You have a new outpatient client who had a stroke a year ago and has aphasia. Mrs. R. is 63 years old and multilingual. She was born in Colombia, South America, and spoke Spanish as a child and young adult. She studied and learned to speak French in college and spent a summer in France. She moved to North America when she was 22 and married an American who speaks only English. In which language will you provide treatment? Would you provide treatment in Spanish, the first acquired language, or in English, the language learned later but the one the client has been speaking with her immediate family?

More speech-language pathologists face this sort of conundrum every day as they work with the country's growing multilingual population. The challenge here is that research to date has only begun to uncover the optimal rehabilitation for multilingual clients with aphasia (Gitterman, Goral, & Obler, 2012). But the literature is still sparse in terms of interpreting impairment and recovery in this group (Faroqi-Shah, Frymark, Mullen, & Wang, 2010; Lorenzen & Murray, 2008). What we do know from existing research on the influence of multilingualism on language processing can help us evaluate and treat multilingual individuals with aphasia.

Some key points about multilingual language processing have direct implications for aphasia services to multilingual people.

Constant Change

Research has convincingly demonstrated that language processing in a multilingual person is a dynamic process. Thus, learning a second language (L2) changes representations and access not only for that language but also for the first-acquired language (L1) (Cook, 2003; Goral, Libben, Obler, Jarema, & Ohayon, 2008; Linck, Kroll, & Sunderman, 2009; Witford & Titone, 2012). For instance, Linck and colleagues (Linck et al., 2009) demonstrated that immersion experience in L2 results in attenuation of L1, indicating that increased dominance in L2 may result in attrition or decreased accessibility to L1, at least in the short term.

Implications for aphasia: The dynamic nature of language dominance, including factors such as attrition and immersion, needs to be considered when evaluating impairment in the multiple languages of multilinguals and deciding which language to treat. It is reasonable to expect that Mrs. R., because she has been living in the United States for the past 40 years, may have experienced some attrition in her Spanish, and that English may have become her dominant language. In this scenario, it may be reasonable to administer treatment in English, which may be the client's stronger or more used language.

All Systems Switched On

All language systems are active during language processing tasks—during code-switching and translation, but also during word recognition and production and during sentence processing. Therefore, the non-target language is never actually switched off across tasks, even when the task requires only one language output (Costa, 2005; Kroll, Bobb, & Wodniecka, 2006; Marian, Spivey, & Hirsch, 2003). Language systems operate concurrently in highly proficient multilinguals and also in differentially proficient multilinguals (Christoffels, Firk, & Schiller, 2007; Costa & Santesteban, 2004; Herman, Bongarts, De Bot, & Schreuder, 1998). In these individuals, L1 must be inhibited to produce a less dominant language; therefore, there is a lag in the reactivation of L1 when switching from L2 to L1.

Implications for aphasia: Understanding that all languages are active during language processing tasks can greatly affect an SLP's approach. If target words and structures are active across languages, then training in one language should result in changes in all languages—that is, language treatment should facilitate cross-language generalization. Several recent treatment studies have demonstrated this general finding (Edmonds & Kiran, 2006; Goral, Levy, & Kastl, 2010; Kiran & Roberts, 2009; Laganaro & Overton Venet, 2001; Miertsch, Meisel, & Isel, 2009). For Mrs. R., if treatment can be provided in only one language—Spanish or English—the clinician can reasonably expect positive cross-language generalization.

However, if all languages are always active, a greater likelihood of interference among the languages may be more likely when the client is experiencing word-retrieval difficulties in one language. This premise may be particularly true when treatment is in the less dominant language—that is, when treatment requires inhibition of the more dominant language (Abutalebi, Rosa, Tettamanti, Green, & Cappa, 2009; Goral, 2012). In the example of Mrs. R., if one of her two main languages—English or Spanish—seems to be markedly more accessible to her, it may make sense to work in that language. Working in the language that is less accessible may interfere with performance in the more accessible language, at least for the duration of the treatment.

This premise would be true as well for Mrs. R.'s French (L3), her least proficient language. Therefore, if Mrs. R. demonstrates good recovery of her two stronger languages and expresses interest in recovering her abilities in her third language, it is reasonable to explore additional work in that language. However, treatment in French is less likely to advance her stronger English and Spanish.

Overlapping Neural Systems

There are structural and functional consequences of being multilingual. Functional MRI studies have shown that the neural systems engaged by bilinguals' two languages overlap to a great extent (Abutalebi & Green, 2007; Indefrey, 2006; Sebastian, Laird, & Kiran, 2011). We also know that activation patterns during multilingual language processing tasks are driven by language proficiency (Abutalebi, 2008; Sebastian et al., 2011). When the degree of proficiency in multilinguals is very high, a common language system comprising the left hemisphere language network appears to be responsible for the processing of both languages (Chee, Hon, Lee, & Soon, 2001; Perani et al., 2003). In low-proficient languages, however, there is a more extended network of activations that includes brain regions related to speech motor planning and cognitive functions (Abutalebi et al., 2009; Stein et al., 2009).

Implications for aphasia: It is likely that after brain damage, the same mechanisms involved when predicting the neural substrates of language recovery in monolingual aphasia are also true in multilingual aphasia. Multilingual Spanish-English clients with aphasia activate perilesional left-hemisphere regions for their more proficient language and a broader network, including right-hemisphere regions, for the less proficient language (Sebastian, Kiran, & Sandberg, 2012). The redundancy of the neural regions may support language recovery in ways that are facilitative. Therefore, engaging in treatment tasks that tap into multiple languages may also strengthen recovery-related changes at the neural level.

Cognitive Consequences

Research on multilingual language processing indicates that multilingualism may offer cognitive advantages throughout the lifetime. In language processing, multilinguals activate all languages, but they must suppress the non-target language(s). In doing so, multilinguals recruit the executive control system in a way that monolinguals do not, and this system is reinforced throughout the lifetime as the languages continue to be used. Researchers suggest that multilinguals' inhibitory processes may afford advantages in attention control and postpone the decline of cognitive control (Bialystok, Craik, Klein, & Viswanathan, 2004; Gollan, Sandoval, & Salmon, 2011; Kavé, Eyal, Shorek, & Cohen-Mansfield, 2008).

Additionally, multilinguals have been shown to develop dementia later than monolinguals, suggesting that multilingualism contributes to cognitive reserve (Craik, Bialystok, & Freedman, 2010). However, this cognitive reserve may allow for greater neuropathy accumulation before symptoms manifest, a premise suggested by the greater atrophy found in bilinguals with probable Alzheimer's disease as compared to monolinguals (Schweizer, Ware, Fischer, Craik, & Bialystok, 2011).

Implications for aphasia: The cognitive implications of multilingualism can influence the approach to rehabilitation in multilingual people with aphasia. For instance, it may be possible that working with them on cognitive exercises targeting inhibition and suppression may help them control the language in which they are conversing. Such exercise could perhaps also limit cross-language interference or pathological switching (Penn , Frankel, Watermeyer, & Russell, 2010). However, it is also possible that if multiple languages are represented on the same neural substrates, a multilingual person may be more susceptible to aphasia and may have greater language deficits relative to monolinguals (Wong & Kiran, submitted).

Considerations for Clinicians

Studies that systematically examine assessment and intervention with multilinguals with aphasia are ongoing. The results are preliminary, and many questions remain unanswered. Nevertheless, some suggestions may help clinicians working with these clients.

Understand clients' pre- and post-stroke language use and background. A clinician's thorough understanding of the client's history and patterns of language acquisition, mastery, and use for each of his or her languages may improve assessment and intervention. Gather information from the client and people familiar with the client's history [PDF] and integrate the evidence (e.g., Adrover-Roig, Marcotte, Scherer, & Ansaldo, 2012; Kiran & Roberts, 2012).

Multilingual aphasia is not categorical: Deficits in each language lie on a continuum. Use terms of relative impairment and relative recovery in determining whether all the clients' languages were similarly affected and consider pre-morbid proficiency when determining parallel patterns of impairment post-stroke (Gray & Kiran, submitted; Paradis, 2004). For example, clients who were not equally proficient in all their languages prior to the stroke, and do not demonstrate comparable abilities in all their languages following the stroke, can be considered to show parallel impairment, despite differences in the absolute impairment, if the relative proficiency post-stroke mirrors their relative abilities pre-stroke.

Understand the limits of assessment tools. Assessment results may be influenced by the limitations of and differences among assessments. Translated tests may introduce biases of items and levels of difficulty. Even tests carefully adapted to multiple languages (the Bilingual Aphasia Test; Paradis & Libben, 1987) have limitations (see special issues in Clinical Linguistic and Phonetics, 2011, and Journal of Neurolinguistics, 2012). Use open-ended assessments, such as verbal fluency, picture narrative description, and personal narratives to augment or replace translated tests (Altman, Goral, & Levy, in press; Roberts & Kiran, 2007). Be aware of measurement errors introduced when working with a translator who is a family member or a caregiver of the client.

Choosing a Language

Using everything you know about the multilingual patient's language history, you will need to make an informed decision about which language to train.

• Treating one language does not mean you are jeopardizing the other language. Clinicians may decide to treat one or all languages of the clients. Positive cross-language effects may occur in some cases. However, a given treatment session should have a clear target language to reduce confusion and frustration. If the same brain regions are likely to process all the languages in a multilingual, then the brain regions engaged in treatment are probably the same for the client's second or third language, especially for multilinguals who achieved high proficiency in all their languages.
• Consider the client's language background and relative strength of languages. If you select one language for treatment, consider many factors, including the client's choice and relative proficiency/accessibility. In contrast, treating one language may facilitate languages of comparable accessibility (Edmonds & Kiran, 2006; Kohnert & Peterson, 2012). Thus, a stronger language may be an ideal choice when only one language is treated.
• Make informed decisions about code-switching and translation. Code-switching and translation may be useful for some clients (Ansaldo, Saidi, & Ruiz, 2010; Goral, Levy, Obler, & Cohen, 2006), but may promote cross-language competition and interference in others. Nevertheless, don't discourage code-switching and code-mixing if they help the client maximize communication and the communication partners can communicate in all languages. There are no clear findings on what variables determine these patterns, so clinicians may want to adjust intervention strategies for individual clients on the basis of their own experience.
• Consider clients' personal preferences, as well as cultural and sociolinguistic contexts. Decisions about treatment language should respect the linguistic context and preferences of the client. If maintaining multilingual communication abilities is important to the client, there is no need to limit treatment and communication to only one language (Penn, 2007).

Future Trends

We need to know more about the complexities of accurate assessment, maximizing treatment efficacy, and cross-language influence in multilingual aphasia. Future studies would contribute to our understanding and rehabilitation of multilingual aphasia by focusing on:

• The circumstances—including relative language proficiency and language similarities—under which intervention in one language benefits the untreated languages.
• The time-course and maintenance effects of cross-language facilitation and inhibition.
• Best strategies to incorporate natural multilingual behaviors, such as code-switching, to maximize treatment efficacy.
• Effects of treating multiple languages during the same period and during a single session in the context of multilingual and monolingual sessions.