Low-educated immigrants and the social relevance of second language acquisition research

I Introduction

What social benefits does research on the representation of language in the second language learner’s mind confer on the learners studied? Against a backdrop of well-studied foreign language learners with primarily instrumental reasons for learning an L2, there exists a population of low-educated adult immigrants who have received relatively little attention over the last several decades for whom successful acquisition of a second language and literacy in that language has far-reaching social benefits. ¹

Four decades ago when second language acquisition (SLA) ² began to emerge as a field of inquiry in its own right and ceased to be a branch of foreign language pedagogy, such learners were the focus of several major studies. In the 1970s and early 1980s, a series of publications emerged from a radical new line of research whose focus was on the most disadvantaged of all L2 learners: working class adult immigrants past the age of compulsory schooling. In West Germany and other northern European countries post-Second-World-War labour shortages led to recruitment of unskilled/semi-skilled temporary workers from southern Europe, Morocco and Turkey. Visas were for limited stays and L2 classes were therefore not offered. The study of these uninstructed or ‘naturalistic’ L2 learners planted seeds for socially relevant research on the acquisition of linguistic competence, ³ but these seeds have never properly germinated in generative SLA. Those outside of generative SLA circles express similar views, e.g. Ortega argues that instructed SLA has much to gain ‘when good basic research is inspired by societal and educational concerns’ (Ortega, 2005: 432).

If language is an organ, socially-inspired language acquisition research makes no more sense than socially-inspired research on the human heart. Under strong modularity in generative SLA, the acquisition of linguistic competence is unaffected by non-linguistic factors; see, for example, Schwartz, 1993. Research becomes socially relevant when there are populations of particular interest. Just as there are populations of interest to medical researchers, there may be populations of interest to generative SLA researchers. The aim in the present article is to argue for a re-focusing of attention on a population whose acquisition is closely tied to certain non-linguistic variables. Because this population has suffered from long-term neglect among researchers, generative SLA is currently less socially relevant than it was some 40 years ago. This is unfortunate because in post-industrialized countries there are now many more adults who fit the profile of those studied in 1970s West Germany. In formal linguistics-based SLA in general – not just in generative SLA – little attention has been paid to such learners, and we therefore know much less about the L2 acquisition of this population in comparison to what we know about educated, middle class learners. Lack of a research base on this population has serious consequences for these L2 learners, who often fail to develop sufficient language and numeracy skills for living wage employment and participation in society. Nations around the world contravene Article 26 of the Declaration of Human Rights that ‘Everyone has the right to free, basic education’ by limiting the quantity and failing to control the quality of adult basic education for low-educated adults who immigrate past the age of compulsory schooling. When asked which classroom practices have been shown to be most useful for such learners, clear answers are not forthcoming. There is too little systematic research. ⁴ One reason for this research gap is that generative SLA has long been freed of the need to provide practical implications of study findings. Basic research need not involve searching for relevance – social or otherwise – and researchers will not direct their attention to a population of learners simply because their lives might be improved by participation in studies. Research must also be relevant to SLA theory.

When it comes to social relevance, the situation is dire and likely to get worse as climate change produces more refugees from poor low-lying countries with inadequate education. In 2005, the UN Millennium Development Goals Report (United Nations, 2005) reported that over 115 million children worldwide were not enrolled in school, a situation predicted to worsen due to HIV-AIDS-related decline in teacher numbers and school attendance. When some of these individuals become refugees, one third will at some point immigrate to a highly literate country (Refugee Council UK). It then falls on host countries to address the educational needs of children, adolescents and adults who have little or no native language literacy. In post-industrialized countries, literacy level is closely tied to the economic productivity of that country (Coulombe et al., 2004), and lack of literacy (and numeracy) to social exclusion, poor health, young parenthood, children’s poor school performance, increased likelihood of bearing learning-disabled children, and criminality (Bynner, 2001; Dalglish, 1982). Based on immigrants’ self-reported second language proficiency data, Dustmann and van Soest (2002) conclude that oral L2 skills allowing communication with members of the adopted country is ‘probably the most important single alterable factor contributing to their social and economic integration’ (2002: 473). Low-educated adults are the least equipped of all immigrants to be able to communicate with members of their adopted communities.

The existence of adult uninstructed L2 learners in the 1970s/1980s was serendipitous and symbiotic: before the introduction of UG and the poverty of the stimulus in the 1980s, studies sought to eliminate the variable of instruction for a more valid comparison of child and adult learners. The German research (eg Clahsen and Muysken, (1986, 1989), which drew on longitudinal and cross-sectional studies of uninstructed adults from Italy, Portugal, Spain and Turkey, intensified an on-going debate on the interpretation of paths of development for first language (L1) children and uninstructed L2 adults. However, for the past several decades, investigation of poverty of the stimulus effects (see, for example, Schwartz and Sprouse, 2013) to demonstrate post-puberty operation of universal grammar has obviated the need for naturalistic adult learners as a population of research interest. Our assumptions about low-educated adult immigrants’ L2 acquisition are, however, based on studies in which non-linguistic factors that vary for this population such as literacy were not isolated. In addressing two issues of current interest – the status of functional morphology and access to L2 functional features absent from the learner’s L1 – research can benefit from a reconsideration of naturalistic adult learners without classroom exposure to the L2 prior to arrival. For example, we know little about the acquisition of tense by wholly naturalistic Chinese learners of English. Ab initio naturalistic learners might display fundamentally different patterns in their mental representation of tense during acquisition when compared with the instructed Chinese learners studied thus far.

In what follows I review four publications that point to an unbroken line of research that is inherently interesting and socially relevant in its inclusion of non-linguistic factors but which has borne little fruit since the 1970s. These publications are: one of the German reports on the Heidelberger Pidgin Project, Kurvers’ in-depth study of non-literate immigrant adults (the first such study) in the Netherlands; Tarone, Bigelow and Hansen’s similar study of non-literate immigrant adults in the USA; the most recent issue of a journal in Spain entirely devoted to reporting on research and pedagogical practice with respect to immigrants. There is method in this selection: these publications represent nearly four decades of research on low-literate adult immigrants, much of which has not appeared in English. The authors of these publications were not pressed to more widely disseminate their findings because relevance is to those who teach their native language to L2 adults. This, I hope, will be a clarion call for those pondering unresolved questions to consider the role of external variables. Social relevance and research relevance might, as in the 1970s, again enjoy a symbiotic relationship in generative SLA.

The non-linguistic variable considered in these publications, namely literacy, is one that has received less attention in SLA in general than have, for example, motivation, attitude, personality or learning style. Without more research it will remain unclear how literacy might operate on the acquisition of linguistic competence. I conclude this review article with reference to the Low-Educated Second Language and Literacy Acquisition (LESLLA) forum, established in 2005, whose aim is to encourage more research that addresses the dual concerns of research and social relevance, and whose main activities are an annual symposium and proceedings. At present, there is far too little basic research to speak of application of research findings to the much-needed improvement of educational provision for low-educated immigrant adults in their quest to develop the skills that will allow them to participate fully in the economic and social life of their adopted communities.

II Research on uninstructed adult immigrants

Becker et al. (1977) describe the results of the earliest study – the Heidelberger Pidgin Projekt – of uninstructed adult immigrant learners of German. The study – and its title – was motivated by the idea that the many migrant workers in West Germany might be developing a pidgin (as it turned out, they did not), given large numbers of workers who had been recruited to fill labor shortages, e.g. 1977: Italy (267,000), Spain (104,000), Greece (169,000), Yugoslavia (377,000) and Turkey (516,000). The study was cross-sectional and involved 24 Italian- and 24 Spanish-speaking foreign workers in Germany (16 male/8 female L1 Italian; 16 male/8 female L1 Spanish), all over 18 years old. The researchers categorized the participants by period of residence: up to 2 years, 2–4 years, 4–6 years and more than 6 years. Directed conversation techniques were used to elicit oral data and, on the basis of 100 successive utterances produced by each learner for analysis, the team placed them into four proficiency groups. The study included a control group whose performance on these measures was mirrored by the highest L2 group. The lowest group produced utterances without a finite element, a main verb or a subject. Analysis of the data reveals the common developmental progression of morphosyntax that has been corroborated by various studies since; see summary in Vainikka and Young-Scholten (2011). However, some of the lower literate learners in the study appear to take side routes not followed by the more educated learners. Level II speaker Tomá A illustrates a pattern followed by learners with similar educational profiles where modals were used to mark tense.

(1) Ich muss gesehen (= yo lo he visto ‘I have seen it’) Tomá A, L1 Spanish

I must see-past

(Ich habe das/es gesehen.)

‘I saw that/it.’

Becker et al. suggest the overgeneralization of muss is due to its frequency of use in the workplace, but a common pattern of such overgeneralization is also found in more recent data from a Dutch corpus of low-educated adult immigrants’ oral production compiled by van de Craats and colleagues. The data show overgeneralization of a functional element – auxiliaries this time – to mark functions in a non-target manner. Without further research, it will remain unclear which of the factors proposed accounts for these patterns.

The adults in the Heidelberg study were manual workers with at least some primary school education who were socially, politically and linguistically isolated. This project was the first to demonstrate that (apart from the sub-stage mentioned above), naturalistic adult learners ‘do not behave idiosyncratically but rather stages of acquisition are more or less common for all learners, and they pass through them in a well-defined path’ (1977: 45; translation, MYS). But in comparing proficiency level to length of residence and other non-linguistic factors, learners’ rate of development was found to vary. Variation in rate after the first 2 years of residence was accounted for by type of job, location of residence, intensity of contact with German speakers, family status, mobility, sex, age at immigration, attitudes and amount of formal education. For those with higher levels of education, the relationship between schooling and linguistic competence was no longer linear; however, those learners who did not finish primary school belonged overwhelmingly to the lowest proficiency group.

Adding to the accumulating wealth of information on instructed, middle-class learners, several other large-scale cross-sectional as well as longitudinal studies in the 1970s and 1980s in Germany and elsewhere also investigated the linguistic competence of immigrant adults. These included Cancino et al.’s (1978) 10-month longitudinal study of Spanish learners of English (2 children, 2 adolescents and 2 adults), the German ZISA study of speakers of Italian, Portuguese and Spanish (12 adults over 2 years; 45 adults in a cross-sectional design) in Germany (see Clahsen et al. 1983) and the 1980s’ 30-month longitudinal ESF (European Science Foundation) project of 40 adult speakers of 6 native languages learning 5 Europe L2s. By 2001, Hawkins could state that second language learners follow a common path in their development of morphosyntax regardless of age of initial exposure, native language, input or educational background.

In the early 1980s, social factors were no longer being integrated into accounts of the L2 acquisition of morphosyntax, and while generative SLA researchers were not denying their importance, they were of ‘minor interest for this research program since [they] can, at best, explain what is seen as individual variation’ (Nicholas and Meisel, 1983: 81). If social and other non-linguistic factors can explain individual variation, this would be no small accomplishment. While sociocultural research does consider such factors, they are, however, not considered in relation to how language is represented in the mind (e.g. Firth and Wagner, 1997). Factors such as socio-economic stratum and literacy might well determine how far an individual learner progresses on a common, narrow path. There is indeed evidence from various studies that low socio-economic stratum immigrants tend to remain at the earliest attested stage of development; e.g. Klein and Perdue’s (1997) Basic Variety. Literacy has rarely been directly investigated. Where participants are reported to have had only limited education, literacy levels, particularly in the native language, were not determined. This is unfortunate. While the ZISA study established a common route of acquisition for all the learners studied, learners were additionally placed into two groups depending on whether they (over-)supplied or omitted functional morphemes. ⁵ It is possible that level of literacy was a decisive factor (see Tarone et al., 2009).

III Low-educated adults’ cognitive development: Kurvers (2002)

The acquisition of literacy in an alphabetic script is a metalinguistic activity that requires a sophisticated level of awareness. Literacy, along with vocabulary, are main determinants of initial progress in reading as is syntactic competence for later progress in literacy (see, for example, Goswami and Bryant, 1990; Morais et al., 1979; Yücesan Durgunoğlu and Verhoeven, 1998). In her 2002 book Met ongeletterde ogen [With non-literate eyes], Kurvers takes up a topic later addressed in Tarone et al. (2009): are non-literate adults a different population from literates? While there is plentiful research on school-age immigrants (with studies often revealing a lag for such children in comparison to their native-speaking peers), there is far less on adult immigrants. The study on which Kurvers’ book focuses examined various aspects of the metalinguistic awareness of unschooled adults (n = 25), low-educated (4 years of primary school) literate adults (n = 23) and pre-reading children (n = 24) from Moroccan Arabic, Papiamento (Curaçao), Somali, Sranan Tongo (Surinam), Tarafit Berber and Turkish language backgrounds to address the question of whether age or education determines metalinguistic awareness. The non-literate adults had less than 2 years of primary education and could only sight read some words learned from attending Dutch literacy classes for between 4–6 hours a week. The literate adults could read and write simple texts either in their native or second language, but had no more than 6 years of primary schooling. The children were in their last term of pre-school. All had lived in the Netherlands between 1–20 years and most were female.

The test battery included a range of tasks to measure syllable rhyme awareness, word awareness, and word and sentence segmentation. The tasks were translated into the learners’ native languages and conducted either in one of these languages or in Dutch, depending on the participants’ dominant language, which was Dutch for some in the literate adult group. Statistical analysis showed that only for the word judgment tasks could the results not be aggregated over languages (Dutch vs. the other languages), and thus the results refer to data in either the L1 or L2. The non-literate adults differed significantly from the literate adults in all language-awareness tasks, and the children from the literates but not from the non-literate adults in nearly all word tasks. On the rhyme tasks, the scores of the non-literate adults were low compared to those of the children and literate adults. Incorrect responses on the rhyme production task showed that non-literates produced responses indicating onset awareness (alliteration); judgment of rhymes, however, showed no inter-group differences. Compared to the children and the adult literates, non-literate adults struggled with word segmentation, word referent differentiation (naming qualities of something when its name changed). For word length judgment and for sentence segmentation, children and non-literates patterned together. The results from the latter task have interesting implications for acquisition of morphosyntax: the non-literate adults segmented on the basis of content rather than form, and did not isolate function words. The children, however, often segmented based on syllable boundaries. Kurvers concludes that literacy makes more of a difference than age: there were more differences between literate and non-literate adults than there were between the non-literate children and the non-literate adults. This suggests that while literates are a somewhat different population, age is no barrier. Indeed, studies by Young-Scholten and Strom (2006) and by Young-Scholten and Naeb (2010) point to low-literate adults’ development of phonological awareness along the same lines as pre-school children’s (for immigrants in the USA and UK with native language backgrounds including Dari, Farsi, Kurdish, Mipuri, Panjabi, Pushto, Tamil, Tigrinia, Vietnamese and Urdu).

Patterns of adult immigrants’ pre-literacy development are similar to those observed for young L1 children. There is one crucial difference between children learning to read for the first time and low-educated adult immigrants learning to read for the first time. The former have acquired much of the morphosyntax and phonology of their language, and they already have a sizeable lexicon. While there is variation in the rate at which low-educated adult immigrants acquire linguistic competence in their second language, low levels correlate strongly with undeveloped basic reading skills. In Young-Scholten and Strom, the Vietnamese and Somali adults who were at the lowest stage of Organic Grammar, namely the Bare VP stage (see, for example, Vainikka and Young-Scholten, 2011), in English, could not read simple texts or decode familiar words in isolation.

These studies’ findings led to a further question: if these adults possess the cognitive capacity to learn to read for the first time, in a second language, how long will this take? Kurvers et al. (2010) investigated how many classroom hours an uneducated, non-literate adult immigrant requires to reach the lowest Basic User Level (A1) of the Common European Framework of Reference (Council of Europe, 2001) and to demonstrate the ability to read short and simple texts on familiar topics. The study examined the achievement of 322 students from 39 countries of whom 80% were women and 61.3% had experienced no schooling prior to immigration and found considerable heterogeneity in reaching A1, from 300 hours to 2,700 hours. Kurvers et al.’s study corroborated what Condelli et al. (2003) found in their study of 495 adult immigrants in classrooms in seven US states: learner success is tied to varied opportunities for active, individual and relevant learning in the classroom. Among the significant extra-classroom variables in the Kurvers et al. study was contact with Dutch native speakers.

In considering factors that might contribute to rate of development, Kurvers et al. are rendered nearly speechless when it comes to the observed individual variation: it is ‘tremendous’ and they caution against introducing achievement benchmarks tied to programme funding (2010: 77). Kurvers et al. further note lack of understanding by outsiders of such learners’ educational needs and complex profiles (including learning disabilities and trauma) and the implications this has for current stringent residence and citizenship requirements in the Netherlands. Kurvers’ and other studies in Europe have led to a promising development: inclusion of sub-A1 CEF levels to capture development. In the Netherlands, these are Literacy A and B (see Janssen-van Dieten, 2006; Stockmann, 2006); see the Low-Educated Second Language and Literacy Acquisition proceedings for information on levels in other European countries and similar use of benchmarking in North America (http://www.leslla.org). Establishment of these levels is tied to the governmental allocation of funding for immigrants’ basic skills education. If attaining a level equates to achieving a qualification, this might secure funding in countries such as the UK where adult education is qualifications-driven. Applying basic research in this manner does not require re-prioritizing research aims. Rather, it entails carrying out research on a population who stands to benefit when more is known about their language and literacy development. What we need is for practitioners and policy makers to be more accurately informed in their expectations about the learning of these individuals in the same way that Lightbown’s (1985) ‘Great Expectations’ article served to do this for L2 learners in general.

IV Literacy and the acquisition of morphosyntax

Where Kurvers and colleagues look at the development of metalinguistic awareness in relation to literacy, Tarone et al. (2009) consider how literacy might influence the acquisition of linguistic competence. This line of research is likely to be of greater interest in generative SLA. One long-standing issue is the post-puberty operation of UG and a relatively recent issue the L2 learner’s initial state (see Schwartz and Sprouse, 1996). Neither has been investigated in any great depth with non-literate immigrant adults. We assume that the operation of UG in L2 acquisition has nothing whatsoever to do with literacy: if UG operates similarly in children and adults, literacy is irrelevant; see Vainikka and Young-Scholten (2007). That literacy is not a relevant source of initial state knowledge for acquisition of L2 linguistic competence is also assumed. Whether literacy affects acquisition of linguistic competence is a testable hypothesis which has, however, received little attention apart from a growing body of research in L2 phonology (on orthographic input, see, for example, Bassetti, 2009).

In Literacy and oracy, Tarone et al. (2009) note that because ‘we know next to nothing about [non-literate learners’] processes of oral second language acquisition’ and are forced to base ‘an SLA theory of universal cognitive processes on data drawn only from literate learners’ the theory is insufficient (p. 1). Here they point out that in the USA, the percentage of low-educated, non-literate adult immigrants ranges from 3% to 15%, depending on age. The book begins with an overview of research on the relationship of phonological awareness to the development of Roman alphabet literacy and moves on to summarize recent research on low-literate immigrant adults. The authors then present their study, the purpose of which was to challenge the idea that conscious processing plays a major role in the acquisition of morphosyntax for all learners (Schmidt’s 1990 Noticing Hypothesis) in the context of question formation (see, for example, Pienemann and Johnston, 1987). A larger study included 35 participants, but those who were of particular interest were 8 Somali immigrants between the ages of 15 and 27 who had been in the USA between 3–7 years. Their oral proficiency was established as ranging between 30–50 on the SPEAK test; there were no significant group differences with respect to this measure. Their L2 English literacy was tested and established as ‘low’ or ‘moderate’. Only one of the participants (in the ‘low’ group) had had any schooling (7 years) prior to immigration. The test battery was oral and comprised spot-the-difference story completion, story recall and elicited imitation tasks. The authors also included a task that operationalizes the notion of noticing: if a learner is consciously aware of the inaccuracy of what he or she has just produced, he or she will accurately repeat the interlocutor’s correct recast. Results for question recasts showed that those with moderate literacy were significantly better at recalling and accurately repeating recasts. Length of utterance did not turn out to be significant for either the low or moderate literate group. Analysis of data from the story retelling task also showed that verbs erroneously lacking inflection were significantly more frequent in low literates’ oral production (50% of the time) than moderate literates’ (36%.) The authors conclude that alphabetic literacy has an undeniable effect on the acquisition of L2 morphosyntax. ⁶ Tarone et al. also ask how working memory capacity (Gathercole and Baddeley, 1989) might be implicated, and how capacity differences might relate to development of literacy and acquisition of linguistic competence in various domains. Addressing these issues with this population has the potential to offer new perspectives on adult L2 acquisition, but logistic and other challenges must first be addressed; see Juffs and Rodríguez’s (2008) exploratory study.

Like Tarone et al. (2009), Mishra et al. (2012) note the need to question findings based on studies of middle-class, educated populations, pointing to Henrich et al.’s (2010) novel use of non-literate adult populations in psychology experiments that had heretofore been conducted only on Western university students. Mishra et al. compared university students to low-educated ⁷ adults who were all native speakers of Hindi in Uttar Pradesh. They measured eye gaze in response to pictures that depicted orally-delivered declaratives with gender marked adjectives and particles. Unlike the university students, the low-educated adults did not shift their eye gaze to the target upon hearing the adjective. Mishra et al. reject as an explanation the idea that speakers’ grammars differed; rather the tentative conclusion is that processing is faster for educated individuals and they suggest this is due to considerable time spent reading.

Dąbrowska (2012) is another foray into the relationship between education and linguistic competence which also focuses on native speakers. She argues that sub-university education is connected to acquisition of grammars which are dissimilar in important ways to the grammars of more educated members of a speech community, suggesting that the same reasoning applies to L2 learners. In the keynote special issue in which Dąbrowska appears, the 14 responses – most of which take issue with her conclusions – turn out to reveal a marked absence of generative-based research on the relationship of education and particularly literacy and the acquisition of linguistic competence.

Whether different grammars are acquired and/or processing differs qualitatively or quantitatively for low-educated non-native vs. non-native-speaking and native-speaking adults are questions that require both the replication of the above studies as well as newly designed studies. For low-educated L2 adults, answers are urgently needed. If we find that processing is simply slower, this means that basic literacy programs must consider materials used and hours of instruction provided for these learners. If we find that processing is not qualitatively different, both learners and practitioners can expect success with persistence.

V Undiscovered research on low-literate learners

Segundas Lenguas e Inmigiración (http://segundaslenguaseinmigracion.es) is a relatively new journal that represents work on the younger and older immigrant population in Spain and neighbouring Portugal. The existence of Segundas Lenguas e Inmigiración (likely unknown to most readers of this journal) serves to underscore two points: first, compared with mainstream SLA and its sociocultural and formal-linguistics-based camps, there is less protectionism among those who work with the population of low-literate adult immigrants. This can be observed in the array of the journal’s articles whose focus is on coping with the teaching of both younger and older immigrants from Latin America and – for adults – primarily unschooled, from sub-Saharan Africa. Reference is regularly made to the psycholinguistic aspects of reading development and to the acquisition of linguistic competence. Second, there is on-going research on naturalistic, low-educated L2 adults of which mainstream SLA is unaware because the findings appear in languages other than English, in local journals whose target readership is native-language L2 teachers.

Two examples are Álvarez Álvarez (2012) and Villanueva Roa and Ramírez Ortiz (2012) whose articles discuss the training of teachers and development of materials in relation to what is known about immigrants in Spain among whom are now Chinese, Moroccan and Senegalese low-educated adults. In Spain as elsewhere, there is scant basic research to inform educational practice. Here, however, there is potential for sharing in this multidisciplinary journal the now considerable work on the Spanish-speaking population in North America by the researchers involved.

VI The Low-Educated Second Language and Literacy Acquisition (LESLLA) forum

The establishment in 2005 of a forum to bring together researchers and practitioners was motivated by two related concerns mentioned above: (1) little was known about non-literate adults’ L2 reading development or about the developmental interaction of literacy and linguistic competence because (2) little attention was being paid to the language acquisition and literacy development of un/low-educated adult immigrants in comparison to that of school-age immigrants (see, for example, Genesee et al., 2006). The latter have regular opportunities to interact with native L2 speakers; for adult immigrants who resettle in a post-industrialized society after the age of compulsory schooling, there is far less native-speaker interaction, and there are fewer educational opportunities.

Adequate funding for basic skills education for uneducated adult immigrants past the age of compulsory schooling is not guaranteed in many countries, yet these are the immigrants who remain on the margins of society. If LESLLA’s success can be measured by hundreds of participants at annual symposia whose venues alternate between North America and Europe, concern (2) above is being successfully addressed. However, concern (1) remains: there are few studies of such learners’ acquisition of linguistic competence in parallel with their development of literacy. Hawkins’ (2001) assertion that L2 learners follow a predictable route of development largely independent of – among other things – type of exposure and educational background needs to continue to be questioned, treating literacy as a variable. The potential exists, as noted above, for fresh perspectives on a range of much-debated issues such as the status of inflectional morphology in the development of L2 syntax (Prévost and White, 2000) and the role of orthography in the development of L2 phonology (Bassetti, 2009).

VII Conclusions

The recruitment of low-educated immigrant adults as study participants represents a set of challenges unfamiliar to researchers who rely on middle-class, educated learners in primary or secondary schools of university language centres and departments. Generative SLA researchers can take advice from those who carry out ethnographic studies with adults from backgrounds very different from the researchers’ (some of whom are involved in the LESLLA forum), including those in socio-cultural SLA despite fundamentally different research paradigms. Access, along with ethical approval can be a barrier, but the more experience researchers have, the more smoothly such processes will go. Bringing low-educated immigrant adults into generative SLA requires more cooperation with the wider community than researchers have become used to. But this is what socially relevant research is all about.