Bilingual children are quite capable of keeping their languages apart. When you hear them speak in one language, you often don't notice that they can also speak another language. However, it occasionally happens that one language influences another. For example, when children use a word order from one language in another, without mixing the words of the languages. A French-Dutch bilingual child might say, ‘That's a cat angry’. In French, this is the correct sentence structure (c'est un chat en colère), but it is ungrammatical in Dutch (*Dat is een kat boos).
Such situations in which children only use a property of the sentence structure (syntax) or word structure (morphology) of one language in another language are referred to in the literature as cross-linguistic influence. 'We already know a lot about cross-linguistic influence in children's language production, but we know very little about cross-linguistic influence in children's language comprehension’, says linguist Chantal van Dijk(verwijst naar een andere website), who will receive her PhD on this subject on 17 September.
Online experiments
We do not know how often one language of bilingual children influences the understanding of the other language. It is also unclear what the role of language overlap, language dominance and age is. ‘Moreover, with one exception, the existing studies are offline’, says Van Dijk. ‘That means that only the interpretation of a sentence at the end of that sentence was measured. But research shows that we don't wait to interpret a sentence until we've heard it in its entirety. It may therefore well be that cross-linguistic influence already takes place while listening to sentences.'
For her research, Van Dijk conducted three online experiments that measure sentence comprehension during the processing of a sentence. She specifically focused on three factors that may play a role in cross-linguistic influence: language overlap (how similar are languages syntactically and/or morphologically), language dominance (which of the two languages is stronger?) and age.
Sentences in pieces
The first experiment examined how quickly children process Dutch sentences. 40 Dutch monolingual, 42 German-Dutch and 39 English-Dutch bilingual children took part in the experiment. The children heard a sentence in pieces. After hearing a piece, they had to press a button to listen to the next piece. This way it was possible to measure how long it took children to understand the different parts of the sentences: the longer it took to press the button, the more difficult it was to understand the sentence. The second experiment was essentially the same as the first, but it tested (young) adults. 25 monolingual Dutch and 25 German-Dutch and 26 English-Dutch bilingual (young) adults between the ages of 15 and 43 took part.
The results of the two experiments provide evidence for cross-linguistic influence during sentence processing. If there is some degree of overlap between the languages of German-Dutch children, they processed the sentence more slowly. However, the results are subtle; cross-linguistic influence is not always visible to the naked eye. Moreover, language dominance seems to play an important role in the occurrence of cross-linguistic influence during sentence processing. In both experiments, the effects of cross-linguistic influence were stronger as the children and young adults were more dominant in the other language. ‘It is striking that we did not find a general negative effect of bilingualism’, says van Dijk. ‘The first experiment even shows that bilingual children actually process sentences faster than monolingual children. The second experiment shows that cross-linguistic influence during sentence processing remains present in (young) adult bilinguals, although this influence is more subtle than in children.'
Offline vs Online
The last experiment was designed to investigate whether there is a difference between online and offline findings. 17 Turkish-Dutch, 23 German-Dutch bilingual and 14 Dutch monolingual children participated in the study. For this, Turkish-Dutch bilingual children had to listen to Dutch sentences. For each sentence, the children saw three pictures on a laptop screen that a personal pronoun (he/she) could potentially refer to in the sentence. The moment the children heard the pronoun, an eye tracker at the bottom of the laptop screen registered which picture the children were looking at. At the end of each story, the children were given a control question to measure offline sentence comprehension. Van Dijk: 'This experiment showed that the influence of languages in offline sentence comprehension is the same in online sentence comprehension. But the offline effect was not significant, while the online effect was. This ties in with my expectation that offline cross-linguistic influence is a weaker reflection of online cross-linguistic influence and that it is therefore useful to test sentence processing online as well.'
Block activation
Since existing models of interlingual influence cannot explain the results, Van Dijk proposes her own model. It describes, among other things, the process required to control the activation of both languages: inhibitory control. Van Dijk: 'The basic principle is that bilingual children and adults try to counteract the activation of the other language. This explains the online results in the three experiments: blocking the activation of the other language results in (partial) blocking of meanings and abstract word orders that are associated with both languages.' As a result, bilinguals are less able to select these representations. This then leads to delays during sentence processing and fewer eye fixations on the interpretation that suits Turkish.
Part of growing up
Van Dijk's research shows that the languages of bilingual children influence each other. That influence between languages can be very subtle and sometimes only becomes apparent when looking at what happens during language processing. At the same time, her results suggest that bilingual children can keep their languages apart: on the one hand, the languages of bilingual children are always in interplay, but on the other hand, children seem to be able to control and inhibit this co-activation quickly and efficiently. Moreover, it appears that the more dominant a child's language, the greater the chance that it influences the processing of the other language (visibly). Also, a certain degree of overlap between languages seems to be needed to influence each other online. This could mean that languages that are more similar (syntactically and/or morphologically) influence each other more strongly than languages that differ a lot from each other. Moreover, this study shows that cross-linguistic influence persists in (young) adults. Van Dijk: 'It seems that cross-linguistic influence is part of when a child grows up bilingually. Languages may influence each other even more than we realize. But that influence is often very subtle and invisible. Bilingual children can use their languages very efficiently.’
In this podcast episode(verwijst naar een andere website) of Kletsheads, Chantal van Dijk further discusses her research.
Photo: Torsten Dederichs via Unsplash