Abstract
Introduction
Recently we observe an increasing interest in individual differences in skill acquisition, including second language learning skills. Huge evidence has been accumulated describing functional or structural brain variations accompanying second language proficiency level differences, caused by various variables (Reiterer 2005). However, one variable causing individual differences in language learning has been largely neglected so far: the aspect of “talent” or language aptitude. Despite few notable studies (e.g. Golestani 2007, Amunts 2004) – language aptitude has not been investigated so far using brain imaging tools, although one can observe that people vary greatly in their capacity and speed of learning the various subcomponents (e.g. accents) of foreign linguistic systems. We therefore investigated language aptitude from a neuropsycholinguistic perspective, focussing mainly on the outstanding skill of “pronunciation talent” in a second / foreign language
Methods
We recorded 138 German-speaking subjects during imitation of foreign sentences in a language they had no experience with: Hindi. Using an internet rating database (Jilka 2009) we asked 30 native speakers living in India to evaluate the “goodness” of the imitation. Based on this we investigated 60 participants of this large pool with fMRI during foreign (English, Tamil) and native language (German) speech production. From these participants with varying degrees of “phonetic imitation aptitude” we selected the top 10 with the highest scores (talent group) and 10 with the lowest scores (poor talent) for performing group analyses (main effects – task versus baseline, random effects model) using the statistical software SPM5. The fMRI experiments were performed on a 1.5 T scanner employing a sparse sampling paradigm (TR12s) for acquiring speech production in the scanner.
Results
The group analyses show differences between these extreme groups, mainly for how they activate the sensory-motor-acoustic cortices adjacent to the Sylvian Fissure, left insula, the supplementary motor areas, Broca's area and the cerebellum (see Fig.1). Talented participants show restricted, less intense activations in these areas, whereas activations for the “low talent” group are stronger in intensity and cluster extent, regarding more extended networks within and around the aforementioned areas. If the low talent group is compared to the high talent group by direct group comparison, the first additionally shows significantly higher activations in left middle and inferior frontal areas (Broca) as well as left insula and right superior temporal gyrus. The patterns characterize the groups in similar ways across all language tasks, however group differences are less pronounced in the task “word imitation” as compared to “sentence imitation”.
Conclusions
Preliminary analyses of our data showed that individual differences in pronunciation aptitude are accompanied by differences in cortical tissue recruitment. The main difference between good and poor language “imitators” seems to lie in the way how they employ their articulation-relevant areas. We found that individuals with good pronunciation skills (“pronunciation talents”) focus more precisely on the areas classically known as relevant for speech production and articulation (Ackermann 2004), whilst the poorer pronouncers activate the same brain regions, but more intensely and rely on more extended networks involved in speech production. We hypothesize that this reflects a phenomenon of higher versus lower levels of processing effort.
