The input hypothesis was devised by Stephen Krashen and argues that being exposed to vast amounts of input alone is necessary and sufficient for language learning. What does "input" mean? Basically consumming media extensively, like reading books, watching series, etc. This became a very popular method of language study because it does provide amazing gains in understanding vocabulary.
The idea is similar to the way some computerized systems learn vocabulary for language aspects of AI. These systems are based on John Firth's idea that you shall know a word by the company it keeps: If I tell you I got two bottles of JURUPINGA as a birthday gift, and got super drunk, even if you never heard the word jurupinga before, you can infer through the context that it is some alcoholic beverage.
https://multithreaded.stitchfix.com/blog/2016/05/27/lda2vec/#topic=38&lambda=1&term= word2vec2.gif, downloads
But is this sufficient for learning a language? What does learning mean? The first thing we have to remember is that these AI systems are not intelligent, even if they're called "AI". The words are grounded purely to each other. There's no instinct, no emotion, no notion of truth, no conscience, no ontology, nothing beyond a probabilistic self-organization. There's a very important concept in psycholinguistics called "priming" which measures how fast you can read a word after reading something else. Some situations can heavily affect priming and some words activate others. So if I give you "it's raining cats and", you'll read "dogs" much faster. Or if I give you some text talking about loads of furniture, sofa will come much faster to your mind. Now here's the the catch that an most language learning methods fail to consider: priming is affected by WAY MORE than the immediate adjacency of other words. When following a story our brain keeps track of a lot of things. Not only our brain keeps track of words within some subject, we also follow characters, their emotional state, how events follow in time, where things happen, and a lot more stuff. All of that is also impacted by how interested we are on the text, how focused we are, but more importantly, the biggest thing here is our kwoledge of the world. To sum up, something cognitively dissonant, nonsensical or that feels unreal will take longer to grasp.
real chatbot conversation with state-of-the art computerized language models, downloads gp8.png
If I tell you that "a waiter dropped wine over a very angry boy and then the boy XXXX the waiter". You'd infer XXXX differently if I said the boy was very calm. Or you'd take longer to grasp if I kept "angry" and said he apologised to the waiter. "I had lunch and then I went for a walk" is faster to grasp than "I went for a walk after I had lunch", because in the second option the text presents the events in a different order than that of the actual sequence of events. If a character moves to a different place, objects that were in the previous place are less activated than objects in the current place, even if they are closer in number of words to the current word you're reading. There's a MULTITUDE of things that gather context.
In case of doubt, how do you know? If I point to a dog and say "Look, a wee'yar!", how do you know it means dog, mammal or animal? You don't. Our brains are likely to attribute wee'yar to a concept neither too specific, neither to general, and readjust it next time we hear it again. And that requires hearing the same word in a different context, until enough information is added that you know what it is.
But as I was saying before, world knowledge is also closely attached to all that. According to psycholinguistics, when we learn our first language, acquiring those two things are happening more or less simultaneously. Suppose I tell a child "Sweetie, could you fetch me the splork?" and point to a table containing two objects: a book, which the child knows what it is and knows it's called book; and an unknown object they had never seen before. They will make the intelligent decision to fetch me the not-book, the only thing that could be the splork.