Why choose R for an implementation?

Mostly because I'm familiar and comfortable with it. A standalone application (C or python or whatever) is not really an option at this point as a lot of time (more than what I have) would have to be spent reimplementing basic tools that are already available in R.

Installing R

Details on how to get R are available from CRAN. Your distribution may have a packaged version already. To compile R yourself on a Linux/UNIX platform, my suggestion would be to get R via rsync (so that future incremental updates are easy to make). For instance

$ cd /opt/
$ mkdir r-src
$ mkdir r-build
$ rsync -avz --delete franz.stat.wisc.edu::r-devel/ r-src/
(will download the R sources)
$ cd r-src/
$ ./tools/rsync-recommended
$ cd ../r-build
$ ../r-src/configure
$ make

I would typically not run make install, but instead create a link in my PATH to /opt/r-build/bin/R

Anyway, please email me if you need more help with this.