I have seen two comments on this post elsewhere, (viz. badconlangideas), and I figured I should probably address the comments.
I invite people to comment here if they have objections, questions and likewise.
Interesting! I imagine the same analysis would apply to teaching your children a natural language that you are not yourself fluent in. This is why I used to worry about bringing my children up speaking Italian, since it was important to me, but I don’t have the same level of fluency as a native speaker.
I intend to deal with that later on, in a post whose topic also is applicable to exactly that situation. We have some fairly hard data on the results of doing that, and it seems it's not recommendable. Most speakers are less spontaneous in their language use in languages they have acquired after their childhood, and this reduced spontaneity affects the interaction with children in ways that might stunt their emotional development.
But that is only a problem when you are unwilling to alter the language in reaction to what the childs does to it. When a child acquires a language it does not simply copy it but it adjusts it. By speaking with your child in your conlang it would automatically adjust it to its brain structure, overhaul the mistakes and make it a speakable language. This is how creole languages evolved. When two parents are speaking two different languages in a special environment and are mixing them up the child gets a plethora of inconsistencies and errors. But still, what the child will start to speak is a complete, elaborated and speakable language. Teaching a child a possibly incomplete language wouldn’t lead your child to have a deficiant way of thinking. The child would make your language complete automatically, as language doesn’t shape speakability but our language facility shapes language and makes it the way it needs it. Making a conlang an active language would require the step being acquired by a child as a mother tongue. That’s how Esperanto or creoles got shaped to be speakable today.
I have italicized two points that I think fail to appreciate parts of the problem.
Yes, but languages have evolved over _large sample sizes_, whereas you'll be using a sample of one or at most a dozen or so if you're a prolific breeder. However, I already pointed out that you'll need to be somewhat consistent in how you adjust your language - children will do things 'wrong' anyway ("cutted"), so just because a kid does not get your construction right away doesn't mean it's impossible to get - you can't adjust them too early (because then your kids first attempt consistently ends up deciding how your language should be), and on the other hand you have unlearnable shit going on.
Children's language acquisition is somewhat slow - even when learning natural languages, they make errors systematically at first. They fail to place the negation particle in the right place, they fail to use morphology systematically, they overgeneralize morphology ... and only after a while do they end up constructing sentences right, and the bits fall in place slowly.
So, your child not learning the right structures might not be a sign of it being unlearnable - it might be a sign of normal linguistic development. Thus, adjusting your language too quickly is a problem, as it also reduces the consistency in the data that the child learns from. But then again, you don't know whether it's learnable until you might already have stunted the kid's development. Trying to find a balance there is not a thing you should even be doing, because there's no reasonable methodology by which it could be done without risking the child's linguistic development.
As for teaching a child an incomplete language not affecting its way of thinking, we have some pretty solid evidence that this is not so. Children that are taught a language by not-very-competent speakers are more likely to be severely stunted when it comes to emotional development, as well as the ability to discuss more complicated issues. The next post, in fact, will deal with this (and will come with sources).