Linguistic device acquisition in adults versus in children

In Learning to coin agent and instrument nouns, Clark and Hecht explain how children, relying on the directive of transparency, productivity, and conventionality, refine their lexicon and repertoire of word formation devices.

What I find most interesting about this developmental process is that there can be a good amount of variation across the trajectories of all children. For example, there is no order – agent-instrument, instrument-agent, or both – in which a child must begin to coin nouns with the affix –er. There may be trends, but the acquisition process is not discrete so much as conforming to the constraints of a certain model. Children must acquire mastery of derivational word formation, but it is almost as if the process is parameterized in some way.

Language acquisition is similarly parameterized – every child will have a distinct process by which they learn to speak English, or Chinese, or Swahili – but we know, and have read, that our capacity for Language is also heavily parameterized. For example, we are all born with the innate ability to speak a SOV or SVO language, a subject or topic employing language, etc., but no two adults (I would presume) share the exact same capacity for understanding and communicating in any given language known or unknown to them.

These two parameterizations, I know, are very different. But I am curious to know whether how a child learns any aspect of his native language manifests in how he speaks or communicates in the future. Will he have more difficulty with linguistic tools or methods he learns later than those he learns sooner? Or is it irrelevant, because what he is learning is not a method, but that a method can be applied to a certain context?

That being said, I am also interested in how the process by which adults learn a new language, can affect their competency in the new language. Do we have the same instinctive know-how, or do we get caught in logical pitfalls, where the (wrong though seemingly correct) assumptions we make about the manifestation of certain linguistic devices lead us astray?

I’m more confident that if we were able to use a logical model, given a certain knowledge base, that we would make minimal mistakes learning a language, but when we sacrifice precision for speed and fluidity, are there benefits to relying on (temporarily) or learning first certain rules? Children seem quick to implement the rules once they knew where and when to apply them.

But how do we define this knowledge? Can I simply tell a child he can use –er with both instruments and agents, or is it something he becomes accustomed to? How does the development of this knowledge differ in adults?