Krashen (1982, 1985) was very much against grammar teaching as he felt that conscious attention to grammatical structures would not result in learners being able to use these structures in real-world communicative environments. In his Acquisition-learning Hypothesis, Krashen (1985) made a distinction between acquisition, the unconscious absorption of linguistic knowledge, and learning, the study of language rules. He believed that learning could never become acquisition, that learners would not assimilate taught structures, and that the explicit teaching of grammar was of no benefit. In other words, conscious learning through the study of grammar would be quite limited, and is unlikely to drive acquisition forward. For Krashen, meaningful interaction in the target language, with an emphasis on message transmission, and not on accurate use of rules, would lead to successful acquisition. He believed that if learners were exposed to 'comprehensible input' (ie language at a level slightly above their current ability) they could understand and use this to drive their own acquisition process forward. Krashen was not the first to note the problems of trying to learn a language exclusively via the grammar:

From what has already been said about the nature of language it will be evident that any attempt to acquire a foreign tongue by the learning of paradigms and rules must meet with failure. When the pupil appears to be learning from a rule, he is in reality learning from the example which accompanies it. (Kittson, 1918: 23)

Krashen's views influenced proponents of Communicative Language Teaching (CLT), a methodology which has as its basic premise the idea that language teaching should be message-focused and should concentrate on the development of communication skills. (In one language school where I taught, Krashen's ideas were so influential that one of our teachers would not teach language at all, and would only play Monopoly with his students. At the end of a few months of Monopoly, all of his students were proficient in saying, 'It's your turn', and 'Can I have the dice, please'!). For Miller (2002: 144) the dissemination of Krashen's views led to the orthodoxy which held sway for much of the 1980s, according to which both error correction and grammar teaching were banished (or, more accurately, forced underground; many teachers went on doing both, but felt guilty about it).

Although Krashen's ideas remain influential, and CLT is still a popular approach among teachers and researchers, his theories have been criticized as being largely untestable, and based on research that has been carried out primarily with naturalistic rather than classroom learners (see Larsen-Freeman and Long, 1991; Gregg, 1984). Indeed, it is difficult to see how we could even test Krashen's assertion that learned rules cannot become proceduralized into automatic unconscious knowledge. In addition, Krashen's emphasis on input as the driving force behind acquisition fails to consider the role of learner output (for a discussion of how learner output influences acquisition, see Swain, 1985, 1995).

From a pedagogic perspective, Grauberg (1997: 36) argues that '[t]he view, so often expressed, that one can just learn to engage in conversation "without bothering about grammar", is fallacious'. Grauberg (1997) points out that this communicative 'revolution' was not so favourably received by linguists and teachers on the Continent, and it does seem to have been more influential in English Language Teaching (ELT) than in Modern Language Teaching (MLT). Grauberg notes that

In Germany a three-phase procedure is normally followed: first a grammatical point is met within a story or a factual account [...] next it is elicited by pupils and formalized by the teacher, and finally it is practised [...]. In the Soviet Union and probably in its successors the conscious-practical method involved pupils studying the language theoretically and at the same time learning to use it in practice […]. And in France the Baccalauréat examination in foreign language still contains one paper specifically testing grammar. (Grauberg, 1997: 34)

In recent times, proponents of the use of authentic texts (AT) (Little, 1997; Little et al, 1989, and Little and Singleton, 1992) have put forward arguments not dissimilar to Krashen's: an authentic text is one devised to transmit a message in a real-world situation, and not designed for pedagogic purposes. The belief in pedagogic applications of AT derives largely from the idea that language teachers can 'better serve their students through activities in L2 communication and comprehension than by grammar practice and direct instruction' (Pica, 1997: 50-51). One of the advantages of using AT is that it 'entails a pre-requisite engagement in meaning before attention to linguistic features can be expected to be effective' (Doughty and Williams, 1998b: 3).

For Little (1997: 225), 'authentic texts have the capacity to draw language learners into the communicative world of the target language community [... and] support the communicative purpose of language teaching.' Little believes that AT acts as a substitute for a native-speaking community and thereby provides a linguistically rich environment.

It is not entirely clear what the role of grammar is in AT. Some suggest that it allows for grammar teaching in context (Larsen-Freeman, 1991), although there are few specifications as to how this can be achieved in methodological terms. In other words, although AT is a seen as a rich source of language input, there is nothing in this approach that indicates whether or how grammar should be taught explicitly.

Apart from theoretical positions against grammar teaching, there are pedagogic problems in using textbook grammar rules. We have already noted that textbook grammar rules are often partial and incomplete, and do not show how structures actually occur in authentic communication. As Willis (1990) notes, the textbook rule that 'any' should be used in negatives and interrogatives is a misleading one:

Do you smoke? Do you have any problem with your lungs?
They don't have any rights then, as prisoners.
(From COBUILD Bank of English)

He points out that 'any' is in fact often used in declarative statements:

If the number of titles on offer are any indication, Mr Parish has shown it doesn't mean to say there are any more offences

Maestri (1995: 92) provides the following example of learner output to show how this misleading rule is applied to produce error:

She didn't take any books, but dozens of them.

As Maestri, (1995: 95) states: '[...] teachers often resort to a series of oversimplifications and false generalisations which lay stress on the formal system of the language and thus gradually mislead students into approaching the language as a system of absolute rules.' Kittson noted that 'the nature of language is such that, however skilfully we devise our rules, they will fail to cover all its features. The learner will always be meeting difficulties in which no rule can guide him' (Kittson, 1918: 23).

Carter feels that 'there are many features of real, naturally occurring, spoken standard English which are not recorded in the standard grammars of English language' (Carter, 1998: 43). For Andrews, 'textbook categories are erroneous because they stress rudimentary and decontextualized structural features and not meaning as it emerges through contextualized and authentic language use' (Andrews, 1998: 30).

Most textbook rules derive from authors' intuitions, as well as pedagogic reference sources of various kinds, and as many textbook writers are themselves language teachers, they are greatly influenced by their own teaching experience, and also by their experience of language textbooks they have used themselves. What this often means is a perpetuation of the grammar list syllabus.

For many language teachers and textbook writers, the target language (TL) is epitomized by such a focus on lists of grammatical structures to be learned in succession, and these structures are very often lists of tenses, eg once the Present Continuous tense is mastered, learners are taken on to face the Present Simple, and so on. The idea is that learners should gradually accumulate a series of possibly inter-related and contrasted structures (Rutherford, 1987). Research has suggested that learning a language involves, at least to some extent, meaning-focused communication in that language, rather than just studying one structure after another (see, for example, Larsen-Freeman and Long, 1991; Skehan, 1998).

The main problems associated with traditional grammar rules and syllabus, then, are:

1. the presentation of lists of structures to be learned in sequence does not reflect second language acquisition processes (Rutherford, 1987);
2. grammar rules as presented in traditional textbooks fail to reflect how language is used authentically in real-world communication;
3. the rules are often incomplete and may mislead students.