Today, there is a consensus that bilingual learners develop two separate language systems early on. This assumption is supported by the evidence gathered in longitudinal studies (De Houwer 1995; Genesee 2002; Lanza 1997; Meisel 1989; Tracy 1994/5) in which it was shown that “[c]ontrary to the unitary language system hypothesis, current evidence indicates consistently and clearly that bilingual children can use their developing languages differentially and appropriately with different interlocutors from the earliest stages of productive language use” (Genesee 2001: 3). Crucially, the acquisition of more than one language does not affect the quality of the development in terms of the developmental sequence identified for monolingual learners (Meisel 2004). This finding has also been corroborated in studies on the acquisition of sign language and oral language in hearing children (Petitto et al. 2001; Petitto & Holowka 2002).
Bilinguals’ pooling of resources
While the issue of a separate development has been settled, some scholars have turned their attention to the evidence of language mixing in young bilinguals and concluded that both languages may temporarily interact in the course of the bilingual development (Gawlitzek-Maiwald & Tracy 1996; Hulk & Muller 2000; Genesee 2002; Muller et al. 2002). Example (7), an utterance of an English-German bilingual child reported in Tracy and Gawlitzek-Maiwald (2000), provides further illustration of the type of interaction encountered when both languages do not develop at parity. At the time of its production, the structure available to the child in English was a bare verb phrase, while more sophisticated grammatical structures, including constructions with periphrastic verb forms, were available in German. By merging both structures in this utterance the child skilfully pools her resources (Gawlitzek-Maiwald & Tracy 1996). From a developmental perspective, the possibility of a pooling of resources might also have an accelerating effect on the language that lags behind as the properties that have already been acquired in one language might trigger the corresponding ones in the other language (hence the term of bilingual bootstrapping as proposed by Gawlitzek-Maiwald & Tracy 1996). The assumption is confirmed by the observation that the frequency of the type of mixing exemplified in (7) decreases after the child’s acquisition of English modal and auxiliary verbs.
Structural borrowing as in (7) is easy to detect given that the child uses lexical material of both languages. What needs to be kept in mind, however, is that the interaction of two languages in language mixing may not involve all levels of linguistic analysis (that is, the lexical, phonological, syntactic, and semantic). The range of potential combinations of elements of two languages (contact continuum) in bilingual speech suggests that different degrees of co-activation and co-production of information from different levels of linguistic analysis need to be conceived of (cf. Grosjean 1997; Tracy 2000).
The abstract combination of morphosyntactic features of two languages, commonly referred to as interference (Muysken 2004) or cross-linguistic influence (Winford 2003: 12; Kellerman & Sharwood-Smith 1986), involves lexical material from one language only which is the reason why this type of mixing often goes unnoticed (Muysken 2004: 149). In the domain of adult second language acquisition, particular attention has been paid to structural borrowing from the L1. Consider, for example, the utterance of an Italian adult learner of L2 German in (8). The target-deviant arrangement of constituents in this utterance seems to follow word order characteristics of the learner’s L1 Italian (an SVO language). Notice that in target German, the object would appear inside the verb bracket in the main clause, and the infinitive verb would appear sentence-finally in the embedded clause (Plaza-Pust 2000: 177).
Two further observations concerning language mixing in this acquisition situation are important for present purposes, namely, (a) structural borrowing in L2 acquisition is a temporary phenomenon to the extent that learners succeed in restructuring the target-deviant properties borrowed toward the target language, and (b) reorganisation in L2 grammars is commonly tied to variation, i.e. there is an apparent coexistence of target-like and target-deviant properties (which is in line with dynamic view of change sketched previously, section 22.214.171.124). For example, in the development of L2 German by the Italian learner mentioned previously, we observe the alternate production of target-deviant and target-like constructions with periphrastic verbs (cf. (9) and (10) produced during the same recording session, Plaza-Pust 2000: 183) prior to the eventual implementation of the target German word order.
In conclusion, the progressive convergence of the different lines of research in the domain of bilingualism has provided further insights into the role of language mixing in the organisation of multilingual knowledge in child and adult learners. The sophisticated combination of two distinct grammars in mixed utterances indicates that bilinguals (tacitly) know, by virtue of their innate language endowment (i.e. UG), that grammars are alike in fundamental ways. Thus, “language mixing, either temporarily as a help and bootstrapping mechanism in acquisition or as the permanent potential of the proficient bilingual is only a natural consequence of that (tacit) assumption” (Tracy 1994/5: 484).