Cognitive Processes in Second Language Learning

Vivian Cook

Vivian Cook 
Writing Home

IRAL 1977, 15, 1-20

While some applied linguists have attempted to use the tools of psycholinguistics to investigate second language learning, the results achieved have been far from providing a unified picture. Part of this work has explored the hypothesis that second language learning is basically similar to first language acquisition. One set of articles broadly confirms this hypothesis; Dulay and Burt (1972) found that mistakes made by children in learning a second language could be explained more readily in terms of first language acquisition than in terms of interference from the mother language; Natalico and Natalico (1971) showed that the acquisition of plural inflections by children in a second language followed the same sequence as in first language acquisition; Cook (1973) claimed that foreign adults repeated sentences in similar ways to native children and that they followed the same stages in learning the comprehension of certain 'deep' structures as native children; Kessler (1971) found that bilingual children learnt both languages by progressing from linguistically simple to linguistically complex structures. A second set of articles, sometimes by the same authors, does not confirm the hypothesis: Dulay and Burt (1974) found that children acquired "functors" in a different order when they were learning English as a second language and as a first language; Bailey et al (1974) continued this line of research by establishing that the order of acquisition of functors was the same in foreign children and foreign adults; Politzer (1974) used a developmental scoring test to show that the syntactic structures of foreign children did not develop in the same way as those of native children; Boyd (1975) found general similarities between native children acquiring Spanish and foreign children but certain specific grammatical differences. All in all, these results can be said to establish that first language acquisition and second language learning are similar processes, but differ in specific content and order of acquisition; baking a cake and baking a loaf of bread may utilise the same process but require different ingredients, oven temperature and cooking time.

A partial explanation for these apparent differences between first language acquisition and second language learning can be found in the features of the situation. Native children and foreign learners usually find themselves in vastly different situations. Native children for example acquire their first language 'informally' - they pick it up without specifically being taught it. Most foreign learners acquire a second language 'formally' in a structured teaching situation (Ingram, 1975). Any differences that emerge could equally well be caused by differences in the situation as differences in the learning process; they may be 'accidental' products of the situation. A fair comparison of first language acquisition with second language learning would have to control this variable and cancel out the situational differences in one way or another by, say, comparing something that is learnt informally by the native child with something that is learnt informally by the foreign learner. It is interesting to note that in the research where the situational factor is kept constant, the usually accepted advantage of children over adults in second language learning is not only cancelled but reversed: Asher and Price (1967) found that adults were superior to 10 and 14 year-olds who were in turn superior to 8 year-olds when they were taught Russian by the same teaching technique of the total physical response; Olson and Samuels (1973) found that older children and adults were better at learning German phonemes than younger children when they were taught by the same drill method. To study second language learning, these situational features must be discounted.

A further explanation for this disparity goes back to the distinction between learning language and learning a language. The child acquiring his first language has to find out what language itself is - that it uses symbolic representation, that it has grammatical and phonological structure, and so on. To quote Halliday (1975) 'A child who is learning his mother tongue is learning how to mean.' In a second language this is presumably unnecessary: the learner already knows the potential of language and can go straight on to discovering how that potential is realised in the second language. Partly this gives the second language learner an advantage since he is already aware of what language is. Partly, however, it puts him at a disadvantage since he may not be aware which parts of his knowledge are about 'language' and which are about 'a language'. He may assume, not just that languages are all the same in general terms, but that they are the same in specific details. Hence the problem of interference. One can speculate that this separation of 'language' from 'a language' required for second language learning will facilitate the learning of third and subsequent languages; the learner does not have the same problem of realising that his own language is not the sole system through which linguistic meaning can be communicated after he has learnt his first foreign language. A third explanation for these differences is that they are not caused by language but by other psychological attributes of the second language learner that are necessarily different from those of the native child. Such attributes might be the second language learner's greater memory capacity, his more advanced stage of conceptual development, the larger range of communicative functions for which he employs language, and so on, all of which undoubtedly have some effect on his learning and performance in a second language. This explanation is, however, inadequate when these attributes are left poorly defined and are not themselves investigated. The present article tries to remedy this deficiency by looking more closely at how these 'other' mental attributes are involved in second language learning. However, since the applications of this area of psychological research to second language learning are comparatively new, many theoretical and methodological aspects are unresolved and the results and discussion must be considered highly tentative. In general the methodology adopted consists of taking experimental paradigms well-tested with native children and applying them to foreign adults. This answers the question 'Are foreign adults similar to native children?' but, in the case of a negative answer to this question, does not necessarily go very far towards answering the question 'In what ways are they different?' [FN1]

Speech processing memory

Let us first look at the kind of memory process used in the production and comprehension of speech, which can be conveniently labelled speech processing memory. [FN2]. In spoken language we cannot perceive the whole sentence or utterance simultaneously; instead we have to store part of it while we are processing the rest. Linguists who have looked at this area have mostly felt that the outstanding characteristic of speech processing memory is its limited capacity for syntax: when we find a grammatical sentence difficult to understand it is likely to be because our speech processing memory is overloaded. Yngve (1960) for instance claimed that "his mother's brother's son's daughter's hat" is more difficult to produce than "the hat of the daughter of the son of the brother of his mother"; this was due to its greater syntactic 'depth' in terms of Yngve's model and to the greater load it consequently put on the memory. Savin and Perchonock (1965) found that the more 'transformations' in a sentence the more memory capacity it took up. Kimball (1973) described a parsing system for speech involving a short term memory whose chief characteristic is its limited capacity for syntax. It seems likely that speech processing memory in the adult will be dependent upon syntax whichever language he speaks: syntactic complexity will lead to loss of comprehension whether the sentence is spoken in Japanese or in English. It is, however, open to doubt whether syntactic complexity can be defined in a universal sense, whether in other words the type of complexity that proves difficult for the native speaker of Japanese in his own language will be difficult for the native speaker of English in his own language. Research by Just and Carpenter (1975) shows indeed that the difficulty of verifying sentences involving negation is approximately the same in Chinese, Norwegian and English, though this is not a straightforward matter of the processing of surface syntax. The following experiment was constructed to see whether the syntactic features of an English sentence that strain the memory of the foreign learner are different from those that strain the memory of the native adult or the native child.

Experiment I.

This experiment has been fully described elsewhere (Cook, 1975). Briefly the hypothesis was advanced that comprehension errors are caused by perceptual strategies that are resorted to when the normal capacity of speech processing memory is exceeded. This was tested by comparing the comprehension of English relative clauses by three groups - native children, native adults, and foreign adults. It was found that, when the load on the memory was too great, all three groups tended to use the strategy that the first Noun Phrase in the sentence was the Subject and the first Noun Phrase after the verb the Object. This overload was created in most foreign adults and native children, and in some native adults, by sentences such as "The cat the dog bites likes the horse" where the strategy led to the subjects saying that the cat was biting the horse; in most native adults this overload was created by sentences with the same syntactic structure but with two centre embedded relative clauses rather than one, as in "The cat the dog the man sees bites likes the horse." The syntactic complexity involved was that the relative clause was qualifying the Subject rather than the Object and that the same Noun Phrase had two distinct functions in the sentence: "the cat" in these two sentences is at the same time Subject of "likes" and Object of "bites"; this point has been made in more general terms by Bever (1970) who sees it as reflecting a general constraint on perception that an object can only with difficulty be perceived in two ways simultaneously. This experiment therefore makes it clear that the same syntactic point gives trouble to all people processing English; the difference between native adults and foreign adults or between native children and native adults is one of degree rather than of kind; as one's speech processing memory expands so one finds certain syntactic complexity easier to understand, as is shown by the native children's gradual improvement on this structure compared to the more clear-cut stages of acquisition found for other structures. It seems then that speech processing memory works in the same way in all speakers of English and that, at least with respect to this syntactic point, the foreign learner's behaviour is similar to that of a native child because he has a more limited capacity for syntax in speech process­ing than the native adult.

Primary memory

A second type of memory process is called primary memory, a survey of which can be found in Craik (1971). Primary memory processes information for only short periods of time measured in seconds; if information is to be stored for longer periods secondary memory processes take over. One important characteristic of primary memory is that it processes information in terms of 'sounds'; visual information is recoded into sounds and semantic information is largely absent. A second characteristic is that it is extremely limited in capacity and can process only 3 to 4 'units' of information at a time [FN3]. So far as child development is concerned, little is yet certain about how primary memory develops; capacity as measured in digits has, however, formed part of the classic tests of I.Q. (Binet and Simon, 1913) and shows an increase according to the mental age of the child; it has been established that up to the age of 5 children do not use 'sounds' in primary memory (Conrad, 1971) but some type of visual coding (Conrad, 1972). We can ask whether the second language learner's primary memory capacity is substantially reduced in the second language, in what way this capacity is linked to his level of knowledge of the second language, and whether from the beginning the second language learner tends to use sound recoding, like native adults, or some other form of coding, like native children. Some answers to the first two questions are suggested by Robert Lado (1965) and Daniel Glicksberg (1963) who report a series of experiments in which various aspects of what is called here primary memory were investigated in native adults and adult foreign learners. Among Lado's conclusions were that 'Memory span is shorter in a foreign language than in the native language' and that 'Memory span increases with mastery of the language.' In order to investigate the question posed about the form of coding it was necessary to confirm Glicksberg's results and accordingly the following experiment was carried out.

Experiment II.

The people who were tested numbered 54 and were adult foreign learners of English as a Foreign Language at technical colleges in the London area [FN4]. They divided into two equal groups: 'beginners', drawn from Stages 1 und 2 at Ealing Technical College, and 'advanced', drawn from classes for the Certificate of Proficiency of Cambridge at Waltham Forest Technical College and Ealing Technical College. They spoke 14 different languages. The 'beginners' had been in England an average of 11 months and had studied English an average of 8 months; the 'advanced' had been in England an average of 1 year 4 months and had studied English for 6 years 3 months. Each student was tested individually and was given a battery of tests, the first of which formed Experiment II. In this test the students were asked to repeat strings of randomised digits, starting with a length of 4. If the student repeated the string correctly he went on to a string that was one digit longer, and so on till he made a mistake. Since his mistake might have been accidental, he was given one or two further strings of the same length. If he repeated 2 out of 3 correctly he went on to a longer string; if he was unsuccessful for 2 out of 3 the test finished. The intention was to establish the maximum number of digits that the student could repeat in English and this maximum was defined as the last length at which the student had more than 1 out of 3 correct. The strings consisted of numbers from 1 to 9, randomly jumbled with no number occurring more than once in a given string.

The results of the experiment were that the beginners had an average maximum capacity for digits of 5.9, the advanced group of 6.7 (t.test, p<.05); the average difference between beginners and advanced was therefore 0.8 digits. These results and those of later experiments are shown in Table 1. These figures are close to those for Glicksberg's students which were 6.4 for students near the beginning of an eight week course in English and 6.7 near the end. The normal digit span is about the magical figure 7 for native speakers and was found by Glicksberg to be 7.1 for American graduate students. The results for both sets of

Table 1. Maximum average capacity for different strings (Experiment II and III)

                    Digits     Non-Homophones (NH)     Homophones (H)
Beginners        5.9               4.5                                   3.5
Advanced        6.7               5.0                                  3.7
Difference       0.8               0.5                                   0.2

experiments suggest that capacity for digits is not greatly impaired in a second language. Assuming a span of 7.0 in the native language, a foreign adult beginner is only reduced by 1.1 in the present experiment; Glicksberg explicitly compared the span in the native and second languages and found an initial deficit of 0.8 digits. It may be that students who were closer to being absolute beginners than those tested here might have had further reduction in span; however, they would probably at that level not have understood the test instructions or even have known the words for numbers in English.

Thus so far as digit span is concerned it can be seen that the adult foreign learner behaves like an adult not like a native child. His capacity is not limited to the same extent as the child's. A native child of three for instance, who has already mastered a good deal of English syntax, has a digit span of about 2 and an eight-year old has a span of 5 (Binet and Simon, 1913); the foreign adult beginner who knows little of English syntax has a capacity of about 5.9. The difference between these results and those for Experiment I provides one reason for distinguishing between speech processing memory and primary memory in second language learning: where the memory process depends on features of syntax the foreign adult is subject to the same type of restrictions as the child; where the memory process is minimally dependent on language the adult can transfer his memory capacity to the task substantially unimpaired; in the case of digit span he simply has to remember the vocabulary items in the new language without any further language coding.

Having established a base-line for digit span in Experiment II, an attempt can be made to see what form of coding is used by foreign adults in primary memory.

Experiment III

This experiment was designed to establish the maximum span for strings of words of different types. It has been shown in Conrad (1964) that native adults use phonological coding in primary memory because they are more likely to confuse letters with similar sounds than letters with similar shapes, regardless of whether the letters are presented orally or visually; Conrad (1971) also showed that children develop phonological coding about the age of 5. However, the design of both these experiments rendered them impossible to duplicate with foreign adults for the reason that they did not directly require a language response from the subjects; the fact that they were using language for memory purposes was inferred from the confusions in their memory. Foreign adults, however, have a second language available for their memory processes; it would be almost impossible to tell whether they were using visual coding, phonological coding in English, or phonological coding in their native language without an exceedingly complicated experimental design. Indeed, a pilot attempt to use the card-matching method of Conrad (1971) with foreign adults showed that, while they initially made many mistakes, their results would suddenly shoot up to a high level of accuracy; the obvious explanation was that they realised the test, did not call for a response in English and switched over to their native language for memory coding. Experiment III then, though based on the work by Conrad, does not directly tackle the same point of phonological versus visual coding but rather tries to establish whether primary memory capacity in the foreign adult is affected by phonological and lexical coding.

The same students and method were used as in Experiment II. However, the strings that were used consisted not of digits but of nouns taken from Conrad (1971). These nouns formed two groups: homophones (H) and non-homophones (NH). The H Group were eight similar sounding nouns: "cat, bat, mat, bag, man, hat, rat, tap". The NH group were dissimilar sounding nouns: "bus, spoon, fish, clock, horse, train, girl, hand." Randomised strings were constructed from these two sets, ranging in length from 3 to 8 and the identical method was used to arrive at the maximum memory capacity in each subject for H words and for NH words as was used to establish digit span.

The results of this experiment are combined with those for Experiment II in Table 1. The average capacity for NH words in beginners was 4.5, in advanced 5.0: both beginners and advanced remembered less NH words than digits (t test, p. < .05). The average capacity for H words was 3.5 in beginners and 3.7 in advanced; both beginners and advanced remembered less H words than NH words (t test, p. < .05). The difference between beginners and advanced was therefore 0.8 for digits, 0.5 for NH words, and 0.2 for NH words: while the first of these differences is statistically significant, the latter two are not. Primary memory capacity then varies according to the type of item involved: capacity for digits is greatest; capacity for NH words is second greatest; capacity for H words is least. This tends to support the assertion that foreign adults, even beginners, used more phonological coding in primary memory (and hence had more mistakes) than young native children, though this result may be biassed by the repetition task used compared with the cardmatching task used in Conrad (1971).

Secondary memory and language memory

The mind also has to store information for longer periods of time than a few seconds. The memory processes that store this kind of information have usually been termed secondary memory and are believed to work in terms of semantic networks and meanings (Kintsch, 1970). One form of information that requires long-term storage is the speaker's knowledge of the language, his linguistic competence: we have to remember the grammatical rules, the phonological rules, the lexical items, and the other types of information that make up the language. Therefore we need a memory that can store a variety of linguistic information: this can be called language memory. Now there is an important distinction to be made between language memory and other secondary memory processes: language memory stores the system of the language, the rules that make up the language; it does not store individual examples of the language, which are rather stored in a process that can be called sentence memory. This distinction is similar to that made in Piaget (1973) between the conservation of schema and the memory for instances; Piaget insists that the conservation of the schema means that the schema can be used when necessary, not that a particular instance of its use is brought to mind. So far as language is concerned we store the grammatical rule for, say, comparatives in English so that we can produce and understand comparative sentences; we do not necessarily have to store any particular comparative sentence. In language learning there are therefore two different areas that could be studied; one is the long-term storage of the language itself; the other the long-term storage of individual instances of the language, i.e. sentences. Here we will look further at the question of language memory rather than sentence memory. The language memory of the second language learner has to cope with storing the information about a second language and this leads to two immediate questions: what is the ultimate relationship between these two sets of language memories; how does language memory for a second language develop in the learner? The answer to the first question raises the issue of the different types of bilingualism and will not be pursued further here [FN5] The answer to the second question can be looked at more closely within the framework employed in this article.

Let us look at a part of language memory in greater depth - the organisation of lexical items. A word is not stored as an isolated item but as part of a network of meanings: it contrasts with other words ("good" v. "bad"); it collocates with certain items rather than with others ("hot jazz" but not *"hot waltz"); it includes the meaning of other items ("animal" includes "dog", "kangaroo", "whale", and so on). The dictionary definition of the word is a small part of its meaning in terms of the meaning network of the language; this approach has been confirmed by much psychological work in secondary memory (Kintsch, 1970; Rumelhart, Lindsay and Norman, 1972; Atkinson and Shiffrin, 1968). Looked at in terms of acquisition some fairly consistent trends emerge: while children have 'syntagmatic' word associations linking, say, "blue" and "sky", adults have 'paradigmatic' associations linking "blue" and "red" (Ervin, 1961); the older children get the more they use verbal rehearsal for memory (Daehler, Horowitz, Wynns & Flavell, 1969); the adult differs from the child in the amount of 'clustering' he gives to words and also in the form of clustering employed (Rossi, 1964; Denney and Ziobrowski, 1972). In particular it has been claimed that the development of memory in the child reflects, not the acquisition of new processes, but learning which process to use (Hagen, 1971); even young children can use verbal rehearsal if they choose to but do not make the choice unless they are guided to it (Bernbach, 1967). It is then particularly interesting to see how language memory develops in the second language learner as he will already be able to choose among alternative memory strategies with greater ease than the native child.

Some relevant research has already been carried out that has a bearing on this question. Experiments described in Stolz and Tiffany (1972) show that the shift from syntagmatic to paradigmatic associations may be a function of the frequency of occurrence of the items: adults tend to have syntagmatic associations to rare words compared to familiar words. Thus second language learners may behave similarly because the words are relatively unknown to them. Henning (1973) explored the parallels between first language acquisition and second language learning by comparing the development of vocabulary; starting from the premise that native children progress from predominantly acoustic storage of words in secondary memory to predominantly semantic storage, experiments with foreign adults showed a similar development, the beginners having a greater proportion of acoustic errors, the advanced learners of semantic errors. While his experiment tested the development of the form of coding in memory for vocabulary, it did not directly examine the development of clustering behaviour and accordingly the following experiment was designed to see whether foreign adults tended to cluster items more as they increased their knowledge of a second language.

Experiment IV

The same 54 students were tested as in Experiment II and III, who took it as the last part of their battery. Two lists of 20 words were prepared; one list was taken from Rossi (1964) with minor adaptations and consisted of 5 words for parts of the body ("head, mouth, nose, leg, hand"), 5 words for animals ("bear, cat, pig, horse, sheep"), 5 words for clothing ("dress, shoe, coat, hat, tie"), and 5 words for food ("milk, cake, soup, bread, meat"); a second list using equivalent words was constructed for the present experiment and consisted of 5 more words from each group ("ear, foot, hair, arm, eye; bird, mouse, dog, lion, deer; sock, boot, suit, cap, shirt; beer, tea, jam, wine, cheese"). Each list was jumbled so that a word was never adjacent to a word from the same group. The two lists were read to the students individually using an intonation pattern consisting of a fall on each item and a pause between items. Then after each list they were asked to write down as many items as they could remember and were given as much time as they liked in which to do so. Finally they were shown the two lists and asked if any of the words were unfamiliar to them.

The results from this experiment are shown in Table 2.

Table 2. Memory for word lists (Experiment IV)
          The results refer to each wordlist of 20 words:
                                                         
Beginners Advanced
1.
Average words remembered correctly     5.27         7.05
2.
Clustering ratio for correct words        
0.246         0.302
3.
Average guesses written down                  7.8            9.8
4.
Clustering ratio for all guesses (= RRJ)   0.261       0.207

Beginners remembered less words than advanced students, whether in terms of correct words only or guesses, i.e. all words written down; the results for correct words are statistically significant (Kruskal-Wallis, p < .01). The amount of clustering was calculated by counting every word that was written down adjacent to a word from the same group. Then two clustering ratios could be established: a clustering ratio consists of the proportion of total words recalled to the total of clustered words. So far as all guesses were concerned there was slightly more clustering by the beginners; so far as correct words were concerned, there was little difference between the groups. To sum up the results for this test: beginners remember fewer words than advanced students correctly; in clustering behaviour they do not differ significantly from advanced students. A comparison is possible with the results of Bousfield et al. (1956) where a measure was used called the RRm (no. of repetitions based on meaning) which is the same as the clustering ratio for all guesses used here. The RRm for 9 year-old children was 0.253, close to the figures of 0.261 and 0.207 found for foreign adults here while the comparable figure for adults was 0.421. However, as different experimental techniques were employed and the results for all guesses were not significant here, too much trust should not be placed in this comparison. What the present experiment does show is that clustering of vocabulary is not well-established even in advanced learners of English; further research would have to show whether in the near or complete bilingual clustering behaviour would approximate to that of a native speaker. Henning's results show that second language learning may be accompanied by the same kind of shift in the form of coding as first language acquisition: the present experiment suggests that organisation in terms of clustering does not increase in a straight line in second language learners and that in this respect they differ from children acquiring their first language.

Language and Cognition

Perhaps the most important attribute of the second language learner that necessarily distinguishes him from the native child is that almost without exception he is at a later stage of cognitive development than the native child. In the extreme case of the foreign adult learner, he has passed through the normal stages of conceptual development and thinks in ways that are quite different from those of the child: it has indeed been suggested by Rosansky (1975) that the attainment of the Piagetan stage of Formal Operations is a more valid explanation for the critical period theory of language acquisition than the usual physiological explanation in terms of hemispheric lateralisation (Lenneberg, 1967). Even in the less extreme case of the foreign child learner, he is usually in the situation where his conceptual development has progressed further than his language development in the second language. The question is what difference this cognitive advantage makes to second language learning; before this can be tackled some account must be taken of how conceptual development relates to first language acquisition.

R. Cromer (1974) in a survey of current research concluded that the evidence suggests that cognition precedes language in development: in other words the child learns the operation of 'seriation' before he learns the language of 'comparison'; teaching him the appropriate language has only a marginal effect on his knowledge of seriation (Sinclair-de-Zwart, 1969; Cook, 1971). The child also has to be able to abstract the relevant perceptual and functional aspects of physical objects before he can correctly describe relationships between them; he may be able to put things "in" a cup when the cup is presented to him in its normal position but be unable to cope when it is presented on its side or upside down (Sinha and Walkerdine, 1974). Nevertheless the fact that cognition precedes language in development only applies when there is a relationship between the two; not all language is determined by cognition; parts of it may be independent and follow a purely linguistic development. This claim that some language, if not all, depends on non-linguistic processes is by now a common-place (Bruner, 1975; Halliday, 1974) and it may provide a useful starting point for looking at second language learning, even if there exist some as yet unexplained contrary results (Kelly, Tenezakis & Huntsman, 1974).

Let us therefore take an instance where conceptual development has to precede language acquisition and look at the way that the concept of time develops in relationship to sentence structure in first language acquisition. In English the child has to learn that the sequence of events in real life does not necessarily correspond to the sequence of events in the sentence. If the child hears the two sentences

(1) Before Bill dances, Bob reads.

(2) After Linda wrote, Susan phoned.

at an early age he will treat (1) as a sequence of Bill dancing followed by Bob reading and (2) as a sequence of Linda writing and Susan phoning: in other words he adopts a strategy in which the order of mention of events in the sentence corresponds to the order of events in the world. This order of mention strategy also applies to sentences where the time clause comes second in the sentence such as

(3) Mary talks before Susan shouts.

(4) Caroline sings after Sally dances.

Cromer (1968) and Clark (1970) suggest that this strategy is dominant in the child up to the age of about 4. Clark (1971) established the further point that, leaving aside order of mention, children understand "before" before they understand "after"; she explains this in terms of a model of language acquisition in which the child acquires semantic components of meaning one at a time: "after" has more components than "before" and is therefore learnt later. Clark (1970) also found that children learn coordinate structures such as

(5) George smoked and Bob drank.

before they learn sentences with a time clause. To make the following discussion easier to follow, sentences 1-4 will be labelled with the terms given in Clark (1971), namely "before"1 (sentence 1), "after"1 (sentence 2), "before"2 (sentence 3) and "after"2 (sentence 4). To sum up: children learn "after"1, and "before"2 before they learn "after"2 and "before"1 (order of mention strategy); they learn co-ordinate structures before they learn "before" and "after" clauses; they learn "before" before they learn "after". So far as second language learning is concerned we can investigate whether this sequence of acquisition occurs or whether the second language learner's more mature conceptual grasp of time leads him to a different sequence.

Experiment V

36 foreign adult students of English who were near beginners were given the following test in groups. They had studied English on average for 1 year 11 months and had been in England for 8'/2 >months. Though the students were attending classes at Waltham Forest Technical College these were not the same students that took part in experiments II—IV. The materials consisted of 24 sentences, 4 each of identical construction to each of sentences 1-5 above and 4 of identical construction to the following sentence

(6) Bill dances. Bob reads.

where the co-ordination was implied solely by intonation. The students were first shown the sentence "The man fell down and broke his leg" and were told that they had to place "1" against whichever of the two events happened first: if the two events happened simultaneously they were instructed to place "1" against both of them. On the answer sheet the two parts of each sentence (without "before" or "after") were presented one above the other, the two parts being in a random order. The sentence were jumbled in blocks of six. They were all read with an intonation pattern consisting of a low rise on the first clause and a high fall on the second clause.

The results of this experiment are given in Table 3 and are statistically significant (x2, p < .01). The results show that the students did not find "after".

Table 3. Results of Experiment V.
                                       
Correct answers (max. 144)
"before," (Sentence 1)                    101
"after," (Sentence 2)                         93
"before2" (Sentence 3)          
           104
"after2" (Sentence 4)                        66
Co-ordinate "and" (Sentence 5)        98

Co-ordinate intonation (Sentence 6)  94

and "before"2 easier than "after"2 and "before",: they did not simply follow an order of mention strategy. Also they did not find the co-ordinate sentences easier than the sentences with time clauses. However, they do appear to find "before" easier to understand than "after", 205 correct answers compared with 159, and this difference is mostly due to "after"2 which had only 66 correct answers compared to 93 for "after", (x2, p. < .01). Clark (1971) found the following stages of development:

A.   Order of mention strategy
Bl.  Correct on "before" and "after," but wrong on "after2"
B2. Correct on "before" but wrong on both forms of "after"
C.   Correct on all forms.

The difference between stages Bl and B2 is that Bl are using an order of mention strategy on "after" and B2 are interpreting "after" as if it were "before". In terms of these stages the adult foreigners are mostly at the B1 stage where they are using an order of mention strategy only with "after2". Two explanations are possible. One is that the difference between "before" and "after" is part of English, not part of 'language': the students have not learnt the extra semantic components that distinguish between them. The alternative explanation is that the students were behaving like English children and happen to have been tested at the moment when they were at stage Bl of the normal progression of the native child.

General discussion

To bring together the threads of the argument, mostly the research has confirmed the hypothesis that second language learning is like first language acquisition to the extent that other attributes of the mind are not involved and has given more precision to the nature of some of these attributes. Each area that was tested provided some support for the hypothesis: speech processing memory, which depends on syntax, showed the same type of constraint operating in foreign adults, native adults, and native children; primary memory capacity was most impaired in foreign adults with near-homophones and least impaired with digits; language memory for vocabulary showed the relative lack of clustering of the child rather than the clustering of the adult; the test of the relationship between the concept of time and the sequence of clauses in the sentence produced a more ambiguous result which does not disconfirm the hypothesis. Two further conclusions from Lado (1964) were that 'The difference between the native and the foreign language memory span is greater when the material contains the pronunciation and the grammatical contrasts between the languages' and 'The relationship of memory span to foreign language learning is greater for contextual material than for digits.' The present experiments have reinforced this conclusion and extended it to other attributes of the second language learner's mind in greater depth. This conclusion does of course rest on the assumption that, in spite of the links between language learning and other mental processes, language learning has unique properties of its own, an assumption that is probably still shared by many researchers.

What we are left with, however, is an extremely complex multiprocess model of second language learning and one which even then does not take into account other attributes that are possibly crucial. One of these is the question of language functions; several linguists have now described how language functions develop, (Brown, 1973; Halliday, 1975); the question has to be resolved whether the second language learner starts off with a relatively mature set of language functions or whether he has in some way to retrace the steps he has taken in his first language. A further problem is the question of whether there is a memory process for language acquisition distinct from the processes that have been described here; if the necessity for such a memory becomes recognised, this, too, will have to be taken into account. Compared to this multi-process model of second language learning, models such as 'interlanguage' (Selinker, 1972) seem to lack sufficient dimensions: as well as the learner progressing from native to foreign language in one dimension, he also changes in memory processes and verbal organisation in others and is subject to his previous conceptual apparatus in yet others, all of them interacting to a greater or lesser extent with language learning per se. The difficulty with the multi-process model is deciding the extent to which each process is concerned with language and how it affects language learning. The present experiments go a small way towards showing this but a great deal of work remains to be done to establish the relationship of each process to the others. It would be very convenient if the borderlines were clearly marked. It is unfortunately wellnigh impossible to assess the contribution of all the processes in any given case without extensive investigation. The development of the second language learner reflects not just the differences between the first and second languages or the progression of the native child, but also the complex interaction between language learning and mental processes.

Because of this complex interaction it is dangerous and premature to apply this area of research directly to language teaching; a long chain of argument and research has to be connected before anything but the broadest of guidelines can be offered to the language teacher. In particular the discussion of language learning and cognitive development points to a danger in attempts to organise language teaching in terms of a set of 'concepts' or 'notions' as they are termed in Wilkins (1973). Wilkins provides an outline of the notional categories 'that the European learner will expect to be able to express through the target language.' One example is the 'semantico-grammatical category' of "Time" which he divides into "Point of time", "Duration", "Time relations", "Frequency", "Sequence" and "Age". At one level these notional categories can be taken simply as a pragmatic guide for language teaching which is justified by its results in the teaching situation, and this is, broadly speaking, the attitude taken by Wilkins. At another level, however, these categories can be taken as attempts to draw a conceptual map of the speaker of a European language similar to Whorf's map of the Standard Average European speaker (Whorf, 1956); in this case the notional categories refer to conceptual knowledge and the application to language teaching assumes that there is a set of 'ideas' common to all speakers that can be expressed through the appropriate linguistic form in each language. Now there may well be universal concepts, or indeed concepts shared by the speakers of one language family, and one of the central goals of transformational grammar has been to establish them (Sampson, 1975); many linguists and psychologists would nevertheless feel that this goal had been far from achieved and that those universals that seem plausible are in terms of high abstraction, such as Piagetan operations. Such a concept as "Time" is too imprecise to be useful even when broken down into 6 subdivisions; if European speakers have a common concept of time that they express through language for example why do all foreign students have difficulty with the concept of 'present relevance' as used in the English present perfect? Any analysis of concepts on a theoretical level implies distinguishing those concepts that have nothing to do with language from those that are linked to language in general from those that are linked to a particular language. Until this is done - and at the moment we can barely conceive how it could be done - notional categories are a convenient system for organising teaching in a practical way without any greater significance, a rule of thumb rather than a scientific theory.

Despite the dangers of applying this research directly to language teaching, the main conclusion that has been reached is nevertheless important to the language teacher: second language learning is like first language acquisition to the extent that mental processes other than those involving language are not concerned. In other words the more learning depends on general psychological processes, the less similar first and second language learning will be. This conclusion should result in a re-examination of our everyday teaching practices to see the extent to which each of them involves language processes rather than other cognitive processes. Take the example of structure drills. We should at least be aware whether a particular drill strains the language learner's speech processing memory, which is constrained like the native child's; whether it utilises primary memory, which will be less limited in the second language learner; or whether it employs a 'concept' that is independent of any language, and hence part of the learner's conceptual apparatus, or one that is part of the language being taught and new to the learner. The teacher and coursewriter must in principle be able to assess the contribution of language and non-language processes to any learning activity in the classroom and to adapt the teaching methods and materials accordingly to the proportion of language processes involved. More detailed conclusions must await more detailed research.

REFERENCES

Asher, J. T., and Price, B. S., 'The Learning Strategy of the Total Physical Response: Some Age Differences,' Child Development, 38, 1219-1227 (1967).

Atkinson, R. C, and Shiffrin, R. C, 'Some Speculations on Storage and Retrieval Processes in Long-Term Memory,' Technical Report No. 127, Institute for Mathematical Studies in the Social Sciences, Stanford University, California (1968).

Bailey, N., Madden, C, and Krashen, S., 'Is there a natural sequence in adult second language learning?' Language Learning, 24, 235-243 (1974).

Bernbach, H. A., 'The effect of labels on short-term memory for colors with nursery school children,' Psychonomic Science, 7, 149-250 (1967).

Bever, T. G., 'The Cognitive Basis for Linguistic Structures,' in J. R. Hayes (ed.), Cognition and the Development of Language (1970).

Binet, A., and Simon, T., A Method of Measuring the Development of the Intelligence of Young Children (1913).

Boyd, P. A., 'The Development of Grammar Categories in Spanish by Anglo Children Learning a Second Language,' TESOL Quarterly, 9, 125-135 (1975).

Bruner, J. S., 'From Communication to Language: A Psychological Perspective,' Cognition, 3/3, 255-287 (1974/75).

Bousfield, W. A., Esterson, J., and Whitmarsh, G. A., 'A Study of Developmental Changes in Conceptual and Perceptual Associative Clustering,' Journal of Genetic Psychology, 92, 95-102 (1958).

Clark, E., 'How young children describe events in time,' in G. B. Flores d'Arcais and W. J. Levelt (ed.) Advances in Psycholinguistics (1970).

Clark, E., 'On the Acquisition of the Meaning of "Before" and" After"',' Journal of Verbal Learning and Verbal Behavior, 10, 266-275 (1971).

Conrad, R., 'Acoustic Confusions in Immediate Memory,' British Journal of Psychology, 55,75-84 (1964).

Conrad, R., 'The Chronology of the Development of Covert Speech in Children,' Developmental Psychology, 5, 398-405 (1971).

Conrad, R., 'Form and Color as short-term memory codes in pre-school children,' Psychonomic Science, 27, 225-226 (1972).

Cook, V. J., 'Which came first: the language or the concept?' Multiracial School, I, 22-24 (1972). online version

Cook, V. J., 'The Comparison of Language Development in Native Children and Foreign Adults,' IRAL, XI, 13-28 (1973). online version

Cook, V. J., 'Strategies in the Comprehension of Relative Clauses,' Language and Speech, 18, 204-212 (1975). online version

Craik, F. I. M., 'Primary Memory,' British Medical Bulletin, 27,232-236 (1971).

Craik, F. I. M., and Lockhart, R., 'Levels of Processing: A Framework for Memory Research,' Journal of Verbal Learning and Verbal Behavior, 11, 671-84 (1972).

Cromer, R., 'The development of temporal reference during the acquisition of language,' unpublished doctoral dissertation, Harvard University (1968).

Cromer, R., 'The Development of Language and Cognition: The Cognition Hypothesis,' in B. Foss (ed.), New Perspectives in Child Development (1974).

Daehler, M. W., Horowitz, A. B., Wynnes, F. C, and Flavell, J. H., 'Verbal and nonverbal rehearsal in children's recall,' Child Development, 40, 443-452 (1969).

Denney, N. W., and Ziobrowski, M., 'Developmental Changes in Clustering Criteria,' Journal of Experimental Child Psychology, 13, 275-282 (1972).

Dulay, H., and Burt, M. K., 'Goofing: an indicator of children's second language learning strategies,' Language Learning, 22, 235-252 (1972).

Dulay, H., and Burt, M. K., 'A New Perspective on the Creative Construction Process in Child Second Language Learning,' Language Learning, 24, 253-278 (1974).

Ervin, S., 'Changes with age in the verbal determinants of word association,' American Journal of Psychology, 74, 361-372 (1961).

Ervin-Tripp, S., 'Is Second Language Learning Like the First?', TESOL Quarterly, 8, 111-127 (1974).

Glicksberg, D. H., 'A Study of the Span of Immediate Memory among Adult Students of English as a Foreign Language,' unpublished doctoral dissertation, University of Michigan (1963).

Hagen, J. W., 'Some thoughts on how children learn to remember,' Human Development, 14, 262-271 (1971).

Halliday, M. A. K., Learning How to Mean (1975).

Henning. G. H., 'Remembering Foreign Language Vocabulary: Acoustic and Semantic Parameters,' Language Learning, 23, 185-196 (1973).

Ingram, E., 'Psychology and Language Learning,' in J. P. B. Allen and S. Pit Corder (ed.), Papers in Applied Linguistics (1975).

Just, M. A., and Carpenter, P. A., 'Comparative studies of comprehension: An investigation of Chinese, Norwegian, and English,' Memory and Cognition, 3, 465-473 (1975).

Kelly, M., Tenezakis, M., and Huntsman, R., 'Some unusual conservation behaviour in children exposed to two cultures,' Educational Psychology, 43, 181-2 (1973).

Kimball, J., 'Seven principles of surface structure parsing in natural languages,' Cognition, 2,15-47 (1973).

Kintsch, W., Learning, Memory, and Conceptual Processes (1970).

Lado, R., 'Memory span as a factor in second language learning,' IRAL, III, 123-129 (1965).

Lenneberg, E., The Biological Foundations-of Language, M.I.T. (1967).

Macnamara, J., 'Bilingualism and Thought,' in J. E. Alatis (ed.) Georgetown University Round Table on Languages and Linguistics (1970).

Miller, G. A., 'The magical number seven, plus or minus two; some limits on our capacity for processing information,' Psychological Review, 60, 81-97 (1956).

Natalico, D. S., and Natalico, F. S., 'A comparative study of English pluralisation by native and non-native English speakers,' Child Development, 42, 1302-1306 (1971).

Olson, L. L., and Samuels, S. J., 'The Relationship Between Age and Accuracy of Foreign Language Pronunciation,' Journal of Educational Research, 66, 263-268 (1973).

Politzer, R. L., 'Developmental Sentence Scoring as a Method of Measuring Second Language Acquisition,' Modern Language Journal, 58, 245-250 (1974).

Piaget, J., Memory and Intelligence (1973).

Rosansky, E., 'Neurophysiological and Cognitive Developmental Factors and the Critical Period for the Acquisition of Language,' paper presented at the Third International Child Language Symposium (1975).

Rossi, E., 'Development of Classifactory Behavior,' Child Development, 35, 137-142 (1964).

Rumelhart, D.E., Lindsay, P. H., and Norman, D. A., 'A Process Model for Long-Term Memory,' in E. Tulving and W. Donaldson (ed.), Organisation of Memory (1972). Sampson, G., The Form of Language (1975).

Savin, H. B., and E. Perchonock, 'Grammatical Structure and the Immediate Recall of English Sentences,' Journal of Verbal Learning and Verbal Behavior, 4, 348-353 (1965).

Selinker, L., 'Interlanguage,' IRAL, X, 219-231 (1972).

Sinclair-de-Zwart, H., 'Developmental Psycholinguistics,' in D. Elkind and J. H. Flavell (ed.), Studies in Cognitive Development: Essays in Honor of Jean Piaget (1969).

Sinha, C, and Walkerdine, V., 'Functional and Perceptual Aspects of the Acquisition of Spatial Relational Terms,' mimeo (1974).

Stolz, W., and Tiffany, J., 'The Production of "Child-like" Word Associations by Adults to Unfamiliar Adjectives,' Journal of Verbal Learning and Verbal Behavior, 11, 38-46 (1972).

Waugh, N. C, and Norman, D. A., 'Primary Memory,' Psychological Review, 72, 89-104 (1965).

Whorf, B. L., Language, Thought and Reality (1956).

Wilkins, D. A., 'The linguistic and situational content of the common core in a unit/credit system,' in Council for Cultural Co-operation Council of Europe, Systems Development in Adult Language Learning: A European unit/credit system for modern language learning by adults (1973).

Yngve, V., 'A model and an hypothesis for language structure,' Proceedings of the American Philosophical Society, 104, 444-466 (1960).

Footnotes

1. I am grateful to Paul Meara and Carl James for helpful criticism of an earlier draft of this article.

2. Throughout the discussion, the distinctions between types of memory are seen in terms of depth of processing rather than as discrete stores, as argued in Craik and Lockhart (1972).

3. The contradiction between this figure and the well-known 'magical' number of 7 given by Miller (1956) is explained in  Waugh and Norman (1965).

4. I would like to thank Mrs. Norma Brewer and her staff at Waltham Forest Technical College and Brian Abbs and his staff at Ealing Technical College for putting up with the disruptions that these tests caused to their classes.

5. See Macnamara (1970) for an appropriate discussion.