Acquisition of Ditransitive Structures in Croatian Child Language: A Corpus Study


Although Croatian is a free word order language, the freedom of word order is influenced by pragmatic factors such as animacy (the referent of the object is either Human or Animate), givenness (the object was mentioned in the discourse 5 lines above the target utterance), and object form (NP, PRonoun, and CLitic) (de Marneffe et al. 2012). In this talk, I explore how those factors influence word order in children’s productions of ditransitive sentences.

The data used in this study comes from the Croatian child language corpus available in CHILDES; it was inserted in a Double Object DataBase (DODB), where all the relevant occurrences were classified according to their object order (IO-DO or DO-IO) and form, and all the objects were marked for the relevant properties they possessed, such as animacy and givenness. This allows us to see the dynamics of the gradual production of these structures, and to identify which object properties guide the relative choice of object order. A preliminary study was also conducted for Child Directed Speech in the same corpus, in order to determine an adult baseline and compare the children to the adult data.

Table 1 shows the distribution of object order throughout the child data. Half of the occurrences involve omissions, while, in the non-omission group, the IO-DO order is strongly predominant. This was also the most frequent order in the adult data.

The properties of these objects are provided in Tables 2 and 3, and reveal the following: animacy (Human + Animate) is related to IO and also licenses the object in first (relative) position; given is more relevant for the first object. Overall, the data shows good indications that the children are moving steadily towards the target-like relevance of properties, and are paying significant attention to animacy and to givenness.

Children also display a very limited variation in ditransitive structures, and prefer a CL-NP combination of object forms. This poses some limitations to the pragmatic analysis of properties, since the Croatian clitic is obligatorily in second position. Most of the utterances with no omissions are of the structure daj mi DO (give me DO). However, daj mi is not acquired as a chunk. We have conducted corpus searches in order to find the progression of the ditransitive construction with age, and also on the lemmas daj and mi, in order to see whether they appear separately.

Overall, the research has given us good insights on how children construct their first ditransitives. They start with what is most frequent and slowly build up to more variation. The proportion of animacy and the IO-DO order point towards the upmost relevance of this property in ordering the two objects. These data provide us with starting points towards future research, including comparisons with the adult preferences, and how the relevant properties become adult-like. The data obtained by the database were also crucial for a successful design of an elicitation experiment that was already conducted.

Selected References:

de Marneffe, M. C., Grimm, S., Arnon, I., Kirby, S., & Bresnan, J. (2012). A statistical model of the grammatical choices in child production of dative sentences. Language and Cognitive Processes27(1), 25-61.

Mykhaylyk, R., Rodina, Y., & Anderssen, M. (2013). Ditransitive constructions in Russian and Ukrainian: Effect of givenness on word order. Lingua137, 271-289.

Siewierska, A., & Uhlirova, L. (1998). An overview of word order in Slavic languages. EMPIRICAL APPROACHES TO LANGUAGE TYPOLOGY20, 105-150.