I don't think that we disagree all that much. We both agree that there are some people who are able to learn structural rules implicitly without explicit instruction. We typically call these people "good at languages" or "good at music". Our main disagreement therefore, is how large that set of people is. I happen to think that it is very large given that everyone learns the grammatical rules of their first language this way, and a fair number of polyglots learn their second language this way as well (Unless you deny the usefulness of Pimsleur like approaches). If I understand you correctly, you think that the group of people who are able to properly learn a language/music this way is smaller, because it often results in bad habits and poor inferences about the structure of the language. I would endorse this as well - grammatical texts are useful for refining your understanding of the structure of a language.
Well in that case, what do you need a phrasebook for? You can, after all, learn a language simply by immersion, with nothing other than the data itself to guide you. If you're going to have any preliminary or supplementary instruction at all, it surely may as well be in an organized fashion, aimed at increasing the efficiency of the learning process by directing one toward correct theories and away from incorrect ones -- which is exactly what grammar books do and phrasebooks don't do.
Because it is scary to learn to swim without arm floats even if there is someone else helping you (I think that phrase books are analogous to arm floats). Other than that I would agree with most of this. If you want secondary instruction in a language then you should probably use a grammar book and not a phrase book and I may return to Westergaard after I have taken some composition lessons. Also I would go one step further and say that not only is it possible to learn a language via immersion, it is necessary, and any other tools you may use to learn a language should help to support this goal.
I would endorse this as well - grammatical texts are useful for refining your understanding of the structure of a language.
Tentatively-- grammatical texts have a complex relationship with language. They can be somewhat useful but still go astray because they're for a different language, with the classic example being grammar based on Latin being used to occasionally force English out of its normal use.
I suspect the same happens when formal grammar is used to claim that casual and/or spoken English is wrong.
We recently established a successful Useful Concepts Repository. It got me thinking about all the useless or actively harmful concepts I had carried around for in some cases most of my life before seeing them for what they were. Then it occurred to me that I probably still have some poisonous concepts lurking in my mind, and I thought creating this thread might be one way to discover what they are.
I'll start us off with one simple example: The Bohr model of the atom as it is taught in school is a dangerous thing to keep in your head for too long. I graduated from high school believing that it was basically a correct physical representation of atoms. (And I went to a *good* high school.) Some may say that the Bohr model serves a useful role as a lie-to-children to bridge understanding to the true physics, but if so, why do so many adults still think atoms look like concentric circular orbits of electrons around a nucleus?
There's one hallmark of truly bad concepts: they actively work against correct induction. Thinking in terms of the Bohr model actively prevents you from understanding molecular bonding and, really, everything about how an atom can serve as a functional piece of a real thing like a protein or a diamond.
Bad concepts don't have to be scientific. Religion is held to be a pretty harmful concept around here. There are certain political theories which might qualify, except I expect that one man's harmful political concept is another man's core value system, so as usual we should probably stay away from politics. But I welcome input as fuzzy as common folk advice you receive that turned out to be really costly.