Much evidence in usage-based research on language acquisition suggests that when acquiring a language, children use low-level generalizations instead of abstract patterns (Ambridge & Lieven, 2015). Bilinguals could rely on low-level-generalizations even more since the ready-made patterns are more easily processed. First data with bilingual children also suggests that code-mixed utterances consist of specific chunks (Quick et al., 2018).
The present study aims to further investigate the use of low-level-generalizations in German-French bilingual children (4, 6 and 8 years, N= 118) in comparison to monolingual control groups (N=119) focusing on the expression of caused motion. The corpus consists of elicited data from object displacements of four different types (Hickmann & Hendriks, 2006).
First, the verb use in each type of caused-motion event was analyzed in order to check whether one prototypical verb emerges for the different caused-motion events, which in turn can be indicative of a low-level-generalization with a lexicalized prototypical verb. Results showed the frequent occurrence of one specific verb for some events. High variation was observed, especially regarding the factors language (French vs. German) and age.
Secondly, the variation in the slots of verb and directional phrase was measured, counting the number of different combinations in each subject, in order to draw conclusions about the degree of abstraction.
Low variation was considered an indication for the use of slot-and-frame patterns. The goal was to investigate whether variation in lexical content increases with age and whether differences between mono- and bilinguals are observed. While results showed no variation when considering age, the groups differed in German, where the monolinguals showed more variation.
The results enable us to take a developmental perspective on the use of low-level-generalizations in bilingual children and will be discussed in the light of current research in usage-based language models.