What is the benefit of using Julia over R?
You can use R and Python packages from Julia. You can call Fortran and C code from it. Unlike R and Python you don't need to have a module rewritten to C or Fortran code to get high enough performance.
In fact Julia will often outperform C and Fortran code. There are researchers which have switched from Fortran to Julia to get a higher productivity programming language expecting to get a performance drop only to realize they got a 3x performance increase.
The main downside of Julia is that it is a very new language. R and Python and old and hence well known, but Julia is a rapidly growing and taking over serious areas. Next generation climate models are built using Julia e.g.
I've used R for a lot of data-visualization (and become familiar with the idiosyncrasies of many packages) so it's nice to know that it's compatible.
Julia really demonstrates the difference between "theoretically fast" and "actually fast in practice". In theory, maybe I could outperform my Julia program in C with careful thought about cache strategies and so forth, the same way I could outperform my C code with hand tuned assembly. In practice, the ability to prepend a 'for' loop with 'Threads.@threads' and suddenly the loop runs in parallel - or the ability to declare an array as type 'cuArray' and suddenly all operations on it run on the GPU - means that it's vastly easier for me to write fast Julia code than fast C code.