Understanding Second Language Acquisition (2nd ed)

by Rod Ellis

  Correlational studies

  Much of the learning-strategy research has used questionnaires (for example, Oxford’s, 1990, Strategy Inventory for Language Learning). These ask learners to self-report which specific strategies they use or how frequently they use them. Quantitative scores derived from these questionnaires are then correlated with measures of L2 achievement/proficiency. Studies that have employed such an approach (for example, Wharton 2000) generally report that higher-proficiency learners use more strategies than lower proficiency learners. Such studies are inherently problematic, however, as it is impossible to tell whether it is strategy use that causes learning or whether the ability to use particular strategies is dependent on proficiency. For example, more advanced learners have been found to make greater use of metacognitive strategies (Macaro 2006), but this might simply be because their greater language proficiency gives them access to these strategies

  Effects of strategy instruction

  More interesting is the research that has investigated the effects of strategy instruction on learning. Given the uncertainty about which strategies are important for learning, perhaps not much should be expected of such studies. Another reason for being doubtful is the lack of standardization in both the instructional packages and the manner in which learning was assessed (Macaro 2006). However, there is evidence that strategy instruction is effective.

  Hassan et al. (2005) carried out a review of 25 strategy training studies and concluded that 17 reported positive results, five mixed results, and only two negative results. However, the effectiveness of the training varied according to the different language skills, with the most robust effect in reading comprehension and writing skills. It was not possible to determine which types of strategy training or which particular training techniques were effective. The studies investigated mainly non-school populations so Hassan et al. did not comment on the effectiveness of strategy training with school-based learners. Nor were they able to determine whether the learners’ stage of development was a significant factor influencing the success of training.

  A more comprehensive, meta-analytic review of strategy instruction studies (Plonsky 2011) provides a clearer endorsement of strategy training. Plonsky’s meta-analysis included 61 studies involving 6,791 learners in his meta-analysis. He calculated the average ‘effect size’ of the strategy instruction in all the studies and also investigated how moderating variables such as the age of the learners, the kind of the strategy instruction, and the type of learning outcome impacted on the overall effect. The overall effect size was ‘small to medium’. He argued that this compared favourably with the effect sizes reported in meta-analyses of strategy training studies in general education, but was modest compared to meta-analyses of other types of L2 instruction (see Chapter 10). He found that instruction directed at cognitive strategies was more effective than instruction directed at metacognitive strategies. This casts some doubt on the emphasis that other researchers have placed on metacognitive strategies. The effects of the strategy instruction were clearly evident in the case of reading, writing, speaking, vocabulary, and pronunciation, but weak or not evident at all in listening, grammar, or general proficiency.

  There are some major theoretical problems with learning strategy research. Dörnyei (2005) noted that whereas research on learning strategies was popular in general educational research in the 1980s, it declined dramatically in the 1990s, and that language learning strategies should be re-conceptualized and investigated in terms of self-regulation. Tseng, Dörnyei, and Schmitt (2006) argued that an approach based on self-regulation provides a more satisfactory way of empowering learners than traditional strategy training as the real goal should be that of assisting learners to achieve self-regulation, not to use specific strategies. They developed an instrument to measure self-regulatory capacity for vocabulary learning.

  Another rather surprising limitation is the failure of research on learning strategies to relate strategies to the processes that SLA researchers have proposed are important for learning. An exception is Purpura (1999). Purpura distinguished strategies in terms of whether they related to comprehending processes, storing/memory processes, or using/retrieval processes. The advantage of such an approach is that it aligns the study of learning strategies with mainstream thinking about the nature of language learning and would provide learning strategy research with a stronger theoretical base. It might be possible to identify not just which specific strategies are important for learning, but also when, and in what way.

  Age and psychological factors

  It is interesting to consider whether the impact of the psychological factors we have considered in this chapter depends on the age of the learners. In the case of language aptitude, there is clear evidence that it does. Whereas short-term phonological memory may be of special importance for young learners, language analytical ability appears to be more important for older learners. Granena and Long (2012) showed that language aptitude mitigates the negative effects that increasing age has on the acquisition of lexical knowledge.

  Age and motivation are also related. Kormos and Csizér (2008) investigated Hungarian school pupils, university students, and adult language learners and found that there were differences in what motivated these three groups. The school learners were primarily motivated by interest in English-language cultural products whereas the two older groups were more influenced by their attitudes to English in the globalized world. Relationships have also been found between age and language anxiety, with older learners tending to be more anxious than younger learners (for example, Onwuegbuzie, Bailey, and Daley 1999). Not surprisingly, given their mediating role, the learning strategies that learners prefer vary with age. Age differences in the speed of learning and ultimate attainment that we noted in the previous chapter may be—in part at least—traceable to the part played by key psychological factors.

  Conclusion

  I have elected to examine the different psychological factors separately, reflecting the general approach adopted by researchers. However, the danger of such an approach is that it loses sight of the ‘whole’ learner. As Tudor (2001) pointed out, learners are ‘complex human beings who bring with them to the classroom their own individual personality’ (p. 14). Dörnyei (2010), too, is now critical of a modular approach to investigating individual differences on the grounds that the factors exert a cumulative effect that is not simply the sum of their individual effects and also they are continuously fluctuating in nature.

  There is an interesting branch of early research into individual differences that I did not consider in this chapter, but that can be seen as addressing the ‘whole’ learner. The ‘good language learner’ studies (for example, Naiman, Fröhlich, Stern, and Todesco 1978) sought to identify the characteristics of successful learners mainly in terms of the learning strategies that they reported using. These indicated five major aspects of successful learners: (1) a concern for language form; (2) a concern for communication; (3) an active task approach; (4) awareness of the learning process; and (5) flexible use of learning strategies. These studies, however, took little account of the role of language aptitude, motivation, or language anxiety. A more complete picture of the efficacious learner could perhaps incorporate the insights provided by research into these three areas.

  One way of viewing how the ‘parts’ make up the ‘whole’ might be to consider what constitutes self-efficacy. The construct of the ‘ideal language self’ can also be seen as a composite construct that involves more than just motivation. Dörnyei (2010) argued that ‘the effective functioning of the ideal L2 self is dependent on the operation of several underlying cognitive components, most notably on the learners’ self-appraisal of their capabilities and evaluation of the affordances of their personal circumstances in order to anchor their vision in a sense of realistic expectations’ (pp. 257–8). In other words, there is a cognitive and emotional dimension to this construct.

  I have also not considered how motivation is trea
ted in neuropsychological SLA which seeks to relate the mental processes involved in language learning to brain activity. Interestingly, this approach also points to the need to recognize the interconnections between emotion and cognition. Schumann’s (1997, 2004b) neurobiological theory is built around the twin notions of ‘stimulus appraisal’ and ‘mental foraging’. Learners appraise the stimuli they are exposed to in terms of their novelty, pleasantness, goal/need significance, coping potential (i.e. whether they feel able to cope), and self-social image (i.e. whether the stimuli conforms to the individual’s sense of social and personal values). The appraisals they form are stored in memory. Mental foraging is the search for knowledge, which Schumann suggests involves the same neural mechanisms as foraging for food or for a mate. Learners are more inclined to engage in mental foraging if the appraisals they have stored provide them with an incentive for doing so. Learner differences arise as a result of differences in stored stimuli-appraisals and in the learners’ preparedness for mental foraging. From the perspective of the brain, then, no clear distinction can be made between cognitive and affective memories and the actions of learners. This again points to the need to examine differences in terms of the whole learner.

  The challenge facing researchers who wish to investigate the ‘whole’ learner is to determine how this can be undertaken. It is clearly much easier to investigate the effect of specific factors such as language aptitude, motivation, and language anxiety as there are instruments available to measure each. Dörnyei (2010) proposed applying Dynamic Systems Theory to the study of individual differences. This involves investigating the interactive effect of different factors and how this effect changes over time. Clearly, this is a challenging undertaking. Dynamic Systems Theory is considered further in Chapter 5 and Chapter 9.

  Notes

  1 Dörnyei and Skehan (2003), in their review of psychological factors in SLA, considered that only two factors—language aptitude and motivation—have the ‘capacity to generate research programs’ (p. 622). I have included language anxiety as a separate factor as research has consistently shown it is also an influential factor.

  2 Pimsleur (1966) produced a rival language aptitude test—the Pimsleur Language Aptitude Battery (PLAB). This was designed for use with younger learners and emphasized auditory ability more. It also included a test of inductive language learning ability, which was missing from the MLAT.

  3 A very similar construct to Dörnyei’s (2009) Ideal L2 Self is Yashima’s (2002) international posture. This was defined as ‘a general attitude towards the international community that influences English learning and communication among Japanese learners’ (pp. 62–3). The Ideal L2 Self construct is, however, broader—international posture being just one ‘possible self’ that may motivate a learner.

  4

  The development of a second language

  Introduction

  In Chapter 1, I introduced the term interlanguage to refer to the mental system of a second language (L2) that a learner constructs and that is different from the target language system. We can talk about ‘an interlanguage’ to refer to the system that a learner has constructed at a particular point of time; we can also talk about the interlanguage continuum to refer to the series of systems that the learner constructs over time. Our concern in this chapter is with the interlanguage continuum—that is, with how the interlanguage system is constructed and reconstructed as acquisition takes place. By and large, the research that has investigated this has focused on how learners acquire the grammar of a second language, but there have also been studies of the acquisition of pronunciation, vocabulary, and pragmatic features (for example, requests and apologies).

  I will begin by considering the methods researchers have used to investigate interlanguage development and then go on to examine what the research has shown about the nature of L2 development. A key issue in this chapter is whether it is possible to identify a ‘natural route’ that is common to all learners of a second language.

  Order of acquisition, sequence of acquisition, and usage-based accounts of L2 development

  When we talk about interlanguage development we are necessarily concerned with ‘change’ (i.e. how a learner’s interlanguage is modified over time). A key issue, however, is how this ‘change’ is conceptualized. We will consider three ways of looking at this—in terms of the ‘order of acquisition’, ‘sequence of acquisition’, and the more recent usage-based accounts of ‘learning trajectory’.

  Order of acquisition

  To determine the order of acquisition it is necessary to investigate when learners achieve mastery of different linguistic or pragmatic features. Mastery is defined in terms of the learner’s ability to produce specific grammatical forms accurately (i.e. in accordance with target language norms). As native speakers do not typically achieve 100% accuracy all the time, especially when speaking, researchers have taken the 80% or 90% criterion level as indicating mastery (i.e. if learners use a specific feature accurately at least 80% of the time they are considered to have mastered it).

  The ideal way to investigate order of acquisition is in longitudinal studies which show when learners reach the 80% criterion level for different linguistic features. However, many studies have been cross-sectional (i.e. they collect samples of learner language at just one time) and determine the order of acquisition by equating it with the accuracy order: they assume that if one feature has reached the criterion before another it must have been acquired earlier. However, as we will see, there are reasons for believing that accuracy order may not be a valid way of investigating acquisition order.

  To determine the level of accuracy of different grammatical features researchers make use of obligatory occasion analysis. This involves identifying contexts that require the use of a specific linguistic feature (for example, English plural -s) and then establishing whether learners supply this feature in these contextsNOTE 1. For example, if a learner says ‘*My father live in London’, an obligatory occasion has been created for use of third-person -s—but in this case, it has not been supplied. But if a learner says ‘My father lives in London’, it has been correctly supplied.

  This approach to investigating interlanguage development falls foul of the comparative fallacy. Bley-Vroman (1983) argued that interlanguages are unique systems and should not be investigated in terms of whether learners have achieved native-like mastery. Other researchers (for example, Seidlhofer 2001) have been even more critical of treating L2 systems as failed approximations of native language systems. They point out that in multilingual situations in particular, learners are more likely to be concerned with acquiring the diverse linguistic resources needed to function effectively in their speech community rather than with trying to speak like a native speaker. However, despite these objections, much of the SLA research—both old and more recent—has investigated interlanguage development in terms of accuracy.

  Sequence of acquisition

  An alternative is to investigate the sequence of acquisition. Both syntactic features (for example, negation) and morphological features (for example, English regular past tense) are acquired gradually, with learners passing through a number of stages of development. Thus, development is determined not in terms of target-like accuracy, but in terms of whether there is evidence of the learner progressing from an early stage to a later one.

  This requires undertaking a frequency analysis. Ellis and Barkhuizen (2005) provide a detailed account of this type of analysis. It involves selecting a specific linguistic feature for study (for example, negatives) and then identifying occasions when learners attempt to use this feature and documenting the various linguistic devices they employ and how these change over time. Such analyses can show how learners shift from the predominant use of one device at one time (for example, negative + verb as in ‘*No coming today’) to the use of another at another time (for example, ‘don’t’ + verb as in ‘*I don’t coming today’). The advantage of investigating sequence of acquisition is that it
can show that development is taking place even if the learner has not achieved target-like use.

  Usage-based accounts of learning

  Frequency analysis figured in much of the early research (for example, Cancino, Rosansky, and Schumann 1978). However, it is not without problems. One is the well-attested phenomenon of formulaic speech. Learner language, especially in the early stages, is characterized by the use of formulaic sequences such as ‘I don’t understand’ and ‘I don’t know’. Such chunks appear to show that the learner has reached a relatively late stage in the development of negation, but in fact they show nothing of the kind because the learner has simply learned some fixed lexical units. For this reason, researchers investigating the sequence of acquisition only examine learners’ creative speech (i.e. utterances that have been constructed out of separate linguistic units). The problem here is that it is not always easy to tell whether an utterance constitutes a chunk or has been creatively constructedNOTE 2.

  Usage-based theories of second (and first) language development (N. Ellis 2002) reject the view that chunks and creative constructions constitute a simple dichotomy. They claim that linguistic knowledge develops in response to the frequency of exposure to and use of exemplars of the language. Development involves a continuum from words and fixed expressions on the one hand to schematic templates that allow for utterances to be creatively constructed on the other. The continuum stretches from words and multiword chunks to partially analysed schemas or patterns and then on to increasingly more generalized schematic constructions. In all linguistic systems, these different representations co-exist so that an utterance such as ‘I don’t know’ might be accessed as a whole or be partially or fully constructed.