The study of code-mixing (CM) has yielded a lot of debate in early bilingualism, particularly in heritage language (HL) acquisition. Several works tried to shed light on the internal and external factors that promote CM use (cf. Deuchar 2020 for a review of the literature). Concerning internal factors, language proficiency, language dominance and age have been examined for different language combinations. Most studies attribute a link between these factors and CM use (from the more proficient/dominant language to the less proficient/dominant, cf. Schlyter 1993; Bernardini & Schlyter 2004). For example, Schmeißer et al. (2016) analyzed the relation between inter- and intrasentential CM and language proficiency/dominance in four French-German children (1;4-4;0) and showed that whereas the former is employed by balanced bilinguals, the latter reflects a child's strategy not to accommodate to the contextually desired language. However, other empirical studies play down this connection since evidence for CM from the less proficient/dominant LB to the more proficient/dominant/used LA has also been observed (Cantone 2007). Regarding the external factors promoting CM, numerous authors have analysed different family language policies (FLP). For instance, those families that keep the community language (also called majority language or MaL) outside home help improving proficiency in the non-MaL and, therefore, CM is less frequent (De Houwer 2007, 2009; Patuto et al. 2014, Poeste et al. 2019; ; Arnaus Gil & Jiménez-Gaspar 2022; Arnaus Gil 2022, among others).
Sixteen German-Catalan bi- trilingual children (mean age 5;7) being raised in Germany participated in our cross-sectional study that consists of a 30-minutes recording in every child's L1s in a spontaneous game situation. Moreover, we administered a questionnaire of linguistic input based on the work by Torregrossa & Bongartz (2018) to examine external factors of input quantity and quality. The aim of our study is to give insights from a less studied language pair which is, in turn, comparable to other studies that dealt with CS in typologically similar language combinations, such as Spanish/Italian/French-German in early bi- and trilingualism (Patuto et al. 2014; Müller et al. 2015; Schmeißer et al. 2016; Poeste et al. 2019, Arnaus Gil et al. 2019). Our contribution investigates the role of (i) child-internal factors such as age, language dominance and language use (measured with the aid of MLU and words per minute, respectively), and (ii) child-external factors that might promote CM use, e.g., FLP, family language and the siblings' directed speech.
The general results on CM use show that bi- and trilinguals behave monolingually in 90.4% of the cases, and thus CM only represents 9.6% of the data. CM significantly declines with age and children who are dominant in the MaL code-mix more frequently than the other groups. Interestingly, balanced and HL dominant children present instances of intrasentential CM (particularly insertions), while intersentential CM is frequent across all groups. When families stick to OPOL and do not have a family language, CM is almost absent. Finally, those sibling pairs using the HL and the MaL in their interactions show low CM rates.
