Learning a new language changes brain structurally and functionally according to research study done by Yang et al. However, their results also show that second language learning reflect individual differences in learning ability as well.

Introduction

Second language learning has become a much sought-after skill in the 21st century. Not only does it open new doors in the job market, it also broadens one’s cultural horizons. An increasingly global word is an increasingly lingual world. With this in mind, Yang and colleagues[1] conducted an fMRI study to study the changes taking place in the brain as learning progresses. They studied 39 native English speakers’ brains over a six week period in which they were taught Chinese vocabulary. Ultimately, Yang and colleagues found that those who were better able to learn Chinese vocabulary showed a stronger inter-connected network in different regions of the brain after six weeks that was more efficient than those who were not able to learn the Chinese vocabulary as well. This suggests that second language experience shapes neural network changes. Not only this, but neural changes reflect individual differences in learning success.

Review

Yang et al. are certainly not the first researchers to study the links between second language learning and neural changes in the human brain. They preface their study by giving a brief overview of the literature. “The extent of neuroimaging literature indicates significant neural plasticity both functional and structure, as a result of [second] language experience.”[1] For example, earlier studies done by Wang, Sereno, Jongman, and Hirsch[2] trained adult speakers of English to learn Chinese tones. By examining cortical activation before and after the training (using Blood Oxygen Level Dependent or BOLD signals from fMRI), their results showed that increased tone-identification performance was associated with an increase in the spatial extent of activation in multiple areas of the brain. Some of these areas were the same areas of activation before they had learned the new Chinese tones. Therefore, Wang et al. showed an expansion of pre-existing language related areas. Furthermore, their work revealed that successful learners treated the learned Chinese tones as linguistic differences rather than just acoustic differences. Wang et al. also found distinct neural activation patterns between the successful learners and the unsuccessful learners. Unsuccessful learners were found to have a more diffuse network in the frontal and temporal regions of the brain for Chinese tone learning.

Taking Wang et al. as an example of the literature, Yang and colleagues summarize from the previous literature that cognitive functions arise from the interactions between and within distributed brain systems. “In short, it looks at not just activation of individual brain regions, but also the spatial and temporal relationships between multiple brain regions during cognitive and linguistic processing.”[1]

However, what do researchers in neurolinguistics mean by efficiency in brain networks? We can define brain networks in terms of the average number of connections (edges) between regions of interest (nodes). Therefore, the fewer number of edges that need to go from one node to the next, the more efficient the network is deemed to be. Successful learners, on average, had reduced local efficiency (within a brain region) but an increase in global efficiency (between brain regions) compared to unsuccessful learners. This shows a more automatic processing of language for successful learners. These studies reviewed by Yang and colleagues employ a “functional connectivity analyses”[1] which reveal correlational relationships between different brain regions. However, this study desires to perform a “effective connectivity” analyses of the data which involves the direction of the influences between brain regions and how strong these connections are. Therefore, Yang et al. use uSEM and euSEM effective connectivity analyses on their fMRI data as they study the functional changes in the brain that occur in the second language (L2) word learners.

Method

Thirty-nine right-handed adults from Pennsylvania State University participated in this study over a six week period. They were separated into two groups: the learner group and the non-learner group. This is not the same as successful learners and unsuccessful learners; it just means that the non-learner group was used as a control group. All the participants had no experience with Chinese beforehand, and their task was to take part in three Chinese training sessions per week for six weeks. The non-learners did not receive the same training sessions, however.

Learners underwent 18 training sessions in total and learned 48 Chinese pseudowords which were composed of 16 monosyllabic morphemes. Three different pitch contours were superimposed on each of the 16 syllables resulting in 48 different possible monosyllables as the stimuli. Each stimulus was paired with a picture of a familiar object. Each of their training sessions included matching the pictures with the words. Each session also had a study phase in which they were shown the picture and the word, and then a test phase which included recognition and recall tasks. Their response accuracy in the recognition task during the test phase of each training session was used to indicate their L2 learning success.

Learners and non-learners alike participated in a sound discrimination experiment during the pre-learning phase and then again after training. These sound discrimination tasks tested the participants ability to distinguish between tone (T), onset (O) or difference in initial consonant of a word, and pitch (P). These tasks were essentially trying to see the participants’ sensitivity to the segmental features of the learned words, and to see whether learners processed learned words using tonal information.

Effective Connectivity Analysis

After the participants completed their full training, they underwent an fMRI analysis again. The learning group was now separated into two categories: successful learners (SL), those who achieved over 96% accuracy across tests, and less successful learners (LSL) who did not achieve 96% accuracy across tests. Successful learners also performed more accurately in all three conditions of the sound discrimination task after their training. Results also showed that the LSL were significantly slower than both the SL and non-learners (NL). Yang and colleagues will use this evidence that brain networks may differ in efficiency and connection patterns across individuals affecting how well one can learn L2.

Regardless of type of tone discrimination task, learners, both SL and LSL, showed increase in neural responses and connectivity which is consistent with the literature before. However, different from NL, learners showed less activation in the left hemisphere of the brain which suggest that the learners are now more efficient in processing the tones as linguistic units. Yang and colleagues state that NL show more activation in the left hemisphere of the brain, including left inferior frontal gyrus, premotor area, and left superior temporal gyrus, because NL use a rehearsal strategy more than the learners during the tone discrimination tasks.

What essentially separates the SL from both the LSL and NL is that the ROIs were not much connected or related for the brain networks of the LSL and NL groups. On the other hand, for the SL, the nodes of the brain network were highly connected even before the SL began learning the L2. After learning L2, their brain regions showed an increase of connections amongst an already highly connected network among nodes in their network. Their data shows that SL have a more “coherent” and “multi-path” network compared to LSL and NL.

Interestingly enough, SL and NL seemed to show similar behavioral results more so than SL and LSL. Although they exhibited similar behavioral results, SL and NL showed different results from the fMRI “under the hood.” Increased activation of the angular gyrus for the learners after the training session suggests that learners are treating tonal information as lexical information with semantic cues. Essentially, they “understand” the new Chinese words that they’ve learned. LSL on the other hand, showed that they treated the tonal information as just acoustic information because they had more neural responses in their left superior temporal gyrus compared to both the SL and NL.

Why would SL have more activation in certain areas of the brain such as the left superior temporal gyrus, than both LSL and NL? Yang and colleagues cite studies showing that the decrease of brain activation in certain areas can be associated with the increase of cognitive skills including “enhanced language proficiency and motor skills.” However, previous studies have also found the opposite: certain studies have found an increase of activation in the left superior temporal gyrus with increased proficiency in tone.[3] Yang et al. point out that these could be differences within the design of the stimuli, perhaps the length of the given stimuli or the amount of given words to be learned.

Discussion

Essentially, this study reveals the plasticity of human brain networks when it comes to language learning on an individual level. SL and LSL recruit different networks of the brain to handle the same lexical task. Their research more interestingly shows that even before having any experience with L2, the two groups show differences in their brain networks and how intricately they are connected. The SL group exhibits a better and more connected brain network both before and after learning L2. Yang et al. cite this as a way to possibly predict how successful an individual can be when it comes to second language learning. They also state that it is more likely that a better connected brain is more flexible brain. Results showed that the multipath feature of the SL group provided a more efficient and flexible ability to learn a new language.

Therefore, Yang et al. state that when it comes to tasks such as tone discrimination, SL were able to make this a more automatic process as they were able to retrieve lexical knowledge in the newly formed path between the IPL and the MFG. LSL on the other hand, were not able to make this new connection, meaning that instead of a more extensive, connected network, the already existing networks for the LSL showed more activation or had to “work harder.”

However, can we really consider SL brains to be truly more plastic when they already had many connections before learning L2? It seems as though a brain network that already has many connections would find it easier to learn, and therefore easier to make connections that start to include more linguistic cues such as semantic cues. Perhaps this is a positive feedback loop where connections create more connections. Furthermore, more studies are needed to test whether less activation in different of the brain region means greater efficiency.

Learning new languages is by no means an easy task, but certainly a doable one. This systematic approach to identify the brain networks in all types of language learners shows the differences of human brains on an individual level, but also how much goes into processing a language. This is by no means a conclusive study, and more research ought to be done.

References

  1. “Neural changes underlying successful second language word learning: An fMRI study” by Jing Yang, Kathleen Marie Gates, Peter Molenaar, and Ping Li in Journal of Neurolinguistics. doi:10.1016/j.jneuroling.2014.09.004.
  2. Wang, Y., Sereno, J. A., Jongman, A., & Hirsch, J. (2003). fMRI evidence for cortical modification during learning of Mandarin lexical tone. Journal of Cognitive Neuroscience, 15(7), 1019e1027.
  3. Wang, Y., & Zhang, Y. (2007). Neural plasticity in speech acquisition and learning. Bilingualism: Language and Cognition, 10(2), 147e160.