Let's clarify one thing here; both the pro- and anti-grammar camps share the same goal: learning the grammar. It's the how that is in question. The most extreme pro-grammar position would have you start with a grammar and not do anything else until you have the grammar down. The most extreme anti-grammar position would have you parachute right into the middle of the language zone with no background and have you learn it all by observing the language in use.
Neither approach optimizes efficacy. Many high school kids in the U.S. have studied 3, 4, or even 5 years of a language, but just try talking to them and you'll realize that the hidebound grammar-centric approach most of those schools are using is a complete failure. On the other hand, learning completely by induction requires huge amounts of time and leaves gaps for things that are less commonly encountered, especially when time is limited, as it is for most language learners.
Ramses did make a good point:
It’s just a pity to see that many people in the pro-grammar and anti-grammar camp just focus on LEARN grammar or DON’T LEARN grammar, and don’t come up with alternatives.Let me see what I can do about that. My alternative, after the jump.
As mentioned above, the goal of both camps is actually how, not whether, to learn grammar. And we learn grammar rules with good reason; memorizing a short rule—say, how -ar verbs are conjugated in the present tense in Spanish—is a heckuva lot easier than memorizing individually all of permutations of all of the -ar verbs separately. This of course even applies to irregular grammatical rules, such as irregular Spanish verbs, many of whose forms are actually regular and whose irregularities even follow some patterns (such as an irregular -g- in some first-person singular irregular verbs, stem-changing verbs, etc.).
What's more, grammatical rules tend to follow patterns. and, as I recently noted in relation to music and language learning, "The brain has a strong propensity to organize information and perception in patterns". Rick Aurtus has a nice summary of how patterns help:
Your mind tends to organize the impressions it receives, and to reduce them to simple formulas wherever possible. This saves it, and you, a lot of trouble, because the knowledge that something fits into a certain pattern gives you a head-start in trying to remember it. ...This can, of course, easily be applied to languages.
How does this tendency toward pattern act upon your ability to remember? Well, for one thing, rhyming lines of poetry are more easily memorized than are sentences of prose. Words are easier to memorize than nonsense syllables. Sentences are easier to memorize than groups of unrelated words. Try naming all the letters of the alphabet, stating them at random without relying on the order in which you've learned them!
Furthermore, you'll find that it's easier to remember things in groups than singly, and less difficult to memorize lists when they are placed in alphabetical order, or in size place, or chronologically, or in any established pattern that will lend itself to your list.
The inductive method to learning grammar, however, is not the way to go. There are two reasons for this. First, the inductive method is inefficient in its use of time and effort. Second, you're likely to be left with gaps in your understanding for things that don't appear often, especially if you're time is limited, as most language learners' time is.
The inductive method is often billed as "learning like a child". What could be more simple, more effortless, than learning like a child? The problem with this is that you need to read the fine print. Think about how children learned their native language. They spent years and years in an immersion environment, and on top of that spent years and years in classes that aimed to refine their understanding of their native language. If you've got time for that kind of exposure, you can probably learn like a child, but chances are that kind of time is not available to you.
Limited exposure due to time restrictions exacerbates another problem with learning by induction; the risk that you'll miss out on less common (but necessary) rules or that you'll think you've gotten how a rule works when in fact your understanding of it is flawed. You might argue that if the rule isn't that common, why bother? I'd counter with if a native speaker knows it, so should you.
So, finally, as an alternative, I propose something of a compromise position that's somewhere in the middle of the two camps. Get the grammar rules in front of you. Read them. Organize them into a way that makes sense for you, whether in your head or otherwise. Understand them. This allows you to know what's out there and what to expect. Those declensions in Russian or those verbs in Spanish won't seem quite so mysterious, even if you forget what a particular ending means. What's more, you'll have a complete picture of any given rule, so you won't need to wonder if there are any gaps in your understanding or blatant misunderstandings that you picked up by trying to guess how the rules work.
Once you've got the rules in your head, even if only lightly so, jump into exposure. By seeing the rules in action rather than starting from a rule and going through abstract and inane drills, you'll learn how native speakers use the rules in practice and end up sounding more natural than grammar-centric learning would leave you. This beats learning like a child because you don't need to fumble around for the rules, but you do get the same exposure. Repeated exposure will then help cement the rules. If a rule gets fuzzy, go back to the rule and review it, and then go right back to the exposure. Then all you need to do going forward is to repeat this process. This will result in an ad-hoc spaced-repetition system that will eventually result in you knowing the grammar rules, and with a lot less time spent than using an inductive method.
To give an example of this approach, I'll actually turn to Josh of Language Geek. Although Josh's post would seem at first glance to put him squarely in the anti-grammar camp, he's actually practicing exactly what I'm preaching:
I’m finding that I grasp grammar more fully after learning the grammar points via the Penguin course, and then seeing the grammar in use repeatedly in the Assimil course.Bingo.