2. Intentional language learning and older adults

There is agreement across the scientific and lay communities that older adults likely learn differently than young adults – although not necessarily more poorly – due both to cognitive aging and differing motivations for undertaking language study (Cox, Reference Cox, Sanz and Lado2013; Ramírez Gómez, Reference Ramírez Gómez2016). However, SLA studies on the ‘age question’ have tended to be framed in terms of, on the one hand, the Critical Period Hypothesis (e.g., DeKeyser, Reference DeKeyser2013) and on the other, the effects of starting age in L2 learning in childhood and adolescence (e.g., Muñoz, Reference Muñoz2011; Pfenninger & Singleton, Reference Pfenninger and Singleton2017). Turning to an under-studied age group can further our understanding of the cognitive basis of SLA and also inform teachers and learners on best practices for this population. Since in many regions, older adults already gather in senior centers or other community groups, and thus may be interviewed or surveyed relatively easily, one task can be to start with their specific language-learning needs. When working with communities in which older adults do not have established gathering places, it can be possible to reach older adults through local media or organizations not specific to older adults. For example, my research team and I regularly appear on our local bilingual (Spanish/English) radio station to talk about research on bilingualism in general and our studies in particular, which often results in older adults contacting us to participate. Although less personal, recruitment flyers could also be disseminated through institutions such as schools (targeting pupils’ grandparents) or healthcare centers.

To know whether what is happening in the classroom will meet learners’ needs, we first need verification of what those needs are. Task-based language teaching (TBLT) offers a theoretically grounded framework for conducting needs analysis that yields specific tasks that learners want to be able to do in the target language and learning to complete those tasks becomes the framework of the language course (Long, Reference Long2015). Needs analyses will take us the next step beyond current surveys of general beliefs that older adult language learners hold about their language classrooms, which to date have been mostly focused on non-migrant retirees (Bernal Castañeda, Reference Bernal Castañeda2016; Mora & Abad, Reference Mora and Abad2016; Słowik, Reference Słowik2016, Reference Słowik2017; Lipińska, Reference Lipińska2017; Derenowski, Reference Derenowski and Gabryś-Barker2018; Grotek, Reference Grotek and Gabryś-Barker2018). Hubenthal (Reference Hubenthal2004) reported findings from similarly broad interviews with migrants and Pot, Keijzer, and de Bot (Reference Pot, Keijzer and de Bot2018a) interviewed health consultants who work with aging migrants to investigate the needs of the latter group.

Moving from general beliefs to specific needs is a crucial step because general beliefs can be misguided. For example, older adults may have negative self-perceptions of their learning abilities due to cultural stereotypes of aging that are not necessarily accurate representations of their actual abilities (Levy & Langer, Reference Levy and Langer1994). Tasks, on the other hand, are things that people do with language or that learners want to be able to do in the target language: ‘[activities in which] the target language is used by the learner for a communicative purpose (goal) in order to achieve an outcome’ (Willis, Reference Willis1996, p. 23).

Needs analysis may be especially fruitful for older learners, who likely have specific motivations and goals and thus may be better able to articulate their needs than non-migrant young adults who may be undertaking language study with only vague ideas of its future usefulness (Cox, Reference Cox, Sanz and Lado2013; Arxer, Ciriza, & Shappeck, Reference Arxer, Ciriza and Shappeck2017). By definition, a needs analysis will be specific to the local population and circumstances and thus will need to be replicated when undertaking research in a new geopolitical area or profile of L2 learner (e.g., migrant versus non-migrant retiree).

Developing a needs analysis consists of several stages, beginning with collecting general information about what tasks individuals need to be able to complete in the target language (often collected via several semi-structured interviews) and then corroborating and quantifying those needs via a survey administered to a larger sector of the population (Long, Reference Long2015). Surveys should include broad questions (e.g., ‘What activities do you do on a daily or weekly basis?’) as well as questions that offer specific tasks (e.g., ‘call the doctor's office to make an appointment’) that the participants rank in order of importance. Prior to conducting a Web-based survey, researchers should speak with individuals who often interact with the targeted population to determine whether the older adults in question will be sufficiently familiar with surveys to be able to complete it on their own (a potential issue when cultural differences come into play) and whether they uniformly have reliable access to the internet to do so. Alternative methods could include distributing paper copies of the survey in class or at a senior center event, perhaps preceded by the researcher briefly presenting their goals and motivation. A second option would be to continue with semi-structured interviews, using follow-up questions to lead interviewees to the desired level of specificity in their answers. Time and resources permitting, the most informative needs analysis will make use of multiple methods of collecting information (survey, interview, stimulated recall) and triangulation between informants: learners, teachers, and in the case of language classes for specific purposes, domain experts (Long, Reference Long2015). An example of this type of triangulation outside of TBLT is Liu and Wen's (Reference Liu and Wen2010) analysis of older adults’ reactions to comprehensible input and the silent period in an English class in China.

Regardless of which methodology (or methodologies) is chosen for the needs analysis, a subset of questions should be dedicated to means analysis, defined as an analysis of additional constraints on learning that might exist in the community, such as cultural expectations for student/teacher roles in the classroom and sociocultural valuations of the first language (L1) and second language (L2) (Long, Reference Long2015; Bernal Castañeda, Reference Bernal Castañeda2017). Means analysis should be conducted concomitantly with a needs analysis, not retrospectively, so as to anticipate potential challenges to the implementation of a task-based curriculum and thereby ‘work with the system, rather than against or in spite of it, to the greatest extent possible’ (Holliday & Cooke, Reference Holliday and Cooke1982). Potential challenges and constraints identified in the means analysis can then be addressed by shifting priorities to avoid roadblocks that cannot be changed, using creative critical thinking to address potential roadblocks that are malleable, and taking full advantage of cultural features that will work in favor of the desired curricular changes.

One of the data sources for triangulation in both needs and means analysis can be classroom observation. An equally valid approach could be to take classroom observation and posterior interviews with learners and teachers as its own research task. Rather than using class observation to define the needs and constraints of the population, the goal of this task would be to evaluate current classroom practices with older adult learners.

In locations ranging from Europe to Japan to Ecuador, some classes for older adults (or including older adults alongside other adult ages) are already underway, whether they are in conjunction with or independent from universities (e.g., Mackey & Sachs, Reference Mackey and Sachs2012; Bernal Castañeda, Reference Bernal Castañeda2016; Mora & Abad, Reference Mora and Abad2016; Ramírez Gómez, Reference Ramírez Gómez2016; Arxer et al., Reference Arxer, Ciriza and Shappeck2017; Gabryś-Barker, Reference Gabryś-Barker and Gabryś-Barker2018). These courses may tend to attract non-migrant retirees, although the European Union has also developed initiatives to increase language instruction for migrants at all ages (Majhanovich & Deyrich, Reference Majhanovich and Deyrich2017). Older migrants, whether newly arrived or aging in place, might attend class at a literacy or adult-education center. In any of the above cases, researchers have the opportunity to evaluate current language instruction, to be able to ground any innovations in what has already been established. Additionally, this task could be helpful to the administrators or teachers of current programs, although care must be taken to establish rapport with stakeholders to clearly establish the researcher's intentions with the study (i.e., to provide an outsider's perspective, not to criticize or impose).

Ideally, multiple researchers or research centers in different regions will observe existing classrooms that either consist entirely of or at least include different profiles of older learners (newly arrived migrants, migrants aging in place, and non-migrant retirees). Moreover, there will likely be differences between practices and their outcomes in classrooms specifically for older adults (as in Ramírez Gómez, Reference Ramírez Gómez2016) versus those that include a variety of adult ages (as in Andrew, Reference Andrew2012), so the two should be investigated separately. Observations should include attending class and taking field notes in person while also audio or video recording the class (Mackey & Gass, Reference Mackey and Gass2015). To lend weight to the researchers’ interpretations of what is happening in class, recordings of specific language-related events should be used in retrospective interviews with teachers and students to elicit their appraisals of challenges and successes in the classroom through the procedure of stimulated recall (Mackey & Gass, Reference Mackey and Gass2015).

Besides demonstrating what is currently occurring in classrooms with older adults, as well as what older adults’ needs are, an additional approach can be to test whether a teaching practice is, in fact, successful with the population at hand. While the teaching practice or methodology that a researcher chooses to investigate may arise from classroom observations or interviews with learners and teachers, it should also be informed by SLA theory and the psychology of aging, to increase the likelihood that it will successfully minimize any challenges particular to older adults while also capitalizing on their strengths. This type of empirical study will build on the current research that we have on older adults’ ability to learn new words (Gallardo del Puerto & Martínez Adrián, Reference Gallardo del Puerto, Martínez Adrián and Jiménez Catalán2014; Marcotte & Ansaldo, Reference Marcotte and Ansaldo2014), phonology (Ristin-Kauffman & Gullberg, Reference Ristin-Kauffman and Gullberg2014; Ingvalson, Nowicki, Zong, & Wong, Reference Ingvalson, Nowicki, Zong and Wong2017) and artificial grammars (Schwab et al., Reference Schwab, Stillman, Howard, Schuler, Newport and Howard2016).

To ensure control over both dependent and independent variables and limit the effects of potential confounding factors, I recommend laboratory studies for this research task. However, laboratory studies are more resource-intensive, since they generally require learners to make appointments with the researcher outside of their other obligations, which may not be feasible with all populations. Such studies also generally require a relatively large sample size: at least one experimental and one control group, with more than the minimum number of participants in each group to anticipate attrition. Also, groups should be either balanced in important factors such as socioeconomic status and gender or researchers should develop a statistical plan to take those additional variables into account. As a result, this task might be best suited to aging in place migrants or non-migrant retirees, as new arrivals might not have the resources to make this type of commitment. The study should follow a repeated-measures design of pretest, instructional treatment with practice, post-test immediately following the treatment and, preferably, delayed post-test some weeks after treatment. Comparing scores on pretest, immediate post-test and delayed post-test will reveal participants’ intake from the instruction and retention of that information.

Laboratory studies can measure learning via behavioral or neurophysiological methods. Behavioral methods could be, but need not be, computer-based (see, for example, Mackey & Sachs, Reference Mackey and Sachs2012 for a study with older adults that did not use computers). If instruction and practice are computerized, then researchers should be sure that their participants have the necessary experience with said technology before beginning. If funding permits, neurophysiological methods can yield fine-grained data on L2 processing without requiring the learner to interact with the technology. As noted by Kliesch, Giroud, Pfenninger, and Meyer (Reference Kliesch, Giroud, Pfenninger, Meyer and Gabryś-Barker2018), electroencephalogram (EEG) and specifically, event-related potentials (ERPs) in response to specific linguistic stimuli, would be preferable for older adults over other neurological methods, such as fMRI, since it is relatively quiet and less stressful to the participant.

This task also moves the field forward by focusing on the extent of older adult success in what and how older adults learn (Ramscar, Hendrix, Shaoul, Milin, & Baayen, Reference Ramscar, Hendrix, Shaoul, Milin and Baayen2014; Hartshorne & Germine, Reference Hartshorne and Germine2015; Amer, Campbell, & Hasher, Reference Amer, Campbell and Hasher2016), instead of possible failure to learn as much or as quickly as young adults. As a result, the study should not include a young adult comparison group; instead, comparisons could be done between older adults who receive two different types of instruction or instruction and no instruction. Results from this task can help combat negative stereotypes of aging and may be more relevant for teachers of classes dedicated to older adult students, even laying the groundwork for training teachers to work with older adults (Gabryś-Barker, Reference Gabryś-Barker and Gabryś-Barker2018).

Some researchers have identified a few methodologies as potentially appropriate for older adult learners, but they have not been thoroughly tested yet. I have advocated for Processing Instruction (PI) and TBLT (Cox, Reference Cox, Sanz and Lado2013): PI because it instructs learners on target-language processing (drawing attention to common processing errors through structured input and practice with feedback) and TBLT because it teaches communicative competence for specific linguistic tasks. Both methods likely play to older adults’ strengths and needs by prioritizing communication over rote memorization. I have shown that PI-inspired instruction yielded comparable learning for non-migrant retirees compared to young adults, although bilinguals had an advantage over monolinguals in the more metalinguistic condition (Cox & Sanz, Reference Cox and Sanz2015; Cox, Reference Cox2017). Using a similar paradigm, Lenet, Sanz, Lado, Howard, and Howard (Reference Lenet, Sanz, Lado, Howard, Howard, Sanz and Leow2011) found that older adults who received yes/no feedback showed significant learning, but those who received metalinguistic feedback did not. Also working with non-migrant retirees, Ramírez Gómez (Reference Ramírez Gómez2016) proposed a critical foreign language geragogy that includes limiting extraneous processing demands, focusing on tasks that are most relevant to older learners, and ‘retraining’ or teaching effective language-learning strategies. She reported positive student evaluations of a retraining class. However, TBLT with older adults has not yet been tested, nor have the other aspects of Ramírez Gómez’ critical foreign language geragogy. Moreover, findings to date from an immigration context (Mackey & Sachs, Reference Mackey and Sachs2012; Kliesch et al., Reference Kliesch, Giroud, Pfenninger, Meyer and Gabryś-Barker2018) showed that learners with the highest cognitive capacity showed the greatest linguistic gains; in the case of Mackey and Sachs’ study, only the highest-capacity learners maintained gains at a delayed post-test. Therefore, systematic research with both populations is warranted.

Given the immense societal pressure on migrants to acculturate and the economic advantages that are contingent on mastery of the host country's majority language, classes targeting migrants are perhaps where the societal impact of research on older adults can be the greatest. At the same time, classes of migrants, where they exist, are likely to have the most diversity in learner backgrounds, including not only variability in cognitive and socioaffective individual differences, but also in the number of L1s present in the classroom, learners’ level of education prior to immigration, and their L1 literacy. Thus, we need to consider how different learner-internal individual differences interact with different types of instruction (teaching methodologies). I suggest literacy and social networks as two important variables that could have a substantial impact on older migrant learning.

We know from studies of young adult learners that some instructional conditions are more advantageous for some learners, due to their working memory capacity, personality, or any number of other individual differences (e.g., Stafford, Bowden, & Sanz, Reference Stafford, Bowden and Sanz2012; Tagarelli, Ruiz, Moreno Vega, & Rebuschat, Reference Tagarelli, Ruiz, Moreno Vega and Rebuschat2016). Some individual differences that are present in groups of older migrants, such as variability in L1/L2 similarity, will be similar to those in studies of young adult migrants, whereas other individual differences may differ in their nature and effect on learning for older adults. Literacy and social networks are likely to be key factors for older adult migrants because both low literacy and L2 socialization could have large effects on older adults’ intake from the input they receive, intake that is already potentially lesser than that of younger adults due to aging-related declines in sensory acuity and working memory (Park, Reference Park, Park and Schwarz2000).

Literacy impacts multiple areas of cognition beyond the literal ability to read, such as phonemic awareness and memory (Tarone, Bigelow, & Hansen, Reference Tarone, Bigelow and Hansen2009; Young-Scholten, Reference Young-Scholten2013), so challenges in the low-literate L2 classroom are not entirely resolved by simply avoiding print materials in instruction – they likely will also appear in learning and retention of vocabulary and grammar patterns, even in oral/aural modes. For these reasons, Tarone et al. (Reference Tarone, Bigelow and Hansen2009) and Young-Scholten (Reference Young-Scholten2013) argue persuasively that avoiding variability in adult L1 literacy and education level severely limits the generalizability of SLA theories. Questions such as whether and how low-literate older adults can make use of constructs familiar to SLA researchers, such as the structured input that is the crux of PI, largely remain open empirical questions. To test these questions, literacy and social networks could be measured via questionnaires (e.g., Pot, Keijzer, & de Bot, Reference Pot, Keijzer and de Bot2018b for potential questions about social networks); however, the questionnaires will need to be completed orally, in an interview with a researcher. Literacy could also be tested through the Native Language Literacy Screening Device, which is available for 29 languages and could be adapted for others (Tarone et al., Reference Tarone, Bigelow and Hansen2009). Operationalizations of literacy should take into account modern technological developments in reading and writing and differentiate between reading novels and newspapers versus texting and email.

As for social networks, while it is relevant to investigate young adults’ willingness to communicate in L2, older adult migrants may be willing to communicate in the target language, but lack opportunities to do so due to L1-based social networks or social isolation (for example, grandparents who stay at home to care for grandchildren). In turn, their limited opportunities for practice can feed into greater language anxiety and L2 avoidance (Arxer et al., Reference Arxer, Ciriza and Shappeck2017; Sevinç & Backus, Reference Sevinç and Backus2017). Individuals with limited L2 social networks also tend to come from more marginalized groups (women, less educated, and/or older in age), which may compound their difficulties in learning L2 (Adamuti-Trache, Reference Adamuti-Trache2013; Pot, Keijzer, & de Bot, Reference Pot, Keijzer and de Bot2018c). In sum, by studying the combined effects of these two sociocognitive variables in older adult language-learning contexts, we will gain a better idea of how both social and cognitive factors might play out for older learners.

This research task would follow a repeated-measures design similar to that described for the previous task. Quantitative analyses could proceed in one of two ways: if the population being researched is consistently low literacy and/or low in L2 networks, then the tasks from the previous paragraph serve to describe the population and frame the discussion of linguistic gains resulting from instruction. Alternatively, if the population includes a spectrum of literacy levels or social networks and multiple levels are present in the study sample, then these factors can serve as predictors of any linguistic gains that result from instruction. Discussion could then posit how each factor affects L2 learning in the population under study.

In either case, to enhance quantitative findings, qualitative stimulated-recall interviews can incorporate insight from participants’ perceptions of the instruction. Thus, as with Research task 3, this study might be best undertaken in laboratory settings rather than classrooms, to allow for individualized contact between participants and researchers. As noted for Research task 3, this can be difficult with migrant populations. It can be possible, though, especially if the laboratory can be mobile and travel to participants instead of requiring them to travel. Participant recruitment for such a study will be most successful if the researchers are already known to the migrant community, so time must be taken to establish relationships and trust. Additionally, a member of the migrant community who is willing and able to work as a cultural liaison, and interpreter if necessary, can facilitate communication between researchers and prospective participants (Tarone et al., Reference Tarone, Bigelow and Hansen2009). Taking a laboratory approach also makes this task feasible for areas that do not yet have language classes with many older migrant learners.

In addition to variability in learners’ individual differences that affect their intake from instruction, such as literacy and social networks, older learners likely also vary in what they do outside of class. One advantage of laboratory studies outside of the immigration context is that the participant (presumably) does not have exposure to the target structures outside of the laboratory, but that is not necessarily the case when participants are migrants, since depending on their circumstances, they may continue to hear and use the target language (the language of their new country) around them outside of the laboratory. Therefore, an additional area of research needs to address the factor of migrants’ target-language interaction in daily life. For young adults, this variable has long been incorporated into research on study abroad (Kinginger, Reference Kinginger2009) and, to a lesser extent, at-home environments that range from language cafes to online tutoring (Reinders & Benson, Reference Reinders and Benson2017). Although this could be a subcomponent of a laboratory study, for simplicity's sake, I discuss conducting this research task with classroom learners, because it could be undertaken in a classroom setting without being part of a larger project.

As outlined by Reinders and Benson (Reference Reinders and Benson2017), LBC can be intentional or incidental, explicit or implicit, and deductive or inductive. Since the analysis of L2 social networks from the previous research task indexes incidental learning opportunities, I focus here on intentional LBC. Although LBC can be relevant to learners of all ages, older adults are of particular interest because they are likely to be more self-motivated and autonomous learners than young adults (Cox, Reference Cox, Sanz and Lado2013). In fact, Blumenfeld, Quinzon, Alsol, and Riera (Reference Blumenfeld, Quinzon, Alsol and Riera2017) found that the number of classes attended did not predict learning outcomes for older adults, suggesting that practice with language materials (provided by the researchers) was able to make up for missed class. At least one study to date reports the offerings and challenges coming from a language-learning center that attracted seniors (Murray, Reference Murray, Benson and Reinders2011), but not how LBC contributes to success for classroom learners. Furthermore, LBC is relevant for each of the profiles of older adult learners (newly arrived or aging migrants as well as non-migrant retirees) and may be key to reaping cognitive advantages from language learning (Bak, Long, Vega-Mendoza, & Sorace, Reference Bak, Long, Vega-Mendoza and Sorace2016).

This task will depend on self-report, so steps should be taken to ensure participants’ accurate recall to the greatest extent possible. One approach for longer courses (e.g., traditional semester length of 10–15 weeks) could be to distribute the relevant questionnaire multiple times, thereby obtaining multiple snapshots of the learners’ practices rather than asking them to recall and average across 15 weeks at the end of the semester. Questionnaires should include items that target both quantity of independent study (that is, how much time the participant spends studying on a daily or weekly basis) as well as quality (what activities or strategies they take part in). Possible strategies to include on the questionnaire could be compiled from other work on strategies and older learners (Ramírez Gómez, Reference Ramírez Gómez2016; Blumenfeld et al., Reference Blumenfeld, Quinzon, Alsol and Riera2017; Pawlak, Derenowski, & Mystkowska-Wiertelak, Reference Pawlak, Derenowski, Mystkowska-Wiertelak and Gabryś-Barker2018; Piechurska-Kuciel & Szyszka, Reference Piechurska-Kuciel, Szyszka and Gabryś-Barker2018). Further guidelines for LBC questionnaires can be found in Reinders and Benson (Reference Reinders and Benson2017). Results from the LBC questionnaire could then be tied to learning outcomes in the language class to see if certain LBC activities, or a threshold of amount of LBC activity, are related to higher levels of L2 development.

3. Language, cognition, and aging

Older adult retirees (non-migrants) often report starting L2 study as a way of ‘exercising’ their mind à la the ‘use it or lose it’ philosophy that keeping one's mind active will attenuate cognitive declines due to aging (e.g., Pfenninger & Polz, Reference Pfenninger and Polz2018). Older adults who enroll in L2 classes also often include goals of increased social interaction, firstly through attending class and secondly, through L2 proficiency that can expand their social networks at home or through travel, as reported in the popular press (Mohn, Reference Mohn2012). Some research in cognitive psychology shows evidence of an advantage in the component of cognition that ‘actively maintain[s] task goals and goal-related information’ (Miyake & Friedman, Reference Miyake and Friedman2012), that is, general executive function, for some bilinguals. Such an advantage may be more evident in older adults, due to the general declines that come with cognitive aging, as opposed to young adults, who are at the peak of their cognitive capacity. Many early studies in this field used monolingual/bilingual comparisons and found a bilingual advantage for older adults (e.g., Bialystok, Craik, Klein, & Viswanathan, Reference Bialystok, Craik, Klein and Viswanathan2004; Salvatierra & Rosselli, Reference Salvatierra and Rosselli2011; Cox et al., Reference Cox, Bak, Allerhand, Redmond, Starr, Deary and MacPherson2016; Gathercole et al., Reference Gathercole, Thomas, Guasch, Kennedy, Prys, Young and Jones2016); notably, though, others have not found such an advantage (e.g., de Bruin, Treccani, & Della Sala, Reference de Bruin, Treccani and Della Sala2015; Antón, Fernández García, Carreiras, & Andoni Duñabeitia, Reference Antón, Fernández García, Carreiras and Andoni Duñabeitia2016).

To resolve this discrepancy in findings, we need to take into account that real differences between bilingual individuals, such as age of acquisition, frequency of use of each language, and proficiency in each language, all imply different cognitive experiences and therefore may have different effects on executive function. As a result, ignoring differences between bilinguals may obscure results. Moreover, the effects of language-experience variables on cognition may be different at different lifespan stages, so we need to continue studying older adult bilinguals. For example, lifelong use of two grammars and lexicons may result in language systems that are intertwined, rather than in competition, whereas young adult bilinguals of similar proficiency level may have more separate language systems (Goral, Campanelli, & Spiro, Reference Goral, Campanelli and Spiro2015). Dual language systems that are not in competition would not be expected to yield the same kind of cognitive benefits as would systems that need to be actively managed (Green & Abutalebi, Reference Green and Abutalebi2013). Moreover, the potential public health benefits of any bilingual advantage in executive function would be greatest if they led to lesser negative effects of cognitive aging (e.g., Bialystok, Reference Bialystok2016), which necessitates testing older adults.

For most research interests under the umbrella of cognition, aging, and multilingualism, detailed questionnaires on language background and previous language-learning experiences are necessary, but many existing questionnaires are biased toward young adult respondents. For example, older adults have a longer, potentially more complex language history to report and may spend their time in different domains than young adults, all of which needs to be reflected in the questionnaire.

Established language-background questionnaires in both applied linguistics and cognitive psychology implicitly have younger adults in mind. For example, the Language Experience and Proficiency Questionnaire or LEAP-Q (Marian, Blumenfeld, & Kaushanskaya, Reference Marian, Blumenfeld and Kaushanskaya2007) does not include a question on current classroom exposure to the language, because in their sample overall classroom exposure was highly correlated with current reading exposure in the L1 and L2, respectively. However, it is unknown whether amount of previous language instruction would play a distinct role in the language profile of someone who completed school many years earlier. The LEAP-Q does ask for the number of years of education in each language, but that indicates enrollment at an L1 or L2-speaking school, not amount of (foreign) language instruction. As a second example, although the defining factor of bilingual proficiency for young adult bilinguals in migrant contexts may be their use of the non-majority language, the same is not necessarily true of older adults (Anderson, Mak, Chahi, & Bialystok, Reference Anderson, Mak, Chahi and Bialystok2018). In fact, older adult migrants who at one point used the majority language frequently in the workplace may experience a drop in their use of, and therefore proficiency in, that language after retirement – or it may be that increased errors in the L2 are merely artefacts of general cognitive aging (Schmid & Keijzer, Reference Schmid and Keijzer2009; Keijzer, Reference Keijzer, Schmid and Lowie2011). Lastly, some studies of bilinguals across the adult lifespan show a greater role for length of residence in childhood than that of usage patterns at the time of testing (that is, adulthood or older adulthood), at least in the case of highly proficient simultaneous bilinguals in Europe (Kupisch & van de Weijer, Reference Kupisch, van de Weijer, Silva-Corvalán and Treffers-Daller2016).

In sum, a language-background questionnaire for older adults should include opportunities to report changes to the language system(s) not only with age of onset of each language, but also changes in use and proficiency in response to life events such as moving, marriage, or changing jobs (de Bot & Schrauf, Reference de Bot and Schrauf2009). A questionnaire for older adults should also refer to contexts and interlocutors that are relevant to them. These innovations hold for both migrant and non-migrant retiree populations and are necessary for accurate portrayals of participant samples, precise measurement of language-experience variables (for example, for investigating effects on executive function in Research task 7), and comparability across studies.

Models can be found in the Bilingual Dominance Scale (Dunn & Fox Tree, Reference Dunn and Fox Tree2009) and Anderson et al.’s (Reference Anderson, Mak, Chahi and Bialystok2018) Language and Social Background Questionnaire (LSBQ), which ask about language use at multiple life stages, not just at the time of testing. However, both measures are limited by only allowing for two languages, thus excluding multilinguals with knowledge of three or more languages, and have the specific goal of calculating one ‘bilingualism score’ per participant, which may not be appropriate for all research questions.

As with Research task 1, technological and cultural concerns should be taken into account in questionnaire development. If the targeted older adult population is less experienced with technology and/or with detailed questionnaires, then it may be best for the researcher to conduct an interview in which they ask the participant the relevant questions orally and record the participant's responses in a Web-based questionnaire themselves. Indeed, Anderson et al. (Reference Anderson, Mak, Chahi and Bialystok2018) recommend an interview approach even for young adults, due to the complex nature of language histories.

Moving beyond task development to investigations of cognition itself, as aforementioned, there are natural links between multilingualism, cognition, and aging. Language is generally one of the best-preserved cognitive abilities in older age (Wingfield & Grossman, Reference Wingfield and Grossman2006), and multilinguals additionally have the ability to use language control to select languages as appropriate for their context (Green, Reference Green1998). Thus, does maintenance of multiple languages and practice in language control change domain-general control, such as executive function? And, more precisely, which aspects of multilingual practice yield such effects?

This task is relevant to each profile of older adult multilingual, but only one profile should be investigated at a time, as the very different lived experiences of migrants versus non-migrant retirees likely correlate with substantial differences in cognition outside of language experience, such as trauma related to the migration experience. Replication with different older adult profiles and in different geopolitical areas is essential, as bilingual environments vary greatly and we do not expect the same findings in each area (Bak, Reference Bak2016; Ooi, Goh, Sorace, & Bak, Reference Ooi, Goh, Sorace and Bak2018). In migrant contexts, social variables such as sociolinguistic generation are also likely to be relevant and should be included in analyses.

In analyses, language-experience factors should be kept as continuous variables, not divided into levels of categorical variables, to avoid losing information and to reflect the reality that there is not a definitive dividing line between a ‘more’ proficient bilingual and a ‘less’ proficient bilingual or an ‘early’ versus ‘late’ bilingual. Whenever possible, I advocate for the use of objective measures of language proficiency, such as grammaticality judgment tests, oral proficiency interviews, or standardized vocabulary tests, to minimize the effects of bias in self-report. Additionally, comparable objective measures should be used to test both languages, rather than assuming that one language is ‘nativelike.’ In a recent review of studies that compared bilinguals to monolinguals, less than half of the studies reviewed (47%) included an objective proficiency measure and 38% of those only included an objective measure for one of the two languages (always the majority language of the region) (Surrain & Luk, Reference Surrain and Luk2017). Using objective language proficiency measures will likely limit the number of language pairings a given study can include, but as alluded to above, the study will have better internal validity if diverse bilingual groups are analyzed separately. Of course, for some variables it may only be feasible to use self-report, such as how frequently a participant uses each language.

In some areas, it may be difficult to recruit a sufficiently large number of older adults for the type of quantitative analyses outlined above. In such a situation, it may be preferable instead to conduct a case study of an individual or small cohort of older adults. Instead of looking for quantitative differences in reaction times on an executive function task, researchers could instead analyze how cognition changes qualitatively as a result of a multilingual intervention, such as engaging in languaging. Languaging is the process in which one ‘talk[s] one's way into understanding’ (Swain, Reference Swain2013, p. 5) and as such, suggests roles for both cognitive effort (working to reach that understanding) and affective change (pride in having worked out something difficult). Since languaging occurs in interpersonal interactions, recording multiple interactions over a set time period results in dense data that offers insight into the process as well as the product of the intervention, and yields more holistic results than traditional pre/post- or cross-sectional testing (Barkaoui, Swain, & Lapkin, Reference Barkaoui, Swain and Lapkin2011).

This research task is inspired by Swain and colleagues’ work on languaging with five older adult residents of a long-term care facility, all of whom were treated as functional monolinguals (summarized in Swain, Reference Swain2013). However, the same type of languaging activities could also be used in multilingual settings, with a bilingual or multilingual researcher acting as interlocutor. The activities were cognitively complex (relative to the participant) yet logistically simple, such as reading and discussing a newspaper article or movie. Analyses differed by researcher-participant pairing, and included discourse structure (Swain, Lapkin, & Deters, Reference Swain, Lapkin and Deters2013) and zones of proximal development (Lapkin, Swain, & Psyllakis, Reference Lapkin, Swain and Psyllakis2010).

Swain and colleagues reported generally positive changes in both cognition and affect as a result of their languaging interventions, and it stands to reason that benefits to multilinguals could be even larger. For example, multilingual languaging would give older adults an opportunity to continue active use of both languages, which may be important for reaping the type of cognitive benefits outlined in the previous research task. Additionally, for multilinguals, improved affect may go beyond simply feeling successful to include active maintenance of cultural ties to both languages, building an identity as a ‘language expert’ who can facilitate interactions between individuals from both language groups (Müller, Reference Müller, Plejert, Lindholm and Schrauf2017), and offering different forms of self-expression for bilinguals who associate each language with different contexts, linguistic repertoires, and potentially parts of their identity (Pavlenko, Reference Pavlenko2006).

One limitation to this task is, as noted, that the researcher and participant need to share at least two languages, so it will not be possible in all multilingual environments. However, in areas where there is a large concentration of migrants from one language group, such as Spanish speakers in some areas of the United States or Arabic or Turkish speakers in some parts of Europe, it may not be difficult to satisfy this requirement and even take into account the participant's preferred dialect, too. Or, if the participant and researcher use different dialects, contrastive analysis of the two varieties could be one of the languaging activities.