CORDIS - EU research results

Executive resources and bottom-up speech input in the first and second language of bilinguals

Final Activity Report Summary - TOP-DOWN BOTTOM-UP (Executive resources and bottom-up speech input in the first and second language of bilinguals)

It is more difficult for non-native listeners to understand speech in a noisy background than it is for native listeners. Furthermore, it has been shown that native listeners are better able to use linguistic, contextual information to compensate for sub-optimal speech input. Studies on this topic have used sentence-level tasks, which do not allow determining the level of language from which the native language benefit arises.

In this project, we used a new paradigm which isolated semantics. We tested bilinguals with an auditory version of the retroactive word priming paradigm. We presented pairs of semantically related or unrelated words, the first of which was embedded in different signal to noise ratios and was to be subsequently recognised. Our findings fit with predictions. Performance was better with less compared to more noise and in the first (L1) compared to the second language (L2). Importantly, we found a benefit of context, i.e. better performance during related compared to unrelated trials, in L1 compared to L2. Furthermore, we found that the benefit of context in L1 was proportional to the amount of noise, further supporting the finding that semantic context, specifically, contributed at least to the intelligibility advantage of L1 over L2 when listening to speech in noise.

We then repeated this study with new participants in a functional magnetic resonance imaging study. The results also fit with predictions. There was a main effect of relatedness in L1 but not in L2. Specifically, there was an influence of relatedness in L1 and not in L2 in higher-level components of the language network and in attention or executive brain regions during more top-down, context-driven processing, i.e. related trials, as well as in lower-level parts of the language system during more bottom-up, stimulus-driven processing (unrelated trials).

These results had important implications for understanding language and communication in bilinguals that could be generalised to real life situations, where speech and communication often took place in noisy external and internal environments.

Regarding brain structure and behaviour relationships and brain structural plasticity, as an extension of previous work, we recently published papers showing that brain structural differences in the left auditory and parietal cortices predicted individual differences in non-native speech sound learning. The papers also showed that brain structure in the left insula / pre-frontal cortex as well as in parietal cortices bilaterally predicted individual differences in how well individuals could articulate foreign speech sounds. We also recently submitted a paper on individual differences in behavioural measures of foreign speech sound learning. Furthermore, work included better understanding aspects of white matter anatomy using Diffusion tensor imaging (DTI).

The above results were exciting and innovative and led to the question of relative influences of innate predisposition versus experience-dependant plasticity on brain structure. One way to address this issue was to compare the brain structure of 'language experts' to that of non-experts. A relationship between brain structure in the experts and the amount of expertise would suggest that experience 'shaped' the brain. In a recent study, performed in collaboration with Dr Cathy Price, we scanned 17 phoneticians and 17 controls using structural Magnetic resonance imaging (MRI) and DTI protocols. We identified differences between the phoneticians and non-experts in the morphology of Heschl's gyri and of the temporal poles bilaterally, as well as of the left pars opercularis.

This line of work had predictive value. It suggested that information about brain morphology might help to evaluate domain-specific aptitudes, or the need for domain-specific training for particular individuals. This latter avenue had prevention, diagnostic and intervention clinical applications.