Understanding Second Language Acquisition (2nd ed) Page 3

  The initial response was to downplay the effect of the first language. Some researchers (for example, Burt 1975) argued that negative transfer played a relatively minor role in L2 acquisition, accounting for only five per cent of the errors that learners made. However, other researchers continued to acknowledge that it played a significant role. Selinker (1972) saw it as one of the five ‘central processes’ in interlanguage development. Language transfer began to be seen as just one of several factors that contributed to L2 acquisition, rather than as the single, all-important factor.

  The research that followed was directed at identifying the conditions that governed when transfer occurred and when it did not. As Kellerman (1983) put it ‘now you see it, now you don’t’. Kellerman proposed a number of factors that could explain when transfer was likely to manifest itself. One of these was language distance: learners whose first language was very similar to the second language—for example, Dutch learners of English—were more likely to draw on their L1 than learners whose language was very different—for example, Chinese learners of English. But language distance was not the only factor. Learners also had an inbuilt capacity to assess which features were likely to be transferable. Researchers also recognized that first language transfer did not just manifest itself in learner errors but exerted its influence in other ways—for example, in avoidance of structures that differed from the first language. Other researchers (for example, Ringbom 1987) emphasized the role of positive transfer, especially in the acquisition of vocabulary.

  Researchers have continued to show a strong interest in language transfer right up to today. The focus of this research has broadened, however. It is not focused so exclusively on pronunciation, vocabulary, and grammar but also addresses how the first language influences the expression of politeness, discourse conventions, and even gesture. There is now wide acceptance that the L1 plays a significant role in L2 acquisition, but that to understand this role it is necessary to examine how language transfer functions as a cognitive process alongside other cognitive processes. The study of language transfer has also broadened its scope to consider conceptual transfer (i.e. how the concepts associated with one language affect the linguistic choices made in another language). Language transfer is examined in detail in Chapter 6 and again in Chapter 8.

  Input and interaction

  The recognition that second language learners manifest a definite developmental route—as shown by the existence of order and sequences of L2 acquisition—together with the rejection of L1 interference as the sole factor influencing L2 acquisition led researchers to look for an alternative explanation of L2 acquisition. One possibility they examined was the role of input and interaction. The term ‘input’ refers to the samples of the oral or written language a learner is exposed to. This constitutes the ‘data’ that learners have to work with to construct their interlanguage. The term ‘interaction’ refers to the oral exchanges a learner participates in—with native speakers or with other learners—which provide both ‘input’ and opportunities for ‘output’ (i.e. use of the L2 in production).

  Early research on input focused on foreigner talk. This is the special register that native speakers adopt when talking to non-native speakers. It is characterized by a number of ‘modifications’ to the normal talk that native speakers use when communicating with each other—i.e. when native speakers address learners they typically speak more slowly; pause more; use simpler high-frequency vocabulary; use full forms rather than contractions (for example, ‘She is coming’ rather than ‘She’s coming’); move topics to the front of a sentence (for example, ‘John, I like him’); and avoid complex subordinate constructions. Researchers like Hatch (1983) suggested that such modifications might help to make grammatical features more salient and thus help learning. Krashen (1985) argued that input modifications make input ‘comprehensible’ for learners and, more controversially, that comprehensible input is all that was needed to activate the learner’s built-in syllabus.

  One possibility that researchers considered was whether the frequency of different forms in the input learners were exposed to matched the order of acquisition. However, the results were somewhat disappointing as the correlation was not a close one. Nevertheless, input frequency has continued to be viewed as one of the main factors that determine acquisition. N. Ellis (2002), for example, claimed that it was ‘an all-pervasive causal factor’ (p. 179) although he also acknowledged that ‘there are many other determinants of acquisition’ (p. 178). In fact, ‘frequency’ is not a straightforward concept. For a start, it is necessary to distinguish token frequency and type frequency. The former refers to the number of times a specific linguistic form occurs in the input. The latter refers to the nature of the linguistic forms that can occur in a particular slot in a construction. For example, verbs like ‘know’, ‘understand’, and ‘like’ occur frequently in the construction ‘I don’t ___’. The early research investigated token frequency, but it is now thought that it is type frequency that is important as it helps learners to develop more abstract categories.

  Humans have an innate desire to communicate with each other—what Lee, Mikesell, Joaquin, Mates, and Schumann (2009) called their ‘interactional instinct’. Learners learn a language in order to interact and also they learn language through interacting. In the case of SLA, however, this fundamental truism took a while to take hold. Hatch (1978) advanced what, at that time, was a radical idea: she suggested that learners do not first learn syntactical features and then try to use them in interaction, but rather syntax emerges as a result of the kinds of interaction they typically engage in. Drawing on the methods employed in discourse analysis, she described the interactions involving both child and adult second language learners and illustrated how syntax grew out of their attempts to communicate. This idea was then developed by Long (1983b) who proposed that when there was a communication problem which speakers attempt to resolve, negotiation of meaning takes place which both helps learners to comprehend and directs their attention to specific linguistic features.

  Long’s proposal, which was eventually formalized as the Interaction Hypothesis, has proven enormously influential, leading to numerous studies investigating whether the interactional modifications that occurred in the negotiation of meaning assisted comprehension and whether they helped learners to progress along a sequence of acquisition (for example, Mackey 1999). However, the initial formulation of this hypothesis was limited as it saw negotiation only as a source of comprehensible input and allowed no role for learner output.

  The role of output in L2 acquisition proved a controversial issue. Krashen’s (1985) Input Hypothesis claimed that L2 acquisition is entirely input-driven; that is, output (speaking or writing) plays no role in acquisition. However, drawing on research that showed that immersion language learners (i.e. learners who received all their school subjects through the medium of the second language) still failed to acquire a target-like grammar even though they had experienced ample comprehensible input, Swain (1985) advanced the Comprehensible Output Hypothesis. This proposed that, as well as comprehensible input, learners needed opportunities to produce ‘pushed output’ (i.e. output where they struggled to make themselves comprehensible). She pointed out that it is possible for learners to comprehend input without having to process it linguistically (for example, they could use context to guess its meaning), but that to produce concise and comprehensible output they had to engage in syntactical processing.

  It is now generally acknowledged that both comprehensible input and comprehensible output are needed if learners are to achieve high levels of linguistic accuracy in a second language. Long (1996) revised the Interaction Hypothesis to take account of this. He now proposed that the negotiation of meaning facilitated acquisition in three major ways: (1) by providing the learner with comprehensible input; (2) through feedback that showed the learner when an error had been made; and (3) by prompting monitoring—i.e. encouraging the learner to make changes to the utterance that
had triggered negotiation. Subsequent research has focused on (2) and (3). In particular, there are an increasing number of studies investigating corrective feedback (i.e. the effect that different ways of correcting learners’ errors has on acquisition).

  The Input Hypothesis, the Comprehensible Output Hypothesis, and the Interaction Hypothesis have had an enormous influence on SLA. They have spawned countless studies investigating how input is made comprehensible, the role of production in L2 acquisition, and the role played by interaction in facilitating acquisition. While the Input Hypothesis has not changed since its initial formulation, both the Comprehensible Output Hypothesis and the Interaction Hypothesis have evolved over time, in particular as a result of the importance that researchers began to attach to the role of consciousness in language learning.

  The roles of input, interaction, and output are examined in Chapter 7.

  Consciousness and L2 acquisition

  It was at this point that SLA researchers turned to cognitive psychology to explain how input was processed by learners and thus how acquisition—viewed as an internal, mental phenomenon—took place. What motivated this development in SLA was a rather unique case study of an individual learner. Schmidt and Frota (1986) investigated the first author’s acquisition of Portuguese during a five-month stay in Brazil. Schmidt kept a diary of his observations and experiences and his interactions with native speakers were audio recorded. One of the main findings was that he only learned what he had first noticed in the input. Schmidt ‘subjectively felt … that conscious awareness of what was present in the input was causal’ (p. 281). He also reported noticing-the-gap when he compared a non-target form he had produced with the target form that appeared in the input. This often occurred when Schmidt received corrective feedback on his own utterances.

  Schmidt’s claims about the role of ‘noticing’ and ‘noticing-the-gap’ concern the role that attention plays in acquisition. In subsequent publications (for example, Schmidt 1994; 2001), he developed what has become known as the Noticing Hypothesis. The crucial question for Schmidt was ‘Can there be learning without attention?’ The conclusion he reached was that while some learning might be possible without conscious attention (i.e. noticing), the more that learners noticed, the more they learned. However, Schmidt was careful to point out that noticing did not imply intentionality on the part of learners; it could occur incidentally while they were primarily concerned with comprehending input. He also emphasized that learners did not notice ‘higher-level categories’ but rather exemplars of them. For example, they might notice the /s/ sound on ‘boys’ and also consciously register that it signalled ‘more than one’. But this did not constitute conscious awareness of the rule for making nouns plural.

  Schmidt’s claims about the importance of conscious noticing are controversial, however. Krashen’s Input Hypothesis is based on the assumption that L2 acquisition is a subconscious process; that is, he claimed that learners automatically acquire new L2 features simply as a result of comprehending the input they were exposed to. Some studies have shown learners are able to pick up some L2 features without any conscious awareness. We will consider this possibility more closely in the next section. In general, however, SLA has accepted that noticing is facilitative of acquisition. Both Long’s (1996) revised Interaction Hypothesis and Swain’s (1995) later account of the Comprehensible Output Hypothesis have incorporated the idea of noticing. Conscious attention is seen as the key mechanism that connects input to acquisition.

  Schmidt’s Noticing Hypothesis has informed an increasing number of studies investigating whether (1) learners do notice linguistic forms in the input, and under what conditions, and (2) whether this results in learning. For example, researchers have experimented with various ways of highlighting problematic features in the input that learners are exposed to (for example, by putting them in bold type in a reading passage) to see if this results in their acquisition. These studies have shown that learners tend to notice some features (for example, lexis and word order) but are less likely to notice others (for example, morphological features such as third-person -s). This can explain one of the central puzzles of L2 acquisition—why learners often fail to fully acquire some grammatical features even though these occur with high frequency in the input. Features such as third-person -s (for example, ‘comes’) are not noticed because they are semantically redundant (i.e. they are not necessary for understanding the meaning of an utterance) and not very salient (i.e. they can easily be overlooked) and, as a result, they are not acquired. Left to their own devices, learners are much more likely to notice those features that are meaning-bearing—individual words or grammatical structures like plural -s.

  Implicit and explicit learning

  The influence of cognitive psychology on SLA is perhaps most evident in the research and theorizing related to the distinction between implicit and explicit learning. N. Ellis (1994) defined implicit learning as ‘acquisition of knowledge about the underlying structure of a complex stimulus environment by a process which takes place naturally, simply and without conscious operations’ (p. 1). Explicit learning is a ‘more conscious operation where the individual makes and tests hypotheses in search of structure’ (p. 1). As defined, implicit learning is incidental (i.e. there is no intention to learn), whereas explicit learning is intentional (i.e. the learner makes deliberate attempts to learn an L2 feature). This distinction, then, echoes the earlier one that Krashen (1981) made between ‘acquisition’ and ‘learning’.

  Theories differ in how they conceptualize the implicit knowledge that results from implicit learning. Symbolic theories adhere to the idea that implicit knowledge consists of ‘rules’ (i.e. abstract representations of the features and patterns that underlie the actual use of the second language). This is probably how most people conceptualize ‘knowing a second language’ and also constitutes Krashen’s view of implicit knowledge. N. Ellis (1996), however, drew on a radically different theory of implicit knowledge. Connectionist theories claim that although learners may appear to behave in a rule-like way, they do not acquire rules but rather construct a web of connections in the neural structure of the brain. Implicit learning is an associative process whereby combinations of sounds, words, and larger units of language are internalized in accordance with the frequency with which these combinations occur in the input. For example, through exposure, learners perceive that the sounds /s/, /t/, and /r/ constitute a chunk that occurs in words like ‘strong’ or that the chunk ‘I don’t’ can co-occur with words like ‘know’, ‘understand’, and ‘like’ or that a larger chunk such as ‘I don’t know’ can combine with another chunk such as ‘where is it?’ to construct what is traditionally called a complex sentence (‘I don’t know where it is’). Learning a language according to connectionist theories is not a matter of learning rules, but of learning the possible associations between chunks of varying sizes that occur in the input. Knowing a language, then, is a matter of gradually building an elaborate network of such associations through the implicit registration of their occurrence in the input. It is a slow, organic process that requires exposure to massive amounts of input.

  On the face of it, a connectionist view of learning contradicts the Noticing Hypothesis as it assumes that an associative network is constructed without consciousness. However, N. Ellis (2005) acknowledged that the initial registration of a linguistic feature might occur consciously but ‘thereafter there is scope for its implicit learning on every subsequent occasion of use’ (p. 321). In other words, explicit learning serves as a foundation for implicit learning. Thus, although—like Krashen—N. Ellis saw implicit and explicit learning as distinct, involving separate neurological mechanisms, he did not see them as entirely unconnected. R. Ellis (1994) suggested ways in which the explicit knowledge resulting from explicit learning can assist the processes involved in implicit learning. He suggested that if learners have explicit knowledge of a grammatical rule, they are more likely to pay attention (i.e. ‘notice’) e
xemplars of this rule in the input they are exposed to and—through noticing—fine-tune their developing, implicit knowledge-system. Explicit knowledge of rules could also prime ‘noticing-the-gap’. Furthermore, learners could use their explicit knowledge to construct sentences in the L2, which then served as ‘auto-input’ that fed into the mechanisms responsible for implicit learning. Thus, whereas Krashen saw little value in explicit knowledge, both N. Ellis and R. Ellis argued that learners could utilize their conscious understanding of underlying rules in ways that facilitate implicit learning.

  The distinction between implicit and explicit learning is important for understanding the role played by form-focused instruction in L2 acquisition. A key question is whether teaching learners ‘rules’ can assist the development of learners’ implicit knowledge, or whether it simply results in explicit knowledge. Another key question is what type of instruction is needed to facilitate implicit learning.

  Dual-mode system

  It should be clear from the foregoing explanation that the learner’s second language knowledge can be both explicit and rule based or implicit and exemplar based—i.e. consists of ready-made stored chunks. In other words, learners may possess a dual-mode system. For example, the utterance ‘I don’t know’ can be processed as a single chunk or can be computed on the basis of the learner’s knowledge of the rules for negation in English. Skehan (1998) claimed ‘two systems co-exist, the rule-based analytic, on the one hand, and the formulaic, exemplar-based on the other’ (p. 54).