Date of Degree


Document Type


Degree Name



Speech-Language-Hearing Sciences


Loraine K. Obler

Committee Members

Mira Goral

Lisa Edmonds

Subject Categories

Communication Sciences and Disorders | Speech Pathology and Audiology


Aphasia, Multilingual, Interference, Attrition, VNeST, Treatment


Background: Successfully assessing and treating aphasia in multilingual people requires a detailed understanding of the mechanisms underlying language processing in the multilingual population, and the potential impairments to those mechanisms. The balance between spreading activation of language processes via treatment and controlling interference of competing items within the lexico-semantic networks appears to be a key factor in determining whether treatment effects generalise within and across languages in multilingual people with aphasia (Kiran, Sandberg, Gray, Ascenso, & Kester, 2013).

This balance can be exploited through treatment, which, if carefully chosen, should maximise potential within- and cross-language generalisation. One treatment that has been shown to consistently result in within-language generalisation, to varying degrees, is Verb Network Strengthening Treatment (VNeST), in which thematic role assignment to given verbs is repeatedly trained, thus strengthening semantic verb networks (Edmonds, 2016). Due to the shared semantic network across languages of multilingual people (e.g., Paradis, 1993), VNeST should result in generalisation across languages of multilingual individuals with aphasia, in addition to within-language generalisation, especially when trained verbs share argument structure across languages, and when basic sentence structure is similar across languages. To date, conflicting evidence has been observed in multilingual individuals with aphasia regarding within-language and cross-language generalisation relative to the underlying and competing mechanisms of spreading activation and interference control.

Aim: We investigated under which conditions generalisation is likely to occur in multilingual individuals with aphasia, using a treatment with high potential for generalisation (VNeST), in a language pair with overlapping basic word order and mostly overlapping verb argument structure. Furthermore, we investigated whether any treatment gains were maintained after treatment was discontinued.

Method: Three multilingual participants with aphasia whose first-acquired language was English, and who all acquired Modern Hebrew in elementary school and reached moderate-high pre-stroke proficiency in adulthood, participated in this study. All participants received VNeST in each of their languages, in consecutive treatment blocks. English and Hebrew abilities were tested before and after each treatment block, and 4-5 weeks after treatment was discontinued, using a large battery of language tests that included comprehension and production tasks for single-words, sentences, oral connected speech and written narratives. Functional communication skills in each language were also assessed via questionnaire.

Results: We found that direct treatment effects were measured in both languages, for all participants with moderate-severe aphasia in any given language, but not in mild aphasia. Within-language generalisation was also observed for all participants, but not equally for both languages. Rather, the amount and type of generalisation was qualified by order of acquisition, relative proficiencies, attrition, aphasia type and severity, and motivational factors. Cross-language generalisation was observed in each participant in one direction only, with contradictory patterns across participants. For two participants with pre-stroke high proficiency in both languages, we found support for the strong suppression of interference in the less impaired English during treatment of the more impaired Hebrew, resulting in either no cross-language generalisation to English, or a decrease in post-treatment English language performance, which we attribute to the involvement of damage to the language control network (Ansaldo & Saidi, 2014). Conversely, in the same two participants, cross-language generalisation was observed in the more impaired Hebrew after treatment in the less impaired English, likely due to a weak suppression of interference of the more impaired Hebrew, and therefore a stronger effect of spreading activation from treatment in English (Kiran et al., 2013). We observed the opposite pattern in a participant whose attrited Hebrew had never reached full proficiency pre-stroke, with treatment in his more impaired Hebrew demonstrating cross-language generalisation to his less impaired English. We attribute this to strong spreading activation of an attrited language, both generally through exposure as well as specifically through treatment. Conversely, a decrease in performance in the more impaired Hebrew after treatment in the less impaired English was attribute to rarely using Hebrew in the environment once treatment in English began, together with fluctuating motivation. Treatment gains began to decline for all participants after treatment was discontinued, with the most widespread decline in the least communicative participant, in his rarely-used language (Hebrew).

Conclusion: Our study supports the competing mechanisms theory of Kiran et al. (2013), relative to factors such as order of acquisition, damage to the language control network, language of the environment, attrition, and motivational factors. Clinically, we found that VNeST is a valuable treatment option in multilingual participants with aphasia, resulting in direct treatment effects and within-language generalisation, including for a moderately proficient language that had undergone attrition for many years. Notably, we found that when treating a multilingual participant with aphasia in one language only, not only can cross-language generalisation occur or not occur, but treatment in one language can also result in a decrease in performance in the untreated language, especially if (a) the language control network is damaged, and (b) treatment is provided to the more impaired language only. Therefore, carefully monitoring language gains and losses throughout treatment is essential, in order to modify treatment plans as therapy progresses. Finally, it is necessary to consider a low dosage maintenance treatment plan relative to participants’ language and communicative environment, so that treatment gains can be appropriately maintained allowing multilingual patients with aphasia to maximise their potential in each language.