The original article is by Leo Brodie. I bought his book Starting FORTH (which is now available online) shortly after reading a 1980 Byte Magazine article on FORTH.
It's not hard to implement FORTH. I found that R. G. Loeliger's book Threaded Interpretive Languages: Their Design and Implementation allowed me to easily build my own FORTH inspired system.
It's long out of print, but used copies are cheap, and scans are readily findable on the net.
I wrote a comment a while back outlining how to do a FORTH-like language from scratch by starting with a simple calculator and expanding it, writing in C with optional assembly optimizations. This was aimed at people who have less knowledge of FORTH-like systems than you do, I think, but if you are curious here it is .
I think Forth is still useful for embedded systems, often even more so than C, especially given that it can go places that C can't. I would personally choose it for my own embedded projects, but the paradigm is sufficiently different that I would worry about finding other people who have the mental capability to write Forth that doesn't immediately devolve into something unmaintainable, or who even want to learn it in the first place. I still write little scripts with GForth sometimes too, but it's not nearly as productive as just writing some Perl or Python. If you're interested in playing with a slightly more modern concatenative language, Factor is a good choice.
I'd actually love an annotation for Python that ran it under the null-collector. Lots of times in short run or steady state programs one doesn't generate any garbage and the constant GC or ref count over head could done away with.
Did you look @ the Bob language? It was the spiritual seed for Java by David Betz http://www.xlisp.org/
I really like http://www.amazon.com/Threaded-Interpretive-Languages-Design...
Fresh book recommendations delivered straight to your inbox every Thursday.