In English, unstressed words compress. “What do you want to do” doesn't come out as five separate words: it becomes something like “whaddya wanna do”, two syllables shorter with the auxiliary gone entirely. This isn't laziness or bad speech; it's how the rhythm of the language works, and everyone does it — in a job interview and on the news alike.
That is why you can read a Guardian article without effort and lose the thread in a pub conversation. Your vocabulary is fine. What fails is that you learned the words in their careful form, nobody taught you their fast form, and your ear is waiting for something that never arrives.
The good news is that reductions are a closed and fairly small set. They aren't infinite and they aren't random: there are about forty, they follow rules, and once you've consciously caught a few of them, the ear starts doing it on its own. Below they're grouped by what happens to the sound.
Why you can't understand spoken English even though you know the words →
Fast English dictation test: type what you hear and find out which words your ear drops →