abstract. Classroom language by aiming at 'competence' has neglected many 'performance' factors that are vital to the use of language. Among these are: the conversion of message into language in speech production; the use of selection and attention in perception; the interaction of production and perception processes; the roles of primary and secondary memory in processing and storing language. In addition the bias towards competence has ignored the process of language learning: specifying a syllabus should involve looking at learning issues as well as the description of competence, even if that is broadened to include communicative aspects.

As originally formulated linguistic competence, the ideal speaker's knowledge of the language, was contrasted with performance, the use of language in particular situations (Chomsky 1965). Though this formulation was initially appealing, it led to problems when it was applied to the areas of psycho-linguistics and sociolinguistics. The solution that was commonly adopted was to construct an intermediate term with some of the properties of performance, some of competence; examples of this are 'abstract performative grammar' (Watt 1970) and 'communicative competence' (Hymes 1971). What implica­tions does Chomsky's original formulation and its later adaptations have for language teaching? A convenient starting point is to compare the language of the classroom with a 'natural' use of language outside the classroom. As a classroom let us take a beginners class in English as a Foreign Language; as a 'natural' use the language of the students chatting to each other in their coffee-break in whatever language they choose.

The lesson starts, say, with a dialogue about the adventures of a young couple in London. The students listen to what the characters say; they repeat the sentences; they read them aloud. Then they practice particular functional and grammatical points arising from the dialogue; they do drills and repeat sentences for pronunciation practice. Halfway through the morning they go to the canteen for a coffee-break.

One group of students sit together and talk about pop music; another group discuss their money problems; a girl describes the symptoms of her cold; a man tries to persuade a girl to go out with him.

The difference between these two are striking. The type of classroom situation in which the students were placed is heavily controlled: nothing happens without the consent of the teacher or the materials writer. The student is not expected to initiate activities: he does what he's told. This control covers not only the choice of activities but also variations within the activities themselves and the language the student hears or is expected to produce; if the student is asked to repeat, he must repeat word for word without varying the syntax, the vocabulary or the pronunciation; if he is asked to vary the struc­ture of the sentences in a drill, he must not take the initiative and decide where or how to vary it: there is one correct answer decided in advance. The coffee-break situation is comparatively uncontrolled. The activities that go on, the topics that are discussed may, indeed, be limited; it would be unlikely that the students would start dancing a tango or discussing whether electrons are made out of quarks. But these limitations are imposed by the students, not by the teacher; the student himself decides that dancing and nuclear physics are not appropriate to the coffee-break. The classroom forces the student to encounter a carefully selected and controlled form of language behaviour; the coffee-break leaves him free to talk about whatever interests him, using the structures and vocabulary he has at his disposal. In the classroom the student is external­ly controlled, in the coffee-break internally controlled.

These differences between classroom and coffee-break reflect an approach to language similar to the distinction between competence and performance. Competence—the ideal speaker's knowledge—is independent of speech production or perception; it contains no errors, meaningless repetitions, and so on. The language of the classroom is often intended to have the same properties. Performance—the processes of speech and examples of actual speech—depends upon mental processes of perception, production, and memory; it tolerates all kinds of mistakes and distortions that are idealised out of competence: so does the coffee-break. Competence is not concerned with the situation at the moment of speaking; neither is the classroom. Performance concerns a specific person talking and listening at a particular time, motivated by his own needs and urges; so does the coffee-break. The typical classroom situation seems to involve an abstraction, from 'natural' language similar to the abstraction of competence from performance.

This superficial resemblance, however, conceals important differences. These can be explored by looking at one aspect of the competence/performance distinction, that in which competence represents static knowledge, performance dynamic process. A corollary of this is that the forms in which knowledge is stored in the mind or the rules by which the linguist represents this knowledge are not necessarily the same as the processes that speakersuse when handling specific bits of language. Linguistic competence is deli­berately distinct from mental process.

While there may be similarities between competence rules and speech processes, every attempt to use competence rules directly as a model of speech processes has been doomed to failure. Take, for instance, the sentence "There's a book on the table." In terms of competence rules it is derived from an underlying structure paraphrasable as "A book is on the table" (Burt 1971). It seems highly unlikely that this derivation corresponds to the order of speech production or perception, in other words that the speaker starts with "A book is on the table" and then changes it into "There's a book on the table" or that the listener hears "There's a book on the table" and has to change it into "A book is on the table" before he understands it. Rules of competence are not the same as processes of performance; the best one can say is that they may provide some clues about production and perception but no more. In addition performance requires more than linguistic competence. When the speaker is actually processing speech, as well as his competence, he makes use of memory processes, physiological processes, general cognitive operations, and so on: the description of an actual speaker in an actual situa­tion requires much more than an account of his competence. But, to come back to the classroom, both these points have been largely ignored in language teaching: the language has been selected and graded in terms of competence; the activities have made nothing but the vaguest concession to the other components of the speech situation.

The classroom implies that the student is the ideal speaker of competence and that performance processes are irrelevant; what happens in the classroom is an approximation to what native speakers would do if they used solely their competence in speech. There are then two implications of the classroom/ coffee-break distinction that go against competence/performance. One is that competence rules are treated as performance processes; the other is that linguistic competence alone is taken to be sufficient for speaking and hearing.

Let us now look at some processes of performance that are specifically excluded from competence and from the classroom. The logical point to begin is speech production. At one time research in this area tended to take refuge in painstaking studies of pauses; their location, duration, frequency, and rhythm were used as evidence for underlying mental processes (Goldman-Eisler 1968). More recent work has postulated a variety of internal processes and usually is based on the assumption that speech production consists first of deciding to say something and then converting this 'something' into speech by selecting the appropriate syntax, lexical items, morphology, and muscular movements (Fry 1968). One model of production, developed by I.M. Schlesinger, starts with the speaker's intentions—what he wants to say; those of his intentions that are linguistically relevant form an 'I-marker' that represents relationships without any information about order or word category.

The speaker converts the I-marker into speech by 'realisation rules' that assign order and word-categories (Schlesinger 1971). The initial input is then crucial: you don't get a ticket out of the machine without first putting in some money; you don't say something unless you have some kind of intention to convey. The distinct processes of production depend on the initial 'message' and cannot be considered in isolation from it. Clearly this view of production has been thought irrelevant to the classroom. The student has been dis­couraged from having intentions or messages of his own; he has not been permitted to follow the normal order of first having something to say and then producing it. Rather the processes have been disconnected; he practises producing utterances without having anything to say. Thus the basic process of speech production—the conversion of message into speech—has been neglected. It seems that the teacher is aiming at a competence-like knowledge of the language independent of the ability to produce speech.

Alongside production are the processes of speech perception, which have shown themselves more amenable to investigation. Some of the factors involv­ed can be seen by taking a brief extract of spontaneous speech. In the following passage an adult educated native speaker of English is describing a cartoon. "Yes well er we got here we have a story er a story in three pictures about um there's a psychiatrist and one of his patients typical American story and um we have the patient lying on the traditional couch er with his pad and er pencil and um obviously he's incredibly bored by the whole the whole thing." Characteristic features of performance illustrated in this extract are the hesitation noises ("er" and "um"), the broken constructions ("a story in three pictures about um there's a psychiatrist"), and the repetitions ("the whole the whole thing"). These are typically excluded from linguistic competence, and edited out of teaching materials.

So far as perception is concerned, their presence means that part of the listener's task in perceiving speech must be the filtering out of irrelevancies. The hearer knows that hesitation noises are not significant; he realises that some repetitions such as "very very" are significant and others are not. He must therefore possess a means for setting aside part of what he hears as irrelevant. This filtering process is allied to the general psychological process of attention—how the hearer selects only the relevant information out of all the information he has been exposed to. Research on this 'cocktail party' problem showed that the hearer can attend to only one message at a time even when he hears several competing messages; he is oblivious except to the message he is attending to. What is still uncertain is the point at which the listener decides between conflicting messages. Some research seemed to show that the choice was made very early in the process of perception and lay between actual physical channels; the hearer can attend to a message coming to his right ear but be unaware of a message simultaneously coming to his left (Broadbent 1958).

Other research suggested that the choice is made somewhat later and that all messages receive some kind of analysis even if they are later rejected (Treisman 1964). In the classroom the student has had the problem of attention and filtering solved for him: he hears speech in idealised conditions where the performance features have already been filtered out and where there is

Only one distinct message; he is rarely exposed to 'noisy' conditions; he does not have to filter out irrelevancies or to select one of several competing messages. Again the processes of performance have been ignored in favour of competence-like language and the student has not been required to use the normal processes of perception. Curiously enough this exclusion of perfor­mance features may well increase the difficulty for the student. While he no longer has to cope with broken constructions, repetitions, and hesitation noises, their absence means that the message is effectively shortened and a certain amount of redundancy cut out. Trying to perceive and comprehend edited normalised speech may well be more difficult than dealing with natural speech, for all its alleged imperfections.

Other processes of perception are similarly excluded from the classroom. It is known, for instance, that there is no simple correspondence between phonemes and acoustic signals; greatly varying acoustic forms can be heard as the same phoneme (Liberman, Cooper, Shankweiler and Studdert- Kennedy 1967). Whatever the hearer uses to perceive speech, it is not primarily the phoneme (Savin and Bever 1970); rather he works out the relevant sounds from a variety of clues : speech perception consists of 'decoding' the vast range of speech sounds into a limited number of significant speech elements. Yet this process of decoding has not been practised in the classroom. In pronunciation classes the learner has been presented with material organised, implicitly or explicitly, in terms of phonemes: he repeats tongue twisters exemplifying particular phonemes; he discriminates minimal pairs. While the native speaker is of course capable of doing this, it can be disputed whether it forms part of the normal process of speech perception. Again a performance process has been left out of the classroom.

- As an illustration of processes of greater abstraction one may take analysis-by-synthesis (Halle and Stevens 1964). In this model perception is subordinat­ed to production. Rather than analysing the speech he hears, the listener continually constructs a message of his own, which he compares with it: we understand speech by matching a message of our own to what we hear. This accounts for the ease with which we can supply the very word for which the speaker is fumbling or can see the way his sentence is heading almost before he does. Analysis-by-synthesis suggests that perception uses the same processes as production in the same order but adds a further matching process.

Consequently we perceive what we can produce; we cannot perceive what we cannot produce. This model has far-reaching implications for lan­guage teaching; for instance, it reverses the normal priorities among the four 'skills' by making the 'passive' skill of listening dependent on the 'active' skill of speaking. Analysis-by-synthesis does have certain grave defects; it is difficult to see how any loophole is left for learning new items since anything new would automatically not be perceived; it is hard put to cope with evi­dence that people who cannot produce speech can nevertheless understand it (Lenneberg 1962). But in a sense this model is simply the extreme statement of a common theme in models of perception, namely the emphasis on the listener's active contribution. He may, for instance, impose upon a sentence degrees of stress that are only present in his own mind(Lieberman 1965); he may utilise deep structures that are not present in the surface (Bever, Lackner and Kirk 1969). The processes involved in listening are far from passive. But the classroom has set this aside; the student does not process speech in a 'natural' way.

Perhaps the most critical aspect of performance that is eliminated from competence and absent from the classroom is memory. Competence is timeless and hence memory-less; the speaker does not take time to produce or understand sentences. In a real situation speech is involved with time; the whole sentence cannot be uttered or perceived simultaneously.

Hence memory is a vital component in the processes of production and perception. Current research commonly recognises at least three types of memory process (Craik and Lockhart 1972). First there is the extremely brief process that registers the sensory input momentarily. Then comes primary memory, which has been extensively studied in recent years (Craik 1971). Primary memory processes information for only a few seconds and then discards it or passes it on elsewhere. It processes language primarily as sounds rather in terms of syntax or semantics; even material that is presented visually is received as sounds. This link between primary memory and sound coding develops in the child about the age of 5 (Conrad 1971). Primary memory can only process a few items at a time, current estimates being between 2½ to 3½ words. So far as the classroom is concerned this is another process conspicuous by its absence. Some attempt has been made to investigate its properties in the second language learner, summarised in Cook (1978) but little has yet been done to apply the results to language teaching.

However advocates of teaching methods that insist on oral presentation before written may draw some consolation from the important role that sound appears to play in primary memory.

The other major form of memory is secondary memory, the process that stores long-term information. Mostly this information is derived from primary memory possibly through 'rehearsal' (i.e. repeating the material audibly or inaudibly), or by organising the information into a form that fits with what is already present in secondary memory. For instance a minister is supposed to have remembered his friends' phone numbers in terms of hymns: Egham 851 became "All things bright and beautiful." There is, how­ever, some evidence that information can reach secondary memory directly without passing through primary memory (Shallice and Warrington 1970). The form in which information is stored in secondary memory seems to be largely semantic and various proposals for semantic features (Kintsch 1970) and semantic networks (Rumelhart, Lindsay and Norman 1972) have been made not unsimilar to developments in transformational grammar. In language teaching, however, the organisation of the material has been made by the teacher on syntactic or logical grounds; the teacher has neither benefit­ed from recent research in the area nor allowed the student to fend for himself. In general it is true that this work on memory processes can as yet provide few concrete points of guidance for the teacher; the precise details of the processes involved in the perception and storage of language are far from established. Nevertheless the teacher should be aware of the centrality of memory processes to performance and to normal language processing: memory is not something we use simply for learning poems by heart but is employed every time we produce or perceive an utterance.

So far we have contrasted competence and performance and have argued that performance processes have been almost entirely neglected in the class­room. But another type of process is involved in language behaviour, namely language learning.

This learning process is not necessarily reflected in the grammar of competence, nor is it necessarily the same as performance processes. The elimination of performance processes from the classroom leads to unreal competence-like language; the only possible justification for it is that it facili­tates second language learning. The teacher seems to be adopting the hypo­thesis that language learning proceeds more effectively when the learner does not have to use normal performance processes. One may well question what evidence there is for this hypothesis. Take the editing of 'mistakes' out of teaching materials for example: does this in fact aid learning? Braine (1971) reports an experiment in which adults learnt an artificial language; the proportion of errors to correct sentences they heard did not affect their learning. Perhaps this could be dismissed as too remote from second language learning. However, direct evidence about second language learning is almost non-existent on this point and in any case does not sufficiently distinguish the effects of language teaching from language learning. Research in first language acquisition suggests that there are indeed differences between the language addressed to children and the normal language of adults, as a recent collection of papers suggests (Snow and Ferguson 1977): certain 'babytalk' features are used (Ferguson 1964); the syntax may be 'simpler' and more redundant (Snow 1972); the functions of interrogatives may be more varied (Holzmann 1972). Language to foreigners may have similar characteristics in terms of simpli­fication and the adoption of special lexical forms (Ferguson 1975). These 'simplified' forms of language seen in babytalk and foreigner talk may reflect something about instinctive adaptation to a more basic form of the language suitable for learners (Corder 1977).

But these do not add up to the peculiarities of classroom language either in quality or quantity. The greatest effect on the language the child produces is the child's mind itself: what he understands, what he says, are limited far more by his own conceptual apparatus and language processing systems than by the interfering adult. Those who have tried to 'improve' the child's language have often found that the best results are achieved, not by restricting the syntax and vocabulary of the language the child hears but by increasing its variety and richness (Cazden 1965). The teacher's hypothesis about second language learning is not confirmed by evidence from first language acquisition.

The present ignorance about the processes of second language learning should not provide an excuse for the teacher to take learning for granted. Suppose that the teacher decides to introduce performance processes into the classroom to a greater extent than hitherto; this would still expose him to two possible dangers. One is simply that of being too selective about the processes he uses, by stressing, say, 'intention' and omitting 'attention'; the present article has tried to give some idea of the complex processes involved and the extent to which they are interdependent. The second danger is that of assuming that by describing the content he has automatically solved the problem of learning. It may be that perceptual strategies are identical to learning strategies (Bever 1970) but this is not logically necessary.

The listener perceiving speech is 'decoding' the message using a code he already knows; the learner acquiring the language is 'codebreaking' the message to retrieve a code he does not know. Codebreaking may well use different processes and memory systems than decoding. So the introduction of' communication into the classroom or the adoption of a notionally-based syllabus does not provide an automatic answer to the question of how language is learnt or how it is to be taught: the 'communicative' teaching of English implies not just using communication in the classroom but teaching how to communicate. The present article has suggested that there is a case to be made for a hypothesis completely the reverse of the assumption that perfor­mance processes hinder learning, namely that the use of performance processes actually helps second language learning. Supporters of the two hypotheses need to find more evidence that will help to decide between these positions since they are so diametrically opposed and have such far-reaching conse­quences for the teaching situation.

REFERENCES

Bever, T.0.1970. The cognitive basis for linguistic structures. In J.R. Hayes (ed.) Cognition and the development of language.

-- , Lackner, J.R. and Kirk, R. 1969. The underlying structures of sentences are the primary units of immediate speech processing. Perception and Psychophysics. 5:4.

Braine, M.D.S. 1971. On two types of models of the internalization of grammars. In D. Slobin (ed.) The ontogenesis of grammar.

Broadbent, D.E. 1958. Perception and communication.

Burt, M.K. 1971. From deep to surface structure: an introduction to transformation syntax.

Cazden, C. 196S. Environmental assistance to the child's acquisition of grammar. Ph.D. thesis, Harvard.

Chomsky, N. 1965. Aspects of the theory of syntax.

Conrad, R. 1971. The chronology of the development of covert speech in children. Develop­mental Psychology. 5:3

Cook, V.J. 1978. Second language learning: a psycholinguistic perspective. Language Teach­ing and Linguistics Abstracts (to appear).

Corder, S.P. 1977. Language continua and the interlanguage hypothesis. In S.P. Corder and E. Roulet (eds.) The notions of simplification, interlanguages, and pidgins and their relation to second language pedagogy. AIMAV, 1977.

Craik, F.I.M. 1971. Primary memory. British Medical Bulletin. 27:3.

--. and Lockhart, R. 1972. Levels of processing: a framework for memory research.

Journal of Verbal Learning and Verbal Behaviour. II.

Ferguson, C.A. 1964. Baby talk in six languages. In J.J. Gumperz and D. Hymes (eds.) The ethnography of communication.

--  1975. Toward a characterization of English foreigner talk.   Anthropological Linguistics. 17:1.1-14.

Fry, D.B. 1969. The linguistic evidence of speech errors. Brno Studies in English. 8.

Goldman-Eisler, F. 1968. Psycholinguistics.

Halle, M. and Stevens, K.N. 1964. Speech recognition: a model and a program for research. In J.A. Fodor and J.J. Katz (eds.) The structure of language: readings in the philosophy of language.

Holzman, M. 1972. The use of interrogative forms in the verbal interaction of three mothers and their children. Journal of Psycholinguistic Research. 1:4.

Hymes, D. 1971. Competence and performance in linguistic theory. In R. Huxley and E. Ingram (eds.) Language acquisition: models and methods.

Kintsch, W. 1970. Models for free recall and recognition. In D.A. Norman (ed.) Models of human memory.

Lenneberg, E. 1962. Understanding language without ability to speak. Journal of Abnormal and Social Psychology. 65.

Liberman, A.M., Cooper, F.S., Shankweiler, D.P. and Studdert-Kennedy, M. 1967. Perception of the speech code. Psychological Review. 74:6.

Lieberman, P. 1965. On the acoustic basis of the perception of intonation by linguists. Word. 21.

Rumelhart, D.E., Lindsay, P.H. and Norman, D.A. 1972. A process model for long-term memory. In E. Tulving and W. Donaldson (eds.) The organisation of memory. Savin, H.B. and Sever, T.G. 1970. The nonperceptual reality of the phoneme. Journal of Verbal Learning and Verbal Behaviour. 9.

Schlesinger, I.M. 1971. Production of utterances and language acquisition. In D. Slobin (ed.) The ontogenesis of grammar.

 Shallice, T. and Warrington, E.K. 1970. Independent functioning of verbal memory stores: a neuropsychological study. Quarterly Journal of Experimental Psychology. 22.

Snow, C.E. 1972. Mother's speech to children learning language. Child Development. 43.

--. and Ferguson, C.A. 1977 (eds.) Talking to children. C.U.P.