Consciousness-Raising and the Second Language ...

22 downloads 0 Views 641KB Size Report
The notions 'explicit knowledge' and 'implicit knowledge' have occurred in ... such hard-line implicit methods will swiftly come to appreciate the fact that.
Consciousness-Raising and the Second Language Learner1 MICHAEL SHARWOOD SMITH University of Utrecht

Applied Linguistics, VoL 11, No. 2

Downloaded from http://applij.oxfordjournals.org/ at University of Edinburgh on April 5, 2016

The notions 'explicit knowledge' and 'implicit knowledge' have occurred in various forms both in the literature on language instruction and within a more strictly learning-oriented context. Explicit knowledge, broadly speaking, denotes a conscious analytic awareness of the formal properties of the target language whereas implicit knowledge means an intuitive feeling for what is correct and acceptable (Bialystok 1978). Although, as will become clear in the course of this paper, this binary distinction needs a great deal of qualification, it does serve to highlight generally different kinds of outcome resulting from the process of learning a new language. The ultimate, most highly prized goal of learning, i.e., spontaneous, unreflecting language use, is uncontroversial. How this is achieved is, of course, a matter of considerable debate. Certain trends in language learning research suggest that promoting conscious awareness of language structure is at best a luxury and does not lead in any meaningful way to the attainment of this ultimate goal. On the face of it, this lends support to the Direct Method and more recent attempts to focus on getting the learner to perform in life-like communicative ways rather than in a strictly formal classroom context of the traditional type. A closer look at the issues, however, reveals how simplistic such pedagogical inferences are and how dubious the distinction is between two theoretically distinct types of knowledge where no allowance is made for different degrees of explicitness and the possibility of interaction between different types of competence. It is a basic problem in teaching to know how much one has to tell a learner about the language and what to do with the language, and to what extent mere practice will invoke the appropriate learning mechanisms to cope with the task in hand. It takes a committed audiolingualist or believer in the Direct Method in its extreme and most consistent form to deny all learners even occasional explanations of linguistic structures. People who attempt such hard-line implicit methods will swiftly come to appreciate the fact that they require an inordinate amount of time and energy: in other words, fullscale intensive teaching programmes are necessary for any success to be guaranteed. This also goes for a methodologically inspired refusal to use the native language in the classroom: where one translation equivalent would resolve the problem, long complicated paraphrases are needed to show the learner, in the target language, what a given word or expression means and how it is used. It is difficult to assess exactly how many people stick to such methodological principles in practice: one suspects that many deviations and compromises occur. For example, it is notoriously difficult to deny adult learners

160

CONSCIOUSNESS-RAISING

Downloaded from http://applij.oxfordjournals.org/ at University of Edinburgh on April 5, 2016

explicit information about the target language (TL) since their intellectual maturity as well as their previous teaching/learning experience makes them cry out for explanations. Teachers, one suspects, often resort to explanations (and the native language, for that matter). If explanations about the structural properties of the TL directed towards younger learners are rather less common this is probably more due to the lack of intellectual maturity in the learners involved than to the particular methodological leanings of the teacher. Teachers, and doubtless many learners as well, view explanations as shortcuts. It may be 'naturalistic' to learn languages in a purely intuitive manner but how long will it take to amass a sufficient amount of implicit knowledge and the appropriate skills for using it? It may even be rewarding to discover formal regularities in a more or less conscious manner on one's own, without the aid of the teacher or textbook, but, again, what time is needed to accomplish this in more than just a piecemeal manner? The short cut, a ready-made a priori explanation (partial or otherwise), is attractive: at the very least, it provides an insight into the task and means of labelling and specifying the problem. By revealing some pattern or system in the target language, the teacher holds out the promise of a short cut as far as learning is concerned, in other words a shorter and more effective way of mastering a structure (via practice, of course). The fact that the young child may not have this (hypothetical) possibility to the same degree, i.e., learning via explicit knowledge, puts him/her at a disadvantage when compared with the mature learner. This is presumably the motivation for the cognitive-code approach to language teaching. Armed with explicit information about particular linguistic tasks, the learner can use conscious applications of rules to practise in and out of class and to communicate in the target language at a higher level of proficiency, albeit without the speed and spontaneity associated with the notion of 'fluency'. Fluency is assumed to come later and as a result of practising TL structures in formal and informal, naturalistic ways. It is therefore quite reasonable to state that the more mature the learner, the greater the variety of resources that he or she can exploit in learning (see Faerch and Kasper 1980: 106-108). The onus is surely on those who wish to deny the value of explicit teaching techniques to show that explicit teaching is a waste of time and energy: the evidence will have to be very convincing. What might be called language consciousness-raising in the classroom is sometimes assumed to consist of the pedantic giving and testing of rules and lists of vocabulary items, that is, a complete and unrelenting focus on the formal structure of the TL. This impression is probably the result of what people associate with the grammar-translation method where learners were required to learn by rote and produce rules and lists of words almost as much for intellectual exercise as for learning to express meaning in the target language. The conveying of a rule or any other kind of information about the language can, however, be more or it can be less reduced to the familiar metalinguistic prescriptions of traditional grammars. And it may or not be required of the learner to actually produce those prescriptions himself. The relevant information can vary in the degree of elaboration or conciseness with which it is presented, as well as the degree of explicitness or intensity in the way attention is drawn to the relevant regularities. Strictly speaking, the discovery of regularities in the target language whether blindly intuitive or

MICHAEL SHARWOOD SMITH

161

10 E L A B A B O R A T C D I O N »— 0 10 EXPLICITNESS (LESS TO MORE OVERT) Fig. 1. Consciousness-raising in language learning. i

To begin with Type D, the most familiar and traditional type, this form of highly overt consciousness-raising may be found in the standard school grammar and is characterised by fairly concise prescriptions couched in a metalanguage that is supposedly within the grasp of the teacher and learner alike. This assumption is safer where mother tongue teaching has involved relevant terminology like 'adjectives', 'nouns', 'clauses' and so on, less safe where mother tongue grammar instruction has been accomplished in less formalistic ways. The accuracy and effectiveness of such prescriptions is often limited. There may be strictly linguistic reasons for this or the reasons may be of a psychological/didactic character. An accurate, technically sophisticated formulation may help a linguist to appreciate a theoretical point or an insight into a given language but may be lost on the learner and the teacher as well. A technically simple formation which somehow manages to retain the insight may nevertheless fail because it becomes clumsy, vague and ambiguous: the conveying of the insight is not guaranteed. Also, even if the statement of the information is clear, comprehensible and applicable, its effectiveness has to be assured via practice in TL. Type C—brief, indirect 'clues'—if done well, may give the learner a greater feeling of self-discovery in that the regularity is only hinted at, using some

Downloaded from http://applij.oxfordjournals.org/ at University of Edinburgh on April 5, 2016

conscious, or coming in between these two extremes, will always be selfdiscovery. The question is to what extent that discovery is guided by the teacher. The guidance, where consciousness-raising is involved, can take more or less time or space and it can be more or less direct and explicit. It is one thing, for example, to set up an illustrative pair of examples and draw the learner's attention to the relevant distinctions using verbal or non-verbal (visual) 'hints' and quite another thing to give a formal rule couched in traditional metalinguistic terms and thereby appeal also to the learner's cognitive analytic capacities.2 In both cases the learner is being made conscious of some aspect of the language itself but the manner varies. We may usefully speak of four basic types of manifestation as far as language-consciousnessraising is concerned. These are represented in Fig. 1: a hypothetical 10-point scale is used; the types are represented within the four cells:

162

CONSCIOUSNESS-RAISING

Downloaded from http://applij.oxfordjournals.org/ at University of Edinburgh on April 5, 2016

linguistic or non-linguistic perceptual clue. Also it is easily incorporated into some naturalistic exercise. It may however be necessary to confine this type of consciousness-raising to relatively simple regularities in the language or to combine it with other techniques. One might imagine some perceptually attractive way of emphasizing the third person singular endings, say, but it is difficult to see how this could be done effectively in an unelaborated manner for semantically complex phenomena such as English aspect. A safer way of ensuring that the appropriate insight was gained in such cases would be to opt for Type B, namely elaborated and explicit guidance. One obvious way of doing this would be along the lines developed by Lev Landa (see Landa 1976 and also Sharwood Smith 1978 for further discussion) in which explanation is broken down algorithmically into easy, highly structured stages giving the relevant differentiations and decisions that the learner must make in order to use the particular pattern or rule correctly. This elaborate manner of presentation can be done in a less explicit manner, that is, more covertly as Type A or there can be a teaching sequence in which the more explicit Type B explanations are followed up with less explicit versions (Type A) where the teacher's direct assistance is gradually reduced via substitute symbolic devices (such as Engels' mediators: see Engels 1970, Sharwood Smith 1978) that serve as mnemonics and 'summarisers' of what was previously explained in full and explicit terms. These four basic types of consciousness-raising, which summarise what are actually two dimensions along which it can vary, should together provide evidence that what people sometimes call teaching about the language can be accomplished in a great number of ways ranging from covert clues 'hidden' in the input organised for the learner to abstract statements and varying also in the time and space devoted to drawing attention3 to the structures in question. The problem of whether learners should themselves be able to verbalise rules (see Lawler and Selinker 1971) or to use Bialystok's term, articulate (see Bialystok 1979) is another matter. Consciousness-raising can clearly be accomplished without requiring of learners that they talk about what they have become aware of. It seems fairly certain that only some learners are able to do this anyway. One of the major flaws in the grammar-translation method, whatever its goals, may indeed have been to emphasise this rather special metalinguistic ability. Another flaw may have been the extent to which emphasis was placed on the rote learning of rules and vocabulary out of context There may be a place for getting learners to articulate rules but this should be tied to the type of learner and the general learning context. There may even be a place for some rote learning. But it is also clear that these questions should be considered apart from the general consideration of whether or not to draw the learner's attention to structural regularities in the course of teaching. This aspect of the grammar-translation method may stand up better to criticism than the other two. Teaching about the language is currently relevant to the kind of debate going on in second language acquisition concerning the role explicit and implicit knowledge play in developing the competence of the learner. There has been a tendency to lay stress on the natural language learning ability that every human being has, irrespective of colour or class. The ability to analyse language in a conscious manner is seen as a different kind of skill, fostered in

MICHAEL SHARWOOD SMITH

163

(a) Learners can only profit from learned knowledge after a certain age, that is, roughly puberty (see Krashen 1979: 153). (b) Learned knowledge is normally only accessible given time and focus on form. (c) Some learners hardly ever (and some never) use learned knowledge (see also Lawler and Selinker 1971 for a related discussion).

There is also the point that only certain easy rules can become part of the learner's 'mental baggage' (see McLaughlin 1978: 319): i.e., there are capacity restrictions on the Monitor. It would seem that, following this line of think-

Downloaded from http://applij.oxfordjournals.org/ at University of Edinburgh on April 5, 2016

formal classroom teaching (by association an educated middle class phenomenon) hence the apparent attractiveness4 of the creative construction hypothesis developed by Dulay and Burt and others which emphasizes the universal developmental patterns supposed to exist in second language acquisition over and above those patterns which are due to various differences in learners' language backgrounds. It seems reasonable to suppose that there are universal patterns, but their extent and significance are still a subject of much discussion (see doubts raised in Rosansky 1977, Larsen- Freeman 1977, Hatch 1978 and Hyltenstam 1978). Of greater interest here is the theory developed by Krashen which incorporates the notions of explicit knowledge and implicit knowledge, what he calls, respectively, learned competence and acquired competence. The evidence for the distinction is inferred or investigated via, amongst other things, studies on morpheme acquisition: differing language backgrounds seem to play no real role when the learner has to perform spontaneously without focussing on form; learned competence (explicit knowledge) is tapped only when there is time and focus on form (see Krashen 1979, Sharwood Smith 1979 and 1980). This theory is dynamic or flexible in the sense that it is under constant revision and more recent versions have played down the generality of learned competence (not everybody has it) and its importance (it does not aid acquisition). Krashen's recent versions of the model are significant for teaching in that there is absolutely no 'interface' between the two types of knowledge. If a given learner 'learns' a rule at one stage and then, at a later stage gives evidence of being able to use it swiftly and without reflection, we are not permitted to say that learned knowledge has been transferred to acquired knowledge. Rather, the learner has in the intervening time, acquired the rule by another (intuitive) route; that is, in the course of exposure to the target language, the acquisition device has been stimulated to (reconstruct the target rule from the data and not via some kind of 'osmosis' obtaining between the two knowledge sources in the mind of the learner. Thus there are rules which learners 'learn' (consciously) and rules which learners 'acquire'. From learner performance in tests and interviews we may observe that some rules are acquired but the learner is quite unable to account for them in any explicit way, i.e. they are not part of learned competence. In other cases, rules are verbalised by the learners but may not turn up in spontaneous speech and in other cases rules may turn up in spontaneous speech and also in the explicit verbalisations. This and other research (see Bialystok 1979 for example) is strongly suggestive of the fact that the two knowledge sources exist. In Krashen's view, what is here called 'consciousness-raising' would be a luxury of highly dubious value since he holds that:

164

CONSCIOUSNESS-RAISING

'The Monitor' LEARNED COMPETENCE

ACQUIRED COMPETENCE-

OUTPUT

Fig. 2. The Monitor Model.

In Krashen's view, McLaughlin's ideas are just speculations whereas his, Krashen's, are at least, he claims, supported by empirical evidence; at least the question as to how learners seem to be able to perform automatically in the target language without being able to give any account of the rules does indeed need to be answered. We may note in passing that the fact that learners are not able to verbalise rules or even indicate in some indirect ways facts about the target language does not mean that at some time in their learning career they were not dimly or very clearly aware of structural problems to be overcome. Another model, the one which actually employs the terms 'explicit' and 'implicit' knowledge has been developed by Bialystok (see Bialystok 1979 for a discussion of empirical evidence). Like Krashen, Bialystok has two separate knowledge sources but unlike Krashen she has an interface between them (although she does not seem to draw attention specifically to this difference (cf. Bialystok 1978, 1979)). The model is represented in Fig, 3:

Downloaded from http://applij.oxfordjournals.org/ at University of Edinburgh on April 5, 2016

ing, time taken up to make learners aware of linguistic regularities would be better spent fostering acquisition (see Fig. 2) without consciousness-raising. There have been other attempts to specify the role of explicit and implicit knowledge in second language performance which have not adopted the extreme 'no interface' position taken by Krashen. McLaughlin, in a critical review of Krashen's model (in McLaughlin 1978) claims that the relevant phenomena can be more economically accounted for by using an information-processing model along lines favoured by cognitive psychologists. In McLaughlin's suggested account, there are two types of performance behaviour depending upon whether 'controlled processes' are being used, or 'automatic processes'. The general idea is that one first begins slowly, haltingly, sometimes with a great deal of conscious awareness and then, in the course of time, we are able to automatise the whole process and execute the relevant programmes and routines swiftly and without reflection. Krashen's reaction to this (see Krashen 1979) is to categorise this as a 'learning precedes acquisition' account, which is not allowed in Krashen's model, since learned knowledge acts as a 'Monitor', i.e. a corrective mechanism with acquired output as its input (see Fig. 2). Krashen claims that strings (utterances) are always initiated by acquired competence (that is L2 competence or, where there are gaps, perhaps LI acquired competence).

165

MICHAEL SHARWOOD SMITH LANGUAGE EXPOSURE

INPUT

luSOWLbUOc

Other knowledge

Explicit linguistic knowledge

4

Implicit linguistic knowledge

-I

OUTPUT

Type II Fig. 3. Bialystok's Model (Bialystok 1978).

Only certain aspects of this model will be discussed here. The continuous lines ( ) refer to inevitable processes, in other words obligatory relationships. The broken lines ( ) represent optional strategies in language learning. For example the lines going from R (Response/output), from Other Knowledge (cultural information, etc) and from Implicit Knowledge and (all) terminating in Explicit Knowledge represent the language learning strategy known as 'inferencing', the result of which is information, not previously known, coming to reside in Explicit Knowledge. Of special relevance here is the broken line which connects' Implicit and Explicit Knowledge going in the other direction, that is, towards Implicit Knowledge. This denotes the strategy of practice, that is, formal practice which has the result of allowing information to move from Explicit Knowledge to Implicit Knowledge via automatisation (Bialystok 1978: 77). This represents a similarity between McLaughlin and Bialystok and a link between language learning research and the cognitive-code style of teaching referred to earlier. The pedagogical implications are, indeed, that consciousness raising is not a time-wasting procedure. We may note, also, that no implications are made in Bialystok's model as to the time taken to learn something depending on which route is taken. Thus, although the route from Language Exposure to Implicit Knowledge can be a direct one, there is no necessary conclusion to be drawn which says that it will necessarily take a shorter time to amass information in Implicit Knowledge this way than via -Explicit Knowledge using a strategy of formal practice. One reason why a multi-faceted approach to learning might be preferred is that it takes into account the possibility that second language learners can employ methods that are not open to the child LI learner. The increased cognitive maturity and knowledge of the outside world is brought into play. Krashen's more radical view would be that short cuts via explicit teaching (formal practice) do not exist and that such activities are a waste of time inasmuch as they do not foster intuitive acquisition. Data purporting to show this may however be reinterpreted as evidence of fossilisation whereby learners are not prepared to invest the extra time and energy to automatise what is currently exclusively explicit knowledge. At certain points in their learning career some learners may be prepared to live with this partial deficit (see Vigil and Oiler 1976).

Downloaded from http://applij.oxfordjournals.org/ at University of Edinburgh on April 5, 2016

Type I

166

CONSCIOUSNESS-RAISING

Explicit knowledge

Ic

Output

Input

Other speakers' utterances

Implicit knowledge

Fig. 4. Linguistic input and output: Three potential sources of feedback.

The exclusive use of explicit knowledge is inevitably going to be restricted, especially where 'hard rules' are involved. It may be considerably more likely in foreign language learning where the learners are of appropriate age and maturity and their instruction is exclusively formalistic, that is, focussing on Type D teaching. However, the point at issue here is not whether it is frequent but rather whether it is in principle possible. This is presumably an empirical question and experiments using learners' introspections might help to shed light on this. But even when we consider situations where learned knowledge and acquired knowledge interact to produce utterances (see arrow 2 in Fig. 4) the ultimate, unified results are presumably available to the speaker as potential feedback into acquired.(as well as learned) knowledge. Thus utterance arrived at partly via explicit knowledge can come to affect implicit (acquired) knowledge. This means that, when considering the input to either of the two knowledge systems, we should include not only incoming utterances from other speakers or writers but also the utterances produced by the learner. The

Downloaded from http://applij.oxfordjournals.org/ at University of Edinburgh on April 5, 2016

Whatever the view of the underlying processes in second language learning are concerned, it is quite clear and uncontroversial to say that most spontaneous performance is attained by dint of practice. In the course of actually performing in the target language, the learner gains the necessary control over its structures such that he or she can use them quickly without reflection. Now let us suppose, looking at Fig. 4, that some aspects of second language performance can in principle be planned from the start entirely on the basis of explicit knowledge: a good oral example of this would be preparing a short question, a speech or telephone conversation where certain things can be predicted in advance. You know what you will have to say. You have attended an effective course of formal instruction giving you a range of procedures which allow you to put together utterances in a completely conscious manner. Let us also suppose that this type of activity is repeated again and again. In such situations, it is surely reasonable to suppose that a certain number of structures planned and performed slowly and consciously can eventually develop into automatised behaviour. Utterances initiated by explicit knowledge (see arrow no. 1) can provide feedback into implicit knowledge. A good example would be placing the verb at the end of a subordinate clause in German; bad examples would be linguistic phenomena so complex and diffuse as to defy pedagogically attractive explanation: here conscious attention may be evoked but with no productive result.

MICHAEL SHARWOOD SMITH

167

NOTES 'This paper is a much extended and adapted version of a paper given at the 2nd Annual Conference on Tacquisition d'une langue etrangere: problemes et perspectives' entitled: 'Alternative perspectives on the notion 'explicit knowledge' in second language learning' at the University of Paris-Vincennes. April 1980. The paper has hopefully benefited from useful comments made during the discussions. Thanks are due to Vivian Cook, Eric Kellerman, Bill Littlewood, Barry McLaughlin and Give Purdue, amongst others who attended the Colchester Second Language Acquisition Workshop 1st June, 1980. 2 An interesting example of this technique, in a strictly learner-research context, is the test in Bialystok and Frohhch (1980) designed to assess inferencing ability by indirectly suggesting cross-language correspondences between Danish and English using pairs such as hvad-what.

Downloaded from http://applij.oxfordjournals.org/ at University of Edinburgh on April 5, 2016

verb placed consciously at the end of the clause under orders from the Monitor can become part of a total utterance produced in some natural or naturalistic context. If the utterance is judged successful (see Vigil and Oiler 1976, Sharwood Smith 1979) it is reasonable to assume that it can partake in the general input into either of the knowledge sources. The flow of information from one source to others is formalised, in Fig. 4, as a process mediated by the learner's own output. In this model this is represented as a potential movement via output without specifying at the moment whether only the final or latter stages of the process of production are involved or whether any intermediate output during utterance planning might be involved as well. The model represents the storage and movement of linguistic information but it is assumed that a full account should formalise the role of the assessment made by the learner of the cognitive and affective impact of his or her utterances and any feedback where the learner has observed and is sensitive to assessment by interlocutors. In conclusion, consciousness-raising cannot or should not be treated simplistically. In the same way extremist approaches to language teaching whether or not they are based on such hypotheses and others neither reflect the great truths nor indeed the practice and interpretation by the average language teacher. While the empirical evidence for the impermeability and primacy of the acquisition device in the second or foreign language learner is hotly contested, there is every reason to accept the older, intuitively attractive version which says that explicit knowledge may aid acquisition via practice; learners who cannot articulate rules may still have access to the relevant information in explicit knowledge; learners who do not have such access may well at one time have had access to such information when the relevant rule had not yet been automatised; learners who do not appear to automatise rules that they have had in explicit knowledge for long periods of time may simply not be disposed to spend the extra time and energy transferring the information to implicit knowledge: i.e. fossilisation has occurred in that part of their learning development where 'learning' has in fact preceded acquisition. This account does not exclude pure acquisition but it allows for more than what either Krashen seems to allow for or indeed what recent approaches to language teaching methodology seem to allow for. There is nevertheless no reason to assume that consciousness-raising by the teacher and conscious learning by the learner cannot be investigated in a systematic way using some less simplistic model as a guide. (Received June 1980)

168

CONSCIOUSNESS-RAISING

3 For a corresponding discussion of the different degrees of consciousness or 'attention* in the learner, see Tarone 1979. *Kellerman makes a similar inference in his article, Transfer or no transfer, where we are now', (1979) in Studies in Second Language Learning, 2: 1, p. 38.

REFERENCES Bialystok, E., 1978. 'A theoretical model of second language learning*. Language Learning 28, 69-84.

Bialystok, E. and Frohlich, M., 1980. 'Oral communication strategies for lexical difficulties', Interlanguage Studies Bulletin, 5.1, 1-31. Engels, L. K., 1970 The function of grammar in the teaching of English as a foreign language', ITL, A Review of Applied Linguistics 10: 11-24. Faerch, C. and Kasper, G., 1980. 'Processes and strategies in foreign language learning and communication'. Interlanguage Studies Bulletin, S.I, 47-119. Hatch, E. M., 1978. 'Apply with caution', Studies in Second Language Acquisition 2.1, 123-143. Hyltenstam, K., 1978. 'Variation in interlanguage syntax', Working Papers, Phonetics Laboratory, Department of General Linguistics, Lund University, 18. Krashen, S. D., 1979. 'A response to McLaughlin, "The Monitor Model, some methodological considerations*", Language Learning, 29, 151-167. McLaughlin, B., 1978. "The Monitor Model, some methodological considerations'. Language Learning 28, 309-332. Landa, L. N., 1976. Instructional Regulation and Control. Englewood Cliffs, Educational Technology Publications. Larsen-Freeman, D., 1977. 'An explanation for the morpheme order of learners of ESL\ in Hatch, E. M. (ed.) Second Language Acquisition, Rowley, Mass: Newbury House. Lawler, J. and Selinker, L. 1971. 'On paradoxes, rules and research in second language learning', Language Learning, 21, 27-43. Rosansky, E. J., 1977. 'Explaining morpheme acquisition orders: focus on frequency' in Henning, C. (ed.) Proceedings of the Los Angeles Second Language Research Forum. Los Angeles: UCLA English—TESL Department. Sharwood Smith, M. A., 1978. 'Applied linguistics and instructional psychology: a case for transfusion?" in Studies in Second Language Acquisition 1, 2. Sharwood Smith, M. A., 1979. 'Optimalizing interlanguage feedback in the foreign language learner'. Studies in Second Language Acquisition, 2, 2, 17-29. Sharwood Smith, M. A., 1980. 'On the Monitor Model', forthcoming in Cornucopia 1.2. Department of English, University of Utrecht, Holland. Tarone, E., 1979. interlanguage as Chameleon'. Language Learning 29, 1. 181-193. Vigil, N. and Oiler, JU 1976. 'Rule fossilization: a tentative model'. Language Learning 26, 2, 281-295.

Downloaded from http://applij.oxfordjournals.org/ at University of Edinburgh on April 5, 2016

Bialystok, E., 1979. 'Some evidence for the integrity and interaction of two knowledge sources', paper given at the TESOL conference, Boston, 1979, to be published in Anderson, R. (ed.) New Directions in Research on the Acquisition and Use of a Second Language, Rowley, Mass: Newbury House (in press).