The world is a difficult place for a language learner. With so much advice out there, how do you know which to take? Unfortunately, you can't take anyone's advice without a pinch of salt. The usual response to this is to use your "common sense" or your "critical faculties", but neither of those is going to get us very far.
Before getting into the meat of the question, I'd like to quote Marcel Pagnol:
<<Telle est la faiblesse de notre raison : elle ne sert le plus souvent qu'à justifier nos croyances.>>
(Such is the weakness of our reason: most often it serves only to justify our beliefs.)
Pagnol wrote this in his autobiography La gloire de mon père, in reference to the rigorous debates between anti-clerical schoolmasters and the clergy, something of an everyday occurrence in post-revolutionary France. Both sides were well educated and well read, and both could set forth a good argument, but for the most part their arguments were always built on selective evidence.
This is a problem that bedevils research even to this day. In writing a paper, many researchers will quote research that supports their view, and will only cite research that disagrees with it when they can counter the point raised.
It the field of language learning, this is a particularly vexing problem. People on a regular basis go back and cite sources from the 60s and 70s, completely ignoring decades of research that run counter to it. Here our critical faculties as individuals fail us: it is not that we cannot be critical, but we are not presented with all the relevant information. Even if it was, we would be incapable of processing all the relevant information due to the sheer volume of research carried out.
We can be "blinded by science".
So much for critical reasoning. What about common sense?
Well, if it really was a question of "common sense", you wouldn't even need to ask, right? Common sense is just another word for a person's beliefs. At one time the existence of gods was common sense, and today the argument for or against gods boils down to the same thing -- all sides consider their view "common sense" and "logical".
Nope, common sense is no good, and reasoning can trick you, so you have to be careful.
Regular readers will know that my favourite piece on language is Wilfried Decoo's lecture On the mortality of language learning methods. One of the key points of Decoo's argument is that most methods have a broad basis of similarity, and differ in the inclusion or exclusion of one or two particular features, or even just in declaring that one particular feature is made more prominent.
In short, for all their arguing, most methods are incredibly similar. For example, Assimil claims that you doesn't teach rules, and that you learn by "natural assimilation", yet more of a typical Assimil book is dedicated to grammatical explanations than dialogues. On the other hand, any grammar-heavy course may say that the rules are the important bit, but most use lots of examples. Even the proponents of grammar-free "natural" learning produce courses with a structured introduction to grammatical features.
Problem 1: One all-important concept
Decoo says, "A new method draws its originality and its force from a concept that is stressed above all others. Usually it is an easy to understand concept that speaks to the imagination." Traditionally, this could be disregarded simply as marketing, and not directly harmful to the student. However, the internet has changed that, and these claims are becoming downright dangerous.
Well, people are going out and trying to recreate these methods for themselves, not based on the content of the methods, but on these stated principles. So while Krashen advises lots of listening, he offers a structured lesson. Yet a self-teacher can't produce a structured lesson, and listening to poorly selected material won't get you anywhere.
Problem 2: Nobody really knows what they're doing
As a general rule, when we're involved in doing something, we're not normally aware of exactly what it is we're doing. For example, can you describe how you walk? How you ride a bike? Probably only very superficially. So how are we to trust someone's claims about how they learn languages.
Aside from this general observation, we've got another problem -- the human brain can make insignificant things seem very significant indeed.
For example, if you live in a big city, you'll walk past thousands of people in a week. Yet run into one old schoolmate, and you'll comment on how unlikely it is, or how it's a small world. Statistically, that one friend is insignificant, but psychologically, that friend is more significant than every stranger in your town.
This phenomenon inhibits our individual ability to evaluate the effectiveness of our learning techniques. If you learn 3 things from a given technique and forget 300, by definition you only remember the things you learned. The 3 things become used as proof that you learned effectively from that technique.
In fact, I find that if I can remember a few specific examples that I learned in a particular way, it usually means quite the opposite: if a technique is effective, I rarely remember specific examples of things I learned with it. If I remember one or two examples, they're often the only things I learned well using that technique. So quite unhelpfully, the things that stick in my mind are actually the least helpful techniques. (I previously wrote about a similar phenomenon: the unspoken value of student feedback.) (This is part of something called recall bias, and it is one of the reasons very little science relies on survey responses these days.)
As a consequence, I'm generally sceptical when someone advises something as what they do, because, quite simply, how do they know?