Interrupted Time Series Analysis of the Efficacy of Bilingual Abstract Semantic Associative Network Training

In response to the need for more appropriate anomia treatment materials for bilingual persons with aphasia (BPWAs), we have created linguistically and culturally appropriate versions of AbSANT (Abstract Semantic Associative Network Training) for several languages that we call BAbSANT (Bilingual Abstract Semantic Associative Network Training). BAbSANT has now been conducted with 15 BPWAs in six different language combinations. The purposes of this study were to determine whether BAbSANT produces (a) similar within-language treatment effects as AbSANT and (b) cross-language treatment effects, and to determine whether language dominance affects (a) within- and (b) cross-language generalization.

We employed interrupted time series models to analyze all available (published and unpublished) BAbSANT data.

Model 1 replicated existing work showing both direct training effects and generalization to concrete words when abstract words are trained, regardless of the language trained. Model 2 revealed robust cross-language generalization effects, regardless of the language trained. Models 3a and 3b revealed differential effects of language dominance on within- and cross-language generalization patterns.

These results suggest that BAbSANT is a linguistically and culturally appropriate word retrieval therapy that is effective in improving word retrieval not only for directly trained items but also for related words in the same language and in the untrained language. Thus, BAbSANT is a viable and robust therapy for bilingual persons with aphasia.

https://doi.org/10.23641/asha.34189701.