There is an emerging literature suggesting that speaking two or more languages may significantly delay the onset of dementia. Although the mechanisms are unknown, it has been suggested that these may involve cognitive reserve, a concept that has been associated with factors such as higher levels of education, occupational status, social networks, and physical exercise. In the case of bilingualism, cognitive reserve may involve reorganization and strengthening of neural networks that enhance executive control. We review evidence for protective effects of bilingualism from a multicultural perspective involving studies in Toronto and Montreal, Canada, and Hyderabad, India. Reports from Toronto and Hyderabad showed a significant effect of speaking two or more languages in delaying onset of Alzheimer’s disease by up to 5 years, whereas the Montreal study showed a significant protective effect of speaking at least four languages and a protective effect of speaking at least two languages in immigrants. Although there were differences in results across studies, a common theme was the significant effect of language use history as one of the factors in determining the onset of Alzheimer’s disease. Moreover, the Hyderabad study extended the findings to frontotemporal dementia and vascular dementia.
Recent studies from Canada [
Although the mechanisms underlying possible protective effects of bilingualism on dementia onset are unknown, these may relate to cognitive reserve [
In the case of bilingualism, cognitive reserve may arise from reorganization and strengthening of neural networks due to enhanced executive control. There is substantial evidence that language processing in bilinguals involves ongoing activation of both spoken languages. Thus, there is a need for management of attention to the two activated languages, continuous monitoring to determine which language is appropriate, and rapid switching between languages in response to changes in the environment [
Evidence for enhanced executive control in bilinguals comes from studies in children and adults (see Bialystok et al. [
The initial study designed to test the concept that bilingualism may delay onset of dementia was carried out at Baycrest Centre for Geriatric Care in Toronto, Canada [
In Study 1, two hundred twenty-eight consecutive charts of patients from the Sam and Ida Ross Memory Clinic were reviewed [
There were speakers of 25 different first languages in Study 1 and 21 different languages in Study 2. The most common were Polish, Yiddish, Hungarian, German, and Romanian (Study 1) and Polish, Yiddish, Hungarian, Italian, and French (Study 2).
Both studies showed a significant delay in onset of symptoms of dementia in bilinguals (Table
Means (and standard deviations) for background measures and age of onset of symptoms of dementia from the two Toronto studies.
Group |
|
Age at first appointmenta | Years of education | MMSEb | Age at onsetc |
---|---|---|---|---|---|
Study 1: [ | |||||
Monolingual | 91 | 75.4 (9.3) | 12.4 (3.8) | 21.3 (6.4) | 71.4 (9.6) |
Bilingual | 93 | 78.6 (8.4) | 10.8 (4.2)1 | 20.1 (7.1) | 75.5 (8.5)2 |
|
|||||
Study 2: [ | |||||
Monolingual | 109 | 76.5 (10.0) | 12.6 (4.1) | 21.5 (5.7) | 72.6 (10.0) |
Bilingual | 102 | 80.8 (7.7) | 10.63 (5.1) | 20.4 (5.6) | 77.7 (7.9)4 |
bMMSE: Mini-Mental State Examination (first appointment); maximum score = 30.
cAge at which symptoms were first reported by family, year.
The proportion of immigrants differed between bilinguals and monolinguals in both studies. Whereas most bilinguals were immigrants, the minority of monolinguals were immigrants. However, immigration status does not appear to account for the results. In Study 1, analysis of immigrants separately showed a significant delay of 11.5 years in onset of dementia in bilinguals compared to monolinguals. In Study 2, controlling for immigration status in the analysis did not change the results.
There was no difference in severity of dementia between monolinguals and bilinguals at initial visit based on MMSE scores. Moreover, rate of cognitive decline was assessed in a subgroup of patients in Study 1 who had follow-up MMSE scores over 4 years, that is, 24 bilinguals and 25 monolinguals, and these rates were also equivalent for monolingual and bilingual patients. Finally, there was no suggestion that the delay in onset of dementia in bilinguals was an artefact of their waiting longer than monolinguals before consulting a physician. In Study 2, the time interval between first symptoms and first clinic visit was significantly shorter in bilinguals compared to monolinguals. In Study 1, the interval was also shorter for bilinguals although the difference was only marginally significant (
In addition to the above studies suggesting that bilingualism delays the onset of dementia by up to five years, there are two Toronto neuroimaging studies supporting a protective effect of bilingualism on brain function. Schweizer et al. [
Chertkow et al. [
Age at diagnosis was available in all cases and served as the main outcome measure (Table
Age of diagnosis of Alzheimer’s disease.
Number of languages spoken | Native English | Native French | Immigrants |
---|---|---|---|
1 | 78.0 (7.0) | 72.7 (9.1) | 71.4 (8.1) |
( |
(289) | (66) | (23) |
|
|||
2 | 77.9 (7.5) | 75.9 (6.5) | 76.5 (8.2) |
( |
(62) | (24) | (81) |
|
|||
3 | 79.8 (5.6) | 79.5 (2.5) | 77.8 (6.4) |
( |
(24) | (4) | (39) |
|
|||
≥4 | 80.7 (3.2) | — | 80.9 (5.9) |
( |
(3) | — | (15) |
Adapted with permission from Lippincott Williams and Wilkins/Wolters Kluwer Health: Alzheimer’s Disease and Associated Disorders, [
Three levels of language ability were defined: monolingual, bilingual, and multilingual. Multilingualism was defined differently based on the language groups being compared. For comparison of those speaking one versus multiple languages, multilingualism was defined as speaking two or more languages. For analyses of effects of increasing numbers of languages, bilingualism was defined as speaking two languages, whereas multilingualism was defined as speaking three or more languages. When examining bilingualism in the nonimmigrant population, only those who spoke both French and English since youth were considered. The monolingual cohort was made up of only English or French speakers.
There was no direct information of immigrant/native status. Therefore, immigrants were defined as individuals whose first language was not native to Canada, that is, neither English nor French. In contrast, individuals whose first language was English or French were considered native-born. Bilinguals and multilinguals were defined according to the criterion set out by Bialystok and colleagues for bilingualism, that is, those who spent the majority of their lives, at least from early adulthood, regularly using at least two languages [
There were 253 multilinguals and 379 monolinguals. Twenty-five first languages were spoken. The most common were English, Polish, French, Yiddish, and Hungarian.
There was no significant effect of language status (monolingual versus multilingual) overall on age of dementia diagnosis. However, there was an impact of number of languages spoken when assessed by regression analysis. Considering all language groups, there was a significant positive relation between number of languages spoken and delay in diagnosis of dementia. Education or sex did not account for the findings. Analysis of number of languages spoken showed that those who spoke four or more were diagnosed at a significantly older age than those who spoke one or two languages. There was also a trend for those who spoke three languages to be diagnosed later than those who spoke one or two languages. There was no difference between those who spoke one or two languages. These results suggest that speaking four or more languages is protective, speaking three languages has marginal benefit, and speaking two languages has no benefit. In contrast to the protective effect of speaking more than two languages, being an immigrant led to earlier onset of dementia and thus had a negative impact. However, within the immigrant group, bilingualism did in fact have a protective effect in delaying onset of dementia.
One motivation for the study was to examine the effect of bilingualism without the potential confound of differences in cultural and life experience between individuals born in Canada and immigrants. Thus an additional analysis that was restricted to Canadian born subjects was carried out in English and French monolinguals compared with French/English bilinguals. There were 356 monolinguals (290 English-speaking, 66 French-speaking) and 43 bilinguals (19 with English as first language and 24 with French as first language). Therefore, the monolinguals were 81% English speaking compared with only 19% French speaking, and in the whole sample, there were 89% monolinguals and 11% bilinguals, so the results must be interpreted with caution. In this group, onset of dementia was significantly earlier in bilinguals as compared to monolinguals. Thus being bilingual appeared to have an adverse effect on age of diagnosis in individuals born in Canada.
Further analysis showed that among monolinguals, French speakers were diagnosed 5.3 years earlier than English speakers, while in bilinguals there was no difference between language groups. In the larger native English and French group, including speakers of more than two languages and bilinguals who spoke additional languages other than English or French, there was no difference in age of diagnosis for the native English group based on number of languages spoken when controlling for education and gender, while in the native French group, there was a trend towards significance (
The immigrant group was examined separately to see if there was a similar pattern compared to the nonimmigrant group. Monolinguals were diagnosed 5 years earlier than bilinguals, 6.4 years earlier than trilinguals, and 9.5 years earlier than those speaking four or more languages. Also, there was a significant difference between multilingual speakers of four or more languages and bilinguals, with multilinguals being diagnosed 4.5 years later on average. In the immigrant subgroup, there was an impact of number of languages spoken, both at the level of bilingualism and at the level of four or more languages. However, monolinguals had significantly less education than all other language groups, a factor that could have contributed to earlier onset of dementia.
To examine the patterns uncovered by separating the cohort into Canadian born subjects whose first language was English, Canadian born subjects whose first language was French, and immigrants, the difference in age of diagnosis between these groups within each linguistic group was analyzed. In the monolingual group, there was a significant difference in age of diagnosis. Canadian born subjects whose first language was English were diagnosed significantly later than Canadian born subjects whose first language was French (5.4 years later) and immigrants (6.6 years later). There was no significant difference between Canadian born subjects whose first language was French and immigrants. In the bilingual and multilingual (three or more languages) groups, there was no significant difference between the three groups. This suggests that native English speaking monolinguals are diagnosed later than either French-speaking native Canadians or immigrant monolinguals. However, this difference disappears in bilinguals and multilinguals. One interpretation of this was that there was a protective effect of multiple languages which emerged much more strikingly in the French-speaking and immigrant populations, but that monolingual English speaking individuals were already “protected” in some way. An alternative explanation is that lack of significance was due to small sample size which was 24 in native Canadians whose first language was French and who spoke two languages and four in those who spoke three languages. None spoke more than three languages. Only 27 native Canadians whose first language was English spoke more than two languages. Fifty-four immigrants spoke more than two languages.
Higher occupation status and more intellectually stimulating work are associated with retained cognitive function in old age [
India offers an appropriate environment to study the association between bilingualism and age of dementia onset due to its high degree of linguistic diversity [
Several epidemiological studies from India demonstrate a high burden of dementia and cognitive disorders [
In Hyderabad, Telugu is spoken by the majority group who are primarily Hindus, whereas the language of a minority group of Muslims is Dakkhini Urdu, a variety of Hindi spoken in the Deccan plateau that includes Hyderabad. As in other parts of India, in the state of Andhra Pradesh (for which Hyderabad is the capital) English is gradually acquiring more and more functional roles in education, administration, and mass media. In addition, Hindi is spoken as the official national language and is taught as a subject at school level. Thus, most people in Hyderabad are exposed to Telugu and Urdu in informal contexts and Hindi and English in formal contexts. A study exploring patterns of multilingualism in Hyderabad suggested that no single language catered to the needs of both Telugu and Urdu speakers’ day-to-day life in Hyderabad. A language-use history questionnaire study on Telugu and Dakkhini Urdu speakers indicated that while the former group manage most of their communication needs using Telugu and English (and are functionally bilingual) the latter group are trilingual in that they tend to use Dakkhini Urdu, Telugu, and English in different situations and for different purposes [
In the Hyderabad Memory Clinic study, case records of 648 consecutive patients in the dementia registry were reviewed for age, gender, age of onset of dementia, education, and age when the diagnosis was made [
Four-hundred twenty-four of the 648 consecutive dementia cases were men (65.4%). The mean age of the group at presentation was 66.2 years (range 32–92) and duration of illness ranged from 6 months to 11 years (mean 2.3; SD 1.8). AD was diagnosed in 240 (37.0%), VaD in 189 (29.2%), FTD in 116 (17.9%), DLB in 55 (8.5%), and mixed AD with cerebrovascular disease in 48 (7.4%). Three hundred ninety-one cases (60.3%) were bilingual [
The age of onset of dementia among bilinguals was 4.5 years later than in monolinguals [
Demographic measures and age at onset of symptoms of dementia and its subtypes in Indian study.
Group |
|
Durationa | Years of education | Age at onset of dementiab | Age at onset of AD | Age at onset of FTD | Age at onset of VaD |
---|---|---|---|---|---|---|---|
Monolinguals | 257 | 2.1 (1.7) | 5.9 (5.1) | 61.1 (11.4) | 65.4 (10.0) | 55.6 (10.5) | 57.0 (10.7) |
|
|||||||
Bilinguals | 391 | 2.3 (1.9) | 12.9 (4.9)1 | 65.6 (10.0)2 | 68.6 (9.6)3 | 61.6 (9.0)4 | 60.7 (9.7)5 |
AD: Alzheimer's disease.
FTD: frontotemporal dementia.
VaD: vascular dementia.
aDuration of elapsed time between age at onset and age at first appointment.
bAge at which symptoms were first reported by family.
Exploratory analyses were carried out to determine whether speaking three or more languages conferred an added benefit over two languages. No added benefit was found.
The Toronto and Hyderabad studies showed a significant protective effect of speaking two or more languages in delaying onset of AD. In addition, the Hyderabad study extended this finding to FTD and VaD.
In contrast, the Montreal study failed to show an overall protective effect of bilingualism on delaying onset of dementia. However, consistent with the findings of Kavé et al. [
The differences among the studies may reflect the influence of several factors that may offer protection against dementia, including those that relate to culture, socioeconomic status, and linguistic relatedness between languages, demographics, and policies leading to differential status among languages. Differences in oral, literate, and metalinguistic dimensions of language acquisition and use [
The Toronto and Hyderabad studies defined onset of dementia based on age when symptoms first developed, whereas the Montreal study used age at diagnosis as the main outcome measure in the majority of statistical comparisons. However, a subset of Montreal cases was analysed using age of symptom onset as the outcome measure. There was a protective effect of multilingualism compared to bilingualism or monolingualism in this latter group but there was no effect of bilingualism compared to monolingualism. However, sample size was relatively small with only 54 cases speaking more than one language, a factor that might have led to negative results when comparing bilinguals to monolinguals.
There was a surprising finding in the Montreal study related to immigration status. Whereas bilingualism was protective in immigrants, this effect was not found in native-born Canadians who spoke English. In contrast, immigration status had no effect in the Toronto study. The conflicting findings might at least in part relate to the definition of immigrant versus nonimmigrant status in the Montreal study. As acknowledged by Chertkow et al. [
Although studies from Canada and India suggest a protective effect of bilingualism on dementia, caution must be observed in view of negative findings by others. Zahodne et al. [
Although the mechanisms by which bilingualism might delay the onset of dementia are unclear, it has been suggested that this relates to the development of cognitive reserve. Recent neuroimaging data support this concept. Abutalebi et al. [
In conclusion, multicultural studies from Canada and India suggest that there is a protective effect of bilingualism in delaying onset of dementia. In all three studies, the language history of patients had a significant association with onset of dementia but the details of that language history, as well as other contextual factors, mattered and produced different results. Thus, in some contexts bilingualism provided protection, whereas in the context of specific cultural and immigration factors, only multilingualism provided protection. It is clear that bilingualism alone is insufficient to guarantee the postponement of dementia. Future cross-cultural studies are needed to determine the contexts in which bilingualism offers these protective effects and the other factors with which it interacts in order to resolve this important issue.
The authors declare that there is no conflict of interests regarding the publication of this paper.
Morris Freedman receives support from the Saul A. Silverman Family Foundation as a Canada International Scientific Exchange Program and Morris Kerzner Memorial Fund. Toronto Study 1 was supported by a grant from Canadian Institutes of Health Research (CIHR) (Ellen Bialystok, Fergus I. M. Craik); Toronto Study 2 was supported by grants from CIHR (Ellen Bialystok, Fergus I. M. Craik), Alzheimer Society of Canada (Ellen Bialystok, Fergus I. M. Craik, Morris Freedman), and the SHaRna Foundation; the Montreal Study was supported by a grant from CIHR (Howard Chertkow); the Hyderabad study was supported by Cognitive Science Research Initiative, Department of Science and Technology (DST), Government of India. Shailaja Mekala, DST Research Fellow, assisted with data analysis. This work was presented in part at the World Federation of Neurology Research Group on Aphasia and Cognitive Disorders meeting in Hyderabad, India, December 2012.