Experimental Approaches Applied to Two Areas of Second Language Learning Research: Age and Listening-based Teaching Methods

Vivian Cook 
Spelling data
Writing Home
SLA Home

(Chapter 2 of Cook, Experimental Approaches 1986)

The aim of the present chapter is to give the reader a feel for research carried out within an experimental approach to second language learning. Rather than imposing an arbitrary overall classification on the whole field, it covers two areas in some depth—age and listening-based teaching methods—to see the range of techniques that have been used and to examine their advantages and disadvantages.

Age in L2 Learning

People commonly believe that success in L2 learning depends on the age of the learner. Chomsky, for instance, commented in his celebrated review of 'Verbal Behavior' "It is a common observation that a young child of immigrant parents may learn a second language in the streets, from other children, with amazing rapidity . . . while the subtleties that become second nature to the child may elude his parents despite high motivation and continued practice" (Chomsky, 1959, p. 49). Many others have claimed explicitly or implicitly that adults learn a second language less efficiently than young children. The Critical Period Hypothesis advanced by Lenneberg suggested that the ability to learn language naturally atrophied after the early teens (Lenneberg, 1967). The Monitor Model, too, sees a change as learners become able to monitor their output at about the same time, which is not always to the older learner's benefit (Krashen, 1981). One linguist has allegedly issued a challenge to present ten thousand dollars to anyone who can produce a foreign learner who started learning the language after their early teens and cannot be distinguished from a native speaker. Anecdotes about successful and unsuccessful learning of children and adults abound; on the one hand there is the stereotype of Hyman Kaplan, still speaking a unique language of his own after many years; on the other, any public discussion of this invariably yields at least one person present who knows a possible claimant for the ten thousand dollars. Such anecdotes are bound to be coloured by the comparative ability of the immigrant who has not lost his or her foreign accent; the Maurice Chevaliers or the Henry Kissingers stand out but the Robert Maxwells and Laurence Harveys merge with the crowd.

The age question has also become inextricably interwoven with language teaching. There is a common belief that children need different language teaching methods from adults. However, many of the techniques used with adults in the communicative language teaching method popular in recent years, such as communication games, are derived from activities used in primary schools and implicitly deny the difference. The optimum age for starting a second language has also been a perennial issue in education and figured highly for example in the British controversies of the 1960s over whether French should be taught in the primary school.

The consensus of opinion among both linguists and the general public seems to be that adults are worse than children at second language learning. Why should this be? Normally we think of adults as being more mature, more capable than children; universities prefer students of 18 rather than 8. There is something anomalous about an area of human activity where it is a handicap to be an adult. Several explanations have been advanced for this oddity. It may be that the brain loses "plasticity" at some point of development (Penfield and Roberts, 1959), although the evidence for this seems lacking (Whittaker, 1981). It might be that adult cognitive processing inhibits language learning (Rosansky, 1975). Or the specialization of the brain's two hemispheres may hinder language learning, though opinions as to when this happens are various (Lenneberg, 1967; Krashen, 1973). It might be differences in the language or social situation that adults encounter. It may be that emotional changes associated with adolescence hamper language learning. Or it may be that the establishment of a language "ego" in the first language causes the mind to be defensive when meeting new languages (Guiora, et al, 1972). Perhaps one of these explanations is correct. Perhaps, however, it would be as well to check that the original observation of children's superiority at L2 learning is in fact correct: are children indeed better than adults at learning a second language?

To achieve this, ideally, we need to compare the success of two groups of second language learners who differ in no other way except that one group are children, the other adults, through a manipulative method with a controlled situation, or a difference method where the differences between two groups are assessed. Bearing in mind the arguments of the last chapter, this means narrowing the question down to a form that can be answered. What aspects of language should be dealt with? Is it learning over a period of years that is crucial or learning over a period of hours or weeks? Should they be learners in their home countries or learners in the country where the language is spoken? Should the language be picked up naturally or learnt in a classroom? Perhaps all children in all circumstances do better at L2 learning than adults. To show this, however, would mean testing a variety of learners and circumstances. The great danger is to compare adults with children who differ in some other way than age. In particular, it is important to ensure they are in the same type of situation; there may be many differences in situation that could favour the child. As Smith and Braine comment about the "alleged facility of children", "All that these anecdotes demonstrate is that an adult who lives in a sub-culture where his native language is known need not be motivated to learn, and a child who is sent to the local school or nursery in a foreign country is subject to what are surely very effective conditions for learning: massive exposure to the language together with overwhelming pressure to learn" (Smith and Braine, n.d., p. 21). A proper answer to the question involves specifying as closely as possible all the factors that could conceivably make a difference to the performance of the two groups.

Let us now see how this subject has been tackled in L2 learning by examining some of the relevant research. At the end of this section (p. 30) there is a list of twenty-two important papers dealing with age in L2 learning; the first twelve are collected in a single book, Age Differences in Second Language Learning (Krashen, Scarcella, and Long, 1982). These will be referred to by number and, when appropriate, by the authors' names.

First of all, particular features of the syntax or phonology of the different L2s might influence the age at which they are best learnt. It might be, say, that Chinese is difficult for English children but not for adults. Out of the twenty-two articles, nine are concerned with the learning of English (1-5, 10, 12, 22), three with German (6, 13, 15), three French (19-21), two Dutch (8, 9), two Hebrew (2, 16), one Swedish (11), one Russian (7), one Esperanto (18), and one nonsense syllables (14). Or it might be that features of the learners' L1s make learning particular L2s difficult or easy at certain ages. Ten papers concern native speakers of English (2,6-8,13,15,18-21), five speakers of mixed languages (5, 10, 11, 16, 17), three Italian speakers (2, 3, 1), one Swedish (12), one Finnish (12), and one Japanese (22). About three-quarters are therefore concerned with the learning of an Indo-European language as the second language, three-quarters with learning of second languages by speakers of Indo-European languages, not counting five studies where native languages were mixed or not given. The papers concentrate on one language, English, and one group of languages, Indo-European. Whatever they show may not be true for languages other than English or language families other than Indo-European.

A further problem is the definition of age. The crucial comparison is between the language proficiency of learners of two age groups who have learnt the second language for the same period of time; time has to be taken into account not only as the age at which learning started, but also as the duration of learning. Two basic possibilities present themselves. One is to take two groups of different ages and to get them to learn the second language for the same amount of time, a manipulative method; Asher and Price (7) for example taught Russian to four groups aged 8, 10, 14, and adult; Politzer and Weiss (21) taught French to children from first to ninth grades; Locke (13) taught German sounds to kindergarten and first grade children. While this approach deals admirably with short-term learning, it is hard to utilize for the period of 10 years or more that may be necessary to approach near-native performance in a second language. The second possibility is to take a large group of learners and to sort them out according to the duration of second language learning and to the age of starting, a difference method; a group of mixed ages are subdivided into those who started learning French below 11 and above 11, which can in turn be sub-divided into those who have learnt it for more than 5 years or less than 5 years. Patkowski (5) tested immigrants to the USA and looked at age of arrival and years in the USA as factors in their proficiency; Smith and Braine (16) took census returns in Israel (Bachi, 1956) and compared the amount of use of Hebrew for different ages of arrival. The advantage of this approach is that it can settle issues about long-term learning; it substantially relies, however, on the learner's recollections of dates and duration; nor is the type of learning and the situation controlled as in the first solution.

Finally the age of starting can be used in two ways; on the one hand it may be the actual moment when the learner started learning the language; on the other it may be the moment when an immigrant arrived in the target country. Gomes da Costa et al. (15), for instance, looked at the ages at which British university undergraduates had started to learn German. But many studies take age of immigration as the starting point, as we saw in Patkowski (5) and Smith and Braine (16). Using immigrants could confuse several possibilities; the immigrant may have learnt some of the language before arrival; he may have been adopted into an immigrant community in the host community and not started using the target language for some time.

The question of immigrants raises the issue of other learner variables than age itself. Immigrants are only one type of language learner; they meet a particular situation in a particular country. Conceivably, immigrants have different personalities from non­-immigrants; they are selected in various ways from the population of their mother country by political, racial or religious reasons. Their encounter with the host country reflects the status of immigrants with their background at that particular time. A Cuban immigrant to the USA in the 1950s may be a very different person and encounter different people and situations from a Cuban immigrant in the 1970s. Research into age differences among immigrants has to be qualified in terms of the possible limitations reflected in the use of immigrant learners. Eight of the articles concern immigrants to the USA (1-5, 10, 17, 22), two concern immigrants to Israel (2, 16), three concern immigrants to Sweden, Switzerland, and Holland (9, 11, 20).

The other major type of L2 learner is the non-immigrant, that is to say, somebody learning the foreign language in a country where it is not spoken, perhaps the majority of L2 learners in the world today. Some of the studies concern learners of this type. Snow and Hoefnagel-Hoehle (8) for example look at the imitation of Dutch sounds by English speakers in England whereas Gomes da Costa et al. (15) look at English people learning German in England; five studies are about the learning of European languages in the United States (6, 7, 13, 19, 21), one is about children learning English in Sweden (12).

Crisscrossing these varieties of learner is the question of education. Many of the articles choose learners that are highly educated—"they were taken from the upper educational groups" (3), "all subjects . . . had had some college education"(4), and "Most subjects either held professional positions or were continuing their education" (5). The sample is biased towards literate middle-class learners rather than being typical of all learners or even of all immigrants. But education is also involved as a situational variable: have the learners been taught the language in a formal situation or have they picked it up informally? Learners taught in a structured, organized way could show many differences from informal learners. Some of the articles are indeed concerned with actual teaching methods, for instance Total Physical Response (7). Most of the subjects have encountered formal teaching in one way or another; Oyama (3, 4) deliberately selected learners who had had the maximal opportunities to learn the language. Other accounts either report learners taught by mixed teaching methods (15), or use specific teaching methods as part of the experimental technique (13). In most it is probably impossible to sort out the roles played by formal and informal situations; Ekstrand (12), for example, looked at "all immigrant pupils in school ages who were registered as needing special tuition in Swedish and consequently were given such tuition"; that is to say, the children were not only in a "natural" immigrant situation but were also receiving specific instruction.

Finally, sex may also be a factor. Oyama (3) observes that women immigrants have restricted language experiences in the host country, compared with men. A popular belief is that girls are better at language learning than boys, confirmed by Asher and Garcia (1) but not by Gomes da Costa et al., (15) or by Snow and Hoefnagel-Hoehle (8).

In most of the articles roughly equal weighting is given to the sexes, though some are exclusively about males (3, 4, 13).

Next the aspects of language that are compared. No overall linguistic framework can be applied to the experiments. Most deal with various aspects of pronunciation: three evaluate the foreignness of "accent" (2, 3, 11); seven look at diverse aspects of phonological production (1, 6, 8,9,12,13, 21); three look at perception of sounds (4, 9, 21); one looks at the learning of phonological patterns (14). Some deal with syntax: the production of grammatical morphemes (7, 20) and syntax (5, 9, 10); and the comprehension of syntax (7). Amount of language use (16), and telling and understanding of stories (9, 11) are also tested. The aspects are therefore heterogeneous and have notable omissions. There is little connection with discourse or pragmatics, with stress or intonation; only two experiments concern vocabulary (17, 18) and one of those is Esperanto (18). While the learning of phonology may be peculiarly susceptible to age differences, the case is not proved if it is not contrasted with other areas.

Most of the actual techniques used in the experiments inevitably concern phonology. Many consider it in terms of "accent"; commonly this is ascertained by asking the learners to produce specimens of speech and then by evaluating foreignness of accent according to the judgement of native speakers (1), or graduate linguistics students (3); sometimes the learners are simply asked to report on their own accents (2). Degree of accent is essentially a lay person's rather than a linguist's measure in that it does not specify whether the difference between accents is caused by, say, phonemes, stress, intonation or even, in those studies which ask the learner to speak spontaneously, by grammar or discourse (3). Nor is accent necessarily a measure of comprehensibility to native speakers. Reporting one's own accent may also be unreliable in that perhaps adults are less confident about their accents, regardless of how good they are. A more linguistically acceptable way of looking at phonology is to test particular phonemes, ranging from a complete phonetic transcript (11), to a combination of discrimination and production tasks (21), to thirty-three phonemes of German (6), down to three sounds of German (13).

Production of syntax and vocabulary is tested by repetition (1, 9), and by various adaptations of the "wug" test which involves seeing how people attach grammatical morphemes to words (9, 10). Comprehension is tested by responses to commands (7) and choice of pictures (11). Some research uses complex test batteries from outside the experiment itself—Foreign Service Institute levels (5, 15) or other standardized batteries (17, 19). Two use social measures: the number of friends who speak the native language (2); or the amount of perceived language use(16). The remainder use a variety of measures; grammaticality (9) and the learning of strange phonological sequences (14), translation (9, 12), self- and teacher-rating of fluency(22), among others. Overall, though, it can be said that few use the accepted techniques of language testing developed by language teachers in the past few years: the multiple choice tests of the 1970s; the perennial cloze test; or the more recent communicative type tests. Little of the testing involves aspects of language that most language teaching methodologists currently consider important; with one or two exceptions, such as Patkowski (5), they concern linguistic competence dominated by syntax and phonemes rather than communicative or pragmatic competence relating language to its use or situation—grammatical accuracy rather than communicative fluency.

Hardly surprisingly the results of these investigations are by no means in accord with each other. Let us recognize three main groups: those which compare older with younger children, those which compare children with adults, and those which look at the eventual proficiency of adult immigrants. Three apparently paradoxical statements can be made.

1. Older Children are Better than Younger Children at Learning a Second Language

For phonology, this conclusion is advanced by Locke for first grade children versus kindergarten children (13); by Asher and Price (7) for children of 8, 11 and 14; by Snow and Hoefnagel-Hoehle (8-9) for children aged 5-15; by Ekstrand for children aged 8-17 (11) and 8-15-year-olds (12); by Ervin-Tripp (20) for 4-9-year-olds; and by Politzer and Weiss (21) for Grades 1-9. For other aspects of language the conclusion is true for 8-14-year-olds for responses to commands (7), for 6-10 versus 11-15 for "wug" tests (10), for 8-11-year-olds for various syntactic tests (11). The only contrary result is Fathmann's finding that 6-10-year-olds were rated more highly on pronunciation than 11-15-year-olds (10), and perhaps Walberg et a/.'s evidence that amount of experience is more important than age (22). To sum up, older children were better than younger children in all tests but one in which they were directly compared. This applies even to phonology, contrary to the notion of young children as excellent mimics. Most, though not all, of these are short term studies where the language was taught for brief periods to different groups e.g. Asher and Garcia (1) or Locke (13). All but four (7, 8, 13, 21) concern the learning of an L2 in the country where it is spoken.

2. Adults are Better than Children at Learning a Second Language

Thus Snow and Hoefnagel-Hoehle found that "Older subjects were considerably better than younger subjects at pronunciation and only after a period of about a year did the younger subjects begin to excel" (8); in their second paper the results were less compatible with the theory that adulthood confers an advantage, in that the 12-15s outshone the adults (though they were also better than the younger age groups) (9). Asher and Price (7), however, found consistent advantages for adults over all other age groups; Olson and Samuels (6) claimed "Our evidence suggests that adults are superior to children in foreign language pronunciation", supported by Thorndike (18), who is convinced 20-40-year-olds are better than 8, 10 or 12-year-olds, Cheydleur, who found adults better than high school students (19), and Smith and Braine(16), who found an optimum age around 30. Most of the supporting evidence for this statement comes from shorter-term second language learning, usually in the learner's own country.

3. Immigrants Who Start Learning a Second Language Younger End up Better Speakers than Those who Start Older

Again, this is supported by many of the studies — Patkowski's battery of tests (5), Oyama's phonological work (3, 4), work with accents (1, 2, 4), comprehension (4), and mixed tests (17). Mostly, these utilize the technique of retrospectively looking at adults, though Ramsay et al. show the amount of accent depends upon age at arrival for child immigrants (17). Snow and Hoefnagel-Hoehle (9) however found contradictory if complex results with English immigrants to Holland, and Smith and Braine found reported language use increased with age of arrival; "any change would be better located at pre- and post-30 than pre- and post-13" (16, p. 26).

At first sight, then, those papers seem to have established that, on the one hand, the older you are when you start learning a second language, the better; on the other, the older you are, the worse. How can this contradiction be resolved? A direct paradox has been avoided in these three statements by making the first two refer to learners of any type, the third to immigrants. The crucial difference may be whether the learner is in his own country or is starting life in a new country; some feature of immigration, whether in the learner or in the situation, causes the apparent inferiority of adults. But this is only one interpretation of the results and rests on the coincidental factor that much of the evidence for statements 1 and 2 comes from non-immigrant settings, much of that for 3 from immigrants. Krashen, Scarcella, and Long (1982) sum up the articles collected in their book (1-12) in three statements that are superficially similar to the present three, with one crucial difference. Their first claim is that adults are faster than children in the early stages of learning; their second is that older children are faster than younger ones; their third is that those who begin during childhood end up superior to those who start as adults. The crucial difference is that they pick out of the evidence for each statement one feature which they consider to be crucial—the question of short-term versus long-term learning. Children are better in the long term, adults in the short. Most of the evidence for the first two statements is concerned with learning over months rather than years, though with some exceptions; most of the evidence for the third is based on the experiences of those who have learnt a language for many years. They argue that while adults start off better, they tail off rapidly and are soon passed, as for instance the results of Snow and Hoefnagel-Hoehle's research (8-9) suggest, with the adults' advantage diminishing as the year went on.

To sum up, the area of age has been tackled in a number of painstaking articles. Many of them have an orientation towards teaching methodology, though some are concerned with the psychological and linguistic implications of age. The methods used have varied between manipulative, observational and difference. Thorough as this research has been, it is still far from conclusive.

The original question—are adults better than children at learning a second language?—has produced contradictory answers. The contradictions are argued here to lie somewhere in the design of the research—perhaps the factor of immigration, perhaps short versus long-term learning, perhaps the narrow aspects of language that have been tested, perhaps some other factor that is not immediately apparent. How can we settle which is right? One approach would be to re-analyse the existing studies in painstaking detail to see if the important factor can be isolated. The only real way, however, is to undertake new research that specifically controls for the factor we suspect is responsible, whether this may be immigration, length of learning, or whatever.

Nevertheless, the bulk of the research has shown that the facile assumption that children are better than adults at second language learning is far from true. Indeed, most of the research that directly compares them in controlled conditions found that the reverse was true. Only in retrospective research with immigrants, using date of arrival and uncontrolled learning situations, does adult inferiority come to light. While the issue is far from settled, it seems that the advantages of children in most forms of second language learning over the short term are nothing but a myth.

Numbered List of Selected Articles on Age in Second Language Learning

1.Asher, J. and Garcia, G. (1969) The optimal age to learn a foreign language, Modern Language Journal, 38, 334-341.

2.Seliger, H., Krashen, S. and Ladefoged, P. (1975) Maturational constraints in the acquisition of second languages, Language Sciences, 38, 20-22.

3.Oyama, S. (1976) A sensitive period for the acquisition of a non-native phonological system, Journal of Psycholinguistic Research, 5, 261-285.

4.Oyama, S. (1978) The sensitive period and comprehension of speech, Working Papers on Bilingualism, 1978, 1-17.

5.Patkowski, M. (1980) The sensitive period for the acquisition of syntax in a second language, Language Learning, 30, 449^72.

6.Olson, L. and Samuels, S. J. (1973) The relationship between age and accuracy of foreign language pronunciation, Journal of Educational Research, 66, 263-267.

7.Asher, J. and Price, B. (1967) The learning strategy of total physical response: some age differences. Child Development, 38, 1219-1227.

8.Snow, C. and Hoefnagel-Hoehle, M. (1978) The critical period for language acquisition: evidence from second language learning, Child Development, 49, 1114-1128.

9.Snow, C. and Hoefnagel-Hoehle, M. (1977) Age differences in the pronunciation of foreign sounds, Language and Speech, 20, 357-365.

10.Fathmann, A. (1975) The relationship between age and second language productive ability, Language Learning, 25, 245-253.

11.Ekstrand, L. (1976) Age and length of residence as variables related to the adjustment of migrant children, with special reference to second language learning, in G. Nickel (ed.) Proceedings of the Fourth International Congress of Applied Linguistics, Stuttgart, Hochschulverlag, 3, 179-197.

12.Ekstrand, L. (1978) English with a book revisited: the effect of age on second language acquisition in formal setting, Didakometry, 60.

13.Locke, J. L. (1969) Experimentally elicited articulatory behaviour, Language and Speech, 72/3, 187-191.

14.Kuusinen, J. and Salin, E. (1971) Children's learning of unfamiliar phonological sequences, Percept, and Motor Skills, 33, 559-562.

15.Gomes da Costa, B., Smith, T. M. F. and Whiteley, D. (1975) German Language Attainment, Julius Groos, Heidelberg.

16.Smith, K. H. and Braine, M. D. S., Miniature languages and the problem of language acquisition, mimeo, no date.

17.Ramsay, C. A. and Wright, E. N. (1974) Age and second language learning, J. Soc. Psych., 94,111-121.

18.Thorndike, E. L. (1928) Adult Learning, NY, Macmillan.

19.Cheydleur, F. D. (1932) An experiment in adult learning of French at the Madison Wisconsin Vocational School, J. Ed. Res., 26, 259-275.

20.Ervin-Tripp, S. E. (1972) Is second language learning like the first? TESOL Quarterly, 8, 111-129.

21.Politzer, R. L. and Weiss, L. (1969) Developmental aspects of auditory discrimination, echo response and recall, Modern Language Journal, 53, 75-85.

22.Walberg, H. J., Hase, K. and Rasher, S. P. (1978) English acquisition as a diminishing function of experience rather than age, TESOL Quarterly, 72/4, 427-437.

Listening-based Teaching Methods and Second Language Learning

Let us now take a second area of second language learning research to which an experimental approach can be applied. The past few years have seen a dramatic change in attitudes among language teaching theorists towards the skill of listening. Numerous reports have attested to the success of language teaching methods that emphasize listening rather than speaking as the main priority in the early stages of language learning. Theoretical support comes from the ideas of Krashen and others that the most important element in language learning is meaningful input—listening to language that carries an actual message to the learner; "all other factors thought to encourage or cause second language acquisition only work when they provide comprehensible input" (Krashen, 1981, p. 57). Let us call the overall approach to language teaching that emphasizes listening as the primary skill "Listening First".

The essential common creed of Listening First is that the learner should listen to the language without being forced to speak; in the words of Gary and Gary (1981a, p. 332) "foreign language instruction should give primary emphasis to the comprehension skills of listening and reading and should relegate the productive skill of speaking to a later and less conspicuous role in the typical classroom ..." In some ways this is a reaction against the audiolingual emphasis on oral production through dialogue repetition, drills, and so on. But it is also in opposition to other more recent methods, as diverse as communicative language teaching and Community Language Learning (CLL), that insist that a major component in language teaching should be to get the student to create new utterances that express his own ideas and feelings rather than simply reacting to those of others. In communicative language teaching, usually, the students aim at exchanging information with other people in a two-way process rather than being only at the receiving end; in CLL at taking part in the social setting of the learning group. It should perhaps be pointed out that Listening First is a loose label for several methods that stress the importance of listening but otherwise differ quite widely in their assumptions. There seems a world of difference, for example, between those that ban the written word, such as Total Physical Response (Asher, 1981), and those that actively rely on it, such as Gary and Gary (1981a); between those that stress meaningful situations, such as Gary and Gary (1981a), and those that teach words or sentences with no more context than an illustrative picture, such as Reeds, Winitz and Garcia (1977).

The overall advantages of Listening First have been neatly summarized by Gary and Gary (1981b) as:

—cognitive. It is better to concentrate on one skill at a time; learning, listening and speaking simultaneously places too many demands on the learner.

—affective. A major handicap for some learners is that speaking in public in the second language embarrasses or frightens them; they should only have to speak when they feel they are ready for it.

—communicative. Listening is inherently communicative in that the listener tries to work out a message from what he hears; speaking can be parrot-like repetitions or manipulations of sounds.

—media compatability. Listening lends itself to the use of tape-recorders, films, and even computer-controlled language labs; speaking does not fit in so easily with modern technology.

—utility. Students can, in many circumstances, use listening outside the classroom because of the availability of films, television programmes, etc., while they are unlikely to be able to speak in the foreign language outside the classroom itself.

The reason why Listening First is discussed here is not, however, found in such teaching advantages, real as they may be. Rather, Listening First is almost unique among teaching methods in seeing itself within an experimental approach; indeed those who advocate it often come from a background of psychology. Its claims, unlike those of most methods, are based on better evidence than the teacher's experience and hunch. Let us then look at the evidence that has been cited in favour of Listening First as an example of how a teaching method may be treated within an experimental approach. Two major types of evidence are usually advanced; controlled studies of language teaching; and research into first language acquisition.

1. Controlled Studies of Listening First in the Classroom

The usual research paradigm is to have an experimental group who are taught by the teaching method in question and a control group who are taught by a more conventional method, a classic manipulative method with controlled data. For example, Postovsky (1974) divided intakes of military personnel studying Russian into two equivalent groups who had exactly the same hours of instruction, length of study, teachers, and so on, except that for the first 4 weeks of their 12-week course the Experimental group made written rather than spoken responses. Tested in the same way at the end of 6 weeks, the experimental group were ahead of the Control on speaking, reading, and writing skills; after 12 weeks they were still ahead on listening comprehension. A similar design is found in Kunihara and Asher (1965) which compared groups taught by a control and a Listening First Method, namely Total Physical Response, in which the students respond physically to commands; results for retention of "behavioural units" immediately after training and 6 weeks later showed the superiority of the experimental group. An alternative to a two-group design is to test a single group. Winitz and Reeds (1973) carried out a pilot study with two learners to establish the feasibility of their particular Listening First method, which consisted of the presentation of pictures and oral words at the same time, before extending it to two groups learning German for 2 and 3 weeks respectively (Reeds, Winitz and Garcia, 1977); they then measured the extent to which the students could cope with a translation test of written German which they had never encountered before; this established that there was a carry-over of about 80 per cent from the spoken to the written language. Obviously the claims for a one group experiment are narrower than for a two group design; rather than proving the superiority of the method over others it shows, as they point out, that "skill in reading need not be emphasized in language courses from the outset", i.e. Listening First does not handicap reading, even if it is not provably better at teaching reading. Other work that can be cited includes numerous studies by Asher (1981), Gary and Gary (1981a,b), Nord (1980) or indeed the whole volume edited by Winitz (1981) and the paper by Winitz and Garcia in this book.

As a body of research this is undoubtedly impressive in quantity and in quality. One problem is the short duration of many of the experiments, for instance the 8 hours teaching of Reeds et al. (1977) or the 12 weeks of Postovsky (1974); short-term gains do not necessarily correspond to long-term advantages. A more serious design fault is the choice of the control teaching method. Either this is not made explicit, or else the chosen method often does not seem representative of good contemporary teaching methods today; none of the control groups to my knowledge used communicative teaching methodology or Suggestopedia, for example. Mostly, the control method seems to be a hard-line audiolingualism, where the students drill and repeat; for instance "The Cs from the beginning followed the regular DLIWC Russian program with great emphasis on habit-forming drills and oral practice" (Postovsky, 1974, p. 234). Also, the Hawthorn effect may have influenced both teachers and students; any new method may show some gains simply because it is new.

The overall question, however, is whether this research actually shows the paramount importance of listening. What has been demonstrated is the efficiency of teaching methods incorporating something called "listening", either in isolation or in comparison with other methods. Teaching methods are notoriously loose labels for a complex of techniques; Cook (1982) argues that structure drills work because they involve cognitive processing and present a model of conversation, rather than because they belong to an audiolingual method. In other words, a teaching method is not a pure and straightforward "treatment" for an experiment; everything that happens in the classroom involves many different aspects of the learner's mind—memory, motivation, aptitude, cognitive style, and so on. The choice between a method that relies on "listening" as opposed to one that involves production represents a choice in many other dimensions. In addition the results refer to groups, not individuals; it has not been shown that for an individual learner, listening comes before production or is necessary for production. The Listening First approach has at some point to show how listening and production are related in the mind of the individual learner; the comparison of groups must eventually yield to the analysis of the two skills in one learner.

2. First Language Acquisition

First let us dispose of a pseudo-issue—the point that listening must come first because a person cannot utter something before having heard it. In a logical sense, perhaps, someone cannot produce, say, a present perfect "I have been to Paris" if he or she has never heard an example of a present perfect. A counter-argument is that, according to Chomskyans at any rate, some language may already be present in the child's mind and external stimuli only present "triggers" for activating this rather than stimuli for learning it. Part of the problem is defining what counts as hearing something or producing something; the usual argument against stimulus-response theories of learning was, indeed, that people can produce sentences they have never heard before; hearing an example of the present perfect may enable a speaker to produce sentences that are novel, e.g. "I have never met a Martian who didn't like gorgonzola". In the case of a second language, the necessity for having heard examples is not total; something in the new language may be based on the first language; transfer as well as language experience can be a source. More importantly, perhaps, the belief that it is necessary for someone to hear a form before he can use it does not prove that it is necessary to understand it before he can use it. I can happily use the word "irrefragible" in conversation, with slight idea of what it means—a habit not unknown in student essays. The main issue, which will be developed later, is that listening and speaking both take place at many levels. Minimally, it is necessary to have encountered the form at a phonetic and phonological level to be able to use it later. But uttering something does not necessarily involve understanding it.

It cannot be said that the relationship between listening and speaking has been discussed very frequently in first language acquisition research in recent years; there is not a great body of evidence one way or the other, even if there are numerous opinions.

Listening First advocates normally cite a particular set of first language sources as their evidence for the priority of listening, in particular Lenneberg (1962) and Fraser, Bellugi and Brown (1963). the Lenneberg article, for instance, is discussed in Gary and Gary (1981a) and Nord (1980). It described the case of an 8-year-old boy who had "never been heard to use any words" but showed by nodding that he could understand a range of questions about a story, including those "couched in complex grammatical constructions, such as the passive voice". This constitutes a neat refutation of the extreme position that to understand you have to be able to speak. But a single instance of a pathological condition may bear witness to the mind's extraordinary adaptability rather than to the priority of listening over speaking in normal development.

The second experiment, by Fraser, Bellugi and Brown (1963), is often cited by Listening First advocates, e.g. Postovsky (1974), Winitz and Reeds (1973), Nord (1980), Winitz (1981). This experiment asked children aged 3 to choose between pictures illustrating grammatical distinctions such as "The sheep is jumping" versus "The sheep are jumping". The children were tested in three ways: imitation, in which they repeated the experimenter's sentences; comprehension, in which they had to point to the appropriate picture; and production, in which "After repeating the names of the pictures, E points to one picture at a time and asks S to name it". Results showed children were best at imitation, next best at comprehension, and worst at production. Lovell and Dixon (1967) confirmed this with children aged 2-6 and ESN children. But there are basic problems with this experiment. One, pointed out by Fernald (1972), is that the scoring system itself biased the results against production; if he used the Fraser et al. scoring system, Fernald found that comprehension was better than production, but if he used his own scheme, they came out the same. A second problem is the nature of the tasks. Fraser et al. carefully defined production as what happened in the production task, i.e. repetition of the correct sentence. This hardly counts as normal production; Fernald claims it is "highly artificial and more closely resembles delayed imitation than spontaneous speech production". The child's production depends upon his prior comprehension of the two adult sentences; by the nature of the task, production could not exceed comprehension and is likely to fall short, for many accidental reasons—loss of attention, memory span, etc. The tasks themselves can also be attacked because they are unequal in terms of memory; production involved storing two sentences whereas comprehension involved storing one. In addition, Cocking et al. (1981) have shown that picture tasks produce biases in the language; Baird (1972) has shown that the statistical interpretation is insecure and the results could well be due to chance. So several question-marks hang over the original research by Fraser, Bellugi and Brown. For the purposes of the present discussion two final points should be made. One is that the research compares performance, not learning—how comprehension related to production at a single moment rather than how they related over time. Secondly, it should be remembered that what the children were actually best at was imitation; an audiolingualist could claim this as support for getting the students to repeat sentences as much as an advocate of Listening First could cite it to support making the students listen.

Indeed, some first language studies show the opposite—children could produce things they could not comprehend. Keeney and Wolfe (1972) tested children's command of Subject-Verb agreement in English under three conditions: spontaneous speech to the experimenter, elicited imitation of correct and incorrect sentences, and comprehension of noun and verb inflections by sentence completion and pointing to pictures. Children between 3 and 4:11 almost always had correct agreement in imitation and production but scored no better than chance in comprehension. The authors' explanation is that production can rely solely on syntactic rules, such as Subject-Verb agreement; comprehension is based on semantics, since it relates the rule to the situation. If their results are taken as seriously as Listening Firsters have accepted those of Fraser et al. they might make up a powerful argument for teaching production first. However, again it depends on the meaning of "listening"; Keeney and Wolfe are making the same point as Gary and Gary about the inherent communica­tiveness of listening: the different levels of listening need to be separated.

Another counter-example in the first language literature can be found in experiments with reading. Here it has been claimed that "generating" sentences or words oneself improves memory compared with just reading them aloud (Graf, 1980), though there is some controversy over this (Ghatala, 1983). Work with speech functions also showed that producing a sentence oneself helped one to remember it compared with hearing someone else say it (Jarvella and Collas, 1974).

So the basis for the view that listening precedes speaking in first language learning is far from well-founded. In a review, Ingram (1974) rejects the hypothesis that "all comprehension of language is complete before any production occurs" as far too strong; he also rejects the view that "complete comprehension of a specific grammatical point is complete before it is ever produced" because of evidence that children use constructions with only a partial grasp of their meaning; instead he accepts that "Some comprehension of a specific grammatical form or construction occurs before it is produced", a much weaker alternative.

Even if it is accepted that listening precedes production in LI acquisition, using this as evidence for Listening First depends upon a close similarity between first and second language learning. Yet after a decade of articles, this issue remains substantially unresolved. Cook, Long and McDonough (1979) suggest that there is no simple, overall answer; specific areas of L2 learning need to be compared with specific areas of LI learning. It is true that most comparisons have come up with similarities rather than differences, for example the survey by McLaughlin (1977); but this cannot be taken for granted in areas that have not been explored. LI research can only be applied to L2 teaching if it has in effect been replicated with L2 learners. What is missing is direct evidence from L2 learning, as we saw in another connection in the last chapter.

The general problem with the evidence from teaching studies and from first language acquisition is the term "listening": has the question been posed precisely enough? What does it in fact mean to say someone "listens" to something or "hears" something or "understands" something? Fraser et al. (1963) point out, "Any infant that can clap its hands can appear to its parents to understand 'pat-a-cake'; any animal that can run to its food dish can appear to its owner to understand 'Come and get it'". Throughout the discussion we have come across the vagueness of "listening" in particular contexts; for example, the difference that is often made in the Listening First research between "hearing" at a mechanical level and "understanding" meaning. Mostly, listening is equated with listening for meaning, with full understanding, rather than with the dog's ability to respond to "Come and get it". But is it being compared with oral production at the same level of meaningfulness? Gary and Gary (198 lb) insist that the student must respond actively to what he hears; "By listening we mean active listening, a process whereby the learner is actively attempting to understand and respond to oral communication carefully presented in a meaningful context". This they contrast with statements about production, such as "much oral practice—at least as commonly used in the classroom—is manipulative rather than communicative". In terms of methods, as we have already pointed out, it may be that the comparison unfairly contrasts listening techniques involving deeper levels of meaning with audiolingual or grammar-translation methods that effectively ban meaningful use of the language. Turning to L2 learning rather than teaching, the same point can be put in terms of a "depth of processing" model. Craik (1973) proposes that memory works at different levels, going from a superficial phonological level of processing through progressively deeper levels of syntax and semantics; Perfetti (1978) has devised a model of listening based on seven such levels. The deeper the processing, the more effective; semantically-processed information is learnt better than that which is just processed phonologically, as for example the research by Graf (1980) and Jarvella and Collas (1974) suggests. The Listening First evidence is open to the interpretation that listening and production are not being compared at equivalent depths of processing—"deep" listening is contrasted with "shallow" production. Anything that engages the mind sufficiently will facilitate learning, regardless of whether listening or production is involved. To rephrase Krashen, all teaching methods work in proportion to the extent that they involve the students' minds in deeper levels of processing.

However, as with age, this is only one interpretation of the research. Only because of the painstaking use of experimental approaches by the Listening First proponents is it possible to pose such questions; methods that have not been researched in this way cannot even begin to be questioned. While it is possible to take issue with the conclusions arrived at in Listening First, it is on the right track in seeking to provide research support for its proposals. It would be a pleasant surprise if other factions in language teaching methodology sought such rigorous justification of their ideas, rather than taking refuge in assertions and anecdotes.

References

Asher, J. (1981) The total physical response: theory and practice, in H. Winitz (ed.), Native and Foreign Language Acquisition, New York Academy of Sciences.

Bachi, R. (1956) A statistical analysis of the revival of Hebrew in Israel, Scripta Hierosolymitana, 3,179-247.

Baird, R. (1972) On the role of chance in imitation- comprehension- production test results, JVLVB, 11, 474-477.

Chomsky, N. (1959) Review of B. F. Skinner's "Verbal Behavior", Language 35, 26-58.

Cocking, R. R. and McHale, S. (1981) A comparative study of the use of pictures and objects in assessing children's receptive and productive language, J. Ch. Lang., 8, 1-14.

Cook, V. J. (1982) Structure drills and the language learner, Canadian Modern Language Review, 38, 321-329.

Cook, V. J., Long, J. and McDonough, S. (1979) First and second language learning, in G. E. Perren (ed.),The Mother Tongue and Other Languages in Education, London, CILTR, 1-22.

Craik, F. I. M. (1973) A levels of analysis view of memory, in P. Pliner, L. Kramer and T. Alloway (eds.),Communication and Affect, Academic Press.

Fernald, C. D. (1972) Control of grammar in imitation, comprehension, and production: problems of replication, JVLVB, 11, 606-613.  

Fraser, C, Bellugi, U. and Brown, R. (1963) Control of grammar in imitation, comprehension, and production, JVLVB, 2, 121-135.

Gary, J. and Gary, N. (1981a) Comprehension-based language instruction: practice, in H. Winitz (ed.),Native Language and Foreign Language Acquisition, N.Y. Academy of Sciences.

Gary, J. and Gary, N. (1981b) Caution: talking may be dangerous to your health, IRAL, XIX/1.

Ghatala, E. S. (1983) When does internal generation facilitate memory for sentences?, American Journal of Psychology, 96, 1, 75-83.

Graf, P. (1980) Two consequences of generation of information on memory performance: increased inter-and intra-word organisation of sentences, JVLVB, 19, 316-327. Guiora, A. Z., Brannon, R. C. L. and Dull, C. Y. (1972) Empathy and second language learning, Language Learning, 22, 111-130.

Ingram, D. (1974) The relationship between comprehension and production, in R. L. Schiefelbusch and L. L. Lloyd (eds.), Language Perspectives—Acquisition, Retardation, and Intervention, University Park Press.

Jarvella, R. J. and Collas, J. G. (1974) Memory for the intentions of sentences, Mem. Cog., 2, 185-188.

Keeney, T. J. and Wolfe, J. (1972) The acquisition of agreement in English, JVLVB, 11, 698-705.

Krashen, S. (1973) Lateralisation, language learning, and the Critical Period; Some New Evidence, Language Learning, 23, 63-74.

Krashen, S. (1981) The fundamental pedagogical principle in second language teaching, in B. Sigurd and J. Svartvik (ed.), Studia Linguistica, 35, 1/2.

Krashen, S., Scarcella, R. C. and Long, M. H. (eds.) (1982) Child-Adult Differences in Second language Acquisition, Newbury House.

Kunihara, X. and Asher, J. (1965) The strategy of the total physical response: an application to learning Japanese, IRAL, 3, 291-300.

Lenneberg, E. (1962) Understanding language without ability to speak: a case report, J. Ab. Soc. Psych, 65,419-425.

Lenneberg, E. H. (1967) Biological Foundations of Language, New York, Wiley.

Lovell, K. and Dixon, E. M. (1967) The growth of the control of grammar in imitation, comprehension, and production, J. Child Psychiat., 8, 31-39.

McLaughlin, B. (1977) Second language acquisition in childhood, Psychological Bulletin, 84, 438-459.

Nord, J. R. (1980) Developing listening fluency before speaking: an alternative paradigm, System, 8,1-22.

Penfield, W. and Roberts, L. (1959) Speech and Brain Mechanisms, Princeton University Press.

Perfetti, C. (1978) Levels of language and levels of processing, in L. Cermak and F. Craik (eds.), Human Memory.

Postovsky, V. (1974) Effects of delay in oral practice at the beginning of second language learning, Modern Language Journal, 58.

Reeds, J. A., Winitz, H. and Garcia, P. A. (1977) A test of reading following comprehension training, IRAL, XV/4, 307-319.

Rosansky, E. (1975) The critical period for the acquisition of language; some cognitive developmental considerations, WPB/TRB, 6, 92-101.

Smith, K. H. and Braine, M. D. S., Miniature languages and the problem of language acquisition, mimeo, no date.

Whittaker, H. (1981) Neurolinguistic aspects of bilingualism, in H. Winitz (ed.) Native and Foreign Language Acquisition, New York Academy of Sciences.

Winitz, H. and Reeds, J. (1973) Rapid acquisition of a foreign language (German) by the avoidance of speaking, IRAL, A7/4, 295-317.

Winitz, H. (ed.) (1981) The Comprehension Approach to Foreign Language Instruction, Newbury House.