A Good Head for Grammar

Some people learn foreign languages faster, easier and better than others. What makes them so lucky - and so different? Read about the findings of linguist Olga Kepinska.

A Good Head for Grammar

The common view of language processing is that it takes place in the left hemisphere of the brain. But when it comes to language learning, linguist Olga Kepinska discovered that it is the right hemisphere that distinguishes the excellent from the average grammar learner.

Some people learn foreign languages faster, easier and better than others. They just seem to have a knack for languages. Why is that? What makes them so lucky – and so different? Neurolinguist Olga Kepinska, herself an avid language learner, has been intrigued by this question since she was a bachelor student of Dutch at the University of Wroclaw, Poland and later used it as the starting point of her PhD research, which she did at Leiden University. Her thesis, entitled ‘The Neurobiology of Individual Differences in Grammar Learning’, is now finished and she will defend it on 1 June, 2017.

What makes up a good brain for grammar

‘Language learning is very complex’, Kepinska says. ‘Some people have a really good memory for words, others have an analytical ability; they see grammatical patterns quickly and use them without much difficulty. Others again have a talent for pronunciation. And very few are good at everything. I had only four years, so I had to choose which aspect to focus on. She decided to focus on grammar, the analytical side of language learning. What makes up a good brain for grammar? The PhD candidate recruited 240 students who were non-bilingual native speakers of Dutch, and made them take a language aptitude test in the computer lab of the Faculty of Humanities. ‘I was particularly interested in the 80-100% scorers’, says Kepinska. ‘We wanted to find a group of participants who are outstandingly good, and then use neuroimaging techniques to discover how this is reflected in their brains.’ 

Artificial language

Quite a few students scored excellently, so she now had more than 40 high-performing participants at her disposal, whom she could compare with 40 average performers. She divided the group of approximately 80 students into two cohorts: one cohort for an EEG experiment and the other for a number of different MRI experiments. They all had to figure out the grammar rules of an artificial language. While the participants were solving the puzzle, Kepinska measured activation levels of brain areas and networks, to find out what happens in the brain during the process of learning. However, she also looked at white matter, the part of the brain lying underneath the cortex, connecting all its parts. ‘There are pathways that connect the language regions of Broca and Wernicke, and I wanted to know if their structure and density is different in highly skilled grammar learners. This would give an indication about how efficient their brain is.’

foto 2 Kepinska artificial language
An example of a trial from the test phase of the artificial grammar learning task

Good learning is hard work

Her first finding was that the brains of the excellent performers show more activation than the brains of the average performers, which indicates that they work harder. This seemed a bit surprising in the light of the so-called ‘neural efficiency hypothesis’, according to which the skilled brain uses less, not more, resources for a specific task than the unskilled brain. Kepinska: ‘But the neural efficiency hypothesis is about learning of tasks of average difficulty. We also know that in difficult tasks highly skilled people use more neural resources than others. That is to say: until they have mastered the trick. After that point, they use less resources than others.’ And this was precisely what she saw: the higher activation levels disappeared as soon as the excellent learners had detected the grammar rules and started working with them.

Right hemisphere

However, the most surprising finding was the crucial role the brain’s right hemisphere plays in grammar learning of very good performers. Kepinska: ‘The common view is that for grammar processing we use the left hemisphere. If you have a stroke in the left hemisphere, you typically get difficulty producing grammatical sentences and understanding complex sentences. However, we discovered that the areas working harder in excellent grammar learners are in the right hemisphere, not the left. This is only the case in the first stages. After the initial stages of learning, the language centres in the left hemisphere take over. Kepinska: ‘It looks as if the right hemisphere is helping the left.’ Both EEG results and analysis of structural MRI data corroborated this finding. Analysis of the structure of white matter showed that specific white matter pathways connecting the frontal and temporal lobes of the right hemisphere are more coherent and better organised in excellent learners.

Foto 3 Kepinska more effort brains 428x400

Brain activation in response to ungrammatical as compared to grammatical items was significantly greater for the High than for the Average LAA participants


DSC 0514 450x300

Olga Kepinska and her research

A linguist’s way into brain research

‘I am a linguist, not a neurobiologist’, Kepinska emphasizes.  But highly complex and innovative neuroimaging methods did not scare her off.  What is needed for a linguist to work her way into the world of brain research?Kepinska summarises her learning process: ‘After my BA in Poland, I came to Leiden for a master in Dutch studies but I also picked up on experimental linguistics. And afterwards, I did an additional  research master in Brussels, where I learned functional MRI. But in my research, I also used structural MRI data, in order to look at white matter. And that was something completely new for me! I found a workshop in King’s College, London dealing with the technique but you also really need to know a lot about neuroanatomy. So, I took a first-year bachelor course for medical students in the LUMC. I actually went to the dissecting hall and held a post-mortem brain in my hands, which was very special. Each brain is so different!’

‘We have just the guy for you’

Help from others was of crucial importance. Kepinska: ‘My Leiden MA-supervisor Johanneke Caspers introduced me to my promotor Niels Schiller, professor of psycho- and neurolinguistics. The Leiden University Centre for Linguistics has its own EEG labs, and when I started my PhD research, my colleagues there helped me learn EEG. One of the MRI research scanners in Leiden University Medical Centre is administered by LIBC, and Mischa de Rover from the LIBC support team helped me figure out how to analyse my MRI data. Also, Egbert Lakke, my neuroanatomy teacher, helped me enormously. Together, we made virtual dissections of 42 brains, using diffusion tensor imaging, which gives you an image that approximates the structure and density of white matter.’
For a specific, little-researched, part of her EEG research, Kepinska contacted the authors of an Austrian paper about the differences between proficient and non-proficient native German speakers of English. ‘They said “We have just the guy for you, he works in Tenerife”. So, I went there, which was great, and he and I wrote a paper together. This kind of collaboration between people from extremely different backgrounds is so fruitful! It moves the field ahead.’

More about Olga Kepinska’s research

The LIBC hotspot Language covers all areas of human (neuro-)cognition that are related to language processing. Language is part of human behavior and probably one of the most complex cognitive skills.
Read more