(Again I'm thinking of medicine. My own post grad work is in astronomy so I'm very much a lay reader when it comes to medicine, and when I've tried to read medical papers it's usually been an interesting excercise)
I think there are three main hurdles to comprehending scientific literature:
1) Obtuse grammar. This is universal. Why describe something in five words when you can use twenty?
2) Jargon: Every field has its jargon, and may co-opt words from the vernacular and give them very specific meanings. This gets in the way of a simplified description.
3) Intuition: Quite a lot of papers don't properly explain the intuition behind what they do. This is particularly rampant in fields that depend strongly on math. The reader is often expected to recognize the form of an equation without any explanation whatsoever. If you can do this, the intuition often turns out to be surprisingly simple. If you cannot do this (say, you're a new grad student) it looks like an impenetrable wall of Greek letters.
We can do something about (1) by journals forcing submitters to simplify their language. But fixing (2) by avoiding jargon would interfere with meaning. And fixing (3) would make papers much longer. So it's a tough problem.