Using SLA research in language teaching by Vivian Cook

Using SLA research in language teaching	Vivian Cook Spelling data Writing Home SLA Home
Version of IJAL 9/2, 1999, 267-284

This paper discusses ways of using second language acquisition (SLA) research in teaching. It argues that the proper use of SLA research should meet a set of overlapping requirements: validity of the research, ethics in obtaining results, generality of research and teaching use, matching of languages in research and teaching, matching of learners and students, and breadth of coverage of language learning. These requirements are applied to three sample areas of research: the question of age-effects in L2 learning, the acquisition of phonology and the acquisition of the writing system. The conclusion is that SLA research can be a source of provocative ideas to use in different levels of teaching and can form a bridge between psychological theories or linguistic descriptions and language teaching.

When second language acquisition (SLA) research began to emerge as a separate discipline, one of the hopes was that it would benefit language teaching (Corder 1973; Stern 1983). In recent years SLA research has mostly seen its role as raising teachers' awareness of SLA concepts rather than affecting teaching directly; Markee (1993) for example sees SLA research as "a resource for changing teachers' professional cultures". However, channelling SLA research through teachers leaves the other people involved in language teaching such as course-book writers and students out in the cold and prevents SLA research from being used in the practical design of coursebooks, examinations and so on (Cook in progress)

This paper suggests putting the usefulness of SLA research to language teaching on a firmer base. It concentrates on the more 'linguistic' aspects, partly for space, partly because these are already covered in books such as Lightbown & Spada (1993). The first section discusses requirements for using SLA research in language teaching, which are applied in the second section to brief samples of SLA research. This paper is not attacking or defending any contemporary SLA positions. SLA research is today a large and heterogeneous area, ranging from Vygotskyan psychology (Lantolf 1994) to connectionism (Broeder & Plunkett 1994); the presentation here is necessarily selective. Seeing SLA research as a valuable resource for teaching does not mean denying its status as an autonomous field with its own goals and interests.

This section suggests some overlapping requirements that any use of SLA research must meet, building on the discussion in Cook (1992, 1996).

SLA research is still a young field embracing a range of methods and theories; it cannot be taken for granted that any piece of research is necessarily sound. The main question is the validity of the SLA research in terms of standard research criteria: is the methodology sound; are the data adequate; do the conclusions come from the actual research; and so on. Though in one sense SLA research has to discipline itself, its proper use depends on these requirements being met. One example is the number of L2 learners involved in the research. Small-scale case studies can provide insights, such as the six Hispanic learners of English in the Cazden, Cancino, Rosansky & Schumann (1975) corpus, which have been reanalysed many times in the SLA literature, for example Gass & Lakshmanan (1991) and Hilles (1991). Some research designs, particularly in the Universal Grammar area, still rely on single subjects (Schwartz & Sprouse 1996) in the belief that what is true for a single learner must be true for all. But such small numbers may lack sufficient people to cover individual variation in learning, sufficient L1s or L2s to counteract variation between languages, a sufficient range of situations, and so on, to be discussed below. Even the mammoth European Science Foundation (ESF) project (Perdue 1993), which covered five different L2s, only took in 26 learners.

In principle, the SLA research should be clearly based on evidence from second language acquisition rather than, say, from L1 acquisition, animal behaviour, or computer simulations, all of which have figured in the past. The validity of research is problematic when it depends upon disciplines that those concerned with teaching or SLA research are not qualified to discuss. Take the example of brain lateralisation. Albert & Obler (1978: 254) interpreted the existing research as saying that the right hemisphere of the brain is more involved in the L2 than in the LI; so "it might be useful to develop a program of second language teaching that emphasises so-called 'right hemisphere strategies' ... through nursery rhymes, music, dance, or techniques emphasising visuo-spatial skills". Paradis (1992) claims the original research itself did not bear out this conclusion and that there is still no certain evidence that L2s are stored in the brain differently from L1s. The lack of validity of the research undermines the teaching application. Any benefits from nursery rhymes are unrelated to the involvement of the right hemispheres of students' brains. Validity leads in to the question of generality discussed below.

While the extreme ethical issues involved in animal research hardly apply, SLA research should nevertheless not inflict damage on the learners involved and should be conducted with their consent, to the extent that consent is possible. Teaching English pronunciation through artificial bridges in the mouth needs careful sanctioning by the learners and dentists (Wenk 1982), just as the use of hypnosis needs appropriate medical ethics (Schumann et al. 1978). But the participants in more straightforward experiments also have the right not to have their language-learning goals or educational achievement impaired by the research. To students, L2 learning is part of their lives, not an arbitrary research task done at a researcher's whim. For instance, one popular research design splits students into two groups, one of which is given the treatment the researcher believes is effective, the other a control treatment predicted to be less effective. To a degree, as in drug trials, the researchers have withheld what they consider the optimum treatment from some students and thus have disadvantaged them for the sake of improving conditions for future students. While this hardly threatens the students' lives, the consequences for their careers or examination success could be severe in a research design that called for students to be deliberately put in an inappropriate teaching group to demonstrate they do not succeed.

It should also be possible to take for granted that the research accepts the Labovian dictum about difference rather than deficit that governs most other areas of language research (Labov 1969); people of various nationalities, cultures, sexes, religions, ethnic minorities, classes, ages, and so on have differences rather than deficits. Since L2 learning involves transforming a person with a single culture into a person with a multiple culture, the research should not imply that one group or one language is intrinsically superior to the other. Not that language teaching itself is immune to the charge of cultural imperialism by forcing particular languages or teaching methods on students in diverse circumstances (Phillipson 1992; Holliday 1994). More controversially, it has been argued that it is as biased to use native speakers as the goal for non-native speakers as it is to judge women by men, whites by blacks, or any group by the standards of another group to which by definition they can never belong (Cook 1999); that is to say, L2 users, too, are different from native speakers rather than deficient.

A perennial problem for any research is the extent to which generalisations can be made from small amounts of evidence: can the fall of an apple tell us about the forces within stars? At one level the risk of overstating the conclusions should be minimised by statistical tests, now usually applied to SLA research: it should be clear within the research report itself whether the results stand up or not. At another level, complete generality is impossible to achieve in SLA research given the diversity of the learners' L2s, L1s, situations, personalities, and so on.

SLA research includes both general overall theories such as the Universal Grammar (UG) model and specific results such as Arabic learners’ use of epenthetic vowels to pad out consonant clusters in English such as /bi l

stik/ for plastic (Broselow 1992). Teaching, too, ranges from overall goals for language teaching to specific teaching techniques used with a particular group of learners (Cook 1992). Linking research to teaching means employing both these scales of general to specific, on the one hand using general SLA theories for either general or specific aspects of language teaching, on the other using specific SLA results in specific teaching areas. While the match between SLA research and language teaching can never be exact, the user has to bear the relationship constantly in mind.

The most dangerous procedure is to use one specific piece of research as the basis for general teaching implications. Dulay & Burt (1973) looked at the acquisition of grammatical morphemes such as past tense -ed and articles the/a by 294 5-8-year-old Hispanic children learning English in California and New York City to settle the question "Should we teach children syntax?" Whatever its other merits, this research concerned grammatical morphemes rather than other aspects of syntax, Hispanics learning English rather than other groups acquiring English or other languages, the USA rather than other countries, and children aged 5 to 8 rather than adults. Though subsequent research amply showed that the results could be generalised to other learners and situations (Makino 1993), using the original research as the rationale for not teaching syntax was a case of a mouse giving birth to an elephant.

We have suggested that one habit in SLA research is to borrow general theories of human learning rather than to rely on SLA theories or research or indeed on linguistics. The classic example was indeed audiolingualism, which took a learning theory based on animal behaviour, applied it to LI acquisition with very little evidence, and generalised it to L2 learning with almost none (Rivers 1964). A contemporary example is the theory of connectionism, the SLA papers on which are mostly presentations of the theory itself for L2 researchers, sometimes with a computer simulation, for example Gasser (1990) and Sokolik & Smith (1992); they seldom have a specific basis in L2 research, apart from Broeder & Plunkett (1994). To use connectionism as a rationale, say, for reviving drill-like activities in the classroom depends on the strength of the theory itself and its general relevance, not on SLA research as such. The reliance on general learning theories also usually means accepting that language is no different from any other aspect of the human mind, rather than being a unique independent module, a conclusion that is anathema to many linguists and SLA researchers or, indeed, to many language teachers, who often insist that language has to be taught in different ways from other school subjects because it is different in nature.

The languages involved in the research need to match those involved in the teaching situation in relevant ways. At one extreme some research is so general that it concerns all languages everywhere, as in the UG theory. Even there it cannot be assumed that something that is true for one group of learners carries over to another. Take the example of the pro-drop parameter which separates languages in which subjects are not needed in the sentence, such as Italian Parla ('speaks'), from languages in which they are required, e.g. He speaks not *Speaks in English. L2 learners should carry this variation across from their first language to the second; this turned out indeed to be true for the English of French and Spanish learners (White 1986) but not to be true for the Spanish of English learners (Liceras 1989).

It is useful, then, to reiterate a few of the major differences between languages which have to be taken into account in L2 learning, taken from Cook (1997):

Tone languages use change of pitch to convey lexical meanings, e.g. Chinese and Yoruba; intonation languages such as English use it to express a variety of factors in attitude, grammar, and discourse function. SLA research on these includes Leather (1990) and Cruz-Ferreira (1986).

Two widely-used writing systems relate written symbols either to the sounds of language, whether alphabetic as in English or syllabic as in Korean or Japanese kana, or to units of meaning as in Chinese characters. There is a continuum of orthographic 'depth' from 'shallow' languages in which the script is closely connected to the sounds, e.g. Spanish, to 'deep' languages in which it is closely related to meaning, e.g. Chinese (Katz & Frost, 1992). This forms a growing area of SLA research, to be discussed below.

Languages vary in the extent to which they express grammatical relationships by changing the forms of words through inflections. Russian is a highly inflected language, with each word averaging 3.33 morphemes; Vietnamese has no grammatical inflections and averages only 1.06 morphemes per word; English has a handful of inflections for verbs and nouns and comes at the low end, averaging 1.68 morphemes per word. SLA research has concentrated on variational aspects of inflectional morphology (Young 1994) and on the presence or absence of inflections in early L2 speech (Klein & Perdue 1997) but has not yet tackled this area in depth.

Languages differ over whether the main word order in the sentence is SOV (Subject Object Verb) (45% of languages), SVO (42%), VSO (9%), VOS (3%), OVS (1%) or OSV (<1%) (Tomlin 1986). In SLA research, the basic word order of the sentence has chiefly been explored through the Competition Model (MacWhinney 1997), with some attempt at other types of word order universals (Cook 1988).

The L2 learner may speak an L1 with any combination of these features and be learning an L2 with any combination. SLA research may be straightforward when a specific limited area of language is involved, say the acquisition of English intonation. But it becomes extremely difficult when the features interact with each other; the very combination of Chinese being uninflected, having tones and using a character-based script could lead to some Chinese difficulties with English rather than any one feature taken separately.

Research based on specific L1s or L2s does not necessarily apply to all languages. English is still the L2 that dominates SLA research, with French and German featuring more in recent years, as introductions to SLA research such as Larsen-Freeman & Long (1991) or Ellis (1994) readily attest. Using English as the L2 may bias SLA research in various ways. Taking the above features, English is an intonation language with a fairly 'deep' sound-based script, fairly sparse inflectional morphology, and SVO order. These facts are not just separate but interact; for example the non-alphabetic aspects of English orthography relate to its morphology in such conventions as the past tense 'ed' being spelt as <-ed> despite its three usual pronunciations /t/, /d/ and /id/ as in watched, played and waited. Together they make English a unique combination of features, just as Chinese is unique, or any language is unique. Any pair of L1 and L2 involved in SLA research will involve at least as many variables.

The concentration on English may also highlight features that are comparatively rare in other languages. The pro-drop parameter mentioned above has been widely studied in UG based research from White (1986) to Davies (1996), despite frequent changes in the syntactic analysis. The 'fact' that languages fall into two groups seems potentially useful for language teaching, particularly when languages with opposite values are involved, such as Spanish and English. One use might be the shaping of the language input the learner encounters to provide the relevant clues to establish the parameter setting. For example, the clues that English requires subjects may be the use of dummy subjects like there and it and the fact that English verbs are uninflected for person with the exception of third person singular present tense verb forms. Yet languages such as English, French and German that require the subject form a small minority, possibly about 7%. Is it useful to base language teaching on something affecting only one in fourteen of the world's languages? The applicability of Universal Grammar or any other model to language teaching should not depend on unusual constructions limited to a few languages.

In LI acquisition research it is axiomatic that any normal human child taken at random could learn any possible human language; though this is a question of belief rather than research, no evidence has ever been produced to contradict it. L2 acquisition is another matter: L2 learners come in all shapes and sizes and reach varying levels of L2 competence. The nature of their progress depends on features of their individual personalities, ages and situations. Before using SLA research it is necessary to see how well the profiles of the learners involved in the research match the profiles of the projected students.

Francois Grosjean (1997) has enumerated seven factors involved in choosing bilinguals for an experiment: language history, language stability, languages known, competence in each skill, functions of the languages, language modes and amount of code-switching. It cannot be anticipated in advance which of these may turn out to be relevant.

There are at least as many varieties of language student. To go over some of the more obvious variations, SLA research has claimed that at least the following factors affect the outcome of L2 learning:

The motivation of L2 learners varies in many ways, for example the integrative motivation to take part in the target culture and the instrumental motivation to use it for one's own ends - career, examinations etc., elaborated in Gardner (1985). Even this one theory of motivation has revealed cultural variation from India (Lukmani 1972) to Hungary (Dornyei 1990).

All over the world it seems that, when they have a choice, language students are more likely to be women than men. This may be cultural stereotyping or may reflect a sex difference; for example, girls are better at L2 listening than boys (Assessment of Performance Unit 1986).

Regardless of the teachers' wishes, students choose the type of learning strategy individually, whether broad Good Language Learner strategies such as "Involve yourself in the language learning process" (Naiman et al. 1978) or 'metacognitive' strategies such as deciding in advance to concentrate on one aspect of a task (O'Malley & Chamot 1989).

Cognitive factors affect L2 learning even if each area that has been studied seems to yield contradictory results. Cognitive style, for example, reflects the difference between field-dependent people whose thinking requires a context and field-independent people whose thinking does not. Some SLA research has shown that this interacts with teaching method (Stansfield & Hansen 1983), other research that it is a cultural variable (Hansen 1984).

SLA research reveals even wider possible differences between students than even the language teaching profession had suspected. Teaching aimed at one type of student may not work with other types. Teaching designed for students with integrative motivation may antagonise those with instrumental motivation; 18-year-olds may rebel if taught in the same way as 12-year-olds; and so on for all the other factors. At minimum the results of SLA research into learner variation have reinforced the view that no single teaching solution suits all students in all circumstances. In general then the uses of SLA research must take into account the parallels between the learners studied in the research and the projected students.

SLA research is not a monolithic whole but covers many areas of language. Some research may be specific to one small area: the way you learn to pronounce /t/ may not be the same as the way you learn conditional clauses. Other research may embody a principle that applies everywhere, say the use of a meta-cognitive learning strategy such as monitoring one's speech (O'Malley & Chamot 1989). Before using particular SLA research, it is necessary to see how specific the language learning area is and how well it matches the proposed application.

L2 learners have to learn at least:
vocabulary: the meaning relationships between words
phonology: the sound systems of the language
discourse: the connections between, and within, utterances
syntax: the 'computational system' that sequences and modifies units of language to convey meanings.

listening: comprehending the spoken language
speaking: conveying meanings in spontaneous speech
writing: conveying meanings in writing
spelling: using the orthographic system of the language
reading: extracting meaning from written texts.

However, standard introductions to SLA research such as Ellis (1994) or Larsen-Freeman & Long (1992) are virtually dominated by syntax and barely recognise that areas such as spelling, vocabulary or phonology exist. This dominance of syntax is no longer representative of SLA research itself, which now covers most of the above areas, for instance spelling (James et al. 1993), vocabulary (Schreuder & Weltens 1993) and pronunciation (James 1996). These lacunas in the introductory treatments of SLA research not only limit the direct use of SLA research but also hinder language teaching professionals from seeing its usefulness, since recent views of language teaching have by and large not given syntax a central place (Grauberg 1997).

SLA research into syntax may be appropriate for the teaching of syntax, or at least for those aspects of syntax it covers, but it is doubtful whether it has implications outside this area. It may be crucial to the Universal Grammar theory to know how learners tackle principles such as subjacency that rule out sentences such as *What did John hear the news that the mayor would do? At best this research describes how learners tackle core issues of UG syntax. It provides no help with the many peripheral areas of grammar that are not part of that core or with the many areas of grammar that are not studied by UG theory. It says nothing about the rest of language, be it the performance aspects not covered by competence or the other areas that UG theory pays no attention to, such as discourse, or indeed the many areas of syntax UG theory does not tackle.

Overall, the answers to these questions mostly come down to matters of appropriateness. To be useful to language teaching, SLA research has to concern appropriate learners, appropriate languages, appropriate areas of language learning and appropriate situations. The research has to match the teaching use. A piece of SLA research must either relate specifically to the area of teaching in question or be so general that it relates to all of language teaching. The danger has often been that the devotees of a successful SLA theory have applied it to all aspects of teaching despite its partial coverage. A proper use of SLA research has to include many aspects if it is to link to more than limited aspects of language teaching. And it needs to be filtered through the research that specifically deals with the features of the classroom itself, such as Lightbown & Spada (1993).

Let us then take three pieces of SLA research to explore briefly some of the implications of this argument. Each example first presents the particular SLA research, then applies the requirements discussed above and finally suggests some uses. The selection represents an arbitrary choice out of the hundreds of research areas that could have been chosen. The intention is to take one general topic — age; one specific piece of research — the acquisition of English pronunciation by Austrian children; and one area virtually untapped by language teaching — the writing system.

A perennial question for linguists, SLA researchers and the person in the street is whether children are better at acquiring second languages than adults. The proliferating research into the question of age is surveyed in Singleton (1989). The question affects language teaching in that educational systems have to decide on the optimum age for starting an L2 in the classroom.

A common assumption is that children are better than adults. One set of results challenges this view. Some of this work concerns naturalistic settings: adults and 11-15-year-olds learnt Dutch better than children aged 3-10 for most of the first year in the Netherlands (Snow & Hoefnagel-Hohle 1978); by the end of the year the most successful learners were the 8-10-year-olds and the least successful the 3-5 year-olds. Sometimes it covers classroom acquisition: Ekstrand (1978) studied 2400 children acquiring Swedish as an L2 and found that the older the child, the better.

A second set of results, however, shows that children are better. Johnson and Newport (1989) tested the grammaticality judgements of people who had immigrated to the USA between the ages of 3 and 9 and found a steady decline in this ability with age of arrival in the USA. Patkowski (1980) came to similar conclusions by looking at the Foreign Service Institute ratings of written language by 67 immigrants to the USA. The contradictions between these two sets of results have usually been resolved by saying that adults have an initial advantage, children an advantage over the longer term (Krashen, Scarcella & Long 1982; Singleton 1989).

Much age research has concerned foreign accent, which is a slightly different issue, being related not only to how people produce sounds physically but also to how speech shows membership in social groups. Despite the prevalent belief that only children acquire an L2 without a foreign accent, recent studies have revealed a group of Dutch speakers of English who are indistinguishable from native speakers (Bongaerts, Planken & Schils 1995). Indeed Neufeld (1978) managed to produce adult L2 learners of Japanese whose pronunciation could not be distinguished from native speakers after eighteen hours of language teaching.

Applying the six requirements fully to such a large area of research would be impossible; the following seems generally true of much of the research. The first three requirements are easily met because of the sheer range of the research. The research is valid in that it includes a vast amount of data with large numbers of learners; it is ethical in that none of it appears to affect the learners adversely; it is general in that it takes an overall issue of L2 learning; and it is specific in that it deals with particular issues of syntax and pronunciation.

The remaining requirements pose more of a problem. The breadth of languages covered is one issue. The L2 is almost invariably English and the majority of learners speak European languages, as in Johnson & Newport (1989), Bongaerts et al. (1995) and Patkowski (1980). There are therefore dangers in generalising to other L1s and other L2s. The appropriateness of the learners is also suspect. The majority are adults or children arriving in the USA as in Johnson & Newport (1989) or Patkowski (1980), sometimes the Netherlands (Snow & Hoefnagel-Hohle 1978). Immigrants, however, form a particular group in social or psychological terms and their results may not generalise to non-immigrant groups. The bulk of the research, furthermore, concerns learning situations where the second language is in active use within the community, whether English in the USA, Dutch in the Netherlands or Swedish in Sweden, unlike situations where a foreign language is taught in a classroom but not readily available in the world outside, say French in the USA, Japanese in the Netherlands or English in Sweden. Finally, the coverage of learning areas is rather limited. Usually the concern is syntax and accent; vocabulary studies in fact have found no age difference (Singleton 1995); the works of Conrad, Nabokov or Beckett provide strong evidence that older L2 learners can master the writing system. Claims have to be restricted to a narrow range of language activities.

Age research could be used in language teaching in several ways. One is the original debate about the age when a second language should be started. Ultimately the decision about the optimal age to start L2 teaching in schools is a matter for the whole educational system in a country. SLA research can at best provide hints to inform this decision. If the educational system aims at a quick basic ability to use a second language, this is best acquired as an adult; if it aims at a high level of performance, this is better acquired as a child (Krashen, Scarcella & Long 1982). However, this does not take into account any of the benefits of cultural understanding, language awareness, etc. that are unrelated to the ability to speak and talk the L2 and that may be particularly valued by the students or by the educational system. The UK national guidelines for language teaching (HMSO 1995) include the student's self-development, the training of cognitive processes and insight into the LI, as well as the usual communicative goals.

Another approach is through the attitude towards foreign accent. The dire prediction that older people cannot lose their foreign accent seems unwarranted. Teachers may be more optimistic about teaching pronunciation and convince themselves and their students that a native-like accent is indeed attainable for some students whatever their age. However, if the goal is L2 use rather than native speaker use, it does not necessarily follow that a foreign accent is undesirable and that L2 users have no right to express their identity in the second language; as a French wine-maker said, "My English is not good but my French accent is perfect" (Cook 1999).

Let us now take a specific example of detailed research from phonology. In recent years, several alternative theories have been mooted for the acquisition of phonology in an L2. Major's ontogeny model (1990, 1994), for example, claims that phonological transfer decreases over time while developmental factors first increase, then decrease.

A more traditional type of study was carried out by Wieden & Nemser (1991), who looked at the acquisition of English phonology by 384 Austrian school children in grades 3-11 (two classes in four regions at each age) to establish the stages of development and the characteristic problems of Austrian learners through a battery of tests of perception, imitation and elicitation. This yielded a range of results. One interesting feature was a contrast between phonemes such as the diphthong /

/ in boat that improved over time and other phonemes such as the schwa in folder and the dark /l/ in ball which did not improve after eight years of school. That is to say, some phonemes develop over time, others do not. Three distinct stages of development were found: a pre-systemic stage in which the children knew the sounds in individual words but not in all circumstances; a transfer stage in which they systematically used the L1 categories in the L2, for example using the single Austrian sound /b/ for both English /p/ and /b/; and an approximative stage in which they restructured the L1 sounds into a new system, i.e. a native-like RP system.

In terms of the requirements, the research seems valid in that it uses appropriate methodology with large numbers of child learners; it is ethical in that no harm came to the children through the testing. On a scale of generality, it is specific to the acquisition of English phonemes by a certain group of learners in one classroom situation. A full match between languages and learning situation occurs only if the prospective students are young Austrian learners of English. It may or may not be possible to generalise to other students. Since the research looks at the phonetic detail of English and Austrian German rather than phonological theory, it may be hard to apply the categories to other languages. The match between L2 learners and students again comes down to assessing whether the characteristics of the learners are commensurable with the students in age, personality etc. The results might be useful for teaching teenage children English in foreign language situations but could only be tentatively extended to other students in other places. The coverage of language learning is confined to segmental or non-segmental phonology. Even if the stages turn out to be true of other learners, this does not mean that they are necessarily true of other areas of language such as vocabulary acquisition.

Given these caveats, the research could be used in teaching in several ways. One choice is for the person designing the teaching sequence. Sounds like the diphthong /

/ that can improve could be concentrated on as this may have the maximum effect. Or difficult sounds like schwa /

/ could be taught since the easy ones will be learnt anyway. The reason for choosing one or the other approach is unlikely to be found in the original research but only emerges when the two alternatives are followed in practice; nevertheless, it is the research that has provided the impetus. A second choice concerns whether or not to draw the students' attention to phonology. An interventionist teacher might tell the students about the sounds that have been established to be a problem. A non-interventionist might decide not to squander precious teaching time on things that are unalterable and so minimise any attention to 'difficult' sounds. The same piece of research could also be used to affect the design of pronunciation materials, the details of the syllabus, and the specification of the final examination, provided the requirements were used appropriately.

The third example comes from the area of writing systems. Much processing of written text is a balance between sound-based processes for converting letters into sounds and visual-based processes for handling whole symbols, both between languages and within the same user. The main SLA research thrust has been on the transfer from a character-based system to a sound-based one, and vice versa. Chinese L1 students have difficulty in processing non-words in English, showing their phonological processing is under-developed (Holm & Dodd 1996). Speakers with meaning-based LI writing systems in general are better at visual reading tasks in English than those with sound-based L1s (Brown & Haynes 1985).

A recent piece of research in this tradition is Chikamatsu (1996), which looked at the transfer from L1 to L2 orthographic systems by contrasting L2 learners of Japanese syllabic signs (kana) from two different L1 systems – Chinese characters and the English alphabet. Results were indeed that Chinese people relied more on visual information, English more on phonological. L2 learners are carrying over the characteristics of their LI writing system to the L2.

In terms of the six requirements, this research seems valid methodologically, particularly in contrasting two groups of learners rather than relying on one, ethical in not interfering, and reasonably general in its scope in that it concerns overall characteristics of writing systems rather than language-specific details. In terms of coverage, English, Japanese and Chinese represent three language types, even if within each type there is great variation - English is a relatively 'deep' orthography compared to the 'shallow' orthography of Italian, though both are alphabetic sound-based writing systems (Katz & Frost 1992). The learners are appropriate if they are taken to represent older academic types of learner. The area of learning covered is in one sense narrow as it concerns a single aspect of orthography; in another sense it affects all the learners' processing of written language even at an advanced level.

The main point for teaching is simply to draw attention to the fact that some L2 students have to shift from one overall writing system to another, whether English people learning Japanese, Chinese learning Spanish, or whatever. The presentation of early written material in coursebooks, for instance, should be influenced by whether or not the students are transferring across this great divide of languages or are shifting in the amount of 'depth'; the testing of reading comprehension should pay attention to the different degrees of difficulty in writing for students from different LI scripts. Specific teaching exercises may be utilised to direct the students' attention towards the nature of the new script they are acquiring. Yet virtually none of this seems to have been mentioned in language teaching.

As we see from these three briefcase studies, SLA research can provide teaching with provocative ideas - is an accent avoidable in adults; should pronunciation teaching concentrate on 'improvable' sounds? If nothing more, these ideas can stimulate the teaching profession into re-examining its current practice. As always, the real-world success of these in the classroom doubtless depends on other variables in the teaching/learning situation that have little to do with SLA research directly. This is, then, where such insights need to be combined with the classroom-based approaches explored in, say, Lightbown & Spada (1993) or Harley et al. (1990).

The main argument here is that research into language learning and teaching can provide useful input for language teaching. Obvious as this may seem, it nevertheless constitutes a rejection of several popular approaches to the applied linguistics of language teaching. One consists of taking a theory from some other area and applying it willy-nilly to language teaching, for example the behaviourist theories that underpinned the audiolingual approach or the theory of connectionism. Without a basis either in the acquisition of second languages or in classroom research, such ideas have no more to do with language teaching than, say, theories of knitting. A second approach is the trend towards presenting teachers, and indeed students, with data about the language of native speakers, whether vocabulary, discourse or syntax, often derived from large corpora. Such descriptions are not relevant to language teaching unless they can show firstly how these performance data relate to the knowledge in the speakers' minds, secondly how L2 learners acquire such knowledge, thirdly how non-native speakers relate to native speakers. These two approaches have no more or less chance of succeeding intrinsically than, say, teaching students the difference between plain and purl stitches. Their success is accidental; the links between linguistic or psychological theories or descriptions and language teaching can only come through the mediation of SLA research. Willis (1998: 45), talking of concordances, remarks: "Once we begin to view the process of language description in this way it is a short step to apply the process pedagogically." This "short step" is, however, a big leap for teaching: SLA research provides the missing link between the description and the classroom application.

The three brief examples here have only scratched the surface. SLA research has something to say about the acquisition of vocabulary, phonology, syntax, discourse, classroom interaction, and much more: all aspects of language and language acquisition are covered somewhere in SLA research. Each of these examples suggested ideas for language teaching, whether through the attitudes towards foreign accent, the choice of what sounds to teach, or the importance of the writing system. Other insights are crying out to be used — what syntax should be taught to beginners; what level can we realistically expect L2 learners to achieve; do pragmatic functions of language and communication strategies transfer from the L1; should the L1 really be minimised in the classroom? Others can be gleaned from surveys of SLA research such as Cook (to appear). The very lack of a unified theory of SLA may be a strength for language teaching by suggesting a diversity of possibilities rather than a single party line and thus leaving teaching more free to pick and choose. By mining this rich vein, teaching may be able to base its future developments in part on how people learn second languages rather than on the description of native speakers or the changing fashions in the classroom.

Albert, M. & Obler, L. (1978) The bilingual brain. New York: Academic Press. Assessment of Performance Unit (1986) Foreign language performance in schools: report on 1984 survey of French. London: Dept. of Education and Science.

Bongaerts, T., Planken, B. & Schils, E. (1995) Can late starters attain a native accent in a foreign language? A test for the Critical Period Hypothesis. In D. Singleton and Z. Lengyel (eds), The age factor in second language acquisition. Clevedon; Multilingual Matters 30-50

Broeder, P. & Plunkett, P. (1994) Connectionism and second language acquisition. In N. Ellis (ed.), Implicit and explicit learning of languages. London: Academic Press. 421-53.

Brown, T. & Haynes, M. (1985) Literacy background and reading development in a second language. In T.H. Carr(ed.), The development of reading skills. San Francisco, CA: Jossey-Bass. 19-35.

Broselow, E. (1992) Non-obvious transfer: on predicting epenthesis. In S. Gass & L. Selinker (ed.), Language transfer in language learning. Amsterdam: Benjamins. 71-86.

Cazden, C, Cancino, E., Rosansky, E. & Schumann, J. (1975) Second language acquisition in children, adolescents and adults: final report. Washington D.C.: National Institute of Education.

Chikamatsu, N. (1996) The effects of L1 orthography on L2 word recognition. Studies in Second Language Acquisition 18: 403-32.

Cook, V.J. (1988) Language learners' extrapolation of word order in phrases of micro-artificial languages. Language Learning 38.4: 497-529.

— (1992) Relating second language acquisition research to language teaching. Die Neueren Sprachen 91: 115-30.

— (1996) Second language learning and language teaching [2nd edition]. London: Arnold..

— (1999) Going beyond the native speaker in language teaching. TESOL Quarterly, 33, 2,185-209

— (to appear) Linguistics and second language acquisition: one person with two languages. In M. Aronoff & J. Rees-Miller (eds), The Blackwell handbook of linguistics. Oxford: Blackwell

Cruz-Ferreira, M. (1986) Non-native interpretive strategies for intonational meaning: an experimental study. In A. James & J. Leather, Sound patterns in second language acquisition. Dordrecht: Foris.

Davies, W.D. (1996) Morphological uniformity and the null subject parameter in adult SLA. Studies in Second Language Acquisition 18: 475-93.

Dornyei, Z. (1990) Conceptualising motivation in foreign-language learning. Language Learning 40.1: 45-78.

Dulay, H. & Burt, M. (1973) Should we teach children syntax? Language Learning 3: 245-57.

Ekstrand, L. (1978) Age and length of residence as variables related to the adjustment of migrant children with special reference to second language learning. In G. Nickel, Proceedings of the fourth international congress of applied linguistics, Stuttgart: Hoch-schulverlag. 179-97.

Ellis, R. (1994) The study of second language acquisition. Oxford University Press

Gardner, R. (1985), Social psychology and second language learning: the role of attitude and motivation, London: Edward Arnold

Gass, S. & Lakshmanan, U. (1991) Accounting for interlanguage subject pronouns. Second Language Research 7:181-203.

Gasser, M. (1990) Connectionism and universals of second language acquisition. Studies in Second Language Acquisition 12: 179-99.

Grauberg, W. (1997) The elements of foreign language teaching. Clevedon: Multilingual Matters.

Grosjean, F. (1997) Processing mixed languages: issues, findings and models. In A. de Groot & J. Kroll, Tutorials in bilingualism. New Jersey: Erlbaum. 225-54.

Hansen, L. (1984) Field dependence-independence and language testing: evidence from six Pacific Island cultures. TESOL Quarterly 18.2: 311-24.

Harley, B., Allen, P., Cummins, J. & Swain, M. (eds.) (1990) The development of second language proficiency. Cambridge University Press.

Her Majesty's Stationery Office (HMSO) (1995) The National Curriculum. London: HMSO.

Hilles, S. (1991) Access to Universal Grammar in second language acquisition. In L. Eubank, Point counterpoint: universal grammar in the second language. Amsterdam: Benjamins. 305-38.

Holliday, A. (1994) Appropriate methodology and social context. Cambridge University Press.

Holm, A. & Dodd, A. (1996) The effect of first written language on the acquisition of English literacy. Cognition 59: 119-47.

James, A. (1996) Second language phonology. In P. Jordens & J. Lalleman, Investigating second language acquisition. Berlin: Mouton de Gruyter.

James, C., Scholfield, P., Garrett, P. & Griffiths, Y. (1993) Welsh bilinguals' spelling: an error analysis. Journal of Multilingual and Multicultural Development. 14.4: 287-306.

Johnson, J. & Newport E. (1989) Critical period effects in second language learning: the influence of maturational state on the acquisition of ESL. Cognitive Psychology 21: 60-99.

Katz, L. & Frost, R. (1992) Reading in different orthographies: the orthographic depth hypothesis. In R. Frost & L. Katz, Orthography, phonology, morphology and meaning. Amsterdam: Elsevier. 67-84.

Klein, W. & Perdue, C. (1997) The basic variety (or: couldn't natural languages be much simpler?). Second Language Research 13.4: 301-47.

Krashen, S., Scarcella, R. & Long, M. (1982) Child-adult differences in second language acquisition. Rowley, Mass.: Newbury House.

Labov, W. (1969) The logic of non-standard English. In J. Alatis, Georgetown monographs on language and linguistics 22: 1-44.

Lantolf, J.P. (ed.) (1994) Socio-cultural theory and second language learning. Special Issue, Modern Language Journal 78.4.

Larsen-Freeman, D. & Long, M. (1991). An introduction to second language acquisition research. London & New York: Longman.

Liceras, J.M. (1989) On some properties of the 'pro-drop' parameter: looking for missing subjects in non-native Spanish. In S.M. Gass & J. Schachter (eds). Linguistic perspectives on second language acquisition. Cambridge University Press. 109-33.

Lightbown, P. & Spada, N. (1993) How languages are learned. Oxford University Press.

Lukmani, Y. (1972) Motivation to learn and language proficiency. Language Learning 22:261-73.

MacWhinney, B. (1997) Second language acquisition and the Competition Model. In A. de Groot & J.F. Kroll, Tutorials in bilingualism. Erlbaum. 113-144.

Major, R. (1990) L2 acquisition, LI loss, and the critical period hypothesis. In J. Leather & A. James, New sounds 90: proceedings of the Amsterdam symposium on the acquisition of second-language speech. University of Amsterdam. 14-25.

— (1994) Current trends in interlanguage phonology. In M. Yavas, First and second language phonology. San Diego: Singular. 181-204.

Makino, T. (1993) Perspectives on second language acquisition. Tokyo: Yumi Press.

Markee, N. (1993) Second language acquisition research: a resource for changing teachers' professional cultures? Modern Language Journal 97: 80-93.

Naiman, N., Frohlich, M., Stern, H. & Todesco, A. (1978) The good language learner. Toronto: The Ontario Institute for Studies in Education [Research in Education Series No. 7]. Reprinted in 1995, Clevedon: Multilingual Matters.

Neufeld, G. (1978) On the acquisition of prosodic and articulatory features in adult language learning. Canadian Modern Language Review 34.2: 163-74.

O'Malley, J. & Chamot, A. (1989) Learning strategies in second language acquisition. Cambridge University Press.

Paradis, M. (1992) Neurosciences et apprentissages des langues. In D. Girard, Enseignement des langues dans le monde d'aujourd'hui III. Paris: Hachette. 20-31.

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

Perdue, C. (ed.) (1993) Adult language acquisition: cross-linguistic perspectives [2 volumes]. Cambridge University Press.

Rivers, W.M. (1964) The psychologist and the foreign language teacher. Chicago University Press

Schumann, J., Holroyd, J., Campbell, R. & Ward, F. (1978) Improvement of foreign language pronunciation under hypnosis: a preliminary study. Language Learning 28: 143-8.

Schwartz, B. & Sprouse, B. (1996) L2 cognitive state and the Full Transfer/Full Access model. Second Language Research, 12, 1, 40-72

Singleton, D. (1989) Language acquisition: the age factor. Clevedon: Multilingual Matters.

— (1995) A critical look at the Critical Period Hypothesis in second language acquisition research. In D. Singleton & Z. Lengyel, The age factor in second language acquisition: a critical look at the critical period hypothesis. Clevedon: Multilingual Matters. 1-29.

Snow, C. & Hoefnagel-Hohle, M. (1978) The critical age for language acquisition: evidence from second language learning. Child Development 49: 1114-28.

Sokolik, M.E. & Smith, M.E. (1992) Assignment of gender to French nouns in primary and secondary language: a connectionist model. Second Language Research 8.1: 39-58.

Stansfield, C. & Hansen, J. (1983) Field dependence-independence as a variable in second language cloze test performance. TESOL Quarterly 17: 29-38.

Stern, H. (1983) Fundamental concepts of language teaching. Oxford University Press.

Tomlin, R.S. (1986) Basic word order: functional principles. London: Croom Helm.

Wenk, B. (1982) Articulatory setting and the acquisition of second language phonology. Revue de Phonetique Appliquee 65: 51-65.

White, L. (1986) Implications of parametric variation for adult second language acquisition: an investigation of the pro-drop parameter. In V.J. Cook, Experimental approaches to second language acquisition. Oxford: Pergamon. 55-72.

Wieden, W. & Nemser, W. (1991) The pronunciation of English in Austria. Tubingen: Gunter Narr.

Willis, J. (1998) Concordances in the classroom without a computer: assembling and exploiting concordances of common words. In B. Tomlinson, Materials development in language teaching. Cambridge University Press

Young, R. (1994) Functional constraints on variation in interlanguage morphology. Applied Linguistics 14: 76-97.