The first time we learn a language is when we learn to speak as little kids. We learn the language because we constantly hear people around us speak that language, and we have no other means of verbal communication which introduces as strong reinforcement learning feedback loop. 

When we have to learn how to ask someone to give us food or to turn on the TV or to bring us a toy, we have to learn the associated grammar and the vocabulary. Otherwise, we'd be forever pointing to things and making random noises. 

As kids, we are immersed in our world and this enables us to learn the language that our parents and others around us speak in.

When learning a new language as an adult, we don't have nearly the same level of immersion with respect to that language. At the slightest sign of difficulty, we can switch back to our default language that we already communicate in. 

And our learning is primarily through translating from the language that we know. 

This is the reason one of the best ways to learn a new language is to live in a city where it is hard to get around without knowing that language, or by living with a person or people who only speak the language we wish to learn. 

Of course, it isn't always practical to do that. So, the next best thing is to simulate it. For instance, we can have specific times of day when we only speak in the language we are trying to learn, no matter how hard it gets. 

Immersing ourselves in a new environment is a sure way to learn the ways of that environment.