A recent piece on the Scientific American website highlighted a couple recent studies, one of which I read a few months ago. The headline reads as follows:
Some People’s Brains Are Wired for Languages
Brain scans may offer clues to a person’s natural aptitude—and help those less gifted learn how to study better
For example, in the first study in the SA article (Qi et al., 2015), Mandarin learning was predicted by right hemisphere measures (basically, differences in the structure of white matter). Mandarin learning is based on a classroom Listening/Reading/Speaking test, a pinyin dictation task, and a vocabulary test. Few details were provided about the classroom test, other than that it was presented mostly in Mandarin (so Chinese characters?) and scored by the teacher. Part of me likes the inclusion of the classroom test, which brings in some communicative skills- it's also reflective of real-world language learning and outcomes that people care about. But often, teacher-made and scored classroom tests focus on discrete bits and highly-practiced materials- so studying the night before (or in the 10 minutes before class) could be just as influential on results as actual acquisition of linguistic forms and development of communicative competence. The dictation task required students to write 10 (yes, just ten) words/phrases in pinyin, a phonetic script with Latin letters. The obvious confound here is that pinyin knowledge could be confounded with Mandarin phonetic discrimination. The vocabulary test involved 30 words presented in English (how often were these words presented in the curriculum?) and a choice of two Chinese characters, one of which was a correct translation and the other a similar-looking character (imagine choosing between star and stare in English)- so, a 50% chance to get it right just by guessing. Scores for the latter two measures were percent correct. To some extent, the limited coverage of the learning measures understandably reflects the limited learning experience in the study (60ish hours of classroom learning). It could be that the brain measures relate to a general analytic capacity/phonetic discrimination capacity rather than language learning/aptitude for language learning, and the psychometric properties of the tests potentially introduce a fair amount of noise on top of that.
The second study (link here) French learning was predicted (substantially- 60% of variance explained) by resting EEG readings. French learning was operationalized by progress through an interactive/game-based French learning course (about 8 or so hours worth of study, from what I can tell). The course actually sounds really cool (for lack of a better word)- you have to carry out real-world-esque tasks in a virtual French immersion environment. But progress through a self-paced language course could be based as much on personality and general analytic/reasoning skills as an aptitude for SLA. In the SA article, one of the study authors admits as much, which was nice to see. In a Duolingo project I've worked on (as a participant-researcher), I outpaced a few colleagues in learning Turkish, but that alone doesn't mean I actually acquired more Turkish. The authors did get a bunch of general cognitive measures from the participants, none of which had significant correlations- which could be support for the resting EEG having a meaningful connection with language learning, or it could just be the case that not many correlations are going to be significant when you have 16 people.
Although I'm critical here, I do think there's a lot of potential in this work. I definitely think these authors are on to something, but I hope future work can clarify what that is- and on the language side, better-defined and operationalized measures of linguistic competence/communicative ability could be a great help.