palmer, F. R. (2001). Mood and Modality. Cambridge: CUP.
Pisarkowa K. (1972). Tryb przypuszczający i czas zaprzeszły w polszczyźnie współczesnej
(formy i funkcje). Język Polski, LII(3), 183-189.
Puzynina, J. (1971). Jeden tryb czy dwa tryby? Biuletyn Polskiego Towarzystwa
Językoznawczego, XXIX, 131-139.
Quer, J. (1998). Mood at the Interface. (The LOT International Series). The Hague: Holland
Academic Graphics.
Quer, J. (2006). Subjunctives. In M. Everaert, H. van Riemsdijk, R. Goedemans, and B.
Hollebrandse (Eds.), The Blackwell Companion to Syntax, vol. IV (pp. 660-684). Malden,
MA: Blackwell.
Ruda, M. (2014). The impersonal subject -n/-t construction in Polish and the typology of voice
heads. Studies in Polish Linguistics, 9(4), 203-243.
Sadowska, I. (2012). Polish: A Comprehensive Grammar. New York: Routledge.
Stowell, T. (1993). Syntax of Tense. (Unpublished manuscript). University of California, Los
Angles, USA.
Tokarski, J. (1973/2001). Fleksja polska. Warszawa: Wydawnictwo Naukowe PWN.
Tomaszewicz, B. (2009). Subjunctive mood in Polish. In G. Zybatow, U. Junghaus, D.
Lenertova, and P. Biskup (Eds.), Studies in Formal Slavic Phonology, Morphology, Syntax,
Semantics and Information Structure: Proceedings of FDSL 7 (pp. 221-233). Frankfurt am
Main: Peter Lang.
Willim, E. (1989). On Word Order: A Government-Binding Study of English and Polish.
Kraków: UJ.
Witkoś, J. (1995). Wh-extraction from clausal complements in Polish: A minimality/locality
account. Folia Linguistica, 39, 223-264.
Wurmbrand, S. (2007). Infinitives are tenseless. U. Penn Working Papers in Linguistics, 13(1),
407-420.
Marcin Orszulak
University of Wrocław
Institute of English Studies
Kuźnicza 22
50-138 Wrocław, Poland
e-mail: [email protected]
Unauthenticated
Download Date | 6/19/17 6:19 AM