Do Croatian Clitics Have to Be Verb-Adjacent?

Andrea Stiasny, University of Michigan

Recent research on adult cliticization in Croatian suggests a shift towards the hypothesis that there is a very tight relation between clitic and verb. Boškovic (2001), Franks (1999) and Progovac (Forthcoming) propose (using different mechanisms) clitic adjunction to a verb (or to a category carrying verbal features, possibly with multiple copies). I test the plausibility of this hypothesis to explain child acquisition data. Children’s early grammar may show more limitations or less variation than we find in adult grammars, and constitutes an important testing ground for grammatical theory.

Croatian clitics are not always adjacent to overt verbal forms, different from Spanish clitics, which do not allow anything in between a clitic (proclitic or enclitic) and a verb. In Croatian, appearing in the second position of an intonational phrase is the only phonological requirement a clitic has; otherwise, word order can vary widely.

Two questions arise then: (i) whether the rate of adjacency between clitics and verbal elements constitutes enough evidence for the verb-adjacency hypothesis; (ii) what explains the cases in which clitic-verb adjacency does not obtain. We hope child data will shed the light on different analyses of cliticization in Croatian.

We analyze clitic production of three Croatian children aged 1;10 to 2;8 (28807 utterances from CHILDES, with 1242 clitics). Children from the age of 1;10 show almost adult-like production of clitics, with variation in clitic structures similar to what is found in adult language.

Children data show clitics overtly adjoined to verbs in 64%. Adding other elements that would count as verbs in languages that have been analyzed as verb-adjacent languages (auxiliaries) the rate increases to 86%. This is significant, as only 14% of cases would need to be explained otherwise. These correspond to 138 utterances: 47% do not have a verb at all, but most of them are not finished utterances; the remainder ha a verb in the fourth or fifth position. Constituents intervening between a clitic and the overt verb/auxiliary include subjects (60%) and adverbs (23%). However, objects, eventive to, negation and the particle evo are also found. The proposals I review claim that even in these cases the clitic is adjacent to copy of the verb that ends up not being pronounced, forced by other mechanisms.

I argue that this preference for clitic adjacency to a verb/auxiliary is due to precise mechanisms determining the syntactic structure affecting both elements, and not accidental to clitic structures. The child data provide strong support for the theoretical hypothesis that the clitics are attached to verbs across different positions. In fact, it supports the view that the clitics are overtly adjacent to verbal elements in child language, provided less common data such as XP+cl+XP+V constructions can be appropriately accounted for in a consistent way.


Boškovic, Željko. 2001. On the nature of the syntax-phonology interface. Amsterdam:
Franks, Steven. 1999. The pronunciation of verb and clitic. Ms. Princeton University.
Progovac, Ljiljana. Forthcoming. Syntax of Serbian: Slavica Publishers.