The Placement of Clitics in Serbo-Croatian

Introduction

This paper is supposed to be a short overview of the basic evidence on clitics in

Serbo-Croatian. Serbo-Croatian is a language with a virtualy free word order due

to ist rich morphological heritage in form of inflections for case, gender and tense

marking. In this paper, I am basically concerned with one exception to this general

rule- the position of clitics. Serbo-Croat (nowdays formally divided into three

standard languages Serbian/Croatian/Bosnian) has a rich system of clitic forms,

including Dative and Accusative pronominal clitics; verbal clitics; which are

unstressed forms of finite auxiliary verbs; and the interrogative marker “li“. On

the course of this paper I will confront some competing wievs on the positon of

clitics in Serbo-Croatian. Since a real study of syntatic phenomena takes a large

amount of material and a great deal of in-depth syntatic knowledge, I will limit

my efforts to a mere stocktaking of some of the existing wievs- those of native

speaker linguists- Progovac, Cavar/Wilder and Boskovic.

1. Progovac

The underlying word order in Serbian/Croatian is essentially the same as in English, SVO. However, Serbian/Croatian is a so-called free word order language in that it allows almost any permutation of constituents 1

(1)Stefan crta An-u

Stefan draws Ana-ACC

For emphasis, however, constituents can be freely scrambled, and appear in any order:

(2) Stefan Anu crta

(3) Crta Stefan Anu (4) Crta Anu Stefan (5) Anu Stefan crta (6) Anu crta Stefan ...etc.

`Stefan is drawing Ana`

2

Word order freedom is possible because of the rich inflectional morphology in Serbo- Croatian.

Like in German, accusative and dative NPs are marked with dinstinct Case endings, so that a fixed word order is not necessary to determine the grammatical function of an NP. Clitics in Serbian/Croatian are the only NPs that show fixed word order (Browne (1974) Although direct (DO) and indirect object (IO) can freely permute if full NPs, the DO must follow the IO if they are cliticized 2 :

(7) Ana joj ga- ACC daje.

Ana her-Dat it-ACC gives (8)*Ana ga joj daje (9) Ana mu ga daje (10)*Ana ga mu daje

Traditionally, clitics in Serbian/Croatian have been claimed to have two possible landing sites: either the second constituent position, or the second word position. Hence, clitics should appear either after the first constituent, or after the first word (Browne (1974), Comrie (1981)) (11) Anina drugarica mu nudi cokoladu.

Ana´s girlfirend him offers chocolate Ana´s girlfriend is offering him chocolate.

(12) Anina mu drugarica nudi cokoladu.

Ana´s him girlfriend offers chocolate

Progovac assumes that clitics in Serbian/Croatian adjoin to the head of CP, Comp, given that clitics always immediately follow a complementizer in subordinate clauses, and they can stack up

(13) Stefan tvrdi [da mu ga je Petar poklonio]

Stefan claims that him it is Peter given „Stefan claims that Peter has given it to him as a present“ She argues that in a matrix clause, where there is no complementizer, clitics in Comp cannot remain clause-initial, since they are phonologically dependent, and need a host. One posibillity for clitics to satisfy this requierment, as Progovac suggests, is to lower down and attach to the right side boundary of the first constituent or word. Another alternative suggested by Progovac, is that clitics stay in Comp (or wherever their landing

3

site turns out to be), and that support for clitics should be achieved by moving constituents to the Spec of CP. Thus in (11), Anina drugarica is the Spec of CP, and the clitic is adjoined to Comp. The phenomenon of clitic placement is claimed to be reminiscent of verb-second in Germanic, where a verb in Comp is always preceded by the constituent in the Spec of CP.

As shown in Progovac (1994), a wh-phrase in the Spec of CP can support clitics, which is also reminiscent of verb-second:

(14) Koga je Milan zbunio whom is Milan confused Who did Milan confuse?

It has often been claimed that word-second clitic placement in Serbo-Croatian violates the most basic principle of UG-constituent dependency , see Comrie (1981) 3 . This is, as Progovac argues, because in examples like (12) the clitic is placed after the first word, which does not look like a constituent. This is, she claims, only at first sight. Both possessive and other adjectives behave as constituents in Serbian/Croatian in that they can be displaced from the head noun, and wh-questioned 4 ; (15) Anina dolazi sestra Ana´s comes sister

(16) Cija dolazi sestra?

whose comes sister Whose sister is coming?

Hence, Progovac assumes that in (12) “Anina” is a constituent and that it moves to the Spec of CP to support the clitic in Comp; “thus the difference between (11) and (12) reduces to the difference between preposing a full NP versus proposing the possessive specifier of the NP” This way she tries to reject the wide-spread assumption that clitics in Serbo-Croat either follow the first constituent or the first word. Though, a little correction should be made, since the examples in (15) and (16) sound rather peculiar and strongly emphasized. The neutral form would be:

(15)a Anina sestra dolazi.

Ana´s sister comes (16)a Cija sestra dolazi?

Progovac also gives examples that clitics cannot be hosted by a preposition: (17)Na sto ga ostavi.

on table(loc) it leave “Leave it on the table.” (18) *Na ga sto ostavi.

4