|Background: Globally, there are 9.9 million new cases of dementia each year, meaning one new case is diagnosed every 3.2 seconds (Prince et al., 2015). The risk of Alzheimer’s disease (AD) increases with increasing age, and so with the rising aging population, the prevalence of AD is projected to increase in the years to come. Cognitive resilience has been a focus of preventative strategies against AD/dementia as cognitive resilience is hypothesized to delay or prevent the onset of clinical symptoms of dementia despite the presence of Alzheimer neuropathology (see review by Stern, 2002). Cognitive resilience has two components: cognitive reserve refers to an ‘active’ process of using neural networks efficiently to compensate for brain damage, and brain reserve relies on structural advantages within the brain that increase the capacity to tolerate brain damage. Multilingualism (i.e., speaking more than one language) may contribute to cognitive resilience against dementia/AD, as it can improve cognitive flexibility and executive function through constant switching between languages and the use of inhibition and attention control.
Objectives: The first two objectives of this study were to investigate whether there was an association of cognitive resilience with (1) multilingualism or (2) type of language. The second two objectives of this study were to explore whether (3) cortical atrophy (an indicator of brain reserve) or (4) education (an indicator of cognitive reserve) modified the association of cognitive resilience with multilingualism or type of language.
Methods: Data were used from the Nun Study, a longitudinal study on aging with religious sisters, aged 75+ at baseline. Multilingualism was determined through convent archival records. Neuropathological diagnosis of AD was based on both Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) and the National Institute of Neurological Disorders and Stroke-Alzheimer’s Disease and Related Disorders Association (NIA-RI) criteria. Dementia status was assessed by the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) criteria at the last cognitive assessment before death. Analyses included logistic regression models adjusted for the presence of apolipoprotein E-ε4 allele, age at death, education, primary language, and immigration status. Firth regression was used to test for interactions between the exposure variables and cognitive resilience and for any subsequent stratified models.
Results: Speaking four or more languages (versus speaking fewer languages) was significantly associated with cognitive resilience based on NIA-RI criteria only, and remained so after adjusting for all of the covariates (OR=5.00, 95% CI=1.08-27.56). Speaking German significantly reduced the likelihood of cognitive resilience based on CERAD criteria only, even after adjusting for all covariates (OR=0.50, 95% CI=0.27-0.94). With respect to ceiling and floor effects, adjusting for education or atrophy in multivariable models did not cause any substantial changes to the association between multilingualism and cognitive resilience. When using type of language as an exposure variable, adjusting for education did not substantially change the association between type of language and cognitive resilience; however, when adjusted for atrophy the statistically significant association observed between speaking German and cognitive resilience became nonsignificant (OR=0.55, 95% CI=0.27-1.09). Lastly, no statistically significant associations were observed between multilingualism or type of language and cognitive resilience when stratified by education or the presence of atrophy.
Discussion: Overall, speaking four or more languages significantly increased the likelihood of cognitive resilience. However, no evidence of a benefit of bilingualism on cognitive resilience was observed in this study. Previous research indicates that only the studies that were cross-sectional and had used a clinic-based sample observed a bilingual benefit, while studies that were longitudinal and had used a population-based sample found a protective effect against dementia, only when higher number of languages were spoken. Since this current study was longitudinal and had used a population-based sample, the findings were consistent with the previous studies that did not observe a bilingual benefit against dementia. Other than the methodological differences, this current study perhaps did not observe a bilingual protective effect because of the measure of multilingualism used, which was less stringent than in some previous studies. For example, the participants in this study only had to self-report the number and the type of languages they were proficient in and thus, this study could not incorporate other important aspects of multilingualism, such as reading ability, language comprehension, frequency and intensity of language use, and age at language acquisition, in the assessment of multilingualism. However, this current study did use a stringent measure of the outcome, cognitive resilience, based on both the dementia status at the last cognitive assessment and the presence of Alzheimer neuropathology. Moreover, after controlling for various confounders, such as the presence of apolipoprotein E-ε4 allele, age at death, education, primary language, and occupation, a significant benefit from speaking four or more languages on cognitive resilience was still observed.
Secondly, previous research has shown mixed findings on the cognitive benefits of linguistically similar or linguistically dissimilar languages, where there have been studies that observed cognitive benefits from speaking either similar or dissimilar languages or no significant cognitive benefits at all. Because the outcomes of most previous studies were on cognitive benefits rather than on dementia/AD/cognitive resilience, the findings from previous studies cannot be extrapolated fully to the findings from the current study. Although speaking similar languages has been found to benefit cognition presumably because of greater levels of cross-linguistic interference experienced while speaking, in this current study, speaking similar languages such as German and English did not significantly contribute to cognitive resilience. While speaking German and English was initially shown to significantly reduce the likelihood of cognitive resilience, since German speakers in this study had higher levels of atrophy present than those who spoke other type of languages, it is likely that the presence of atrophy instead of speaking the German language itself was responsible for reducing the likelihood of cognitive resilience.
Finally, when the models were adjusted for atrophy or education to explore the floor and ceiling effects, respectively, no substantial changes were observed between multilingualism and cognitive resilience. Between the type of language spoken and cognitive resilience, only the significant association between speaking German and cognitive resilience became nonsignificant, as a higher number of German speakers had atrophy present. As there were no significant interactions between the exposure variables and atrophy or education and no significant trends supporting a ceiling or floor effect were observed in the stratified analyses, this current study did not provide supportive evidence for the presence of a ceiling or floor effect for cognitive resilience.
Overall, speaking four or more languages was shown to increase the likelihood of cognitive resilience, whereas there was no evidence of benefit of speaking two or three languages on cognitive resilience. Since those who spoke four or more languages were likely to have experienced higher levels of intellectual stimulation than those who spoke fewer languages, the results may indicate that a higher degree of intellectual stimulation is the underlying factor that is required to significantly build up cognitive resilience. It is also possible that those who spoke four or more languages were different in the sense that they had higher levels of intelligence or had a greater tendency to engage in intellectually stimulating activities than those who spoke fewer languages. Moreover, the context within which the languages were spoken might have influenced the results. In addition, this research project did not find any evidence on type of language being important for contributing to cognitive resilience; this topic has not often been studied and warrants further investigation. Finally, multilingualism can be considered as one modifiable way to enhance cognitive stimulation in order to build up cognitive resilience and thus reduce the impact of AD, thereby improving the quality of life of aging populations.