Although the field of language education has seen many advances during the last two decades, especially when it comes to acknowledging the roles social interaction and cultural identity play in the language teaching and learning process, I fear the same cannot be said for teaching grammar. There seem to be just as many question marks now as there were when I was in college twenty years ago. In fact, the situation may be grimmer now than it was when I first studied these issues because at least back then, there was a fair degree of certainty over what "grammar" was (those were the heydays of generative grammar), whereas today, theoretical linguists cannot even agree on whether grammar rules exist at all (at least not in any psychologically real way). This is one of the reasons I am so hesitant to teach grammar explicitly in the classroom--no matter what I tell my students, chances are that much of what I tell them would be factually inaccurate. Consequently, to the extent that I approach grammar in a direct way, I carefully avoid mention of "rules" and instead focus on helping students recognize patterns and tendencies and assigning meaning to those patterns.
Perhaps the greatest challenge when teaching grammar in any instructional context is that so much of the language acquisition process still takes place in an impenetrable black box, that it is impossible to know with any degree of certainty what the "right" way to teach grammar is. Debates that began in earnest in the 1970s are still largely unresolved today. Which approach is more effective--inductive, deductive, or a mixture of both? Is there a difference between language learning and language acquisition? How much interface, if any, is there between implicit and explicit language knowledge? Should language input or learner output be emphasized in the classroom? Do grammar rules even exist? I could go on and on. For most of these and related questions, for every published article in favor of one position there is another arguing for the opposite. And of course, the fact that I experienced naturalistic foreign language learning and found it to be far more effective than any classroom instruction I ever encountered will constitute an enduring bias against explicit grammatical instruction that I will likely never be able to erase. So it would seem then that my greatest challenge when teaching grammar is that I am conflicted over whether I should be teaching it at all, mostly because the field itself is conflicted and has no definitive answers to my questions.
Luckily for me, the curriculum used in my current teaching context adopts a strictly inductive approach to language instruction, allowing me to minimize the impact of this intellectual conflict on my instruction (it still deals an occasional glancing blow, but one parried by a methodological framework that permits evasion of grammatical explanation). According to the tenets of our method, grammatical principles should be exemplified but not explained (unless absolutely necessary for learners to understand a target structure or form), which for the most part vitiates the question of how we should teach grammar since in the strict sense our method militates against teaching grammar in the first place. According to the details of this method, as long as students notice a grammatical structure, they should be able to acquire it provided the input is comprehensible and their output is reacted to appropriately, sans mention of grammar. This is not to say that we do not provide students feedback about whether their output is grammatical, only that we do not actually say "your output is not grammatical." We negotiate form, not grammatical rules.
I strongly believe in a meaning-based approach to language instruction, for we communicate to express meaning. Form is but a means to an end, not an end in itself. Form without meaning is of almost no value (as revealed by the Corandic Activity), whereas meaning with imperfect form is often of great communicative value. Whether planned focus on form is necessary at all is debatable, in my opinion, since authentic interactions will often produce more than enough incidental focus on form to go around. As pointed out by Ellis et al. (2001), incidental focus on form, especially when initiated by the learner, appears to be more effective than planned focus on form in generating uptake, and it may be more useful to the learner since it is directly relevant to the communicative needs of the moment. The strongest argument for planned focus on form is that complete reliance on incidental focus on form may not adequately address structures or vocabulary that students intentionally avoid, and so for this reason (and for this reason alone) I can begrudgingly concede that some planned focus on form may be required depending on the circumstances. However, throughout the long course of human history people have been learning second and foreign languages naturalistically and with great success, and in such authentic learning contexts (which must numerically dwarf all combined classroom language instruction ever carried out) most if not all focus on form is of the incidental type. Yes, I admit that it is at least within the realm of possibility that at least once or twice, some Greek struggling to speak Latin in ancient Rome walked up to a native Latin speaker and asked him to explain the ablative absolute, just like a native Spanish speaker asked me last week to clarify the difference between "hassle" and "hazard," but such interactions in naturalistic language learning settings are decidedly rare. My point is that since we seem to get so much more mileage out of incidental focus on form, why take valuable class time away from meaning-centered instruction by planning a focus on form that is of questionable value to start with?