Search

• Bilingual persons with aphasia produced more language mixing when telling narratives in their post-aphasia weaker L2 compared to their stronger L1. • During L1 narrative production, bilingual persons with aphasia inserted L2 nouns which increased communicative efficiency. • During L2 narrative production, bilingual persons with aphasia inserted L1 discourse markers and function words which improved fluency. • Language abilities, not cognitive control, predicted language mixing frequency in bilingual persons with aphasia. Language Mixing (LM) occurs among neurotypical bilinguals as well as among bilingual persons with aphasia (BiPWAs). The current study aimed to investigate whether LM in BiPWAs stems from a linguistic impairment, an impairment in cognitive control, or both. Twenty Russian-Hebrew-speaking BiPWAs were split into two groups based on aphasia severity (Severe/Moderate vs. Mild). Frequencies and patterns of LM in narrative production by BiPWAs in L1-Russian and in L2-Hebrew were analyzed. To investigate the underlying mechanisms of LM, all participants completed linguistic background questionnaires, the Bilingual Aphasia Test (BAT) in both languages, and a battery of 10 cognitive tests. The results indicated an effect of aphasia severity and an effect of language. Higher LM frequency was observed in BiPWAs with severe/moderate aphasia symptoms as compared to BiPWAs with mild symptoms. In both groups, higher LM frequency was observed in L2-Hebrew narratives, the weaker post-stroke language for most participants in the sample. The results also showed qualitative LM differences in L1-Russsian and L2-Hebrew contexts. In L1-Russian narratives, BiPWAs mainly switched to L2-Hebrew nouns, while in L2-Hebrew narratives, they mainly inserted L1-Russian discourse markers and function words. Linguistic factors such as pre- and post-stroke self-rated language proficiency and level of language impairment due to aphasia were found to predict LM frequency in L1-Russian and in L2-Hebrew. Cognitive abilities did not predict LM frequency. Based on our findings, we suggest that LM behavior in BiPWAs might be primarily related to language skills in L1 and L2, rather than to cognitive control impairments. [ABSTRACT FROM AUTHOR]