In the final segment of our e-interview with Natascha Müller she presents some testable assumptions, which she has corroborated in her own research.
“The two assumptions for delay effects are testable, since they make different predictions. In my research I have argued against Serratrice, Sorace & Paoli (2004) and I have assumed that the presence and direction of the influence is not a question of MORE-or-LESS constraints, but a question of whether pragmatics decides on syntactic options or not. The invasive function of pragmatics is complex (Italian, Spanish), non-invasiveness is derivationally neutral (English, German, French). If delay is not due to a default strategy (processing load in one language) but motivated by cross-linguistic influence where the linguistically more complex analysis of language A is avoided in favor of the less complex analysis of language B, we can make the following predictions which have been corroborated in the child data we have analyzed:
a) The effect should only be observable in bilingual children with particular language combinations, i.e. it is not due to the fact that the children acquire two languages, one with more, the other with less constraints, generally speaking. Only an approach to delay effects which takes into account the structure of the two languages involved will be able to account for the following observation: one and the same language A (in the case of null-subjects Italian) is affected by delay effects if and only if the surface strings of the two languages A and B are analyzable in terms of the syntactic derivation of language B (which is less complex, in the case of null-subjects German).
b) The effect should be independent of language dominance, in the sense that the linguistically more complex language is affected, not the weak language. We have even been able to show that the effect is most visible in balanced bilingual children. Rare are the studies which observe acceleration effects in bilinguals. Notice that the research by Bialystok (2001) has shown that bilinguals have, for certain cognitive tasks, advantages over monolinguals. Acceleration means from a developmental perspective that speed of acquisition is accelerated as compared to monolinguals. The question, however, is whether acceleration is an outcome of cross-linguistic influence or an effect of processing. The less complex grammatical analysis (complexity defined as the coordination of information from different modules, pragmatics and syntax for example) is used and bootstraps the analysis of the more complex grammatical property in the more costly language. Trivially, acceleration can never lead to target-deviant grammatical representations because acceleration is a positive effect and erroneous representations would have to be corrected during the course of acquisition. The idea that complexity can also lead to acceleration is strange for the following reasons. Let’s assume that language A and language B exhibit different (linguistic) complexity for a particular grammatical phenomenon. If this constellation leads to delay of the more complex language, how can the same situation lead to acceleration? Is acceleration due to a default strategy adopted when one of the grammatical analyses requires less processing load and is therefore preferred over the more complex analysis? If yes, we would make the following prediction which again has been corroborated by the child data we have analyzed:
c) The effect should be observable in bilingual children independent of the language combination. The structure of the two languages involved will only matter to the extent that complexity is being defined. One and the same language A (finite verb placement in German for example) is affected by acceleration effects, even if the surface strings of the two languages A and B are not analyzable in terms of the syntactic derivation of language B (which is less complex, in the case of finite verb placement Italian and French).
The study of bilingual children enables us to tangle apart the issue of competence and performance and it can be used to show the effects different languages have on one and the same language because bilingual children are individuals who develop one performance system (with all social variables being constant) but two competence systems. This fact makes the study of bilingualism interesting and important for a theory of language in general and for a theory of language acquisition.