...and yet, a child with no prior experience of human language can pick it up without much fuss. So what advantages does a child have over a stereotypical adult learner? I'd say these are the key points:
1. Lack of assumptions. Words often have multiple meanings, don't assume you know all ways a word is applied, instead pay attention to how context alters the meaning(s) you are already aware of.
2. Immersion, a.k.a. throwing yourself in at the deep end.
3. Openness to new information. Could also be highlighted with the word 'playfulness'. Enjoy your progress.
4. The plasticity of a human child's brain is far greater than that of an adolescent or an adult. And humans have among the most protracted childhoods in the animal kingdom.
It's not only immersion, but also sheer duration. That child takes _years_ fulltime to learn a language. Few adults are willing to make such an investment.
Language learning in children is one of the hardest to explain phenomena in psychology, linguistics and neuroscience. Trying to sum it up in three bullet points doesn't do it justice and is a gross oversimplification.
What do you find hard to understand about it? We have limited understanding of how our brains work, that I grant you, but if you're looking at the progression we follow in learning a language (i.e. at a more top level view than at the neuroscience level) then the process is much simpler to understand, it's all to do with recognition, association and conversation. Recognition in both the sense of copying sounds we hear and in recognising facets of the world around us. Association in learning to attribute certain sounds to what we experience. Conversation in learning that by making specific sounds and noticing the reaction of others, we show ourselves that the sounds have purpose in the world around us.
As our skill with language develops, so too does our self awareness, which is a whole other can of worms, but the skills we take to start off are simple to summarise, in my opinion.
Sometimes a view can be so high-level that you don't see anything meaningful anymore.
All three steps you mention are not nearly understood from either a philosophical or a biological view.
We don't know how an infant recognizes sounds: Is that some innate ability in humans or is that something that has to be learned as well? If the first, what are the basic building blocks of recognizing things? If the later, how is it learned? Think about that what we perceive as language is highly post-processed. A lot of filtering is done on sound before it becomes conscious language. Are those filtering rules innate? Are they learned? How does an infant know the difference between language and other sounds?
Association is even harder. We cannot really start to imagine how association and memory work.
Conversation is less tricky but still struggles with the question: Is it a priori knowledge we use to recognize a reaction?
All your three steps require seem obvious, but are incredibly hard to explain in full. I think you are oversimplifying the problem and only look at the most obvious parts of it. Consider a simple question: Why can infants learn language but not arithmetic given the same process you just described? Basic arithmetic is arguably a lot simpler and about as natural as language, but still won't be learned by most infants. If your answer somehow includes language as a fundamental building block of what makes us human, then you are also taking a stand on some of the things I pointed out earlier and have to unite your theories.
"Sometimes a view can be so high-level that you don't see anything meaningful anymore."
High level abstractions are valid tools for building a map of a specific domain. Please understand that having a high level map does not prevent you from exploring deeper if you find the subject interesting, nor does it rob you of the tools to do so... all you need to do is keep on asking 'Why?', 'How?', 'When?', etc... until you're satisfied.
"We don't know how an infant recognizes sounds: Is that some innate ability in humans or is that something that has to be learned as well?"
"How does an infant know the difference between language and other sounds?"
"Why can infants learn language but not arithmetic given the same process you just described?"
I grouped these questions, as I wish to answer them all together. Firstly, I would propose that the recognition of sensory stimuli is innate, which includes sounds. The roots of recognition can be traced back to the sensory stimuli we experience early on. Consider the sound of a baby screaming, which is typically the first sound they make after being born. A baby screaming in pain will typically be physically comforted shortly afterwards. Linking screaming with a subsequent response of comfort could easily be the start of our understanding of language, the senses working in tandem helping our early understanding of the world grow.
Arithmetic is more complex than the relationship between scream and comfort.
"We cannot really start to imagine how association and memory work."
We can 'start to imagine' that though, through neural networks, brain scans, etc....
Please understand, if you are interested in this subject, I appreciate other approaches (such as neuroscience) can help build a fuller picture, but high level approaches are still valid when building up intuition for a subject.
1. Lack of assumptions. Words often have multiple meanings, don't assume you know all ways a word is applied, instead pay attention to how context alters the meaning(s) you are already aware of. 2. Immersion, a.k.a. throwing yourself in at the deep end. 3. Openness to new information. Could also be highlighted with the word 'playfulness'. Enjoy your progress.