Grammar for Teachers (Seminar)
All ideas expressed in Thor's Stories and The Passionate Skeptic are entirely those of the author, who has no aim to influence, proselytize or persuade others to a point of view. He is pleased if his writing generates reflection in readers, either for or against the sentiment of the argument.

When Grammar Doesn't Help

Thor May
South Korea 2007


Abstract : This paper questions the role of grammar in language teaching and learning. Firstly it identifies the constituencies in academic language teaching, and their often conflicting notions of language programs. Several kinds of learners are discussed, with particular attention to the large group who are uncomfortable with any technical analysis, including formal grammars. Some conventional ideas about what a natural language grammar actually is are challenged. The consequences of a connectionist view of language processing are briefly explored. The power of collocation sets is identified as a key to language acquisition. Language is set in the broader cognitive context of memory processes and patterns of generalization. Pedagogical grammars are viewed as forced external generalizations with little organic presence in memory, but some suggestions are made about how to make use of them. Actual student language memory, as well as teacher self-insight into L1 are both contrasted with the idealized patterns assumed by academic language programs. Finally, the stubborn problem of average teacher behaviour is set against the real ways in which people appear to use grammars and learn languages.

Sections : 1. Grammar as a Cultural Artefact / 2. Types of Language Learners / 3. The "Two Cultures" Theory for Every Civilization / 4. The Nature of Grammar in Natural Language / 5. Abstraction and the Acquisition of Language / 6. Grammar and Abstraction / 7. Language Teaching and Forced Generalizations / 8. The Uses of Pedagogical Grammars / 9. How Students Really Remember a Language / 10. What Do Native Speakers Already "Know" About Grammar ? / 11. The Monday Morning Question / 12. Conclusion



1. Grammar as a Cultural Artefact

The role and nature of language grammars depends upon who uses them, when, where and for what purpose. In that sense grammar is a cultural artefact. This paper looks at grammar from within the world of language teaching. Firstly it will discuss the people who use grammars in this context. Secondly it will explore why book grammar is like a respected elder sister who always has to get a party invitation, but is rarely invited onto the dance floor where the real action takes place.

Several cultures attach to language teaching and learning. There are the cultures of various kinds of language learners, both in a national sense and the narrower sense of language learning style. There are the cultures of teachers, of teacher trainers and of the publishing industry. There is the academic culture of applied linguistics. There are also cultures of community expectation about language teaching and learning.

Historically, grammar has had a shifting status amongst all of these groups. Often the groups are in conflict about the uses to which grammar may be put, and sometimes in their very understanding of what grammar actually is. Also, although language teaching has become increasingly internationalized amongst professionals (especially in TEFL), the main body of local teachers, and of course students, remain strongly influenced by local conditions. What is agreed wisdom in Europe or America isn't necessarily what happens on the ground in say, East Asia. References to teacher/student behaviour in this paper are mostly drawn from my experience in China and South Korea.

Despite the confusion of competing interests, it is rare to find a language teacher training course which does not include units on grammar. It is also fairly rare to find a language learning textbook for any language which doesn't contain grammatical explanations. Every bookshop with a language learning section contains pedagogical grammars, so presumably somebody buys them. Worldwide most foreign language teaching (including teaching English as a foreign language) is done by non-native speakers, and overwhelmingly the approach they use is some form of grammar translation. Sometimes it is unfashionable to admit this in conferences, but it is the classroom reality. The effective uses, if any, to which pedagogical book grammars are put will be questioned by the analysis which follows.


2. Types of Language Learners


In most countries, most inhabitants from birth fail to learn a foreign language in schools, although they may very well do so in their homes. That is, school work fails to give them a working competence in a second language. In countries like Australia, America, Japan or South Korea that failure may be 90% plus of all those who start some kind of course (Asher 2004). Therefore, one very interesting question we can ask is what kind of people are generally successful at second language learning? What is success? There is also a closely related question : what kind of people become foreign language teachers?

We know that under the right conditions all kinds of people can and do learn a second language. The "right conditions" in this sense usually means people who already live in bilingual or multilingual communities where there is a demand for those multilingual skills, or people who move to another country and have to learn a second language to prosper. "Have to" is the key item here : Bangladeshi labourers in Korea may be more likely to learn some working Korean than expatriate foreign English teachers. The bureaucratic/military elite of many nations also require a second language (usually English) as a class membership qualification, and their children are typically sent to English speaking countries for enculturation.

Increasingly also we can find a kind of fractional use of English worldwide in workplaces by very large numbers of people. Such people, for example in a shipping office, may have subroutines in their daily employment which use an English component (maybe quite minor), and which they develop stratagems to handle. In such cases, the L2 element is effectively encapsulated as a specialized set within L1. This is a little studied phenomenon, and the path of these users to adequate job competence surely needs thoughtful examination.

In addition there are certain groups of people who learn a second language more or less successfully although they may have very little contact with native speakers of this language. Firstly, there are some professionals, like doctors or engineers or scientists who may find that most textbooks for their subject are in, say, English. These folk are usually trained learners with efficient memories to begin with, and they apply great discipline to their task. They tend to be analytic, and can make good selective use of the pedagogical grammar material in traditional language learning publications. Very often they arrive at a competent reading knowledge of L2 in their subject area, at a minimum.

Another major group of non-contact successful L2 learners are humanities type people, especially folk who finally become language teachers themselves. I am always mildly astonished to meet such individuals with a working competence in English when they have never lived within reach of a live, English speaking community. Over many years I have had a lot of contact with successful learners of this kind. Certain characteristics stand out:

a) On the whole they are uncomfortable with technology, unless it has been made idiot proof for mass marketing;
b) on the whole they dislike mathematics;
c) on the whole they are baffled by strict formal logical argument;
d) on the whole they dislike and are baffled by linguistic analysis (i.e. formal grammar). I am certain of this last point because I have had to teach large numbers of them "grammar" in teacher training courses.
e) in the language studies arena, there is also a gender imbalance in this group, with a preponderance of women. Exploring the nature/nurture origins of that would take us too far off-topic here. We can note in passing though language specialization is culturally acceptable for women in many parts of the world, where other roles might not be.

There is a caveat here. The preferences just listed are what I have interpreted people to feel and do, not necessarily what they say. What any group *say* they prefer tends to be strongly influenced by public ideology. For example, if it seems cool to be a techno geek in some sub-culture, then large numbers of people will talk the talk, even if they can't walk the walk. Similarly, significant numbers of these people, especially the older and more solemn, may lament "the failure of schools to teach proper grammar", whatever that means within their cultural code.


3. The "Two Cultures" Theory for Every Civilization


Perhaps we can sum up the relevant characteristics of humanities type people I have just described in this way : they tend to dislike any abstraction which can't easily be attached to an emotional key. That is, some of them might wax lyrical about (say) 'freedom' or the meaning of life - both highly abstract concepts, but layered with emotion and cultural embroidery. A mathematician's fascination with superstring theory, or more poignantly for our argument here, a theoretical linguist's pursuit of, say, a universal grammar, is likely to leave them confused, or even resentful.

These observations are not new. A generation ago the British writer & scientist, C.P. Snow made famous the concept of 'two cultures' (1959 Rede Lecture). His analysis of the problem, which he linked closely to education and even political preference, may have been deeply flawed, but the existence of those two contrasting orientations does create a destabilizing tension. I see that destabilizing tension as an engine driving all human groups, although a particular culture may favour one orientation above another. I also see the divide more as a double bell curve with a smaller number of individuals between the peaks who can mediate. The equation is further confused in our environment of frantic social change by all kinds of specialized knowledge, and even generational divisions. For example, you will find many older automotive mechanics appalled by car computers, which they regard as black boxes from hell. Nevertheless, Snow's original dichotomy is a useful, crude starting point when we think about not only what people learn, but how they learn.


4. The Nature of Grammar in Natural Language


The classic generative description of syntax is that it is a finite set of phrase structure rules applied to a finite set of words to generate an infinite number of sentences; (there are frills, but that is the crux of it). Pedagogical grammars share this notion of a finite set of rules applied to a finite vocabulary. The schema is essentially a linear programming process, sometimes diagrammed as recursive phrase structure rules. Within such a schema, sentences are either right or wrong according to the rules. Of course there are organizational patterns in language at the textual level too, not merely at sentence level or below. Some systemic and discourse grammar models have attempted to map these patterns. Discourse level grammar is much less categorical than sentence level grammar, and it is relevant to this discussion that humanities type people, including most language teachers, tend to find it more congenial than the strict formulations of phrase-structure grammar.

In fact, human brains seem to be parallel processing devices, not linear binary computers. Connectionist/parallel processing models (Rumelhart & McClelland 1986, and many others since) may not necessarily describe quite what happens in a human brain, but they create better analogies for exploring that process (including the process of language creation) than the linear schema that preceded them. For example, one intriguing property of connectionist/parallel processing systems is that they are self-teaching (i.e. the grammars are "emergent").

From an information systems perspective, parallel processing has some important properties, notably :

1) It is unpredictable within certain degrees of freedom; (mathematical Chaos Theory is the formal expression of this). For example, you often predict the meaning substance of what someone is about to say, though you may not be able to predict the exact phrasing. Likewise, we may know there will be a certain number of cyclones each season, but not their precise strength and timing.
2) It is robust at the margin. Thus, with the examples just given, our predictions of a friend's utterance and the meteorologist's predictions of cyclones are pretty reliable.
3) Parallel processing systems can only be understood as patterns of probabilities, not as sets of categorical rules.
4) Related to 1-3 above, parallel processing is not very vulnerable to catastrophic failure. Linear computing programs (as in your office computer) often crash. In a parallel system, some workable solution is usually available.

Consider the consequences of parallel processing models when applied to language learning. They seem to predict pretty well what happens : we produce sentences which may conform or be deviant from a norm, but their validity depends upon successfully decoding the speaker's intention, and that is a probability game, not a game of applying categorical right/wrong rules. In such a parallel processing environment, the absolute phrase structure rules of generative grammars, or the more eclectic but still strict rules of pedagogical grammars have much less descriptive power. At best traditional grammar rules (of syntax, phonology, morphology etc) are rough abstractions from typical patterns. It is quite likely that those output patterns could be arrived at by somewhat different cognitive routes by different speakers, or even by the same speaker on different occasions.

Another of the foundation assumptions of traditional grammars, and linguistics generally, has been that language is a system of symbols which somehow represents "the real world". I have challenged this elsewhere (May 1994; the concept of Generative Oscillation), arguing that whatever we mean by "real", language is only partly representational. A great deal of what we say, including much inner talk, is (I think) driven by the unstoppable dynamic of the cognitive language machine itself, unmotivated by social purpose or a shared objective world. In other words it is extremely difficult *not* to talk, either socially or internally (try meditating!). This semi-autonomy has consequences for language learning too extensive to explore here, but also for the notions of grammars which could describe it. It appears that connectionist type models and chaos theory with their emergent grammars form a much better explanation of non-representational language than traditional phrase structure grammars with their notions of a static symbolic system.

The fuzziness of the grammars just suggested has been rather unattractive to the ordered and logical minds of many researchers drawn to the formal study of natural and synthetic languages. This may explain much about the dominance of linear models in the grammar marketplace, and has certainly been a misfortune from the viewpoint of many language learners and teachers.


5. Abstraction and the Acquisition of Language

Whatever the configuration of language processing in the human brain, our focus in this paper is on how different classes of language learners actually develop enough mastery to produce decodable communications in a second language. I will argue here that their success or failure in acquiring this skill is influenced by how abstract they find the task to be, and that abstraction is intimately related to memory.

Without memory we are no longer human : cultural identity and of course language has no residence in a creature without memory. The internal structure of human memory is itself controversial. Here I will assume something like Ausubel's theory of subsumption (Ausubel 1962, 1978).

Immediate experience is memorable according to its emotional resonance. This emotional content may take many forms, coloured by the association of physical sensations, earlier memories and all kinds of cultural inferences. That mix will vary greatly amongst individuals, and we could also guess that the varying biochemistry of individuals may have a big influence on what threshold of stimulation will lead them to accept one kind of immediate experience or another for longer term memory storage.

On the whole we don't accept individual experiences for long term memorization. There are obvious exceptions to this, but the second by second, lifetime flood of sensations would quickly overwhelm us without filtering. The normal process is to seek patterns in experience, and abstract those patterns as generalizations, or as Ausubel argues, subsume them into already known patterns. Later, we even make generalizations about generalizations. Eventually we may attach broad emotional attitudes to some special abstract concepts, not all of them. (Some political concepts such as democracy and freedom would be instances of emotionally laden abstractions). In general however, abstractions are drained of most emotional colour. This has consequences for memory.

Generalizations which we extract on the basis of personal experience have a very organic presence in our memory. They are quite resistant to change or to challenge. However, pre-packaged generalizations, offered externally on the basis of authority, are usually not easily accepted for personal memorization (although individuals vary in accepting external authority : another acquisition variable). Even when stored under duress, pre-packaged generalizations are frequently not deployed for practical application. The further from immediate experience a pre-packaged generalization is, the less adhesive it is in memory. There is therefore a tradeoff between storing a mass of specific but (perhaps) emotionally charged material, and storing compact but emotionally dead generalizations.

Prototype theory (Eleanor Rosch 1978) draws on this tension to predict that in hierarchies of types, our "default" recall tends to cluster around an intermediate level of generalization. Thus we find the word token 'dog' more generally useful than "Mary's terrier", or even "terrier". On the other hand "canidae" (of which the class of domestic dogs is a member), is both too general and too emotionally abstract to be very memorable to us. Isn't it easier to think of your favourite dog, than to think of your favourite canidae ? Interestingly though, the notionally more abstract species term in this particularly hierarchy, "mammal", does have popular currency. For that we can probably thank our own star membership of the group, and a generation of emotional nature programs on television.


6. Grammar and Abstraction

The history of science is essentially a record of how human societies have come to recognize general truths which had previously been hidden behind a torrent of individual experiences. We have always sought general truths of this kind, sometimes through inducting from the combined experience of village or tribal elders, sometimes by putting faith in the guesswork or insight of influential individuals, sometimes through the medium of religion or magic. Experimental science gave us a replicable tool for arriving at generalized insights while minimizing the idiosyncrasy of individuals.

Just as the understanding of generalized patterns in the public domain has yielded great power to groups of people - that is to societies -- so the search for general truths is characteristic of individuals everywhere. These personal generalizations, whether sound or faulty, give each person the power to operate confidently as an autonomous being. We call it learning from experience, and its expression may be 'prejudice' or 'wisdom', depending upon your point of view. Clearly individuals vary greatly in the sophistication of the generalizations they arrive at.

Although our conscious generalizations in life may be unreliable, a unique characteristic of natural languages is that even the least sophisticated individuals, from early in life, are able to employ apparently generalized patterns of symbols which enable them to exchange complex messages with other people. We call it language. That is, those generalized patterns are not just private. They are similar enough to the ones used by other speakers to become part of a wide-area-network.

The wide-area-networks of natural language have many intriguing properties. They are robust and reliable, in spite of constant local transmission failures. Thus our personal language is full of small slips, and other people frequently fail to fully understand us, but this is a near-enough-is-good-enough system. We also use a mass of contextual and other clues to guess speaker intent where the transmission pattern of language is not adequate by itself.

Perhaps the most puzzling aspect of natural languages though is that no one, from the least to the most sophisticated users, is able to describe the deepest general patterns which underlie and support their individual utterances. These general, or master patterns are called 'grammar' in language; (the master pattern underlying any complex system could be called its grammar). That is, grammar is the most extreme abstract description of how words collocate (hang together in clusters). My own suspicion, drawing on connectionist ideas, is that "the deepest general patterns" do not have any fixed presence, but are rather probable outcomes generated by the relation between mathematical properties of the cognitive parallel processor and a large array of "lexical collocational weights" which we do store in memory. By collocational weight I mean the likelihood that two or more words will co-occur. This is knowledge that every speaker demonstrably has .

Accumulating collocational knowledge within a connectionist system is very efficient. It seems to me to be not only a more credible source of language acquisition than Chomsky's problematic innate Language Acquisition Device (Chomsky 1979), but to render the LAD redundant. If I am right, then it is hardly surprising that most language learners feel little emotional bond with the book grammars they are confronted with, and find it difficult to employ them productively. Generalizing about all the pointy headed professors you have met is just not the same as generalizing about the grammatical rules of your language.


7. Language Teaching and Forced Generalizations

There is an assumption behind the use of all pedagogical grammars. Teachers believe that by giving students explicit statements about the most abstract patterns underlying language (grammar), then those students will be able to reverse engineer language acquisition, and apply the pre-packaged abstract grammar rules to efficiently produce language in L2. This is even though speakers (including the teachers) are unable to articulate similar abstract patterns when you ask them to describe the operation of their own mother tongue. It is also although nobody is sure whether the pedagogical rules for grammar which students are told about resemble in any way the real system of cognitive abstractions used by any speaker of the target language..

What is the net result for language acquisition of forcing a set of grammatical abstractions on students? We do have empirical knowledge about this:

a) where there is no compelling reason for students to keep studying the language academically, the vast majority withdraw from studying it altogether;
b) where students do have a compelling reason to continue studying, such as an examination requirement or an employment requirement, then they will indeed memorize and learn to formally manipulate the pedagogical grammar generalizations they are given;

c) the manipulations which students from situation (b) learn may resemble live language strings sometimes, but when faced with a genuine communication situation these students overwhelmingly are unable to produce intelligible language. Countries such as South Korea are huge laboratories, involving millions of learners, for demonstrating the (b) and (c) outcomes.


8. The Uses of Pedagogical Grammars

Are pedagogical grammars entirely useless then? No, it is clear that some people, not all, find them useful occasionally in a particular way. The way that pedagogical grammars can have some power is in what we might call the 'aha!' factor. Any learner of an unfamiliar language is constantly faced with baffling arrangements of words. The puzzlement remains even after the learner has memorized the dictionary meaning of individual words. He may have no confidence in his ability to produce similar patterns.

In fact learners who spend a lot of time in genuine communicative situations with the new language do eventually become familiar with "ways of saying things". That is, they become familiar with certain collocations of words and will gradually form the subconscious generalizations which make up their real grammar of L2. However, just as some individuals philosophize consciously about, say, the meaning of life and will grasp at a prepackaged answer which seems to sum it up (a religion, an ideology .. ), so in the right frame of mind an individual might grasp at a pedagogical grammar rule which seems to package a generalization they were seeking to make. In both instances, the ideology or the pedagogical grammar rule, the generalization might be genuinely valuable, or faulty. Regardless, it becomes a tool for those who voluntarily accept it.

Successful teaching, like successful advertising, is largely about creating a targeted need in the minds of others. Here perhaps there is an opening for 'teaching grammar'. I have just argued that pedagogical grammars can be of some value when they prove a shortcut to sorting out felt puzzlements. Academically bright students may spontaneously employ book grammars as puzzle solvers in this way. With less able or less logically inclined students, there is scope for a teacher to work behind the scenes in setting up classroom communicative problem situations where recourse to some rule of pedagogical grammar may be one of the resources available. Some of the studies in language teaching literature of "a focus on form" in various kinds of communicative classes could be examples of this kind of grammar salesmanship.


9. How Students Really Remember a Language


Eleanor Roche's Prototype Theory (ibid.) made the point that we find it most useful to store generalizations at an intermediate level of abstraction. This certainly seems to be true with the system of language itself. Thus, we store a memory of words, because our brains find these more useful to work with than the minimal unit of meaning, the morpheme. Also, their variety gives us more creative flexibility than storing our memory of language in larger chunks such as phrases.

On the other hand, we do retain some knowledge of morphemes (different for different people). Even the linguistically unsophisticated are probably aware, if asked, that -un in English means "not". Explicit morphemic memory may also vary amongst languages. For example, Chinese words are mostly two-morpheme sets, and Chinese ideographic writing basically assigns one character to each morpheme. The characters themselves are also integrated double sets (a radical plus a meaning element). The individual components of these double sets are reshuffled into thousands of word combinations, so we could expect a Chinese speaker to become thoroughly familiar with the fairly small number of base units.

In addition to words, we also retain some memory of many thousands of what I will call "loose phrases" : collections of words which we feel very often go together (May 1994). The technical name for this chumminess is collocation. Some other common names for collocation sets are chunks, formulaic phrases and clichés. Sometimes collocation is quite rigid (as in idioms), but mostly it is far more open to modification. Our feelings about collocation are mostly "instinctive" (to use that word in its street sense). That is, we are usually confident of recognizing whether two or more words comfortably go together, but find it difficult to give a reasoned argument why. We do retain some emotional bond to these collections of words.

Take the phrase "our/your/their .. feelings about what is right or wrong". All of us can probably think of real contexts where somebody might say that. In your mind's eye, you can mock up the right video scene without too much trouble. You can also shuffle this deck of words in a familiar way. For example, instead of saying "feelings about" you might say "sense of" ... and so on. A native speaker would not think of this set as a formulaic phrase, but would immediately sense the alien character of " they know the good way to do it" (a possible L2 formulation).

Traveller's phrase books offer the non-L2 speaker a way to rough communication through formulaic phrases, and there is evidence that chunking unanalyzed phrases in memory is a survival and early acquisition stratagem used by learners in a foreign language environment (Fillmore 1979, and others. I've certainly done this myself, struggling to get by in China and South Korea). In other words, the Rosch prototypes may actually change for learners over time as their language acquisition advances.


10. What Do Native Speakers Already "Know" About Grammar ?

At a level of abstraction beyond loose phrase collocations, native speakers do have some instinctual feeling for "grammar". This organic grammar is a generalization about the collocations of collocations. Somewhere in our mental processing we need this information. We need to know that the collection of words, "the man bit the dog", means something different from that other collection, "the dog was bitten by the man". However, if asked to mock up a mental video of Subject-Verb-Object, let alone the passive inversion, then all but the linguistic cognoscenti will be left speechless. This abstraction is just too far off the emotional radar. And this is in our FIRST language.

The Chomskyan perspective has always been that this knowledge of well-formedness in sentence strings establishes ipso facto that their IS a subconscious grammar. My earlier connectionist discussion implied that such sentence judgements are really just a fancy assessment of collocation weightings. Anyone who doubts their innate mathematical ability to perform at this level should stop to wonder at the subconscious calculations they make in a split second to slip their car through a small gap in fast moving traffic. This feat, like speech, comes equally to engineers and poets.

Enter the language teacher. Foreign language teachers are generally parrot-trained to repeat the following proposition : if we give our students a bunch of words plus a list of book grammar rules, then their brains will henceforth be able to produced all the sentences they need in a second language. This article has suggested that such a proposal is just wrong because our brains don't work that way, in spite of their apparently awesome subconscious computing abilities. Human brains have the utmost difficulty in making the enormous leap from those fuzzy emotional things known as words across to the bleak skyscraper scaffolding of pre-packaged abstract book grammar. There are no easy mental tracks to connect them up. What is missing is that enormous middle kingdom of collocations, the thousands of vaguely familiar phrases that we all have in our first language, and in any other language which we can genuinely manipulate.


11. The Monday Morning Question

The analysis developed in this paper might or might not be sound. It might also seem practically irrelevant to the people most affected. For a working language teacher, there is always the Monday morning question : "so how do I keep my classes busy for the next few hours, weeks, years (preferably without pre-scripting a whole movie production in my scarce leisure hours) ?" Sadly but realistically, for most teachers in most places, the answer will be working from a textbook that requires the least amount of preparation, offers the least personal risk, and packages tasks into tidy sets that can be unambiguously be marked right or wrong. This is a perfect specification for the thousands upon thousands of language learning classroom text books that clutter the foreign language sections of bookshops. In the case of English, it also fits with the vast industry of standardized language tests like TOEFL, TOEIC, IATEFL and G-TELP which are taken by employers, universities and governments as the base measure of foreign language proficiency. In East Asian countries, for example, this paradigm is integrated from pre-school infant classes through to university graduation.

What emerges from the language sausage factory are battalions of foreign language cripples, who can barely utter a phrase in their second language. Authorities wring their hands and say there must be a better way. However, there are few real plans to overcome the enormous vested interests involved in the present industry. Nor has anybody come up with a workable way to genuinely teach a foreign language to millions of students, usually in large classes, using teachers whose bottom line is the Monday morning question.


12. Conclusion

This paper cannot offer a packaged solution to the untidy problem of mass language learning failure. However, for that greater enterprise to even begin, the many interest groups involved in the language teaching industry must first understand why the pre-packaged presentation of abstract book grammars does little to help real language acquisition for most learners. I have tried to outline some reasons for that here.

Those interest groups also need a persuasive and workable substitute for the present arrangement. The "communicative language movement" has largely failed to provide that substitute, for a variety of reasons; (that could be the subject of another paper). The real solution to effective mass language teaching, if there is one, would have to be built around the psychology of how our brains acquire, remember, store, analyse and process language. There are many abstract theoretical models of language which try to explain just that process. Few of them are wholly persuasive even to professional linguists. None of them offer a working solution for ordinary language teachers.

For a layman's answer to the language learning enigma, in the end we have to turn to those people who have actually mastered a foreign language. We need to especially study those individuals who acquired language competence against all the odds, the non-analytic humanities types who never lived amongst native speakers of the target language, and somehow weathered the barrage of grammar text book exercises with little personal aptitude or liking for formal linguistic generalization.

I have tried to question groups of these successful language learners in an informal way. They are mostly contemptuous of the whole TOEIC type test culture and pedagogical grammar exercises when applied to their own learning, although as teachers many will go on to uncritically use just those materials in the classroom. As individuals they tend to find it difficult to be explicit about their own "methods". The single thing which comes through is that at some point they found the environment of foreign language learning emotionally rewarding, and sought to experience more of it. They spent a lot of time with the language, mostly as non-active participants, such as in watching videos again and again. They began to become familiar with words, not as isolates in lists, but as members of familiar phrases in emotionally interesting contexts. They were in fact learning the collocation game. Perhaps that is the base clue to our language learning enigma.


_______ End ______


References and further reading

Asher James (2004) TPR after forty years.

Ausubel D.P. (1962) A subsumption theory of meaningful verbal learning and retention. The Journal of General Psychology, 66, 213-244

Ausubel D. P., Novak J. D. & Hanesian H. (1978) Educational Psychology: A Cognitive View. 2nd ed. New York: Holt, Rinehart and Winston. Reprinted, New York: Warbel & Peck, 1986.

Chomsky N (1979) Language and Responsibility (based on conversations with Misou Ronat). New York: Panthenon Press p. 180

Christiansen M. H. and Chater N. (1999) Toward a connectionist model of recursion in human linguistic performance. Cognitive Science, 23:157--205.

Damuth Michele (2000) A brief summary of some theories of cognitive psychology with links

Fillmore Lily Wong (1979) Individual differences in second language acquisition. Charles J. Fillmore (ed.) Individual Differences in Language Ability and Language Behavior. NewYork: Academic Press.

Green Christopher D. (2001) Scientific models, connectionist networks, and cognitive science. Theory & Psychology, 11, 97-117

Hadley Gregory Sensing the winds of change: an introduction to data-driven learning.

Halliday M.A.K. (1985) An introduction to functional grammar. London ; Baltimore, Md., USA : Edward Arnold

Lee, Namhee & Schumann J. (2004) The nature of grammar : a neurobiological perspective. Applied Linguistics Association of Korea Newsletter, fall 2004

Lowe Mark (2005) is grammar innate? at

May Thor (1994) Postsupposition and pastiche talk: mediating chaos and order in language.Working Papers In Linguistics, Vol.14 : 22pp. University of Melbourne;

Meehan Paul (2005) Lexis - The new grammar? How new materials are finally challenging established course book conventions.;

Novak Joseph D The theory underlying concept maps and how to construct them.

Rae Greg (2003) Chaos Theory : A brief introduction

Rosch Eleanor (1978) Principles of categorization. E. Rosch and B. Lloyd (eds.) Cognition and Categorization, pp. 27- 48.

Rumelhart DE & McClelland JL (eds.) (1986) Parallel Distributed Processing, volume 1. MIT Press, Cambridge, MA; citation links :

Searle John (1995) The Construction of Social Reality. New York: Free Press

Smith K . (1999) Cognitive linguistics and connectionist models of language acquisition. Master thesis, Department of Linguistics, University of Edinburgh.

Snow C.P. (1959) Rede Lecture at Cambridge University :The two cultures and the scientific revolution; Wikipedia summary at ;

Thompson Thomas L. (1999) The learning theories of David P. Ausubel : the importance of meaningful and reception learning.

Wikipedia on Chaos Theory

Widmayer Sharon Alayne Schema theory: an introduction


Bio: Thor May has been teaching English to non-native speakers, and lecturing linguistics, since 1976. This work has taken him to seven countries in Oceania and East Asia, mostly with tertiary students, but with a couple of detours to teach secondary students and young children. He has trained teachers in Australia, Fiji and South Korea. His PhD is in language teaching productivity. Many of his papers, essays and stories may be seen on his website at ; e-mail .


When Grammar Doesn't Help
   copyrighted © Thorold (Thor) May 2007
   all rights reserved,
thormay AT  


The Passionate Skeptic 
[and what this website stands for ..]

Doubt well, do what you can, then let it be. Presidents, priests, wage slaves, hustlers, men and women, kids, we all live by the grace of those we love to despise...

I don't care what you believe in, so long as you don't believe in it too strongly. A belief is a weapon in the armoury of your heart, and its razor edge will murder the innocent. The ice, the fire of your passion will seduce mundane men and women. Your clarity will excite respect. And the first demagogue who comes along with a key to your heart's armoury will wrest the weapon from your moral grasp. The first cause which wears the colours of your belief will enlist you as a soldier in ravaging crusades. Peace friend. Keep your passion to doubt with. Our civilization is a simple matter of live and let live, of giving dreams a go, but stepping back with a wry smile when we get it wrong. Let the fundamentalists perish in their own pillars of fire. Spare a dollar for the living, and have a nice day. 

Thor @1 November 1991

Direct Link to Thor's Aphorisms

©2008 Thor May