Didactics - 9 ; Critique of Krashen V
The Input Hypothesis
A : Recap
We have looked at the concept of interlanguage and of fossilisation, and we have seen that according to theorists such as Selinker and Ellis, language learning proceeds through a series of intermediary languages, with elements being at times in competition with each other. Through a process of elimination and elaboration, the learner progresses, testing his or her hypotheses about the L2 in a variety of ways. The great danger lies in fossilisation - and it appears sometimes that all learners reach a stage at which they fossilize unless there is a radical change in aspects of their environment related to SLA. We may derive from these propositions the idea that mistakes on the part of the learner are a necessary part of the learning process, but that they should be surveyed by the teacher for signs of fossilization.
B : The input hypothesis
A correlate of this position is that when teachers correct output, they do not help the student. Krashen quotes numerous studies that tend to suggest that error-correction has no or little effect upon learner's competence. He cites a study by Cohen and Robbins (1976) that concluded that correction of advanced ESL students' papers over a period of 10 weeks had no significant effect on student errors. In a study by Lalonde, the teacher followed a policy of total error correction, followed by special exercises - students kept track of their errors on an 'error-awareness' sheet. According to Krashen, the effect was very small, with 10 of the 30 students actually getting worse. Nevertheless, Lalonde goes on to recommend a policy of total correction - in the control group, using grammar-based instruction including error-correction, 22 out of 30 actually got worse.
Krashen acknowledges that learners themselves often say that they want to be corrected, both when in a learning situation and when in ordinary conversations with native speakers (Cathcart & Olsen, 1976). Chenoweth et al found that learners said that they liked to be corrected by native speakers during ordinary conversation. However, Cathcart & Olsen went on to look at the real effect of error correction, and found that when teachers tried to provide constant correction, the learners did not like the resulting communication.
As we might imagine, Krashen's position is hotly contested. There are broadly two kinds of objection to the input hypothesis.
- a) He appears sometimes to mean that the input should be written or spoken in such a way that the language itself is comprehensible to the student - hence he refers to Motherese, caretaker language and foreigner talk. This kind of speech, he says, is 'roughly tuned' to the learner's language level, and tends to get more complex as the learner progresses. In this case, it is the language input itself that is modified.
- b) On other occasions, he stresses that the language used must be backed up with a scaffolding of environmental clues - pictures, gestures, objects and so on, which make the meaning clear. In this case, it is the context, the environment, that is modified.
- - K claims that this is what happens in natural language learning situations - however, as we have seen, in some cultures, children are not addressed directly until they are capable of 'giving information' themselves - and yet they manage to learn the language.
- - if the theory is taken seriously by language teachers, it may lead to the learner's being subjected to substandard or language poor samples of the L2 - the Mrs Plornish syndrome.
- (i) Foreigner talk is often of an ungrammatical and misleading nature. 'You comprenez, monsieur?'.
- (ii) Ferguson has noted that many of the features found in Foreigner talk are also found in pidgins.
- (ii) It also happens that, just as we tend to slide into the accent of the person we are talking to, so native speakers adopt the errors of the foreigners that they are speaking to, as a kind of semi-conscious goodwill gesture.
- - moreover, learners are often irritated by 'foreigner talk', as they feel that it expresses contempt. Lynch found that some learners objected to the speech of one of their teachers because they felt that he was 'talking down' to them.
- There is a danger that teachers who oversimplify their language for the learner will not sufficiently well equip her for the experience of full communication with native speakers. Learners sometimes complain that, after their first contact with native speakers, they realise that they have learnt nothing, even though they used to get good marks for their classroom language.
- - the learner gets by by behaving as if they have understood the language, whereas in fact they have read the environment - this appears to be what happens in some immersion programmes - producing classroom pidgin
- This may be a problem with techniques such as TPR. Some of the learners do not listen to the language used to give the orders, but watch the teacher and their classmates, and use the clues and cues given by them to work out what they should do.
- 2. Although most theorists now accept that it is desirable to offer a far richer range of input, and that teachers should be spending more time on the receptive skills than has heretofore been the case. But - many feel that to abandon output altogether would be throwing the baby out with the bathwater. Output of some kind is seen as a necessary phase in language acquisition. There are two kinds of arguments put forward here :
- a) - on the one hand, the teacher needs the output from the student in order to be able to judge the learner's progress, and adapt material to her needs - this is, as it were, a minimalist argument. It is only by correcting your written work, that I can come to some idea of which students need to revise which grammar points.
- b) - on the other hand, it could be argued that it is when the student is called upon to produce language that he will feel a real need to reorganise and elaborate upon his knowledge of the L2 . Jacqueline Boulouffe, for example, holds that :
- - it is only through the student's production that we can check whether she has fully understood the input or not, and that without this verification, there are a number of errors, particularly 'avoidance' errors, that are never cleared up. According to her, these errors are deep-rooted, and come about because language cannot be understood as simply a surface phenomenon.
- The learner will understand the simplest form (l'association simple) - thus 'You must listen to the music' is understood as 'Vous écoutez la musique' - when confronted with an utterance that demands processing of a more complex kind.
- In particular, she points to the issue of modality, or the speaker's specifying her position vis-à-vis what is being said. (This she refers to as 'l'association double')
- But not only does production allow the teacher to check the learner's competence. Boulouffe suggests that personal production, which allows the student to reorganise his knowledge of the language in an authentic attempt to communicate, is essential. It is only when the learner speaks or writes that she is compelled to take modality into account.
Il ressort de tout ce développement que les deux activités de compréhension et de production gagnent à s'imbriquer plutôt qu'à se succéder. L'une prendra le pas sur l'autre selon que l'association simple ou l'association double est en vue : la compréhension favorise l'association simple, liée à l'objectif, au lexical, au pragmatique; la production favorise l'association double, liée au subjectif et à l'intériorisation. La même dichotomie se retrouve, d'ailleurs, dans les instruments de mesure : les tests de compréhension tendent à évaluer l'interprétation pragmatique, alors que les tests de production tendent à évaluer la capacité structurale (Flynn 1986).Le langage est constamment tissé d'associations simples et d'associations doubles. S'il est naturel que les associations simples ne soient produites qu'après avoir été comprises, il est peut-être moins naturel, mais tout aussi probable, que les associations doubles ne soient vraiment comprises après avoir été produites : c'est en ce sens que la production peut rejaillir sur la compréhension. L'appropriation s'enclenche plus sûrement au lieu réel de rencontre d'une demi-expression produite avec un demi-souvenir reproduit, qu'au lieu aléatoire de rencontre entre ce qui est compris avec ce qui reste à comprendre. Si tant est qu'on vise la maîtrise totale du langage et non la seule compréhension, on voit mal comment la position de l'enseignement axé sur la compréhension peut se justifier : remettre la production à plus tard équivaut à postposer l'accès à la subjectivité sans laquelle le langage reste hors d'atteinte. Même si la compréhension crée le terrain de l'apprentissage, la production en constitue probablement le moteur principal.
J. Boulouffe, Les évitements de l'enseignement axé sur la compréhension, dans 'Comprehension-based Language Teaching', Courchêne et al, p. 192.
- Similarly, Swain argues that learners need the opportunity to use the L2 meaningfully, because when they are faced with communication failure, they are forced to make their output more precise, coherent and appropriate. She claims that when students focus on comprehension, most of the work is done at the semantic level - top-down processing - while when they focus on production, they must work at the syntactic level - bottom-up processing.
Swain bases her argument on the experience of immersion classes. When she and her colleagues observed interaction in a French immersion classrooms in 9 grade 3 and 10 grade 6 classrooms, they found that less than 15% of students' utterances were of more than a clause in length, and only 19% of grammatical errors were corrected 'often in a confusing unsystematic way'.
- These arguments suggest that, if comprehensible input is necessary, then so also is comprehensible output. The learner should be forced to produce comprehensible language, and this in turn forces her to focus upon form.
- - demand clarification.
- - demand confirmation - 'Do you mean ...?'
- - repetition
- Pica found that of the three, the first was the most useful, as it ensures that the learner has to reformulate their utterance. Pica and colleagues also found that comprehension increased when meanings were negotiated. This implies that it is important that input and output stages are not radically separated - the learner needs to be able to interact with the input, either through direct negotiation, in the case of oral input, or through mediated negotiation. Pica et al compared the effects of three types of input on the ability of 16 low-intermediate ESL learners to understand oral instructions :
- - unmodified input - the authentic text or recording of an authentic dialogue.
- - premodified input - simplification and greater redundancy - the specially created text, or dialogue, or the highly adapted text.
- - interactionally modified input - begin with unmodified input, but given the chance to demand clarification. (This should not be understood as simply reading the text through in class, and listening to the teacher's explanations).
- It was the third of these that resulted in the greatest comprehension. Some care needs to be exercised about the result, because it may simply be that in the third case, learners benefited from greater quantity rather than greater quality.
- Tanaka & Yamazaki discovered that opportunities to modify input interactionally not only increased comprehension, but also resulted in greater vocabulary acquisition. This kind of finding is backed up by the fact that active students usually acquire language faster than passive ones - although the reasons why this should be so are not altogether clear.
- We will remember that one of the points made by Boulouffe was that without output, teachers cannot identify learners' problems. Lydia White is of the same opinion - she suggests that whereas learners do hear or read what can be done through processing input, they get negative evidence about what cannot be done - the learner cannot simply assume that if she doesn't hear some particular structure or usage, that it does not exist ; certain of their overgeneralizations, for example, will not be disconfirmed. It is only when they produce language, and are corrected by a teacher or a native speaker, that they will discover that they have made an incorrect inference.
- She also suggests that if input is too easy to understand, the students will not make progress. It is through recognising that they do not understand a passive sentence, for example, that learners are forced to make the effort to acquire more language - input should by no means always be comprehensible.
Conclusion
(If you wish to comment or ask a question, please write to tmason@timothyjpmason.com)
| Home | Didactics | Top of Page |