Understanding Second Language Acquisition (2nd ed)

by Rod Ellis

  An extension of the ‘breken’ study (Kellerman1979) provides support for the idea of psychotypology. Kellerman compared Dutch learners’ judgements regarding the translatability of the ‘brekens’ into L2 German (a language close to Dutch) with their judgements regarding their translatability into L2 English (a language more distant from Dutch). The results showed that, in general, whereas they accepted the sentences in German, they sometimes rejected them in English. In other words, there was a clear effect for perceived language distance.

  Further evidence comes from Singleton’s (1987) case study of Philip, an English-speaking learner of French. Philip displayed a high level of communicative efficiency in French, despite the fact that he had minimal opportunities to learn it. He demonstrated well-informed notions about which languages would assist him most in learning French. He utilized his knowledge of Romance languages (i.e. those that were close to French) and was often able to attribute the forms he borrowed to a particular language. Clearly, learners do have well-defined perceptions regarding the similarities and differences between languages.

  However, learners’ psychotypology is not fixed. Rather, it is revised as they obtain more information about the target language. Kellerman (1979) showed that Dutch learners of German may start out with the assumption that the target language is very similar to their mother tongue, but later on come to adjust this perception as they recognize differences. According to Kellerman, learners’ psychotypologies interact with their intuitive feel for prototypicality. Their intuitions about prototypicality determine what learners are prepared to risk transferring, while their developing psychotypology determines what is actually transferred at any given moment.

  Contextual factors

  The likelihood of transfer occurring also depends on the context of learning. This influences the input the learner is exposed to and, crucially, the extent to which transfer is manifest in L2 implicit or explicit knowledge. A distinction can be made between the macro-contexts of learning (for example, classroom vs naturalistic) and the micro-contexts (for example, the learner’s interlocutor or the specific task used to elicit samples of language use).

  Macro-contextual influences

  There is abundant evidence that transfer effects occur in both informal (i.e. natural) and formal (i.e. instructed) settings. However, the setting can influence the likelihood and extent of the transfer effects. This influence is related to the extent to which the context caters to the development of implicit—as opposed to explicit—L2 knowledge and also the effect that the context has on the type of language use—i.e. whether it encourages monitoring, using explicit knowledge, or spontaneous communication, involving implicit knowledge.

  Odlin (1989) suggested that negative transfer is less common in classroom settings than in natural settings because, in the former, learners are more likely to treat L1 forms as intrusive and even stigmatized. This can be explained by the fact that learners employ their explicit knowledge of L2 forms and rules to monitor their use of the L2 and thus to inhibit transfer of first language forms and rules. However, this is only likely to occur when the conditions of language use permit monitoring: that is when learners have explicit knowledge of the relevant feature, have time to monitor, and are focused on accuracy (Krashen 1981). When the conditions call for spontaneous, meaning-based production, monitoring will be difficult and, as a result, learners may not be able to use their explicit L2 knowledge to inhibit transfer.

  The situation is more complicated in natural settings. In general, such settings cater to the acquisition and use of implicit knowledge, so negative transfer is likely. However, this will depend on whether the natural setting is of a ‘focused’ or ‘unfocused’ kind. A focused natural setting is one where the learners have a clear idea of what constitutes a language (i.e. they make a clear distinction between their first and second language). An unfocused setting is one where two (or more) languages are mixed without much concern for what is ‘grammatical’ or ‘ungrammatical’. Negative transfer is more likely to occur in unfocused than focused natural contexts.

  These macro-differences can explain the differences in the results obtained by Abdullah and Jackson (1998) on the one hand, and Sridhar and Sridhar (1986) on the other. The first of these studies examined a foreign language setting in Syria. Abdullah and Jackson reported that the learners were reluctant to transfer L1 idiomatic expressions. Sridhar and Sridhar, however, claimed that learners in India and Nigeria often create English idioms based on their L1. The formal foreign language setting inhibited negative transfer. The unfocused, natural setting promoted it.

  Generalizations regarding the effect of macro contexts on L1 transfer are dangerous, however. First, they do not take account of forms of transfer other than negative transfer. Positive transfer is likely in both formal and informal settings, even though the nature of the input learners receive may differ. For example, in formal settings learners are likely to be exposed to substantial written input, making it easier to detect similarities between the source and target languages (for example, cognates). On the other hand, avoidance may be less likely to occur in informal settings than in formal settings where errors are more stigmatized. As Jarvis and Pavlenko (2008) concluded, ‘the formality of the context does affect transfer, but it includes forces that both foster and constrain the occurrence of transfer’ (p. 208). For this reason, an examination of the effect of micro-contexts on L1 transfer is more revealing.

  Micro-contextual influences

  In Chapter 5, we noted that variationist sociolinguists distinguish between careful and vernacular styles. Tarone (1982) argued that negative transfer is likely to be more evident in learners’ careful style than in their vernacular style, on the grounds that when learners are paying greater attention to how they speak, they are more likely to make use of all their potential resources, including L1 knowledge. On the face of it, this contradicts the general claim that negative transfer effects are less likely to occur in formal, instructed contexts, which prioritize the careful style, than in natural contexts where the vernacular style is primary. Jarvis and Pavlenko (2008), however, suggested how these contradicting positions can be reconciled. They proposed that instructed L2 learners have two goals—(1) to avoid errors and (2) to make use of those linguistic forms that serve as markers of formality. If they prioritize (1), negative transfer is likely to be inhibited, but if they prioritize (2) they may elect to employ L1 features that are markers of formality (for example, they will draw on L1 pragmatic devices to appear polite). In other words, much depends on how they orientate to the use of the L2 and on their perception of the importance of social factors in whatever situation they find themselves.

  Evidence for this comes from Beebe (1980). She showed that learners do not always manifest the target-language variant more in their careful style. Sometimes they transfer forms from their L1, if these forms have prestige value in their speech community. Beebe found that her subjects (adult Thai learners of English) produced fewer instances of the target sound, /r/, in formal language use than in informal language use. This was because they used a prestige Thai variant of /r/, which they associated with formal language use in their own language.

  Pragmatic transfer of the kind Beebe investigated also depends on the specific situational context. Olshtain (1983) found that—in the case of an apology situation involving backing into someone’s car and causing damage—English speakers of L2 Hebrew apologised in much the same way as in their L1. In contrast, in a situation where they had insulted someone at a meeting, they did not transfer their first language strategies, but behaved in a similar way to native Hebrew speakers. She concluded that in a situation that calls for careful face-work transfer was more likely to occur than in a situation where politeness was deemed as less crucial.

  More generally, the nature of the task learners are asked to perform will affect whether or not transfer occurs. Translation tasks, for example, are more likely to elicit both more positive and more negative transfer th
an tasks that involve spontaneous production. Tasks that allow for learners’ use of their explicit L2 knowledge are likely to inhibit L1 transfer; conversely when performing tasks that require spontaneous production learners are more likely to draw on all their linguistic resources, including their L1, and thus transfer effects will be stronger.

  Performance-related versus learning-related transfer

  A key question regarding the effects of the macro- and micro-linguistic contextual factors on L1 transfer that we have considered in this section is whether the effects are just performance related, or also learning related—i.e. are the transfer effects simply a manifestation of performance conditions, or do they reflect the influence of the L1 on actual learning? Jarvis and Pavlenko (2008) argued that they are just performance related. However, the distinction between performance-related and learning-related effects is not easy to make, and may not be theoretically justified. I will discuss this point later in the chapter.

  Developmental factors

  In Chapter 4, I examined research that showed there are universal tendencies in L2 acquisition. This research suggests that the learner’s L1 plays only a minimal role and thus contradicts the research we have reviewed in the previous sections of this chapter. This contradiction, however, can be resolved if it can be shown that transfer works alongside other universal and developmental factors.

  While simplification and overgeneralization are universal features, the L1 does affect the extent to which these occur. All learners, for example, omit prepositions in the early stages of L2 acquisition, but as Schumann (1986a) found, this happens much less frequently in the case of Spanish learners of English than in the case of Chinese and Japanese learners. Whereas Spanish has prepositions that are broadly equivalent in meaning to English, Chinese and Japanese do not. Arguably, overgeneralization and transfer are manifestations of the same underlying psychological process—that of prior learning facilitating new learning. Taylor (1975), for example, documented a transition from L1 transfer to overgeneralization. The elementary and intermediate learners in his study made similar kinds of errors, but whereas the elementary learners made more transfer errors, the intermediate learners made more overgeneralization errors. In other words, beginners rely on what they know (their L1) but intermediate learners draw on what they have already learned about the L2.

  Some morpheme studies have provided evidence of L1 effects. Japanese learners of English, for example, have greater difficulty in acquiring plural -s than learners whose first language includes a plural marker. As we noted in Chapter 4, Hakuta (1976) found that a five-year-old Japanese girl acquired plural -s later than was predicted by the morpheme order reported in cross-sectional studies (for example, Dulay and Burt 1973). He attributed this to the lack of an equivalent morpheme in the learners’ first language. Jiang et al.’s (2011) study, which we considered earlier in this chapter, also testifies to the difficulty that Japanese learners have with plural -s in comparison to Russian learners whose L1 includes plural markers. Clearly, then, the morpheme order is not entirely ‘universal’. The ‘natural’ order of acquisition is disturbed in the case of grammatical morphemes that are meaning bearing in the L2, but have no equivalent in the L1. However, non-meaning bearing morphemes—such as English third-person singular—appear to be equally difficult for all learners.

  The first language can also impact on the sequence of acquisition of specific target language features. However, its impact appears to be developmentally constrained: that is, it only occurs if and when learners have reached a stage of development which can provide a ‘crucial similarity measure’ (Wode 1976). In the case of negation, for instance, the German children that Wode studied began, like other learners, with pre-verbal negation (i.e. ‘no’ + verb). But later—after they had discovered the negative particle could follow the verb ‘be’ in English (for example, ‘Maria is not well’)—they began to produce post-verbal negation in L2 English, but extended this to main verb negation (for example, ‘Mary comes not’) as in German. In other words, once they obtained evidence to suggest that post-verbal negation was possible in English they transferred the German rule for negation.

  Further evidence of developmental constraints on transfer comes from studies based on Processability Theory (see Chapter 4 and Chapter 9). According to this theory, the processing operations responsible for the sequence of acquisition apply to all learners. It would seem, then, that there is no place for language transfer in Processability Theory. In fact, Pienemann (2005) recognized that transfer does have a role, but one that it is developmentally moderated. The fact that the L2 has to be ‘reconstructed’—i.e. learners need to master each processing operation from the beginning for each language they learn—can block transfer occurring. Thus, for example, Polish learners of L2 English hold no advantage over Vietnamese learners when it comes to subject-verb agreement marking, even though Polish has such marking and Vietnamese does not. This is because the processing procedure required for this grammatical feature is developed late and thus the presence of the same procedure in the first language confers no advantage. However, once a specific procedure has been developed for the L2, the L1 can come into play.

  However, not all theories support Pienemann’s contention that L1 transfer is constrained by the processability of grammatical structures in the L2. As we will see in Chapter 8, other theories propose that L2 learning is a continuous process of restructuring: initially, learners are strongly influenced by their L1 and only learn to attend to target language forms gradually, as a result of exposure to them. Taylor’s (1975) study, mentioned above, lends support to such a position.

  These different positions are reflected in two different principles that govern L1 transfer. Andersen (1983) formulated the Transfer to Somewhere Principle, which proposes that some degree of congruity between the first language and the target language is needed for transfer to take place. The ‘somewhere’ concerns the learning that results from the application of universal developmental principles. Kellerman (1995), however, advanced the Transfer to Nowhere Principle to reflect the fact that ‘there can be transfer which is not licensed by similarity to the L2, and where the way the L2 works may very largely go unheeded’ (p. 137). In other words, transfer is not necessarily developmentally constrained. The evidence suggests that these two principles are best seen as complementary. L1 transfer can occur at any time, but in some cases, it will only be evident when the learner has reached the requisite stage of development.

  Finally, it should be noted that the effects of transfer can be seen not just in the linguistic forms observed in interlanguage, but also in the rate of development and ultimate attainment. The L1 can speed up or retard the development of an L2. Where there is similarity between the source and target languages, positive transfer can occur and learners can make rapid progress. Conversely, where there is difference, acquisition can be slowed down. Spanish learners of English, for example, tend to spend longer on the early ‘no’ + verb stage when acquiring English negation than German learners. This can be explained by the fact that this early stage corresponds to the L1 negative construction in Spanish, but not in German. Similarly, learners whose L1 includes an article system—for example, Spanish learners of English—are more likely to achieve a high level of accuracy in this system than learners whose L1 lacks an article system—for example, Japanese learners of English.

  Individual factors

  In Chapter 2 and Chapter 3, we saw that learners’ starting age and various psychological factors—such as language aptitude and motivation—affect L2 learning, in particular the rate of learning and ultimate attainment. Here, we will consider whether these same factors moderate the nature and extent of transfer, focusing on age and language aptitude. There are almost no studies that have investigated motivation and transfer and those that have done so have reported no difference in transfer effects for more and less motivated learners.

  Age and L1 transfer

  Learners who start learning an L2
at a young age generally achieve a more native-like accent than those who start later. Some child learners—especially those who start before the age of three—become totally native-like in their pronunciation of the L2. This can be accounted for—in part, at least—by the fact that younger learners rely less on their first language and exhibit less transfer than older learners. As Guion, Flege, Lieu, and Yeni-Komshian (2000) put it, ‘the more established the L1 is at the time of L2 acquisition, the greater the influence it will have on the L2’ (p. 206). All learners have two primary sources of information to draw on—their first language and the second language input they are exposed to. If the L1 is not yet firmly established, it is less available as a resource for learning so learners will need to depend more on L2 input. Thus, one explanation for why children starting before the age of three are able to develop a totally native-like pronunciation is that their acquisition of target language sounds is not blocked by their L1.

  Transfer of L1 grammatical forms is also less likely in younger than in older learners. Older learners may have an advantage when the source and target forms are similar and positive transfer occurs. However, when they are different, it can impede acquisition. Czinglar (2012) reported an interesting longitudinal study of two untutored Russian learners of L2 German. One was aged seven and the other 14. She focused on the acquisition of German word order rules—see Chapter 4, Table 4.3. She found that the two learners differed in their acquisition of the difficult verb-end rule, which has no equivalent in Russian. The younger learner fully acquired this rule in nine months, but the older learner took much longer. Czinglar argued that this difference in the rate of learning could be explained by the fact that the younger learner did not form hypotheses about German verb placement on the basis of the L1 whereas the older learner did. Once again, then, age can be seen to influence the extent to which learners depend on the L1 or L2 input.