Of Minds and Language (62 page)

Although this study provides some clues about the relative weight of age of acquisition vs. amount of exposure, it is not possible to identify the importance of the different language subsystems in the observed results, due to the way the authors measured L2 proficiency. Indeed, as mentioned, not all aspects of a non-native language are equally easy to learn. In what particular dimensions did good and poor L2 learners differ – that is, was it a question of vocabulary, pronunciation, or syntax? What kind of linguistic (or nonlinguistic, cognitive) processes are responsible for the differences that the authors observed in grey matter in the left inferior parietal cortex? The following studies that I will discuss have focused their interest on more specific aspects of learning a new language.

Although learning new words is an ability that most human beings retain throughout life, the fact is that some people are clearly better at it than others. The next study was designed to address this particular issue. Breitenstein and colleagues (2005) showed participants in their study pictures of known objects while they heard new words being pronounced at the same time. Each picture (e.g., a book) appeared ten times with the same new word (“bini”). However, since in real life learners do not hear words in isolation every time they see a particular object, pictures also appeared, only once, with each of ten varying new words (e.g., “enas,” “alep,” etc). Subjects had to intuitively learn that the most frequently occurring couplings were the correct pairs.

The participants' brain activity was measured while they were learning the new words (using functional magnetic imaging). Several brain areas were differently activated in individuals who were good and poor at vocabulary learning. Three areas are of particular interest. First, good learners showed greater activation in the left inferior parietal cortex, an area very close to the one reported by Mechelli et al. (2004). In fact, Breitenstein et al. (2005) were able to observe an increase in the activation of this area as a function of learning: there was a strong correlation between the increase of correct responses during the experiment and the activation of this area. The other two areas were the left fusiform gyrus (an area previously reported to be
involved in visual word recognition) and the left hippocampus (responsible mainly for short-term memory).

Short-term memory, and in particular, phonological short-term memory has often been associated with the capacity to process syntax and language comprehension in general. Chee and coworkers (2004) wanted to test the impact of differences in this particular cognitive component on the ability to learn a foreign language. In this case, Chinese–English early bilinguals were compared. The participants in this study had learned both languages before the age of 5, but they differed in their second language proficiency. They were tested in a phonological short-term working memory task. Subjects listened to French words (an unknown language for all of them) and were required to perform three different tasks. In the first task, they had to press a button every time they heard a particular word. A more difficult task required them to press a button whenever there were two successive words that were the same. Similarly, in a still harder third task, subjects were asked to compare two non-adjacent items. That is, they had to press a button whenever, in a sequence of three items, both item 1 and item 3 were the same. Behaviorally speaking, all participants performed similarly in the three tasks (no differences were observed between good and poor L2 learners). However, important differences were observed in the brain areas activated in the two groups of participants. The brain activation scans showed increased activity in the areas where the good L2 learners were better than the poor L2 learners. One of these activated areas was the left insula, a language area associated with phonological processing. However, out of the four different areas that were shown to be activated differently in the two populations, most were not typically associated with language. One area that was more activated in good than in poor L2 learners was the left cingulate. This area is known to be involved in mechanisms that control the ability to inhibit information. Therefore, the performance of individuals who managed to do the tasks better can be partly accounted for by the fact that there was enhanced processing not only in some language-related areas, but also in different areas of the brain that are responsible for attentional and inhibition processes.

The final group of studies that I want to mention is related to the ability to learn non-native contrasts. In a series of experiments conducted by Golestani and her colleagues (Golestani et al. 2002; Golestani et al. 2007), the authors tested the differences in brain structure between good and poor learners. Golestani and coworkers taught different groups of monolingual native English listeners the alveolar retroflex /da-dha/ distinction, which is very difficult for these participants to perceive. The results showed differences in white-matter volume in some parietal areas, actually very similar to the ones reported in the study of Mechelli et al. (2004). In the present study, fast learners showed greater
white-matter volume than slow learners. In a subsequent study, this time with French listeners, the same findings were replicated and also, using a different analytical technique, anatomical differences between fast and slow learners were obtained in the Heschl gyrus in the left hemisphere. Approximately two-thirds of this gyrus is primary auditory cortex, which is thought to be involved in the processing of rapidly changing stimuli, and therefore supposed to be active in processing consonants, like the ones being learned in this particular experiment. But part of it is considered secondary areas, so probably related to language. It should be stressed that in this study the vast majority of the differences reported were between fast and slow learners and not between good and poor ones. That is, participants were classified according to the speed with which they reached a learning criterion. Furthermore, in these studies exposures were very short. Indeed, in one of the experiments, the training lasted 15 minutes. So, it is impossible to determine to what extent the differences reported were caused by better auditory processing, phonological processing (the ability to create new phonetic categories), or attention.

The final study that I want to present is one that we are carrying out in Barcelona. In this case, a major difference is that we did not train our participants to perceive or learn anything. We were just testing a population of individuals who had all been exposed to two languages very early in life to see whether we could find any differences in brain structure between those who managed to learn the phonology of the L2 with native proficiency and those who did not. We tested our University of Barcelona Psychology students in a variety of behavioral tasks and selected those who fell below native level in all tasks, and those who performed like natives in all tasks. To these ends, the subjects performed a categorical perception task, a gaging task, and an auditory lexical decision task, using procedures we had employed successfully in earlier experiments. The capacity of Spanish–Catalan bilinguals, who had been exposed only to Spanish at home, to perceive the Catalan-only /e/–/ε/ vowel contrast was explored. In previous studies we have shown that this is a contrast that native Spanish listeners have great difficulty in perceiving. As mentioned, we chose for the study individuals who failed at all tasks (23 percent of the population), and individuals who succeeded in all tasks (12 percent), ending with twenty participants in each group.

The scans revealed differences in brain areas not directly related to speech processing, including the fact that the right frontal operculum was much more myelinized in the poor perceivers than in the good ones. This came as a surprise, and we wondered why poor perceivers should have more myelin in this area, which is involved in an auditory attention network. Studies have shown that it is activated whenever we hear something where (a) there is a minimal difference
between two different sounds, and (b) the difference is difficult, but not impossible, to perceive. That is, we have to pay attention to it, and this is what activates the area. If the difference is too small to perceive, then the area is not activated; if it is very easy to distinguish, it is not activated either. What if our poor perceivers in the L2 contrast were also poor perceivers in their L1? To test this, we measured the electrophysiological activity of our two groups of good and poor L2 perceivers. In this study participants listened repeatedly to the very same stimulus, and then from time to time one was inserted that was different, to enable us to observe electrophysiological differences in the brain when the different stimulus was heard. There is a measure known as mismatch negativity that shows how well we perceive these “odd” stimuli. Because the hypothesis was that poor L2 perceivers were going to be poor perceivers generally, we tested them with an L1 contrast: the /e/–/o/ contrast that exists in both languages. Indeed the results showed that trials with good L2 perceivers elicited a larger amplitude of mismatch negativity than trials with poor L2 perceivers, indicating that the former can perceive the difference between /e/ and /o/ more accurately. It has to be emphasized that our poor L2 perceivers exhibited a totally normal, though reduced, mismatch negativity. This suggests that the latter group probably cannot learn the L2 contrast because of a very mild deficit that is not important for the learning of the L1, but is catastrophic for learning some particular aspects of the L2.

What, then, are the reasons why some people are so poor at learning an L2? I realize that answering, even tentatively, is very premature, because there have been very few such studies, but all the data so far indicate that while there are differences in brain structure and function in the groups tested, generally speaking the differences tend not to be in language-related areas. So in most cases we find that it is not the language faculty that keeps us from learning an L2 proficiently, but our general cognitive capacities. One puzzling question for which no clue is provided by these studies is why such differences are not at all important for L1 learning. Remember, in all the Barcelona studies, the participants were University students, supposedly the percentage of the population that has proved successful, not only in passing through the educational system, but in learning language in particular. None of them reported having any difficulties in learning their L1 or difficulties in learning reading, which is also a domain where problems with the processing of speech can give rise to difficulties. Why is it, then, that the differences detected are so important to learning an L2, but not an L1? Given the way the experiments and studies reported in this talk were conducted, it is not very likely that differences in amount of exposure can play a crucial role.

In another study, we tested simultaneous bilinguals (people exposed to two languages from the very first day of their lives) (Sebastián-Gallés et al. 2005). In this case, the age of first exposure was exactly the same day. Of course there were differences in amount of exposure to the two languages, but in our experiment this was as controlled as it can be. The results demonstrated that what is heard in the first year is truly critical. Participants whose mother was a Catalan speaker performed much better than those whose mother was a Spanish speaker. Since young infants are mostly taken care of by their mothers (as opposed to their fathers), in the first months of life babies usually hear the mother predominantly, and that is when the phonemic categories are being established. The participants in our study were young men and women in their twenties, but the impact of very early Li exposure can be traced back.

To conclude, let us return to the Tower of Babel. Genesis provides an answer to the question of the origin of language diversity, but what finally happened to the Babelians? Many of them ended up on this continent, where so many different languages are spoken. Clearly, we need to be able to talk to each other, we need to be able to build up the unity of humanity, symbolized by the Tower. Whether or not we will succeed is still an open question. Perhaps Europe will end up like the Tower in a disastrous fall. I don't know. In any case, we have no choice because we live on this particular continent, but there is some hope for us non-native speakers of English.

P
IATTELLI
-P
ALMARINI
: Several years ago Elissa Newport and Jacqueline Johnson had a very thorough study of native Korean speakers who immigrated into the US at different ages, and already it was very clear that the age at which they started acquiring the language was the only crucial factor (among e.g. motivation, intensity of exposure, etc.).
¹
They also noticed that there were difficulty differences. For example, there was no big problem in learning very different word order. The really serious problem was determiners, the English definite article, which is puzzling to us who are not native speakers. So I wonder if you have data of this kind – that is, which components of language are hardest for L2 learners to acquire.

S
EBASTIÁN
-G
ALLÉS
: It is not determiners, by themselves, that are difficult. The literature on second-language learning has shown that equally important are the properties of your first language. If your first language has determiners, then it is easy. [audience reaction] Okay, I didn't want to go into this, because then things
become extremely complex in the sense that, if something does not exist in your L1, then at the beginning it is very difficult because you need to acquire something that your language doesn't have. But later, you may benefit as you don't have interference from your first language either, because there was nothing before. In any case, this description is very general. The question is that learning determiners also means to learn many more things. I am not a linguist, but I am pretty sure you all know that learning a determiner means not only that you have to add a determiner in certain positions. It means that you have to adjust the whole system, and some adjustments are going to be easier than others. Weber-Fox and Neville (1996) have a paper showing specific aspects of the L2 that are particularly difficult for Chinese learners of English, and even the studies of Elissa Newport are interesting in this regard. I think that the first study was done with people speaking Asian languages. Jim Flege (Flege et al.1999) and others used the same materials but tested Italian and Spanish natives. Although the analyses of the results were not done in terms of what specific linguistic structures are difficult, you can deduce from the results reported in these latter studies that there were important differences between speakers of Asian languages and speakers of Romance languages, when learning some specific properties of English. The overall picture is that it is clear that the first language imposes very important constraints on the learning of the second language. So the message to take home is that you cannot make a universal foreign language textbook. You have to consider what the learner already knows.