by Rod Ellis
Looking back over some 50 years of SLA research, it is clear that there have been huge advances in the methods researchers have employed, reflecting the general tendency to look beyond SLA itself and borrow from other fields, in particular from cognitive psychology and branches of sociology.
The goal of SLA
Given the disputes over the boundaries of the discipline, it is not surprising that there is also no clear agreement over the goal of SLA. For much of the history of the discipline, the goal has been the description and explanation of how people acquire a second language, viewed as separate and distinct from the first language. This is how I defined it in Chapter 1—L2 acquisition is the ‘learning of any language after the first’. I also noted, however, that there are moves to reframe it as the study of bi/multilingualism. Ortega (2012) argued that L2 acquisition should not just be contrasted with L1 monolingual acquisition from birth, but also with bilingual acquisition from birth. She made the case for what she called the ‘bi/multilingual turn for SLA’—that is, making the goal of SLA not just the study of a second language, but of how later-learned languages figure in making a person bi- or multilingual. In a similar vein, Cook (1991) has argued for investigating multicompetence on the grounds that the L1 and the L2 should not be treated as separate linguistic systems but as intertwined, each affecting the other. In general, however, SLA has continued to focus on how learners acquire a second language.
For some SLA researchers, SLA is seen as ‘central to the wider goal of understanding the ontogeny of the human language capacity’ (Ortega 2012: 8). In other words, SLA is seen as contributing alongside other language sciences to an explanation of the special human capacity for language—what this consists of; how it has evolved; and how people draw on it when acquiring a specific language. This has not been a major goal of this book, but there are several areas of research I have covered that are relevant to it. In Chapter 2, we examined the critical period hypothesis, which addresses the key issue of whether the human language faculty is still available for late-starting L2 learners. Chapter 4 reported research that had investigated learner varieties, which—according to Klein (1998)—are a reflection of our innate, genetically endowed capacity for learning a language. Other branches of SLA have also proposed that language learning is only possible because of a highly specialized capacity for language—see, for example, the account of Universal Grammar in Chapter 9. In contrast, the connectionist view of language outlined in Chapter 1 proposes that there is no special language faculty and that language learning depends on the same cognitive abilities as those involved in all learning and that language learning is primarily driven by learners’ experiences of language. By and large, then, just as there is no consensus about the boundaries of SLA, so there is no agreement about what the human language capacity consists of. The contribution that SLA has made to our understanding of this issue, however, is not negligible as it has served as a useful testing ground for the competing nativist and environmentalist theories of language.
There is another goal that has figured strongly in SLA: to provide guidance about how second languages can be most effectively taught. This was the goal that motivated much of the early research in SLA. It is premised on the reasonable assumption that for language instruction to be effective, it must take account of how learners acquire a language. As Long (2006) noted:
Many SLA researchers have witnessed firsthand the relatively few successes and the widespread failures of even the best-intentioned classroom instruction, and many were first motivated to undergo training as SLA researchers with a view to improving that state of affairs.
(Long 2006: 156)
The hope was that SLA would help to identify ways of teaching that were compatible with how learners learn. In Chapter 10 and Chapter 11, we examined a number of proposals for language instruction that have emanated from SLA and have been empirically investigated by SLA researchers. In this respect, SLA constitutes an applied rather than a pure discipline. It aims to utilize theoretical understanding of L2 acquisition and the results of empirical research to propose instructional approaches that will foster learning. However, exactly how SLA can best inform language pedagogy remains uncertain, and is the subject of the final section in this chapter. First, though, I will attempt a summary of what SLA—as a discipline—has told us about the nature of L2 acquisition.
What do we know about L2 acquisition?
The conclusion chapter to the previous edition of this book provided a summary of the main theories of L2 acquisition. However, such an approach no longer seems appropriate as I have already examined the key theories in earlier chapters. Ideally, I would like to conclude this book with a general theory of L2 acquisition but—as will have become clear—this is not possible given the diversity of the theoretical positions on offer and their fundamental epistemological differences as, for example, in how ‘acquisition’ is conceived in the cognitive and social theories discussed in Chapter 8 and Chapter 9. Instead, therefore, and at the risk of over-simplification, I will attempt to formulate a set of general statements that encompass what SLA has shown us about the nature of L2 acquisition.
L2 acquisition is complex
All theories acknowledge the complexity of L2 acquisition. It is complex because a multitude of factors influence how an L2 develops: individual difference factors, such as language aptitude and motivation; the learner’s L1; the nature of the input that learners are exposed to; the kinds of interactions they participate in; the social context in which learning takes place; and whether or not form-focused instruction is available. These factors interact in different ways with different learners, influencing the rate at which learning takes place, the ultimate level of achievement, and also the process of acquisition itself. The sheer complexity of the factors involved and of the interactions among them is the main reason why no general theory of L2 acquisition has been forthcoming. Instead—as we have seen—theories have been developed to account for the influence of specific factors.
Language itself is a complex adaptive system (Larsen-Freeman and Cameron 2008). It incorporates phonological, lexical, and grammatical subsystems which are interlocked in complex ways. These linguistic systems are connected to meaning systems that enable us to behave in pragmatically appropriate ways. The process of acquiring this complex adaptive system is a slow and gradual one. The initial L2 system is a simple one—it consists of formulaic sequences and vocabulary, which learners use as best they can to serve their communicative needs. Gradually, as the system becomes more complex, grammar emerges. This process is not a linear one. Learners do not accumulate discrete grammatical elements in a neat and ordered way. As new linguistic forms are internalized, earlier acquired forms are re-organized and the developing system is restructured. Development is dynamic, uneven, and at times chaotic. Unsurprisingly, acquiring an L2 involves hard work and persistence on the part of the learner—as Hatch (1978) pointed out many years ago.
There are universal patterns of L2 acquisition
In Chapter 4, I reviewed research that testifies to a relative fixed order of acquisition of grammatical morphemes, such as verb + -ing and third-person -s, and to sequences of acquisition involving identifiable stages in the development of grammatical structures, such as negation. Klein and Perdue (1997) documented remarkable consistency in the language varieties that learners construct as they progress from the pre-basic variety to the basic variety and finally to the post-basic variety. Pienemann, Johnston, and Brindley (1988) drew on earlier research to show that there was an invariant order in the acquisition of different English grammatical structures. This and other research points to universal patterns in the acquisition of a second language. Long (1990) considered this one of the ‘well-established findings’ that a theory of L2 acquisition needs to explain:
… learners of different ages, with and without instruction, in foreign and second language settings, follow similar developmental sequences for such items as English negation.
(Long 1990: 659)
However, not all SLA researchers are in agreement with what Lantolf called the Universal Acquisition Hypothesis. It is clearly incompatible with sociocultural theory which emphasizes the importance of mediation in enabling learners to overcome the limitations of their own cognitive resources. For example, the theory claims that mediation in the form of language instruction can help learners ‘beat’ the natural route of developmentNOTE 1. The hypothesis is also disputed by Dynamic Systems Theory, which views learning as unpredictable with individual learners following distinctively different trajectories (see Chapter 5). Usage-based approaches to investigating L2 acquisition—for example, Eskildsen (2012)—also question the existence of universal sequences for specific grammatical structures. However, they do recognize that there is a general pattern of development common to all learners involving a ‘developmental sequence from formula, through low-scope pattern, to construction’ (N. Ellis 2002: 170).
These differences can be reconciled if we accept that the Universal Acquisition Hypothesis constitutes an idealization and serves as a useful simplification of the complexity evident in the real system that is empirically adequate within accepted limits (Ellis 2014). There are clearly regularities in how all learners acquire the linguistic systems of an L2. In general, learners do follow a similar developmental pattern and it is important to acknowledge this.
L2 systems are variable
At any stage of development, learners deploy a variety of forms for realizing a particular grammatical structure. For instance, a learner may have begun to produce negative utterances containing an auxiliary—for example, ‘Maria is not coming today’—but at the same time continues to use the earlier acquired ‘no’ + verb construction as in, ‘Maria no coming today’. Much of the variability in learner language is systematic: that is, it is possible to identify the factors that predict which forms a learner will deploy in accordance with the situational or linguistic context. More advanced constructions are likely to occur in the learner’s careful style when attention is paid to form; earlier acquired forms occur more frequently in the vernacular style where the focus is on spontaneous communication. Some linguistic contexts elicit the use of the more advanced form, others the use of the more primitive form. Learners also make systematic use of their variable linguistic resources to realize different functions. Schachter (1986), for example, observed that one learner systematically used the ‘no’ + verb form to express denial and ‘I don’t know’ for the ‘no information’ function. Learners deploy their resources communicatively, mapping whatever forms are available to them onto the functions they need to perform. Over time the configuration of these form-function mappings changes.
Not all L2 variability is systematic, however. When new forms first enter the L2 system they are likely to be used as alternatives to existing forms. The result is free variation. Researchers working within the framework of Dynamic Systems Theory have documented the chaotic nature of learner language. They emphasize that those points at which variation appears most chaotic are precisely the times when development is likely to occur. Learners look to maximize the efficiency of their linguistic resources by rapidly organizing and then re-organizing them into functional systems that serve their communicative purposes.
To claim that L2 systems are variable is, at first sight, to deny of the existence of a natural route of L2 development. However, the fact that L2 systems are variable does not necessarily contradict the claim that there are orders and sequences of acquisition. Some grammatical forms do typically emerge before others. All learners, for example, start on the path to acquiring English negation with the ‘no’ + verb construction. Over time, early forms are supplanted by later forms, but this is a gradual process with the result that the variable use of the different forms is evident at any particular stage of development.
Processability Theory (see Chapter 8) seeks to account for both developmental regularities and variability. It proposes that some features are developmental in nature and are acquired in sequence as learners master a set of hierarchically ordered processing procedures. As each procedure is mastered, a range of grammatical structures governed by that procedure become available. But individual learners may not demonstrate immediate acquisition of all the structures associated with a procedure. There is individual variation in the structures activated by the procedure. The theory also proposes that some grammatical features—for example copula be—are not subject to processing constraints and thus will be acquired by learners at different times. This theory, then, affords an account of both the universal and variable aspects of acquisition.
The effects of the L1 are pervasive
The learner’s L1 influences all aspects of L2 acquisition—phonology, lexis, grammar, and pragmatics. It is also evident in all stages of L2 development. Transfer—both negative and positive transfer—is linguistic, semantic, and conceptual in nature. Linguistic transfer takes place when learners incorporate an L1 structure into their L2 system as, for example, when Spanish learners of English persist in using the ‘no’ + verb negation structure over time because it is the same as the L1 structure. Semantic transfer involves the mapping of a concept shared by two languages onto a translation equivalent. as when Finnish learners of English use ‘tongue’ in place of ‘language’. Conceptual transfer occurs when the concept itself differs in the two languages as for example when an English learner of Russian refers to a paper cup as ‘chaska’ (= cup) rather than ‘stakanchik’ (= little glass) as a result of differences in how the two languages conceptualize ‘cup’ and ‘little glass’. Transfer is also evident in the gestures that learners employ when speaking.
Much of the research on L1 transfer has focused on the issue that Kellerman (1983) identified many years ago—’now you see it, now you don’t’. Various factors have been shown to influence when transfer occurs: the relative markedness of the equivalent L1 and L2 forms; the distance between the two languages; the learners’ intuitions about the transferability of specific linguistic forms; the context in which learning takes place (for example, the formal language classroom may inhibit transfer); the learner’s age (for example, older learners tend to transfer more than younger ones); and psychological factors (for example, learners with larger working memories are better able to overcome negative transfer).
The pervasiveness of transfer effects raises once again the legitimacy of claiming universal patterns in L2 acquisition. At the most general level, however, L1 transfer affirms universality: all learners do it. The strength of L1 representations blocks attention to L2 forms in the input. More interestingly, however, research has shown that transfer effects are themselves developmental. They interact with the universal processes responsible for the relatively fixed orders and sequences of grammatical features in L2 development. Transfer tends to occur when learners have reached a stage of ‘natural’ development that allows them to access an L1 form that is similar or equivalent to an L2 stage. Thus—in recognizing the prevalence of transfer—it is not necessary to abandon the idealization that L2 acquisition is subject to universal constraints.
The learner’s starting age plays an important role
The differential achievement of learners who commence learning at different ages—i.e. as young children or as adolescents/adults has been an abiding topic of interest in SLA (see Chapter 2). Granena and Long (2013) reported that there have been over 100 empirical studies published during the last 50 years that have investigated this issue. This reflects its importance for both understanding the nature of the human capacity for language and for educational decision making.
The research demonstrates conclusively that starting earlier results in higher levels of L2 proficiency. The crucial issue, however, is whether learning needs to start before a certain age for learners to achieve native-like ability. According to the Critical Period Hypothesis native levels of ability are only possible if learning commences within a critical period. However, the existence of a ‘critical’ period
is disputed on the grounds that there is no clear discontinuity between learning during the period and after it ends, but rather a gradual falling-off in abilityNOTE 2. For this reason, many researchers prefer to talk about a ‘sensitive’ period. That is, they claim that children are more likely to achieve native-like ability but acknowledge that learners who start as adults may also sometimes succeed in doing so. Long (2013) defined the sensitive period as consisting of a ‘peak’—i.e. a period of intense sensitivity to language that starts at age zero—followed by an ‘offset’ (a period of gradual but not sharp decline) and thereafter a sharper ‘discontinuity’ in language learning ability. Thus, for example, learners who start learning before the age of six are likely to achieve a native-like pronunciation and can still do so if they start between the ages of six and 12, but are very unlikely to do so after that.
Why are children generally more successful in learning an L2 than adults? There is no consensus on the answer to this question. The explanation may be social—children are more likely to seek membership of a native-speaking peer group. It is tempting, however, to look for a cognitive explanation. Children are more likely—and perhaps better equipped—to engage in implicit learning while adults are more likely to learn explicitly. Even if adult learners still possess the capacity for implicit learning—and there are reasons for believing that this is the case—this ability may decline with age. Also, their greater analytical skills can interfere with ‘natural’ learning.