Key Issues in SLA
DO SECOND LANGUAGE LEARNERS HAVE A COGNITIVE DEFICIT?
Vivian Cook, Essex University
In Cook 1986 Experimental Approaches to Second Language Learning 1986, 73-80
The relationship between language and cognition has been of perennial fascination to linguists and psycholinguists. Their conclusions range all the way from seeing language as dependent on cognition, through seeing cognition as dependent on language, to seeing language and cognition as independent (Cromer, 1974). One major reason for this diversity of opinion has been the difficulty in deciding where to draw the line between cognition and language or, indeed, if one exists. In particular, it is hard to separate cognitive and language development in the first language, since they both happen simultaneously. Second language learning has a special interest of its own because potentially it is immune to cognitive development. Older learners who have already passed through all the stages of cognitive development are learning a language from an adult viewpoint and their language development is unaffected by cognitive immaturity. With L2 learning it is possible, then, to investigate language acquisition on its own, or alternatively to use L2 learning research as a touchstone for separating language and cognition; if we find the same phenomena in adult L2 learning as in LI learning it may be a safe bet that they are related to language rather than cognition, and vice versa. For instance, Sinclair-de-Zwart showed a link between children entering the Piagetan stage of concrete operations and the acquisition of syntactic structures; children who can "conserve" use slightly different syntax from children who cannot (Sinclair-de-Zwart, 1967). If this breakthrough is the consequence of cognitive growth, we would expect L2 learning in adults who are already past the concrete operations stage to be different from that of LI children in that they should be capable of using the "advanced" structures much sooner. If, on the other hand, their development is similar to that of native children, this should indicate that the syntactic progression is not related to cognition.
In the contemporary discussion of first language acquisition the relationship between language and thinking has re-emerged recently within the Chomskyan paradigm. The overall framework that Chomsky postulates sees the child as having an in-built Universal Grammar, consisting of certain principles that apply to all human languages and certain parameters that vary within preset limits from one language to another. This aspect of Chomskyan thinking is developed more fully in the context of L2 learning in the paper by White in this volume and in Cook (1985). The aspect that is relevant here is Chomsky's separation of language development from cognitive development; he has often claimed that there is no convincing evidence for subordinating language to cognition; language is a separate organ with its own internal structure and obeys its own laws of growth. However, in actual language use, language combines with many other things. Any speech actually produced by an adult speaker, i.e. performance, is subject to constraints upon memory, attention, and physical processes such as breathing that have nothing to do with the speaker's knowledge of language—his competence. In terms of language acquisition, this is complicated by immaturity of the other cognitive and perceptual processes employed in speech possible therefore to separate language "development" (the historical process by which language develops in a child), from language "acquisition" (the abstract idealised model of how the child acquires language itself) (Cook, 1985). Any study of actual children's speech provides evidence of development but not necessarily good evidence of acquisition; Chomsky, for example, claims that "much of the investigation of language development is concerned with something that may not properly belong to the language faculty .. . but to other faculties of the mind that interact in an intimate fashion with the language faculty in language use" (Chomsky, 1981). The sequence in which a child learns syntax may have little to do with acquisition itself but may reflect the way that the child develops the processes for handling information; "suppose it is a fact that children generally acquire the use of simple one-clause sentences before compound sentences; there is no reason to assume that this fact must follow from a particular principle of the theory of grammar, as opposed, say, to some developing property of the perceptual maturation of the short-term memory system" (Hornstein and Lightfoot, 1981, p. 15).
But L2 learning involves a different relationship between acquisition and development. A mature adult learning a second language is not affected by cognitive development; hence L2 learning is that much closer to the acquisition model than the child developing his first language; "second language acquisition data provide a clear window for the investigation of and verification of linguistic universals" (Gass and Ard 1980, p. 451). L2 learning is theoretically useful for language learning in general as it disengages language and cognitive development. If a certain sequence is found in adult L2 learning, it seems likely that it is not based on cognitive development; if it is the same as the L1 sequence we may conclude that it is not related to cognition there either. L1 children go through a particular transition in acquiring structures such as "easy/eager to please" (Chomsky, 1969) at about the age of 7; this might easily be seen as an offshoot of cognitive development, like the structures studied by Sinclair-de-Zwart. But L2 learners of different ages and nationalities also go through much the same stages (Cook, 1973; d'Anglejan and Tucker, 1975; Trosberg, 1983). It would seem, therefore that the simultaneity of the cognitive and the linguistic transitions is coincidental and that the sequence is basically linguistic.
This assumption that L2 development displays acquisition free from the confusing factor of cognition is true, however, in only one or two senses of cognition. Certainly the L2 adult is free of maturational factors, both physical and mental; in so far as first language development depends upon the availability of certain physical abilities to hear or to produce sounds, or upon the possession of late overall stages in a Piagetian sequence, the older L2 learner is free from encumbrances. There may also specifically linguistic maturational stages that the L2 learner does not have to pass through, whether the Hallidayan idea of learning how to mean (Halliday, 1975) or the Gleitman semantic to syntactic shift (Gleitman, 1981). But it is evident that many of the performance processes of cognition are not totally transferred to a second language, in particular what can be called channel capacity—the other mental and physical processes necessary to speech but not part of the development of cognitive stages. A fairly large body of evidence suggests that L2 learners are in fact handicapped in terms of channel capacity, or indeed, may never gain full native capacity. Let us review some of these before looking at a particular experiment.
One area is that of short-term memory (STM)—the ability to store pieces of information for a few seconds. Much of the research on STM in an L2 shows two common features. One is that beginners or low level users of an L2 are nevertheless still able to use a comparatively large proportion of their STM. Miller postulated the famous magical number of seven plus or minus two as the amount of bits of information the mind could store (though this is believed to conceal a component of long term memory as well) (Miller, 1956). The capacity of beginners in an L2 is already quite high; Glicksberg (1963) found an initial capacity of 6.4 digits in low level L2 students; my own research came up with a capacity of 4.9 (Cook, 1977). A small unpublished study of mine found a digit capacity for German digits of 4.5 in English-speaking learners who had only learnt the German words for the digits the same morning. This suggests that in some ways L2 development can draw on superior cognitive powers in adults, whose L2 channel capacity is already far greater than that of native children even after many years of learning. Hence in this respect L2 development is closer to acquisition.
The second common feature to STM research in L2 learning has been the discovery that, despite this initial good start, even at an advanced stage L2 learners still have smaller capacity than native speakers; Glicksberg's students at the end of the course had a capacity of 6.7, as did the advanced students I tested myself. While the L2 learner starts with an initial advantage compared with the native child, capacity equivalent that of native adults may be a long time coming or may indeed never come. A paper by Marsh and Maki (1976) described a similar situation with regard to arithmetic and found that the L2 adult was slower at mental arithmetic in the second language: the authors postulated a cognitive deficit that operated in the second language.
Another form of memory is the long-term memory for the meanings of sentences, which is also part of the cognitive processes involved in language use. Again, it might be that the L2 learner can use this effectively from the beginning; however, no research has addressed this problem directly. At the other end there is, however, the same type of evidence as for STM that advanced L2 learners have a cognitive deficit. Long and Harding-Esch (1977), for example showed that English-speaking learners of French remembered 70 per cent more factual information about passages presented in English, and that they made three times the amount of mistakes in remembering French. Advanced as they were, they were suffering from a handicap in the L2 so far as their ability to extract information from discourse was concerned. Some contradictory evidence has, however, recently been produced by Brenzel (1984).
Let us now look in more detail at a particular experiment set up to explore the relationship between L2 learning and cognitive processes. The intention was to test a particular type of processing claimed to form part of cognition rather than language and to see how this was affected in handling a second language. A convenient starting point was Wason's work with negation (Wason, 1965). Wason suggested that in addition to the normal differences in linguistic processing time between affirmative and negative sentences, negation also required the person to relate the sentence to the situation, i.e. it took additional cognitive processing time according to the relationship between the sentence and the situation. Some negatives are expected to occur in a particular situation, others are not; there are what Wason termed "plausible contexts of denial". As an example let us take the following black and white squares:
We can make four types of statement about these squares:
—similar affirmative (SA): "Square 1 is white", i.e. a positive statement about one of the several similarly coloured squares;
—similar negative (SD): "Square 1 is not black", i.e. a negative statement about one of the several similarly coloured squares;
—dissimilar affirmative (DA): "Square 3 is black", i.e. a positive statement about the only square of that colour;
—dissimilar negative (DN): "Square 3 is not white", i.e. a negative statement about the only square of that colour.
Wason predicted that there would be a difference between positive and negative statements in terms of the time taken to judge whether they are true or false, due to straight linguistic processing of the syntactic structure. He also predicted, however that the similar negative sentence is "implausible" while the dissimilar affirmative is "plausible"; given one square that is different, we expect it to be the exception and described negatively; given a square that is one of several identical squares, we expect it to be described positively. To quote Wason "Given a set of similar stimuli, xl, x2, … and a stimulus y which is perceived to differ from those in one important attribute it is more plausible to assert that y is not x than to assert that x1 is not Wason's original research used circles and the colours red and blue rather than black and white squares; it indeed showed a delay for native speakers processing the similar negative sentences. This seems, then, a useful process to examine in L2 learning because of its apparent independence from language. The prediction is that, if it were a language-independent cognitive process, while both the syntactic and the cognitive delays might be more extreme in an L2, there should be the same difference between the types of negation.
The method involved using Wason's design with a group of bilingual learners in their first and second languages. The equipment used was an Acorn BBC B microcomputer. Subjects were presented with displays of seven red or green squares, in each of which one square was a different colour from the others. They had to judge whether a sentence taken from one of the four types was true or false by pressing T or F, or V or F, depending on language. The time they took to answer each sentence was measured, as was the accuracy of the response, and stored on the computer.
The subjects were twenty adult native speakers of Spanish studying English at University of Essex or the Bell College, Saffron Waldon; a small control group o adult native speakers of English was also used.
The sentences numbered twenty-four, six of each type; half were true, half false. They were presented in randomized order. Equivalent sets of sentences and instructions were prepared in English and Spanish. Half the subjects were given the English version first, half the Spanish version first.
The results are displayed in the following graph:
Let us look at the results in three different ways:
The experiment has then produced a curious result. The expected effects of plausibility were not found in Spanish in Spanish speakers and only to a limited degree in English in Spanish speakers. The conclusion is then that plausibility is not cognitive universal for speech processing that applies to all speakers of all languages all times. Two possibilities suggest themselves:
- that the design is at fault. It might be that there were limits on the machinery; the BBC B may not be an appropriate machine for measuring times, since there is variation in keyboard scan time, and so on. However the time differences in the experiment are of the order of tenths of a second rather than milliseconds and so measurement problems could not really account for the results. It might be that the results are a product of the particular pair of languages employed; there may be something odd about the relationship between English and Spanish as Meara's paper later in this volume suggests, or there may be some critical difference between the forms of negation in the two languages. Or it may have been something to do with the particular subjects—say the fact that they were tested after they had spent some time in the L2 environment rather than in their home country. Magiste (1979), for instance, found a handicap for bilinguals in their first language compared with monolinguals.
- that the plausibility of negation relates to particular languages. Even given the explanations discussed above, it is nevertheless curious that the effect is not replicated. One would expect that a result such as Wason's that seems solidly to demonstrate cognitive processing as opposed to linguistic processing would be confirmed regardless of variations in subjects, languages, machinery, and so on. It might be that the original effect is an artifact of the particular experimental technique employed; or it might be that the interpretation is wrong: it is not cognitive processing that is involved but something peculiar to the particular language that is being tested. Rather than showing how people relate language to situation it shows how people relate a language to a situation. One possibility worth exploring is to relate the Wason paradigm to the model of negation proposed by Klix and Hoffman (1979). They suggest that the meaning of the negative sentence is first compared with the picture; if it does not agree, the meaning is "transformed" into the "complementation of the negated argument” and compared with the picture once again. They predict differences between negative and positive sentences and also between sentences according to truth value, but not, apparently, differences according to plausibility. It seems that Spanish speakers adopt a Klix and Hoffman model but English speakers adopt a Wason model.
Clearly, then, further research is necessary to eliminate some of these possibilities. The present research has shown once again the difficulties of separating linguistic from cognitive processes; while you would expect something that was cognition to transfer to a second language, as for instance the adult transfers rehearsal strategies (Cook, 1981) this is not always the case, as we find here, or for instance in the failure of L2 adults to cluster vocabulary to any great extent (Cook, 1977). The present experiment illustrates the use of L2 learning as a touchstone for distinguishing cognition from language; its results pose the interesting question of whether plausibility is related to language or to cognition.
d'Anglejan, A. and Tucker, G. R. (1975) The acquisition of complex English structures by adult learneers. Language Learning, 25/2.
Brenzel, E. (1984) Summarising in a second language, in H. Dechert, D. Mohle and M. Raupach (eds) Second Language Productions, Gunter Narr, Tubingen.
Chomsky, C. (1969) The Acquisition of Syntax in Children from 5 to 10, MIT Press.
Chomsky, N. (1981) Principles and parameters in syntactic theory, in N. Hornstein and D. Lightfoot (1981).
Cook, V. J. (1973) The comparison of language development in native children and foreign adults, IRAl, XI/1, 13-28.
Cook, V. J. (1977) Cognitive Processes in Second Language Learning, IRAL, XV/1, 1-20.
Cook, V. J. (1981) Some uses for second language research, in H. Winitz (ed.), Native and Foreign Language Acquisition, New York Academy of Sciences.
Cook, V. J. (1985) Chomsky's Universal Grammar and second language learning, Applied Linguistics,6
Cromer, R. (1974) The development of language and cognition: the cognition hypothesis, in B. Foss (ed.) New Perspectives in Child Development, Penguin.
Gass, S. and Ard, J. (1980) L2 data: their relevance for language universals, TESOL Quarterly, XIVI A.
Gleitman, L. (1981) Maturational determinants of language growth, Cognition, 10, 103-116.
Glicksberg, D. H. (1963) A study of the span of immediate memory among adult students of English a: foreign language, PhD dissertation, University of Michigan.
Halliday, M. A. K. (1975) Learning How to Mean, Edward Arnold.
Hornstein, N. and Lightfoot, D. (eds.) (1981) Explanations in Linguistics, Longman.
Klix, F. and Hoffman, J. (1979) The method of sentence-picture-comparison as a possibility for analysing representations of meaning in human long-term memory, in F. Klix (ed.) Human and Artificifial Intelligence, North Holland.
Long, J. and Harding-Esch, E. (1977) Summary and recall of text in first and second languages: some factors contributing to performance differences, in H. W. Sinmaiko and D. Gerver (eds.), Proceedings of NA Symposium on Language Interpretation and Communication, Plenum Press.
Magiste, E. (1979) The competing language systems of the multilingual: a developmental study of decoding and encoding processes, JVLVB, 18.
Marsh, L. G. and Maki, R. H. (1976) Efficiency of arithmetic operations in bilinguals as a function of language, Memory and Cognition, 4/4, 459-464.
Miller, G. A. (1956) The magical number seven, plus or minus two, Psychological Review,60.
Sinclair-de-Zwart, H. (1967) Acquisition du langage et developement de la pensée,Paris.
Trosberg, A. (1983) The acquisition of some complex syntactic structures in LI and L2 learners, Rassegna Italiana Di Linguistica Applicata, XV, 2/3, 261-283.
Wason, P. (1965) The contexts of plausible denial, JVLVB, 4, 7-11.