A Processing View of Teaching English Grammar
Foreign language teaching is dominated by three types of approaches. Traditional approaches are form-oriented. They believe that explicit grammar instruction is useful and that exercises must be designed that help prevent errors. In contrast, approaches that believe in an inborn Universal Grammar represent a radical alternative. They maintain that all explicit grammar teaching is useless and insist that all learners need is comprehensible input. A middle position is taken by constructivist and task-based approaches.
In this essay Processability Theory is discussed, which is a constructivist approach. Its central message is that grammar instruction can only be useful if teachers teach what learners can learn. Errors are to a certain degree inevitable because learners, on their way to language competence, all pass through the same sequence of stages of grammar acquisition and make errors typical of the individual stage. Based on psycholinguistic and empirical studies of second language acquisition, the theory explains why learners make what type of errors at each stage. That allows a prognosis of what parts of grammar are teachable for individual learners at a given stage of their language development.
0. Outline of the status quo
Everyone knows that learning to speak a second language (a) requires that we learn enough “words” (vocabulary) to give lexical expression to what we want to say; (b) that we learn to put the words in a morphological form and syntactic order that express what we mean to say and that are “correct” from a native-speaker perspective; (c) that we learn to pronounce the words and phrases more or less correctly. Leaving aside that there are other, pragmatic and cultural aspects to second language acquisition (SLA), the above three components are central to it, and most people consider grammar learning to be the most difficult part. Sophisticated theories of language learning, too, centre on grammar acquisition. Correspondingly, foreign language teaching (FLT) has traditionally based its methods on syllabuses that prescribe which forms of grammar must be taught in what sequence. The definition of which forms are easy or difficult to learn was based on formal linguistic descriptions of grammar, and on teaching experience. That is why traditional approaches to FLT are all variants of the same type: form-oriented FLT.
In the classroom, form-oriented FLT takes the form of the PPP-Model (Presentation – Practice – Production): First, the grammar forms are correctly presented; teachers next provide many opportunities for their practice in “guided exercises” (pattern drills) and aim to stamp out learner errors; finally learners are expected to use the practised forms creatively and correctly in spontaneous and meaningful communicative interactions. It is well known, though, that the hoped-for learning transfer from the practice to the production phase (communication) mostly does not take place. This is nicely captured by the title of D. Allwright’s (1984) essay: “Why don’t learners learn what teachers teach?” Many researchers have therefore suggested that a new approach to FLT be developed. But, depending on their views of how grammar is learnt, suggestions for reforms differ. They can be divided into three groups.
1. Three types of approach to FLT
One group of researchers base their suggestions for reforms on pragmalinguistic concepts of language, and on the experience of teachers that learners need plenty of opportunity for using the L2. As a result, communicative and meaning-oriented approaches to FLT were developed. They stress that learners, in class, need to want to say what they say, or else they will not learn to use the target language creatively. At the same time however, communicative approaches, in this country, basically stick to a form-based syllabus that prescribes which grammar forms must be taught when, though underlining that teachers should practice error tolerance and rate fluency higher than accuracy. That however (a) leaves it to the individual teacher to decide how to teach grammar and when to tolerate errors, and (b) brings teachers into a situation that requires them to manoeuvre the learners into a situation which allows them to say what they mean to say while ensuring that the language forms they need for that “happen to be identical” with the ones prescribed by the syllabus.
Quite obviously that is difficult to achieve. It requires of learners a “willing suspension of disbelief”, that is they must imagine themselves to be in a situation in which the use of the prescribed lexical and grammatical items makes sense, and it requires of teachers to find tasks that are both motivating and manageable with the limited linguistic skills that a particular group of learners have. Such tasks, though, provide only a poor substitute for the existential need to learn the L2 in an L2 environment with its many contextualized and communicatively meaningful situations; even more importantly, task-based learning does not change the fact that in an FLT context the learners have much less time and opportunity for autonomously figuring out (by trial and error, hypothesis formation and hypothesis testing) how the L2 grammar functions. That puts limits to task-based approaches.
A second group of researchers bluntly denies that grammar can be taught. They argue that learners do not need formal instruction nor an explicit knowledge of grammar rules, since they acquire grammar implicitly if they receive enough comprehensible language input (Krashen, 1985). These researchers point, for a support of their hypothesis, to a language acquisition theory which was designed to explain first language acquisition (Chomsky ,1980). It argues that children neither get nor understand grammatical instruction; they nevertheless manage to acquire their first language, even though the language input they get from their parents may be grammatically incomplete or corrupt. That is called “the logical problem of language acquisition”. It is speculated that children are able to solve it because they are born with a so-called language acquisition device (later called Universal Grammar [UG]). That theory stimulated SLA researchers to speculate on whether or not inborn UG-knowledge is available for SLA too or “used-up”, so to speak, with first language acquisition, or available for SLA up to a critical age only. Other researchers pointed out that “comprehensible input” is not enough for language acquisition to take place because learners also need to produce comprehensible output (Swain, 1985).
One of the attractions of UG-based theories of SLA is that they offer a seemingly elegant way of explaining “the logical problem of language acquisition”, and for the fact that L2 grammars are acquired in universally similar developmental stages (Pienemann, 1984; Long, 1988). If we assume a UG that unfolds following a “biologically built-in syllabus”, that would nicely explain these stages. The logical consequence would be, though, to give up syllabus construction, grammar teaching and error correction; we should instead focus on providing learners with plenty of comprehensible language input, leaving it to nature to see to the rest.
There are some serious problems with UG-based acquisition theories, however. They start with the fact that neurobiologists and neuroanatomists say that from an evolutionary and neurophysiological perspective there is no evidence for the existence of a genetically inherited UG (Deacon, 1997); similarly many psychologists object to UG-theories as empirically unwarranted (Hörmann, 1976, p. 320). Apart from that, the evidence for parameter setting processes in SLA is not conclusive, nor has it been shown that better results are achieved in SLA if we stop to give learners support in grammar acquisition by explaining the rules of grammar. Furthermore, the fact that there are universal similarities in SLA processes is not of itself proof of the existence of an inborn UG; they call, though, for a non-nativist explanation. That is provided by the constructivist view of learning (Wolff, 1994). It sees language acquisition as the result of general learning strategies that apply to all domains of knowledge and skills, from learning to swim or ski, for example, to playing chess and solving mathematical problems, and to SLA (Bates & MacWhinney, 1989).
The challenge is, however, to come to a theoretically consistent explanation of such constructive processes in SLA, because only if we do, can we give more precise advice than recommending “awareness raising” as the best strategy to help learners with grammar acquisition. We therefore need a psycholinguistic theory that can explain (a) why there are universally similar developmental stages in SLA, (b) why a late beginning SLA does not lead to native speaker competence, and (c) why explicit rule instruction cannot change the developmental sequence in SLA. The theory should (d) be compatible with what we know from the cognitive sciences about the distributed storage of knowledge and the parallel spreading of activation in the neural network.
The progress made in that direction in recent time is considerable, although we are still some way from a generally accepted theory of how language is mentally represented. I want to show, however, that there is a theory that can explain the universally similar developmental stages in SLA without assuming an inborn UG. The theory is called Processability Theory (Pienemann, 1998); it derives from the well-known teachability hypothesis (Pienemann, 1984), which is famous for the dictum “teach what is teachable”. Note, though, that “teachable” must not be confused with “prescribed by the formal syllabus”. After presenting Processability Theory, I will discuss its consequences for FLT, focusing in particular on how it relates to task-based and holistic learning (Timm, 1995a).
2. Constructivism and Processability Theory: A psycholinguistic account of SLA
Processability Theory (PT) has an empirical basis in the analysis of several sets of data from studies of SLA in different languages and by different researchers (Pienemann, 1998, pp. 117-230). They are interpreted against the background of (a) a concept of human information-processing which shows that Working Memory (Baddeley, 1986) is the bottle-neck through which all information must go on its way in and out of Long Term Memory, both in learning and doing, and (b) a social-functional model of grammar, Lexical Functional Grammar (Kaplan & Bresnan, 1982; Kempen & Hoenkamp, 1987), and a corresponding model of speech production (Levelt, 1989). The speed with which goal-directed activities like speaking can take place depends on the availability of knowledge patterns, including skills that are fast and incrementally activated (Levelt, 1989, p. 24) to serve the communicative goal. Their processability is a function of the strength of the associative (neural) links from conceptual knowledge units to language-specific lexical and grammatical and articulatory processing units that serve to express the concepts. Frequent use of a processing unit (knowledge pattern) enhances the chances of its fast (automatized) processability.
PT shows that in speech production the mind is faced with a number of key grammatical encoding tasks. They are tackled in language acquisition in a developmental sequence that reflects a hierarchy of processing procedures, which is universal (Pienemann, 1998, p. 87). If a learner fails to successfully take a “low down” step in that hierarchy of processing procedures, it has a negative effect on all “higher up” processing procedures. For FLT that would be a point for stepping in with supporting (preventive) measures. That does not mean, though, that there is no individual variation. It is restricted, however, to how fast and well the developmental steps are taken by a specific learner, which is a function of his/her learning motivation and social identity. Before going into details, however, it is useful to first discuss the “mechanisms” that underlie language learning processes.
2.1 From experience and fuzzy concepts to automatized processes
Learning processes transform the flood of incoming sensory data into knowledge patterns, using for their ordering the fact that human beings experience things and events in meaningful contexts which help to structure the data. The contexts prompt learners to remember “things” in terms of their qualities, such as “eatable” or “not eatable”, “solid” or “soft”, “mobile” or “immobile”, “good-for-me” or “bad-for-me”, “usually having consequence X”, and so on. That indicates how such concepts relate to predicate argument structures and an implicit, meaning-based understanding of linguistic categories like “noun”, “verb”, “adjective”, which are essential for grammar acquisition. The point here is, though, that they are not inborn linguistic knowledge but created in learning processes. That is not in conflict with the fact that in the course of the self-organization of knowledge, linguistic units develop a preference for activating other linguistic forms (modular organization). L1 forms generally activate other L1 forms; and L2 forms need to develop equally strong “associative cohesion” before they can be used for the expression of meaning.
Memories of the conceptual qualities of “things” are first stored in Episodic Memory. That is a sort of biographical memory that stores information in a distributed but connected form across many parts of the brain. The different parts of a specific concept mutually activate each other if one of them receives activation from outside. Thus, in an animalistic fashion, instinctive reactions to the appearance of certain sensory stimuli can be triggered.
On the basis of its episodic memories the human mind can create, in what is called Semantic Memory, cognitive concepts of things and events. They are the result of processes that abstract away from the qualities of the individual “thing”, constructing instead a concept of the type of thing/event they represent. They are “fuzzy concepts” because they comprise, besides their core or defining semantic features, other, possible or likely features, which make them fuzzy on the edge (Timm, 1995b; Multhaup, 1999). Concepts are, over time, subject to a restructuring and a re-evaluation of the number and kind of “things” they stand for (mean).
It is the fuzziness of concepts that makes them powerful tools in learning processes. They allow people to adapt to new situations by mental processes which Jean Piaget (1980) calls assimilation and accommodation. From a processing perspective, it means that mental processes are probabilistic calculations, and not based on the “application” of explicit rules and invariable symbols as used by standard computers. From a neurological perspective that means that connectionist models come closest to what goes on in neural networks (Bechtel & Abrahamsen, 1991). With regard to language learning, including SLA, a model of how they work in acquisition processes has been put forward in form of the Competition Model by MacWhinney (1989).
The probabilistic nature of mental processing procedures explains how people come from a conceptual (pre-linguistic) understanding of what “things” are like and can do, to a semantic understanding of their roles in predicate argument structures. In fact children use a whole range of processing strategies to get their grammar acquisition processes started. They range from semantic bootstrapping (Pinker, 1989), that is interfacing semantic structures with syntax, and syntactic bootstrapping, that is guessing the meaning of new language forms from their relationship to known lexical and other items in a syntactic frame, to homing in on the sound quality of utterances (rhymes, stress, intonation). Linked with this is the fact that semantic learning is exemplar-based: Children first learn how a verb like give functions before other verbs are (tentatively) fitted into the mould created by it, thus inheriting a preformed syntactic “behaviour pattern”. Because of the probabilistic nature of neural processing patterns, they can be modified without a break-down of the old system.
In late beginning SLA, learners profit from the knowledge of how language functions, which they acquired in L1 learning. But they pay a price for that, which is interference. The semantic mappings from non-linguistic concepts to language forms tend to transfer from L1 forms to seemingly equivalent L2 forms. In a way, one of the most difficult things in SLA therefore is the un-learning of such automatized mappings. Pronunciation is a typical example; new speech production procedures, which need to establish themselves against the competition from deeply ingrained L1 speech habits, are initially at a disadvantage, and it takes plenty of time and practice for them to gain ground. Late beginning L2 learners rarely get rid of their L1 accent. A good description of the nature of the mental processes involved in such language learning processes is provided by the neurolinguist Paradis (1994); it, too, gives a good explanation of the roles of implicit and explicit learning in language learning processes:
Learners are conscious of the sounds they attempt to produce and they practise until they are satisfied that their performance sufficiently approximates the intended sound. The articulatory (and hence motor) gestures, when often repeated, end up leaving proprioceptive kinaesthetic traces, i.e. automatically available articulatory movement programmes. But even in this borderline case, although the sounds are consciously practised, it is the proprioceptive data (i.e., the cortical record of the position of each organ involved, from the position of the tongue relative to the palate, the teeth and the lips, to the movements of the vocal cords) that are stored in implicit memory. The speaker is not really aware of the simultaneous or sequential position of all the phonatory organs. The speaker is only conscious of the result obtained, not of how it is obtained. The organs’ position, of which the speaker is not conscious, is what is stored implicitly and becomes automatic through practice, just as with any other motor skill. What the speaker is conscious of, is the sound that is produced and practised. (Paradis, 1994, p. 401)
This shows that it is futile to debate whether explicit learning always precedes implicit learning or vice versa.
Returning to the competition of newly to be established L2 items with automatized L1 procedures: It is well-known that lexical learning is also affected by interference. With regard to grammar acquisition the picture is complicated by the fact that in core areas of grammar there are both universal developmental sequences and language-specific interference effects. That is of consequence for error analysis in SLA, and for a pedagogical definition of what is teachable at a given stage of a learner’s L2 development.
In this context a note on the relation of imitation to cognition seems appropriate, because some language acquisition theories decry imitation. Nobody denies that language learning can take place only if the learners come into “contact” with the target language, that is, have a model on which they can base their acquisition processes. But only if they try to imitate the model will they acquire the target language. It is important therefore to see that imitation and cognition are not two mutually exclusive but complementary learning strategies; they are preceded by reception (a silent period), that is, by processes that try to capture both the physical form and semantic function of the language items being learnt. That is so because “words” must be represented in the neural network, as declarative knowledge, before the mind can analyze them. For them to enter the neural net, however, they must first be noticed, and there must be a motivation to further analyze them. That takes us to the crucial role of motivation in learning.
The brain is a biological organ, and its function is to serve the interests of the whole organism. The creation and use of knowledge patterns in the brain is therefore motivated by the interests of the whole organism. That is why new knowledge patterns are created only if a person sees a useful purpose (function) in their construction. With regard to the recall/use of knowledge patterns it must be stated, though, that because in a given situation different knowledge patterns (i.e., options for action, plans or desires) compete, what is needed is a sort of executive committee that decides which of the competing options are realized (Faw, 2000). In the brain (Figure 1), the role of such a central control is reserved for the joint operation of the Prefrontal Cortex and neighbouring parts of the Limbic System. The Prefrontal Cortex is the place where cognitive plans and moral decisions are made; the Limbic System some people call the Emotional Brain (Goleman, 1995).
The Prefrontal
Cortex is not itself a seat of knowledge but registers which areas of
the cortex are “on” but in competition with each other, all “calling for attention”.
They are “specialists”, and together they represent the complete range of
actions that could be taken at a given 
moment.
They meet, though, with affective messages from the Limbic System, and it
is only by the joint operation of cognition- and affection-based factors that
specific processing routines receive “permission to fire”. For FLT the message
is: no processes without a motivation to reach a specific goal, and no processes
without the knowledge patterns (processing routines) that know how to reach
that goal. This underlines the need for a functional approach to the study
of language and learning, which is in keeping with what teachers who take
a pragmatic approach to FLT say on the basis of their classroom experience.
It is essential to remember, however, that the processing capacity of the Prefrontal Cortex is limited. It actually is identical with what is called Working Memory, which can handle no more than the famous 7±2 units known from memory research. It is well-known, too, that the speed and power of mental processes can be enhanced if individual bits of knowledge are compiled in large chunks of knowledge, which are stored in Long Term Memory as processing routines (schemata/skills). The chances of their automatized recall are enhanced through practice. With regard to such processing routines it must be pointed out, though, that their initial construction is subject to constraints: Some processes presuppose that certain other processing routines have been established prior to them. It is impossible for instance, to learn to jump or dance before learning to walk. What that means for SLA is spelled out by PT, which is discussed in the next section.
2.2 PT and the implicational hierarchy of processing procedures
The goal of general-purpose courses in FLT is to enable learners to fluently use the target language in communication. Considering, however, that oral communication requires a much faster processing of language than writing, what L2 learners say in oral communication is the acme test of their language competence. In other words, although it is possible to train learners to do well in set formal grammar tests, they do not provide reliable information on the real stage of their language competence. Speech requires automatized processing routines, or else, with the limited capacities of Working Memory, it would become impossible to find, within fractions of a second, the words and morpho-syntactic structures that can translate concepts into linguistic forms appropriate to them.
PT therefore takes the productions of L2 learners in (fluent) meaning-oriented oral communication as the measure for assessing their stage of development in SLA. Speech production rests on the fast interaction of several components of the language faculty. Their interplay is described in the well-known model of speech production by Levelt (1989; see Figure 2). It shows that, given that a speaker has decided what he wants to say, which is the work of the conceptualizer, he next has to find in his mental lexicon words (lemmas) that can express that meaning. It is part of the entry of a lemma in a mental dictionary that it contains a set of semantic features with information on the role that it can play in a syntactic frame (predicate argument structure) (Levelt, 1989, p. 191). If we take, for example, the verb “give”, its entry in the mental dictionary makes clear that it is a verb that requires three arguments: One argument provides information on who does the giving (agent), a second says who is the recipient, and a third argument names the object given (theme). The lemma “give” therefore contains information on the morphological form which its arguments must take with regard to the expression of case, tense, aspect, person, number, etc. Correspondingly, the speaker must find in his lexicon the specific morphological forms that can assure that all the words of a sentence agree with the intended message. Seeing to that is the job of the formulator. Finally, the mentally encoded speech plan must be articulated to become a communicative message.
Traditionally linguists consider the process of sentence generation to be strictly serial in time: First, the conceptual context is specified by the conceptualization process; next the syntactic structure is built for the whole utterance; finally this structure is realized phonetically. But that is, as Kempen and Hoenkamp (1987, pp. 202-203) have shown, a wrong conception of speech production. It really proceeds incrementally, not in a strictly serial order, because otherwise the high speed with which speech is produced could not be explained. In incremental processes a chain of highly automatized processing procedures is set in motion in which “the next processor can start working on the still-incomplete output of the current processor” (Levelt, 1989, p. 24). In other words, speakers mostly start with naming the topic of a sentence before they have a complete plan for how to finish it.
Considering that speakers, in fractions of a second, must (a) find the right words (lemmas) for the different parts of a speech plan, and (b) put them in a (syntactic) order that represents their communicative intention (question, statement, negation, etc.), and (c) put the words in a morphological form that agrees with the rest of the utterance, it becomes clear why, in that interlocking chain of processing tasks, the whole procedure can come to a stand-still if one of the processors is not available as a processing routine.
In that context the evidence from SLA studies is that the different processing routines develop in a universally similar sequence that is plausible from a psycholinguistic perspective: At the lowest level, a word must have been entered in the L2 lexicon before a grammatical category can be assigned to it. The latter is needed for the word to find its proper place in a phrase structure with its head and modifiers. Only if such a phrasal procedure has been completed and its value is returned can the function of the phrase (predicate, type of argument) be determined; and only if the function of the phrase has been determined can it find its proper place in a complete sentence. Given that this is the “logic” underlying the development of speech processing routines, the next question is: What happens when an element is missing in this implicational hierarchy?. The answer is that the hierarchy will be cut off in the learner grammar at the point of the missing processing procedures, and the rest of the hierarchy will be replaced by a direct mapping of conceptual structures onto surface form, if there are lemmas that match the conceptually instigated searches of the lexicon.
With regard to English, that leads to the implicational hierarchy of processing procedures represented in Table 1. The table shows that language acquisition begins (Level 1) with the use (reproduction) of holistic forms (unanalyzed forms). On Level 2 the learners start with a grammatical analysis of the holistic forms and come to first results, which enable them to distinguish basic word categories and to use semantically motivated word combinations like “many money”, or “my house”, or “two child”. As far as the (syntactic) arrangement of the words is concerned, learners stick to a conceptually instigated sequencing of the words (canonical order).
Table 1. The processing hierarchy applied to English (Modified following Pienemann, 1998, p. 7)
| Stage | Processing procedure | L2 process | Morphology | Syntax |
Examples |
| 6 | subordinate clause procedure | main and subordinate clause procedure | Cancel INV | I wonder
why he sold that car. |
|
| 5 | S-procedure/ WO-rules/ -salience | inter-phrasal info-exchange | SV agreement (= 3sg –s) |
Do2nd, Aux2nd | Why did he sell that car? |
| 4 | S-procedure/ WO-rules/ +salience | inter-phrasal info-exchange | tense agreement | Yes/No inversion, pseudo inversion
|
Has he seen you? Can he come? |
| 3 | phrasal procedure | phrasal info-exchange | NP agreement |
|
two dogs / two womans |
| 2 | category procedure | lexical morphemes | plural marking possessive pro |
canonical word order | many houses my house, two childs I like spaghetti |
| 1 | word/ lemma | “words” | invariant forms | single constituent | Good morning / How are you / Sit down, please / yes / no / one / two / three |
Once a learner has sufficient experience (practice) in the fast production of such strings of words and knows enough “words” to express his basic communicative intentions without long searches, the availability of such processing routines frees Working Memory from the task of a consciously controlled and therefore resource-consuming construction of each basic utterance. At this stage the next step in language development can be taken, and that is (Level 3) that learners now add, more or less regularly, a Plural –s to nouns, like in “two dogs”; but it is typical of this stage that over-generalization of this rule happen, like in “two womans” or “two childrens”. In psycholinguistic terms: the formulator has now learnt to react to conspicuous (salient) semantic aspects of lexemes like “many” or “two” by automatically adding a plural suffix to the corresponding noun in the same phrase. That is an intra-phrasal information exchange or intra-phrasal unification of forms. Parallel to that, learners begin, on the syntactic level, to modify the overall meaning of an utterance by the fronting of either an Adverb, or a Negation-Particle or a Question-Indicating “Do”, while not otherwise changing the canonical word order. This leads to sentences like “Yesterday we to home”, and “There children play”, and “Do he going home?”
After the formulator has learnt to arrange words in this way, the next step is (Level 4) that processing capacity becomes available for now tackling the task of a unification of the grammatical features of items on an inter-phrasal level, that is across the boundaries of a single Noun-Phrase or Verb-Phrase. Generally learners start with supplying conceptually salient items with formal markers that reflect their agreement; a typical example is that a verb is given a plural (singular) marking in agreement with the corresponding noun in the nominal Noun-Phrase (the Subject). Parallel to such a morphological marking of matching items across phrase boundaries, learners begin, on the syntactic level, to use inversion of constituents as a means of indicating questions. That produces sentences like “Has he seen you?” and “Have you done it?” and “Do he sees it?”
Only after such processing routines have been established in Long Term Memory, which spares Working Memory the task of consciously producing such a formal unification across phrase boundaries each time anew, do learners manage to extend inter-phrasal information exchange to non-salient parts of phrases, like 3rd person singular present –s (Level 5). Parallel to that, but on the syntactic level, the formulator now can do mental operations that lead to the inversions that are typical of wh-fronted questions, like “Where has he gone?” or “What is she eating?” (instead of sentence structures like “Where he has gone?” or “What she is eating?” which are typical of Level 3).
And, again, only after sufficiently automatized procedures for the production of such structures have been established, can learners tackle the next task (Level 6), which eventually puts them in the position to reverse, in indirect questions, the inversion they have learned to use in questions. They now learn to produce sentences like “I wonder where he has gone” and “I wonder what she is eating”, instead of utterances like “I wonder where has he gone”, and “I wonder what is she eating”.
The examples illustrate the nature of the implicational hierarchy of processing procedures that is typical of all SLA processes. What needs to be seen is (a) that at each stage of development a learner will make typical – and predictable – (types of) errors; that makes it (b) possible to distinguish errors that a learner should no longer make at that stage – and which teaching therefore should attend to – from errors that are harmless because they are, from a developmental perspective, “inevitable”; it (c) provides teachers with crucial information on “what is teachable”, which differs from the prescriptions of what must be taught when which were based on structural syllabi; it (d) allows a distinction of serious deficits in core areas of language development, which teachers must attend to, from what is due to individual variation.
What must be stressed is that “universally similar developmental stages” does not mean that there is a biologically pre-programmed guarantee that all learners steer the right course. On the contrary, things can go wrong, and not all learners reach near-native speaker competence sooner or later. The basic thesis of PT is that different outcomes of acquisition processes are, at least partly, due to different developmental dynamics. The basic mechanism behind them is
the principle that developmentally early decisions bias the further development of the interlanguage system. This percolation of structural properties in developmental processes is known in biology and philosophy and has been termed “generative entrenchment”… (Pienemann, 1998, p. 316)
For FLT that means (a) that it is possible, beyond a vague call for “comprehensible input”, to define which linguistic forms can be taught at what stage of a learner’s language development (either implicitly, by input enhancement, or explicitly by formal instruction), and (b) it means that if a specific learner fails to produce the structures which he should produce at the given stage of development, that is the time for teaching to step in with specific supportive measures.
Another important aspect of PT is that it underlines the need for a meaning-and-form oriented approach, because
[f]irst of all, there is no reason to assume that learners acquire a structure just because they can process it. A functional need would have to be present for the structure to emerge. Secondly, functional constraints interact with processing constraints. … In other words, even if the basic processing of the grammatical information related to a given structure is in principle possible some of that information may not be available to the learner because of complex form-function relationships. (Pienemann, 1998, pp. 250-251)
That must be put into relation to well-known approaches to FLT which, on the basis of classroom research and the experiential knowledge of teachers, call for a task-based approach to FLT (Long, 1988; Doughty & Williams, 1998).
3. Processability and task-based learning
FLT has always been faced with questions like: Do we need a formal (structural) syllabus? How should it be constructed? Is grammar best learnt and taught implicitly or explicitly? What should we do about learners’ errors? Should there be a focus on form, and when should it occur in the overall curriculum? Which forms are amenable to a focus on form? Are some forms resistant to focus on form? Can tasks and techniques be designed during which problematic forms are likely to arise so that an opportunity to focus on form can be provided? Should a focus on form occur only if a learner asks for it? And should it be restricted to awareness raising and the training of learning strategies?
In the introductory chapter of this essay I pointed out that under the three types of approach to FLT different answers are given to such questions. Traditional approaches prescribe which forms must be taught when, and they mostly favour explicit instruction in the rules of grammar. At the opposite end we find radical communicative approaches which trust that grammar can be acquired implicitly and incidentally while the learner’s attention is focused on content, with the proviso that SLA may be enhanced by providing comprehensible input. (What makes language input “comprehensible” is a different issue; the options range from structurally and lexically simplified textual input to teacher talk and recasts of learner output in interactions with their teacher.) In comparison, the constructivist approach, with PT as a theoretically consistent part of it, occupies a middle position. What needs to be stressed, though, is that it represents a psycholinguistically well-founded position, and not just a compromise solution.
PT can predict which forms of grammar a learner can process at what stage of development, but systematically distinguishes individual variation from core developmental aspects of SLA (Pienemann, 1998, pp. 234-249). That includes the option, for syllabus construction, to (a) find tasks and materials that contain a crucial amount of the language forms that a learner can integrate into his learner grammar at a given stage of development (“teach what is teachable”); it (b) makes it possible to help learners notice the targeted forms by, for example, input enhancement, that is highlighting them in print or colour, or by making them a point of explicit discussion; it (c) allows to distinguish harmless errors from errors that put learners on the wrong track in all subsequent processes of hypothesis formulation and testing, errors therefore that teachers should not tolerate.
Whether, in this context, the necessary focus on form is best achieved by strategies of implicit or explicit learning must be left to the discretion of the teacher and must be decided on the basis of what is known about the preferred learning style of the individual learner. Both styles of learning are compatible with what I said before on language learning being exemplar-based; the explicit formulation of a rule can support the assembling and compilation of small units of declarative knowledge into processing routines; but rule knowledge cannot replace the development of the relevant processing routines. This position is identical with the one proposed by Paradis (1994) who sums up the case as follows:
What is automatised is not the explicit knowledge of the rule, … but its application. … What is automatised is the ability to produce the correct sequence of words in their proper inflectional form, whatever the processes have been used to reach this result. These remain in fact for ever opaque to introspection. (Paradis, 1994, pp. 401-402)
That is also what Hecht and Green found out in their empirical study of SLA (Hecht & Green, 1992).
Taking a task-based (function-oriented) approach to FLT is not in opposition to having phases of form-focused learning. Tasks turn the learners’ attention to the communicative function (meaning) of the language forms, and – as stated by PT – there is no reason to assume that learners acquire a structure just because they can process it. A functional need would have to be present for the structure to emerge. That however can be provided only by a task-based approach, which has the additional advantage of potentially addressing all the senses of a learner and of tapping motivational resources beyond an intellectual interest in language and grammar (Timm, 1995a). It is a different matter whether to place a task always at the beginning of a learning process, or to have it follow an introductory phase of form-focused activity in a task cycle. Teaching experience speaks for that to be pragmatically decided in the context of specific learning processes and on learner-specific grounds.
References
Allwright, D. (1984). Why don’t learners learn what teachers teach? The interaction hypothesis. In D. M. Singleton & D. Little (Eds.), Language learning in formal and informal contexts (pp. 3-18). Dublin: Irish Association for Applied Linguistics.
Baddeley, A. (1986). Working memory. Oxford: Clarendon Press.
Bates, E., & MacWhinney, B. (1989). Functionalism and the competitive model. In B. MacWhinney & E. Bates (Eds.), The crosslinguistic study of sentence processing (pp. 3-73). Cambridge: Cambridge University Press.
Bechtel, W., & Abrahamsen, A. (1991). Connectionism and the mind. An introduction to parallel processing in networks. Cambridge: Blackwell.
Chomsky, N. (1980). On cognitive structures and their development: A reply to Piaget. In M. Piattelli-Palmarini (Ed.), Language and learning. The debate between Jean Piaget and Noam Chomsky (pp. 35-54). London: Routledge & Kegan Paul.
Deacon, T. (1997). The symbolic species. The co-evolution of language and the human brain. London: Penguin.
Doughty, C., & Williams, J. (Eds.). (1998). Focus on form in classroom second language acquisition. Cambridge: Cambridge University Press.
Faw, B. (2000). Consciousness, motivation, and emotion: Biopsychological reflections. In R. D. Ellis & N. Newton (Eds.), The cauldron of consciousness. Motivation, affect and self-organization (pp. 55-90). Amsterdam: John Benjamins.
Goleman, D. (1995). Emotional intelligence. Why it can matter more than IQ. London: Bloomsbury.
Hecht, K., & Green, P. S. (1992). Grammatikwissen unserer Schüler: gefühls- oder regelgeleitet? Praxis des Neusprachlichen Unterrichts, 39(2), 151-162.
Hörmann, H. (1976). Meinen und Verstehen. Grundzüge einer psychologischen Semantik. Frankfurt: Suhrkamp.
Jackendoff, R. (1983). Semantics and cognition. Cambridge: MIT Press.
Kaplan, R., & Bresnan, J. (1982). Lexical-functional grammar: a formal system for grammatical representation. In J. Bresnan (Ed.), The mental representation of grammatical relations (pp. 173-281). Cambridge: MIT Press.
Kempen, G., & Hoenkamp, E. (1987). An incremental procedural grammar for sentence formulation. Cognitive Science, 11, 201-258.
Krashen, S. (1985). The input hypothesis: Issues and implications. London: Longman.
Levelt, W. J. M. (1989). Speaking. Cambridge: Bradford/MIT Press.
Long, M. H. (1988). Instructed interlanguage development. In L. Beebe (Ed.), Issues in second language acquisition: Multiple perspectives (pp. 115-141). Cambridge, MA: Newbury House.
Long, M. H. (1998). Focus on form: Theory, research, and practice. In C. Doughty & J. Williams (Eds.), Focus on form in classroom second language acquisition (pp. 15-41). Cambridge: Cambridge University Press.
MacWhinney, B. (1989). Competition and teachability. In M. L. Rice & R. L. Schiefelbusch (Eds.), The teachability of language (pp. 63-104). Baltimore: Brookes.
Multhaup, U. (1999). The construction of knowledge: Its conditions and causes. In B. Mißler & U. Multhaup (Eds.), The construction of knowledge, learner autonomy and related issues in foreign language teaching: Essays in honour of Dieter Wolff (pp. 91-108). Tübingen: Stauffenburg.
Nelson, K. E. (Ed.). (1978). Children’s language. New York: Gardner.
Paradis, M. (1994). Neurolinguistic aspects of implicit and explicit memory: Implications for bilingualism and SLA. In N. C. Ellis (Ed.), Implicit and explicit learning of languages (pp. 393-419). London: Academic Press.
Piaget, J. (1980). The psychogenesis of knowledge and its
epistemological
significance. In M. Piattelli-Palmarini
(Ed.), Language and learning. The
debate between Jean Piaget and Noam Chomsky (pp. 23-34). London: Routledge
& Kegan Paul.
Pienemann, M. (1984). Psychological constraints on the teachability of languages. Studies in Second Language Acquisition, 6(2), 186-214.
Pienemann, M. (1998). Language processing and second language development. Processability theory. Amsterdam: John Benjamins.
Pinker, S. (1989). Learnability and cognition: The acquisition of argument structure. Cambridge, MA: MIT Press.
Swain, M. (1985). Communicative competence: Some roles of comprehensible input and comprehensible output in its development. In S. Gass & C. Madden (Eds.), Input in second language acquisition (pp. 235-256). Rowley, MA: Newbury House.
Timm, J.-P. (Ed.). (1995a). Ganzheitlicher Fremdsprachenunterricht. Weinheim: Deutscher Studien Verlag.
Timm, J.-P. (1995b). Die „Fuzziness“ der Sprache als Begründung für einen ganzheitlich-funktionalen, erfahrungsorientierten Grammatikunterricht. In J.-P. Timm (Ed.), Ganzheitlicher Fremdsprachenunterricht (pp. 120-148). Weinheim: Deutscher Studien Verlag.
Wolff, D. (1994). Der Konstruktivismus: Ein neues Paradigma in der Fremdsprachendidaktik? Die Neueren Sprachen, 93, 407-429.