Unlike Yorio's (1976) list and the nine items that synopsized the chapter topics of this book, most of Lightbown's generalizations and myths do more than define a domain. They hypothesize directionality within a domain, and are therefore the subject of debate. Item 6 in the first (Lightbown 1985) list, for example, stems from studies that fail to show that explicit error correction causes a permanent change in language production. Such a claim, however, may be mitigated by many teachers who have gathered observational evidence of the positive effects of theory treatment in the classroom. Nevertheless, all such claims are the beginnings of theory building. As we carefully examine each claim, add others to it, and then refine them into sets of tenable hypotheses, we begin to build a theory.
How do we know if we have the appropriate components of a theory of SLA? One answer to this question may lie in an examination of chaos/complexity theory. Diane Larsen-Freeman (1997), outlining similarities between chaos theory and SLA, argued that SLA is as much a dynamic, complex, nonlinear system as are physics, biology, and other sciences. The pathway that one learner takes in order to achieve success is different, and sometimes markedly so, from another's. Like predicting the patterns of flocking birds or the course of droplets of water in a waterfall, certain laws are axiomatic, but the sheer number and complexity of the variables involved make SLA exceedingly difficult to predict a priori.
Larsen-Freeman (1997) suggested several lessons from chaos theory that can help us to design a theory of SLA. I have synthesized her comments below.
1. Beware of false dichotomies. Look for complementarity, inclusiveness, and interface. We have examined a number of continua in this book; it is important to see them just as that, and not as dichotomies.
2. Beware of linear, causal approaches to theorizing. The "butterfly effect" in chaos theory reminds us that the fluttering wing of a butterfly in the Amazonian forest can have a chain of reactions and interreactions that extend all the way to the path of a hurricane in Hawaii. SLA is so complex with so many interacting factors that to state that there is a single cause for a SLA effect is to go too far.
3. Beware of overgeneralization. Pay attention to details. The smallest, apparently most insignificant of factors in learning a second language may turn out to be important! But on the other hand,
4. Beware of reductionist thinking. It is very tempting, with any chaotic, complex system, to oversimplify by taking some little part of the whole and extracting it from the whole system.
If a theory avoids just these four pitfalls, then perhaps it is on its way to achieving adequacy.
Michael Long (1990a: 659-660) also tackled the problem of theory building in a number of suggestions about "the least" a theory of SLA needs to explain. He offered eight criteria for a comprehensive theory of SLA:
1. Account for universals.
2. Account for environmental factors.
3. Account for variability in age, acquisition rate, and proficiency level.
4. Explain both cognitive and affective factors.
5. Account for form-focused learning, not just subconscious acquisition.
6. Account for other variables besides exposure and input.
7. Account for cognitive/innate factors which explain interlanguage systematicity.
8. Recognize that acquisition is not a steady accumulation of generalizations.
The process of theory building may be best illustrated in the form of several models of SLA that have appeared in recent history. These correspond to the schools of thought introduced in Chapter 1 and reintroduced throughout the book. While there is no viable behavioristic model of SLA (it would be far too limiting), we can identify a major innatist model, two cognitive models, and a social constructivist theory. As you read on, look back at Larsen-Freeman's and Long's lists here and decide for yourself the extent to which each model fulfills the criteria. We begin with Krashen's innatist, or creative construction, model of SLA.
One of the most controversial theoretical perspectives in SLA in the last quarter of the twentieth century was offered by Stephen Krashen (1977, 1981, 1982, 1985, 1992, 1993, 1997) in a host of articles and books. Krashen's hypotheses have had a number of different names. In the earlier years the "Monitor Model" and the "Acquisition-Learning Hypothesis" were more popular terms; in recent years the "Input Hypothesis" has come to identify what is really a set of five interrelated hypotheses. These five hypotheses are summarized below.
1. The Acquisition-Learning Hypothesis. Krashen claimed that adult second language learners have two means for internalizing the target language. The first is "acquisition," a subconscious and intuitive process of constructing the system of a language, not unlike the process used by a child to "pick up" a language. The second means is a conscious "learning" process in which learners attend to form, figure out rules, and are generally aware of their own process. According to Krashen, "fluency in second language performance is due to what we have acquired, not what we have learned" (1981a: 99). Adults should, therefore, do as much acquiring as possible in order to achieve communicative fluency; otherwise, they will get bogged down in rule learning and too much conscious attention to the forms of language and to watching their own progress.
Moreover, for Krashen (1982), our conscious learning processes and our subconscious acquisition processes are mutually exclusive: learning cannot "become" acquisition. This claim of "no interface" between acquisition and learning is used to strengthen the argument for recommending large doses of acquisition activity in the classroom, with only a very minor role assigned to learning.
2. The Monitor Hypothesis. The "monitor" is involved in learning, not in acquisition. It is a device for "watchdogging" one's output, for editing and making alterations or corrections as they are consciously perceived. Only once fluency is established should an optimal amount of monitoring, or editing, be employed by the learner (Krashen 1981a).
3. The Natural Order Hypothesis. Following the earlier morpheme order studies of Dulay and Burt (1974b, 1976) and others, Krashen has claimed that we acquire language rules in a predictable or "natural" order.
4. The Input Hypothesis. The Input Hypothesis claims that an important "condition for language acquisition to occur is that the acquirer understand (via hearing or reading) input language that contains structure'a bit beyond' his or her current level of competenceIf an acquirer is atstage or level i, the input he or she understands should contain i + 1 (Krashen 1981:100). In other words, the language that learners are exposed to should be just far enough beyond their current competence that they can understand most of it but still be challenged to make progress. The corollary to this is that input should neither be so far beyond their reach that they are overwhelmed (this might be, say, i + 2), nor so close to their current stage that they are not challenged at all (i + 0).
An important part of the Input Hypothesis is Krashen s recommendation that speaking not be taught directly or very early in the language classroom. Speech will "emerge" once the acquirer has built up enough comprehensible input (i + 1), as we saw in Chapter 4 in a discussion of the Natural Approach.
5. The Affective Filter Hypothesis. Krashen has further claimed that the best acquisition will occur in environments where anxiety is low and defensiveness absent, or, in Krashen's terms, in contexts where the "affective filter" is low.
The first two of Krashen's hypotheses have intuitive appeal to teachers in the field. Who can deny that we should have less "learning" in our classrooms than traditional language programs offer? Who in their right mind would refute the importance of learners engaging in somewhat unmonitored meaningful communication in the classroom? And the natural order hypothesis is, after all, supported in some research (Larsen-Freeman & Long 1991). Finally, the effectiveness of providing a reasonable challenge (i + 1) to students in a supportive, low-anxiety environment can hardly be denied by any teacher.
It is unfortunate that SLA is not as simply defined as Krashen would claim, and therefore his assumptions have been hotly disputed (e.g., de Bot 1996; Swain & Lapkin 1995; Brumfit 1992; White 1987; Gregg 1984; McLaughlin 1978, to name but a few). McLaughlin (1978, 1990a), a psychologist, sharply criticized Krashen's rather fuzzy distinction between subconscious (acquisition) and conscious (learning) processes. Psychologists are still in wide disagreement in their definitions of "the notoriously slippery notion" (Odlin 1986: 138) of consciousness. McLaughlin (1990a: 627) commented:
My own bias ... is to avoid use of the terms conscious and unconscious in second language theory. I believe that these terms are too laden with surplus meaning and too difficult to define empirically to be useful theoretically. Hence, my critique of Krashen's distinction between learning and acquisition—a distinction that assumes that it is possible to differentiate what is conscious from what is unconscious.
In McLaughlin's view, then, a language acquisition theory that appeals to conscious/subconscious distinctions is greatly weakened by our inability to identify just what that distinction is.
A second criticism of Krashen's views arose out of the claim that there is no interface—no overlap—between acquisition and learning. We have already seen over and over again in this book that so-called dichotomies in human behavior almost always define the end-points of a continuum, and not mutually exclusive categories. As Gregg (1984:82) pointed out,
Krashen plays fast and loose with his definitions. ... If unconscious knowledge is capable of being brought to consciousness, and if conscious knowledge is capable of becoming unconscious—and this seems to be a reasonable assumption—then there is no reason whatever to accept Krashen's claim, in the absence of evidence. And there is an absence of evidence.
Second language learning clearly is a process in which varying degrees of learning and of acquisition can both be beneficial, depending upon the learner's own styles and strategies. Swain (1998), Doughty and Williams (1998), Buczowska and Weist (1991), Doughty (1991), Ellis (1990b), Lightbown and Spada 1990, and Long (1983, 1988) have all shown, in a number of empirical research studies, that Krashen's "zero option" (don't ever teach grammar) (see Ellis 1997: 47) is not supported in the literature. Instruction in conscious rule learning and other types of form-focused instruction, as we saw in Chapter 8, can indeed aid in the attainment of successful communicative competence in a second language.
A third difficulty in Krashen's Input Hypothesis is found in his explicit claim (1986: 62) that "comprehensible input is the only causative variable in second language acquisition." In other words, success in a foreign language can be attributed to input alone. Such a theory ascribes little credit to learners and their own active engagement in the process. Moreover, it is important to distinguish between input and intake. The latter is the subset of all input that actually gets assigned to our long-term memory store. Just imagine, for example, reading a book, listening to a conversation, or watching a movie—in any language. This is your input. But your intake is what you take with you over a period of time and can later remember. Krashen (1983) did suggest that input gets converted to intake through a learner's process of linking forms to meaning and noticing "gaps" between the learner's current internalized rule system and the new input. Others have noted, however, that these processes "are not clearly operationalized or consistently proposed" (Mitchell & Myles 1998: 126). So, we are still left with a theory that paints a picture of learners at the mercy of the input that others offer.
Seliger (1983) offered a much broader conceptualization of the role of input that gives learners more credit (and blame) for eventual success. Certain learners are what he called High Input Generators (HIGs), people who are good at initiating and sustaining interaction, or "generating" input from teachers, fellow learners, and others. Low Input Generators (LIGs) are more passive learners who do little to stick their necks out to get input directed toward them. In two studies of second language learners, Seliger found that "learners who maintained high levels of interaction [HIGs] in the second language, both in the classroom and outside, progressed at a faster rate than learners who interacted little [LIGs] in the classroom" (p. 262).
Such studies, coupled with a great deal of intuitive observation of successful learners, suggest that Krashen's comprehensible input must at the very least be complemented by a significant amount of output that gives credit to the role of the learner's production. While Krashen (1997: 7) staunchly maintained that in the language classroom "output is too scarce to make any important impact on language development," Swain and Lapkin (1995) offered convincing evidence that their Output Hypothesis was at least as significant as input, if not more so, in explaining learner success. In a review of the Output Hypothesis, de Bot (1996: 529) argued that "output serves an important role in second language acquisition ... because it generates highly specific input the cognitive system needs to build up a coherent set of knowledge."
Finally, it is important to note that the notion of i + 1 is nothing new. It is a reiteration of a general principle of learning that we have already discussed in this book (Chapter 4). Meaningfulness, or "subsumability" in Ausubel's terms, is that which is relatable to existing cognitive structures, neither too far beyond the structures (i + 2), nor the existing structures themselves (i + 0). But Krashen presents the i + 1 formula as if we are actually able to define i and 7, and we are not, as Gregg (1984), White (1987), and others have pointed out. Furthermore, the notion that speech will "emerge" in a context of comprehensible input sounds promising, and for some learners (bright, highly motivated, outgoing learners), speech will indeed emerge. But we are left with no significant information from Krashen's theories on what to do about the other half (or more) of our language students for whom speech does not "emerge" and for whom the "silent period" might last forever.
Krashen's innatist model of SLA has had wide appeal to teachers who cry for something simple and concrete on which to base their methodology. It is easy to see its appeal since, on the surface, the claims that are made seem to reflect accepted principles of SLA. But in their oversim-plicity, the claims have been exaggerated. Nevertheless, in the final analysis, oddly enough, I feel we owe a debt of gratitude to Krashen for his bold, if brash, insights.They have spurred many a researcher to look very carefully at what we do know, what the research evidence is, and then in the process of refutation to propose plausible alternatives. We continue now with several of these alternative theoretical perspectives.
It is quite tempting, with Krashen, to conceptualize SLA in terms of conscious and subconscious processes. In explaining the difference between a child's and an adult's second language acquisition, our first appeal is to children's "knack" for "picking up" a language, which, in everyday terms, appears to refer to what we think of as subconscious. But there are two problems with such an appeal: (a) as both McLaughlin (1990a) and Schmidt (1990) agreed, "consciousness" is a tricky term, and (b) younger (child language acquisition) is not necessarily better (Scovel 1999).
So, if we rule out a consciousness continuum in constructing a viable theory of SLA, and we do not hold child first language acquisition up as the ideal model of language acquisition, we must look elsewhere for the foundation stones of a theory. A more sound heuristic for conceptualizing the language acquisition process, one that did indeed avoid any direct appeal to a consciousness continuum, was proposed by Barry McLaughlin and his colleagues (McLaughlin 1978; McLaughlin, Rossman, & McLeod 1983; McLeod & McLaughlin 1986; McLaughlin 1987,1990b). Their model juxtaposes processing mechanisms (controlled and automatic) and categories of attention to form four cells (see Table 10.1).