It's hard to point to a particular "language feature" because all the differences stem from the differing evaluation strategy. Any given feature on its own may be interesting, but you can always come up with some alternative approach/representation/encoding/whatever in $YOUR_FAVORITE_LANGUAGE to diminish the feature's value. You really do need to understand term rewriting to understand the language.
This is all to say nothing of the standard library, but beyond evaluation strategy, what parts are "library" and what parts are "language" is so blurry as to be practically meaningless.
Ok, thanks. I am studying logic now, so I think I am starting to understand things related to term-rewriting. Maybe will look at Wolfram Language some day.
Do you know interesting programs/systems developed with WL?
I'm sure there are some, but think of its use cases to be more like Excel than like a typical programming environment. Most interesting Mathematica "Notebooks" are only ever used by a single person.
There are, however, lots of interesting things that most other languages don't have. For example, extremely powerful pattern matching: https://reference.wolfram.com/language/guide/Patterns.html
And a form of predicate dispatch: https://reference.wolfram.com/language/tutorial/TheOrderingO...
Or a universal data representation: http://reference.wolfram.com/language/ref/FullForm.html
Or the ability to extend the syntax by adding to a 2D operator precedence parser's dispatch tables: http://reference.wolfram.com/language/Notation/tutorial/Nota...
This is all to say nothing of the standard library, but beyond evaluation strategy, what parts are "library" and what parts are "language" is so blurry as to be practically meaningless.