Understanding Second Language Acquisition (2nd ed)

by Rod Ellis

  Sequences of acquisition revisited

  In Chapter 4, we considered evidence suggesting that the acquisition of grammatical sub-systems—such as negatives—characterized by relatively well-defined stages, and that sequences of acquisition are universal. To what extent is this finding compatible with what we have discovered about the nature of variability in learner language?

  Variationist research based on the Labovian and the Dynamic Paradigms does not challenge the existence of developmental sequences. Rather, it affords a more elaborate account of how one variable stage evolves into another variable stage. Stage-like development is not just a matter of learners acquiring a new linguistic form, or of making one form the dominant choice, but of gradually extending (or restricting) the use of the forms at their disposal in accordance with the social, linguistic, and functional norms of the variety the learner is targeting. Stage-like development, then, is reflected in variability within a stage and in shifting patterns of variability from one stage to the next. Table 5.2 is an attempt to capture this as a set of general stages that characterize the acquisition of specific grammatical features.

  But do all learners manifest the same developmental sequence? How universal are the stages of development proposed in Table 5.2? The assumption of researchers in the Labovian and Dynamic Paradigms is that the constraints on the use of linguistics forms, the implicational hierarchies evident in the spread of forms from one variety to another, and the way in which form-function systems are re-organized apply to all learners. It is this assumption that Dynamic Systems Theory questions. Dijk et al. (2011) concluded that ‘no individual behaves as the average person and that all individuals have their own patterns’ (p. 83). From this theoretical perspective, then, we ‘would expect different kinds of variability to show different kinds of development’ (p. 84). In other words, there are no universal sequences of acquisition.

  It is not easy to reconcile these different positions. They are based on different theoretical assumptions and have been researched using different methodological procedures. However, I find it difficult to accept that there are not major commonalities in the way in which all learners acquire a second language. Individual differences in acquisition sequences certainly do occur—in particular due to the influence of the learner’s first language (see Chapter 6)—but the claim that there are universal paths in L2 acquisition constitutes a useful idealization.

  Sociolinguistic competence

  In this chapter, we have examined both learners’ variable use of interlanguage and target-language forms (i.e. Type 1 variation) and their variable use of target-language variants (Type 2 variation). The studies of Type 2 variation are of special value because they remind us that target languages are themselves variable and that—if the learner’s goal is to attain target-language norms—this requires not just the ability to use target language forms correctly, but to use them in sociolinguistically appropriate ways. The study of variability in learner language, therefore, affords insights as to how learners develop sociolinguistic competence. The studies we have considered indicate that even very advanced L2 learners (especially those who started post-puberty or who were dependent on classroom instruction for L2 input) often fail to achieve full mastery of the sociolinguistic norms of the target languageNOTE 5.

  Sources of variation

  In this chapter, I have focused on the social and linguistic factors that can account for interlanguage variation. There is, of course, much more to the study of variability. In particular, psycholinguistic factors related to whether learners have the opportunity to plan before or as they speak will impact on their choice of linguistic forms. Another psycholinguistic factor that has been shown to result in variability in learner language is monitoring (i.e. the use of explicit knowledge to edit output). I will consider planning and monitoring more fully in later chapters.

  Notes

  1 Tarone and Parrish (1988) provide a further, in-depth analysis of the same data. They showed that the discourse mode (narrative or interview) also influenced the learners’ use of the definite article. Target-like accuracy was more evident in the former than the latter.

  2 The greater likelihood of the learners omitting ne with the native speaker interlocutors than with other learners can be explained by Speech Accommodation Theory (Giles and Coupland 1991), which claims that speakers will generally converge on the norms of their interlocutors.

  3 It can be argued that research that has investigated multiple influences on variability in the Labovian Paradigm also examined the interaction of external and internal factors. However, this research sought to predict variation in specific linguistic variables. Researchers drawing on Dynamic Systems Theory do not examine specific variables and reject the possibility of a priori predictions.

  4 Dijk, Verspoor, and Lowie (2011) were not the only ones to re-analyse Cancino et al.’s (1978) data. Earlier in this chapter, we considered Schachter’s (1986) form-function analysis of the data. Berdan (1996) also re-examined the data for one of the learners (Alberto) and found evidence of both linguistic context and style predicting his choice of ‘don’t’ + verb. Berdan also reported that if Alberto used ‘don’t’ + verb in one utterance, he was likely to also use it in the subsequent utterance. Although this study demonstrated that the variability was systematic, there was still a proportion of variation that could not be accounted for.

  5 In some cases, L2 learners have full knowledge of the target forms but may elect not to use them if they do not fit with the self they wish to project. For example, Japanese L2 learners may have learned the different uses of tu and vous in French, but resist using the familiar tu with someone they would be deferential to in their L1 culture.

  6

  The role of the first language

  Introduction

  The role of the learners’ native language in the acquisition of a second language, arguably, has received more sustained attention than any other area of SLA. Jarvis and Pavlenko (2008) list 17 full-length books devoted to it and there have been more since. This intensity of attention is testimony to the prevalence of the influence of the L1, its importance for understanding L2 acquisition, and the complexity of transfer phenomena.

  Initially, this thinking was informed by behaviourist learning theory, which was dominant until researchers began to conduct empirical studies of second language learning. Behaviourist psychologists—such as Skinner (1957)—sought to account for language learning in terms of ‘behaviour’, rejecting any role for internal mental processes. Behaviour consisted of ‘habits’ which were formed when specific environmental stimuli were automatically associated with specific responses. This association arose through imitation (i.e. the learner copies the stimulus behaviour sufficiently often for the response to become habitual) and reinforcement (i.e. the feedback following a response that either ‘rewards’ or ‘corrects’ the learner). The theory claimed that old habits proactively inhibit the development of new habits. Thus—in the case of second language learning—the theory predicts that the habits of the L1 interfere with the development of L2 habits. This led to the Contrastive Analysis Hypothesis (Lado 1957). The strong form of this hypothesis claimed that all L2 errors can be predicted by identifying differences between the target language and the learner’s first language, while the weak form more conservatively proposed that contrastive analysis served to identify which errors were the result of interference (Wardhaugh 1970).

  It soon became clear, however, that many of the errors learners made could be neither predicted nor explained by L1 interference. Dulay and Burt (1974), for example, calculated the frequency of the grammatical errors made by Spanish-speaking children learning English. They classified the errors as (1) interference errors; (2) developmental errors (i.e. L2 errors that were the same as the errors observed in L1 acquisition); (3) ambiguous errors (i.e. those that could not categorized as either interference or developmental); and (4) unique errors (i.e. those that did not reflect the first language and were also n
ot observed in first language acquisition). They reported that 85 per cent of the errors were developmental and only three per cent were interference errors, concluding that L1 interference played only a minor role in the learning of L2 grammar.

  Such studies led to the recognition that behaviourist accounts of L2 learning were inadequate and that it was necessary to take account of the internal mental processing that learners engage in. In other words, learners do not acquire habits through imitation and reinforcement, but rather creatively construct their L2 systems. However, it soon became clear that Dulay and Burt had underestimated the extent of interference. A number of other studies (for example, Tran-Chi-Chau 1975) reported much higher levels of interference errors in the learner language samples they investigated. In short, Dulay and Burt’s minimalist account of first language interference was rejected and, from the 1980s onwards, there was a re-appraisal of the role played by the first language directed at identifying the factors that governed both negative transfer effects (i.e. ‘interference’) and positive transfer (i.e. the role the L1 plays in facilitating L2 learning).

  Currently, there is no theory of L2 learning that does not acknowledge the importance of the first language. However, current theories reject both the maximalist view of interference of behaviourists—such as Lado (1957)—and the minimalist view of researchers—such as Dulay and Burt (1974). They acknowledge that differences between the native and target languages do not always lead to learning difficulties. They emphasize the need to view the first language as one of many factors that shape L2 development and seek to explain how the learner’s first language knowledge interacts with other external and internal factors. Thus, the role of the first language has been re-evaluated in terms of the cognitive processes involved in second language learning. The notion of ‘interference’ has given way to that of ‘transfer’. Thus a good starting point is to consider exactly what is meant by the term ‘language transfer’.

  Defining ‘language transfer’

  Language transfer is said to occur when there is evidence that the linguistic features of one language influence those of another language. This simple definition, however, requires elucidation.

  What constitutes ‘evidence’ of language transfer?

  The most obvious kind of evidence is to be found in the errors that can be traced to another language. But there are other kinds of evidence. Knowledge of one language may facilitate the learning of another as, for example, when the two languages share cognates (for example, ‘cosmopolitan’ in English and ‘cosmopolita’ in Spanish). Also, the influence may be manifest in the overuse or avoidance of some linguistic feature rather than in actual errors. For example, Japanese and Korean learners of L2 English have been shown to avoid the use of relative clauses because of differences in the structure of relative clauses in these learners’ L1s and English (Schachter 1974). Finally—as Ringbom (1992) pointed out—transfer is just as evident in reception as in production. Differences and similarities in the two languages can impede or facilitate the comprehension of both oral and written input. Japanese learners of Chinese, for example, have a huge advantage over English learners in learning to read Chinese because the two languages share a common set of written symbols.

  What linguistic features are subject to transfer?

  Transfer is most clearly evident in pronunciation. When French people speak English, they are likely to sound French. In fact, however, cross-linguistic influence has been observed in every aspect of language—in pronunciation, orthography, vocabulary, grammar, semantics, pragmatics, and discourse. Examples of such transfer effects will be provided later in the chapter.

  In what ways can one language influence another?

  The most obvious way is when the learner’s first language influences the learning and use of an interlanguage. However, a learner may have already acquired one L2 and this can impact on the learning and use of a third language (Cenoz, Hufeisen, and Jessner 2001). Moroccan learners of English, for example, will have acquired Arabic as the L1, but are likely to have also learned French at school. Thus both Arabic and French influences on this learner’s L2 English are likely to occur. Also, the learning of a second language can have an impact on the learner’s first language, leading in some cases to language attrition (i.e. loss of L1 knowledge).

  It should be clear that language transfer is a highly complex phenomenon. It involves much more than ‘interference’. Reflecting this complexity, Sharwood Smith, and Kellerman (1986) proposed that ‘crosslinguistic influence’ would be a more theory-neutral term to refer to the way one language interacts with another. In this chapter, however, our focus is limited to the role of the first language in second language learning and thus I will adhere to ‘transfer’ as the more commonly used term.

  Investigating language transfer

  Some researchers have investigated the overall relationship between L1 and L2 proficiency. Bylund, Abrahamsson, and Hyltenstam (2012), for example, compared the native-like proficiency of Swedish-Spanish bilinguals, all of whom had learned the L2 (Swedish) as children and who had been resident in Sweden for an average of 24 years. They posited three possibilities: (1) there would be no relationship; (2) there would be a negative relationship; and (3) there would be a positive relationship. They reported that native-like proficiency in the two languages was positively correlated, indicating that for these learners, who commenced learning within the Critical Period, maintenance of their first language did not prevent their successful acquisition of a second language.

  Most transfer research, however, has examined transfer in relation to specific linguistic features. In Ellis (2008), I distinguished five different ways in which transfer of specific linguistic features has been investigated. Here, I will consider the two ways that have figured most commonly in transfer research:

  L2 errors are identified and the learner’s L1 is then examined to determine if the error type corresponds to an L1 feature.

  Differences are identified in the L2 of learners with different L1s (for example, Japanese and Arabic learners of L2 English). Transfer (both negative and positive) is held to exist if the differences in the learners’ second languages can be shown to correspond to differences in their first languages.

  The second way is clearly superior to the first. Consider, for example, Spanish learners of L2 English. These have been shown to pass through a stage of development consisting of ‘no’ + verb (see Chapter 4). Negation in Spanish involves ‘no’ + verb. Thus, in a study involving a Type 1 comparison, it could be concluded that the ‘no’ + verb construction in the learners’ L2 English was the result of negative transfer. Now consider a study that compared Spanish and German learners’ acquisition of English negation. German, unlike Spanish, has post-verbal negation (i.e. verb + ‘nicht’). Strong evidence of negative transfer would require demonstrating that the ‘no’ + verb construction occurred in the Spanish learners, but not in the German learners. In fact, German learners have also been found to pass through a ‘no’ + verb stage (Wode 1981). Thus, it would seem that the ‘no’ + verb construction is a ‘developmental’ rather than a ‘transfer’ feature of L2 learning. However, as will see later, a case can still be made for a transfer effect in the Spanish learners. Clearly, though, there are dangers in assuming transfer is occurring simply on the basis of Type 1 comparisons.

  A good example of a study that adopted a Type 2 approach is Eckman and Iverson’s (2013) comparison of Japanese and Korean learners’ acquisition of the English /s/ and /ʃ/ contrast (as in ‘sip’ and ‘ship’). These sounds are phonemic in Japanese (i.e. they distinguish Japanese words as they do in English) but only allophonic in Korean (i.e. the two sounds exist but do not distinguish words in Korean). Eckman and Iverson were able to show that differences in the path of acquisition and error patterns of Japanese and Korean learners of L2 English stemmed from this apparently small difference in their native language phonologies.

  A final methodological issue concerns the distinction b
etween implicit and explicit L2 knowledge (see Chapter 1). Theoretically, a distinction can be drawn between the transfer of these two types of knowledge: that is, transfer may be evident when the L2 is elicited by means of instruments that tap into explicit knowledge, but may not occur in free L2 production, which is more likely to draw on implicit knowledge. Bylund, Abrahamsson, and Hyltenstam’s (2012) study—referred to above—is illustrative of this problem. The instruments they employed were more likely to elicit explicit knowledge. Thus, their study does not tell us whether transfer effects were evident in the learners’ implicit knowledge.

  I will turn now to considering the factors that influence whether, and to what extent, transfer effects are found in L2 learning by drawing on studies that have investigated the role of the L1 in different levels of language; in particular pronunciation, vocabulary, and grammarNOTE 1. Five general categories of factors can be identified:

  Linguistic—i.e. factors relating to the linguistic properties of the native and target languages.

  Psycholinguistic—i.e. factors relating to the learners’ perceptions about the transferability of L1 features.

  Contextual—i.e. factors relating to the nature of the learners’ exposure to the target language.

  Developmental—i.e. how universal tendencies in L2 acquisition interact with L1 transfer.

  Individual—i.e. how individual differences in learners affect the probability of transfer occurring.

  Linguistic factors

  I begin by examining a key claim of the Contrastive Analysis Hypothesis, namely that learning difficulty (and thus the errors learners make) depends on the degree of difference between the native and target languages. I then move on to consider the role of language distance before concluding this section with a discussion of how linguistic markedness affects transfer.