ABSTRACT
NH MOVEMENT AND MULTIPLE QUESTIONS IN CZECH
Ludmila VeselovskA
•In the following work some aspects of Wh Movement in
Czech are presented, with the main fochs on multiple s.
• wh-queStions.
In Chapter 1 the simple Mon-echo wh-questions and
indirect wh-questions are shown. Wh' Movement frdt non-CP
infinitival structures is indicated' as standard in Czech,
while long-distance Wh Movement from the embedded CP clauses
is opJy . marginally acceptable.
In Chapter 2 the multiple wh-questions in Czech are
displayed in more detail with a brief survey of some relevant
literature. The phenomena discussed concern word-Order,
position of inserted elements, clitiC position, and some
substandard possibilities of extraction from the embedded-
multiple wh-questions. The arguments given in Chapter '2 as
a whole support the IP-adjoined analysis of the non-first
wh-words in the multiple wh-questions in Czech.
Dissertation submitted in partial requirement for the Degreeof Master of Arts in Applied Linguistics to School of English
Durham University
1993
fi
ACKNOWLEDGEMENTS
I would like to thank the University of Durham and the
British Council, as well as the PalackY University in Olomouc
and my family at home, which enabled me to Study, and provided
me with all the financial, moral, and emotional support that
was necessary during the course. The following work also would
hardly come into being without the help of my friends in the
Czech RepubliC who never refused to spend their.pvenings going
• through my endless lists of crazy sentences...
To. • be a student at Sehool of EngliSh at the Durham
University was a pleasure for me. I owe thanks and excuse to
all' my teachers: while I was making them bear my occasional
excesses, they have introduced me into spheres of knowledge
which I have found mostly reasonable or/and exciting. The
patience of my supervisor, Joseph E. Emonds, is unforgettable.
None of them can be blamed for any of my faults in the
following work, bUt any of its positive features should justly
be accounted for by their influence.
TABLE OF CONTENTS
INTRODUCTION ............................................
1 SOME CHARACTERISTICS OF WH MOVEMENT IN CZECH
1.1 Simple direct.wh-questions ........................ 3
1.2 Indirect wh-questions .............................. 8
1.3 S-structure position of the Czech wh-elements ...... 10
1.4 LON:distance Wh Movement ......................... 17
1.5 Summary of Chapter 1 .............................. 26
Notes to Chapter 1 ...................................... 28
MULTIPLE QUESTIONS IN CZECH 31
2.1 Introduction ..................................... 31
2.2 Some previous analyses of multiple questions ....... 32
2.3.1 Word-order, Focus, and related phenomena ...... 37
2.3.2 Parentheticals, subject position, . wh-order ..... 41
2.4.1 Wh-words and clitic position ...................... 47
2.4.2 A note about the IP structure ..................... 52
2.4.3 Conjoined wh-words ................................ 57
2.4.4 Clitic affix and the particle fe ....................... 59
2.5 Wh extraction from the multiple wh-question ....... 64
2.6 Summary of Chapter 2 .............................. 67
Notes to Chapter 2 ...................................... 71
Bibliography ............................................ 75
INTRODUCTION
In the following work some aspects of Wh Movement in
Czech are presented, with the main focus on multiple
wh-questions. I have chosen this topic mainly because there is
no other substantial literature about CzeCh in a given
framework, and Wh Movement seemed to be restricted to the
periphery of the sentence. Even if it was not possible not to
mention the word-order distinctions, it was possible to avoid
many equally complex phenomena demanding appropriate analysis:
A number of issues were also mentioned and left unexplained
since they were beyond the limits of this work and its author.
Chapter 1 Some Characteristics of Wh Movement in Czech
In Chapter 1 some general information about Wh Movement
in Czech is presented, demonstrating the simple non-echo
wh-questions in section 1.1, and embedded wh-questions in
1.2. In section 1.4 extraction of the wh-element from
infinitival structures is discussed, and examples of .
long-distance Wh Movement are analysed. The position of the
[+wh) complementisers is discussed in 1.2 and some
characteristics of the bridge structures are suggested in 1.4.
The chapter presents Czech as a language with a syntactic
movement of the interrogative wh-element into a SPEC(CP)
position, with a standard and obligatory movement of the
wh-element from the embedded infinitival structures analysed
as IP's or VP's, and with a restricted substandard
long-distance Wh Movement from the embedded finite clauseS.'
Even if in Chapter 1 no detailed analysis is presented of any
phenomena, as a whole the Chapter represents an introduction
which was required because of the lack of literature
concerning Czech for the following analysis of the multiple
wh-questions.
Chapter 2 Multiple Ouestiens in Czech
Chapter 2 presents an analysis of the multiple
wh-questions in Czech. The relevant literature is briefly
mentioned in section 2.2. In sections 2.3- 2.5 some aspec'ts of
Czech multiple questions are displayed in more detail,
concerning the relation to the focus of the sentence, .
Word-order, position of inserted elements, clitic position in
Czech, and limited possibilities of extraction from. the
embedded multiple wh-questions. The arguments given in Chapter
2, as presented in this work support the analysis of the
multiple wh-elements in Czech as claimed .in e.g. Lasnik and
Saito (1984) for Polish, or Rudin (1988) for polish and 'Czech,
i.e. the SPEC(CP) position for the first wh-element, and the
IP-adjoined position of the subsequent, non-initial wh-words
in the multiple wh-questions in Czech.
In this work no explicit theoretical framework is
presented, but the arguments are baSed on the tasnik-and
Saito's (1984) analysis of Wh Movement and Chomsky's Barriers
(1986). The works by other authorsare cited if relevant for
the text, and are mostly not incompatible with.the same
framework.
CHAPTER 1
SOME CHARACTERISTICS OF WH MOVEMENT IN CZECH
Simple direct wh-questions
Wh Movement in direct and indirect questions has long
been analysed as a kind of Wh Fronting when a wh-element
generated inside the clause moves into the pre-sentence
Position; e.g. in Emonds - (1976:188) Wh Movement is presented
as a ' substitution of a phrase node dominating WEI for the
sentence-initial COMP node'. Since Chomsky's (1977) On Wh
Movement the phenomenon is taken for a kind of constituent
extraction the left, comprising question extraction
together with relative clause extractions, topicalisations,
etc. I am not going to deal with any of the other structures,
even if in the Government & Binding framework they are all
analysed in a similar way: the main reasons being a lack of
structural equivalents of some of the phenomena in Czech and
a,latk pf space for a detailed analysis of the distinctions.
In simple direct questions Wh Movement applies within one
clauSe as e monocyclic transformation. Wh Movement is,
however, a movement that can (at least in English and many
other languages) operate across many clausal boundaries, being
a form of an unbounded dependency construction. As an example
of a more general Move Alpha transformation, Wh Movement is
subject to general rules of movement and language specific
parametric Variation. Its interclausal application ' is
restricted with the COMP-to-COMP Condition ('Move a wh-phrase
to CORP' Van Riensdijk and Williams, 1986:64) to a movement
from the COMP position to a higher COMP position only. The
obligatoriness of Wh Movement is a result of interacting
principles of subcategorization and interpretation.
Van Riemsdijk & Williams (1986:65) give examples
(repeated below) of the obligatory Wh Movement in English
(la), and of the optional Wh Movement in French (lb). (lc)
presents the Czech equivalents showing the obligatory
syntactic• movement of the wh-element in a direct non-echo
wh-question.
(1) (a,b: VR&W,1986:65)
(a) John ate what? : echo-question interpretationWhat did John eat? . : direct non-echo wh•question
(b) Tu as vu qui? / Qui as-tu vu?you have seen who / who have you seen
(c) Jenidek snddl co? : echo-question interpretationJohnny ate what
Ty jsi videl koho? : echo-question interpretationyou saw who
Co OnAdl Jenidek? : direct non-echo wh-questionwhat ate Johnny'What did Johnny eat?'
Koh+) jsi vid41? : direct non-echo wh-questionWho saw-you:'Who did you see?'
The position of the moved wh-elements in direct non-echo
questions is more recently stated as 'pre-C position' for the
specifier-initial languages, such as English or Czech (e.g. in
Radford, 1988:501). In the following sections I am going to
presume the 'C-specifier analysis' of Wh Movement as presented
e.g. in Chomsky's Barriers (1986).
The following examples (2-4) present the variety of
constituents that can undergo Wh Movement in a Czech direct
non-echo wh-question: in (2) a subject NP and its internal
constituents, i.e. AP and PP, in (3) an object NP and its
internal constituents, in (4) an adjunct NP. (The traces are
marked, but without detailed analysis 'of the original
extraction site, simply to aid understanding.)
(2)(a) Kdo t ti to dal?whow tw to-you it gave' Who t gave it to you?'
yterd ddvde t se ti libi nejvice?which girl tw to-you likes best'Which girl t do you like best?'
* S dim [mut t pfinedl Pavlovi to knihu?* with what [man tw brought to-Paul the bookLit: 'With what man t did bring the book to Paul?'
* MIA a aim [ t pfinesl Pavlovi to knihu?* man with what, [tw brought to-PaUl the bookLit: 'Man with what t did bring the book to Paul?'
(2): to extract a wh-subject NP is possible; to question an
adjective phrase within the extracted subject a noun phrase is
possible as well. (2c-d) show that to question a noun phrase
within the prepositional phrase postmodyfiing the subject noun
phrase is ungrammatical.
•3a,b) show the same as (2) for an object noun phrase,
even if to extract an adjective phrase from the object noun
phrase seems to be possible, as shown in (3c).
Koho mma MAtenka nejradéji t ?who has MASenka!Who does Masenka most-likelike most t ?'
Jaky dixm / jak velky diam si chce koupit t ?which housev/how big housew wants-he to-buy tw'Which house/how big a house does he want to buy t ?'
? qaky si chce koupit t dAm?? which
'Whichw wants-he (to)buy tw house
does he want to buy a t house?'
(d) * S aim si chce koupit clAlm t':?* with what, wants-he to-buy houseLit: 'With what does he want to buy
tw
a house t ?'
There :Ls no significant asymmetry between the extractions.
from NP in the positions of subject (2) and adjunct in the
following (4); both are equally wrong.
(4)(a) Kam / Na co jsi to polcAil t ?wherew / on whatw did-you it put . tw'Where / On what did you put it t 7 ,
(b) ? Na jakk jsi to polc;i1 ( t stul7PP •
? On which, did-you it . put ( tw table?Lit: 'On which did you put it t table?'
(c) * Jaky jsi to polofil [ PrIna t* whichw did-you it put [ PPon t , table?
', .14t: Which did you put it on.t table?
Leaving aside the explanation of the distinctions betWeen
acceptability in (3c/d), it is possible to say, that extraction
from within the noun phrase is not possible in Czech in away
comparable with. English: The examples in (2-4) - present.
extraction from the NP which is an external argument (2), an
internal argument (3) and an adjunct (4). They show that the
Nmax presents a barrier for movement. The distinct levels of
acceptability may be stated as the difference between the
extraction of an N complement, whiCh is easier. han the'
extraction of an N'complement. In (3c) the object NP is
L-marked by a verb which subcategorises for a NP complement
and Case-marks it, thus voiding the barrier. 1
Consider than the following (5) repeating the impossible
stranding of prepositions in Czech demonstrated already in
(4c). Evpn if some examples can be paraphrased by a single
verb, pied piping of the preposition is obligatory for .10
(b)
(c)
(d)
0 hem mu bude povidat t ?about what him will-he tell t'About what will he tell him t ?'
(b) * dem, mu bude povidat o t ?* - whatLo d . him will-he tell about two
'What will he tell him about t ?
(5)(a)
Movement and excluded for NP Movement in Czech. Example (5b)
shows preposition stranding in Wh Movement and (5c) in. NP
(passive) movement, whete the stranding becomes 'doubly'
unacceptable because while in (5b) it would be theoretically
,passible to relate the adequately case-marked whpronotn to
the stranded preposition, in (5c) the nominative case demanded
for the subject of a passive verb clashes with the case
required by the preposition. The ungrammatical (5d) then shows
the pass.ivization of a prepositional verbal complement.
(c . ) * 116co bylo dohodnuto na t.* something,' was agreed on t-ACC
(d) Nam (* Na nêco ) bylo dohodnuto t.somethingNom (*on somethingACC) was agreed-ACC- t.
The 'reanalysis' of a verbal complex (of a 'Natural
Predicate'), as presented for NP Movement in English in e.g.
Radford"(1988:431-433 or 496-498) into one 'semantic unit' is
not poSsible with a Czech [verb + preposition].
In Kayne (1981) the impossibility of stranding the
piepositiOn is presented as a result of a different Case
:assignment by a verb and a preposition. ' P can assign oblique
• case • .only to an NP for which it is subcategorized whereas
1 v can assign objective Case somewhat more freely,.
particular to any NP that it governs' and ' reanalysis between
two lexical categories is possible only if the two govern in
-. the same 'way.' ( Kayne, 1981:363-4). This would mean that in
Czech, as in French and contrary to English, prepositions
cannot govern structurally. Kayne's hypothesis relates this
phenomena to the absence of 'exceptional Case-marking' in
French. However, in 1.4. examples of IP infinitival clauses in
Czech are presented, which are analysed as 'exceptional
Case-marking' structures. There are also prepositions' in Czech
( marginally) introducing infinitival (IP) clauses and
therefore Kayne's suggestion for French cannot be applied fOr
Czech without more detailed analysis.
The impossibility of stranding a preposition in Czech
together with the above mentioned NP constraint show thg.t in
Czech direct .non-echo wh-questions it is posSible to extract
only the complete constituents (maximal projections)
immediately dominated by IP or some V projection. Extraction
of a wh-element of any lower maximal constituent in a simple
direct wh-question is never fully grammatical.
1.2. Indirect wh-questions
While direct questions are structures in which the
interrogative clause is an independent sentence (as..e.g. all
the examples in the above section), indirect questions are
complex sentential structures. The interrogative part in the
indirect question is an embedded clause which is an argument
of a matrix clause verb. 2
An indirect question is a kind of indirect speech in
which the reproduced proposition has the form of a question.
Indirect speech is usually introduced by a subordinating
go' I . 'to go'are used
question
subject,
??togo.'
* pfijit-li.* to come •
• behaviour in several structures.
In Bresnan (1970) the following examples (9a,b)
to show that if is ungrammatical when the indirect
appears at the beginning of a complex Sentence'in the
i.e. NP, position.
(10) 'John doesn't know, whether/where he will
(a) Jan nevi, JESTLI/ZDA/ZDALI ptijde. /John not-know-he, ?COMP will-come-he /
(b) Jan rievi, pfijde-LI.John not-know-he, will-come-he-?COMP '/
moved into the sPEC(CP) in the Czech wh-questions (following
Chomsky's Barriers). On the other side, complementisers, e.g.
the [-wh) that, are standardly expected to appear in the head
position, i.e. C, with the exception of [+whl complementisers,
or at least with the exception of some of them: those which
are analysed to be in SPEC(CP) as well. In English the
distinct positions are argued also for the two [+wh)
complementisers if and whether because of their distinct
questions are translated as zd., -1i, zdali or lest21 in
Czech. (see e.g. Petr (1986:220-227). Their use is presented
in (10), which shows that when introducing -finite•
subordinate clause, all of the wh-elements, as in EfigliSh, are
grammatical. With an infinitival verb, however, none of the
complementisers seem to be ideal, while the wh-words (10c) are
fully acceptable. (The complementisers are represented•in the, •
translation only as '?COMP' or '7'.)
(c) Jan nevi, KAM / KDY ptijde. / ptijit,
(9) (Bre, 1970:310f) (a) Whether he'll come is not known.John not-know-he,WHERE/WHEN will-come-he / to come-
(b) * If he'll come is not known.Among other differences between whether and i f in.
Another structure where if, contrary to ; whether- , is not
acceptable is introducing the infinitival clause, as shown
e.g. in Borer (1985:76) on the examples (9c,d).
(9) (B.4)1. 4985:76) (c) John doesn't know whether to leave.
(d) * John doesn't know if to leave..
In a footnote Borer (1985:106) cites B.Palek who claims
that similar distinction appears in Czech. In the following
text I am going to present examples showing that with the.
exception of semantic distinctions, there is no difference
between the syntactic behaviour of the Czech
wh-complementisers, and in the same time that their position
is apparently distinct from that of the other wh-constituents.
The English whether and if introducing indirect yes/no
11
English, as mentioned in Emonds (1985:286-291), the
impossibility of if occurring in case-marked NP positions
following prepositions is discussed. The following (11a) shows
that all Czech complementisers are ungrammatiCal when preceded
by a preposition: (11b) gives the correct form with
a case-marked resumptive pronoun to 'it'. 6
(11)(a) * Patrani po ZDA/zDALI/JESTLI/ptiel-LI, bylo•marne.* investigating of 7/7/? came-he/? was in Vain'Investigation of whether/*if he came, was in vain:'
(b) PAtrâni po tom, zda/zdali/jestli bylo maVle.investigating of it(1,0C); 7/7/? came-he, was in vain.
The following example (12) shows, that all Czech
wh-complementisers also lack the inherent 'positive meaning',
which seems to be contained' in whether but not. in IX.
12
in the Czech translation as
kekla mi; "Zitratold-she me: tomorrow' She told me, "I will
well.
to tady potkäm."you here will-meet-Imeet you here tomorrow."'
rotind
.(6)(a)
conjunction ffe 'that' in Czech, while indirect questions are
' introduced by a wh-element. At least since. Baker (1970),
indirect wh-questions are accepted as presenting the same kind
of mil Movement as the direct non-echo wh-questions.
In Banfield (1973) indirect speech is claimed to be
derived independently from direct speech constructions, and
a number of distinctions are presented concerning structures
and elements which appear in English exclusively in one or the
other form. 3 In (10) (which is Banfield's simplified (1))
e.g. Personal pronoun ( you/me) and spatial and temporal
demonstrative elements ( here/there, tomorrow/today) are
changed according to the new context. The same changes are
( b) kekla mi, 2e dnes me tam pOtkS. •told-she me, that today me there will-meet-she
• ' She told me that she would meet me there today.
In English also the tense is related to the new context
( will/would) following the English tense shift rule for
indirect speech, while in Czech the tenses in direct and
indirect speech are identical (in (10) it is the future).
As for the Subject/Aux inversion in direCt wh-questions,
in Czech at least some auxiliaries or finite verbs appear
preferably in pre-subject position (in case of a standard
elauSe with, no element stressed). A possible example is given
In (7).
( 7) Komu bude KryetAfek (?? bude) pomehat?to-who will KrygtAfek (?? will) help,' Who will KryStilfek help?'
Although the example (8) shows a similar distinction for
an indirect wh-question, the level of acceptability of the
standard affirmative word-order in the indirect question is
much higher than in (7b)
(8)(b) Ptal se, .komu buds EryAtIlfek (?bude) pomehat.asked-he, to-who will Kry6tAfek (?will) helpHe asked who Kryetilfek will help?' •
(7) and (8) then demonstrate that even if the inversion in
Czech questions need not be obligatory, it is, unlike in
English, possible in both direct and indirect questions at the
PF level. 4 Whether this inversion is a syntactic movement of
the auxiliary or finite verb into the C position (as an in
English questions) or some later 'Stylistic Inversion'
adjoining the subject to the VP, which is mentioned in e.g.
Bouchard (1989), is not going to be decided here. 5
1.3. 8-structure position of the Czech wh-elements
All the Czech examples of wh-questions, both direct and
indirect, presented above contain an ;.overt [-1-wh] element in
a 'presentential position'. The presence of a wh-word or
wh-complementiser at the beginning of a sentence is supposed
to indicate the presence of the S-bar, i.e. CP projection,
which is the position of either the wh-element itself or of
some operator binding it. Within Stowell's (1581) concept of
the CP projection, however, two positions are available:
Specifier CP = SPEC(CP) and COMP = C. In the previous section
the extracted wh-words (i.e.constituents) were assumed to be
10
(12) * chtal vedgt ZDA/ZDALI/JESTLI nebo ne prgi.He wondered whether / * if or not it rains.
(10), (1/), and (12) present the structures where the
English if complementiser would be the equivalent of all the
Czech forms. No difference between the behaviour of the Czech
complementisers could be observed, leaving aside the fact that
only zda is used introducing causative and only iest1i
conditional, clauses. The position of the wh-complementisers
was evidenced as distinct from the other wh-words. Therefore
C, i.e. the position of the English if, is assumed here to be
the position of all the Czech wh-complementisers.
The analysis of the -// postfix complementiser supports
the above claim. Assuming the movement of the finite,
inflected verb, i.e. of the verb containing the TENSE and AGR
features, into the I position (as presented for e.g. French in
Emonds (f978:165), or Pollock (1989:366)), there is no reason
to expect the infinitival verb (containing no TENSE and no AGR
features) to undergo the same process. The distance between
the position of the infinitive inside VP and CF may be enough
to prevent a suffix appearing anywhere inside CP to appear on
infinitival forms (as presented in (10b) and (13b)).
Then there are still two possibilities how to relate the
finite verb and the affix complementiser: either to suggest
(13)(a) * Ptal se, NAgenka* Asked-he, Magenka came-?Litt 'He asked, whether Magenka had come.'
(b) Ptal se, pkiSla-LI MagenkaAsked-he, came-? Magenka'He asked whether Mdgenka came.'
The latter variant assumes the movement ofI Containing
the finite verb into C, excluding again the infinitival forms
simply by the fact that infinitival verb cannot become the
member of V+.1 cluster. The movement of I into C, howeVer,
would be blocked by the presence of a complementiser, if -1i
were present in C, because the 0-wh) features are,'contrary to .
that and similar 'neutral' complementisers, supposed to be
present at the D-structure (see e.g. Bresnan (1970:315f). The
solution would be to assume that the [+wh] features are'
present in SPEC(C) but become overtly .realized by the -/i
suffix on the head C later on (as may be suggested by its
position in the end of the word). The postverbal position of
the subject, as presented in (13), would then he predicted.
A similar suggestion is mentioned also in Toman (1992): 7
The analysis presented above, then, may. be enlarged on
the other' Czech [+wh] complementisers as well. S. will assume
their position in C (even it they contained the (+whr-featureS
appearing in SPEC(CP)), and the position of the" other
wh-constituents in SPEC(CP).
both of them in I, or both of them in C. The former variant
would predict that in such structures the subject will precede
the verb. Consider then the following example (13)
demonstrating the obligatory postverbal position of overt
subject in clauses with the -/i postfix complementiser.
13
In Lasnik & Saito (1984) a number of' LF filters are
presented, which define the characteristics of interrogative
complements. Two of these filters are cited here as (141.
(14)
'(183) A 1-4-whj Camp must have a (+whj head.. (184) A [-wh] Comp must not have a (4-wh] head...
14
(15) Polish: (L&S,1984:75)
Spotkalds metczyzne,you met the manLit.: 'Whomi did you
(16) Czech:
(a) * Potkal jai mute, kterl-koho zabil?* you met the man who whom killed
(b) . ? Potkallsi mute, kter7 zabil koho?? You met the man who killed whom
(c) . * Koho i jsi potkal mute, kteil zabil t.* who did you meet en who killed ,et the m h killd ti
•
Comparing the Czech examples (16) with the Polish
embedded 'relative Clause/question (15), it shows that while
'Polish need not have a [i-wh] interrogative COMP specified at
S-structure, i.e. a relative clause can be understood as an
indirect . question as well, the Czech equivalent (16a) is
ktOryi kogo 2who whommeet the man
t1 zabil t2killed
who killed t.?'
(187) if a language L has syntactic Wh Movement, (183)(184) apply at S-structure in L.' ( L&S,1984:287)
and The difference between the Czech verbs taking embedded
questions and embedded propositions is presented in (17) and
.L&5 demonstrate an example of an.interrogative relative
clause in Polish which contains an interrogatiye wh-element
following a relative wh-element. For the authors the relative
•pronoun occupies the SPEC(CP) position and the interrogative
. wh-element is adjoined to IP. That is why they take the
filters. (14) as LF filters and not S-Structure filters for
Polish. Consider the comparison of Polish and Czech below
(18) below (notice the distinct complementisers).
(17) (a) Kdo vi, . jestli / * to si to Kuba koupil?Who knows whether/ * that it Kuba bought?
(b) Kdo vi, co / * to si Kuba koupil?Who knows what / * that Kuba bought?
(c) Kdo vi, jestli / * te' si co Kuba koupil?Who knows whether / * that what Kuba bought what?
(17) shows that in Czech, unlike Polish, if the verb
subcategorizes for an interrogative clausal complement, i.e.
for an embedded question, its COMP must contain a fronted
eleként in S-structure: either [+wh] complementiser
or wh-constituent (17b). The L&S LF filter (183, here in
seems relevant for the S-structure in Czech.
Example (18) shows that if a verb subcategorizes for
a (-wh] complement only (18a), i.e. for an embedded
proposition, its complement must not contain a [+wh] element
in the . S-structure (19b), suggesting that the L&S's LF filter
(184, here (14)) applies in S-structure in Czech as well
((18a) is acceptable as an echo-question).
(18)(a) * Maruka si mysli, ie si Toma§ek koupil co?* MaruAka thinks that TomaSek bought what?
(b) MaruLka si,mysli, to /*jestli /*co si (to) Tomd6ek•koupil.MaruAka thinks that/*whether/*what (it) TomAgek bought.
There are two ways that the [-wh] embedded proposition in
Czech can be questioned, both of them enlarging the scope of
the (+wh] to the matrix clause. The following (19) shows the
standard correct form where the wh-word is syntactically
ungrammatical. .p.6b) is acceptable as an echo question only.
(16c) shows the . impossibility of a movement of the wh-word to
' the . higher SPEC(CP) position, presenting a Complex NP
constraint violation (i.e. Wh extraction from the finite
relative clause modifying a noun).
1516
related to (ir subcategorised and case-marked by) the matrix
verb. The emb dded clause is not an indirect question but
a kind of a elative clause with a resumptive pronoun (in
(19), to 'it' which is a clitic) related to (subcategorised and
case-marked by the verb in the embedded clause.
(19)
0 Lem si [Marugka mysli t (le si to [ Tomggek keupil t ]]]About whati(Marugka thinks ti[that iti.[Tomagek bought til'What does Marugka think thaE Tomagek sought?!
(20) presents the colloquial variant containing long-distance
movement of the wh-word to the matrix clause SPEC(CP).
(20) % Co.1[Marugka mysli (te si [TomAgek koupil ti]]]]
What- [Marugka thinks [that [Tomagek bought ti]l)/What does Marugka think that Tomagek bought?
The characteristics of the matrix-clause complementiser
position in Polish are given by L&S as ' we assumed that
Polish is distinct from English in that the matrix Comp is not
specified for a value of (whj.' ( L&S, 1984:284). (20) shows,
however, that a matrix clause in colloquial Czech may be
specified for a value of [+wh] and that long-distance Wh
Movement is not strictly prohibited in Czech.
1.4. Long-distance
Extraction of the wh-element in Czech is fully standard
from.the infinitival structures. The following (21a,b) shows
infinitival complements of modal and 'want' verbs. Notice the
obligatory climbing of the clitics mu 'him' and to 'it'
17
originating in the embedded clause, and obligatory subject'
control of the infinitival subject. (Clitics are written in
Capital letters.)
(21)(a) Karel MU TO musel/chtgl e odevzdat t tKarel himcl itc2 must-ed/wanted e i to-give tc tc2 :' Karel had/wanted to give it to
( b)Komu TO Karel musel/chtel (*komu) odevzdat t t 7to-whowl itc2 Karel must-ed/wanted (*to-who) to-give t wl ec .2who did Karel have/wanted to give it?'
The following (22) presents Exceptional Case Marking
structures following the verbs of perception and 'believe'
' type of matrix predidates. In (22a) the obligatory plitic
climbing '(of to 'it') and obligatory object control of the
infinitival subject are indicated again.
(22)(a) Karel HO TO videl/nutil e dglat t.Karel him itm saw/made ec to-do tm .Karel saw/made him do it.'
(b) co HO Karel videl/nutil (*co) delat t ?what,, him Karel saw/made (*what) to-do'What did Karel saw/made him do?'
Notice that in neither (21b) nor (22b) is the'wh-element
acceptable as an element introducing the infinitival
Structure. This indicates that neither (21)hr (22). present
a CP clause. Using the concept of the missing CP projection,
the properties indicated above appear natural. If the landing
site for the extracted wh-word is in SPEC(CP) position, than
the only one available is the matrix-clause SPEC(CP). Since no
Subjacency effects are noted (and a clitic movement is
obligatory), I suggest the above structures to be preferably
VP structures in (21), and IP•struCtures in (22).,.8 In.any caso
-the maximal projection is L-marked by a matrix verb and 'as
18
such does not present a barrier. The trace in the embeddeda Czech equivalent acceptable example of English
infinitival structure is assigned a Case by the infinitive,long-distance Wh Movement (see Chomsky's (1986:29)).
the intermediate traces bay be thought of as .adjoined to VP
(following Barriers), and the wh-word appears in the position
of the operator indicating the scope of the wh-question. Both
lexical and antecedent government are possible in (21-22) and
no Superiority distinctions can be observed.
Compare the standard Czech obligatory Wh Movement from
- the structures in (21) and (22) with the infinitival
pomplements in (23). Following the above criteria, (23) is
assumed to be a CP structure. In contrast to (21) and (22)
notice the clitics which in (23) cannot appear inside the
matrix clause, the 'subject of the infinitive of arbitrary
reference, and the possible insertion of the wh-element in
front ot the infinitival clause.
(23)(a)
laigta (*TO) vdddle/vyzvidala, komu TO e odevzdat t t .Miga.I(*it) knew/wondered, to-who,, it,2 BARB give t, 1 t,•44isa knew/wondered, who to give it to.T
( b) komu Mi6a vödela/vyzvidela (TO) odevzdat (TO)?• - * to-who' MiSa knew/wondered (it) togive (it)
Lit: 'Who did Miga know to give it?'
The' matrix clause predicates in (23) are subcategorised for
a [+wh] complement and the embedded infinitival complement
must be overtly marked for [+wh]. (23b) shows that further
extraction of the wh-element is not possible, since the
infinitival clause would lackany overt [-Ewh] complementiser.
The. UngraMmaticality of (23b) resembles the
ungrammaticality of the extraction of the wh-element from the
embedded -finite clause as presented in (24) which gives
(24) * Jak si Jan mysli tys opravil to auto t* how John thinks you're repaired the car t
'How does John think you fixed the car t 7°
Long distance Wh Movement is said (see L&S (1984),
Radford (1981:237), Anyadi and Tamrazian (1993), Riemsdi.jk &
Williams (1986:294)) not to appear in Polish, Russian,
Armenian; or German, and in Chomsky's early works is presented
as restricted (the WHisland condition) also in English. L&S
(1984:274), however, show examples of long distance movement
of the wh.-elements extracted out of subjunctive complements in
Polish. Their Czech equivalents are given in (25a,b) together
with two example's of substandard long-distance Wh Movement
from the embedded finite clauses (25c,d).
(25)(a) ? Co chtel Jakub, aby Lenka koupila t-7What wanted Jakub. that Lenka bought t' What did Jakub want that Lanka buy t 7'
(b) ? Kdo chtal Jakub, aby t koupil chleba?who wanted Jakub that t bought bread'Who did Jakub want that t buy bread?'
(c) ? Co vedal Jakub, to ma 2' Lenka koupit t 7what knew Jakub that should' Lenka buy t'What did Jakub know that Lenka should buy t ?'
(d) 7 Kdo vadel Jakub, 2e t ma koupit chleba?who knew Jakub that t should buy a bread ''Who did Jakub know that t should buy a bread?'
Both (25a,b) and (25c,d) have standard forms connecting the
wh-word directly to the matrix-clause verb with resumptive
pronouns in the embedded clause, as shown in (19) above.
The examples (26) and (27) of an inserted parenthetical
clause (26) and of an acceptable long-distance Wh Movement
20
simple form, if it is followed by a structure analysable as
CP. Both this characteristics are going to be briefly
discussed in the following paragraphs.
In Chomsky (1977) long distance Wh Movement is stated as
' a langvage-specific COMP-CORPwh-pbrase from CORP to a higher(COMP, X, wh-phrase, vbl), wherespecial properties'.
movementCOMP over aX contains
rule (44): movebridge' and '(45)a VP with certain
( Ch., 1977:85)
The COMP-COMP movement can, then , be blocked by conditions on
(27) indicate that the colloquial form of long-distance Wh
Movement presents a distinct structure. 9 Notice the presence
of the complementiser ge ' that', punctuation (a pause), and
the position of a clitic AUX affix -s = jsi :are'.
(26)
(7jsi) tikal jsi, (*te) mu to JardUfka dale t 7(?are) said are-you, (*that) him it aareafka gave twyou said, did Jardafka give it to him?
(27) Rde's tikal, to (*) mu to Jarlafka dale. t 7where'rew said, that (*) him it JarCiafka gave tw'Whore did you say Jardilfka gave it to him?'
In both (26) and (27) the fronted wh-word is case-marked and
belongs to the subcategorisational frame of the second
predicate. In (26) the parenthetical and wh-question remain
syntactically independent (the clitic in the parenthetical
follows 'the first constituent' within the parenthetical, the
complementiser is unacceptable, commas mark pauses). In (27)
the clitic follows the extracted wh-word and the
complementiser marks the second clause as subordinate.
The introductory matrix-clause clause in Czech, however,
compared to the similar structures in English, allows only
little variety of predicates and seems to demand the most
rules of interpretation, ie. 'the bridge conditions', which
are relevant for the structure of the matrix clause and
characteristics of its predicate. In Van RiemSdijk and
Williams (1986:294) the 'bridge verbs' are suggested . tobethe
verbs that can make the following CP A'-transparent, which•.
means that the COMP (non-argument) position in the embedded
clause becomes accessible to external government.
It seems that the criteria restricting .the . number - of
possible 'bridge verbs.' can be derived from the fact that
indirect questions are wh-complements of the matrix-clause
predicates. As stated in Grimshaw (1979),
6 ... Ed• a'predicate-complementPair to' be well formed, threeconditions mist be satisfied. The predicate and, 16s complementmust be semantically compatible; the complement must meet theidiosyncratic selectidnal conditions encoded in the semanticframe of the predicate; and the complement must meet the (alSoidiosyncratic) syntactic conditions . encoded in - -thesubcategorization frame of the predicate."
( Grimshaw, 1979:325,)
The subbategorisation frales of matrix verbs expresS-
codcpurence restrictions on prediCates in terms of syntactic
categories, specifying the optionality or obligatorineSs of
the sister phrasal constituents for which the, predicate As
subcategorized. Embedded questions are,Characterized by'the
Rde,wherew ,' Where,
subcategorization frame containing an optional or Obli:gatery
sentential complement, e.g. find out, V:+ I s t WH].
The Czech verbs of communication are mostly ditransitive
and their complements are both + NP, NP and + NP, clause.
At the same time .the complement selection is only optional and
they may be used as the verbs of action requiring no,
argUment. The vague subcategorisation frames of most of the
semantically acceptable 'bridge verbs' present a problem,
21 22
since they hardly ever exclude the possibility of relating the
fronted wh-,word to the matrix clause predicate, which may be
subcategorised for optional complements of the sate kind. 10
Consider also that there is only a limited number of, .
asyndetIc complex sentence structures in Czech, and that
a subordinate : clause must be introduced by an overt
complementiser `(as seen in the ungraimatical (24)).
Combining the obligatory presence of a complementiser
. with a possible requirement of the matrix clause verb on the
overt [±wh] specification, of its clausal complement, the
results may be predicted and are presented in the following
examples ..(28):: If a connecting eletent is obligatory and
subcategorized as [+wh], then it cannot be removed from the
embedded clause (28a). If the selected connector may be I-wh)
as well, the wh-eleMent may move to the main clause SPEC(CP)
and be'replaced with a neutral that (28b). If the selected
connector must,be [-wh] than the wh-element moves (280. 11
.(28)(a) 'Ptd se, *2e /jestli/komu to delaAsk-he, *that/whether/who it gave-sheHe asks *that/who/whether she gave it to'
.* Komu se ptd, 2e to dela?"* Whom ask-he, that it gave-she?
' Who did he ask she gave it to?'
Aikame, te/komu to 41a ( Krygitiifkovi)SaY7we, that/who it gave-she to (K.)'We say that/who she gave it to (K).'
Komu tikdd, 2e /(*) to dale?WhoM say-I, that/(*) it gave-she?Who do you say she gave it to?'
( c) • Myslim si, 2e/*komu to dela.Think-I that/*who it gave-she.'I think who she gave it to.'
Komu si mysliA, 2e to dela?who think-you, that it gave-she'Who do you think she gave it?'
23
Consider then the example (29) respecting the above
requirements (the embedded clause is introduced with
a subcategorised [4-whj complementiser, while another wh-word
is removed into the SPEC(CP) of the matrix clause)
(29) as ptd, jestli to?? Komuask-he, whether it?? Whomdid he ask whether'Who
The (29) example presents a Wh Island Constraint
violation which is in L&S analysed as resulting from the
position of the [+wh] element in the SPEC(CP) of the embedded
clause, where it blocks a cyclic movement of the other
wh-element ..., The [+wh] features in Czech occupy the SPEC(CP)
position also in Czech, and therefore the LEIS (1984) analysis
can be used for the Czech examples as well.
Comparing the Wh-extraction from infinitival vs. finite
structures, and the restrictions stated above, tha conditionsrestrict
which #ieuMmui* the ,acceptability of the long distance Wh
Movement in Czech are as the following (30)
(30)(A) . the unambiguous analysis of the relations betWeen thewh-word and matrix clause vs. embedded clause predicates,(b) a [±wh] specification of the matrix clause predicate,(c) the presence of the CP projectOn.
The analysis of long distance Wh Movement presented in
e.g. Chomsky (1986) demands traces to be properly governed' and
no more than one bounding node for subjacency to be crossed.
With regard to the levels of unacceptability, the violating
government of the trace, i.e. the Empty Category Principle, is
suggested to present usually worse results than violation of
subjacency.
Following the above concept, the violation of (30a)
24
dale?gave-she?she gave it to?'
results in Czech in structures which are usually not
analysable, while violation of (30b,c) presents only a certain
lower (substandard) level of acceptability.
Since there is no difference between the extracted
obligatory complement (25a,c), subject (25b,d), or adjunct
(26), the standard distinction between lexical vs. antecedent
government . does not seem applicable. In Barriers the
intermediate traces of , the moved wh-constituents are
subsequently VP adjoined and placed in SPEC(CP) positions. If
any of these positions would be defined as accessible for the
government by the matrix clause verb, then the violation of
(28a) could be stated in terms of a 'closer potential
governor' interfering the chain link, which would result in
a violation of the Empty category Principle. 12
The restriction on movement presented by Lasnik and Saito
(1984) for Polish, i.e. prohibition of the syntactic movement
from an A' position, would on the other hand present a kind of
subjacency violation. The bounding nodes for subjacency may be
felt in colloquial Czech to be distinct from standard Czech.
In Barriers the difference between Italian and English is
discussed, and 'the parametric variation is restricted to
subjacency, not government, so that "extra barriers" have no
effect on adjunct movement.' (Chomsky, 1986a:39) Assuming that
standard Czech takes both tensed IP and CP for barriers to
movement, while in colloquial Czech only one of them is
relevant, we may get the observed distinctions. Any such
statement would, however, require more detailed discussion,
and analyses of also other possibilities of extraction, which
is beyond the scope of this work.
25
1.5. Summary of Chapter 1
The syntactic Wh Movement of a wh-element into, the
presentential position in direct non-echo wh-questions in
Czech was introduced in 1.1. The variety ofsnoved wh-elements
comprises the maximal NP, AP, and PP constituents immediately
dominated by IP or some V projection. The difference between
the wh-element in the pre-complementiser position and in situ
(i.e. in echo-questions) is clear and suggests the origin'al
post-verbal extraction site of the removed internal argument
constituents.
In 1.2 and 1.3 some properties of indirect questions were
presented, showing that the complementiser of the embedded
question is obligatorily specified for a (1-Wh) interrogative . •
feature at the S-structure, and the complementiser of .the
matrix clause can be specified for [+wh) in colloquial Czech.
The [-F.wh] complementisers in Czech appear in a form of suffix
realised on a finite verb, and as separate words. The:position
of all the [A-wh] complementisers is assumed to be ..in Comp;
while those of the wh-wordS in SPEC(CP), as in Chomskyr(1986),
In 1:4. some examples of extraction of:the Wh-elements
were demonstrated. The 'Wh Movement from the infinitiVal
compleMents argued to be non-CP structures was shown as
obligatory in standard Czech, while the movement from within
the CP infinitival complements was presented as acceptable
only for some speakers. The CP projection represents a barrier
for external government, movement of clitics and Wh Movement. -
Another restriction for a movement of the Wh-word into-the
26
bridge structures, which -must preferably be analysed as not
allowing any structural relation between the fronted
wh-element and the matrix clause,predicate.
NOTES TO CHAPTER 1
1 Some acceptable counter-examples to the aboveevaluations may be explained by a kind of ambiguityallowing analysis of the postnominal prepositional NPmodifier as an adjunct. (5b) indicates that when suchan ambiguous analysis is semantically excluded, thesame structure is unacceptable.
(a) V jakem obalu ptines1 tu knihu t ?in which cover brought-he the book t'In which cover did he bring the book t ?'
(b) * S jakym ndzvem ptines1 tu knihu t 7* with which title brought-he the book t
'With which title did he bring the book t ?'
2 For more detailed characteristics of the introductory,predicates see e.g. Grimshaw (1979). Some are'preSehted in the following section, too.
3 Banfield (1973) states more distinctions than arepresented in the following paragraphs. While theoccurrence of the 'expressive elements' in direct andindirect speech in Czech if comparable with theirdistribution in English, to compare the grammaticalstructures referred to in her 'last-cyclic or roottransformation' section is complicated by a freeCzech constituent-order, which makes it difficult todistinguish any special constructions fortopicalisation or dislocation.
The differences between grammatical vs. ungrammaticalword-order can in most cases be better viewed asdifferenCes between marked vs. unmarked varieties. Inthe following text I am going to use the marking '?'or '.??' for the marked word-order, where 'marked'means 'pragmatidally marked', i.e. possible and fullyacceptable but in some context only, to make itdistinct from the starred examples, which . areungrammatical, i.e. not acceptable in any context.
main clause .pre-sentential position was derived from the
. obligatory overt [+wh] marking of the clause interpreted as
,a question.
,;The long-distance Wh Movement is presented as
e.,substandard variant motivated by obligatory selection of
[±wh] clausal complements by a matrix clause predicates. The
restrictions on long distance Wh Movement presented above
result from the restrictions on the government of the
intermediate traces, i.e. froM the properties required by
The prethence of a Cp projection represents a barrier that
cannot be crossed without the use of the lower SPEC(CP)
position. The examples given above indicate that Czech
respects the Wh Island restriction, suggesting that the
SPEC(Cp).of the embedded clause is used for cyclic movement.
. The distinction between acceptability of the long-distance Wh
Movement in standerd and colloquial Czech was suggested to be. -
a result of a distinction in perception of the bounding nodes
for Subjacency in each of:them.
Bouchard (1988:155) cites Xayne & Pollock (1978), whoclaim that in French a wh-phrase in COMP triggers thesubject adjunction to the right of the VP. A similarpossible solution is finally suggested here in 2.4.2for Czech as well.
Any complementiser would be ungraMmatical followinga preposition, however, in Czech. A resumptivepronoun would be necessary after the preposition with
28
all of them, suggesting that the case assigned bya transitive preposition must be realised in anadjoined nominal element (i.e. Case is to be realisedmorphologically).
Toman (1992) presents -/i and the finite verb ' asa zero-bar phrase resulting (in somewhat simplifiedterms) from Age or NegPhrase° Incorporation. In thisstruqture 71i is the complementizer, hence it appearsin Cu .l(Toman, 1992:117)
A possible distinction between the VP and IP analysesmay be presented as e.g. the restrictions on thebinding domain for the anaphoric reflexive clitic se(-self). As indicated in the following examples thereflexive anaphors remain within the ECM=IPinfinitival clause (b), but move from the SUBJECTlessVP's (a).
Karel SE musel/chtel vykoupat (*SE)Karel ; selfi/*k had/wanted to-bath (*self)'Karel had/wanted to take a bath himself.'
Karel RAS (*SE) nutil/vid61 t koupat SEKarel; usk (*self) forced/saw t to-bath selfv *i
-"Karel made/saw us to take a bah ourselves.'
In'Lenerz (1985) examples of apparent long-distancemovement in German is presented for the structurescontaining the verb 'weenen "believe' in the matrixclause. The author does not take it for an argumentthat a COMP node in German may become an escape hatchfor the embedded clause element, but claims, that'waen may have lost its verbal character and may
have been regarded as some kind of sentential adverbor particle like bitte, 'please' and danke 'thankyou'. Also it may be the case that speakers did notassume there to be a S'-boundary between the matrixpredicate and its complement V/final 'clause', thelatter being considered as similar to the infinitivalcomplement.' (Lenerz, 1985:113).The examples presented by Lenerz, however,exclusively concern fronted embedded clauses and notWh Movement, and the 'when' structures are presentedas .the only acceptable ones. On the other hand inAnyadi and Tamrazian (1993) Ruhr German is shown asapplying long-distance Wh Movement in a way roughlycomparable with Czech, and several 'bridge verbs'(i.e.glauben or sagen) are used.
An example of such interference between the matrixclause predicate and the fronted wh-word occurs alsoin the following example, with a ' pro crossover' in
29
(c) where the pronominal pro is presented as Aboundby a wh-element, violating thus Principle B of theBinding Theory. similar examples may be mistaken forarguments suggesting the impossibility of subjectextraction, i.e. of the ungoverned position: ofa subject even in Czech. (Notice the indexing of thewh-element indicating its relations to the predicteagreement features.)
Kdo tikal, to ptijde?(a) whoi t i said-he i that proi/A will-come-he i/A
'Who said that he would come?'(0) * whoi pro said-heA , that ti will-come-hei
'Who did hesay (that) will come?'(c) * whoi prof said-hei that ti will-come-hei
,The subject wh-word is morphologically marked fornominative of the 2rd person singular pronoun. If itprecedes a finite verb with AGRfeatures for irdperson. singular with a null-subject, the emptyposition for subject will always be taken for a Whtraceooindexed with the subject wh-word. Thereforethe bridge structure with null subject and 3psagreement is not a structure permitting long distancemovement of subject.
11 The kind of 'obligatory' movement presented here maybe compared with the obligatory movement ininfinitival structures lacking the CP projection. Inthe finite clauses there is a CP projectionavailable; this projection, however, issubcategorised for a [-wh] element. The subjunctiveclauses are always [-wh] clausal complements and thelong-distance Wh Movement from these structures istherefore predicted.
(a)
(b)
12 Recalling the Riemsdijk and Williams concept of 'Al-- transparent' bridge predicates, it may bepreferably the lower SPEC(CP) position that beComesgoverned, given the difference between the [+wh] and[-wh] complement selection as would suggest .(30b).The concept would however have to become more'A'-non-transparent', at least for the Czech examples:presented here, if it is to prevent the intermediate •trace from being interfered with. Alternatively"thefact that the finite verb is in the I position ats-structure in Czech may result in the governed trace .of the moved wh-element adjoined to the matrix clauSeVP, which would capture. the interference . of thesubject wh-trace as well, (if it is ,the INFL positionfrom which subject is governed).
In such a concept - adjunct wh-words are alwaysstructurally ambiguous and their relation to eithera matrix clause or embedded clause , predicate is •a result of semantic or pragmatic factors only.
30
2.1. Introduction
the preceding
a language with syntactic
CHAPTER 2
MX.H.,TIPLE QUESTIONS IN CZECH"
chapter Czech was presented as
Wh Movement placing the wh-elementS
at the beginning of the wh-question. This characteristic makes
czech comparable with e.g. English, but while in English
multiple wh-questions only one wh-element appears in the
vresentential position and all the others remain 'in SitU', in
Czech' multiple wh-questions (MWhQ) all wh-words move to the
'.'_presentential position' at the S-structure. Examples of the
Czech direct and indirect MWhQ's are given in (1)
-(1)(a). Kdo Co koupil?Who, what 2 [t 1 bought t
(b) Kdo koupil co? .................... echo queStionWho l [ ti bought what]
(c) Zajima më, kdo co ptinese.wonder-I whoi what 2 [t1 brings t 2 ]'I wonder who will bring what.'
(d) Zajima m4, kdo pf-inese co. ..wander-I .whol [t1 brings what]
(la) is a direct MWhQ, (lb) is acceptable only
on.Oo 'what' as an echo question, (lc) is an
and (1a) is again an echo form of embedded MWhQ. 1
In dealing with Wh Movement in terms of an operator
binding its variables, all the wh-elements appear at LF in the
positions where they can bind their traces within the scope
required, i.e. at the LF level all the wh-elements are to be
- found in the relevant SPEC(CP) position. The examples of
MWhQ's in Czech may suggest that the wh-elements in
pre-sentential, position are in their LF positions at the
S-structure already, i.e. no subsequent LF movement into
SPEC(CP) is required. Such an analysis would, however,
contradict the observed characteristics of the multiple
wh-words in Czech, and also face certain theoretical problems
mentioned in the following sections.
2.2. Some previous analyses of multiple questions
The abstract question morpheme Q posited as a 'meaningful
deep structure segment with a performative reading in Katz and
Postal (1964) was 'revised' in Baker (1970) and applied to
both direct and indirect wh-questions. Baker's Q-Morpheme was
introduced as an operator binding wh-variables (i.e. the
questioned constituents which were derived by means of the Wh
Insertion inside the clause and co-indexed with the operator).
Baker's Replacement Rule places one of the wh-elements in the
pre-sentential position, where the presence of the rah-element
triggers the Subject/Aux Inversion in,' the following direct
questions in e.g. English. The co-indexing of wh-words with
their traces enables more than one of the wh-elements to be
bound inside the clause, which makes it possible different
structures for sentences with different scopes of separate
wh-words. The Replacement Rule, however, predicts that only
one of the wh-words is fronted. Similar predictions were
derived also from Bresnan's (1970) analysis of'the [+wh]
complementiser, which may be viewed as a more syntactic
echo question
with a stress
embedded MWhQ,
31 32
5'
COMP S
COMP
(2)(Teman,:82:(27))
COMP
WH1
counterpart to the abstract Q-Morpheme.
The operator status of the clause initial morpheme Q was
challenged in e.g. Kuno and Robinson (1972), where the
ambiguity of some MWhQ's was explained by pragmatic factors.
Baker's mechanism of Q-binding was exchanged for the clause
Mate Constraint requiring, multiple •wh-words with the same
scope to be clause mates at the time of application of Wh
Movement. The authors pointed out a relation between the
fronted full constituents and wh-words, and formalised it in
the Double Dislocation Constraint prohibiting the dislocation
of more than one constituent.
In Wachowicz (1974) the 'single Wh question movement'
derived from Baker's Single Q-morpheme Hypothesis was rejected
because of the existence of MWhQ's with more than one
wh-element fronted in e.g. Polish and . Russian. Such multiple
structures seemed to contradict the concept of Baker's
Replacement Rule, since 'after the replacement there is no
Q-morpheme left that could lexjcalize'. ( W,1974:164).
Wachowicz argued that all the wh-elements in the multiple
questions are moved by a question movement (i.e. they are not
'in situ') when fronted. She mentioned the relation between
the 'free word order phenomena' which enables the languages
showing multiple fronted wh-words to also extract more than
one topicalised element into the pre-verbal position, and she
rejects Kuno and Robinson's (1972) Double Dislocation
constraint on movement. Wachowicz did not propose any
formalised concept for the MWhQ's; all restrictions that she
applied on the MWhQ's were based on discourse constraints. The
same analysis was repeated in Wachowicz (1978)•
An attempt to explain the MWhQ's within the . 'concept of
unique Wh Movement can be found in Taman (1982)• The author•:
claims that even if only one position in Comp is available
universally, there are still at least two acceptable analyses
of the MWhQ's whiCh do not contradict the multiple filled'COmp
constraint: either the fronted wh-words form a single complex
constituent, or not all the wh-elements are in - Comp, Toman "
demonstrates examples in Polish, Czech, and Russian, .which
support the second of the possible explanations,i.e.that the
wh-wOrds are not a single constituent, and their parts,do not
appear in the same positions in the sentence. The presence pf
adjoined Wh-elements in pre-sentential position is suggested
to be a result of scrambling, which appears in all Slavio
languages. 'However, the assumption that scrambling' is '
adjunction to S is not particularly clear' ( 41982:299 •). The
structure which Toman offers for the MWhQ is given in the
following (2)
Taman (referring to Reinhart (.1981)) mentions also
a possibility of a COMP analysis which would make more than
one node in Comp available, because such an 'extra' position
could be used for the non-initial fronted wh-words'.
Reinhart's (1981) analysis of the Hebrew COMP is, however,
better understood as an argument for SPEC(CP) and C nodes
33
CP
SPEC(CP) IP
IP
IP
CP
SPEC(CP) IP
( t )WH 2 ( t )WH3 ( )
within the CP projection, and is hardly useful for Toman's
alternative:analysis.
In Lasnik and Saito (1984) Stowell's (1981) concept of CP
'. dontaining. SPEc(cP) and C is used for the analysis of Wh
. Movement. Polish examples of MWhQ's are analysed with the
first wh-word!in the SPEC(CP) position and the other wh-words
' adjoined to IP: This 'IP-adjoined' analysis is followed also
in e.g. Haegemann (1991)as a supporting argument for the
Government and Binding concept of the cyclic Wh Movement
through/into the SPEC(CP) position.
In Rudin (1988) the author argues that the sentence
structure and conditions on movement, as proposed by Chomsky
in Barriers, do not exclude either adjunction to IP or to the
other non-argument maximal projection, i.e. to CP. Rudin
claims that both the possibilities presented in the following
(3) appear, i.e. IP-adjunction (3a) and CP adjunction (3b) of
the non-initial wh-words.
(3)(b)
Rudin argues that Bulgarian and Rumanian MWhQ's represent
the structure ( 3a). She calls those languages
'Multiple-Wh-Fronting Languages', replaces the Doubly-Filled
S-Structure or LF.
Czech and Polish area according to Rudin (1988)
'non-Multiple-Wh-Fronting Languages' which respect the
constraint on SPEC(CP) adjunction at S-Structure and PF. This
specification groups Czech; Serbo-Croatian, and Polish with
English, French, and Chinese (Rudin, 1988:494). What makes all
these languages different is the Wh Movement of the multiple
wh-words at S-Structure: while Czech and Polish move all
wh-words into A'.-positions, English moves only the first one,
and Chinese moves none. The following (4) paraphrases Rudin's
schemes . , for Czech (4a) and English (4b) S-Structure positions
of the multiple wh-elements.
(4)(a) ...Czech (4)(b) ...English
The structure (4a:IP-adj) in Rudin's framework predicts
that Czech prohibits co--occurrence of wh-words with overt
complementisers, respects the WH Island Constraint,' and allows
MWhQ's with more than one wh-word moved to A' positions in
random ordering, since all the traces are A'-hound and
governed at the required levels (Rudin, 1988:495). 2
(3)(R,88:(2c,d))
(3)(a) s'1 1
Comp S Comp Sor or I
SpeJcCP SpecCP WH
'WH - WH WH
S.
. comp Constraint with the 'Condition on SPEC(CP) Adjunction
Constraint' •Rudin, 1988:490)), and claims that the above
In Rudin's (1988) article most
in Bulgarian, some in Rumanian, and
of the examples are given
Serbo-Croatian. Czech is
languages do not respect the Constraint at either PF or in many places mentioned only briefly. 3 In the following
35 36
sections I am going to follow Rudin's (1988) arguments, to
show that even if the choice is not unambiguous, the
preferable analysis for the Czech MWhQ's is the structure
presented above as (4a:IP-adj). I am not going to refer
systematically to any other language, Polish being the only
exceptiqn, since Polish is presented as the closest equivalent
to Czech.
2.3.1. Word-order, Focus, and related phenomena .
To argue that the MWhQ's in Czech are a phenomenon
distinct. from something like English MWhQ's presented in
abundance in e.g. Bolinger (1973), one argument should be
advanced: it is necessary to show that all the fronted
wh-elements in Czech are moved, which need not be the case if
Czech Were entirely a free constituent-order language. 4
In the above Chapter, however, the 'echo-questions' were
presented, and the position of wh-elements in such 'in situ'
structures was taken for the extraction site in both simple
and MWhQ's. Also all the authors dealing with this phenomenon
made a distinction between stylistic reordering and multiple
Wh Movement into some syntactically definable position(s);
such positions being derived from the characteristics of the
wh-words.
The origin of the wh-words is standardly assumed as
inside the questioned constituents. In the pre-Q-Morpheme
analysis in Chomsky (1957) the wh-words were derived from
definite- pronouns and fully-specified constituents.. In Katz
and Postal (1964) from indefinite pro-forms, : and similarly in
Bach (1971) the interrogative pro-constituents are claimed to
be based universally on indefinite pro-forms. The author also
relates the position of wh-words to the position of the theme
of the sentence.. The relation between the focused full --
constituents and multiple wh-words in the sentence structure
is pointed out also in Kuno and Robinson (1972), Wachowicz
(1974), Toman (1982), and Rudin (1988). The relation between..
indefiniteness and the theme, however,. is not`-so obvious.
Czech is taken for a language in which the:' correlations
between the order of syntactic constituencies in a sentence
and the discourse role of the information which a particular
constituent represents' is the main parameter for the surfaCe,
word-order (Green, 1989:128). In Firbas (1992) the notion of
definite vs. indefinite in Czech is discussed, and the
equivalentS of indefinite articles in e.g. English are
unambiguously taken for 'signals of rhemacity', not of the
focus (Firbas, 1992:21).
The interrogative wh-words are then exceptions, since in
wh-questions the 'interrogative word is alwayS the feCus of
the question' (Petr, 1986:595). The ftonting of the wh-words -
may be therefore motivated by a general tendency te.place . the
thematic elements to the beginning of.the claUse. The question
remains, however, which syntactic position (if any) can be,
defined as a position of a focus and .whether such.a position. „ .
is really the position of the fronted wh-elements.
The position of a wh-word in a simple wh-question_ • is
SPEC(CP), and therefore the assumed fronted focus position
would preferably inbe SPEC(CP) as well. The IP-adjoined.
3738
analySis of the Other wh-words would contradict such
a Position of the focus even if the adjunction of the
non-initial wh-words to IP in NWhQ represents a kind of
compromise if SPEC(CP) becomes unavailable.
An exclusive syntactic position for Focus is presented in
e.g. Brody (1991) as an FP projection of the head F, situated
betweehIF and CP. While the [-hwh] elements move to SPEC(CP),
the. focused elements move to the specifier or head of.FP. The
Focus 'Criterion presented by Brody (1991) is a variant of.
Rizzi's - (1991) Wh-Criterion requiring a mutual specifier-head
Configuration of the focused X ° and XP. Within this framework,
howeVer, the non-initial wh-elements are not supposed to befocused or topicalised and are suggested to be standardly
adjoined to a lower IP.5
•In Dimitrova-Vulchanova (19_92) a construct FRONT, is
introduced to apply to the description of a clause-initial
structure in Slavic free constituent-order languages. In both
main and subordinate clause FRONT dominates the top IP
projection (in Dimitrova-Vulchanova's terminology the maximal
functional V-main projection) and seems to be replacing both
SPEC(CP) and C. The standard position of subject, i.e.
SPEC(IP), is suggested to be identical with FRONT, which 'is
to.be,seen as a landing site for all topicalized constituents
and fronted verbs' and it 'can contain material non-specified
for -category'. (D.-V.,1992:2,31). FRONT must be filled with .
a given number of elements only, and clitic clusters
right-adjoin to the filled FRONT. The author states the number
of possible elements in the Czech FRONT as one (in Bulgarian
there are at least three possible). Then, if. such
a one-member Czech FRONT precedes clitics (which seems to
describe the reality perfectly), it cannot be taken for the
place where the non-initial wh-words in Czech appear. For the
non-initial wh-words following clitics in Serbo-Croatian,
Dimitrova-Vulchanova advances another 'secondary fronting site
which immediately precedes the main verb and where 'mildly'
fronted constituents (such as the rest of wh-constituents in
multiple wh-constructions) occur'. (D.-V., 1992:44). The
position of the secondary FRONT is suggested to bp rather
V-main than IP adjunction, but V-main apparently means some
upper functional IP projection as well.
I do ' ,not reject or propose the existence of some
T(opic)P or F(ocus)P following CP, but the attempt to state
a new syntactically unique projection for the topicalised
elements in a free-constituent language, as presented in the
above paragraphs, puts together all kinds of Wh Movement with
topicalization, pragmatic and stylistic reordering, in a way
which is not fully justified (at least it is not justified up
to now). Therefore F am not going to use the notion of Focus,
Topic, or FRONT as terms strictly relevant for the syntactic
position of the other wh-words, presenting the above analyses
as possible explanations for the leftwards movement of all the
wh-werds in the Czech MWhQ's. A similarly vague suggestion is
presented also in Toman (1992), where both the clitic and
multiple Wh movements are claimed to be 'overt manifestations'
of the movements which result in 'obliterating the distance
between the :syntax of Logical Form and that of Surface
Structure' (Toman 1992:117).
39 40
2.3.2 Parentheticals, subject position, and wh-order
The word-order examples used for the analysis of the
Czech and Polish MWhQ's in Toman (1982) and in Rudin (1988),
argue for the (4a:IP-adj) analysis, and they are as follows:
The wh-words in Czech multiple question do not form a single
constituent, and the non-initial wh-words precede the unmarked
subject of the clause. Moreover the wh-words do not follow any
strict ordering, i.e. they are freely exchangeable, which
supports the rejection of a (3a:SP(C)-adj) structure requiring
in a given framework the shared SPEC(CP) position to be
indexed by only one (the first) of wh-elements.
The multi-constituent character of the multiple wh-words
is derived mainly from the position of inserted clitics, which
is presented in more detail in the next section 2.4.1, and
froth the possibility of inserting a parenthetical clause among
the wh-words. The following example (5) repeats Rudin's
examples (49) with an inserted parenthetical clause.
. _l'he evaluation may differ, but in any case the' inserted
parentheticals suggest the multi- (at least two-) constituent
nature of the multiple fronted wh-words, which is
contradictory to the (3a:SPEC(CP)adj) structure. 6.
The pre-subject position of all the fronted wh-words,
which forms one of the main arguments for 'the -(4a:IP-adj)
analysis in TAS (1984), is presented also for Polish in .Rudin
(1988), and its Czech equivalent is shown in the 'following (fi)
(6)(R, 1988:489)
(a) Polish: Kogo komu Jan/on przedstawil?Czech: Koho komu Jan/on ptedstavil?
whom to-who Jan/he -introduced'Who did Jan/he introduce to who?'
(b) Polish: * Kogo Jan/on komu przedstawil?Czech ? Koho Jan/on komu ptedstavil?
whom Jan/he to-who introduce
(6b), unacceptable in Polish, makes Rudin reject the
possibility of adjunction to. VP. 7 Contrary to Polish, the:
Czech equivalent in (6b) with the subject preceding the other
wh-word, is evaluated as only 'marked', and in fact many less
(5) ( R, 1988:469)
(a) Rdo, podia tebe, co komu dal?who, according to you, what to-whom gave'Who, according to you, gave what to who?'
(b) Rd° co (? podle tebe) komu, podia tebe, dal?who what(?according to you) to-whom,according to you,gave
Rudin states that the position of parenthetical clauses
is fully optional in languages , like Czech. It' would be more
precise to say that all the positions are possible, but within
the range of marked vs. unmarked variants, the parenthetical
is unmarked either following the first or the last wh-word.
41
marked examples could be given with subjects preceding the
non-initial wh-words, as e.g. the following (7) and (8).
(7) Co jsem jA komu kdy uddlala?what am I to-whom when done'What have I ever done to anybody?'
(8) (a) ? Co jsem kdy komu - já uddlala?what am when towhom I done
(b) * Co jsem jd uddlala kdy komu?what am I done when to-,whom
(8b) presents a possible echoequestion, showing' that .
pre-verbal position of wh-words is obligatory, while (8a)
42
shows that` in such a sentence the subject following all the.
wh-Words is not especially salient. Notice, however, that
while (7) is certainly a kind of wh-question, the English
translation suggests, that its interpretation-is not likely to
. he a multiple wh-question.
In Bach (1971) 'the indefiniteness assumption' for the .
interrogative words is presented as 'consistent with the fact
that :interrogative words and indefinite pronodns are often
morphologically related or even identical' ( Bach, 1971:158),
and examples are given in Japanese, Greek, German', etc. In
Wadhowicz (1974) the indefinite pronouns in Polish are
presented as being 'almost exactly as the interrogative
pronouns, the only difference being that the indefinite
, prohouns have a specificity marker.' ( Wachowiez; 1974:159). The
.specificity marker has a 'tendency to be dropped' in Polish.
WaOhOwicz shows examples of 'multiple questions' with only one
wh-element moved and all : the-others in post-verbal positions.
The author calls such sentences 'clarifying questions' that
are
interpreted by most speakers as yes/no questions
containing indefinite pronouns. An example with its Czech
structural equivalent is given in (9)
(9) Polish: Nie rozumiem, kto Wyjechal kiedy?Czech.: * NeVim, kdo odjel kdy.
not-understand-I who left: . when'I don't understand: who left when?'
- The Czech variant of the Polish 'clarifying questions'
would be an echo-question, since the second wh-werd does not
move from the post-verbal position.
The Czech indefinite pronouns are derived from the
interrogatives with the ná- 'some' prefix as presented in the
following (12).
(10) Kdo / nakdo Co / ndco Kde / nekd etc.who/somebody What/something Where/somewhere
The relation is moreover not only morphological. In Petr
(1986:568f) MWhQ's are mentioned from the point of view of
appropriate answers, and some (nonspecified) MWhQ's are given,
to show that only the first interrogative pronoun is
interrogative (requires the answer), while the others may as
well be interpreted as indefinite pronouns. Such questions
then do not differ semantically from the questions containing
only one (the first) interrogative element. 8
The indefinite interpretation of the 'non-initial
wh-words in (7) is close to the English any or -ever pronouns.
Moreover the required answer for such questions is negative.
For the example presented in (7) it would be 'Nothing (...to
nobody, never)'. (If the first wh-word is answered, the answer
is fully acceptable).
The wh-words following the 'subject and preceding the verb
in Czech, even if moved, contrary to the Polish 'indefinite
pronouns!, may then be analysed as not interrogative.
Therefore their adjunction to some IP projection lower than
the top one, which seems to be evident in (7), cannot be taken
as a position of the interrogative wh-words in multiple wh
questions in Czech, but as a position of indefinite•(from the
point of view positive/negative?) elements. 9
If the post-subject positions of the non-initial
wh-elements in the Czech MWhQ's may be excluded by the
4344
'Who gave what to who.'
Kdo komu co dal, nevfm.who to-who what gave I don't know.
? Co kdo komu dal, nevfm.what who to-who gave I don't know
?,, Komu kdo co dal, nevfm.„
to-who who what gave I don't know.
? Co komu kdo dal, nevfm.what to-who who gave I don't know
In the aboye example only (12a) is likely to be understOod
a rather unambiguous multiple question; the others seem to
indefinite/negative interpretation of the non-initial
wh-words, than the interrogative wh-words in Czech
obligatorily precede the overt subject. Both interpretations,
then, argue against the (4b:in situ) structured» Czech.
The third argument for the (4a:IP-adj) structure in both
Taman (1982) and Rudin (1988) concerns the mutual ordering of
the fronted wh-words. The authors demonstrate that no such
obligatory ordering in Czech can be found. The following (11)
repeats the Rudin's examples (61)
(11)(R,1988:475) 'Who invited whom when I don't know.'
(a) Kdo kdy koho pozval, nevfm.who when whoM invited I don't know.
(b) Kay kdo koho pozval, nevfm.when who whom invited I don't know
(c) . Koh° kdy kdo pozval, nevfm.whom when who invited I don't know
If all the
examples in (11) are acceptable, then within
the concept of LF Wh Movement, the head of COMP indexing, and
proper government, as used in L&S (1984), in any of
(3a:SP(C)-adj) or (4a:IP-adj) or (4b:in situ) structures all
the wh-traces must be lexically governed at appropriate levels
in Czech- Recalling that no pure Superiority constraints were
observed in the Section 1.4 above for the long-distance Wh
Movement either, such a conclusion is even strengthened.
Rudin (1988) makes optional ordering one of the main
arguments for the difference between the (3a:SP(C)-adj) and
(4a:IP-adj) structures, since when all the wh-words appear
inside one SPEC(CP) node (as claimed for Bulgarian in Rudin
therefore non-lexically governed wh-traces must be "excluded by
appropriate ordering. Rudin's claim, however, need not bb
unambiguously accepted, since the following, (12)"
reveal that with arguMent wh-words the ordering Subject
- indirect object. --direct object seems to be preferred also
in Czech.
require the answer for the first interrogatiVe word -only,
i.e. they may be interpreted as simple wh-questions With -
indefinite/negative pronouns. Since such a possibility is not'-
excluded either for (11b,c), the lexical , government of subject
may be not so clear in Czech, and the Ordering of the fronted
fah-words cannot be used as a decisive argument for or againSt .
any of the structures offered above,
In the following section the mutual position of clitiös
and wh-words is shown, which supports the multiconstituent
character of the multiple wh-words .suggested already by the
inserted ParentheticalS and by the position of the subject.
(1988)), the resulting cluster can get only one index and
45 46
(15) Batka MI TOBdtka me it'Betka gave
dela (* MI TO) / * Bétka TO MI dala (TO MI)gave (* me it) / * Batka it me gave (it me)me it.' 'Betka gave it to me.'
' Wh-words and clitic position
The possible relation between the positions . of weak
: pronouns and wh-words in Polish and Russian MWhQ's was
mentioned already in Wachowicz (1974:160). The author,
- however, presented arguments for the difference between the• -
movement of the weak pronouns (Clitic Movement) and the.
movement of the wh7constituents ( Wh Movement). Her Polish
eXampleslare repeated in the following (13) and (14) with
their. Czech equivalents (clitics are written in capital
letters-in the Czech sentences).
(13)(Pol;W,1974:(20))
Polish: (a) Monika to widziala (b) Monika widziala to.Czech: Monika TO vidala * Monika vidala
Monica it saw Monica saw it
Czech: (c) Videla TO.saw-she it.'She; saw•it'
, For the Polish (13a) Wachowicz claims the movement of the
clitic to 'it' from the. post-verbal position to the pesition
preceding the finite' verb to be optional, i.e. (13b)
acceptable. The Czech examples (13a-c) present the only
grammatical possibilities in Czech, with the clitic folloWing
the finite affirmative verb only if it means becoming the
second constituent, as in (13c) . 10 The example (14) presents
similar structures with a wh-Constituent.
' (14)(Pol:W, 1974:(23)) 'What did Monica give to who?'
Polish: (a) Co komu Mooika dela? (b) * Co Monika komu dala?Czech: - Co komu Monika dada? ? Co 14onika komU dala?
. What whom Monica gave? what Monica whom gave?
The Polish starred (14b) compared with grammatical (13b) makes
'47
Wachowicz claim that ' there is no necessary connection between
pronoun movement and the position of wh-words in the
sentence... (It) indicate(s) that wh-words in Polish and
Russian are moved by a question movement' (W,1974:161).
The Czech examples in (14) show the two structures with
different interpretations of the non-initial wh-elements:
(14a) is a multiple question with both wh-words interrogative,
(14b) is a simple wh-question with the other wh-words
perceived as indefinite/negative pronouns, as discussed in the
previous section:
Both the Polish and Czech examples in (13) and (14) argue
for the (3a.'SR(C)-adj) or (4a:IP-adj) structures, and both of
them exclude the (4b:in situ) variant. They show that the
clitic movement in Czech is more constrained than the movement
of the non-initial wh-elements. Compare the example (14) above
with the following (15)
The clitic position in (15) is one of those quite rare
word-order varieties in Czech where grammaticality judgements
are sharp. (15) shows that the ordering of the clitics is
obligatory [indirect object - direct object], and they move to
the 'second position' obligatorily (other ordering or position
of the clitics are perceived as ungrammatical). Similar
principles are observed by' the wh-words only in unmarked
forms, and with a dropped subject the distinctions between the
different interpretations of the non-initial wh-words become
even more vague.
48
In Toman (1982:295) the concept of the difference between
the clitic and wh-movement is accepted and the Polish examples
demonstrate the mutual positions of the clitics and the other
wh-words. The Czech examples are given in the-following (15)
(16)
(a) Kdo MU TO kdy . dal? (b) ?? Kdo kdy MU TO dal?who him it when gave who when him it gave'Who gave it to him when?'
(c) * Ado MU kdy TO dal?who him when it gave
The impossibility of inserting a wh-word inside the
clitic cluster in (16c) makes evident the distinct position of
clitics and wh-words. The unmarked (16a) then shows that
a clitic position preferably precedes the non-initial
wh-word, thus demonstrating the multi-constituent character of
the wh-,words.
The position of Czech clitics has been repeatedly
mentioned in this work, without any structural suggestions. In
the following paragraphs some brief analyses of the clitic
position in Czech are going to be presented.
Referring to Zwicky, Anderson (1993) states the
differene between the simple ( ' purely phonological in
nature') clitics and special clitics (A,1993:74). The clitics
described in this work are then 'the special clitics', i.e.
accentless, mostly pronominal elements: in Czech weak personal
pronouns, weak reflexive pronouns, some adverbial and AUX-be
elements, and discourse particles. The ' clitic placement in
Czech involves movement into the Wackernagel Position' (Toman,
1982:301), where the clitic elements appear as a group right
after the initial stressed word of the sentence (Petr,
1986:619)•
In Anderson (1993) the relation between the functional
categories and morphology is discussed, and clitic placement
is presented as an example of evident interaction between the
morphology and syntax. The 'second' position of clitics may.be
in various languages defined either phonetically (the first
stressed word), syntactically (the first constituent), or
pragmatically (the focus). In Toman (1982:298) the author
claims that the Czech clitics follow ' the first maximal
be definede •
are analysed
with the Sulle result, and the following (17) is proposed
(17)Let us assume that at least one of{ the following
statments is true:(32) A clitic must be locally Supported by Comp.. .
(33) A clitic must be locally supported by Inn. If the first condition is true, its instantiation in Czech
will result in right-adjunction of clitics to CoMp.-If thesecond condition is true the instantiation in Czech willrequire left-adjunction to Infl
illax . Both instantiationsammount to the 'Wackernagel effect' in terms of stringadjacency.' (Taman 1991:162)
Let us recall the choice relevant here, i.e. between the
paraphrased (3a) and (4a) structures. The possible . positions
of clitics are marked following Toman's suggestions above
(3')(a)
CP CP1
r---- 1 iSP(CP)
1
C' SPEC(CP)t
1i
1 : 1SP(CP) WB 2 C : C
. 11 r-11-1 , (
WH1 0
constituent, i.e. the clitic -position can
syntactically. In Toman (1991) reflexive clitics
Cr
( 4') ( ) '
SP(CP) Will 0 (c1)1 IP
(c1) 2 Iv
t t 22
IP
( c1) 1t -IPI
WH2
(c1) - IP
t1
t2
4950
In (7), analysing the clitic AUX -Isom 'am' as adjoined to C,
the subject in SPEC(IP), and the finite verb immediately
dominated by there is no position available for a maximal•
Assuming that clitics in Czech usually precede the other
wh•words (as shown in (1.6)), the structure (4a) with the other
wh-wordS" left-IP-adjoined would suggest the clitics
right-adjoined to C, i.e. (4a 1 :IP-adj.) above, since such
a structure would make the ordering obligatory. Assuming the
orderingnot unambiguous, the left7IP-adjoined clitics, as in
(1a2:IP-adj.), would allow the other wh-wprds to have
a certain amount of freedom since all the.elements would be
adjoined to some projection of IP. Any of these concepts seems
to exclude the (3a:sp(c)7adj) variant, since it predicts that
the clitics obligatorily follow the wh-words.
However, within Stowell's (1981) concept of the
bar-notation, and restrictions On movement as presented in
Chomsky (1986:4-6), the dooccurrence of clitics and
wh-elements in the same place does not seem possible, assuming
the clitics and wh-words to have a distinct status: wh-words
being maximal projections of a ,given constituent, while
clitics are adjoined to the.heada. Then, the movement of the
wh-constituent can be a movement to some SPEC position, or an
• adjUnction to a nonargument maximal category only. The
movement of a X category, on the other hand, would be a Head
to Read' movement, i.e. in case of the clitics, an adjunction
to another X0 category.
In the above conception, the (4a 2 :IP-adj.) variant; i.e.
both . clitics and wh-words adjoined to IF seems to exclude the
-ordexj.7ng jclitic + non-initial wh-word] since the wh-word
would be adjoined to some maximal projection of IP, while the
clitics would follow, being adjoined to (its) head. It seems
evident that theoretical assumptions also prefer the
(4a:IP-adj) structure, which provides either the COMP or some
INFL head for clitics.
2.4.2. A note about the IP structure
The Czech MWhQ's presented in this Chaptercontain the
finite verb which contains all verbal categories in Czech, and
therefore it Was assumed to appear in the top IP projectiOn
head, m-commanding the subject in the nominative. Since the
interrogative multiple wh-words precede both of them, no
detailed analysis of the IP structure was presented up'to now.
In the above section, however, the concept of the bar notation
and conditions on movement were briefly sketched, which are
accepted in this work, and within such a framework at least
a brief nate concerning the inner structure of IP is needed,
if some definable position is to be 'stated for the two
possible interpretations of the the multiple non-initial
wh-words. 11
Recall those wh-words interpreted as indefinite/negative
pronouns, which were presented as appearing after the subject
but obligatorily preceding the verb. The example (7) is
repeated below
(7') Co jsem já komu kdy udgaala?what am I 'to-whom when done'What have I ever done to anybody?'
51 52
cP
SPEC(CP) C'
IpTOP
(l8)(=-7)'What have I ever done to anybody?'
Isp ( IpTOP) ,,TOP
DpsuB TOP fT+AgrP .
r_L_, I 1 1• ? tct WH2 T+AgrP'
I 1
WH1 0 cl
WH3•
rah-constituent, providing adjunction is restricted to
nonargument maximal projections (Chomsky, 1986:6).
Without dealing with the wider implication of the
following statement, I propose the T+Agr projection being not
the top functional head of the Czech 'split IP', claiming
however, that the subject, which must raise to T+AgrP to get
a Case, may raise to the top IP specifier as well.
In the sentences containing the indefinite/negative
multiple wh-words, like (7) above, the top IP functional head
may be empty (with the trace of the clitic AUX only), its SPEC
containing the subject. The non-initial .wh-words may then
appear adjoined to some lower maximal projection. A suggested
position of the indefinite wh-words is demonstrated in (18).
T+AgrP
tSUB T +Agr'/
T+Agr VP: .Co l . jeem ja
to-whom2 when 3 ugYala? [=.komu 2 kdy 3 V
what1 an I
done t1t2t1
If the top IP is stated as a kind of modal or negation
functional head, then the indefinite/negative wh-word
following such a head can get its negative interpretation just
by being within the scope of such a head, exactly as demanded
for negative elements in Jackendoff (1972:350), as cited .in
foothote9 above. The interrogative wh-words are correctly
excluded from getting such indefinite/negative features since
they are presented as adjoined to the maximal projectien•of
the modal or negation functional head. 12 ,
In Haverkort (1993) clitic movement is presented as an
instance of head movement which adjoins clitics to the highest
functional head accessible to them. The Movement is related to
the raising movement of the verb -(both finite and infinite),
because it is the verb that must 'void the barriers' for the
subsequent movement of a clitic. The distinct positiOn of
clitics in tai' resulting string can be derived from (or often
serves as evidence of) the final position of the verb • on the
split IP scale.
The position of Vfin the above (18) is separated.from
the clitic AUX be by two maximal projections, which is top
many for Haverkort's conditions of clitic movement, 13 - this . '
distance, however, is common in Czech (and not, only with the,
clitic AUX). Therefore either some more vacuous movement's
would have to apply to void the barriers or preferably,
another mechanism of the clitic movement has to be worked out.
I will not present any here.
For the analysis of the 'StyliStic Inversion.' in
wh-questions mentioned in section 1.2 above, similar split IP:,
could be used. The Stylistic Inversion in Frendiveis'in
Haverkort (1993) interpreted as a position of the subject in
the SPEC of some lower, non-top functional IP (in the
author's concept SPEC(TP) appears to be the candidate). The
53 54
following J19a• shows Haverkort's (1993:21) French
Wh7question with a clitic, (19b) a Czech wh-question with
a reflexive clitic si , and (19c) a . multiple wh-question.
Haverkort claims that in (19a) the verb raises to C allowing
the clitic to cross the IP barrier and adjoin to the highest
•accessible head C. (19a) shows a Subject/AUX Inversion, which
is evident in the Czech examples (19b,c) as well.
(19)(a)(Hav,93:21) Quand l'a-t-i1 lu?when it has he read.
(b) ..Co SI bude Kryftlifek (?bude) pf.dt na vanoce?. what REFL will K.. (?will) wish for Christtas
'•What Will K. want for Christmas?'
(c)` Co SI kdy bude (?kdy) Kry6tnfek (?bude) plat?what REFL when will (?when) K. (?will) wish'What will K. want when?'
No structure presented in ..section 2.2, however, would be
acceptable for (19c), supposing the non-clitic AUX bude 'will
be' raised to C, with C allowed to contain one element only.
If the clitics adjoin to functional heads, the one most
plausible for the reflexive si in (19) : is C, which is divided
from-the subject by both wh-Word and bude 'will be', but it
still precedes the subject.
Recalling the IP structure suggested above,
Subject/AUX,Verb Inversion in (19b,c) may instantiate the case
of AUX becoming (obligatorily?) the IPT°P head, with the
- Subject remaining in SPEc(TI-AgrP). The interrogative multiple
wh-words in the examples (19b,c) can be adjoined to the IP T°P
(which is the position in this work stated as IP-adjunction,
i.e. ..the preferable (4a:IP-adj) variant). The example (19b)
may have a structure presented below as (20). The infinitival
verb following the future-AUX is supposed to stay inside VP,
AUX moves through the TA-Agr head to get the Tcnse+Agr
features, its obligatory raising being a result of an
interrogative Subject/AUX Inversion structure.
(20)(=l9b)'What will K. want when?'
prat ti t2to-wish t 1 t2
The 'split IF is introduced here as a 'multiplied' IP,
with the only apparent purpose to get the positions needed for
the analysis. Similar results might be achieved by multiplying
the CP projection, i.e. by CP adjunction or taking CP for .
recursive. I prefer the split IP mainly,because the strings of
elements usually contain a verb or AUX, and I take IF for
a kind of finite verb projeCtion, while the connection . of CP
and the verb is less obvious to me. There are also properties
of the multiple wh-cluster mentioned below which make the
adjunction to CP or the recursive CP less preferable.
The multiple wh-words appear in embedded clauses as well
as in root . clauses. The embedded clauses are s-selected •
arguments of main clause verbs and an adjunction to
a s-selected argument is prohibited by either Chomsky's
CPr 1 1
SPEC(CP) C'
. C1 : 1
TOP: rhTp n TOP
WHI..0 Cl : IP
. . . I mnn: WH2 sp(IpTOP) ',..........
: : 0 r 1
: : 1T+AgrPAl.1Xi T
: DP Agr'I ' /-1Agri VP<-
Co l si kdy 2 bude Kryatafekwhat, REFL when 2will KryStafek
5556
(1986) restrictions on movement or McCloskey's (1992)
Adjunction Prohibition. To take a multiple wh-element as
a result of a recursive CP projection, on the other hand,
would not explain why the number of wh-elements is not limited
to one i.e. why the pattern is iterative, and why at the same
time all the elements which indicate the presence of the comp
node preferably follow only the first wh-eloment. 14
2.4.3. Conjoined wh-words
A 'possible. counter-argument for the multi-constituent
analysis of the fronted wh-words may be be found in conjoined
structures as presented e.g. in the following (21), where the
clitics AUX isi 'you are' and weak pronoun le 'them' follow
both the wh-words. (21-23) show the Czech paraphrases of the
Serbo-Croatian examples given in Browne (1972)
(21)(S-c:Browne,1972:223) Kdy a kde . JSI JE videllwhen and where AUX them saw-you'Where have you seen them when?'
Browne (1972) claims that the wh-words in English must be
'free' constituents, while in Serbo-Croatian they can be
'bound' parts of the sentence, i.e. each wh-word representing
a constituent of a distinct (elliptically omitted)
interrogative clause. The author cites. Czech as having the
same characteristics as Serbo-Croatian (Browne, 1972:226f). He
suggests 'the difference to be a result of the distinct
'distributional characteristics of the verb' which he
demonstrates with the Serbo-Croatian equivalent of the
following Czech example (22) where the Czech verb need not be
57
followed by its subcategorised argument.
(22) Rozbil to eklo? Ano, rozbil.broke-he the glass? - yes, broke-he'Did he break the glass? - Lit: Yes he broke.'
(22) makes Browne argue that while the ver in English demands
all its selected arguments to be realised in the.S-structure:,:
Serbo-Croatian can have the arguments present in the ciepi
structure only:15
It is not clear to me how the author derives
the , possibility of conjoined wh-words from the aboVe 'stated
characteristics of the verb: he probably assumes that ik the." '
verb can represent the whole sentence' structure without' its
arguments, then an argument can represent the whole structure
without the verb as well. Then the conjoined .wh-words 4o not
represent different kinds of constituents (where conjunction
seems unlikely), but conjunction of the Whole clauses within
one constituent. 16
With no further specification of the kind of conjunction
involved, I accept Browne's analysis, assuming that the
example (21) presents two conjoined .wh-words in SPEC(CP),.
i.e. one constituent only, which is 'the first constituent'
relevant for the 'the second position'; -Of the clitics.
' To see that the conjoined . structure is distinct from tli6
MWhQs which are discussed, in this Chapter, compare . .the
position of the clitics isi and ie in (21) with the following
(23).
(23)(a) Xdy JSI JE kde (*JSI JE) videl?when AUX them where (*AUX them) saw-you'When have you seen them where?'
(b) * Kdy a JSI JE (a) kde videl?when and AUX them (and) when saw-you .
58
Example (23a) presents non-conjoined wh-words, where the
poSition of clitics is obligatory after the first wh-element.
Given the second position' of the clitics for obligatory, the
unacceptable (23b) shows that the presence of the overt
conjunction a 'and' is enough for the one•constituent
behaviour of the conjoined wh-words. The multi-constituent
analysis appropriate for the wh-words in (23a) indicates that
the non-initial. whwords in such structures obligatorily
follow the clitics, which is. the main argument for preferring
the (4a111,77adj) analysis.
2.4.4.- Clitic affix and the particle ge
In the following example (24) two affixes are presented,
which can be added to the wh-word fronted in the wh-question:
an affix or particle ge 'that' in (24a) . and 2nd p/sg/pres AUX
be -s 'are' in :(24b). (The particle and affix are written in
capital letters in the following examples.)
(24) C62E : ti Eligka tikala?whatTHAT to.-you Eligka said'What did Eligka say to you?'
. (25) ' KomuS to dal?to-whom're it gave-you!who did you give it to?'
The added particle ge 'that' makes the interrogative word
'expressive', and such ' expreSsive interrogatives are used .
mainly in colloquial speech.' ( Petr, 1986:95). 17 Because the
-function- of connector is assumed in Czech to be obligatory for
the definition of conjunction, ge 'that' is taken for
a 'particle' also when it appears at the beginning of
independent sentences, where it is interpreted as an an
expressive element. In interrogative sentences the particle
suggests their being a reaction to some previous context, and
it presumes some following response as well. The following
(26) shows an example of a sentence where ge 'that' replaces
in fact the yes/no question structure.
(26) to pfijde pozdeji?that will-come-he later'(Have you said/Does it mean) that he will come later?
Because of its position at the beginning of the sentence
and close relation to the ge 'that' complementiser4 I will
assume that the position of the 'particle' ge 'that' which
appears on the interrogative wh-words is related (in (26)
probably identical) to the original position of the neutral
complementiser ge 'that', i.e. Comp,
In HWhQ the particle ge can be added to any interrogative
word, but its standard, least marked position is always with
the first of them, as presented in the following (27a). Far
worse is the result with more than one particle in one
multiple question as shown in (27b,c) /
(27) 'Who brought what to who?'
(a) Coll; komu kdo ptinesl?what-that to-who who brought?
(b) ?? Co komu(E) kdo(tE) pfinesl?what to-who-(that) who(that) brought
(c) * Co2E komu(2E) kdo(2E)? ptinesl?what-that to-who-(that) who(that) brought
The structure (4a:IP-adj) with the first wh-word fronted
6059
is obviously the most suitable for the position of the
affix-particle, since the complementiser is supposed
immediately to follow the first wh-word.
The question remains, whether the . particle is occupying
the complementiser's position, or is added to the SPEC(CP)
position of the wh-word. The following (26a) shows that if the
Consider also the other affix element co-occurring with
the wh-words, which was introduced in (25). The -s 'are' affix .
is a 2nd person/singular AUX be, a part of analytical past •
tense or passive. This form of AUX be behaves as a clitic in
Czech. The following (29) shows that in a sequence of several
ciitics the AUX be is always the first, preceding .all- the
first wh-word is separated with a parenthetiCal, the particle other clitics (all clitics are in capital letters).
is acceptable only with the interrogative word.
(28)(a) Rdo(2E), ptala ae, (*2E) pkijde ptitTsti taiden?Who(that), asked-she, (*that) comes next week?'Who, asked she, comes next week?'
(28b) on the other hand demonstrates that if the first
wh-word. is fronted (to the matrix clause SPEC(CP)), the
particle is possible even when the embedded clause is
introduced by a 2g 'that' complementiser.
•(28)(b) Coft si myslela, 2E ji MikuIdA donese?what-that thought-she, that to-her M. will-bring'What did she think that Mikulag will bring to her?'
Taking the particle for a complementiser, we would have .
to assume . for (28b) the structure where both lower SPEC(CP)
and C were fronted, and then the lower C filled again with
another neutral 2e 'that', which would be rather
unconventional. I present examples (28) as arguments for the
position of ffe 'that' particle in SPEC(CP) position. The affix
form on the wh-word may be a result of the prohibition on two
phonetically realised items inside the CP projection, which
otherwise. is observed in Czech, and which is claimed by Rudin
(1988:494) as a characteristic of 'non-Multiple- -Wh-Fronting
Languages', i.e. Czech.
(29) Vdera JSEM MU TO Oval do vlastnich rukou.yesterday am him it gave to his-own .hands'I gave it yesterday to him, into his own hands`.'
The following example (30) presents the past AUX be
folloWing a complementiser of the embedded clause, which Shows-
that the AUX follows the complementiser. .
(30) Tvrdila, to JSME MU TO nemeli daVat.claimed-she, that are-we him it not-haVe to give'She claimed we should not have given it to him.'
Tvrdila, JSME ( e) MU TO...claimed-she are-we (that) him it
In (31) the past AUX be is following the wh-Vord, showing
that both the full form of AUXand affie AUX precede the other
clitic to 'it'
(31) Romu JSME / Xomu$ TO potom m61(i) dat?to-who are-we / to-who're it then should give'Who were we (you) to give it to, then?'
* Romu TO JSME / * Tomu TOS potom 44.t?to-who it ARE-we / to-who it're it then should give'
(29),(30),and (31) indicate at the same time the position of
the clitics as a right-adjunction to C (the adjuncts May be
'topicalised' in SPEC(CP) in (29) and (30)).18
* Vdera MU JSEM TO daval / * Vdera MU TO JSEM davalyesterday him am it gave / Yesterday him it am gave
61
(36a) shows one (either subject or object) wh-word fronted,
and the sentence is more acceptable than (36b), which fronts
both of them. (36c) on the other side supports. the
affix, which is not the case in (34). The acceptable
of (33) and (34) are offered in (35), with both 2e
'that' and -s 'are' connected.
63
Consider then the position
Presented in (32).
-(32) 'What did you give to who?' ..
(a)' 1omuS co dal? CoS komu dal?to-who're what gave-you? what're to-who gave-you
(b) ?? Koisu COS dal? 7? Co komuS dal?to-who what're gave-you what to-who're gave-you
`Even : if the unacceptability of the ?? examples in (32b) is not
fully comparable with the ungrammatical ones in (31), the
''are' is, certainly preferred right after the first
wh-word. In line with the assumed position of is 'that'
partible, I .suggest the -s ' are' affix to be •a part of the
SPEC(CP) as well. .
There are, however, MWhQ's where both the affixes appear.
In the following (33) three wh-words are presented, and in
(34) the clitic to it is inserted. Both (33) and (34) show
the -s 'are° affix bound to the position assumed for C.
(33) ?? Kdy2E komu coS dal?when-that to-who what're gave-you'When did you give what to who?'
(34) - * Rdy2E TO komuS ( TO) dal?what-that it to-who're (it) gave-you'When did you give it to who?'
.(34) May be presented as an argument of the verbal character
Of the AUX,!since the clitiC adjoined to C may require its•
presence in C. However, the same sentence without any AUX will
be correct, so I prefer the explanation requiring the AUX
glitic.,to obey the clitic movement before it becomes realised
(35)(a) Kdy (*TO) 2ES - TO komu (*TO) dal?what (*it) that're it to-who (*it) gave-you'When did you give it to who?'
(b) 7 Kdy komu 2ES TO dal?what to-who that're it gave-you
.'When did you give it to who?'
Recalling that the spelling is not normative in these
colloquial forms, (35a) may be considered as a variant to
(34), with both the ffe 'that' particle and the -s 'are' affix
added to the first wh-word, and (35b) as equiValent to (32b)
or (27b,c).
No matter how the relation of both the particle and AUX
affix to the SPEC(CR) position can be stated, the examples
(34) and .( 35) strongly support the multi-constituent analysis
of the wh-words, since they require the C being in between the
first and the other wh-words. At the same time the examples
argue against a repeatedly recursive CP projection predicting
sentences like J32b) or (27b,c) are acceptable. 19
2.5. Wh extraction from the multiple wh-question
In the above sections a number of arguments were
presented showing the multi-constituent character of the
multiple wh-words. Here some more are going to be added.
If the wh-words in Czech do not form a single
constituent; it could also be expected that if 'one of them are
extracted, then the others may remain in the embedded clause.
The following (36) indicates that this seems to be correct.
affix in a MWhQ as
64
ktOryi kogo 2ktell kohowho whommeet the man
tl zabil t2zabilkilled
who killed t.?'
one-constituent analysis of conjoined wh-words. clause. Notice the position of the neutral 2e 'that°
Kdo /komu si myslia, to ( mu) bude pomahat?who
s /to-who
o'Who/whom do think-you that (t )(him)will help (to )
you think (he) will help (him)?'
(b) ? Edo komu si mysliA, 2e bude pomahat?whos
. to-whoo think-you that (ts ) will help (to )Lit:'Who whom do you think will help?'
(c)
Edo a komu si mysli4, to (*a komu) bude ponAhat?who,Who
and to-Who, think-you that (*and totwhoo ) will help (to ).do you think will help, and whom?'
In Lasnik and Saito (1984) the IP-analysis of the
wh-words in Polish is based on examples of wh-words following
a complementiser (37a) or a relative pronoun (37b).
(37)(a)(Pol:L&S,1984:11)
Maria mysli, [ s pte co Janek kupil]* Marie si mysli, to co si janek koupilMaria thinks that what Janek bought'What does Maria think that Janek bought?'
( h)(Pol:1A5,1984:75)
Polish: SpOtkales metdzyzne,Czech: * Potkal jai mute,
you met the manLit.: 'whoni did you
As stated in Chapter 1, the Czech equivalents of the
Polish examples in (37) are ungrammatical, because in Czech
either the matrix clause or the embedded clausal "compleMent
must be overtly specified for the interrogative [+wh] feature,
if the sentence is to be interpreted as a question.
The post-complementiser position of the non-initial
wh-word can, however, be observed in Czech in a non-fuliy
standard long distance movement from a multiple question. The
following example (38) presents a wh-extraction from a MWhQ,
where the first wh-word is removed to the matrix clause
SPEC(CP) and the non-initial wh-word remains in the embedded
complementiser, which may be optionally inserted at. the
beginning of the embedded multiple question.
(38)(a) Kdo's fikal, (te) kdy (*te) pozve Mart/Skil?Who're-you said, (that) when (*that) invites Mary?Lit: 'Who did you say (that) when invites Mary?'
In (38a) the only acceptable position of the optional 2e .
'that' complementiser is preceding the other , wh-word." (38b)
shows my suggested structure of (38a).
(K3d:b)'s fikal, [CP t' to [ IP,:kdy [ IP ts pzv Marugku tA ?:Who,Pre-you say, [ CP t'$
s
that[ I' whe
A
nA ( IP ts invites-Mary:tA .?Who do you say invites Mary when?'
In*Rudin (1988) the languages that do not allow
adjunction to SPEC(CP) at S-Structure, i.e. Czech, respect the
Wh Island Constraint, since the lower SPEC(CP) is filled and'
indexed with the first wh-word.. The (38b) analysis assumes-the'
non-initial wh-word in an IP-adjoined'position and'•eavee the
CP projection empty if no overt neutral Complementiser as
preSent; this may be used to explain an apparent violation of
the Wh Island ConStraint in Czech.
As shown in Chapter 1 example (29), the presence of an.
overt [+wh] complementiser makes any/ extraction imposeible.
The following (39a) shows an acceptable embedded. MWhQ, and
(39b) presents the unacceptable extraction of any of the
wh-words from such a clause.
(39)(a) Ptam se, jestli kdy (*jestli) .komu co dal.-
ask-I whether whsh (*whether) to-whom what gave7ho.'I ask, whether he gave what to whom when.°
( b)
* Rdy/komu/co se ptdm, jestli (kdy/komu/co) dal?when/to-whom/what ask-I whether (when/te-whom/what) gave-he.'When/whom/what do I ask, whether he gave (whom/what/when)..1.
(36)(a)
Polish:Czech:
65
The subcategorisational frame of the matrix clause verb
in (39) requires a [+wh] complement. If this [-1-wh] complement
is . [+wh] complementiser, it 'Must precede (as shown in
(.39a)) all the other wh-elements, and blocks the extraction of
any of.them. The analysis assuming that the [+wh] features of
the [+wh] •complementiser occupy the lower SPEC(CP), predicts
such results.
The.structure presented for the non-initial wh-words in
(3a:SP(C)-adj) would explain the obligatory first position of
the [-i-wh] .complementiser only if it were placed in the Spec of.
_soma recUrsive!CP, and even then it would be necessary to
state some rule which Would make the complementiser precede
all the .other wh-words adjoined to or in (some lower?)
SPEc(CP). consider also the position of clitics preceding the
other wh-words, as presented in the following (40), which
'again disfavours the (3a:SP(C)-adj) analysis
(40k Ptal se, jestli. MU TO kdy (*mU to) Eya slibila.asked he whether him it where (*him it) EVa promised'He asked whether Eva had promised it to hiM when.'
The structure (4a:IP7,.adj), with p..[+wh] compleMentiser as
the fitst wh-element,' predictS the obligatory precedence of
the•[+wh] complementiser. The clitics can be then adjoined to
C in (40), preceding the non-initial wh-wordS, which are
argued here to be adjoined to IP.
2.6. Summary of Chapter 2
In section 2.2 several papers discussing multiple
whqueStiens in Czech or other Slavic languages were reviewed,
with the analyses of the multiple wh-elements claimed by their
authors. Following Rudin (1988) three structures for the
position of the wh-words were proposed: first, all the
wh-words adjoined to SPEC(CP) (see the example (3'a) in 2.4.1
above); second, only the first wh-word in SPEC(CP) and the
others adjoined to SPEC(IP) (see the (41) below); and third,
the structure similar to the English multiple queStion, i.e.
With the first wh-word in SPEC(CP) and the others 'in situ'
(see the example (4b') in 2.4.1 above).
In section 2.3 the Wh Movement of all the interrogative
wh-words was demonstrated as obligatory and resulting in
non-marked'` word-order, which is the main argument excluding
the (4b:in-situ). structure. Suggesting that the wh-elements
are inherently 'marked', the free constituent-order derived
from the focus-topic sequence was stated as the motivating
factor for the S-structure position of the wh-elements,
together with the impossibility of placing more than one
element in the SPEC(CP) position.
The arguments presented in section 2.3 can be summarised
as follows:
(i) The possibility of inserting a parenthetical clause
supports taking the multiple wh-words as a multi-constituent
structure;
(ii) the interrogative (but not the 'indefinite/negative')
wh-words in Czech obligatorily precede the Overt subject, and
all non-echo wh-elements precede the finite main verb; and
(iii) the ordering of the wh-elements is not strictly
definable but certain sequences seem to be preferred.
In 2.4 the positions of clitics, the particle ge 'that'
6867
CPt
SPEC(CP) c'
IP
Assuming,
wh-elements int o
L&S conception,
hoWever, the LF movement of all the
the appropriate SPEC(CP), which can, in the
have only one index, the arbitrary Wh word
(41).
-p and the affix -s 'are' supported the multi-constituent
structure of the wh-words with the first of them being
separable from the others, and with the Comp node being in
between: The results were supported by the examples of the
long-distance movement of the MWhQ presented in section 2.5.
The arguments in the above sections indicate the proposed
(4a) structure as the most plausible structure of the Czech
MWhQ with multiple wh-words which are all interpreted as
interrogative. The (4a) scheme is repeated below as (41).
overt complementiser and Wh-word inside the CP projection,
as well as the Oh Island Constraint which disallows extraction
of the wh-word from a CP projection containing the (-teah)
element. To decide whether the Comp Adjunction Constraint
stated in Rudin .(1988:490) really applies Czech at .both.
S-structure and PF, the possibility ,of topicalised elements in
SPEC(CP) would have to be- investigated in more detail. As for
the wh-words, they are analysed here as not adjoiried to ,
SPEC(CP) before LF:
WH1
IP
WH2 WH 3IP
ti t2 -t3
The 'multiple' wh-question with only one interrogative
element and the non-initial wh-words interpreted as
indefinite/negative prpnouns have been argued to be distinct
from the MWhQ. Some suggestions about the position of
indefinite/negative wh-words can be found in section 2.4.2 but
without a . proPerly detailed analysis.
In 2.4.3 conjoined wh-words (with the obligatory
conjunction and) were mentioned, which represent one conjoined
constituent in SPEC(CP), i.e. are assumed to appear in the
position of the first interrogative element.
In the above sections also the following properties of
the Vh. Movement were illustrated: Czech obeys the
Doubly-Filled Comp Filter prohibiting the co-occurrence of an
sequence in the MWhQ's still presents a problem for the - proper
governMent of wh-traces in Czech: If both interrogative and
indefinite/negative wh-words appear in A-bar:positiona, as is
claimed, at least for the tormer, in the above Chapter, the
structure requires both external and internal arguments as
well as adjuncts to be lexically governed in - Czech:
AlternativelY antecedent government must becothe acceptable 'fOr
subject and adjuncts at PF or S-structure, which is basically
the solution used in Rudin (1988) . where the traces are
lexically governed at. PF and A' bound within .their domain at
LF. It is also possible to introduce some other concept of
a governmentrelation between the verb and its complements' for
the free-constituent languages (see e.g: Stowell, .1981:124),
which may change the above analysis since the (4b:in-situ)
structure has been so far excluded mainly because of the
'marked' vs. 'unmarked' distinction, which is. not a reliable
base for any evaluation.
7069
1 0
NOTES TO•CHAPTER 2
Similar examples are presented e.g. for Polish andRusSian in 1 Wachowicz (1974); for Polish, Czech, andRussian in Toman (1982), and for Bulgarian,Serbo-Croatian, Polish, Czech, and Rumanian in Rudin(1988).. •
For the formulation of the Empty Category Principle,which requires empty 'category (here wh-trace) to beProperly governed, Rudin uses the notions of 'lexical
, gaVernment', i.e. head government at PF,' anda ''binding condition' requiring an A' anaphor to be
_ A'-bound in its Domain at LF. The governing relationsare assumed as m-commanding relations; and the Domainis'the ' first claUse (IP or CP) or NP which containsan accessible' c--commanding SUBJECT (R,1988:477).
Moreover RUdin's evaluation of the Czech examples isnot always reliable. E.g. her example (19) showinga 'fully grammatical' long-distance Wh Movement inCzech is probably the only example which may passeven the standard grammar requirements. On the otherhand, Rudin's example (27a) showing the impossibilityOf Wh island violation in Czech would be excluded bythe fact that.the fronted subject wh-word would berelated to the dropped subject of the matrix clause.(See footnote' 0 in the previous Chapter.) The latterexample is, however, hard to evaluate properly, sincethe Czech sentence is not equivalent to . the Englishtranslation offered in Rudin's text.
4 The standard unmarked position of verbal argumentsand sentential adjuncts in Czech can be stated asfollowing the verb, i.e. on the left Side of VP, buttheir equally common, marked position is precedingthe verb.! The multiple question Wh Movement may,then, be only some stylistic reordering which neednot be reflected in syntax at all. If this were true,then there would be little reason to deal withMWhQ's in the way presented in this work.
An apparently similar concept is presented also inMilner and Sternefpld (1993), who introduce TP witha head T (=topic). But the aUthors require thetopicalised elements to be [-wh], therefore their TPis net a suitable landing site for the non-initial•wh-elements, even if topicalisation is dealt with asa phenomenon related to Wh Movement.
6 'According to Rudin (1988:485) such a two-constituent
structure with the first wh-word in SPEC(CP) and allthe others adjoined to Comp, is claimed for Polish inCichocki :(1983). Cichocki's analysis, however, putsthe other;wh-words, on assumed position of the Czechclitics, ! which would contradict the obligatoryordering of the clitics and the optional Ordering ofthe following wh-words, as presented in the next'section, As . well as the distinct character of cliticsand wh-wOrds.
Rudin mentions VP7adjunction as an alternative toIP-adjunction, trying to keep in line with Chomsky(1986), who 'disallows adjunction of Wh-phrases toIP, for reasons internal to his Barriers treatment ofsubjacency and related phenomena.' ( RUdin, 1988:489).
In Kuno and Robinson (1972) the distinctiOn betweenthe minimal apprepriate answer is used asa diagnostic for a unique vs. multiple interrogativeelement in a given sentence.
According to Jackendoff (1972) 'negation and wh aregoverned., by essentially the same semantic rules'(Jackendoff, .1972:350) which for the author meansthey are both subject to his 'Modal projection rule' ,(p.348) defining the scope of the element by the samemodal operator. Whichever operator it may be, the LFposition of all wh-elements would then be equal.However there is no reason to assume the sameposition of both interrogative andarbitrary-negative wh's at the S-structure as well.Comparably, when Jackendoff states that ' Negationmust be'determined from the surface configuration'(Jackendoff, 1972:348) he mentions any (which seemsto be a kind of equivalent to the arbitrary-negativewh's in , Czech) as an lexical item of some/anyalteratien and only structurally dependent on neg. Itappears within its scope, but the same position ofthe neg and any is in fact often excluded by hisright-to-left condition, at least in English.
The Polish postverbal position of the clitics.mightbe explained by the fact that the pronoun is'stressed', as Wachowicz states (W,1974:294): Somepronouns in Czech have the weak and strong forms notunambiguously distinguishable. (E.g. 7//1/ 'her'(ACC/DAT) are the only available forms for the 3 pspronoun ona 'she'.) Such pronouns could appearfollowing a verb as well.
11 In Rivero (1991) a split IP for Czech is presented,containing TP(+AgrP?) AspP(Aux?) - NegP - VP. Theauthor derives the IP structure from the obligatoryordering fnegV INF - Auxfin) in null-subject
naffirmative root clapses. Rivero assumes that V'incorporates with Neg' and t4 whole cluster moves.(Long Head Movement) into C', skipping the Aux
7271
generated'in AspP and risen into T° . Rivero proposesthis concept contrary to the ' traditional Slavicapproaches... focusing on the Aux as a clitic' ( R,1991:323). However, the LHM fails to apply 'if aitem is generated in or moved to a slot precedingwithin CP (i.e. specCPP i.e. 'Wh-phrases,
• Focus-phrases and pre-Aux subjects inhibit LHM (R,1991:325f), and therefore Rivero's concept is notdirectly relevant to the topic of this work. I willpresent the Aux 'traditionally' as a clitic (sinceclitics are in this work shown to adjoin C, it makesno difference), with no stipulations about theposition of the verb in null-subject affirmativeclauses with V - Aux ordering.
12 For Slavic languages usually some kind of ModP isargued as a top IP (in e.g. Dimitrova-Vulchanova(1992). I would suggest such a ModP comprising somenegative features as well because negation is usuallyeneugh, to replace the whole clausal structure inquestion tags and similar constructions in Czech,because in negative clauses both subject and verb arenegated, i.e. they may be suggested to be bothdominated by a potential NegP, and also because ofthe indefinite/negative interpretation of thefollowing wh-words in multiple wh-questions, asmentioned above. Any precise definition would,however, demand more detailed study than can beoffered he and now. Therefore I am going to use themarking IP withoUt any specification. For ARGP andTP (as offered in Rivera (1991),Dimitrovova-Vulchanova (1992), and as follows fromrequirements on null-subject languages presented inJaeggli and Safir (1989:33), a one-noderepresentation is used here.
13 In .case of one functional head barrier, i.e. whene.g an AUX verb does not raise to C and AgrP remainsopaque, Haverkort (1993:39-47) explains the movementof a clitic out from AgrP as a result of a vacuousmovement of the whole AgrP into the lower transparentSpecTP position. Using the concept of specifier-headagreement presented in Barriers, the vacuous movementrenders AgrP transparent.
14 ' Contrary to the multiple wh-words, thecharacteristics which indicate the CP recursion or CPadjunction may be found with many focused elements,as presented e.g. for Welsh fronted constructions inTallerman (1993).
15 Recall also a similar suggestion in Chapter 1, wherethe Czech verbs were claimed to have 'vague'subcategorisational frames when compared withEnglish; such an ambiguity has been .stated asa possible explanation for their limited use in the'bridge structures'.
The conjoined Wh-words are supposed to representa kind of sentential ellipsis also according to Petr(1987:417), where the conjunction is explained asmotivated by a similar form and the same subordinate ,
relation to the verb. Similar ellipsis does not seemto be excluded either for English, at least for_ theconjoined relative PP wh-proferms, as presented ine.g. Gazdar, Pullum, Sag, Wasow ,(1983). Moreover' asfar as•I am aware, adjunct interrogative , words can be'conjoined in English As well. (e.g. °Wheh and wheredid you buy it?'), while conjoined argument Wh-wordswould also he much less acceptable in Czech.
The 'component' 2e 'that' may be written togetherwith the interrogative word; or separately. The - same .component 2e 'that' also forms a part of manycompound conjunctions, often already inseparable.
In Rivero's (1991) concept taking the AUX bV't. 'be':for a non-clitic element, (30) may present theclitics. mu 'him' and to 'it' adjoined to someINFL filctional head, which may be evidenced by - -theposition of the subject in the paraphrased (30):below,
'She claimed we should not have given it:to
Tvrdila, ie my JSME (*my). MU TO (?my)claimed-she, that we are-me (*we) him it (?we)
nemeli davat.not-have to give
and also by the unacceptable (33): and (34). For theordering of the wh-words,, clitics, and the AUX,:thisdifference is not crucial, and I leaVe the status ofthe AUX open.
However, both the particle 2e 'that' and the affix-s 'are' are not fully. standard .,. Neither. are themultiple wh-questions with :More than twointerrogative words common or especially salient. 'Tocombine all these factors and then . -- evaluate theacceptability of the results, is often controversial.Therefore this section could not stand as anindependent argument and its being a part of somebroader context seems inevitable. And the, fullyapplies for the following section, because also nolong distance Wh movement is standard in Czech.
20 The presence of the Le 'that' complementiserintroducing the embedded clause in the sentences like(381 substantially supports the analysis of the firstwh-word in the matrix clause SPEC(CP) as moved fropthe embedded clause, which is relevant.espeolallywith the adjunct wh-words (see also foothote inChapter 1)..
16
17
18
19
7374 .
BroWne .„ E.W.III (1972) 'Conjoined Questiona Limitation on English Surface Structures'Inquiry 3, 223-226.
Chomsky, N, _(1977) 'On WH Movement', in Culicover, P.W.,Wasow, • T., and Akmajian, 'eds., Formal Syntax.`Academic Press, 71-132.
chomskyi . N. (1981) Lectures o Government and Bindin Forispublications, Dordrecht.
Chomsky, N. (1986) Barriers, MIT Press.
Dimitrova-Vulchanova, M. (1992) .' Clitics in'EUROTYP (8) Working Papers, Volume 4, 1-50.
Words andLinguistic
SlaVic't in
Bibliography Emonds, T.E.Syntax,
(1976) A Transformational Approach to EnglishAcademic Press, Inc.
Anderson, S.R. (1993) 'Wackernagel's Revenge: Clitics,Morphology, and the Syntax of Second Po6ition', Language 69, 68-98.
AnYadi, S. and Tamrazian, A. (1993) 'Wh Movement in Armenianand Ruhr German!, in UCL Working Papers in Linguistics 5, Dpt. of Phonetics and Linguistics, University CollegeLOndon.
Bach, E. (1971) 'Questions', Linguistic Inquiry 2, 153 - 166.
Baker, C.L. (1970) 'Notes on the Description of EnglishQuestions .; The ROle of an Abstract Question Morpheme',Foundation of Language 6,197-219.
. _Banfield, A. (19.73) 'Narrative Style and the Grammar of Direct
and. Indirect Speech', Foundations of Language 10, 1-39.
• Bolinger, D.L: (1978) 'Asking More Than One Thing at a Time' ,
in Hiz, H., ed., Questions. Reidel, Dordrecht, 107-150.
Borer, H. (1985) 'Anaphoric AGR', in Jaeggli, O., and SafirK.a., eds., The Null Sublect Parameter, Kluwer Academic'Publishers 1989, 69-109.
'Bouchard,' D. (1984) On the Content of Empty Categories', Foris:Publications, Studies'in Generative Grammar, Dordrecht.
Bresnan, J.W. (1970) 'On Complementizers: Toward a Syntactic .
- Theory of Complement Types', Foundations of Language 6, 297-321.
Brody, M. (1991) 'Economy, Earliness and LF-based Syntax', inUCL Working Papers in Linguistics 3, Dpt. of Phonetics
. and Linguistics, University College London, 25-32.
Emonds, J.E . (1978) 'The Verbal Complex V'-V. in French',Linguis tic Inquiry 9, (151-175.
Emonds, J.E. (1985) A Unified Theory of Syntactic Categories,Faris, Dordrecht.
Firbas, J. (1992) Functional Sentence Perspective in Written and Spoken Communication, Cambridge University Press.
Gazdar, G., Pullum, G.K., Sag, I.A., Wasow, T. (1483)'Coordination and Transformational Grammar', LinguisticInquiry 14, 663-676.
Green, G. M. (1989) Pragmatics and Natural LanguageUnderstanding, Lawrence Erlbaum Associates; Inc.
Grimshaw, T.(1979) 'Complement Selection and the Lexibon',Linguistic Inquiry 10, 279-326.
Haegeman, L..(1991) Government & Binding Theory, Blackwell.
Haverkort, M. (1993) Clitics and Parametrization. Case Studies in the Interaction of Head Movement. Phenomena, EUROTYP(8) Working Paper VIII, 2.
Jackendoff, R. (1972) Semantic Interpretation in GenerativeGrammar, MIT. Press.
Kayne, R.S. (1980) 'Extensions of Binding and Case-Marking',Linguistic Inquiry 11, 74-96.
Kuno, S. and Robinson, 3.3. (1972) 'Multiple Wh Questions',Linguistic Inquiry 3, 463-487.
Lasnik, H. & Saito, M. (1984) '01 -1 the Nature of ProperGovernment', Linguistic Inquiry 15, (235-289.
Lenerz, J. (1985) 'Diachronic Syntax: Verb . Position and COMPin German', in Taman, J., ed., Studies in German Grammar,Foris Publications, Dordrecht, 103-133.
Muller, G. and Sternefeld, W. (1993) 'Improper movement andunambiguous binding', Linguistic Inquiry 24, 461-507.
Petr, J. and col. (1986/1987) Mluvnice destiny II, III (CzechGrammar), Academia, Praha.
Pollock, 3.-Y. (1989) 'Verb Movement, Universal Grammar, andthe Structure of IP', Linguistic Inquiry 20, 365-424.
Riemsdijk & Williams (1986) Introduction to the Theory ofGrammar, MIT Press.
7576
Rivero, M.L. (1991) 'Long Head . Movement and Negation:Serbo-Croatian vs. Slovak and Czech', The Linguistic Review 8, 319-351.
Radford, A. (1981) Transformational Syntax, CUP 1985.
Radford, A. (1988) Transformational Grammar, CUP.
Reinhart, T. (1981) 'A Second COMP Position', in Belletti,Brandi, and Rizzi, eds., Theory ot Markedness inGenerative Grammar (GLOW 1979), Scuola Normale di Pisa:Pisa, 517-558.
Rizzi, L. (1978) 'Nominative Marking in Italian Infinitivesand the Nominative Island Constraint', in Heny, F., ed.,Binding and Filtering, Croon Helm Ltd, London, 1981,129-157.
Rizzi, L. (1991) 'Residual Verb Second and the Wh-Criterion',Technical Reports in Formal and Computational Linguistics 2, Faculte des Lettres, Universite de Geneve.
Rudin, C. (1988) 'On Multiple Questions and Multiple WhFronting', Natural Language and Linguistic Theory 6,445-;501.
Stowell, T. (1981) Origins of Phrase Structure, unpublishedMIT dissertation.
Tallerman, M. (1993) Fronting Constructions in Welsh; paper- presented in LAGB Autumn Meeting, Bangor, Sept. 1993.
Toman, J. (1982) 'Aspects of Multiple wh-movement in Polishand Czech', in May, R. and Koster, J., eds., Levels of Syntactic Representation, Foris, Dordrecht, 293-302.
Toman, J. (1991) 'Anaphors in binary trees: an analysis ofCzech reflexives', in Koster, J. and Reuland, E., eds.,Long-distance anaphora, CUP.
Toman, J. (1992) 'A Note on Clitics and Prosody', EUROTYP (8) working Papers, Volume 4, 113-118.
Wachowicz, K.A. (1974) 'Against the Universality of a SingleWh-Question Movement', Foundations of Language 11,155-166.
Wachowicz, K.A. (1978): 'Q-Morpheme HyPotheSis'i in Hiz,ed., Questions, Reidel, DordreCht, 151-164.
77