In the early 1960s we were satisfied that we knew all about how people learnt a second language. Then the impact of Chomsky's ideas exposed the extent of our ignorance. Some people's reactions were that we should stop interfering with the learner, since nature knew best (Newmark & Reibel, 1968). Others admitted the error of their ways in principle but in practice continued to use the same teaching techniques; by and large the changes in the content of language teaching from 'structures' to 'functions' were not accompanied by any changes in teaching methods. However, in the past few years the question of second-language learning has been reopened; this review attempts to give a critical account of some directions that research in this field is taking. Our ignorance is not quite as boundless as it once was; before long we may be able to base language teaching at last on a solid foundation of knowledge about second-language learning itself.

1. The learner's development in a second language

1.1   Methodological issues

One starting point is the language that the learner produces and understands. Inevitably this involves several methodological problems. One is the relationship of the learner's language to time; to show how learning takes place we need to see development in time: we need pictures of 'before' and 'after'. One way of achieving this is to describe the longitudinal language development of a group of learners or a single learner, as in Hakuta (1975). Another is to relate different points of time in different learners, as in Padill_a and Lindholm (1976). Yet another is to relate the learner's language to the final point in time, the language of the native speaker, the approach often used in 'error analysis'. A second methodological problem is deciding what we count as evidence of the learner's language. The most common approach is to gather more or less natural samples of the learner's language; this has the disadvantage common to corpus-based descriptions of making it hard to know when generalisations can be made to the learner's language system as a whole. Kellerman (1974) has suggested supplementing it with 'lateralisation', that is to say, eliciting further information from the learner about specific points. There are also dangers in relying solely on the analysis of errors since this provides partial information about the learner's language; to describe it adequately we need to know more about it than its differences from the target language (Corder, 1971). An alternative approach consists of designing particular techniques that tap only the part of the learner's language that we are interested in. Also, compared to first-language acquisition, the second-language learner is more aware of learning a language and can be asked direct questions about what he knows.

Already in the previous paragraph it was accepted implicitly that the learner has a language system of his own; though this system is related to both his first and second languages, it has its own distinctive characteristics. This implicit assumption has been called, with slightly different emphases, ' interlanguage' (Selinker, 1972), 'transitional competence' (Corder, 1967), 'approximative system' (Nemser, 1971), and 'language learner's system' (Sampson & Richards, 1973). It parallels the common assumption that a child acquiring a first language has a systematic language that is not simply a defective version of the adult's. In itself the assumption tells one nothing about the system or how it develops except that it is a system. For the description of learning, an important characteristic of the learner's language system is that it is changing rather than static: Corder (1977) calls it a 'dynamic goal-oriented language system of increasing complexity'. However, one should not forget the effect of 'fossilisation', in which some aspects of the learner's language system remain fixed and do not develop (Selinker, 1972).

1.2   Grammatical morpheme studies

An aspect of the learner's development that has interested many people is the order in which particular grammatical points are acquired. A common approach starts from part of Roger Brown's work on first-language acquisition. Brown (1973) discovered that the chronological order in which children start to use grammatical morphemes such as the and on is fairly constant: the child first uses -ing, then in and on, then the plural -s and so on. Dulay and Burt (1973) first tried this approach on second-language learning, using the Bilingual Syntax Measure (BSM) to elicit samples of language from children at one moment of time: they scored the accuracy for grammatical morphemes and established an order of difficulty; though they found a common order among the children, this order had certain differences from Brown's order. This method was extended to show that children with different native languages share the same order (Dulay & Burt, 1974); that adults have a similar order to children (Bailey, Madden & Krashen, 1974; Krashen, Sferlazza, Feldman & Fathman, 1976); that the order varies according to the task involved (Larsen-Freeman, 1975); and that the order is not much affected by one month's classroom teaching (Perkins & Larsen-Freeman, 1975). At first glance, then, this research suggests that second-language learners share a common order of acquisition for grammatical morphemes in English. At closer inspection, the connections between the learner's language and time seem slightly dubious. The criteria used in Brown (1973), in which points of acquisition are arranged into a chronological order, are replaced by criteria of relative success at one point of time: Brown was talking about a 'codebreaking' order, the second-language research about a^a 'decoding' order, to borrow two terms from cryptography. Rosansky (1976) argues that it is necessary to show that order of difficulty is the same as order of acquisition, if these results are to mean anything in terms of development, furthermore, comparison with Brown's results is obscured by the different methods of collecting data, on the one hand elicitation techniques such as the BSM, on the other natural recordings; indeed, native children tested with the BSM do not show Brown's order (Porter, 1977). A naturalistic study has in fact shown similarity between chronological order of acquisition in young bilingual children and Brown's order, but less similarity in older children (Padilla, to appear). So order of difficulty for grammatical morphemes is at present hard to interpret; it certainly does not in itself justify any statements about chronological order of acquisition or about differences from first-language acquisition.

1.3   Studies of syntactic development

Other aspects of syntactic development have been investigated in less controversial ways. Much of the research has dealt with English and has been concerned with children. Many results show a fair amount of variation between learners (Cancino, Rosansky & Schumann, 1975; Bertkua, 1974). Even within the speech of one learner there is great variation; Dickerson (1975) has suggested employing variable rules such as those in sociolinguistics to account for the variation from one situation to another. Out of the plethora of studies one can mention a representative handful concerned with different points of English syntax: negation (Milon, 1974); pluralisation (Natalico & Natalico, 1971); auxiliaries (Cancino et al._} 1975); pronouns (Katz, 1976); possessives (Padilla & Lindholm, 1976); eager/easy to please (Cook, 1973; d'Anglejean & Tucker, 1975). Other languages have been studied to a lesser degree: French relative clauses and indirect objects (Chun, 1976); Spanish verb and noun phrases (Dato, 1975); Swedish negation (Hyltenstem, 1977). Mostly the writers have concluded that there are general similarities between the order of acquisition of foreign learners and native children, though there are some differences; for example a stage has been described in the learning of English negation by Germans in which a negative element occurs immediately after the verb, unlike any stage in first language acquisition (Wode, 1976b).

This wealth of information about the learning of English syntax should not blind us to the paucity of research in other areas of development; semantic development has hardly been touched upon, apart from Young (1973); phon­ological development is equally under-researched, with some exceptions (Dickerson, 1975;Tarone, 1976; Wode, 1976 a). This bias towards syntax leaves the study of learner's language in a curiously isolated position. On the one hand, it is cut off from recent ideas in first-language acquisition which are more concerned with semantic, cognitive and social development than with syntax. On the other hand, it is cut off from recent work in applied linguistics which stresses communicative functions rather than grammatical form. While the grammatical data that have been unearthed are interesting in themselves, one may question where they lead. Ultimately, so far as learning is concerned, the learner's language and the order in which he acquires various items are only interesting as evidence for underlying processes at work: the question to be answered is why his language takes the form that it does. So, rather than attempt to summarise the vast mass of descriptive studies of learner's language that are now available, a task resembling Hercules' in the Augean stables, the rest of this survey looks at some of the underlying causes of the learner's language system.

2.   The learner's contribution to second language learning

2.1   Learning and production strategies

Confronted with a new language the learner applies certain strategies both conscious and unconscious to what he hears, which partly determine his language system (Selinker, 1972; Richards, 1971). The positive effects of the learner's conscious awareness that he is learning a new language have been described in terms of the strategies that the good language learner adopts (Rubin, 1975; Stern, 1975; Naiman, Frohlich & Stern, 1975). Broadly speaking, the profile of the good learner shows that he is actively involved in the learning process: he listens carefully and checks what he says himself; he is eager to participate and to communicate with others; he practices the language of his own accord. Though these strategies necessarily appear 'obvious', any model of second-language learning has to take them into account; if, for instance, adult learners feel that second-language learning is chiefly a matter of vocabulary (Hatch, 1976), this clearly has an effect on their learning, regardless of whether they are right or wrong. At the opposite pole from these conscious strategies are those revealed in an experiment, where, after 27 minutes of exposure to an unknown language, people were able to distinguish pauses located at grammatical boun­daries from those that were not: somehow they had employed a strategy that yielded quite accurate information about the grammatical structure of the language (Wakefield, Doughtie & Yom, 1974). The learner then, consciously or unconsciously, absorbs or extracts information from what he hears. One strategy is to impose the structure of his first language: he transfers rules or items to his learner-language system from his first language and shows signs of 'interference' (Selinker, 1972). Another strategy is to attempt to guess the system of the target language, resulting in ' overgeneralisation'; his use of this strategy increases as he learns more of the language (Taylor, 1975). The learner also pays great attention to word order (Chun, 1976; Lattey, 1975). He may appear to 'simplify' the second language; his sentences have a simpler grammar and use reduced forms, similar in many ways to a pidgin language (Schumann, 1975a). Indeed Schumann claims this similarity is no coincidence: pidgins and learner languages are both produced under the pressure of a need to com­municate with people from another culture. There are nevertheless difficulties in accepting either that the pressures on an individual and on a community are the same, or that the similarities between pidgins are necessarily due to a general communication strategy rather than common source languages. A slightly different explanation for simplification is that it represents an attempt to return to an earlier, more basic form of the language (Corder, 1977), resembling the view that early language is closest to deep structure (McNeill, 1965). There are, however, dangers in the over-literal use of the term 'simp­lification ' since this suggests that in some way the learner is aware of the whole grammar of the language rather than extracting a ' simple' system from what he hears. Finally there are the production strategies that the learner uses in a speech situation. One of these is' avoidance'; Schachter (1974) pointed out that learners often do not make the predicted mistakes because they can avoid producing a form that they are likely to get wrong. The idea of avoidance has been extended to include various types of lexical avoidance (Blum & Levenston, 1977). In general, the conscious and unconscious strategies that the learner employs are of vital importance and we need to know more about them, not only in linguistic terms but also in the cognitive and social terms that are starting to be studied (Chun, 1975; Fillmore, 1976).

2.2   Attitudes and motivation

The ways in which a learner learns a second language are also affected by his attitudes and psychological motivations. The two main types of motivation that have been described are the 'integrative', in which the learner wants to communicate actively with the speakers of the other language, and the 'instru­mental', in which he wants to use the language for some utilitarian purpose; of these two the integrative now seems more important (Gardner, Smythe, Clement & Glicksman, 1976). A way in to the large body of literature on this subject can be found in the annotated bibliography compiled by Desrochers, Smythe & Gardner (1975). The causes of motivation are attitudes; success in second-language learning depends on the attitudes towards the foreign culture, towards learning a foreign language, and towards the classroom situation (Gardner et al., 1976). One's native language is an important factor in one's sense of identity as an individual and as a member of a group: learning a new language means adopting a new identity. Lambert (1974) distinguishes 'additive' bilingualism, in which the learner loses nothing by learning a second language, from ' subtractive' bilingualism, in which he is forced to give some­thing up; Clarke (1976) talks of a clash between the level of 'modernity' in the two cultures; Schumann (1976) draws attention to the 'social distance' separating the learner from the target culture. While these concepts have obvious relevance to contact or immigrant situations, they are less important to language teaching in schools. However, one idea of equal importance to both is the loss of personal identity in learning a second language: culture-shock is like the alienated state of a schizophrenic unable to make sense of his surroundings (Clarke, 1976) and this is supported by similarities between the language of schizophrenics and second-language learners (Meara, 1977). Guiora, Brannon & Dull (1972) postulate a 'language ego' that is threatened by the new language; Green (1977) claims that students regress to an earlier Freudian stage in a foreign language and express ideas that are more childish than their level of thinking in their native language.

2.3   Speech and memory processes

Understanding and producing speech involves psychological processes that are only now starting to be understood. In second-language learning their nature is even more obscure. One issue is whether these processes are substantially the same in all languages; some semantic processes involved in verifying the truth of sentences have in fact been found to be similar in different languages (Just & Carpenter, 1975). Another issue is whether these processes can be transferred to a second language; in terms of syntax it has been shown that foreign adults do not benefit from knowing embedding in a first language but fall back on the same processes as the young native child (Cook, 1975). The work on grammatical morphemes can also be interpreted as showing some of the processing limitations common to learners. At least some speech processes then have to be relearnt in a second language. Indeed some of them may never be so efficient in a second language: even skilled bilinguals are worse at a cloze test in their second language (d'Anglejean & Tucker, 1973); advanced learners cannot summarise texts as well as natives (Long & Harding-Esch, 1977).

Turning now to memory processes, it is convenient to distinguish short-term from long-term memory. The characteristics of short-term memory are that information is stored for only a few seconds, capacity for information is small, and the information is stored in the form of sounds. Nothing is known about the duration of short-term memory in a second language but some research has looked at capacity. Glicksberg (1963) found adult learners of English remem­bered 6-4 digits at a time compared with 7-1 in their first language, improving to 6-7 during five weeks of an intensive course; Cook (1977) found a capacity of 5-9 in beginners and 6-7 in advanced learners. So far as the form of storage is concerned, capacity has been related to syntactic complexity (Harris, 1970) and to phonological encoding (Cook, 1977); Henning (1973) traced a develop­ment from phonological to semantic storage. So short-term memory seems to work in a similar fashion in a second language; we can store slightly less information than in our first language but we store it in similar ways. Long-term memory has also been studied. Lambert (1956) tested word associations and concluded that, while advanced learners were more like native speakers in such aspects as number and range of associations, they were still unlike them in type of associations; the clustering effect in which adults normally remember words from the same semantic category together was not found to any great extent with advanced foreign learners nor was there much difference between advanced learners and beginners (Cook, 1977). Ultimately this line of research leads to the issue of how the two language systems are related in the mind of the bilingual: is there a distinct memory system for each language or are they linked in some way? The evidence is inconclusive: in the same kind of task where bilinguals had to switch languages, Taylor (1971) found the languages were stored separately, while Neufeld (1976) found they were stored together; a task involving naming colours, on the other hand, produced evidence for close links between the languages (Lambert & Preston, 1967).

The answer to the question about transfer to a second language is then complex. The conclusion arrived at in Cook (1977) was that memory processes are transferred more easily the less they depend on language. This has affinities with the claim that 'capacity increases with processing depth since we can make greater use of learned rules at deeper levels of analysis' (Craik, 1973), if language is equated with learned rules. Indeed some recent experiments can be taken to support this claim by showing little difference between beginners, advanced learners, and bilinguals in a deductive reasoning task in their second language (d'Anglejean, Gagnon, Tucker & Winsberg, 1977). A related issue is the extent to which these memory processes can be employed in learning a second language, as opposed to using it, in codebreaking rather than decoding. Some attempts have been made to apply conscious mnemonic techniques to the learning of vocabulary: Atkinson (1975) describes a 'keyword' technique and Paivio (1976) a 'pegword' technique, both of which involve forming mental images. However, as Rivers and Melvin (1977) point out, 'a word learnt out of context is for the most part a useless bauble': mnemonic techniques are only useful if the vocabulary learnt with them is available to the speaker for normal language use. A general application of memory research is discussed in Melvin and Rivers (1976), who describe an information processing model which emphas­ises that procedures for speaking and learning are learnt through meaningful use of language rather than memorisation. A two-stage model of memory has also been proposed in which an initial memory fills up with formulas, and a second-stage memory creates rules to account for the overflow (Gallagher, 1976); this connects with the discovery of 'prefabricated utterances' in a young Japanese learner of English (Hakuta, 1974).

2.4   The learner's age

Many people take it for granted that an important factor in second-language learning is the learner's age; most discussions assume that children are better than adults and then go on to suggest explanations for this. It is useful to consider the evidence that might support this assumption. Some Canadian research indeed showed that children who arrived there under the age of seven fared better at learning English than those who arrived at an older age (Ramsay & Wright, 1974). But on the other hand there is a mass of evidence that adults are either superior to, or the same as, children: adults are better than children at understanding spoken Russian by the total physical response method (Asher & Price, 1967); teenagers are better at learning grammar than younger children whether in English (Fathman, 1975) or Dutch (Snow & Hoefnagel-Hohle, 1975); nine-year-olds learn French morphology and syntax faster than four-year-olds (Ervin-Tripp, 1974); Israeli census returns show that the age of immigration makes no difference to the pattern of acquisition, in terms of reported use, and that, while there is decline with age, the main watershed is at the age of 30 (Smith & Braine, n.d.). Nevertheless, some people might say, the superiority of children is at learning pronunciation rather than grammar. Again some evidence confirms this belief: older learners either retain more foreign accent (Oyama, 1976; Asher & Garcia, 1969) or report they retain it (Seliger, Krashen & Ladefoged, 1975); the younger children tested in Fathman (1975) were better at pronunciation than the teenagers. But, on the other hand, adults learn German sounds better than children in the same teaching situation (Olson & Samuels, 1973); adults have an initial advantage at pronouncing and imitating Dutch (Snow& Hoefnagel-Hohle, n.d.); older children up to ten make fewer mistakes with unfamiliar phonological structure (Kuusinen & Salin, 1971). So far from showing the superiority of children, most of the hard evidence warrants the opposite conclusion: adults are better than children at learning a second language when the tests are conducted under controlled conditions.

This lack of evidence has not prevented many ingenious explanations being offered for children's alleged superiority at second-language learning. The most general is that there is a critical period for language learning; after we have left this period we can learn language only with difficulty. The reasons for this critical period may be ' lateralisation', that is to say, the tendency for brain functions to become specialised to one side or the other in the early teens (Scovel, 1969). The evidence for locating lateralisation in the early teens is, however, debatable (Krashen, 1975), and the causal connection between lateralisation and language learning necessarily rests on evidence from pathological cases. A second explanation distinguishes 'acquisition', the normal process through which children learn a first language, from 'learning', an alternative process that can be used by adults because of their greater maturity (Krashen, 1976). This distinction has been enlarged into the 'monitor model' (Krashen, 1977), which claims that the distinctive feature of adult 'learning' is that adults can monitor what they are producing; thus tasks which show similarities between adults and children test 'acquisition' and do not provide sufficient time for the adult to monitor what he is doing. Obviously independent evidence of this monitor needs to be supplied; advanced learners, for example, have not been found to be better at correcting their own mistakes than beginners, as might have been expected (White, 1977). A third type of explanation is cognitive development: Rosansky (1975) suggests that the crucial point is the transition to the Piagetian stage of formal operations in the early teens; Tremaine (1975) correlates the transition to the earlier Piagetian stage of concrete operations at about seven with the syntactic development of bilingual children.

Interesting as these explanations are, it is not clear they are needed. What differences there are between children and adults can be explained without recourse to some unique feature of language learning that changes with age. One simple cause is differences in situation: Macnamara (1973) distinguishes succinctly between the demands of the street and of the classroom; often the adult meets the classroom, the child the street. The types of social interaction also vary; the language play described in Peck (to appear) is rarely allowed to adults; children talk about different topics from adults (Hatch, to appear); the peer-group pressures on children are considerably different from those on adults (Macnamara, 1976) and the motivations and attitudes of the learner vary according to age (Schumann, 1975 b). The common factor to the research cited above that found no disadvantage to adults was precisely that the situation of children and adults was kept the same (Asher & Price, 1967; Olson & Samuels, 1973). Another factor that the research on accents does not take into account is historical changes in the status of immigrants: age of arrival confounds moment of arrival. While it is certainly possible to explain some of the differences in terms of cognitive maturity, this does not necessarily involve linking them to the acquisition of formal operations or to some special process such as monitoring without further evidence. Given all the factors that distinguish adults from children, it would be strange if we found no differences between them but it is not necessary to invoke some peculiar property of language learning to explain them. While one does not wish to deny the strong impression that many people have that some adult immigrants speak their new language poorly, this may be ascribed not so much to an intrinsic defect in the adult's mind as to differences in situation, in motivation, in willingness to surrender part of one's identity, and so on, that separate children from adults.

3.   The learner’s situation

3.1        The language the learner hears

One important element in the learner's situation is the language he hears. Ferguson (1975) described 'foreigner talk', that is to say, the language variety addressed to foreigners, typified by its 'simpler' grammar and by certain vocabulary items. In a natural setting the learner may well encounter this variety; indeed it is not entirely unknown in the classroom (Hatch, 1978). While there are resemblances between 'simple' foreigner talk addressed to learners and 'simple' pidgin-like language produced by learners, there is no evidence that foreigner talk assists second-language learning any more than there is proof that babytalk helps the native child. Both these simple varieties reflect partly the adult's intuition about what learners find easy to understand, partly the language used by learners themselves, partly the sociolinguistic convention of what is appropriate for learners. Aside from foreigner talk, some work has linked the learner's language to what he hears and the types of interaction in which he takes part (Larsen-Freeman, 1976; Hatch, 1976). This link between input and learning is vital to language teaching. A feature of most teaching methods is that they control the language the learner hears in several dimensions - situationally, functionally, grammatically, notionally, and in other ways. Yet evidence for the effects of input hardly exists. Only Valdman (1975) has shown the possibility of basing the sequence of presentation of grammatical items on learner's errors. Though a central feature of most classrooms is the structuring and sequencing of input, we still have no solid evidence to challenge the assertion that a natural unstructured input is as effective (Newmark & Reibel, 1968).

3.2        The learner's social interactions

In second-language learning, Evelyn Hatch has pioneered the study of con­versational interactions between learners and native speakers (Hatch, to appear; Hatch, 1976). The central feature of interaction she sees as the need to find something to talk about: adults interpret children's utterances as naming a topic for discussion; foreign adults struggle hard to establish a topic of conversation. Adults in particular need 'repairs' to be able to check that they and the native are talking about the same thing (Hatch, 1976). The order of natural second-language acquisition is derived primarily from the learner's communicative needs rather than from grammatical complexity or frequency: the learner's language is a consequence of his interactions (Hatch, 1978). Some of the strategies for interaction among children have also been described in Fillmore (1976). As yet these interaction studies offer interesting insights into the learner's behaviour rather than a coherent theory; nevertheless they complement the ideas of integrative motivation and production strategies.

3.3   The language-teaching classroom

Many language learners, however, do not engage in social interactions with native speakers; they sit in a classroom facing a teacher. This contrast is often put in terms of 'informal' learning which takes place in a natural setting and 'formal' learning which takes place inside a classroom. Macnamara (1976), however, denies that language teaching is strictly speaking formal, since the teacher cannot provide formal rules that are linguistically and psychologically valid. Krashen (1976) sees aspects of both formal and informal learning in the classroom, defining 'formal' as teaching one rule at a time and having feedback; he suggests that an adult may benefit from formal teaching since this exploits his adult capacity to 'learn' rather than 'acquire' language. But we still have little idea of what goes on in a classroom in terms of interaction; certainly it is very different from the conversation strategies that Hatch describes. We do have some ideas about correction of errors (Cohen, 1975; Holley & King, 1971) but more information is needed about all the other aspects of classroom behaviour by students and teachers.

4.   General remarks

One obvious conclusion from this research is the complexity of second-language learning; any model has to account not just for grammatical development but also for the contributions made by the learner and by the learner's environment, not to mention the individual differences between learners, and the effects of learning a second language on the learner, a field too vast to include here. If nothing more, we can rule out simple accounts of second-language learning. A further general issue is whether a second language can be learnt in the same way as a first. An adequate discussion of this can only take place if we settle in advance which model of first-language acquisition we are comparing to second-language learning: it is hardly satisfactory to try to prove or disprove the similarity between the processes by choosing select examples from either field. At present this issue can only be discussed in fairly limited terms because the research in second-language learning is restricted in scope. Nevertheless there have been several discussions of this issue (Cook, 1969; Ervin-Tripp, 1975; Macnamara, 1976; McLaughlin, 1977; Spolsky, 1977). The general tenor of these emphasises the broad similarities between first- and second-language learning but draws attention to some specific differences. Much of the grammatical evidence has already been mentioned in section 1.3 above. In addition, one can mention an analysis of learner's errors that found most could be explained in terms of first-language acquisition rather than interference (Dulay & Burt, 1972) and the results of an experiment in which foreign adults and native children repeat sentences in similar ways (Cook, 1973).

One firm conclusion, as in any research survey, is that more research is needed; in some areas we have barely started to scratch the surface. Never­theless, since the interest of this research for many people lies in its potential application to language teaching, it seems fair to risk some tentative conclusions. Obviously any teaching syllabus or materials will be more effective if they pay attention to what is known about the developmental sequence the learner goes through, the comprehension and production strategies he uses, his attitudes and motivations, the interactions he wants to take part in, and so on. But what can the teacher do who is unable to change the syllabus or course that he uses ? Above all, the teacher has to recognise the active contribution made by the learner; regardless of what the teacher wants him to do, the learner adopts certain learning and production strategies; success in learning is a product of many different factors in the learner, most of them out of the teacher's control. In particular, the successful learner wants to communicate actively through the language. The teacher's recognition of the learner's contribution, in particular of the integrative motivation, will lead him not only to stress the ultimate aim of interacting with native speakers but also to encourage the students to interact with each other and with him in meaningful ways in the classroom; this probably means providing a variety of interaction patterns, not just the one in which all communication is channelled through the teacher. The teacher's awareness of the learner's contribution will also make him more conscious of the extent to which he can call upon the maturer cognitive processes of the second-language learner in activities where language is less involved. In terms of teaching techniques, it leads to activities that encourage the students to communicate and interact through the new language about things that vitally concern them, here and now in the classroom, rather than with a native speaker on some far-off occasion in the future; thus the teacher will rely heavily on role-play, communication games, and the like. This new emphasis brings us to a point where the goals of language teaching are no longer defined simply by the language syllabus the student is expected to know, or by the usefulness of the language to some conjectured function in the student's later life, but by their contribution to the social and psychological development of the individual; it may restore a central educational role to language teaching in addition to its academic and utilitarian roles.           

April 1978