Adult Second Language Learning Strategies PDF

# Introduction Over the past two decades or so, the number of studies concerned with second-language learning has increased exponentially, as has the number of journals, anthologies, and textbooks dealing with this topic. One reason is practical: there are more people than ever whose economic aspir...

# Introduction Over the past two decades or so, the number of studies concerned with second-language learning has increased exponentially, as has the number of journals, anthologies, and textbooks dealing with this topic. One reason is practical: there are more people than ever whose economic aspirations depend on their learning a second language. In European countries there are large numbers of immigrants and 'guest-workers' who have to learn the language of the home country. A similar situation exists in the United States, which has recently witnessed a huge influx of immigrants, legal and illegal. Consequently, there is a growing need for second-language teachers of both children and adults, for pedagogical information, and for research on the process of second-language learning. A second reason for the increase in research on second-language learning is that advances in the areas of general linguistics, psycholinguistics, and cognitive psychology have prepared the groundwork for the study of second-language learning. During the past two decades there has been an enormous increase in our knowledge of the process of language learning in young children. All over the world, researchers follow young children with their tape recorders or video recorders, gathering information on their linguistic and cognitive development. As more is learned about the process of first-language development, hypotheses are generated that stimulate research on second-language learning in children and adults. This volume is concerned primarily with second-language learning in adults. I have discussed second-language learning in children elsewhere: McLaughlin 1984, 1985. The focus here is on theory - how theory informs and guides research and how theory is evaluated on the basis of research. To begin, I would like to discuss three assumptions that are basic to the scientific research enterprise, including the field of adult second-language learning. ## Three Assumptions In a thoughtful review of research on adult second-language learning and its application to teaching practice, Patsy Lightbown identified three broad categories of research based principally on methodological differences. * **Descriptive Studies**: Such studies begin by collecting speech samples from second-language speakers – either of spontaneous speech or through various elicitation procedures - and then compares these samples to target-language norms. The goal is to account for consistencies or discrepancies between the second-language learners' use of certain linguistic forms and native use. For example, there have been a large number of studies comparing second-language learners' acquisition of certain morphemes in obligatory contexts to acquisition patterns observed in native speakers. * **Experimental Pedagogical Studies**: These are studies in which the attempt is made to manipulate certain variables experimentally to determine their effect on classroom learning. Thus, for example, researchers modify the presentation of linguistic forms to second-language learners, compare subjects learning under different conditions of instruction (e.g., rule-learning versus conversation-based instruction), or attempt to control the complexity of the input language. * **Hypothesis-Testing Studies**: This is Lightbown's term for research in which the investigator, rather than beginning from a language sample or testing certain variables for their effect on classroom learning, begins with a specific hypothesis based on the findings of previous research or theory. Thus an investigator might test the hypothesis that there exist linguistic universals that shape language development in its early stages. Or a researcher might have found certain error patterns in a sample of speech drawn from speakers of a particular first language, and may conduct a study to determine whether these patterns are found in the speech of speakers of other first languages Lightbown's classification scheme is helpful in understanding recent developments in the field of adult second-language learning. However, while acknowledging the usefulness of this typology, there is a sense in which all research fits into the last category. That is, in a most fundamental sense all research involves hypothesis testing, whether this is explicitly acknowledged or not. Every investigator begins with some hypotheses about the phenomena being studied, although these hypotheses may not be stated formally. Even the researcher involved in descriptive research of the most rudimentary nature is testing hypotheses. These hypotheses originate in some theory about the data. In trying to impose order on the data, the researcher organizes the data according to some cognitive schema or theory. Without a theory there can be no hypotheses. This brings us to the first assumption: ### Assumption 1: Research is inseparable from theory. It is important to be clear about what one means by theory. The term 'theory' is used here to refer to a way of interpreting, criticizing, and unifying established generalizations. A theory is flexible and pliant, in that it allows its generalizations to be modified to fit data unforeseen in their formulation. And theory is heuristic, in the sense that the theory itself provides a way of guiding the enterprise of finding new and more powerful generalizations. The generalizations that constitute the basis of a theory derive initially from regularities or constancies in our experience of natural phenomena. Whitehead wrote: "Recognition is the source of all our natural knowledge. The whole scientific theory is nothing else than an attempt to systematize our knowledge of the circumstances in which such recognitions will occur." (Cited in Kaplan 1964, 85) Thus the first step in the scientific enterprise is the marking of enduring or recurrent events in the flow of experience. Hypotheses serve to carry forward the scientific inquiry. They are the scientist's best hunches about the regularities or constancies characteristic of the phenomena in question. To test an hypothesis, the scientist organizes the inquiry so as to facilitate a decision as to whether the hunch is correct. Once an hypothesis has been established, it is said to constitute a 'fact' or a 'law', according to whether it is particular or general in content. These facts and laws, in turn, are the stuff of theory. In a most fundamental sense, then, a theory is a system of facts and laws. But by being brought together in a theory, the facts and laws are altered, reformulated, reinterpreted. The theory is more than the aggregate of the facts and laws; it gives each of them a new meaning. The facts and laws take on a new light from the theory: the theory illuminates facts and laws. Further, good theory is always open, for the set of generalizations making up a theory is never complete. The value of a theory derives not only from the explanations it is constructed to provide, but also from its unanticipated consequences. This is the sense in which theory is heuristic: theory provides guesses as to how the uncontrolled and unknown factors in the area under study are related to known facts and laws. Theory guides the search for further data and further generalizations. In fact, all observation involves theorizing. One does not simply read and record nature's protocol. Inferences occur even at the perceptual level. As Popper (1959) argued, there is no purely 'phenomenal language' distinct from 'theoretical language'; there is no way of talking of something sensed and not interpreted. We cannot free ourselves from theories; theories are part and parcel of the human cognitive makeup. Thus the researcher concerned with adult second-language learning is inevitably testing hypotheses about a theory. The theory may be vaguely or not at all described. But just as on the perceptual level there is no uninterpreted intuition or bare sensation, the researcher is never unencumbered by expectancies and beliefs about the phenomena in question. And even if we could free ourselves from theory, it would be unwise not to make use of theory to interpret what we observe and to generate new hypotheses. We come now to the second basic assumption: ### Assumption 2: There is no one scientific method. Earlier the point was made that theoretically-derived hypotheses are tested via a scientific inquiry designed to facilitate a decision as to whether a given hypothesis is correct. This inquiry process can take many forms. There is no one ‘scientific method'. Thus the traditional research cycle - proceeding from description to correlation to experimentation - assumes that experimental research is the sine qua non for establishing the validity of research. But science does not progress only in the laboratory. It is true that experimentation allows for degrees of control not possible in other situations, yet there are the attendant perils of trivialization of the phenomenon under investigation and lack of motivation on the part of the subject. Experimentation should obviously have its place in the researcher's repertory, but it is no panacea. Hypotheses may be tested via any of the methods Lightbown identified. To use one of her examples, Huebner (1979) examined the patterns of use of the English article by a Hmong speaker and found that this second-language learner's use of the article was strongly influenced by his mistaken assumption that English, like his native language, uses what might be called pragmatic (topic prominence) rather than syntactic (subject prominence) rules for marking definiteness. This confirmed the hypothesis that for some learners there are internal consistencies in language use that may occur without regard to target language norms. The Huebner study was descriptive and longitudinal. It involved a single case. Yet the results are theoretically significant. (There is no need to take the research further to confirm Huebner's hypothesis. One does not need to carry out an experiment to 'prove' the point. Further research may provide further confirmation (or disconfirmation) of Huebner's hypothesis, but careful descriptive-analysis research can stand on its own. It represents an important scientific contribution; just as important as experimental research. Observation and descriptive-analytic work need not be the first steps on a journey where experimentation is the end of the road. In research on adult second-language learning, as in all scientific research, there is the danger of fadism. Abraham Kaplan (1964) has identified what he termed the law of the instrument, which may be formulated as follows: Give a small boy a hammer and he will find that everything he encounters needs pounding. Second-language research has had its instruments, its ways of doing research, and it can happen that the methods researchers use set limits on the questions they can answer. Worse than that, some successful techniques have come to be applied mechanically and other approaches have been denied the name of science. Another danger comes from the mystique of quantity. Numbers have no magical powers; they do not have scientific value in and of themselves. Yet often one senses an exaggerated regard for the significance of measurement, so much so that complex statistical analyses are used to salvage sloppy observation or bad experimentation. Statistics are tools of thought, not substitutes for thought. I have argued strongly for the use of more sophisticated statistical procedures in research on second-language learning (McLaughlin 1980), but if such techniques as multivariate regression analyses are used to mystify, the field will move quickly into the realm of the occult. My own preference is for a multimethod approach, where a number of different methodologies are employed to test theoretical hypotheses. A catholicity of outlook is needed in research on adult second-language learning. This is not to say that one approach is as good as another in a particular context or that all should be used in any single inquiry. A multimethod approach moves the field along; it may not serve the purposes of a given researcher. Data triangulation is achieved most often through the work of different researchers with different interests getting a fix on the same phenomenon from different angles: ### Assumption 3: There is no single scientific truth. The kind of methodological openness advocated in assumption 3 runs contrary to a natural tendency to feel that one's own approach and one's own point of view is better than other people's. Disciplines tend to become fragmented into ‘schools', whose members are loath to accept, and are even hostile to the views of other schools using different methods and reaching different conclusions. Each group becomes convinced that it has a corner on 'truth'. It can be argued, however, that it is counterproductive to believe that there is a single truth that can be known directly and objectively by the application of scientific methods. One philosophical position contends that truth can never be known directly and in its totality. All knowledge, the argument runs, is mediated by the symbol systems used by scientists. If one adopts this epistemological stance (scientific truth can be known only as it is mediated by the perspectives of the scientists and thus there are multiple accounts of what is seen (Multiple ways of seeing result in multiple truths)), scientific progress is achieved as we come to illuminate progressively our knowledge in a particular domain by taking different perspectives, each of which must be evaluated in its own right. The symbol system or metaphor used by a particular scientist may help us see more clearly, but it does not constitute ultimate truth. Nor does the combination of all partial representations of truth add up to ultimate truth. Ultimate truth, it is argued, is only approximated in the shadows cast by the metaphors of our theories. In an insightful discussion of the way in which symbol systems are used by investigators of adult second-language learning, Schumann (1983) identified a number of metaphors underlying the work of various researchers. These include Krashen's (1981) use of the 'affective filter', Selinker and Lamendella's (1976) highly metaphorical neurofunctional perspective, and Adjemian's (1976) notion of the 'permeability' of the interlanguage system. In fact, as we shall see in the course of this book, metaphors abound in the theories that have been developed to account for various aspects of adult second-language learning. This is not to say that all metaphors are equally valid or that all ways of looking at second-language phenomena are equally veridical. Criteria for evaluating the adequacy of different theoretical points of view can be specified, and in the final section of this chapter, such criteria will be spelled out. (The point is simply that, regardless of one's epistemological beliefs about the nature of truth, it is useful to look at adult second-language learning from as many perspectives as possible. At least in the present stage of the development of our knowledge, it seems premature to argue for the 'truth' of one theory over another. ## The Nature of Theory Earlier in this chapter theory was defined as a way of interpreting, criticizing, and unifying established generalizations. In addition, it was argued that a theory is pliant, in that it allows its generalizations or 'laws' to be modified to fit data unforeseen in their formulations, and heuristic in that the theory itself provides a way of finding new and more powerful generalizations. These last two characteristics of a good theory are especially important in considering how theory operates. ### The Functions of Theories A theory is both a summary of known facts and laws and a conjecture about the relationships among them. The purpose of theory is in part to further understanding. Theories help us understand and organize the data of experience. They permit us to summarize relatively large amounts of information via a relatively short list of propositions. In this sense, theories bring meaning to what is otherwise chaotic and inscrutable. A second function of theories is transformation. New theories change the relationship between laws and facts. They enable us to use the empirical data to draw conclusions that are not evident from the data taken in isolation. In this sense, theories transform the meaning of what is known. They go beyond the information given and change both the content and form of our knowledge. Theories also guide prediction. Good theories stimulate research. They are the ground from which hypotheses spring: theories generate new hypothetical laws to be put to empirical test. These hypotheses embody predictions about where the theory is leading. They are not a guess at an answer to a riddle, but an idea about the next step that is worth taking. These functions of theory can be illustrated by an example from research on adult second-language learning. One of the leading theories of second-language learning for many years was what can be called the structural-behaviourist theory. According to this theory, language learning was viewed as a process of habit development to be inculcated by varying contingencies of reinforcement. Behaviourist theories were invoked to to justify such principles as sequential control of the learning process, specification of learning goals, and immediate reinforcement. Structural theory was invoked to justify the use of contrastive analysis: it was felt that by being aware of the structural differences between languages, the teacher could foresee student errors and help in overcoming them. Thus the structural-behaviourist theory provided a framework for understanding the process of second-language learning. Principles such as reinforcement and habit formation and the method of contrastive analysis transformed thinking about second-language learning phenomena, or at least made more explicit old ways of thinking about second-language learning. Furthermore, the theory led to specific predictions about learning. For example, according to structural-behaviourist theory, the errors that learners make in a second language should reflect the structures of their first language. This turned out not to be the case in many instances (see McLaughlin 1984), nor were structural-behaviourist predictions about classroom learning supported in experimental pedagogical studies (see McLaughlin 1985). Ultimately, structural-behaviourist theory was abandoned or modified so extensively as to be nonrecognizable. The theory was useful, however, for although we now think of second-language learning in very different terms, we have evidence that allows us to reject a position that at first glance agrees with common sense. If one asked the proverbial 'man on the street' for his view of the process of learning a second language, the theory would most probably look something like structural-behaviourist theory. That such notions can be rejected as insufficient is a step forward. ## Types of Theories There are at least two different dimensions that can be used to classify theories. One can examine their form or their content. From the formal point of view, many different distinctions can be made (see Einstein 1934; Kaplan 1964; Nagel 1961). For the most part, these distinctions correspond to the difference between more 'deductive' and more ‘inductive’ approaches (Table 1.1). The deductive approach characterizes formal theories in which the concepts of the theory are related to each other in a set of propositions that are assumed to be true without proof (though they may be empirically testable). These constitute the axioms of the theory. Ideally, these axioms are as few as possible, so that the theory is simple and parsimonious. | Beginning Point | Deductive Theories | Inductive Theories | |-------------------|-------------------------|-----------------------| | Network | Interim Solutions | Empirical Data | | | Theoretical Concepts and | Relationships between | | | Constructs | Laws | | Ultimate Goal | Explanation | Explanation | | Advantage | More Interesting Claims | Close to Data | | Disadvantage | Remote from Data | Limited Claims | Given these basic axioms or assumptions, laws of logic are applied to obtain new propositions. This procedure is called 'deducing the consequences of the theory'. The new propositions that follow from the assumptions of the theory are called 'hypotheses'. If the hypotheses of the theory are empirically supported, they become the laws and facts of the theory. In a deductive theory, there are fewer and more general laws at the top of the pyramid, and a greater number of more specific laws as we move to the base. In contrast to the deductive approach, the inductive approach progresses from the accumulation of sets of facts and sets of laws to theory. Rather than beginning from sets of premises or axioms that are assumed to be true, the inductive approach is tightly empirically based. There is no jump to theoretical statements until a large number of empirical relationships have been established. In the inductive approach, hypotheses are derived from the theory as in the deductive approach, but they are the result of gaps in the network of empirical relationships, rather than the consequence of logical deductions. Hypotheses do not come from logical reasoning, but are the investigator's best hunch about a new relationship, given certain empirical facts. The theory guides the investigator, but hypotheses are not as tightly and logically formulated as in the deductive approach. In practice, it is often difficult to determine whether the approach taken is more deductive or inductive. There are few well-developed formal theories outside of the natural sciences, and in each case it is a question of emphasis. Some theorists are more 'top-down' or deductive in their orientation, while others stay closer to the data and are therefore more 'bottom-up' in approach. Both are approaches to theorizing and so both aim at explanation: inductive approaches tend to offer more cautious and descriptive accounts, whereas deductive approaches tend to provide more general causal statements. But here again we are dealing with a continuum along which theories can be placed, rather than with a dichotomy. From the point of view of the content of theories, there are micro or macro theories, depending on the range of phenomena the theory is thought to accommodate. Thus macroeconomics is concerned with the workings of an economy or industry, whereas microeconomics focuses on the behaviour of individual participants in the economic process. The distinction is a relative one and what is micro to one theorist can be macro to another. In the field of adult second-language learning, a macro theory might attempt to deal with the whole range of phenomena involved in the language-learning process, as does Krashen's Monitor Model, treated in chapter 2, whereas micro theories might concentrate on specific phenomena of limited scope, as do the theories of intermediate range discussed in chapters 3 to 6. It is often argued that micro theories are intrinsically more satisfactory, and in many fields, such as sociology and psychology, the tendency of researchers is to shy away from macro theory. It may be true that at a particular stage in the development of a field of research it is overly ambitious to attempt to articulate a macro analysis. The length of the reduction chain connecting the theoretical terms with observable ones can be too long. This is why many theorists restrict themselves to 'miniature theories'. In the end, however, an adequate theory will have to be 'macro' enough to encompass more than a limited range of phenomena. A satisfactory theory of adult second-language learning must go beyond assumptions, from which hypotheses are derived. These assumptions are not clearly stated by Krashen, but will be discussed in the following pages. The theory also has a number of metaphors, one of which, as we shall see, is the Monitor, or mental editor, that utilizes conscious grammatical knowledge to determine the form of produced utterances. The Monitor is thought to play a minor role in the second-language process, however, and in recent years Krashen has apparently abandoned the term 'Monitor theory'. Nonetheless, because this term remains generally used in the literature, it will be retained here. Krashen argued that experimental and other data are consistent with a set of five basic hypotheses, which together constitute his theory. In this chapter, I will discuss these five central hypotheses: 1. The Acquisition-Learning Hypothesis 2. The Monitor Hypothesis 3. The Natural Order Hypothesis 4. The Input Hypothesis 5. The Affective Filter Hypothesis In each instance, I will look at whether the data support the hypothesis. The concluding section reviews the argument of the chapter and evaluates the success of the theory in meeting the criteria for good theory discussed in chapter 1. ## The Acquisition-Learning Hypothesis Krashen maintained that adult second-language learners have at their disposal two distinct and independent ways of developing competence in a second language: acquisition, which is 'a subconscious process identical in all important ways to the process children utilize in acquiring their first language' (1985, 1) and learning, which is 'a conscious process that results in “knowing about" language' (1985, 1). Acquisition comes about through meaningful interaction in a natural communication setting. Speakers are not concerned with form, but with meaning; nor is there explicit concern with error detection and correction. This contrasts with the language learning situation in which error detection and correction are central, as is typically the case in classroom settings, where formal rules and feedback provide the basis for language instruction. Nonetheless, for Krashen it is not the setting per se, but conscious attention to rules that distinguishes language acquisition from language learning. In the natural setting an adult can obtain formal instruction by asking informants about grammar and by receiving feedback from friends. Similarly, language can be acquired in the classroom when the focus is on communication - for example, through dialogues, role-playing, and other forms of meaningful interaction. But if setting is not the distinguishing characteristic of acquisition and learning in Krashen's sense, it is important for him to make clear what he means by 'conscious' and 'subconscious'. Krashen has not provided a definition of these terms, although he did operationally identify conscious learning with judgements of grammaticality based on 'rule' and subconscious acquisition with judgements based on 'feel' (Krashen et al. 1978). The difficulty with such an approach is that it is impossible to know whether subjects in this experiment were actually operating on the basis of rule or feel. Krashen and his colleagues had their subjects state the rule when they made judgements on the basis of rule, but they may have done so because the demand characteristics of the situation emphasized rule articulation. Moreover, subjects may have given feel answers because they were not quite sure as to how to articulate the rule on the basis of which they had operated. Aside from this attempt, Krashen has provided no way of independently determining whether a given process involves acquisition or learning. ### Does learning become acquisition? The slipperiness of the acquisition-learning distinction becomes more apparent when one considers the evidence for the first corollary of the distinction (and one of the central tenets of Krashen's theory): that learning does not "turn into" acquisition' (1982, 83). That is, according to Krashen, what is consciously learned – through the presentation of rules and explanations of grammar - does not become the basis of acquisition of the target language. The argument that conscious learning does not become unconscious acquisition is based on three claims (Krashen 1982, 83-7): 1. Sometimes there is 'acquisition' without 'learning) - that is, some individuals have considerable competence in a second language but do not know very many rules consciously. 2. There are cases where ‘learning’ never becomes ‘acquisition’ – that is, a person can know the rule and continue breaking it, and 3. No one knows anywhere near all the rules. All of these arguments may be true, but they do not constitute evidence in support of the claim that learning does not become acquisition. Kevin Gregg (1984) pointed out that such a claim runs counter to the intuitive belief of many second-language learners, for whom it seems obvious that at least some rules can be acquired through learning. He gave the example of having learned the rules for forming the past tense and gerundive forms of Japanese verbs by memorizing the conjugation chart in his textbook. In a few days his use of these forms was error-free, with no input but a bit of drill. Gregg's point is that he - and his classmates - had 'learned' the rules and these rules had become 'acquired' – in the sense of meeting the criterion of error-free, rapid production - without meaningful interaction with native speakers. Krashen argued that the experience that Gregg described merely 'looks like' (1981, 117) learning causes acquisition. He gave the example of an individual who had learned in a classroom a rule such as the third person singular morpheme for regular verbs in English. This individual was able to apply this conscious rule at Time 1, but it was only later, at Time 2 - after meaningful input from native speakers - that the rule was acquired. Krashen contended that the learner was 'faking it' until his acquisition caught up, or until he arrived at the rule 'naturally' (1981, 118). Until Time 2, he was outperforming his acquired competence. He was able to apply the rule when given enough time and when focusing on form, but had not yet 'acquired' it. At issue here is Gregg's subjective sense that he, and learners like him, apply the rule rapidly and without making mistakes regardless of context versus Krashen's contention that such application occurs only when certain conditions are met. Krashen's argument has a suspicious ad hoc ring to it, but the most troubling aspect of this disagreement is that it is difficult to support one position or the other until it is clear what is meant by 'acquisition' and 'learning'. I had raised this issue in my critique of Krashen's theory (McLaughlin 1978) in commenting on my own inability to determine whether acquisition or learning was involved in my production of certain German sentences. Remarkably, Krashen, in his reply (1979), was able to specify how much of each was involved in the utterances. This was not the point, however. The point is that there needs to be some objective way of determining what is acquisition and what is learning. This Krashen did not supply. Krashen (1979) argued that the acquisition-learning distinction is an abstraction that predicts many observable and concrete phenomena. He compared his hypothesis to hypotheses in cognitive psychology that are based on abstractions used to predict measurable phenomena. But the abstractions used in cognitive psychology, as Krashen correctly pointed out, are defined by special experimental conditions. If the acquisition-learning hypothesis is to be tested, one needs to know what experimental conditions are necessary to bring out the differences between these two processes. ### Can adults 'acquire' a language as children do? As was noted above, Krashen maintained that acquisition is 'a subconscious process identical in all important ways to the process children with which all individuals are in general uniformly and equally endowed. The application of this theory to second-language learning in adults will be discussed in more detail in chapter 4. As we shall see, principles of the Universal Grammar are hypothesized to involve properties described as parameters. Language acquisition is seen as a process of setting the values of the parameters of these universal principles, and not as a problem of acquiring grammatical rules. This is a very different enterprise from the one that concerns Krashen. Thus the second corollary of the acquisition-learning distinction cannot be supported on the basis of the arguments advanced by Krashen. The claim that adults acquire languages in the same way as children do rests on a faulty understanding of the LAD. Like the first corollary - that learning does not become acquisition – no empirical evidence is provided in support of the claim. In fact, both corollaries are not falsifiable because the central concepts to be tested - acquisition and learning – are poorly defined. ## The Monitor Hypothesis According to Krashen, ‘learning’ and ‘acquisition’ are used in very specific ways in second-language performance. The Monitor Hypothesis states that 'Learning has only one function, and that is as a Monitor or editor' and that learning comes into play only to 'make changes in the form of our utterance, after it has been “produced by the acquired system' (1982, 15). Acquisition 'initiates' the speaker's utterances and is responsible for fluency. Thus the Monitor is thought to alter the output of the acquired system before or after the utterance is actually written or spoken, but the utterance is initiated entirely by the acquired system. This hypothesis has important implications for language teaching. Krashen argued that formal instruction in a language provides rule isolation and feedback for the development of the Monitor, but that production is based on what is acquired through communication, with the Monitor altering production to improve accuracy toward target-language norms. Krashen’s position is that conscious knowledge of rules does not help acquisition, but only enables the learner to ‘polish up' what has been acquired through communication. The focus of language teaching, for Krashen, should not be rule-learning but communication. One interesting implication of Krashen’s argument is that learning is available only for use in production, not in comprehension. Krashen provides no evidence for this claim and, indeed, as Gregg pointed out, this is a rather counter-intuitive notion. Gregg, who lives in Japan, gave an example from his own experience: The other day while listening to the radio, I heard the announcer announce wagunaa no kageki, kamigami no kasoware. Knowing that kageki = 'opera' and that kami = either 'god' or 'hair' or 'paper', and knowing that there is a (fairly unproductive) rule in Japanese for pluralizing by reduplication, I concluded that kamigami must be the plural of kami, 'god', that therefore wagunaa must be Wagner and kasoware must mean 'twilight', and that I was in danger of hearing Die Götterdämmerung (Of course I was not quite right: there is no word kasoware, it's tasogare. But the point is that I was using a rule that I had ‘learned’ (and never used productively), and using that rule consciously (and quickly enough to turn the radio off in time). Which suggests that ‘learning’ can indeed be used in comprehension, as no one before Krashen would have doubted. (1984, 82-3). ### The Three Conditions for Monitor Use As we have seen, the Monitor acts as a sort of editor that is consciously controlled and that makes changes in the form of utterances produced by acquisition. Krashen has specified three conditions for use of the Monitor: 1. Time. In order to think about and use conscious rules effectively, a second language performer needs to have sufficient time. For most people, normal conversation does not allow enough time to think about and use rules. The over-use of rules in conversation can lead to trouble, i.e., a hesitant style of talking and inattention to what the conversational partner is saying. 2. Focus on form. To use the Monitor effectively, time is not enough. The performer must also be focussed on form, or thinking about correctness… Even when we have time, we may be so involved in what we are saying that we do not attend to how we are saying it. 3. Know the rule. This may be a very formidable requirement. Linguistics has taught us that the structure of language is extremely complex, and they [sic] claim to have described only a fragment of the best known languages. We can be sure that our students are exposed only to a small part of the total grammar of the language, and we know that even the best students do not learn every rule they are exposed to. (Krashen 1982, 16) Krashen seems unsure of the first condition, time. In his latest book he mentioned only the last two conditions (1985, 2). His thinking in this respect was modified by results of a study by Hulstijn and Hulstijn (1984), who independently manipulated focus on form and time with adult second-language learners (English speakers learning Dutch). When the subjects were focused on form without time pressure, there were gains in accuracy, as the Monitor theory predicts. But when there was no focus on form, giving subjects more time did not make a difference in their performance. Krashen agreed that this was evidence that the availability of extra time does not, alone, involve the Monitor. The second condition for use of the Monitor, focusing on form, is also problematic. In several studies, focusing subjects on form by having them correct spelling and grammar in written composition did not result in use of the Monitor (Houck et al. 1978; Krashen et al. 1978). Krashen’s solution to this problem was to add another condition: the subject must be focused on form and the test must be an ‘extreme’ discrete-point test that centres on just one item or rule at a time. Even this condition does not, however, suffice to produce the Monitor. Fathman (1975) and Fuller (1978) found that instructions to focus on form did not result in the gain in accuracy that use of the Monitor should produce. The test they used to elicit language seems to meet the criterion of an extreme discrete-point test in that subjects were shown pictures and asked such questions as Here is a ball. Here are two … with the expectation that they would produce the plural. It appears to be rather difficult to demonstrate the operation of the Monitor in such studies and Krashen's repeated failures to do so are explained away by changing the requirements. As he put it, ‘Again, I do reserve the right to change my hypothesis in the light of new data’ (1979, 155). This is true, but for the researcher attempting to test the theory, the constant modifications are frustrating. Seliger (1979) concluded that the Monitor is limited to such specific output modalities and requires such carefully confined conditions for its operation that it cannot be thought to be representative of the learner’s internal, conscious knowledge of the grammar. The third condition, that the learner knows the rules, has also been challenged by the research of Hulstijn and Hulstijn (1984). In their study, subjects' rule knowledge, as assessed in an interview, did not relate to how much subjects gained in performance from focus on form and absence of time pressure. Focus on form and time had the same effect on learners who could correctly verbalize the rules, for learners who could not state any explicit rules at all, and for learners who stated partly correct, or even incorrect, rules. Krashen’s attempt to explain away these findings as due to ‘inter-stage fluctuation’ (1985, 22) is shamelessly ad hoc. Once such explanations are invoked, the theory becomes completely untestable. In short, Krashen has not been able to demonstrate that the putative conditions on this issue has demonstrated is either (1) that the Monitor is rarely employed under the normal conditions of second-language acquisition and use, or (2) that the Monitor is a theoretically useless concept. Krashen has argued lately for the first conclusion (1985), but if the Monitor is the only means whereby conscious knowledge of the rules of a second language (‘learning’) is utilized, then why make the learning-acquisition distinction? If learning occurs only under such rarified conditions, what role can it possibly have in gaining competence in a second language? ## Individual Differences In spite of the difficulties of demonstrating Monitor use under experimental conditions, Krashen has based his explanation of individual differences in second-language performance on the Monitor concept. He distinguished three types of Monitor users: 1. Monitor over-users. These are people who attempt to Monitor all the time, performers who are constantly checking their output with their conscious knowledge of the second language. As a result, such performers may speak hesitantly, often self-correct in the middle of utterances, and are so concerned with correctness that they cannot speak with any real fluency. 2. Monitor under-users. These are performers who have not learned, or if they have learned, prefer not to use their conscious knowledge, even when conditions allow it. Under-users are typically uninfluenced by error correction, can self-correct only by