Language loss in bilingual speakers with Alzheimer's disease

Abstract

This study investigated the changes in language and cognition in five bilingual speakers with Alzheimer's Disease over a period of twelve months. The pattern and rate of loss in English was compared to that of Afrikaans. The bilingual behaviour of language mixing was also investigated, as was the interaction between deteriorating cognitive skills and language functions. Data was collected at three time points (0 - 6 - 12 months) employing a battery of neuropsychological and language tests, and conversation analysis. It was predicted that where both languages were automatised to a similar extent, a similar pattern, severity and rate of loss would be evident across languages. This hypothesis was supported by results. It was also predicted that in cases where one language was less automatised than the other, the less automatised language (i.e. the language learnt later in life (L2) anchor the less proficient language) would be more severely impaired and would deteriorate at a faster rate than the fully automatised language (Li). Results revealed that while L2 was more impaired than Li for some speakers, for others, languages were similarly impaired/spared. These discrepancies were attributed to the fact that tests were not sensitive to inter-language differences near floor or ceiling. Results did not strongly support the second prediction that L2 would deteriorate at a faster rate. Ambiguous findings could be artefacts of the time window of examination, insensitive assessment tasks, and the heterogeneous nature of the population. With regards to language mixing behaviour, code switching mainly affected L2 interactions even though the extent of switching varied across speakers. The amount of language mixing increased for two participants over the year. With regards to a possible interaction between language and cognition, complex language tasks appeared to be more compromised by deteriorating neuropsychological support than less complex tasks, but the extent of this interaction varied across languages and across speakers. Finally, the overall profile of results suggested that a language learnt later in life will never become fully automatised, even if high levels of L2 proficiency had been attained in adulthood.