The bilingual advantage
« previous post | next post »
Back in February, I posted about a terrific symposium on bilingualism at the AAAS meeting ("What bilinguals tell us about Mind and Brain", 2/19/2011). Along with the symposium's abstract and a list of the participants, and some complaints about the AAAS's failure to make it symposiums accessible to a broader public by putting them on line, I promised that "If I have time, I'll summarize some of this work in a later post" — which never happened.
One of the most striking topics covered in that symposium was the fact that bilingualism offers, on average, about five years of protection against the symptoms of Alzheimer's, apparently by creating a "cognitive reserve" in executive function that allows people to continue performing at a higher mental level for a given degree of brain degeneration. This research was the focus of a recent New York Times article: Claudia Dreifus, "The Bilingual advantage", 5/30/2011.
The article is in the form of a Q&A between the reporter, Claudia Dreifus, and the researcher, Ellen Bialystok. The content is excellent — clear, to the point, not hyped or spun for effect — and there are links to the research papers!
Update — Lila Gleitman writes:
I just looked at the front page of the NYT (on line) and found two ill-assorted articles. One shows Byalistok's work on "the bilingual advantage" — her research is one of several kinds to appear lately pointing to enhanced cognitive flexibility in bilinguals as opposed to monolinguals (see also, e.g., Kovacs & Mehler, and recent findings from Trueswell & Thompson-Schill). The other article announces that NY State will drop the Regents Exams in foreign languages for a savings of $700,000 a year. Perhaps this isn't as bitterly ironic as it could be, given the doubtful value of classroom teaching of foreign languages. Still, it seems the US continues to be incoherent in educational policies supporting what Obama calls "American competitiveness."
Gavin said,
June 1, 2011 @ 6:27 am
It was great seeing an article advocating for bilingual education and the beneficial effects it can have for future health. I agree, great content, but I still have some reservations about how the executive function operates in how bilingual/multilinguals select and deploy language.
Andrew said,
June 1, 2011 @ 7:16 am
Couldn't it be that the causation goes the other way around? i.e., people with better executive function are better able to pick up a second language and stay bilingual throughout life?
[(myl) In a word, no. At least I don't think this is a plausible story. But if you're curious about this, check out the studies and report back.]
Johanne D said,
June 1, 2011 @ 7:46 am
To me, it's all the more sad that my mother, now 93, has been suffering from Alzheimer's disease for a decade. She speaks and has taught French, English, Spanish and Esperanto. I remember her saying that she was the first woman in Canada to obtain a certificate for teaching Esperanto.
At the home where she lives, several employees are from Latin America and they love to speak Spanish with her.
I can only tell myself that we would have lost her (the person she was) even sooner without her amazing abilities, which she has partly passed on to all her descendants. She even met my father in a Spanish class, in 1943, where they vied for the rank of star pupil!
(Anecdotal of course, but I couldn't stop myself from contributing this.)
[(myl) The implication of the cited studies is that without the cognitive reserve created by her language experience, she probably would have begun showing symptoms about 15 years ago instead of about 10 years ago. The development is a personal and familial tragedy at any age, of course. ]
David L said,
June 1, 2011 @ 9:50 am
A quick glance at the studies suggest they all involve bilingualism acquired in childhood, as opposed to later in life (or does bilingualism in this context strictly mean an ability from childhood versus fluency acquired later?)
I raise the point because I have a vague recollection of reading about studies (fMRI? PET scan?) indicating that 'true' bilingualism operates differently, in terms of the brain areas it engages, as compared to mere adult fluency in another language…
[(myl) In Bialystok et al., "Bilingualism as a protection against the onset of symptoms of dementia", Neuropsychologia 2007, the monolingual vs. bilingual split was determined by the following procedure:
In connection with Andrew's concern about the direction of causation, it's conceivable that people with better executive function were more likely to learn English when they immigrated to Canada, or more likely to keep up usage of their native language, or both. But my impression is that people who immigrated to Toronto from Central and Eastern Europe in 1940-1960 are extremely likely to have learned English; and that whether or not they continued to use their native language(s) during the rest of their life is much more likely to have depended on their social circumstances than on their degree of cognitive reserve.]
J. W. Brewer said,
June 1, 2011 @ 10:25 am
For the 2007 paper w/ Craik and Freedman, the later-dementia-onset bilinguals sample was overwhelmingly made up of adult immigrants to Canada (mostly from Eastern European backgrounds, and generally arriving in the New World many years before the study) who had retained L1 fluency while becoming fluent in English. The monolingual comparison group was overwhelmingly Canadian Anglophones. The comparison I would think one would like to see (if the data was available . . .) would be with a group of immigrants to Canada of similar vintage from similar backgrounds who had lost their L1 fluency over the decades (and/or with immigrants who stayed in some sort of ethnic enclave and never got very good at English and/or with dementia patients in the L1 old countries).
[(myl) There are some other studies with different populations that come to similar conclusions, but I agree that in this particular case there's a danger of confounding with some other factor, like childhood consumption of herring.]
J. W. Brewer said,
June 1, 2011 @ 10:32 am
I was alas posting simultaneously and thus largely redundantly with myl's comment. But his hypothesis that the differences between immigrants to Toronto of that background and generation that affected whether they kept or lost L1 fluency were social/situational rather than cognitive means that the comparison between those subgroups (which the paper doesn't do) would be particularly illuminating in whether active adult bilingualism as such is doing the work here. I'm also not sure whether the reference to many being bilingual before arrival means they already knew English, or simply knew more than one language (e.g. many/most Yiddish speakers from the Old World were fluent in at least one goyische language as well, and others might have been intra-European refugees through DP camps etc for some years before getting across the ocean), at least one of which was maintained in Canada.
Brandon said,
June 1, 2011 @ 11:39 am
Does this effect still show up when controlling for other cognitively-intensive common tasks? Job, IQ, level of education? The causative theory passes a smell test, but if it's true, then many other things should also cause it.
Theo Vosse said,
June 1, 2011 @ 1:12 pm
I can't believe such conclusions (bilingual vs Alzheimer) anymore. I've seen too many studies where the conclusion was based on a GLM; this one probably modelled level of education, gender, etc. as factors in such a model. If there would be computational model that models e.g. cognitive load in a neural network and predicts Alzheimer onset as an emergent property, and it fitted this data, then I might be convinced.
[(myl) There's converging evidence from studies of executive function in bilinguals, as well as a plausible theory (with both behavioral and imaging back-up) about why being bilingual builds executive-function brain-muscles (because of the constant need, in production and perception, to inhibit responses from the language not currently being used).
I'm not sure whether anyone has bothered to construct a pseudo-neural-net model exemplifying that process — but I don't think you ought to find it convincing if someone did, since I'd be happy to bet you a year's salary that someone could, given that the overall story hangs together, and that PDP models (in the broad sense) are basically a universal programming language.]
And then everyone seems to jump to the conclusion that there should be more money for language education, as your quote of Lila Gleitman illustrates. While I personally find it a laudable goal (even classroom teaching, which at least succeeded in teaching me the basics of four foreign languages), there is a risk that it will backfire. Sooner or later, someone will find that bilinguals suffer from the same neurological problems, or that they score worse on some other test, and that will be an argument to reduce language education.
[(myl) You're right that this has little or nothing to do with secondary-school foreign language teaching, given how rare it is for genuine bilingualism to be achieved by that route. But in fact, Lila said that "Perhaps this isn't as bitterly ironic as it could be, given the doubtful value of classroom teaching of foreign languages".]
Mike said,
June 1, 2011 @ 1:26 pm
Are there any cognitive benefits associated with learning a foreign language later in life, say after age 30 or 40? I know it's more difficult. I also I know it's a joy and opens new worlds. But are there positive "side" effects too? Nobody ever seems to talk about this. The talk is always about bilingualism from a young age — something that is, for me unfortunately, impossible to attain at this point!
J. W. Brewer said,
June 1, 2011 @ 2:56 pm
On the side note, what is apparently being defunded is not actual foreign-language instruction in high schools, but language-specific Regents Exams, which have long been a peculiar/unique feature of New York state's public education system (which, speaking as a current NYS taxpayer and parent who grew up elsewhere, seem to have more to do with local tradition and bureaucracy than demonstrated improvement of educational outcomes). Now, having a uniform statewide test is a potentially useful way of comparing the output of different school systems throughout the state (what proficiency does 4 years of French with B+'s mean if done in district X v. district Y?), but perhaps there are different ways to do that. The SAT Subject Tests and the AP tests cover a wider range of languages than the Regents did, and there are national language-specific exams run by various teacher's groups.
There's a separate issue about whether making some degree of foreign-language instruction mandatory for a high school diploma (which is at least what NY is purportedly moving toward) is actually desirable in promoting even modest bilingualism or will, given political reality, simply lead to the level of foreign-language instruction being dumbed down to a point where essentially everyone will pass, and take resources away that could be used to offer more rigorous instruction for a more self-selected subset of students with a genuine interest in the subject matter.
C said,
June 1, 2011 @ 3:27 pm
Although this study is interesting, studies connecting diet and exercise with Alzheimers and dementia prevention are much more relevant if you have a family history of the disease or are serious about trying to reduce your risk.
@Mike
There are benefits associated with constantly learning new things as you age, so I think language learning can fall under that as well. I think the bilingualism advantage would only show up if you actually make constant use of your new language, like having lots of friends who speak it, marrying someone who speaks it, or moving to a country where it is spoken.
I don't think I get the type of interference between languages that the study is talking about (or maybe my executive control system is great at keeping them separate?). My mom tells me that I started speaking in full sentences very young, and that I never mixed English and Spanish the way my brother did and many bilingual kids do.
I sometimes wonder whether I count as a true bilingual. I was raised bilingual, but I only occasionally speak in Spanish with relatives who don't know English. My parents speak to me in Spanish, but I've been responding in English since elementary school. It's definitely weaker than my English, though I have acquired a larger vocabulary than many heritage speakers thanks to reading lots of Spanish novels in high school. One thing I find interesting is that I can translate directly from Japanese to Spanish (or vice versa) without going through English first.
Rubrick said,
June 1, 2011 @ 3:47 pm
@C wrote:
"studies connecting diet and exercise with Alzheimers and dementia prevention are much more relevant"
Not if one believes the recent, rather well-publicized article concluding, essentially, that such studies have failed to turn up anything convincing.
I think that's why a number of commenters here have shown a lot of skepticism: a 5-year difference in symptom onset would be a gigantic effect if it stands up.
Theo Vosse said,
June 1, 2011 @ 3:58 pm
@myl: That's not the kind of model I was thinking of. I might have been too brief. I was thinking more in the line of my (now abandoned) research, where a processing model meant to work properly shows degradation in performance in certain conditions as an emergent property. My own work was centered on syntactic processing, and showed how a dynamic system that builds syntactic representations can break down, e.g. as a result of heavy center embedding or lexical ambiguity in the wrong spot. But that would relate to linguistic phenomena.
For an explanation of biological phenomena, I would want to see a (computational) model of the physical network, and how the biological properties of its constituting elements change over time as a function of the network activity. So, such a model would probably need to be very detailed. If such a model could explain normal everyday functioning of our brain, and as a side effect show difference in long-term effects on properties related to Alzheimer as a function of cognitive load, then I would take it much more seriously. However, such a model is far beyond our grasp. We cannot even model one neuron, and we certainly do not know how to implement something elusive as an "executive function" in such a network, let alone how to model the difference between the executive functions of a monolingual and a bilingual.
I hope this is a bit clearer…
tal said,
June 1, 2011 @ 5:00 pm
Mark, out of curiosity, why don't you think it's plausible the direction of causation is reversed? I'm no (psycho)linguist, but it seems like pretty much any time you're selecting adults based on self-reported (or even directly measured) bilingualism, all sorts of confounds are likely to creep in. That's certainly true in the Alzheimer's study, where much of the bilingual sample appears to be composed of Ashkenazi Jewish immigrants, who differ from native Canadian monolingual speakers in all sorts of relevant ways. But it also seems likely to be true of college samples. To get to age 18 and display fluency in two languages you not only had to have been exposed to two or more languages growing up (which implies a high likelihood that your parents or grandparents were immigrants, that your house was multicultural, etc.), but you have to maintain fluency in both. I realize that pretty much every child placed in a bilingual household where both languages are spoken all the time will learn both languages, but it seems like there's plenty of room for preexisting difference in executive control to play a large role in borderline cases.
For instance, (and I recognize this is anecdotal) I have many friends raised the same nominal language environment (parents both immigrants who spoke their native language at home at least some of the time) who vary wildly in their fluency at the second language. It seems like the kind of person who makes an effort to engage with a language that isn't spoken very often around them (or picks it up more rapidly, or seeks out opportunities to practice it later in life) might differ in substantive ways from one who doesn't. Is that implausible? All it would really takes is a few such cases for a pretty large selection bias to emerge and skew the results…
Ken Brown said,
June 1, 2011 @ 5:41 pm
@tal said [true bilinguism implies] "… a high likelihood that your parents or grandparents were immigrants, that your house was multicultural, etc…."
Maybe where you or I live, yes. But in a lot of places bilingualism is the normal condition. Would we expect people from, say, Nairobi or Asuncion to be less prone to Alzheimers than otherwise similar people from cities where most people only have one language?
It works. Let’s pull the funding | G7Finance.com - Finance News & Personal Finance Resources said,
June 1, 2011 @ 6:14 pm
[…] a busy day, so only time to post Lila Gleitman's sigh of an e-mail to Language Log: I just looked at the front page of the NYT (on line) and found two ill-assorted […]
YankeeTranslator said,
June 2, 2011 @ 12:35 pm
Question: Does being tri- or quadrilingual bring significantly more benefits than just being bilingual?
Lou Hevly said,
June 2, 2011 @ 2:40 pm
An interesting field of study might be to determine if Catalans and Basques, who are nearly all bilingual from birth, would suffer the onset of Alzheimer's more slowly than monolingual Spaniards. Also, if the difference between the two languages were a factor, that is, if the brain's executive control system worked "harder" for languages that were dissimilar, would the Basques have an advantage over both Catalans and Spaniards? Finally, to determine if there is a difference between bilinguals-from-infancy and bilinguals who acquired their bilingualism as adults, these results could then be compared with those of Spaniards who had migrated to either the Basque Country or Catalonia and become fluent in these languages as adults.
Theo Vosse said,
June 2, 2011 @ 2:57 pm
@Lou: if could also be the other way around. Since Catalan and Spanish are very similar, it could be harder for Catalans to separate them. No predictions there…
Lou Hevly said,
June 2, 2011 @ 4:23 pm
@Theo: What you say may well be true, but my prediction would still be that the brain's executive control system of Spaniards who became bilingual in Basque as adults would be the strongest.
Mar Rojo said,
June 2, 2011 @ 5:24 pm
Not true, as far as the Basques are concerned.
Mar Rojo said,
June 2, 2011 @ 5:27 pm
"who are nearly all bilingual from birth"
Not true, as far as the Basques are concerned.
Diana said,
June 2, 2011 @ 6:01 pm
merf, what happened to my post!?
Edward Neale said,
June 3, 2011 @ 6:18 am
This bilingual advantage stuff is very interesting, but the headline claim that bilingualism delays the onset of Alzheimer's by 5-6 years (I seem to remember this originally being reported as 4 years?) is misleading, I think. The data showing that bilingual populations report having the disease later than monolinguals seems convincing. There are a couple of questions I'd have about this though:
1) How is this Alzheimer's being diagnosed? I'm guessing the tests are designed for monolingual speakers originally, so it's possible the bilinguals are just slipping through the diagnostic net, perhaps not because their brains are healthier – it could be that they're just better at answering the questions in the diagnostic tests. (I know nothing about how Alzheimer's is actually diagnosed, so I'm probably way off here.)
It's interesting that Bialystok says:
I think that's probably right, but that raises my second question:
2) What are we talking about when we're talking about Alzheimer's? There's some dissent about whether it's reasonable to use the term Alzheimer's as if it designates a specific, single disease; there's a book by Peter Whitehouse and Daniel George called The Myth of Alzheimers that claims Alzheimer's is really a collection of distinct problems. There's a transcript of Whitehouse in discussion here. Perhaps bilingualism protects against some types of dementia, but not others? Or just some types of diagnosis?
It's this part of the interview that sounds overly hubristic:
The executive control system explanation sounds a little too simple. There's clearly some interesting work to be done to get a good handle on the mechanisms involved. I look forward to hearing about some of the extensions to this work that people are suggesting above.
Theo Vosse said,
June 3, 2011 @ 7:35 am
@Edward Neale: the explanation is not a little, but way too simple. To my (limited) knowledge, Alzheimer involves physical alterations, reportedly involving proteins taking up interneuronal space, throughout the brain. Now, it is assumed, not known for sure, that this is what causes cognitive degradation. I can live with that. If the environment of a neuron changes, it likely has an effect on its functioning.
The other way around though, has no simple physical explanation. Why would a stronger developed executive function (where function usually is taken to mean: localized part of the brain) slow down this protein build-up? In rest state, neurons are firing anyway, and the executive function is also at work in monolinguals. The extra work required for bilinguals would be pretty small. Consequently, the difference in total brain activity between monolinguals and bilinguals is very small. That just doesn't add up.
I don't think this is going to give anyone a handle into Alzheimer research.
Theo Vosse said,
June 3, 2011 @ 7:39 am
@Lou Hevly: there are theories that say the executive control (if such a thing truly exists) works as a big switch. In that case, there is no prediction for differences between Basques and Catalans.
You seem to assume that a larger distance between languages requires more effort. That would predict Basques getting Alzheimer later than Catalans or Gallegos.
On the other hand, it's much easier to differentiate between a Spanish and a Basque word, than between a Spanish and a Catalan word. So, if the executive function kicks in to distinguish competing representations, it would have to work harder for Catalans, thus reversing the prediction.
Norwegian Guy said,
June 3, 2011 @ 3:11 pm
Comparing immigrants and non-immigrants brings in extra uncertainties, and should be unnecessary in Canada, where there must be plenty of bilingual non-immigrants. Why not rather compare people bilingual in English and French (there must be some among Anglophone Quebecers and Franco-Ontarians) with monolingual Anglophones and Francophones?
And for the conclusion to be valid, the Canadian research should be reproducible in other countries as well. For instance, is there less Alzheimer's in the Gaeltacht, or among Welsh, Frisian and Saami speakers?
Norwegian Guy said,
June 3, 2011 @ 3:33 pm
@Theo Vosse: What about people who are bilingual in two varieties of the same language? Code switching between a dialect and some standard variety is not uncommon in many countries. And you even have the intermediate case of being bilingual in two mutually intelligible languages, like Norwegian and Swedish.
Theo Vosse said,
June 4, 2011 @ 8:02 am
@Norwegian Guy (pining for the fjords?): Catalan and Spanish have about the same relation as Norwegian, Danish and Swedish (many words with identical roots, slightly different pronunciation, differences in inflections, idiom, etc.). If you want to run such an experiment in Scandinavia, you might want to include the Swedish population in Finland…
army1987 said,
June 4, 2011 @ 8:44 am
In my experience (I grew up in central Italy with parents from southern Italy who almost exclusively speak Neapolitan at home, though I seldom speak it myself; I studied French in school for three years by the end of which I was somewhat fluent, but I've since almost completely forgotten it; I've just come back from a one-year exchange in Dublin and now I have almost native-like fluency in English — though Irish people can still immediately tell I'm a foreigner from my accent, and I have a smattering of Spanish and Irish), the more closely related two languages are, the less cognitive effort I need to switch between the two. But I'm not sure this can be extrapolated all the way to the case of two completely unrelated languages such as Castilian and Basque: it could well be that there is a degree of separation between languages at which switching between them is hardest, and it gets easier again for even more distant languages.
Monty said,
June 4, 2011 @ 9:37 am
Related, here's a podcast on bilingualism which includes commentary from three participants in the original AAAS syposium.
Antonella Sorace said,
June 6, 2011 @ 12:15 pm
It is true that we need more research on the cognitive benefits of bilingualism (on early vs late bilinguals, on different language combinations, on the causality of the bilingualism-executive function relationship) but it seems to me that the research findings on children are robust enough to be disseminated among families, teachers, health professionals and policy makers. Given how many misconceptions there still are, persuading people that bilingualism doesn't have any negative effects already is a great achievement; persuading them that bilingualism in any two languages may actually be beneficial is very positive for bilingual children, their families, and society in general. In Edinburgh we have set up BILINGUALISM MATTERS (www.bilingualism-matters.org.uk) precisely to bridge the gap between research and the community. There is a great demand for information and people of all backgrounds respond very well, including some politicians. We are setting up branches of BM in Scotland (Western Isles) and in Europe (Norway and Greece). Definitely worth doing!